Diabetes is one of the most common diagnoses in hospitalized patients.1, 2 Hyperglycemia is present in 38% of adults admitted to the hospital, one third of whom had no history of diabetes before admission.3 The impact of inpatient hyperglycemia on clinical outcome in adult patients has been increasingly appreciated. Extensive evidence from observational studies indicates that hyperglycemia in patients with or without a history of diabetes is an important marker of poor clinical outcome.312 Several prospective randomized trials in patients with critical illness have shown that aggressive glycemic control improves short‐ and long‐term mortality, multiorgan failure and systemic infection, and length of hospitalization.1317 The importance of glucose control also applies to adult patients admitted to general surgical and medical wards.3, 6, 18 In such patients, we recently reported that the presence of hyperglycemia is associated with prolonged hospital stay, infection, disability after hospital discharge, and death.3, 6, 18 Despite the extensive data in adult patients, there is little information on the impact of inpatient hyperglycemia in pediatric patients. The few observational studies in critically ill children admitted to the pediatric ICU with severe brain injury or extensive burn injuries have shown a positive association between inpatient hyperglycemia and increased length of hospital and ICU stay and a higher risk of complication and mortality rates.1923 No previous studies, however, have examined the association of hyperglycemia and clinical outcome in children admitted to a general community pediatric hospital. Therefore, in this study we determined the prevalence of inpatient hyperglycemia and examined the impact of hyperglycemia on morbidity and mortality in children admitted to Hughes Spalding Children's Hospital, a large community hospital serving the inner city and indigent pediatric population in Atlanta, Georgia.

MATERIALS AND METHODS

This was a retrospective observational cohort of pediatric patients consecutively admitted to Hughes Spalding Children's Hospital in Atlanta from January 2004 to August 2004. This general community pediatric hospital is part of the Grady Health System in Atlanta, a large health care organization that operates under the auspices of the Fulton‐Dekalb Hospital Authoritythe major counties in metropolitan Atlantato deliver care to their uninsured and underserved populations. Ninety percent of the organization's inpatient cases are either uninsured or dependent on Medicaid. This is a broad‐based pediatric hospital without cardiac surgery, burn, or dedicated inpatient hematology‐oncology units. Patients are managed by members of the pediatric residency program and supervised by faculty members from Emory University School of Medicine. The Institutional Review Board of Emory University and Grady Health System Oversight Research Committee approved the methods for data collection and analysis used in the study and waived the need for informed consent.

The medical records of 903 consecutive pediatric patients admitted to both critical and noncritical care areas were reviewed. For the analysis, patients were divided according to a known history of diabetes prior to admission and according to admission blood glucose concentration. A normoglycemic group included patients with normal plasma glucose and without a history of diabetes. Serum or plasma glucose measured in the laboratory was assumed to be equivalent to blood glucose measured by finger stick at bedside using a glucose meter. Hyperglycemia was defined as an admission or in‐hospital blood glucose level >120 mg/dL. High blood glucose was subsequently divided into those with blood glucose of 120179 mg/dL and those with blood glucose 180 mg/dL. Patient information was collected regarding demographic characteristics, blood glucose level on admission and during hospital stay, concurrent medical diagnoses, medical treatment, and hospital outcome (including mortality and disposition at discharge).

The primary objectives of this study were to determine the prevalence of in‐hospital hyperglycemia and to examine the association of hyperglycemia and mortality in children with critical and noncritical illness in a community pediatric hospital. Secondary end points included length of hospital stay, requirement of intensive care, and treatment of hyperglycemia. In addition to blood glucose level, prognostic variables included sex, age, body mass index, admission diagnosis, presence of comorbidities, and intensive care unit admission.

RESULTS

Of the 903 admitted patients, 342 patients (38%) had no blood glucose measurement during the hospital stay and were excluded from the analysis. Three patients with a length of stay greater than 6 months were excluded. In addition, 16 patients admitted with diabetic ketoacidosis (DKA) and 1 subject with hyperglycemic hyperosmolar syndrome were also excluded from the analysis. The remaining 542 patients constituted the study population. Most of these, 406 patients (75%), had an admission blood glucose concentration 120 mg/dL (mean SEM 98 1 mg/dL, median 93 mg/dL). A total of 103 children (19%) had an admission blood glucose level of 121179 mg/dL (mean 143 2 mg/dL, median 140 mg/dL), and 32 patients (5.9%) had an admission blood glucose level >180 mg/dL (mean 260 18 mg/dL, median 211 mg/dL; Fig. 1).

Figure 1

The clinical characteristics of study patients are shown in Table 1. Most patients in this study were from minority ethnic groups82% were black, 12% were Hispanic, 2% were from other minority groups, and 4.2% were white. There were no significant differences in mean age, sex, racial distribution, or body mass index among the 3 groups. A total of 409 patients (75.5%) were admitted to general pediatric wards and 133 patients (24.5%) were admitted to the surgical unit. There were no differences in the admission blood glucose between patients admitted to general pediatric wards (112.2 mg/dL) and those admitted to surgical areas (115.7 mg/dL, P > .05). The most common diagnoses in the severe hyperglycemia group were trauma/surgery (25%), pulmonary disease (18.8%), metabolic disorders (12.5%), and infection (6.3%). Most children admitted with hyperglycemia had no history of diabetes prior to admission. Among the 135 children with admission hyperglycemia (blood glucose >120 mg/dL), 17 patients (13%) had a known history of diabetes or were receiving therapy prior to admission. The mean admission blood glucose was 162.4 mg/dL (range 121480 mg/dL) in children with new hyperglycemia and 369.8 mg/dL (range 145678 mg/dL) in those children with a known history of diabetes (P < .01). Among children without a history of diabetes, 33 of 118 children (28%) with admission hyperglycemia had 1 or more glucose values >120 mg/dL during their hospitalizations. Twenty‐five children had a blood glucose of 121179 mg/dL (mean 109 5 mg/dL), and 8 children had a blood glucose 180 mg/dL (mean 159 13 mg/dL). Most patients with a history of diabetes were admitted with significant hyperglycemia. One patient (1%) had a glucose level in the 121179 mg/dL category, and 16 patients (50%) had a glucose level >180 mg/dL.

The presence of hyperglycemia on admission in pediatric patients was not associated with increased mortality or with increased length of hospital stay. There was only 1 death reported during the study period, which occurred in a patient with respiratory failure because of bronchiolitis who was admitted with an admission blood glucose of 151 mg/dL. The mean length of stay for patients with normoglycemia was 3.83 0.2 days, which increased to 5.36 1.0 and 5.68 1.8 days for children with blood glucose of 120179 and 180 mg/dL, respectively (P > .05).

Children with hyperglycemia were more likely to be admitted to the ICU and had a longer length of ICU stay. Admission to the ICU was needed by 10% of children with an admission blood glucose <120 mg/dL, 18% of children with a blood glucose of 120179 mg/dL, and 40% of children with an admission blood 180 mg/dL (P < .01). In addition, length of ICU stay was significantly longer for hyperglycemic children, particularly those with a glucose level 180 mg/dL (P < .001). The mean length of ICU stay (ICU) was 0.56 0.1 days for patients with normoglycemia, and 1.1 0.4 days and 3.6 1.9 days for patients with a blood glucose of 120179 and 180 mg/dL, respectively (P < .01).

Newly diagnosed hyperglycemia was frequently left untreated. Only 3 children without a history of diabetes but with hyperglycemia recorded during the hospital stay received insulin therapy. New hyperglycemia patients received regular insulin per a sliding scale as the main insulin regimen in the hospital. In contrast, all patients with a previous history of diabetes were treated with insulin during their hospital stay.

DISCUSSION

Diabetes mellitus represents a significant public health burden on the basis of increased morbidity, mortality, and economic costs. Increasing evidence from observational and prospective interventional studies has shown that inpatient hyperglycemia is a predictor of poor clinical outcome of adult subjects.313, 16, 17 Admission hyperglycemia has been associated with increased morbidity and mortality in patients with critical illness, as well as in noncritically ill adult subjects admitted to general surgical and medical wards.3, 6, 18 In this study we also found that hyperglycemia is a common finding in children admitted with critical and noncritical illnesses and that most children had no history of diabetes before admission. One‐fourth of the children admitted to the hospital had hyperglycemia on admission. Children with hyperglycemia were more likely to be admitted to the ICU and had a longer length of ICU stay; however, inpatient hyperglycemia was not associated with higher hospital mortality or longer hospital stay than was inpatient normoglycemia. Our findings suggest that recognition of inpatient hyperglycemia can be improved because screening for hyperglycemia was not performed in more than one third of patients (38%) during the hospital stay.

The prevalence of inpatient hyperglycemia in children varies according to the severity of the illness and the study population. Ruiz Magro et al.21 reported that 50% of 353 critically ill children without diabetes mellitus had initial glucose values >120 mg/dL. In a study of 942 nondiabetic patients, Faustino et al.20 found that within 24 hours of admission to the ICU, hyperglycemia was prevalent in 70.4% of patients with a glucose value >120 mg/dL, 44.5% of patients with a glucose value >150 mg/dL, and 22.3% of patients with a glucose value >200 mg/dL. The prevalence of hyperglycemia in non‐critically ill children seen in the emergency department was much lower, ranging from 3.8% to 5.0% (based on an initial blood glucose >150 mg/dL).19, 24 In agreement with these studies, we found inpatient hyperglycemia to be a common finding among hospitalized children. Approximately 75% of our patients had a normal blood glucose on admission, 19% had an admission blood glucose of 121179 mg/dL (mean 143 2 mg/dL), and 5.9% of children had an admission blood glucose 180 mg/dL (mean 260 18 mg/dL). Only 13% of our patients had a known history of diabetes prior to admission, suggesting that the hyperglycemia was a result of the stress of the medical illness or the surgery. Stress hyperglycemia, defined as a transient increase in blood glucose level during acute physiological stress, has been reported to occur in 4% of children with an acute non‐critical illness and in more than 50% of children in the ICU.

A few studies have reported on the impact of inpatient hyperglycemia in children with acute critical illness.1015 Three retrospective studies have demonstrated that admission hyperglycemia is also a predictor of adverse outcomes in the pediatric intensive care unit.20, 22 Srinivasan and colleagues22 demonstrated that 86% of patients in their pediatric intensive care unit had a glucose value >126 mg/dL at some point during their stay. In addition, they showed that duration of the hyperglycemia and peak glucose were also associated with mortality. Faustino and Apkon20 demonstrated that hyperglycemia occurs frequently among critically‐ill nondiabetic children and is correlated with a greater in‐hospital mortality rate and longer length of stay in the ICU. They reported a 2.5‐fold increased risk of dying if the maximum glucose obtained within 24 hours of admission to the ICU was >150 mg/dL. More recently, Yates et al.25 reported that hyperglycemia in the postoperative period was associated with increased morbidity and mortality in postoperative pediatric cardiac patients. Other studies in children with traumatic brain or head injury have also shown an association between poor neurological outcome and elevated admission blood glucose.24, 2628 Brain trauma patients with permanent neurological deficits and in a vegetative state were found to have significantly higher admission blood glucose concentrations than children with good neurological recovery or minimal deficits. In addition, the development of inpatient hyperglycemia in children with extensive burn injuries, covering more than 60% of total body surface area, was found to increase the risk of bacteremia and fungemia, reduce skin graft adhesion, and increase the mortality rate.29 These data show an association of initial glucose, peak glucose, and duration of hyperglycemia with increased incidence of morbidity and mortality in children with acute critical illness. We found no association between initial blood glucose and risk of death. This is in contrast to our previous results in adult patients, in whom inpatient hyperglycemia was found to represent an important marker of increased morbidity and mortality among both those critically ill and not critically ill.3 It is important to note that the overall mortality rate reported in children with hyperglycemia relates to severity of illness and is significantly lower than that of adults.30 In most critically ill pediatric series, hospital mortality ranges from 2% to 5.3% and is higher in patients with severe trauma and those who underwent major cardiac surgery.23, 31 The mortality in children without critical illness admitted to general pediatric wards is significantly lower.30

In agreement with the increasing rate of obesity among children with diabetes,32, 33 especially in minority populations, we found that hospitalized children with a history of diabetes and glucose >180 mg/dL had a higher body mass index than those with normoglycemia (P < .001). Obesity in children has been associated with the presence of several comorbidities and an increased risk of hospital complications.34, 35 There is also increasing evidence among patients admitted to the intensive care unit that obesity contributes to increased morbidity and to a prolonged length of stay.35 Because they have a higher rate of hyperglycemia, diabetes, and hospital complications, we believe that obese children should be screened for hyperglycemia and diabetes.

We acknowledge the following limitations of this study. The main limitation was its retrospective nature. The method of blood glucose collection and analysis was not standardized; thus, it prevented uniformity in the determination of serum glucose values of individual patients. We arbitrarily used 3 glucose cutoff values in this study (<120, 120179, and >180 mg/dL). Although similar values have been used in inpatient diabetes studies,2022 there is no uniform definition of hyperglycemia in hospitalized patients, and the clinical significance of these cutoff values in pediatric population has not been determined. The study was conducted in a single institution in Atlanta, whose population and disease spectrum might be different from those at other pediatric institutions. Our study did not address the question of whether treatment of hyperglycemia might improve the outcome of length of hospital stay of patients with hyperglycemia. We believe that newly diagnosed hyperglycemia is usually considered a transient finding in response to acute illness not requiring medical intervention, as indicated by the fact that more than half of these patients did not receive antidiabetic therapy. Another limitation of our study is that we were not able to determine the percentage of patients with latent or unrecognized diabetes because of the lack of hemoglobin A1C testing and follow‐up after discharge. A prospective, randomized trial of strict glycemic control is certainly needed to address these issues.

In summary, inpatient hyperglycemia is a common finding in children with and without critical illness. One‐fourth of the children admitted to the hospital had hyperglycemia, most of them without a history of diabetes prior to admission. Although we found a higher need for ICU admission and a longer length of ICU stay, hyperglycemia in pediatric patients was not associated with higher hospital mortality compared with that in children with normoglycemia. Several observational studies have reported an association of hyperglycemia with poor clinical outcome in critically ill children; however, no prospective controlled studies have assessed the effect of tight glucose control in pediatric populations. These studies need to be prospective, randomized multicenter trials of sufficient magnitude to provide a well‐powered analysis to enable multiple observations and evaluation of subsets of critically and non‐critically ill pediatric patients.

Diabetes is one of the most common diagnoses in hospitalized patients.1, 2 Hyperglycemia is present in 38% of adults admitted to the hospital, one third of whom had no history of diabetes before admission.3 The impact of inpatient hyperglycemia on clinical outcome in adult patients has been increasingly appreciated. Extensive evidence from observational studies indicates that hyperglycemia in patients with or without a history of diabetes is an important marker of poor clinical outcome.312 Several prospective randomized trials in patients with critical illness have shown that aggressive glycemic control improves short‐ and long‐term mortality, multiorgan failure and systemic infection, and length of hospitalization.1317 The importance of glucose control also applies to adult patients admitted to general surgical and medical wards.3, 6, 18 In such patients, we recently reported that the presence of hyperglycemia is associated with prolonged hospital stay, infection, disability after hospital discharge, and death.3, 6, 18 Despite the extensive data in adult patients, there is little information on the impact of inpatient hyperglycemia in pediatric patients. The few observational studies in critically ill children admitted to the pediatric ICU with severe brain injury or extensive burn injuries have shown a positive association between inpatient hyperglycemia and increased length of hospital and ICU stay and a higher risk of complication and mortality rates.1923 No previous studies, however, have examined the association of hyperglycemia and clinical outcome in children admitted to a general community pediatric hospital. Therefore, in this study we determined the prevalence of inpatient hyperglycemia and examined the impact of hyperglycemia on morbidity and mortality in children admitted to Hughes Spalding Children's Hospital, a large community hospital serving the inner city and indigent pediatric population in Atlanta, Georgia.

MATERIALS AND METHODS

This was a retrospective observational cohort of pediatric patients consecutively admitted to Hughes Spalding Children's Hospital in Atlanta from January 2004 to August 2004. This general community pediatric hospital is part of the Grady Health System in Atlanta, a large health care organization that operates under the auspices of the Fulton‐Dekalb Hospital Authoritythe major counties in metropolitan Atlantato deliver care to their uninsured and underserved populations. Ninety percent of the organization's inpatient cases are either uninsured or dependent on Medicaid. This is a broad‐based pediatric hospital without cardiac surgery, burn, or dedicated inpatient hematology‐oncology units. Patients are managed by members of the pediatric residency program and supervised by faculty members from Emory University School of Medicine. The Institutional Review Board of Emory University and Grady Health System Oversight Research Committee approved the methods for data collection and analysis used in the study and waived the need for informed consent.

The medical records of 903 consecutive pediatric patients admitted to both critical and noncritical care areas were reviewed. For the analysis, patients were divided according to a known history of diabetes prior to admission and according to admission blood glucose concentration. A normoglycemic group included patients with normal plasma glucose and without a history of diabetes. Serum or plasma glucose measured in the laboratory was assumed to be equivalent to blood glucose measured by finger stick at bedside using a glucose meter. Hyperglycemia was defined as an admission or in‐hospital blood glucose level >120 mg/dL. High blood glucose was subsequently divided into those with blood glucose of 120179 mg/dL and those with blood glucose 180 mg/dL. Patient information was collected regarding demographic characteristics, blood glucose level on admission and during hospital stay, concurrent medical diagnoses, medical treatment, and hospital outcome (including mortality and disposition at discharge).

The primary objectives of this study were to determine the prevalence of in‐hospital hyperglycemia and to examine the association of hyperglycemia and mortality in children with critical and noncritical illness in a community pediatric hospital. Secondary end points included length of hospital stay, requirement of intensive care, and treatment of hyperglycemia. In addition to blood glucose level, prognostic variables included sex, age, body mass index, admission diagnosis, presence of comorbidities, and intensive care unit admission.

RESULTS

Of the 903 admitted patients, 342 patients (38%) had no blood glucose measurement during the hospital stay and were excluded from the analysis. Three patients with a length of stay greater than 6 months were excluded. In addition, 16 patients admitted with diabetic ketoacidosis (DKA) and 1 subject with hyperglycemic hyperosmolar syndrome were also excluded from the analysis. The remaining 542 patients constituted the study population. Most of these, 406 patients (75%), had an admission blood glucose concentration 120 mg/dL (mean SEM 98 1 mg/dL, median 93 mg/dL). A total of 103 children (19%) had an admission blood glucose level of 121179 mg/dL (mean 143 2 mg/dL, median 140 mg/dL), and 32 patients (5.9%) had an admission blood glucose level >180 mg/dL (mean 260 18 mg/dL, median 211 mg/dL; Fig. 1).

Figure 1

The clinical characteristics of study patients are shown in Table 1. Most patients in this study were from minority ethnic groups82% were black, 12% were Hispanic, 2% were from other minority groups, and 4.2% were white. There were no significant differences in mean age, sex, racial distribution, or body mass index among the 3 groups. A total of 409 patients (75.5%) were admitted to general pediatric wards and 133 patients (24.5%) were admitted to the surgical unit. There were no differences in the admission blood glucose between patients admitted to general pediatric wards (112.2 mg/dL) and those admitted to surgical areas (115.7 mg/dL, P > .05). The most common diagnoses in the severe hyperglycemia group were trauma/surgery (25%), pulmonary disease (18.8%), metabolic disorders (12.5%), and infection (6.3%). Most children admitted with hyperglycemia had no history of diabetes prior to admission. Among the 135 children with admission hyperglycemia (blood glucose >120 mg/dL), 17 patients (13%) had a known history of diabetes or were receiving therapy prior to admission. The mean admission blood glucose was 162.4 mg/dL (range 121480 mg/dL) in children with new hyperglycemia and 369.8 mg/dL (range 145678 mg/dL) in those children with a known history of diabetes (P < .01). Among children without a history of diabetes, 33 of 118 children (28%) with admission hyperglycemia had 1 or more glucose values >120 mg/dL during their hospitalizations. Twenty‐five children had a blood glucose of 121179 mg/dL (mean 109 5 mg/dL), and 8 children had a blood glucose 180 mg/dL (mean 159 13 mg/dL). Most patients with a history of diabetes were admitted with significant hyperglycemia. One patient (1%) had a glucose level in the 121179 mg/dL category, and 16 patients (50%) had a glucose level >180 mg/dL.

The presence of hyperglycemia on admission in pediatric patients was not associated with increased mortality or with increased length of hospital stay. There was only 1 death reported during the study period, which occurred in a patient with respiratory failure because of bronchiolitis who was admitted with an admission blood glucose of 151 mg/dL. The mean length of stay for patients with normoglycemia was 3.83 0.2 days, which increased to 5.36 1.0 and 5.68 1.8 days for children with blood glucose of 120179 and 180 mg/dL, respectively (P > .05).

Children with hyperglycemia were more likely to be admitted to the ICU and had a longer length of ICU stay. Admission to the ICU was needed by 10% of children with an admission blood glucose <120 mg/dL, 18% of children with a blood glucose of 120179 mg/dL, and 40% of children with an admission blood 180 mg/dL (P < .01). In addition, length of ICU stay was significantly longer for hyperglycemic children, particularly those with a glucose level 180 mg/dL (P < .001). The mean length of ICU stay (ICU) was 0.56 0.1 days for patients with normoglycemia, and 1.1 0.4 days and 3.6 1.9 days for patients with a blood glucose of 120179 and 180 mg/dL, respectively (P < .01).

Newly diagnosed hyperglycemia was frequently left untreated. Only 3 children without a history of diabetes but with hyperglycemia recorded during the hospital stay received insulin therapy. New hyperglycemia patients received regular insulin per a sliding scale as the main insulin regimen in the hospital. In contrast, all patients with a previous history of diabetes were treated with insulin during their hospital stay.

DISCUSSION

Diabetes mellitus represents a significant public health burden on the basis of increased morbidity, mortality, and economic costs. Increasing evidence from observational and prospective interventional studies has shown that inpatient hyperglycemia is a predictor of poor clinical outcome of adult subjects.313, 16, 17 Admission hyperglycemia has been associated with increased morbidity and mortality in patients with critical illness, as well as in noncritically ill adult subjects admitted to general surgical and medical wards.3, 6, 18 In this study we also found that hyperglycemia is a common finding in children admitted with critical and noncritical illnesses and that most children had no history of diabetes before admission. One‐fourth of the children admitted to the hospital had hyperglycemia on admission. Children with hyperglycemia were more likely to be admitted to the ICU and had a longer length of ICU stay; however, inpatient hyperglycemia was not associated with higher hospital mortality or longer hospital stay than was inpatient normoglycemia. Our findings suggest that recognition of inpatient hyperglycemia can be improved because screening for hyperglycemia was not performed in more than one third of patients (38%) during the hospital stay.

The prevalence of inpatient hyperglycemia in children varies according to the severity of the illness and the study population. Ruiz Magro et al.21 reported that 50% of 353 critically ill children without diabetes mellitus had initial glucose values >120 mg/dL. In a study of 942 nondiabetic patients, Faustino et al.20 found that within 24 hours of admission to the ICU, hyperglycemia was prevalent in 70.4% of patients with a glucose value >120 mg/dL, 44.5% of patients with a glucose value >150 mg/dL, and 22.3% of patients with a glucose value >200 mg/dL. The prevalence of hyperglycemia in non‐critically ill children seen in the emergency department was much lower, ranging from 3.8% to 5.0% (based on an initial blood glucose >150 mg/dL).19, 24 In agreement with these studies, we found inpatient hyperglycemia to be a common finding among hospitalized children. Approximately 75% of our patients had a normal blood glucose on admission, 19% had an admission blood glucose of 121179 mg/dL (mean 143 2 mg/dL), and 5.9% of children had an admission blood glucose 180 mg/dL (mean 260 18 mg/dL). Only 13% of our patients had a known history of diabetes prior to admission, suggesting that the hyperglycemia was a result of the stress of the medical illness or the surgery. Stress hyperglycemia, defined as a transient increase in blood glucose level during acute physiological stress, has been reported to occur in 4% of children with an acute non‐critical illness and in more than 50% of children in the ICU.

A few studies have reported on the impact of inpatient hyperglycemia in children with acute critical illness.1015 Three retrospective studies have demonstrated that admission hyperglycemia is also a predictor of adverse outcomes in the pediatric intensive care unit.20, 22 Srinivasan and colleagues22 demonstrated that 86% of patients in their pediatric intensive care unit had a glucose value >126 mg/dL at some point during their stay. In addition, they showed that duration of the hyperglycemia and peak glucose were also associated with mortality. Faustino and Apkon20 demonstrated that hyperglycemia occurs frequently among critically‐ill nondiabetic children and is correlated with a greater in‐hospital mortality rate and longer length of stay in the ICU. They reported a 2.5‐fold increased risk of dying if the maximum glucose obtained within 24 hours of admission to the ICU was >150 mg/dL. More recently, Yates et al.25 reported that hyperglycemia in the postoperative period was associated with increased morbidity and mortality in postoperative pediatric cardiac patients. Other studies in children with traumatic brain or head injury have also shown an association between poor neurological outcome and elevated admission blood glucose.24, 2628 Brain trauma patients with permanent neurological deficits and in a vegetative state were found to have significantly higher admission blood glucose concentrations than children with good neurological recovery or minimal deficits. In addition, the development of inpatient hyperglycemia in children with extensive burn injuries, covering more than 60% of total body surface area, was found to increase the risk of bacteremia and fungemia, reduce skin graft adhesion, and increase the mortality rate.29 These data show an association of initial glucose, peak glucose, and duration of hyperglycemia with increased incidence of morbidity and mortality in children with acute critical illness. We found no association between initial blood glucose and risk of death. This is in contrast to our previous results in adult patients, in whom inpatient hyperglycemia was found to represent an important marker of increased morbidity and mortality among both those critically ill and not critically ill.3 It is important to note that the overall mortality rate reported in children with hyperglycemia relates to severity of illness and is significantly lower than that of adults.30 In most critically ill pediatric series, hospital mortality ranges from 2% to 5.3% and is higher in patients with severe trauma and those who underwent major cardiac surgery.23, 31 The mortality in children without critical illness admitted to general pediatric wards is significantly lower.30

In agreement with the increasing rate of obesity among children with diabetes,32, 33 especially in minority populations, we found that hospitalized children with a history of diabetes and glucose >180 mg/dL had a higher body mass index than those with normoglycemia (P < .001). Obesity in children has been associated with the presence of several comorbidities and an increased risk of hospital complications.34, 35 There is also increasing evidence among patients admitted to the intensive care unit that obesity contributes to increased morbidity and to a prolonged length of stay.35 Because they have a higher rate of hyperglycemia, diabetes, and hospital complications, we believe that obese children should be screened for hyperglycemia and diabetes.

We acknowledge the following limitations of this study. The main limitation was its retrospective nature. The method of blood glucose collection and analysis was not standardized; thus, it prevented uniformity in the determination of serum glucose values of individual patients. We arbitrarily used 3 glucose cutoff values in this study (<120, 120179, and >180 mg/dL). Although similar values have been used in inpatient diabetes studies,2022 there is no uniform definition of hyperglycemia in hospitalized patients, and the clinical significance of these cutoff values in pediatric population has not been determined. The study was conducted in a single institution in Atlanta, whose population and disease spectrum might be different from those at other pediatric institutions. Our study did not address the question of whether treatment of hyperglycemia might improve the outcome of length of hospital stay of patients with hyperglycemia. We believe that newly diagnosed hyperglycemia is usually considered a transient finding in response to acute illness not requiring medical intervention, as indicated by the fact that more than half of these patients did not receive antidiabetic therapy. Another limitation of our study is that we were not able to determine the percentage of patients with latent or unrecognized diabetes because of the lack of hemoglobin A1C testing and follow‐up after discharge. A prospective, randomized trial of strict glycemic control is certainly needed to address these issues.

In summary, inpatient hyperglycemia is a common finding in children with and without critical illness. One‐fourth of the children admitted to the hospital had hyperglycemia, most of them without a history of diabetes prior to admission. Although we found a higher need for ICU admission and a longer length of ICU stay, hyperglycemia in pediatric patients was not associated with higher hospital mortality compared with that in children with normoglycemia. Several observational studies have reported an association of hyperglycemia with poor clinical outcome in critically ill children; however, no prospective controlled studies have assessed the effect of tight glucose control in pediatric populations. These studies need to be prospective, randomized multicenter trials of sufficient magnitude to provide a well‐powered analysis to enable multiple observations and evaluation of subsets of critically and non‐critically ill pediatric patients.

Diabetes is one of the most common diagnoses in hospitalized patients.1, 2 Hyperglycemia is present in 38% of adults admitted to the hospital, one third of whom had no history of diabetes before admission.3 The impact of inpatient hyperglycemia on clinical outcome in adult patients has been increasingly appreciated. Extensive evidence from observational studies indicates that hyperglycemia in patients with or without a history of diabetes is an important marker of poor clinical outcome.312 Several prospective randomized trials in patients with critical illness have shown that aggressive glycemic control improves short‐ and long‐term mortality, multiorgan failure and systemic infection, and length of hospitalization.1317 The importance of glucose control also applies to adult patients admitted to general surgical and medical wards.3, 6, 18 In such patients, we recently reported that the presence of hyperglycemia is associated with prolonged hospital stay, infection, disability after hospital discharge, and death.3, 6, 18 Despite the extensive data in adult patients, there is little information on the impact of inpatient hyperglycemia in pediatric patients. The few observational studies in critically ill children admitted to the pediatric ICU with severe brain injury or extensive burn injuries have shown a positive association between inpatient hyperglycemia and increased length of hospital and ICU stay and a higher risk of complication and mortality rates.1923 No previous studies, however, have examined the association of hyperglycemia and clinical outcome in children admitted to a general community pediatric hospital. Therefore, in this study we determined the prevalence of inpatient hyperglycemia and examined the impact of hyperglycemia on morbidity and mortality in children admitted to Hughes Spalding Children's Hospital, a large community hospital serving the inner city and indigent pediatric population in Atlanta, Georgia.

MATERIALS AND METHODS

This was a retrospective observational cohort of pediatric patients consecutively admitted to Hughes Spalding Children's Hospital in Atlanta from January 2004 to August 2004. This general community pediatric hospital is part of the Grady Health System in Atlanta, a large health care organization that operates under the auspices of the Fulton‐Dekalb Hospital Authoritythe major counties in metropolitan Atlantato deliver care to their uninsured and underserved populations. Ninety percent of the organization's inpatient cases are either uninsured or dependent on Medicaid. This is a broad‐based pediatric hospital without cardiac surgery, burn, or dedicated inpatient hematology‐oncology units. Patients are managed by members of the pediatric residency program and supervised by faculty members from Emory University School of Medicine. The Institutional Review Board of Emory University and Grady Health System Oversight Research Committee approved the methods for data collection and analysis used in the study and waived the need for informed consent.

The medical records of 903 consecutive pediatric patients admitted to both critical and noncritical care areas were reviewed. For the analysis, patients were divided according to a known history of diabetes prior to admission and according to admission blood glucose concentration. A normoglycemic group included patients with normal plasma glucose and without a history of diabetes. Serum or plasma glucose measured in the laboratory was assumed to be equivalent to blood glucose measured by finger stick at bedside using a glucose meter. Hyperglycemia was defined as an admission or in‐hospital blood glucose level >120 mg/dL. High blood glucose was subsequently divided into those with blood glucose of 120179 mg/dL and those with blood glucose 180 mg/dL. Patient information was collected regarding demographic characteristics, blood glucose level on admission and during hospital stay, concurrent medical diagnoses, medical treatment, and hospital outcome (including mortality and disposition at discharge).

The primary objectives of this study were to determine the prevalence of in‐hospital hyperglycemia and to examine the association of hyperglycemia and mortality in children with critical and noncritical illness in a community pediatric hospital. Secondary end points included length of hospital stay, requirement of intensive care, and treatment of hyperglycemia. In addition to blood glucose level, prognostic variables included sex, age, body mass index, admission diagnosis, presence of comorbidities, and intensive care unit admission.

RESULTS

Of the 903 admitted patients, 342 patients (38%) had no blood glucose measurement during the hospital stay and were excluded from the analysis. Three patients with a length of stay greater than 6 months were excluded. In addition, 16 patients admitted with diabetic ketoacidosis (DKA) and 1 subject with hyperglycemic hyperosmolar syndrome were also excluded from the analysis. The remaining 542 patients constituted the study population. Most of these, 406 patients (75%), had an admission blood glucose concentration 120 mg/dL (mean SEM 98 1 mg/dL, median 93 mg/dL). A total of 103 children (19%) had an admission blood glucose level of 121179 mg/dL (mean 143 2 mg/dL, median 140 mg/dL), and 32 patients (5.9%) had an admission blood glucose level >180 mg/dL (mean 260 18 mg/dL, median 211 mg/dL; Fig. 1).

Figure 1

The clinical characteristics of study patients are shown in Table 1. Most patients in this study were from minority ethnic groups82% were black, 12% were Hispanic, 2% were from other minority groups, and 4.2% were white. There were no significant differences in mean age, sex, racial distribution, or body mass index among the 3 groups. A total of 409 patients (75.5%) were admitted to general pediatric wards and 133 patients (24.5%) were admitted to the surgical unit. There were no differences in the admission blood glucose between patients admitted to general pediatric wards (112.2 mg/dL) and those admitted to surgical areas (115.7 mg/dL, P > .05). The most common diagnoses in the severe hyperglycemia group were trauma/surgery (25%), pulmonary disease (18.8%), metabolic disorders (12.5%), and infection (6.3%). Most children admitted with hyperglycemia had no history of diabetes prior to admission. Among the 135 children with admission hyperglycemia (blood glucose >120 mg/dL), 17 patients (13%) had a known history of diabetes or were receiving therapy prior to admission. The mean admission blood glucose was 162.4 mg/dL (range 121480 mg/dL) in children with new hyperglycemia and 369.8 mg/dL (range 145678 mg/dL) in those children with a known history of diabetes (P < .01). Among children without a history of diabetes, 33 of 118 children (28%) with admission hyperglycemia had 1 or more glucose values >120 mg/dL during their hospitalizations. Twenty‐five children had a blood glucose of 121179 mg/dL (mean 109 5 mg/dL), and 8 children had a blood glucose 180 mg/dL (mean 159 13 mg/dL). Most patients with a history of diabetes were admitted with significant hyperglycemia. One patient (1%) had a glucose level in the 121179 mg/dL category, and 16 patients (50%) had a glucose level >180 mg/dL.

The presence of hyperglycemia on admission in pediatric patients was not associated with increased mortality or with increased length of hospital stay. There was only 1 death reported during the study period, which occurred in a patient with respiratory failure because of bronchiolitis who was admitted with an admission blood glucose of 151 mg/dL. The mean length of stay for patients with normoglycemia was 3.83 0.2 days, which increased to 5.36 1.0 and 5.68 1.8 days for children with blood glucose of 120179 and 180 mg/dL, respectively (P > .05).

Children with hyperglycemia were more likely to be admitted to the ICU and had a longer length of ICU stay. Admission to the ICU was needed by 10% of children with an admission blood glucose <120 mg/dL, 18% of children with a blood glucose of 120179 mg/dL, and 40% of children with an admission blood 180 mg/dL (P < .01). In addition, length of ICU stay was significantly longer for hyperglycemic children, particularly those with a glucose level 180 mg/dL (P < .001). The mean length of ICU stay (ICU) was 0.56 0.1 days for patients with normoglycemia, and 1.1 0.4 days and 3.6 1.9 days for patients with a blood glucose of 120179 and 180 mg/dL, respectively (P < .01).

Newly diagnosed hyperglycemia was frequently left untreated. Only 3 children without a history of diabetes but with hyperglycemia recorded during the hospital stay received insulin therapy. New hyperglycemia patients received regular insulin per a sliding scale as the main insulin regimen in the hospital. In contrast, all patients with a previous history of diabetes were treated with insulin during their hospital stay.

DISCUSSION

Diabetes mellitus represents a significant public health burden on the basis of increased morbidity, mortality, and economic costs. Increasing evidence from observational and prospective interventional studies has shown that inpatient hyperglycemia is a predictor of poor clinical outcome of adult subjects.313, 16, 17 Admission hyperglycemia has been associated with increased morbidity and mortality in patients with critical illness, as well as in noncritically ill adult subjects admitted to general surgical and medical wards.3, 6, 18 In this study we also found that hyperglycemia is a common finding in children admitted with critical and noncritical illnesses and that most children had no history of diabetes before admission. One‐fourth of the children admitted to the hospital had hyperglycemia on admission. Children with hyperglycemia were more likely to be admitted to the ICU and had a longer length of ICU stay; however, inpatient hyperglycemia was not associated with higher hospital mortality or longer hospital stay than was inpatient normoglycemia. Our findings suggest that recognition of inpatient hyperglycemia can be improved because screening for hyperglycemia was not performed in more than one third of patients (38%) during the hospital stay.

The prevalence of inpatient hyperglycemia in children varies according to the severity of the illness and the study population. Ruiz Magro et al.21 reported that 50% of 353 critically ill children without diabetes mellitus had initial glucose values >120 mg/dL. In a study of 942 nondiabetic patients, Faustino et al.20 found that within 24 hours of admission to the ICU, hyperglycemia was prevalent in 70.4% of patients with a glucose value >120 mg/dL, 44.5% of patients with a glucose value >150 mg/dL, and 22.3% of patients with a glucose value >200 mg/dL. The prevalence of hyperglycemia in non‐critically ill children seen in the emergency department was much lower, ranging from 3.8% to 5.0% (based on an initial blood glucose >150 mg/dL).19, 24 In agreement with these studies, we found inpatient hyperglycemia to be a common finding among hospitalized children. Approximately 75% of our patients had a normal blood glucose on admission, 19% had an admission blood glucose of 121179 mg/dL (mean 143 2 mg/dL), and 5.9% of children had an admission blood glucose 180 mg/dL (mean 260 18 mg/dL). Only 13% of our patients had a known history of diabetes prior to admission, suggesting that the hyperglycemia was a result of the stress of the medical illness or the surgery. Stress hyperglycemia, defined as a transient increase in blood glucose level during acute physiological stress, has been reported to occur in 4% of children with an acute non‐critical illness and in more than 50% of children in the ICU.

A few studies have reported on the impact of inpatient hyperglycemia in children with acute critical illness.1015 Three retrospective studies have demonstrated that admission hyperglycemia is also a predictor of adverse outcomes in the pediatric intensive care unit.20, 22 Srinivasan and colleagues22 demonstrated that 86% of patients in their pediatric intensive care unit had a glucose value >126 mg/dL at some point during their stay. In addition, they showed that duration of the hyperglycemia and peak glucose were also associated with mortality. Faustino and Apkon20 demonstrated that hyperglycemia occurs frequently among critically‐ill nondiabetic children and is correlated with a greater in‐hospital mortality rate and longer length of stay in the ICU. They reported a 2.5‐fold increased risk of dying if the maximum glucose obtained within 24 hours of admission to the ICU was >150 mg/dL. More recently, Yates et al.25 reported that hyperglycemia in the postoperative period was associated with increased morbidity and mortality in postoperative pediatric cardiac patients. Other studies in children with traumatic brain or head injury have also shown an association between poor neurological outcome and elevated admission blood glucose.24, 2628 Brain trauma patients with permanent neurological deficits and in a vegetative state were found to have significantly higher admission blood glucose concentrations than children with good neurological recovery or minimal deficits. In addition, the development of inpatient hyperglycemia in children with extensive burn injuries, covering more than 60% of total body surface area, was found to increase the risk of bacteremia and fungemia, reduce skin graft adhesion, and increase the mortality rate.29 These data show an association of initial glucose, peak glucose, and duration of hyperglycemia with increased incidence of morbidity and mortality in children with acute critical illness. We found no association between initial blood glucose and risk of death. This is in contrast to our previous results in adult patients, in whom inpatient hyperglycemia was found to represent an important marker of increased morbidity and mortality among both those critically ill and not critically ill.3 It is important to note that the overall mortality rate reported in children with hyperglycemia relates to severity of illness and is significantly lower than that of adults.30 In most critically ill pediatric series, hospital mortality ranges from 2% to 5.3% and is higher in patients with severe trauma and those who underwent major cardiac surgery.23, 31 The mortality in children without critical illness admitted to general pediatric wards is significantly lower.30

In agreement with the increasing rate of obesity among children with diabetes,32, 33 especially in minority populations, we found that hospitalized children with a history of diabetes and glucose >180 mg/dL had a higher body mass index than those with normoglycemia (P < .001). Obesity in children has been associated with the presence of several comorbidities and an increased risk of hospital complications.34, 35 There is also increasing evidence among patients admitted to the intensive care unit that obesity contributes to increased morbidity and to a prolonged length of stay.35 Because they have a higher rate of hyperglycemia, diabetes, and hospital complications, we believe that obese children should be screened for hyperglycemia and diabetes.

We acknowledge the following limitations of this study. The main limitation was its retrospective nature. The method of blood glucose collection and analysis was not standardized; thus, it prevented uniformity in the determination of serum glucose values of individual patients. We arbitrarily used 3 glucose cutoff values in this study (<120, 120179, and >180 mg/dL). Although similar values have been used in inpatient diabetes studies,2022 there is no uniform definition of hyperglycemia in hospitalized patients, and the clinical significance of these cutoff values in pediatric population has not been determined. The study was conducted in a single institution in Atlanta, whose population and disease spectrum might be different from those at other pediatric institutions. Our study did not address the question of whether treatment of hyperglycemia might improve the outcome of length of hospital stay of patients with hyperglycemia. We believe that newly diagnosed hyperglycemia is usually considered a transient finding in response to acute illness not requiring medical intervention, as indicated by the fact that more than half of these patients did not receive antidiabetic therapy. Another limitation of our study is that we were not able to determine the percentage of patients with latent or unrecognized diabetes because of the lack of hemoglobin A1C testing and follow‐up after discharge. A prospective, randomized trial of strict glycemic control is certainly needed to address these issues.

In summary, inpatient hyperglycemia is a common finding in children with and without critical illness. One‐fourth of the children admitted to the hospital had hyperglycemia, most of them without a history of diabetes prior to admission. Although we found a higher need for ICU admission and a longer length of ICU stay, hyperglycemia in pediatric patients was not associated with higher hospital mortality compared with that in children with normoglycemia. Several observational studies have reported an association of hyperglycemia with poor clinical outcome in critically ill children; however, no prospective controlled studies have assessed the effect of tight glucose control in pediatric populations. These studies need to be prospective, randomized multicenter trials of sufficient magnitude to provide a well‐powered analysis to enable multiple observations and evaluation of subsets of critically and non‐critically ill pediatric patients.

Smoking represents the single most important cause of premature death and potentially lost years of life in the developing countries. Cigarette smoking causes more than 350,000 deaths each year in the United States and more than 4.9 million premature deaths worldwide.1 Death as a consequence of smoking is by no means limited to the elderly. Tobacco is the largest single cause of premature death and accounts for 3 of 10 of all deaths that occur among smokers and nonsmokers between the ages 35 and 69.2 Because most health professionals deal with different smoking‐related health problems, they make up the sector with the greatest potential to influence reducing smoking among their patients if they can show a positive attitude toward smoking‐cessation intervention.3 Tobacco smoking by health care workers has a negative influence on the general population.3, 4 The World Health Organization (WHO) has advocated that physicians should not smoke cigarettes, and surveys on this issue should be conducted among medical professionals.35 In Jordan, the prevalence of smoking is high and is increasing among women, but there are no data about the prevalence of smoking among physicians and other health care workers (HCWs).5 As members of an antismoking committee working at King Hussein Medical Center (KHMC) we realized that before applying any tobacco control strategy, it was important to understand the prevalence of smoking among HCWs at our center. To our knowledge, no representative survey of smoking among physicians in Jordan has been reported.

This study describes the prevalence of cigarettes smoking among HCWs in the largest tertiary‐care hospital in Jordan.

METHODS

The study was approved by the local ethics committee at KHMC and was conducted between June 1999 and September 1999. The study involved 600 representative samples of HCWs at KHMC. Subjects were divided into 3 groups according to their professions (physicians, nurses, and other professions). Each subject was interviewed personally. Questions were designed to obtain various demographic data and cigarette smoking characteristics. All other forms of tobacco consumption were not included into the questionnaire. Questions addressed various factors such as the age at which smoking was started and its duration and the number of cigarettes smoked per day. We defined smoking status as current smoker, occasional smoker, past smoker, or never smoker, according to WHO's 1995 definitions.4 Current smokers were those who had smoked at least 100 cigarettes and who were currently smoking on a daily basis. Occasional smokers were those who did not smoke daily. Past (ex‐)smokers were those nonsmokers who previously smoked every day for 6 months or more. The rate of cigarette smoking was calculated for each age group and for different medical specialties. Statistical analysis was performed with Statistical Package for Social Sciences 10.0 software (SPSS Inc., Chicago, IL). The ² test was used to determine statistical significance. The 2‐tailed significance level was set at 5% (P 0.05).

RESULTS

Among the 600 respondents, there were 310 women (52%) and 290 men (48%), of whom 260 (43%) were physicians, 250 (42%) were nurses, and 90 (15%) were other HCWs. The total prevalence of smoking was 65%, ranging from 10% in the dermatologist group to 75% in the family practitioner group. We learned that 52% of smokers started before age 21 and that 78% started their habit during the first 2 years of college. The most common motive for starting smoking was pleasure encouraged by peer influence. Eighty‐two percent of male HCWs smoked cigarettes compared with 47% of female HCWs. The prevalence of current smokers was 77% and 33% in men and women, respectively (P = .002). Forty‐three percent of women did not smoke cigarettes, whereas only 14% of men did not smoke (P = .002; Table 1). Smoking prevalence did not significantly differ between age groups (P = .38; Table 2). The highest rate of smoking was among current smokers age 3140 years (58%). Of the 260 physicians, 46% were smokers, (currently or occasionally), 29% were ex‐smokers, and 25% were nonsmokers. Sixty‐seven percent of physicians who were smokers smoked 1120 cigarettes/day. There were fewer current smokers among physicians than among other HCWs (46% versus 74%, respectively). The highest percentage of smokers in the physician group was observed among family practitioners working in the emergency room (75%). On the other hand, dermatologists had the lowest percentage (10%). Women in general had a lower prevalence than men in all categories. Of the female nurses, 17% were smokers, 13% were ex‐smokers, and 70% were nonsmokers. The smoking rate of female nurses fell below their male counterparts (17% and 49%, respectively; P = .002). Seventy‐eight percent of the nonsmoking physicians reported that they do ask their patients routinely about their smoking history and encourage them to discontinue this habit. Only 36% of the physicians who smoked provide such advice during their clinical practice.

DISCUSSION

Tobacco use, notably cigarette smoking, is the leading cause of an array of preventable diseases.12 It is estimated that approximately 30%40% of the adult population worldwide smokes. The situation is particularly alarming in adolescents.5, 6 The prevalence of smoking in developing countries now equals or exceeds the high smoking levels common in the United Kingdom 20 or 30 years ago.6 There is a significant difference in smoking prevalence between socioeconomic groups in the Western world. For professional people the prevalence is now 16%, whereas for unskilled manual workers the prevalence is 48%.7 HCWs are important opinion leaders in the community, and their behavior more than their words has a significant impact on the lifestyle of their patients.3, 89 It is therefore discouraging to learn that so many doctors and nurses still smoke. The smoking habits of health staff members may influence their attitudes toward patients.810 Numerous international studies have addressed the issue of smoking among physicians and other HCWs.816 In a study conducted by Ohida et al.,8 the prevalence of smoking among Japanese physicians was 27.1% for men and 6.8% for women, about half the general population in Japan (male, 54.0%; female, 14.5%). The prevalence of smoking varied in other industrialized countries: in the United States, the prevalence was 3% of men and 10% of women9; in the United Kingdom, it was 4% of men and 5% of women10; in France, 33% of men and 24% of women;11 and in the Netherlands, 41% of men and 24% of women12 Approximately 40% of Italian general practitioners and approximately 45% of their Spanish colleagues also smoke.13 There are limited published data addressing the issue of cigarette smoking among physicians and other HCWs in various Arab countries. Our results showed a higher rate of cigarette smoking among Jordanian physicians compared with that in the surrounding Arab countries.1416 Physicians at KHMC have a very high prevalence of cigarette smokingfar above the results reported in the above‐noted countries. It is comparable with that of unskilled manual workers in the Western world.2, 5 It has been reported that the highest smoking prevalence among young women in the East Mediterranean region occurs in Jordan.17 Our study showed that the smoking rate among women at KHMC, especially among nursing staff, is much lower than that of men, but this might change in the coming years if antismoking measures are not applied and directed toward younger generations. Smoking practice widely varies among the nonmedical KHMC staff and is reaching a very dangerous and worrisome level. This study was the first to be conducted to calculate the prevalence of smoking among HCWs at the largest tertiary‐care hospital in Jordan. A limitation of our study was that the number of responders included in this study might not fully represent the smoking status among HCWs in the country. However, the results raise some important issues to be discussed and analyzed further on a national level concerning this growing health problem. Physicians play an important role in accelerating the process of smoking cessation. Physicians should play an active role in the control of smoking by participating in public debate regarding smoking, both individually and through medical organizations. Nonsmoking physicians at KHMC were more active in asking patients about smoking habits than were those who smoked. The physician smokers were less critical of smoking than were the physician nonsmokers. Jordanian physicians do not fully comply with the rules against tobacco smoking in hospital. Smoking doctors frequently smoke in the hospital and do not counsel patients about smoking.10, 11, 13 Special effort is needed in the educational field concerning the issue of tobacco smoking for Jordanian physicians, and a strong initiative toward smoke‐free hospitals would help spread the message. To promote antismoking measures among doctors and nurses, it will be necessary to scrutinize the smoking habits and behavior of medical and nursing students18 and to conduct effective antismoking and health education activities before they acquire the smoking habit.

Smoking represents the single most important cause of premature death and potentially lost years of life in the developing countries. Cigarette smoking causes more than 350,000 deaths each year in the United States and more than 4.9 million premature deaths worldwide.1 Death as a consequence of smoking is by no means limited to the elderly. Tobacco is the largest single cause of premature death and accounts for 3 of 10 of all deaths that occur among smokers and nonsmokers between the ages 35 and 69.2 Because most health professionals deal with different smoking‐related health problems, they make up the sector with the greatest potential to influence reducing smoking among their patients if they can show a positive attitude toward smoking‐cessation intervention.3 Tobacco smoking by health care workers has a negative influence on the general population.3, 4 The World Health Organization (WHO) has advocated that physicians should not smoke cigarettes, and surveys on this issue should be conducted among medical professionals.35 In Jordan, the prevalence of smoking is high and is increasing among women, but there are no data about the prevalence of smoking among physicians and other health care workers (HCWs).5 As members of an antismoking committee working at King Hussein Medical Center (KHMC) we realized that before applying any tobacco control strategy, it was important to understand the prevalence of smoking among HCWs at our center. To our knowledge, no representative survey of smoking among physicians in Jordan has been reported.

This study describes the prevalence of cigarettes smoking among HCWs in the largest tertiary‐care hospital in Jordan.

METHODS

The study was approved by the local ethics committee at KHMC and was conducted between June 1999 and September 1999. The study involved 600 representative samples of HCWs at KHMC. Subjects were divided into 3 groups according to their professions (physicians, nurses, and other professions). Each subject was interviewed personally. Questions were designed to obtain various demographic data and cigarette smoking characteristics. All other forms of tobacco consumption were not included into the questionnaire. Questions addressed various factors such as the age at which smoking was started and its duration and the number of cigarettes smoked per day. We defined smoking status as current smoker, occasional smoker, past smoker, or never smoker, according to WHO's 1995 definitions.4 Current smokers were those who had smoked at least 100 cigarettes and who were currently smoking on a daily basis. Occasional smokers were those who did not smoke daily. Past (ex‐)smokers were those nonsmokers who previously smoked every day for 6 months or more. The rate of cigarette smoking was calculated for each age group and for different medical specialties. Statistical analysis was performed with Statistical Package for Social Sciences 10.0 software (SPSS Inc., Chicago, IL). The ² test was used to determine statistical significance. The 2‐tailed significance level was set at 5% (P 0.05).

RESULTS

Among the 600 respondents, there were 310 women (52%) and 290 men (48%), of whom 260 (43%) were physicians, 250 (42%) were nurses, and 90 (15%) were other HCWs. The total prevalence of smoking was 65%, ranging from 10% in the dermatologist group to 75% in the family practitioner group. We learned that 52% of smokers started before age 21 and that 78% started their habit during the first 2 years of college. The most common motive for starting smoking was pleasure encouraged by peer influence. Eighty‐two percent of male HCWs smoked cigarettes compared with 47% of female HCWs. The prevalence of current smokers was 77% and 33% in men and women, respectively (P = .002). Forty‐three percent of women did not smoke cigarettes, whereas only 14% of men did not smoke (P = .002; Table 1). Smoking prevalence did not significantly differ between age groups (P = .38; Table 2). The highest rate of smoking was among current smokers age 3140 years (58%). Of the 260 physicians, 46% were smokers, (currently or occasionally), 29% were ex‐smokers, and 25% were nonsmokers. Sixty‐seven percent of physicians who were smokers smoked 1120 cigarettes/day. There were fewer current smokers among physicians than among other HCWs (46% versus 74%, respectively). The highest percentage of smokers in the physician group was observed among family practitioners working in the emergency room (75%). On the other hand, dermatologists had the lowest percentage (10%). Women in general had a lower prevalence than men in all categories. Of the female nurses, 17% were smokers, 13% were ex‐smokers, and 70% were nonsmokers. The smoking rate of female nurses fell below their male counterparts (17% and 49%, respectively; P = .002). Seventy‐eight percent of the nonsmoking physicians reported that they do ask their patients routinely about their smoking history and encourage them to discontinue this habit. Only 36% of the physicians who smoked provide such advice during their clinical practice.

DISCUSSION

Tobacco use, notably cigarette smoking, is the leading cause of an array of preventable diseases.12 It is estimated that approximately 30%40% of the adult population worldwide smokes. The situation is particularly alarming in adolescents.5, 6 The prevalence of smoking in developing countries now equals or exceeds the high smoking levels common in the United Kingdom 20 or 30 years ago.6 There is a significant difference in smoking prevalence between socioeconomic groups in the Western world. For professional people the prevalence is now 16%, whereas for unskilled manual workers the prevalence is 48%.7 HCWs are important opinion leaders in the community, and their behavior more than their words has a significant impact on the lifestyle of their patients.3, 89 It is therefore discouraging to learn that so many doctors and nurses still smoke. The smoking habits of health staff members may influence their attitudes toward patients.810 Numerous international studies have addressed the issue of smoking among physicians and other HCWs.816 In a study conducted by Ohida et al.,8 the prevalence of smoking among Japanese physicians was 27.1% for men and 6.8% for women, about half the general population in Japan (male, 54.0%; female, 14.5%). The prevalence of smoking varied in other industrialized countries: in the United States, the prevalence was 3% of men and 10% of women9; in the United Kingdom, it was 4% of men and 5% of women10; in France, 33% of men and 24% of women;11 and in the Netherlands, 41% of men and 24% of women12 Approximately 40% of Italian general practitioners and approximately 45% of their Spanish colleagues also smoke.13 There are limited published data addressing the issue of cigarette smoking among physicians and other HCWs in various Arab countries. Our results showed a higher rate of cigarette smoking among Jordanian physicians compared with that in the surrounding Arab countries.1416 Physicians at KHMC have a very high prevalence of cigarette smokingfar above the results reported in the above‐noted countries. It is comparable with that of unskilled manual workers in the Western world.2, 5 It has been reported that the highest smoking prevalence among young women in the East Mediterranean region occurs in Jordan.17 Our study showed that the smoking rate among women at KHMC, especially among nursing staff, is much lower than that of men, but this might change in the coming years if antismoking measures are not applied and directed toward younger generations. Smoking practice widely varies among the nonmedical KHMC staff and is reaching a very dangerous and worrisome level. This study was the first to be conducted to calculate the prevalence of smoking among HCWs at the largest tertiary‐care hospital in Jordan. A limitation of our study was that the number of responders included in this study might not fully represent the smoking status among HCWs in the country. However, the results raise some important issues to be discussed and analyzed further on a national level concerning this growing health problem. Physicians play an important role in accelerating the process of smoking cessation. Physicians should play an active role in the control of smoking by participating in public debate regarding smoking, both individually and through medical organizations. Nonsmoking physicians at KHMC were more active in asking patients about smoking habits than were those who smoked. The physician smokers were less critical of smoking than were the physician nonsmokers. Jordanian physicians do not fully comply with the rules against tobacco smoking in hospital. Smoking doctors frequently smoke in the hospital and do not counsel patients about smoking.10, 11, 13 Special effort is needed in the educational field concerning the issue of tobacco smoking for Jordanian physicians, and a strong initiative toward smoke‐free hospitals would help spread the message. To promote antismoking measures among doctors and nurses, it will be necessary to scrutinize the smoking habits and behavior of medical and nursing students18 and to conduct effective antismoking and health education activities before they acquire the smoking habit.

Smoking represents the single most important cause of premature death and potentially lost years of life in the developing countries. Cigarette smoking causes more than 350,000 deaths each year in the United States and more than 4.9 million premature deaths worldwide.1 Death as a consequence of smoking is by no means limited to the elderly. Tobacco is the largest single cause of premature death and accounts for 3 of 10 of all deaths that occur among smokers and nonsmokers between the ages 35 and 69.2 Because most health professionals deal with different smoking‐related health problems, they make up the sector with the greatest potential to influence reducing smoking among their patients if they can show a positive attitude toward smoking‐cessation intervention.3 Tobacco smoking by health care workers has a negative influence on the general population.3, 4 The World Health Organization (WHO) has advocated that physicians should not smoke cigarettes, and surveys on this issue should be conducted among medical professionals.35 In Jordan, the prevalence of smoking is high and is increasing among women, but there are no data about the prevalence of smoking among physicians and other health care workers (HCWs).5 As members of an antismoking committee working at King Hussein Medical Center (KHMC) we realized that before applying any tobacco control strategy, it was important to understand the prevalence of smoking among HCWs at our center. To our knowledge, no representative survey of smoking among physicians in Jordan has been reported.

This study describes the prevalence of cigarettes smoking among HCWs in the largest tertiary‐care hospital in Jordan.

METHODS

The study was approved by the local ethics committee at KHMC and was conducted between June 1999 and September 1999. The study involved 600 representative samples of HCWs at KHMC. Subjects were divided into 3 groups according to their professions (physicians, nurses, and other professions). Each subject was interviewed personally. Questions were designed to obtain various demographic data and cigarette smoking characteristics. All other forms of tobacco consumption were not included into the questionnaire. Questions addressed various factors such as the age at which smoking was started and its duration and the number of cigarettes smoked per day. We defined smoking status as current smoker, occasional smoker, past smoker, or never smoker, according to WHO's 1995 definitions.4 Current smokers were those who had smoked at least 100 cigarettes and who were currently smoking on a daily basis. Occasional smokers were those who did not smoke daily. Past (ex‐)smokers were those nonsmokers who previously smoked every day for 6 months or more. The rate of cigarette smoking was calculated for each age group and for different medical specialties. Statistical analysis was performed with Statistical Package for Social Sciences 10.0 software (SPSS Inc., Chicago, IL). The ² test was used to determine statistical significance. The 2‐tailed significance level was set at 5% (P 0.05).

RESULTS

Among the 600 respondents, there were 310 women (52%) and 290 men (48%), of whom 260 (43%) were physicians, 250 (42%) were nurses, and 90 (15%) were other HCWs. The total prevalence of smoking was 65%, ranging from 10% in the dermatologist group to 75% in the family practitioner group. We learned that 52% of smokers started before age 21 and that 78% started their habit during the first 2 years of college. The most common motive for starting smoking was pleasure encouraged by peer influence. Eighty‐two percent of male HCWs smoked cigarettes compared with 47% of female HCWs. The prevalence of current smokers was 77% and 33% in men and women, respectively (P = .002). Forty‐three percent of women did not smoke cigarettes, whereas only 14% of men did not smoke (P = .002; Table 1). Smoking prevalence did not significantly differ between age groups (P = .38; Table 2). The highest rate of smoking was among current smokers age 3140 years (58%). Of the 260 physicians, 46% were smokers, (currently or occasionally), 29% were ex‐smokers, and 25% were nonsmokers. Sixty‐seven percent of physicians who were smokers smoked 1120 cigarettes/day. There were fewer current smokers among physicians than among other HCWs (46% versus 74%, respectively). The highest percentage of smokers in the physician group was observed among family practitioners working in the emergency room (75%). On the other hand, dermatologists had the lowest percentage (10%). Women in general had a lower prevalence than men in all categories. Of the female nurses, 17% were smokers, 13% were ex‐smokers, and 70% were nonsmokers. The smoking rate of female nurses fell below their male counterparts (17% and 49%, respectively; P = .002). Seventy‐eight percent of the nonsmoking physicians reported that they do ask their patients routinely about their smoking history and encourage them to discontinue this habit. Only 36% of the physicians who smoked provide such advice during their clinical practice.

DISCUSSION

Tobacco use, notably cigarette smoking, is the leading cause of an array of preventable diseases.12 It is estimated that approximately 30%40% of the adult population worldwide smokes. The situation is particularly alarming in adolescents.5, 6 The prevalence of smoking in developing countries now equals or exceeds the high smoking levels common in the United Kingdom 20 or 30 years ago.6 There is a significant difference in smoking prevalence between socioeconomic groups in the Western world. For professional people the prevalence is now 16%, whereas for unskilled manual workers the prevalence is 48%.7 HCWs are important opinion leaders in the community, and their behavior more than their words has a significant impact on the lifestyle of their patients.3, 89 It is therefore discouraging to learn that so many doctors and nurses still smoke. The smoking habits of health staff members may influence their attitudes toward patients.810 Numerous international studies have addressed the issue of smoking among physicians and other HCWs.816 In a study conducted by Ohida et al.,8 the prevalence of smoking among Japanese physicians was 27.1% for men and 6.8% for women, about half the general population in Japan (male, 54.0%; female, 14.5%). The prevalence of smoking varied in other industrialized countries: in the United States, the prevalence was 3% of men and 10% of women9; in the United Kingdom, it was 4% of men and 5% of women10; in France, 33% of men and 24% of women;11 and in the Netherlands, 41% of men and 24% of women12 Approximately 40% of Italian general practitioners and approximately 45% of their Spanish colleagues also smoke.13 There are limited published data addressing the issue of cigarette smoking among physicians and other HCWs in various Arab countries. Our results showed a higher rate of cigarette smoking among Jordanian physicians compared with that in the surrounding Arab countries.1416 Physicians at KHMC have a very high prevalence of cigarette smokingfar above the results reported in the above‐noted countries. It is comparable with that of unskilled manual workers in the Western world.2, 5 It has been reported that the highest smoking prevalence among young women in the East Mediterranean region occurs in Jordan.17 Our study showed that the smoking rate among women at KHMC, especially among nursing staff, is much lower than that of men, but this might change in the coming years if antismoking measures are not applied and directed toward younger generations. Smoking practice widely varies among the nonmedical KHMC staff and is reaching a very dangerous and worrisome level. This study was the first to be conducted to calculate the prevalence of smoking among HCWs at the largest tertiary‐care hospital in Jordan. A limitation of our study was that the number of responders included in this study might not fully represent the smoking status among HCWs in the country. However, the results raise some important issues to be discussed and analyzed further on a national level concerning this growing health problem. Physicians play an important role in accelerating the process of smoking cessation. Physicians should play an active role in the control of smoking by participating in public debate regarding smoking, both individually and through medical organizations. Nonsmoking physicians at KHMC were more active in asking patients about smoking habits than were those who smoked. The physician smokers were less critical of smoking than were the physician nonsmokers. Jordanian physicians do not fully comply with the rules against tobacco smoking in hospital. Smoking doctors frequently smoke in the hospital and do not counsel patients about smoking.10, 11, 13 Special effort is needed in the educational field concerning the issue of tobacco smoking for Jordanian physicians, and a strong initiative toward smoke‐free hospitals would help spread the message. To promote antismoking measures among doctors and nurses, it will be necessary to scrutinize the smoking habits and behavior of medical and nursing students18 and to conduct effective antismoking and health education activities before they acquire the smoking habit.

Current practice patterns among nurses show they are working longer than they ever have.14 The effect of these long hours is that many nurses work in the midst of severe lethargy and sleep deprivation.5 Sleep deprivation jeopardizes not only patient safety but also the safety and general health of the nurses themselves.46 Numerous studies of shift workers in other professions have been done to assess sleepiness using subjective and objective data and also the effect of shift work on work and health.710 Despite the Accreditation Council for Graduate Medical Education (ACGME) mandating work‐hour limitations for medical residents, recent data suggest that sleepiness continues to be a significant issue for medical residents.11 There is a paucity of objective information about the sleepiness and performance of nurses, especially now that most nurses in the United States are working 12‐hour shifts. We hypothesized that nurses working a 12‐hour night shift would have a significant degree of sleepiness. Our objective was to assess the daytime sleepiness of post‐night‐shift nurses using both subjective measures (Epworth Sleepiness Scale [ESS]) and objective testing (Multiple Sleep Latency Test [MSLT]).

MATERIALS AND METHODS

The study was initiated after we obtained institutional review board approval.

RESULTS

Baseline data are shown in Table 1. Nurses in the 2 groups were matched for age, sex, and marital and offspring status. There was a small but statistically significant difference between the 2 groups in BMI (see Table 1). There was no difference between the 2 groups in the average time slept (per night) in the week preceding the test. Seven of 10 ICU nurses had an abnormal ESS (>8) compared with 2 of 10 floor nurses. Mean ESS of ICU nurses was also higher (5.6 2.1 vs. 8.7 3.9 minutes, respectively, P = .042). Nine of 10 ICU nurses had sleep latency values in the severe pathologic range (<5 minutes) for the first sleep period, compared with only 2 of 10 in the floor group. Mean sleep latency in the nap 1 period differed significantly between the 2 groups (Table 2). Overall, however, mean MSLT value did not differ between the ICU nurses and the control group (6.1 3.8 vs. 10.6 7.5 minutes; Table 2 and Fig. 1). Also, during MSLT, the nurses were unaware of their sleep onset in 10 of the 32 periods (31.3%).

Figure 1

DISCUSSION

Our study shows that nurses working night shifts have a pathologic degree of sleepiness. This was especially severe in the ICU nurses as determined by both the ESS and the MSLT studies.

To our knowledge, ours is the first study that has comprehensively evaluated the issue of sleepiness in nurses working night shifts using both the ESS and the MSLT. Previous studies have evaluated subjective sleepiness in nurses. In a cross‐sectional study in 8 large hospitals in Japan, Suzuki et al. found that an estimated 26% of the 4407 nurses surveyed reported excess sleepiness.14 They found key associations of daytime sleepiness with motor vehicle accidents, medication errors, and incorrect operation of medical equipment. Scott et al. randomly surveyed 502 critical care nurses across the US and found that almost two thirds reported struggling to stay awake at least once during the study period and that 22% fell asleep at least once during their work shift.2 Sleep deprivation resulting in impairment in cognitive and psychomotor performance and its association with medical errors have now been well documented in medical residents.1517 This has resulted in the Accreditation Council for Graduate Medical Education mandating a reduction in resident work hours.18 No state or federal regulations restrict the number of hours a nurse may voluntarily work in a 24‐hour or a 7‐day period. Bills prohibiting mandatory overtime for nurses have passed only in California, Maine, New Jersey, and Oregon. No measure, either proposed or enacted, addresses how long nurses may work voluntarily. The recent Institute of Medicine (IOM) report, Keeping Patients Safe, explicitly recommends that nurses' shifts be limited to 12 hours in a 24‐hour period, 60 hours per week, and that voluntary overtime be limited.19

Even with the current ACGME‐mandated reduction in work hours, we and others have reported that sleepiness in medical residents continues to be a major issue.11, 20 In the first year following implementation of the ACGME duty‐hour standards, as many as 43% of interns reported noncompliance with these requirements.21 This demonstrates that mandating work‐hour reductions is only the first step in what is likely to be a long process of effecting change in nurse work hours and fatigue and, in turn, improving patient safety. However, initiating this process is going to be crucial, given the impact even small changes in nursing fatigue could have on patient‐care outcomes.

A nationwide nursing shortage has placed enormous stress on the delivery of patient care in our hospitals. Demands of nursing care requirements have also increased in today's health care scenario because of a variety of socioeconomic factors, and this in turn has forced hospitals to encourage and in many instances insist on nurses working overtime and longer shifts. Rogers et al. examined logbooks completed by 393 hospital staff nurses and found that 40% of the 5317 work shifts they logged exceeded 12 hours. The risk of making an error was significantly increased when the work shift was longer than 12 hours, when overtime was worked, and when the workweek was more than 40 hours.5 There are good data comparing 8‐ and 12‐hour shift lengths among other occupational groups that demonstrate, particularly for the night shift, greater sleepiness during a 12‐hour night shift than during an 8‐hour night shift.22 Emergency room physicians overwhelmingly prefer shifts to last 8 and 10 hours than 12 hours, and longer shifts have been shown to impair their triage decisions in simulation studies.23 The problem is compounded for female nurses, as they also have to carry out their responsibilities as partner and parent along with working, resulting in chronic fatigue and sleep deprivation.24 From these results together with the results of the present study, we suggest enforcing a shift length of no longer than 10 hours for nurses working the night shift in a critical care environment.

In our study, ICU nurses were found to be more sleepy than floor nurses. Sleep quantity in the week prior to the study day did not differ between the 2 groups (405.2 36.6 vs. 416.1 84.73 minutes; P = .364; Fig. 2). For the 24 hours prior to the night shift that was studied, the average amount of sleep was not different between the 2 groups (406 65 minutes for the ICU group vs. 432 107 minutes for the control group; P = .265; Fig. 4). Sleep quality could certainly be markedly different between the 2 groups. A recent study has reported that nearly a third of ICU nurses had severe burnout syndrome, and this has been associated with profound sleep disturbances.25, 26 This could also be attributed to the floor nurses having a less demanding schedule than the ICU nurses.

Figure 2

Our study had some limitations. Our sample size was small, and larger studies may be needed to validate the results of this pilot study. Although our 2 groups were matched for age and sex, the BMI of ICU nurses was slightly but statistically significantly higher than that of floor nurses. Although the mean BMI of ICU nurses was still not in the obese range (one of the exclusion criteria was a BMI > 30), we still cannot definitively rule out that the higher BMI may have conferred a risk of increased upper airway resistance and sleep‐disordered breathing. The control group consisted of RNs in different settingsmedical and surgicaland because of the small numbers of nurses studied, we are unable to further dissect this group and identify differences in degrees of sleepiness. For example, sleep deprivation effects have been shown to be less pronounced in nurses regularly and permanently working night shifts than in nurses working to rotating shifts,27 perhaps a consequence of circadian misalignment being more severe in the latter group, and this factor was not controlled for. We also measured sleepiness after the shift was completed and not during the shift. Not only is this likely reflective of the nurses' sleepiness toward the latter portion of their shift, it also has direct implications for the driving safety of nurses at the end of a shift.

Our MSLT data showed significant differences only for nap 1 but not when combined for the 2 nap periods. We speculate that the reason for this could be that some alertness was recovered by the first nap, as 9 of 10 in the ICU group had at least 15 minutes of sleep during the first nap opportunity compared with only 2 of 10 nurses in the floor group. Incidentally, the only nurse in the ICU group who had no sleep during the first nap period had a sleep latency of 2 minutes during the second nap period (see Fig. 3).

Figure 3

Figure 4

We also did not correlate sleepiness in our study with any clinical performance, and this will be an important variable to focus on in future studies.

In conclusion, our data indicate that nurses working in the ICU are significantly more sleepy than nurses on the floor. Level of sleepiness of ICU nurses is frequently in the pathologic range, comparable to narcolepsy.

Current practice patterns among nurses show they are working longer than they ever have.14 The effect of these long hours is that many nurses work in the midst of severe lethargy and sleep deprivation.5 Sleep deprivation jeopardizes not only patient safety but also the safety and general health of the nurses themselves.46 Numerous studies of shift workers in other professions have been done to assess sleepiness using subjective and objective data and also the effect of shift work on work and health.710 Despite the Accreditation Council for Graduate Medical Education (ACGME) mandating work‐hour limitations for medical residents, recent data suggest that sleepiness continues to be a significant issue for medical residents.11 There is a paucity of objective information about the sleepiness and performance of nurses, especially now that most nurses in the United States are working 12‐hour shifts. We hypothesized that nurses working a 12‐hour night shift would have a significant degree of sleepiness. Our objective was to assess the daytime sleepiness of post‐night‐shift nurses using both subjective measures (Epworth Sleepiness Scale [ESS]) and objective testing (Multiple Sleep Latency Test [MSLT]).

MATERIALS AND METHODS

The study was initiated after we obtained institutional review board approval.

RESULTS

Baseline data are shown in Table 1. Nurses in the 2 groups were matched for age, sex, and marital and offspring status. There was a small but statistically significant difference between the 2 groups in BMI (see Table 1). There was no difference between the 2 groups in the average time slept (per night) in the week preceding the test. Seven of 10 ICU nurses had an abnormal ESS (>8) compared with 2 of 10 floor nurses. Mean ESS of ICU nurses was also higher (5.6 2.1 vs. 8.7 3.9 minutes, respectively, P = .042). Nine of 10 ICU nurses had sleep latency values in the severe pathologic range (<5 minutes) for the first sleep period, compared with only 2 of 10 in the floor group. Mean sleep latency in the nap 1 period differed significantly between the 2 groups (Table 2). Overall, however, mean MSLT value did not differ between the ICU nurses and the control group (6.1 3.8 vs. 10.6 7.5 minutes; Table 2 and Fig. 1). Also, during MSLT, the nurses were unaware of their sleep onset in 10 of the 32 periods (31.3%).

Figure 1

DISCUSSION

Our study shows that nurses working night shifts have a pathologic degree of sleepiness. This was especially severe in the ICU nurses as determined by both the ESS and the MSLT studies.

To our knowledge, ours is the first study that has comprehensively evaluated the issue of sleepiness in nurses working night shifts using both the ESS and the MSLT. Previous studies have evaluated subjective sleepiness in nurses. In a cross‐sectional study in 8 large hospitals in Japan, Suzuki et al. found that an estimated 26% of the 4407 nurses surveyed reported excess sleepiness.14 They found key associations of daytime sleepiness with motor vehicle accidents, medication errors, and incorrect operation of medical equipment. Scott et al. randomly surveyed 502 critical care nurses across the US and found that almost two thirds reported struggling to stay awake at least once during the study period and that 22% fell asleep at least once during their work shift.2 Sleep deprivation resulting in impairment in cognitive and psychomotor performance and its association with medical errors have now been well documented in medical residents.1517 This has resulted in the Accreditation Council for Graduate Medical Education mandating a reduction in resident work hours.18 No state or federal regulations restrict the number of hours a nurse may voluntarily work in a 24‐hour or a 7‐day period. Bills prohibiting mandatory overtime for nurses have passed only in California, Maine, New Jersey, and Oregon. No measure, either proposed or enacted, addresses how long nurses may work voluntarily. The recent Institute of Medicine (IOM) report, Keeping Patients Safe, explicitly recommends that nurses' shifts be limited to 12 hours in a 24‐hour period, 60 hours per week, and that voluntary overtime be limited.19

Even with the current ACGME‐mandated reduction in work hours, we and others have reported that sleepiness in medical residents continues to be a major issue.11, 20 In the first year following implementation of the ACGME duty‐hour standards, as many as 43% of interns reported noncompliance with these requirements.21 This demonstrates that mandating work‐hour reductions is only the first step in what is likely to be a long process of effecting change in nurse work hours and fatigue and, in turn, improving patient safety. However, initiating this process is going to be crucial, given the impact even small changes in nursing fatigue could have on patient‐care outcomes.

A nationwide nursing shortage has placed enormous stress on the delivery of patient care in our hospitals. Demands of nursing care requirements have also increased in today's health care scenario because of a variety of socioeconomic factors, and this in turn has forced hospitals to encourage and in many instances insist on nurses working overtime and longer shifts. Rogers et al. examined logbooks completed by 393 hospital staff nurses and found that 40% of the 5317 work shifts they logged exceeded 12 hours. The risk of making an error was significantly increased when the work shift was longer than 12 hours, when overtime was worked, and when the workweek was more than 40 hours.5 There are good data comparing 8‐ and 12‐hour shift lengths among other occupational groups that demonstrate, particularly for the night shift, greater sleepiness during a 12‐hour night shift than during an 8‐hour night shift.22 Emergency room physicians overwhelmingly prefer shifts to last 8 and 10 hours than 12 hours, and longer shifts have been shown to impair their triage decisions in simulation studies.23 The problem is compounded for female nurses, as they also have to carry out their responsibilities as partner and parent along with working, resulting in chronic fatigue and sleep deprivation.24 From these results together with the results of the present study, we suggest enforcing a shift length of no longer than 10 hours for nurses working the night shift in a critical care environment.

In our study, ICU nurses were found to be more sleepy than floor nurses. Sleep quantity in the week prior to the study day did not differ between the 2 groups (405.2 36.6 vs. 416.1 84.73 minutes; P = .364; Fig. 2). For the 24 hours prior to the night shift that was studied, the average amount of sleep was not different between the 2 groups (406 65 minutes for the ICU group vs. 432 107 minutes for the control group; P = .265; Fig. 4). Sleep quality could certainly be markedly different between the 2 groups. A recent study has reported that nearly a third of ICU nurses had severe burnout syndrome, and this has been associated with profound sleep disturbances.25, 26 This could also be attributed to the floor nurses having a less demanding schedule than the ICU nurses.

Figure 2

Our study had some limitations. Our sample size was small, and larger studies may be needed to validate the results of this pilot study. Although our 2 groups were matched for age and sex, the BMI of ICU nurses was slightly but statistically significantly higher than that of floor nurses. Although the mean BMI of ICU nurses was still not in the obese range (one of the exclusion criteria was a BMI > 30), we still cannot definitively rule out that the higher BMI may have conferred a risk of increased upper airway resistance and sleep‐disordered breathing. The control group consisted of RNs in different settingsmedical and surgicaland because of the small numbers of nurses studied, we are unable to further dissect this group and identify differences in degrees of sleepiness. For example, sleep deprivation effects have been shown to be less pronounced in nurses regularly and permanently working night shifts than in nurses working to rotating shifts,27 perhaps a consequence of circadian misalignment being more severe in the latter group, and this factor was not controlled for. We also measured sleepiness after the shift was completed and not during the shift. Not only is this likely reflective of the nurses' sleepiness toward the latter portion of their shift, it also has direct implications for the driving safety of nurses at the end of a shift.

Our MSLT data showed significant differences only for nap 1 but not when combined for the 2 nap periods. We speculate that the reason for this could be that some alertness was recovered by the first nap, as 9 of 10 in the ICU group had at least 15 minutes of sleep during the first nap opportunity compared with only 2 of 10 nurses in the floor group. Incidentally, the only nurse in the ICU group who had no sleep during the first nap period had a sleep latency of 2 minutes during the second nap period (see Fig. 3).

Figure 3

Figure 4

We also did not correlate sleepiness in our study with any clinical performance, and this will be an important variable to focus on in future studies.

In conclusion, our data indicate that nurses working in the ICU are significantly more sleepy than nurses on the floor. Level of sleepiness of ICU nurses is frequently in the pathologic range, comparable to narcolepsy.

Current practice patterns among nurses show they are working longer than they ever have.14 The effect of these long hours is that many nurses work in the midst of severe lethargy and sleep deprivation.5 Sleep deprivation jeopardizes not only patient safety but also the safety and general health of the nurses themselves.46 Numerous studies of shift workers in other professions have been done to assess sleepiness using subjective and objective data and also the effect of shift work on work and health.710 Despite the Accreditation Council for Graduate Medical Education (ACGME) mandating work‐hour limitations for medical residents, recent data suggest that sleepiness continues to be a significant issue for medical residents.11 There is a paucity of objective information about the sleepiness and performance of nurses, especially now that most nurses in the United States are working 12‐hour shifts. We hypothesized that nurses working a 12‐hour night shift would have a significant degree of sleepiness. Our objective was to assess the daytime sleepiness of post‐night‐shift nurses using both subjective measures (Epworth Sleepiness Scale [ESS]) and objective testing (Multiple Sleep Latency Test [MSLT]).

MATERIALS AND METHODS

The study was initiated after we obtained institutional review board approval.

RESULTS

Baseline data are shown in Table 1. Nurses in the 2 groups were matched for age, sex, and marital and offspring status. There was a small but statistically significant difference between the 2 groups in BMI (see Table 1). There was no difference between the 2 groups in the average time slept (per night) in the week preceding the test. Seven of 10 ICU nurses had an abnormal ESS (>8) compared with 2 of 10 floor nurses. Mean ESS of ICU nurses was also higher (5.6 2.1 vs. 8.7 3.9 minutes, respectively, P = .042). Nine of 10 ICU nurses had sleep latency values in the severe pathologic range (<5 minutes) for the first sleep period, compared with only 2 of 10 in the floor group. Mean sleep latency in the nap 1 period differed significantly between the 2 groups (Table 2). Overall, however, mean MSLT value did not differ between the ICU nurses and the control group (6.1 3.8 vs. 10.6 7.5 minutes; Table 2 and Fig. 1). Also, during MSLT, the nurses were unaware of their sleep onset in 10 of the 32 periods (31.3%).

Figure 1

DISCUSSION

Our study shows that nurses working night shifts have a pathologic degree of sleepiness. This was especially severe in the ICU nurses as determined by both the ESS and the MSLT studies.

To our knowledge, ours is the first study that has comprehensively evaluated the issue of sleepiness in nurses working night shifts using both the ESS and the MSLT. Previous studies have evaluated subjective sleepiness in nurses. In a cross‐sectional study in 8 large hospitals in Japan, Suzuki et al. found that an estimated 26% of the 4407 nurses surveyed reported excess sleepiness.14 They found key associations of daytime sleepiness with motor vehicle accidents, medication errors, and incorrect operation of medical equipment. Scott et al. randomly surveyed 502 critical care nurses across the US and found that almost two thirds reported struggling to stay awake at least once during the study period and that 22% fell asleep at least once during their work shift.2 Sleep deprivation resulting in impairment in cognitive and psychomotor performance and its association with medical errors have now been well documented in medical residents.1517 This has resulted in the Accreditation Council for Graduate Medical Education mandating a reduction in resident work hours.18 No state or federal regulations restrict the number of hours a nurse may voluntarily work in a 24‐hour or a 7‐day period. Bills prohibiting mandatory overtime for nurses have passed only in California, Maine, New Jersey, and Oregon. No measure, either proposed or enacted, addresses how long nurses may work voluntarily. The recent Institute of Medicine (IOM) report, Keeping Patients Safe, explicitly recommends that nurses' shifts be limited to 12 hours in a 24‐hour period, 60 hours per week, and that voluntary overtime be limited.19

Even with the current ACGME‐mandated reduction in work hours, we and others have reported that sleepiness in medical residents continues to be a major issue.11, 20 In the first year following implementation of the ACGME duty‐hour standards, as many as 43% of interns reported noncompliance with these requirements.21 This demonstrates that mandating work‐hour reductions is only the first step in what is likely to be a long process of effecting change in nurse work hours and fatigue and, in turn, improving patient safety. However, initiating this process is going to be crucial, given the impact even small changes in nursing fatigue could have on patient‐care outcomes.

A nationwide nursing shortage has placed enormous stress on the delivery of patient care in our hospitals. Demands of nursing care requirements have also increased in today's health care scenario because of a variety of socioeconomic factors, and this in turn has forced hospitals to encourage and in many instances insist on nurses working overtime and longer shifts. Rogers et al. examined logbooks completed by 393 hospital staff nurses and found that 40% of the 5317 work shifts they logged exceeded 12 hours. The risk of making an error was significantly increased when the work shift was longer than 12 hours, when overtime was worked, and when the workweek was more than 40 hours.5 There are good data comparing 8‐ and 12‐hour shift lengths among other occupational groups that demonstrate, particularly for the night shift, greater sleepiness during a 12‐hour night shift than during an 8‐hour night shift.22 Emergency room physicians overwhelmingly prefer shifts to last 8 and 10 hours than 12 hours, and longer shifts have been shown to impair their triage decisions in simulation studies.23 The problem is compounded for female nurses, as they also have to carry out their responsibilities as partner and parent along with working, resulting in chronic fatigue and sleep deprivation.24 From these results together with the results of the present study, we suggest enforcing a shift length of no longer than 10 hours for nurses working the night shift in a critical care environment.

In our study, ICU nurses were found to be more sleepy than floor nurses. Sleep quantity in the week prior to the study day did not differ between the 2 groups (405.2 36.6 vs. 416.1 84.73 minutes; P = .364; Fig. 2). For the 24 hours prior to the night shift that was studied, the average amount of sleep was not different between the 2 groups (406 65 minutes for the ICU group vs. 432 107 minutes for the control group; P = .265; Fig. 4). Sleep quality could certainly be markedly different between the 2 groups. A recent study has reported that nearly a third of ICU nurses had severe burnout syndrome, and this has been associated with profound sleep disturbances.25, 26 This could also be attributed to the floor nurses having a less demanding schedule than the ICU nurses.

Figure 2

Our study had some limitations. Our sample size was small, and larger studies may be needed to validate the results of this pilot study. Although our 2 groups were matched for age and sex, the BMI of ICU nurses was slightly but statistically significantly higher than that of floor nurses. Although the mean BMI of ICU nurses was still not in the obese range (one of the exclusion criteria was a BMI > 30), we still cannot definitively rule out that the higher BMI may have conferred a risk of increased upper airway resistance and sleep‐disordered breathing. The control group consisted of RNs in different settingsmedical and surgicaland because of the small numbers of nurses studied, we are unable to further dissect this group and identify differences in degrees of sleepiness. For example, sleep deprivation effects have been shown to be less pronounced in nurses regularly and permanently working night shifts than in nurses working to rotating shifts,27 perhaps a consequence of circadian misalignment being more severe in the latter group, and this factor was not controlled for. We also measured sleepiness after the shift was completed and not during the shift. Not only is this likely reflective of the nurses' sleepiness toward the latter portion of their shift, it also has direct implications for the driving safety of nurses at the end of a shift.

Our MSLT data showed significant differences only for nap 1 but not when combined for the 2 nap periods. We speculate that the reason for this could be that some alertness was recovered by the first nap, as 9 of 10 in the ICU group had at least 15 minutes of sleep during the first nap opportunity compared with only 2 of 10 nurses in the floor group. Incidentally, the only nurse in the ICU group who had no sleep during the first nap period had a sleep latency of 2 minutes during the second nap period (see Fig. 3).

Figure 3

Figure 4

We also did not correlate sleepiness in our study with any clinical performance, and this will be an important variable to focus on in future studies.

In conclusion, our data indicate that nurses working in the ICU are significantly more sleepy than nurses on the floor. Level of sleepiness of ICU nurses is frequently in the pathologic range, comparable to narcolepsy.

A 41‐year‐old intravenous drug user (IVDU) was admitted with candidal endophthalmitis 6 weeks after a hospitalization for pneumonia. After discharge from his previous hospitalization which were blood cultures grew Candida albicans, attributed to contamination by a covering physician. The patient described looking through spider webs. Fundoscopic examination revealed fluffy, white string‐of‐pearls opacities with retinal obscuration (Fig. 1). There were no findings of endocarditis (negative echocardiogram) or congestive heart failure. Blood and vitreal cultures grew Candida albicans. The patient underwent a pars plana vitrectomy, and was prescribed chronic fluconazole. He was lost to follow‐up.

Figure 1

Candida albicans is the most common organism identified in endogenous endophthalmitis.1 Predisposing factors include IVDU, indwelling catheters, endocarditis, recent surgeries, immunosuppression, broad‐spectrum antibiotics, and parental nutrition.1 The diagnosis is based on retinal findings of white pinpoint opacities (string‐of‐pearls, Fig. 2), with vitreous involvement and positive cultures. Endocarditis occurs in 15%17% of patients with endophthalmitis.2 This case highlights the importance of physician recognition of the significant attributable morbidity and mortality of candidemia.3, 4

Figure 2

A 41‐year‐old intravenous drug user (IVDU) was admitted with candidal endophthalmitis 6 weeks after a hospitalization for pneumonia. After discharge from his previous hospitalization which were blood cultures grew Candida albicans, attributed to contamination by a covering physician. The patient described looking through spider webs. Fundoscopic examination revealed fluffy, white string‐of‐pearls opacities with retinal obscuration (Fig. 1). There were no findings of endocarditis (negative echocardiogram) or congestive heart failure. Blood and vitreal cultures grew Candida albicans. The patient underwent a pars plana vitrectomy, and was prescribed chronic fluconazole. He was lost to follow‐up.

Figure 1

Candida albicans is the most common organism identified in endogenous endophthalmitis.1 Predisposing factors include IVDU, indwelling catheters, endocarditis, recent surgeries, immunosuppression, broad‐spectrum antibiotics, and parental nutrition.1 The diagnosis is based on retinal findings of white pinpoint opacities (string‐of‐pearls, Fig. 2), with vitreous involvement and positive cultures. Endocarditis occurs in 15%17% of patients with endophthalmitis.2 This case highlights the importance of physician recognition of the significant attributable morbidity and mortality of candidemia.3, 4

Figure 2

A 41‐year‐old intravenous drug user (IVDU) was admitted with candidal endophthalmitis 6 weeks after a hospitalization for pneumonia. After discharge from his previous hospitalization which were blood cultures grew Candida albicans, attributed to contamination by a covering physician. The patient described looking through spider webs. Fundoscopic examination revealed fluffy, white string‐of‐pearls opacities with retinal obscuration (Fig. 1). There were no findings of endocarditis (negative echocardiogram) or congestive heart failure. Blood and vitreal cultures grew Candida albicans. The patient underwent a pars plana vitrectomy, and was prescribed chronic fluconazole. He was lost to follow‐up.

Figure 1

Candida albicans is the most common organism identified in endogenous endophthalmitis.1 Predisposing factors include IVDU, indwelling catheters, endocarditis, recent surgeries, immunosuppression, broad‐spectrum antibiotics, and parental nutrition.1 The diagnosis is based on retinal findings of white pinpoint opacities (string‐of‐pearls, Fig. 2), with vitreous involvement and positive cultures. Endocarditis occurs in 15%17% of patients with endophthalmitis.2 This case highlights the importance of physician recognition of the significant attributable morbidity and mortality of candidemia.3, 4

Figure 2

The appropriate role of specialists in hospital management of common medical conditions has been vigorously debated.13 Few argue that specialists serve an important role as consultants, but whether patients with specific conditions should be admitted to the care of specialists or generalists is unresolved. This is demonstrated by the large degree of hospital‐to‐hospital variability in the proportion of patients with myocardial infarction admitted to cardiologists,4 patients with asthma exacerbations admitted to pulmonologists,5 and patients with renal failure admitted to nephrologists.6

Stroke is another common diagnosis, with variable rates of admission to specialists and generalists. Several prior studies have suggested that outcomes after ischemic stroke are better if a neurologist is the attending physician.710 However, these observational studies could not rule out the possibility that differences in outcome were a result of prognosis at the time of admission rather than improvements in medical care. Although these studies have controlled for known prognostic variables, it is possible that unknown, unmeasured, or inadequately measured variables were different in the groups admitted to neurologists and the groups admitted to generalists. These differences, in turn, might account for outcome differences rather than specialist care.

This form of selection bias, a type of confounding by indication, is a constant threat to validity in observational studies. Randomized trials avoid it because the randomization process balances all prognostic variables, both known and unknown, in the treatment groups.11 Observational studies cannot guarantee the same balance of unmeasured risk factors.12 Multivariate modeling is meant to account for prognostic differences between groups in observational studies, but confounding by indication may remain if all the factors that determine prognosis are not accurately measured. We developed a method to avoid confounding by indication by evaluating individual outcome differences associated with practice variability.13 This technique, termed grouped‐treatment (GT) analysis, is related to the instrumental variable approach developed by economists and occasionally applied to health services research.14

In multivariate GT analyses, the institutional proportion of cases admitted to the care of a neurologist is used as a predictor of outcomes rather than whether an individual patient was admitted to neurology. For example, at a hospital where three‐fourths of acute stroke patients are admitted to neurology, all patients are treated as having a 75% chance of admission to neurology. Rather than denoting whether each patient's specialist attending was a neurologist or a generalist, the 0.75 probability of admission to neurology is used for analysis. If admission to an attending neurologist improves ischemic stroke care, then GT analysis should demonstrate that hospitals admitting higher proportions of stroke patients to neurologists have improved outcomes regardless of whether there is selection bias at the individual patient level. In this way, the method bypasses unmeasured confounders at the individual level in its estimates of treatment effects. The method is susceptible to confounding at the group level; that is, unmeasured prognostic differences in patients admitted to hospitals that rely more heavily on neurologists could bias the GT estimate of treatment effect. The GT estimates are accurate if it can be assumed that a hospital's rate of treatment is not associated (in an unmeasured way) with its patient population's intrinsic, pretreatment prognosis. However, practice variability is very common between hospitals and is generally poorly associated with systematic differences in prognosis of treated patients,15, 16 and in this setting GT provides an independent assessment of treatment effect that may either confirm or refute an association found at the individual level, where confounding is nearly always an important issue.

In this study, we evaluated the impact of admission to a neurologist or generalist on outcomes of ischemic stroke patients treated at academic medical centers throughout the United States. We also compared traditional analysis to GT analysis. In doing so, we demonstrate the influence of unmeasured confounders on observational assessments of specialist care and may provide a more accurate measure of the impact of care by a neurologist on outcomes after ischemic stroke.

MATERIALS AND METHODS

We used the University HealthSystem Consortium (UHC) administrative database, which contained patient information from 84 large academic health centers and their 39 associate hospitals, with more than 2.1 million discharges each year.17 We obtained UHC discharge abstracts for all ischemic stroke patients admitted through emergency departments from 1997 through 1999. Discharge abstracts included patient demographics, urgency status (emergent, urgent, elective), illness severity class, admitting and discharge specialties, discharge diagnoses, procedure codes, in‐hospital mortality, length of stay, and total hospital charges. Patients were identified using International Classification of Diseases, Ninth Revision (ICD‐9) codes that were previously recognized as specific indicators of acute ischemic stroke (ICD‐9 433.01, 433.11, 433.21, 433.31, 433.81, 433.91, 434.01, 434.11, 434.91, 436).1820 We limited the cohort to emergency department admissions in order to reduce the likelihood of referral bias.

Variables in the discharge database were validated by comparison with a detailed medical record review. Between June and December 1999, 42 institutions participating in a quality improvement project identified 30 consecutive ischemic stroke cases. Trained analysts or clinicians abstracted information on demographics, medical history, and treatment. Kappa statistics have been previously reported for all individual characteristics except hospital charges, for which medical record review data were not available.21 Demographic and clinical variables in the administrative database tended to agree well with medical record review, with agreement ranging from 85% to 100% (kappa 0.581.00). Because the admitting attending likely directed acute stroke management, this was used to define a patient's attending physician specialty in all analyses. Administrative coding of tissue plasminogen activator (tPA) use was imperfect, with a sensitivity of 50% but a specificity of 100%.22

Institutional rate of admission to neurologists versus generalists was calculated as the percentage over the entire study duration. Unadjusted logistic regression was used to compare the distribution of patient pretreatment prognostic factors between institutions above and below the 50th percentile to determine a rate of admission to neurology because generalized estimating equations that could account for clustering were unable to support these models as a result of diverging estimates. We calculated the yearly volume of ischemic strokes treated at an institution from discharge abstracts, including admissions from all sources, because all treated cases would be expected to increase physician experience.

In‐hospital mortality was chosen as the primary outcome because of its frequency, importance, and coding reliability. Univariate predictors of in‐hospital mortality were identified using Pearson's chi‐square and the Wilcoxon rank sum tests.23 Length of stay (LOS), total hospital charges, and receipt of tPA were secondary outcomes. LOS and total hospital charges were compared using the Wilcoxon rank sum test. LOS and total charge calculations included only those patients surviving to discharge so that early mortality would not be confused with more efficient care. Similarly, we compared demographics and clinical variables of patients admitted to the care of neurologists with those of patients admitted to the care of generalists. To evaluate variability between institutions, we determined the proportion of patients with specific characteristics and outcomes at each institution and report median values and the 10th‐ to 90th‐percentile range among the institutions. The correlation between institutional rate of admission and institutional rate of mortality was evaluated.

In standard multivariate analysis, we assessed physician specialty as a predictor of in‐hospital mortality of individual patients after adjustment for demographic characteristics, admission status (emergent, urgent, elective), comorbid illness severity score (range 04, from 0 = no substantial comorbid illness to 4 = catastrophic comorbid illness), and annual institutional treatment volume of ischemic stroke. UHC defined severity class to represent an individual's overall calculated risk of illness; its value was dependent on the refinement of the Health Care Facility Administration's diagnosis‐related groups (DRGs) and the Sach's Complication Profiler count of total comorbidities present.24, 25 Effects on LOS and total charges, as well as the ability of physician specialty to predict tPA use in individual patients, were similarly evaluated. Analysis of tPA use was restricted to patients admitted to universities that ever coded tPA use, which increased the sensitivity of the indicator to 57%.22 Residual misclassification error of tPA use would be expected to obscure a true underlying association between its use and physician specialty.

In multivariate GT calculations, we used the institutional proportion of cases admitted to a neurologist as a predictor of outcomes. GT analysis is based on the observation that if a treatment is effective, hospitals that use it more frequently should have better patient outcomes and that this association should persist regardless of whether individual‐level selection bias is present. The method assumes that hospital rates of admission to neurology are independent on the patient population's pretreatment prognosis. Because utilization differences between hospitals likely reflect practice variability rather than differences in patient prognosis,15, 16 the influence of unmeasured confounders at the hospital level is expected to be small. Measured variables that proved significant in univariate analyses or were thought to be responsible for an association between overall patient prognosis and modalities and frequencies of acute stroke treatments used, such as institutional treatment volume, were included in the multivariate GT model in order to isolate the effect of increasing rates of admission to neurologists.

We included both institutional and individual data to more accurately specify individual outcomes and covariates compared with an analysis that simply compared institutions' characteristics and their outcomes.26 Generalized estimating equations (GEE) were used in order to account for institutional clustering of predictor variables and outcomes. GEE is similar to logistic regression but produces broader confidence intervals (CIs) because logistic regression ignores the possibility that individuals at institutions are more similar to each other than would be expected by chance alone. We used a compound symmetry correlation structure, which initiates modeling by assuming a constant correlation between observations within each institution as well as between institutions, and used a logistic link function for binary outcomes in order to mimic logistic regression. The natural log transformations of LOS and hospital charges were modeled to reduce positive skew and approximate a normal distribution, and an identity link function was used in GEE to mimic linear regression for these analyses. To evaluate the impact of adjustment, both unadjusted and adjusted analyses were conducted. Methods to calculate power of GT analysis are not available. The Stata statistical package was used for all analyses (version 8.0; Stata Corporation, College Station, TX).

RESULTS

A total of 26,925 patients with ischemic strokes were admitted to neurologists or generalists through the emergency department at 113 institutions participating in the study. Patients admitted to neurologists rather than generalists (Table 1) were younger and more likely to be male, but less likely to have a serious comorbid illness. Institutions varied widely in the demographics of treated patients as well as in the markers of pretreatment prognosis. Institutional annual case volume of all ischemic strokes ranged from 1 to 741. Mortality rate, mean LOS, and mean hospital charges also varied broadly between institutions (Table 1). Patients treated at institutions whose rate of admissions to a neurologist's care was in the upper 50th percentile were younger and more often male, but did not differ in illness severity class (Table 2).

There were 1760 in‐hospital deaths (7.0%). In univariate analysis, older age (P < .001), white ethnicity (P < .001), emergent stroke (P < .001), and increased illness severity (P < .001) were associated with greater risk of death, whereas African‐American (P < .001) and Hispanic (P = .007) ethnicities were protective. No other patient characteristics were important, and institutional annual case volume showed no association with mortality risk.

Overall, 60% of patients with ischemic stroke were admitted to a neurologist's care. In univariate analysis (Table 3), a lower risk of in‐hospital mortality was observed in cases admitted to neurologists (4.6%) compared with those admitted to generalists (9.5%; P < .001). After adjustment in standard multivariable models, the association between neurologist admission and lower risk of death persisted (OR 0.60; 95% CI, 0.500.72; P < .001).

The institutional rate of admission of ischemic stroke patients to neurologists ranged from 0% to 90%, and higher rates were seen at hospitals with higher institutional case volumes (P < .001). There was no correlation between the institutional rate of admission to neurology and the institutional mortality rate (0.33; P = .73). At the individual‐level, greater rates of admission to neurologists had no significant impact on mortality (OR 1.05; 95% CI, 0.851.31; P = .64; Table 3) in unadjusted analysis. After adjustment for patient demographics, comorbid illness severity score, urgency status, and institutional case volume in GT analysis, there remained no association between death and proportion of ischemic stroke cases admitted to neurologists (OR 1.02; 95% CI, 0.791.30; P = .90), consistent with the absence of an association between neurologist care and in‐hospital mortality.

Patients treated by neurologists were likely to have shorter stays (P < .001) and lower charges (P = .01) in univariate analysis (Table 4). In traditional adjusted multivariable analysis, the same associations were seen for LOS (P < .001) and charges (P = .05). However, in adjusted GT analyses, increased institutional rate of admission to neurologists was not associated with briefer LOS (P = .36) and was associated with greater hospital charges (P = .044).

In 1999, 190 (2.2%) ischemic stroke patients received tPA at the 64 universities that had ever coded tPA use. In univariate analysis, patients admitted to a neurologist were more likely to have received tPA (P = .001; Table 3), and this association persisted after adjustment (P < .001). In adjusted GT analysis, institutions admitting a higher proportion of ischemic stroke patients to neurologists also treated patients with tPA more frequently (P = .033).

DISCUSSION

Several prior studies found that ischemic stroke outcomes were better when an attending neurologist was responsible for patient care.710 Traditional analyses of our data also indicate that care by a neurologist lowers inpatient mortality, LOS, and total charges. By contrast, a GT analysis that bypasses selection bias at the patient level suggests there is no independent benefit of neurologist care on mortality or LOS and actually shows higher associated charges.

The discrepancy between standard and GT analyses suggests that healthier patients may have been preferentially admitted to the care of neurologists. Measured pretreatment prognostic factors in our data present a mixed picture. Patients admitted to a neurologist's care were younger, more often male, more often emergently admitted, and less likely to have serious comorbid illnesses. These patient factors were controlled for in all adjusted analyses. Although traditional multivariate analysis attempts to adjust for variations between the 2 patient populations, it cannot adjust for inaccurately measured or unmeasured differences. Using the institutional proportion of admissions to neurologists as a predictor of patient outcomes, we were better able to control for the selection bias associated with differential distribution of patients to teams led by attending neurologists versus generalists.13, 14

Petty et al.7 studied 299 ischemic stroke patients and showed equivalent survival among stroke patients admitted to neurology inpatient teams versus generalist teams with neurologic consultation. However, patients cared for by generalist teams without neurologic consultation fared worse. Their subjects were treated at both academic and community hospitals. In our study, contributing hospitals were solely academic institutions. Because specialty cross talk may be more frequent at university‐based hospitals, academic‐based generalist physicians may be more familiar with recent stroke literature and guidelines than are their community‐based peers. Further, restricting analysis to academic centers in our study should have reduced the potential confounding influences of differences between other aspects of institutional care. Although no information was available on neurologist consultation in our database, informal consultation is believed to play a large but hidden role at academic medical centers. Thus, the inclusion of a formal consultation variable may be misleading at academic medical centers.

Analyzing claims data on 44,099 Medicare beneficiaries with acute ischemic strokes cared for at both academic and community hospitals, Smith et al.10 also recently reported a 10% lower risk of 30‐day mortality and 12% lower risk of rehospitalization for infections and aspiration pneumonitis among patients admitted to the care of neurologists compared with those admitted to the care of generalists. However, the upper 95% confidence interval limits for these 2 findings nearly crossed 1 (ranging from 0.9980.999). The study also concluded that patients cared for collaboratively by generalists and neurologists had a 16% lower 30‐day mortality risk (hazard ratio 0.84; 95% CI, 0.790.90) than those cared for by generalists alone but simultaneously noted that patients admitted to generalists only had more comorbidities than either the collaborative care or neurologist‐only patient groups. If sicker patients were triaged to generalist admission, as occurs in confounding by indication (also known as channeling bias), then incomplete adjustment for comorbid disease may bias outcomes in favor of neurologist involvement. The GT analysis we employed is specifically designed to overcome this exact type of selection bias.

In our study, patients admitted to neurologists received tPA significantly more often than those admitted to generalists. GT analysis also found that hospitals admitting a higher proportion of strokes to neurologists treated more patients with tPA. This result is consistent with a prior study demonstrating that academic institutions employing a vascular neurologist had significantly higher odds of administering tPA.21 Since tPA must be administered within 3 hours of symptom onset,27 it is commonly delivered in the emergency department prior to admission. Thus, patients may be preferentially selected for admission to neurologic services because of their receipt of tPA, rather than that this association reflects an actual increased use of tPA by neurologists over generalists. Alternatively, institutions with a higher rate of stroke admissions to neurology may simply be more familiar with tPA protocols. Importantly, the poor sensitivity of our data for actual tPA administration may affect the analysis of its use by physician specialty; however, the failure to administratively code tPA use is unlikely to be differentially biased based on physician specialty. Thus, undercoding of tPA use would be expected to bias these analyses toward the null.

The potential advantage and efficacy of stroke centers, stroke units, stroke services, and other institutional processes of care are not addressed by our data. Previously, among academic hospitals, we found that acute ischemic stroke mortality was lower at hospitals employing a vascular neurologist and at those whose guidelines allowed only neurologists to administer tPA.21 A later analysis evaluated the impact of all elements of stroke center care supported by the original Brain Attack Coalition consensus28 and found that no single element improved mortality.29 However, recent studies have found significant mortality benefit associated with stroke units30, 31 and stroke services.32 Clearly, the debate continues over these important questions.

Our study had several limitations. First, generalizability may be lessened because only academic medical centers contributed data and only admissions through the ED were included. However, limiting the study population to academic centers provided a homogenous study population and greatly reduced the potential for confounding at the institutional level. Although the selection of ED cases mitigated the effects of referral bias and the use of only academic hospitals minimized interinstitutional differences, institutions whose rate of admissions to neurology was above the 50th percentile differed from those whose rate admissions to neurology was below the 50th percentile. However, this difference did not consistently result in patients with worse pretreatment prognostic factors being cared for at hospitals with higher rates of admission to neurology. Second, there are important limitations to using administrative data. In our study, patients were selected based on diagnostic coding of records analysts at discharge, and the diagnostic accuracy of such coding for stroke is imperfect.33 Furthermore, missing or incomplete information could have impaired adjustments for patient differences. Third, details of patient treatment were limited. The lack of information about formal and informal consultations may have obscured a true difference in outcomes among specialties.7 Additionally, academic institutions may use systematized care plans more often than do community hospitals, potentially minimizing differences between specialties. Fourth, at the time of our study, tPA had been recently introduced into stroke care. Current rates of tPA use among neurologists and generalists may be more similar. Fifth, the ability of in‐hospital mortality to adequately assess quality of care is limited, and longer‐term and functional outcomes would be better measures and more clinically relevant.

After controlling for selection bias using GT analysis, we found stroke outcomes to be similar regardless of whether a neurologist or a generalist was the admitting physician. This result contrasts with the findings of several previous studies that suggested admitting stroke patients to a neurologist resulted in better clinical outcomes.710 Because only 1 neurologist is employed for approximately every 19.8 generalists in the United States34 and 40% of acute strokes were cared for by generalists, even in this sample entirely restricted to university hospitals, such findings would suggest that many U.S. stroke patients receive inferior care. Because the role of the neurologist as consultant rather than as attending physician is significantly more feasible in most practice settings, the demonstration of equivalent outcomes by both types of physicians is reassuring and certainly reinforces the important role that unmeasured confounders may play in observational studies.

However, these results do imply that it is vital that generalists remain fully trained in the current best practices of acute stroke management in order to maintain the equivalence of care suggested here. Given how common acute stroke is, any proposed future hospitalist training, certification, and recertification programs should include a focus on acute stroke management.

The appropriate role of specialists in hospital management of common medical conditions has been vigorously debated.13 Few argue that specialists serve an important role as consultants, but whether patients with specific conditions should be admitted to the care of specialists or generalists is unresolved. This is demonstrated by the large degree of hospital‐to‐hospital variability in the proportion of patients with myocardial infarction admitted to cardiologists,4 patients with asthma exacerbations admitted to pulmonologists,5 and patients with renal failure admitted to nephrologists.6

Stroke is another common diagnosis, with variable rates of admission to specialists and generalists. Several prior studies have suggested that outcomes after ischemic stroke are better if a neurologist is the attending physician.710 However, these observational studies could not rule out the possibility that differences in outcome were a result of prognosis at the time of admission rather than improvements in medical care. Although these studies have controlled for known prognostic variables, it is possible that unknown, unmeasured, or inadequately measured variables were different in the groups admitted to neurologists and the groups admitted to generalists. These differences, in turn, might account for outcome differences rather than specialist care.

This form of selection bias, a type of confounding by indication, is a constant threat to validity in observational studies. Randomized trials avoid it because the randomization process balances all prognostic variables, both known and unknown, in the treatment groups.11 Observational studies cannot guarantee the same balance of unmeasured risk factors.12 Multivariate modeling is meant to account for prognostic differences between groups in observational studies, but confounding by indication may remain if all the factors that determine prognosis are not accurately measured. We developed a method to avoid confounding by indication by evaluating individual outcome differences associated with practice variability.13 This technique, termed grouped‐treatment (GT) analysis, is related to the instrumental variable approach developed by economists and occasionally applied to health services research.14

In multivariate GT analyses, the institutional proportion of cases admitted to the care of a neurologist is used as a predictor of outcomes rather than whether an individual patient was admitted to neurology. For example, at a hospital where three‐fourths of acute stroke patients are admitted to neurology, all patients are treated as having a 75% chance of admission to neurology. Rather than denoting whether each patient's specialist attending was a neurologist or a generalist, the 0.75 probability of admission to neurology is used for analysis. If admission to an attending neurologist improves ischemic stroke care, then GT analysis should demonstrate that hospitals admitting higher proportions of stroke patients to neurologists have improved outcomes regardless of whether there is selection bias at the individual patient level. In this way, the method bypasses unmeasured confounders at the individual level in its estimates of treatment effects. The method is susceptible to confounding at the group level; that is, unmeasured prognostic differences in patients admitted to hospitals that rely more heavily on neurologists could bias the GT estimate of treatment effect. The GT estimates are accurate if it can be assumed that a hospital's rate of treatment is not associated (in an unmeasured way) with its patient population's intrinsic, pretreatment prognosis. However, practice variability is very common between hospitals and is generally poorly associated with systematic differences in prognosis of treated patients,15, 16 and in this setting GT provides an independent assessment of treatment effect that may either confirm or refute an association found at the individual level, where confounding is nearly always an important issue.

In this study, we evaluated the impact of admission to a neurologist or generalist on outcomes of ischemic stroke patients treated at academic medical centers throughout the United States. We also compared traditional analysis to GT analysis. In doing so, we demonstrate the influence of unmeasured confounders on observational assessments of specialist care and may provide a more accurate measure of the impact of care by a neurologist on outcomes after ischemic stroke.

MATERIALS AND METHODS

We used the University HealthSystem Consortium (UHC) administrative database, which contained patient information from 84 large academic health centers and their 39 associate hospitals, with more than 2.1 million discharges each year.17 We obtained UHC discharge abstracts for all ischemic stroke patients admitted through emergency departments from 1997 through 1999. Discharge abstracts included patient demographics, urgency status (emergent, urgent, elective), illness severity class, admitting and discharge specialties, discharge diagnoses, procedure codes, in‐hospital mortality, length of stay, and total hospital charges. Patients were identified using International Classification of Diseases, Ninth Revision (ICD‐9) codes that were previously recognized as specific indicators of acute ischemic stroke (ICD‐9 433.01, 433.11, 433.21, 433.31, 433.81, 433.91, 434.01, 434.11, 434.91, 436).1820 We limited the cohort to emergency department admissions in order to reduce the likelihood of referral bias.

Variables in the discharge database were validated by comparison with a detailed medical record review. Between June and December 1999, 42 institutions participating in a quality improvement project identified 30 consecutive ischemic stroke cases. Trained analysts or clinicians abstracted information on demographics, medical history, and treatment. Kappa statistics have been previously reported for all individual characteristics except hospital charges, for which medical record review data were not available.21 Demographic and clinical variables in the administrative database tended to agree well with medical record review, with agreement ranging from 85% to 100% (kappa 0.581.00). Because the admitting attending likely directed acute stroke management, this was used to define a patient's attending physician specialty in all analyses. Administrative coding of tissue plasminogen activator (tPA) use was imperfect, with a sensitivity of 50% but a specificity of 100%.22

Institutional rate of admission to neurologists versus generalists was calculated as the percentage over the entire study duration. Unadjusted logistic regression was used to compare the distribution of patient pretreatment prognostic factors between institutions above and below the 50th percentile to determine a rate of admission to neurology because generalized estimating equations that could account for clustering were unable to support these models as a result of diverging estimates. We calculated the yearly volume of ischemic strokes treated at an institution from discharge abstracts, including admissions from all sources, because all treated cases would be expected to increase physician experience.

In‐hospital mortality was chosen as the primary outcome because of its frequency, importance, and coding reliability. Univariate predictors of in‐hospital mortality were identified using Pearson's chi‐square and the Wilcoxon rank sum tests.23 Length of stay (LOS), total hospital charges, and receipt of tPA were secondary outcomes. LOS and total hospital charges were compared using the Wilcoxon rank sum test. LOS and total charge calculations included only those patients surviving to discharge so that early mortality would not be confused with more efficient care. Similarly, we compared demographics and clinical variables of patients admitted to the care of neurologists with those of patients admitted to the care of generalists. To evaluate variability between institutions, we determined the proportion of patients with specific characteristics and outcomes at each institution and report median values and the 10th‐ to 90th‐percentile range among the institutions. The correlation between institutional rate of admission and institutional rate of mortality was evaluated.

In standard multivariate analysis, we assessed physician specialty as a predictor of in‐hospital mortality of individual patients after adjustment for demographic characteristics, admission status (emergent, urgent, elective), comorbid illness severity score (range 04, from 0 = no substantial comorbid illness to 4 = catastrophic comorbid illness), and annual institutional treatment volume of ischemic stroke. UHC defined severity class to represent an individual's overall calculated risk of illness; its value was dependent on the refinement of the Health Care Facility Administration's diagnosis‐related groups (DRGs) and the Sach's Complication Profiler count of total comorbidities present.24, 25 Effects on LOS and total charges, as well as the ability of physician specialty to predict tPA use in individual patients, were similarly evaluated. Analysis of tPA use was restricted to patients admitted to universities that ever coded tPA use, which increased the sensitivity of the indicator to 57%.22 Residual misclassification error of tPA use would be expected to obscure a true underlying association between its use and physician specialty.

In multivariate GT calculations, we used the institutional proportion of cases admitted to a neurologist as a predictor of outcomes. GT analysis is based on the observation that if a treatment is effective, hospitals that use it more frequently should have better patient outcomes and that this association should persist regardless of whether individual‐level selection bias is present. The method assumes that hospital rates of admission to neurology are independent on the patient population's pretreatment prognosis. Because utilization differences between hospitals likely reflect practice variability rather than differences in patient prognosis,15, 16 the influence of unmeasured confounders at the hospital level is expected to be small. Measured variables that proved significant in univariate analyses or were thought to be responsible for an association between overall patient prognosis and modalities and frequencies of acute stroke treatments used, such as institutional treatment volume, were included in the multivariate GT model in order to isolate the effect of increasing rates of admission to neurologists.

We included both institutional and individual data to more accurately specify individual outcomes and covariates compared with an analysis that simply compared institutions' characteristics and their outcomes.26 Generalized estimating equations (GEE) were used in order to account for institutional clustering of predictor variables and outcomes. GEE is similar to logistic regression but produces broader confidence intervals (CIs) because logistic regression ignores the possibility that individuals at institutions are more similar to each other than would be expected by chance alone. We used a compound symmetry correlation structure, which initiates modeling by assuming a constant correlation between observations within each institution as well as between institutions, and used a logistic link function for binary outcomes in order to mimic logistic regression. The natural log transformations of LOS and hospital charges were modeled to reduce positive skew and approximate a normal distribution, and an identity link function was used in GEE to mimic linear regression for these analyses. To evaluate the impact of adjustment, both unadjusted and adjusted analyses were conducted. Methods to calculate power of GT analysis are not available. The Stata statistical package was used for all analyses (version 8.0; Stata Corporation, College Station, TX).

RESULTS

A total of 26,925 patients with ischemic strokes were admitted to neurologists or generalists through the emergency department at 113 institutions participating in the study. Patients admitted to neurologists rather than generalists (Table 1) were younger and more likely to be male, but less likely to have a serious comorbid illness. Institutions varied widely in the demographics of treated patients as well as in the markers of pretreatment prognosis. Institutional annual case volume of all ischemic strokes ranged from 1 to 741. Mortality rate, mean LOS, and mean hospital charges also varied broadly between institutions (Table 1). Patients treated at institutions whose rate of admissions to a neurologist's care was in the upper 50th percentile were younger and more often male, but did not differ in illness severity class (Table 2).

There were 1760 in‐hospital deaths (7.0%). In univariate analysis, older age (P < .001), white ethnicity (P < .001), emergent stroke (P < .001), and increased illness severity (P < .001) were associated with greater risk of death, whereas African‐American (P < .001) and Hispanic (P = .007) ethnicities were protective. No other patient characteristics were important, and institutional annual case volume showed no association with mortality risk.

Overall, 60% of patients with ischemic stroke were admitted to a neurologist's care. In univariate analysis (Table 3), a lower risk of in‐hospital mortality was observed in cases admitted to neurologists (4.6%) compared with those admitted to generalists (9.5%; P < .001). After adjustment in standard multivariable models, the association between neurologist admission and lower risk of death persisted (OR 0.60; 95% CI, 0.500.72; P < .001).

The institutional rate of admission of ischemic stroke patients to neurologists ranged from 0% to 90%, and higher rates were seen at hospitals with higher institutional case volumes (P < .001). There was no correlation between the institutional rate of admission to neurology and the institutional mortality rate (0.33; P = .73). At the individual‐level, greater rates of admission to neurologists had no significant impact on mortality (OR 1.05; 95% CI, 0.851.31; P = .64; Table 3) in unadjusted analysis. After adjustment for patient demographics, comorbid illness severity score, urgency status, and institutional case volume in GT analysis, there remained no association between death and proportion of ischemic stroke cases admitted to neurologists (OR 1.02; 95% CI, 0.791.30; P = .90), consistent with the absence of an association between neurologist care and in‐hospital mortality.

Patients treated by neurologists were likely to have shorter stays (P < .001) and lower charges (P = .01) in univariate analysis (Table 4). In traditional adjusted multivariable analysis, the same associations were seen for LOS (P < .001) and charges (P = .05). However, in adjusted GT analyses, increased institutional rate of admission to neurologists was not associated with briefer LOS (P = .36) and was associated with greater hospital charges (P = .044).

In 1999, 190 (2.2%) ischemic stroke patients received tPA at the 64 universities that had ever coded tPA use. In univariate analysis, patients admitted to a neurologist were more likely to have received tPA (P = .001; Table 3), and this association persisted after adjustment (P < .001). In adjusted GT analysis, institutions admitting a higher proportion of ischemic stroke patients to neurologists also treated patients with tPA more frequently (P = .033).

DISCUSSION

Several prior studies found that ischemic stroke outcomes were better when an attending neurologist was responsible for patient care.710 Traditional analyses of our data also indicate that care by a neurologist lowers inpatient mortality, LOS, and total charges. By contrast, a GT analysis that bypasses selection bias at the patient level suggests there is no independent benefit of neurologist care on mortality or LOS and actually shows higher associated charges.

The discrepancy between standard and GT analyses suggests that healthier patients may have been preferentially admitted to the care of neurologists. Measured pretreatment prognostic factors in our data present a mixed picture. Patients admitted to a neurologist's care were younger, more often male, more often emergently admitted, and less likely to have serious comorbid illnesses. These patient factors were controlled for in all adjusted analyses. Although traditional multivariate analysis attempts to adjust for variations between the 2 patient populations, it cannot adjust for inaccurately measured or unmeasured differences. Using the institutional proportion of admissions to neurologists as a predictor of patient outcomes, we were better able to control for the selection bias associated with differential distribution of patients to teams led by attending neurologists versus generalists.13, 14

Petty et al.7 studied 299 ischemic stroke patients and showed equivalent survival among stroke patients admitted to neurology inpatient teams versus generalist teams with neurologic consultation. However, patients cared for by generalist teams without neurologic consultation fared worse. Their subjects were treated at both academic and community hospitals. In our study, contributing hospitals were solely academic institutions. Because specialty cross talk may be more frequent at university‐based hospitals, academic‐based generalist physicians may be more familiar with recent stroke literature and guidelines than are their community‐based peers. Further, restricting analysis to academic centers in our study should have reduced the potential confounding influences of differences between other aspects of institutional care. Although no information was available on neurologist consultation in our database, informal consultation is believed to play a large but hidden role at academic medical centers. Thus, the inclusion of a formal consultation variable may be misleading at academic medical centers.

Analyzing claims data on 44,099 Medicare beneficiaries with acute ischemic strokes cared for at both academic and community hospitals, Smith et al.10 also recently reported a 10% lower risk of 30‐day mortality and 12% lower risk of rehospitalization for infections and aspiration pneumonitis among patients admitted to the care of neurologists compared with those admitted to the care of generalists. However, the upper 95% confidence interval limits for these 2 findings nearly crossed 1 (ranging from 0.9980.999). The study also concluded that patients cared for collaboratively by generalists and neurologists had a 16% lower 30‐day mortality risk (hazard ratio 0.84; 95% CI, 0.790.90) than those cared for by generalists alone but simultaneously noted that patients admitted to generalists only had more comorbidities than either the collaborative care or neurologist‐only patient groups. If sicker patients were triaged to generalist admission, as occurs in confounding by indication (also known as channeling bias), then incomplete adjustment for comorbid disease may bias outcomes in favor of neurologist involvement. The GT analysis we employed is specifically designed to overcome this exact type of selection bias.

In our study, patients admitted to neurologists received tPA significantly more often than those admitted to generalists. GT analysis also found that hospitals admitting a higher proportion of strokes to neurologists treated more patients with tPA. This result is consistent with a prior study demonstrating that academic institutions employing a vascular neurologist had significantly higher odds of administering tPA.21 Since tPA must be administered within 3 hours of symptom onset,27 it is commonly delivered in the emergency department prior to admission. Thus, patients may be preferentially selected for admission to neurologic services because of their receipt of tPA, rather than that this association reflects an actual increased use of tPA by neurologists over generalists. Alternatively, institutions with a higher rate of stroke admissions to neurology may simply be more familiar with tPA protocols. Importantly, the poor sensitivity of our data for actual tPA administration may affect the analysis of its use by physician specialty; however, the failure to administratively code tPA use is unlikely to be differentially biased based on physician specialty. Thus, undercoding of tPA use would be expected to bias these analyses toward the null.

The potential advantage and efficacy of stroke centers, stroke units, stroke services, and other institutional processes of care are not addressed by our data. Previously, among academic hospitals, we found that acute ischemic stroke mortality was lower at hospitals employing a vascular neurologist and at those whose guidelines allowed only neurologists to administer tPA.21 A later analysis evaluated the impact of all elements of stroke center care supported by the original Brain Attack Coalition consensus28 and found that no single element improved mortality.29 However, recent studies have found significant mortality benefit associated with stroke units30, 31 and stroke services.32 Clearly, the debate continues over these important questions.

Our study had several limitations. First, generalizability may be lessened because only academic medical centers contributed data and only admissions through the ED were included. However, limiting the study population to academic centers provided a homogenous study population and greatly reduced the potential for confounding at the institutional level. Although the selection of ED cases mitigated the effects of referral bias and the use of only academic hospitals minimized interinstitutional differences, institutions whose rate of admissions to neurology was above the 50th percentile differed from those whose rate admissions to neurology was below the 50th percentile. However, this difference did not consistently result in patients with worse pretreatment prognostic factors being cared for at hospitals with higher rates of admission to neurology. Second, there are important limitations to using administrative data. In our study, patients were selected based on diagnostic coding of records analysts at discharge, and the diagnostic accuracy of such coding for stroke is imperfect.33 Furthermore, missing or incomplete information could have impaired adjustments for patient differences. Third, details of patient treatment were limited. The lack of information about formal and informal consultations may have obscured a true difference in outcomes among specialties.7 Additionally, academic institutions may use systematized care plans more often than do community hospitals, potentially minimizing differences between specialties. Fourth, at the time of our study, tPA had been recently introduced into stroke care. Current rates of tPA use among neurologists and generalists may be more similar. Fifth, the ability of in‐hospital mortality to adequately assess quality of care is limited, and longer‐term and functional outcomes would be better measures and more clinically relevant.

After controlling for selection bias using GT analysis, we found stroke outcomes to be similar regardless of whether a neurologist or a generalist was the admitting physician. This result contrasts with the findings of several previous studies that suggested admitting stroke patients to a neurologist resulted in better clinical outcomes.710 Because only 1 neurologist is employed for approximately every 19.8 generalists in the United States34 and 40% of acute strokes were cared for by generalists, even in this sample entirely restricted to university hospitals, such findings would suggest that many U.S. stroke patients receive inferior care. Because the role of the neurologist as consultant rather than as attending physician is significantly more feasible in most practice settings, the demonstration of equivalent outcomes by both types of physicians is reassuring and certainly reinforces the important role that unmeasured confounders may play in observational studies.

However, these results do imply that it is vital that generalists remain fully trained in the current best practices of acute stroke management in order to maintain the equivalence of care suggested here. Given how common acute stroke is, any proposed future hospitalist training, certification, and recertification programs should include a focus on acute stroke management.

The appropriate role of specialists in hospital management of common medical conditions has been vigorously debated.13 Few argue that specialists serve an important role as consultants, but whether patients with specific conditions should be admitted to the care of specialists or generalists is unresolved. This is demonstrated by the large degree of hospital‐to‐hospital variability in the proportion of patients with myocardial infarction admitted to cardiologists,4 patients with asthma exacerbations admitted to pulmonologists,5 and patients with renal failure admitted to nephrologists.6

Stroke is another common diagnosis, with variable rates of admission to specialists and generalists. Several prior studies have suggested that outcomes after ischemic stroke are better if a neurologist is the attending physician.710 However, these observational studies could not rule out the possibility that differences in outcome were a result of prognosis at the time of admission rather than improvements in medical care. Although these studies have controlled for known prognostic variables, it is possible that unknown, unmeasured, or inadequately measured variables were different in the groups admitted to neurologists and the groups admitted to generalists. These differences, in turn, might account for outcome differences rather than specialist care.

This form of selection bias, a type of confounding by indication, is a constant threat to validity in observational studies. Randomized trials avoid it because the randomization process balances all prognostic variables, both known and unknown, in the treatment groups.11 Observational studies cannot guarantee the same balance of unmeasured risk factors.12 Multivariate modeling is meant to account for prognostic differences between groups in observational studies, but confounding by indication may remain if all the factors that determine prognosis are not accurately measured. We developed a method to avoid confounding by indication by evaluating individual outcome differences associated with practice variability.13 This technique, termed grouped‐treatment (GT) analysis, is related to the instrumental variable approach developed by economists and occasionally applied to health services research.14

In multivariate GT analyses, the institutional proportion of cases admitted to the care of a neurologist is used as a predictor of outcomes rather than whether an individual patient was admitted to neurology. For example, at a hospital where three‐fourths of acute stroke patients are admitted to neurology, all patients are treated as having a 75% chance of admission to neurology. Rather than denoting whether each patient's specialist attending was a neurologist or a generalist, the 0.75 probability of admission to neurology is used for analysis. If admission to an attending neurologist improves ischemic stroke care, then GT analysis should demonstrate that hospitals admitting higher proportions of stroke patients to neurologists have improved outcomes regardless of whether there is selection bias at the individual patient level. In this way, the method bypasses unmeasured confounders at the individual level in its estimates of treatment effects. The method is susceptible to confounding at the group level; that is, unmeasured prognostic differences in patients admitted to hospitals that rely more heavily on neurologists could bias the GT estimate of treatment effect. The GT estimates are accurate if it can be assumed that a hospital's rate of treatment is not associated (in an unmeasured way) with its patient population's intrinsic, pretreatment prognosis. However, practice variability is very common between hospitals and is generally poorly associated with systematic differences in prognosis of treated patients,15, 16 and in this setting GT provides an independent assessment of treatment effect that may either confirm or refute an association found at the individual level, where confounding is nearly always an important issue.

In this study, we evaluated the impact of admission to a neurologist or generalist on outcomes of ischemic stroke patients treated at academic medical centers throughout the United States. We also compared traditional analysis to GT analysis. In doing so, we demonstrate the influence of unmeasured confounders on observational assessments of specialist care and may provide a more accurate measure of the impact of care by a neurologist on outcomes after ischemic stroke.

MATERIALS AND METHODS

We used the University HealthSystem Consortium (UHC) administrative database, which contained patient information from 84 large academic health centers and their 39 associate hospitals, with more than 2.1 million discharges each year.17 We obtained UHC discharge abstracts for all ischemic stroke patients admitted through emergency departments from 1997 through 1999. Discharge abstracts included patient demographics, urgency status (emergent, urgent, elective), illness severity class, admitting and discharge specialties, discharge diagnoses, procedure codes, in‐hospital mortality, length of stay, and total hospital charges. Patients were identified using International Classification of Diseases, Ninth Revision (ICD‐9) codes that were previously recognized as specific indicators of acute ischemic stroke (ICD‐9 433.01, 433.11, 433.21, 433.31, 433.81, 433.91, 434.01, 434.11, 434.91, 436).1820 We limited the cohort to emergency department admissions in order to reduce the likelihood of referral bias.

Variables in the discharge database were validated by comparison with a detailed medical record review. Between June and December 1999, 42 institutions participating in a quality improvement project identified 30 consecutive ischemic stroke cases. Trained analysts or clinicians abstracted information on demographics, medical history, and treatment. Kappa statistics have been previously reported for all individual characteristics except hospital charges, for which medical record review data were not available.21 Demographic and clinical variables in the administrative database tended to agree well with medical record review, with agreement ranging from 85% to 100% (kappa 0.581.00). Because the admitting attending likely directed acute stroke management, this was used to define a patient's attending physician specialty in all analyses. Administrative coding of tissue plasminogen activator (tPA) use was imperfect, with a sensitivity of 50% but a specificity of 100%.22

Institutional rate of admission to neurologists versus generalists was calculated as the percentage over the entire study duration. Unadjusted logistic regression was used to compare the distribution of patient pretreatment prognostic factors between institutions above and below the 50th percentile to determine a rate of admission to neurology because generalized estimating equations that could account for clustering were unable to support these models as a result of diverging estimates. We calculated the yearly volume of ischemic strokes treated at an institution from discharge abstracts, including admissions from all sources, because all treated cases would be expected to increase physician experience.

In‐hospital mortality was chosen as the primary outcome because of its frequency, importance, and coding reliability. Univariate predictors of in‐hospital mortality were identified using Pearson's chi‐square and the Wilcoxon rank sum tests.23 Length of stay (LOS), total hospital charges, and receipt of tPA were secondary outcomes. LOS and total hospital charges were compared using the Wilcoxon rank sum test. LOS and total charge calculations included only those patients surviving to discharge so that early mortality would not be confused with more efficient care. Similarly, we compared demographics and clinical variables of patients admitted to the care of neurologists with those of patients admitted to the care of generalists. To evaluate variability between institutions, we determined the proportion of patients with specific characteristics and outcomes at each institution and report median values and the 10th‐ to 90th‐percentile range among the institutions. The correlation between institutional rate of admission and institutional rate of mortality was evaluated.

In standard multivariate analysis, we assessed physician specialty as a predictor of in‐hospital mortality of individual patients after adjustment for demographic characteristics, admission status (emergent, urgent, elective), comorbid illness severity score (range 04, from 0 = no substantial comorbid illness to 4 = catastrophic comorbid illness), and annual institutional treatment volume of ischemic stroke. UHC defined severity class to represent an individual's overall calculated risk of illness; its value was dependent on the refinement of the Health Care Facility Administration's diagnosis‐related groups (DRGs) and the Sach's Complication Profiler count of total comorbidities present.24, 25 Effects on LOS and total charges, as well as the ability of physician specialty to predict tPA use in individual patients, were similarly evaluated. Analysis of tPA use was restricted to patients admitted to universities that ever coded tPA use, which increased the sensitivity of the indicator to 57%.22 Residual misclassification error of tPA use would be expected to obscure a true underlying association between its use and physician specialty.

In multivariate GT calculations, we used the institutional proportion of cases admitted to a neurologist as a predictor of outcomes. GT analysis is based on the observation that if a treatment is effective, hospitals that use it more frequently should have better patient outcomes and that this association should persist regardless of whether individual‐level selection bias is present. The method assumes that hospital rates of admission to neurology are independent on the patient population's pretreatment prognosis. Because utilization differences between hospitals likely reflect practice variability rather than differences in patient prognosis,15, 16 the influence of unmeasured confounders at the hospital level is expected to be small. Measured variables that proved significant in univariate analyses or were thought to be responsible for an association between overall patient prognosis and modalities and frequencies of acute stroke treatments used, such as institutional treatment volume, were included in the multivariate GT model in order to isolate the effect of increasing rates of admission to neurologists.

We included both institutional and individual data to more accurately specify individual outcomes and covariates compared with an analysis that simply compared institutions' characteristics and their outcomes.26 Generalized estimating equations (GEE) were used in order to account for institutional clustering of predictor variables and outcomes. GEE is similar to logistic regression but produces broader confidence intervals (CIs) because logistic regression ignores the possibility that individuals at institutions are more similar to each other than would be expected by chance alone. We used a compound symmetry correlation structure, which initiates modeling by assuming a constant correlation between observations within each institution as well as between institutions, and used a logistic link function for binary outcomes in order to mimic logistic regression. The natural log transformations of LOS and hospital charges were modeled to reduce positive skew and approximate a normal distribution, and an identity link function was used in GEE to mimic linear regression for these analyses. To evaluate the impact of adjustment, both unadjusted and adjusted analyses were conducted. Methods to calculate power of GT analysis are not available. The Stata statistical package was used for all analyses (version 8.0; Stata Corporation, College Station, TX).

RESULTS

A total of 26,925 patients with ischemic strokes were admitted to neurologists or generalists through the emergency department at 113 institutions participating in the study. Patients admitted to neurologists rather than generalists (Table 1) were younger and more likely to be male, but less likely to have a serious comorbid illness. Institutions varied widely in the demographics of treated patients as well as in the markers of pretreatment prognosis. Institutional annual case volume of all ischemic strokes ranged from 1 to 741. Mortality rate, mean LOS, and mean hospital charges also varied broadly between institutions (Table 1). Patients treated at institutions whose rate of admissions to a neurologist's care was in the upper 50th percentile were younger and more often male, but did not differ in illness severity class (Table 2).

There were 1760 in‐hospital deaths (7.0%). In univariate analysis, older age (P < .001), white ethnicity (P < .001), emergent stroke (P < .001), and increased illness severity (P < .001) were associated with greater risk of death, whereas African‐American (P < .001) and Hispanic (P = .007) ethnicities were protective. No other patient characteristics were important, and institutional annual case volume showed no association with mortality risk.

Overall, 60% of patients with ischemic stroke were admitted to a neurologist's care. In univariate analysis (Table 3), a lower risk of in‐hospital mortality was observed in cases admitted to neurologists (4.6%) compared with those admitted to generalists (9.5%; P < .001). After adjustment in standard multivariable models, the association between neurologist admission and lower risk of death persisted (OR 0.60; 95% CI, 0.500.72; P < .001).

The institutional rate of admission of ischemic stroke patients to neurologists ranged from 0% to 90%, and higher rates were seen at hospitals with higher institutional case volumes (P < .001). There was no correlation between the institutional rate of admission to neurology and the institutional mortality rate (0.33; P = .73). At the individual‐level, greater rates of admission to neurologists had no significant impact on mortality (OR 1.05; 95% CI, 0.851.31; P = .64; Table 3) in unadjusted analysis. After adjustment for patient demographics, comorbid illness severity score, urgency status, and institutional case volume in GT analysis, there remained no association between death and proportion of ischemic stroke cases admitted to neurologists (OR 1.02; 95% CI, 0.791.30; P = .90), consistent with the absence of an association between neurologist care and in‐hospital mortality.

Patients treated by neurologists were likely to have shorter stays (P < .001) and lower charges (P = .01) in univariate analysis (Table 4). In traditional adjusted multivariable analysis, the same associations were seen for LOS (P < .001) and charges (P = .05). However, in adjusted GT analyses, increased institutional rate of admission to neurologists was not associated with briefer LOS (P = .36) and was associated with greater hospital charges (P = .044).

In 1999, 190 (2.2%) ischemic stroke patients received tPA at the 64 universities that had ever coded tPA use. In univariate analysis, patients admitted to a neurologist were more likely to have received tPA (P = .001; Table 3), and this association persisted after adjustment (P < .001). In adjusted GT analysis, institutions admitting a higher proportion of ischemic stroke patients to neurologists also treated patients with tPA more frequently (P = .033).

DISCUSSION

Several prior studies found that ischemic stroke outcomes were better when an attending neurologist was responsible for patient care.710 Traditional analyses of our data also indicate that care by a neurologist lowers inpatient mortality, LOS, and total charges. By contrast, a GT analysis that bypasses selection bias at the patient level suggests there is no independent benefit of neurologist care on mortality or LOS and actually shows higher associated charges.

The discrepancy between standard and GT analyses suggests that healthier patients may have been preferentially admitted to the care of neurologists. Measured pretreatment prognostic factors in our data present a mixed picture. Patients admitted to a neurologist's care were younger, more often male, more often emergently admitted, and less likely to have serious comorbid illnesses. These patient factors were controlled for in all adjusted analyses. Although traditional multivariate analysis attempts to adjust for variations between the 2 patient populations, it cannot adjust for inaccurately measured or unmeasured differences. Using the institutional proportion of admissions to neurologists as a predictor of patient outcomes, we were better able to control for the selection bias associated with differential distribution of patients to teams led by attending neurologists versus generalists.13, 14

Petty et al.7 studied 299 ischemic stroke patients and showed equivalent survival among stroke patients admitted to neurology inpatient teams versus generalist teams with neurologic consultation. However, patients cared for by generalist teams without neurologic consultation fared worse. Their subjects were treated at both academic and community hospitals. In our study, contributing hospitals were solely academic institutions. Because specialty cross talk may be more frequent at university‐based hospitals, academic‐based generalist physicians may be more familiar with recent stroke literature and guidelines than are their community‐based peers. Further, restricting analysis to academic centers in our study should have reduced the potential confounding influences of differences between other aspects of institutional care. Although no information was available on neurologist consultation in our database, informal consultation is believed to play a large but hidden role at academic medical centers. Thus, the inclusion of a formal consultation variable may be misleading at academic medical centers.

Analyzing claims data on 44,099 Medicare beneficiaries with acute ischemic strokes cared for at both academic and community hospitals, Smith et al.10 also recently reported a 10% lower risk of 30‐day mortality and 12% lower risk of rehospitalization for infections and aspiration pneumonitis among patients admitted to the care of neurologists compared with those admitted to the care of generalists. However, the upper 95% confidence interval limits for these 2 findings nearly crossed 1 (ranging from 0.9980.999). The study also concluded that patients cared for collaboratively by generalists and neurologists had a 16% lower 30‐day mortality risk (hazard ratio 0.84; 95% CI, 0.790.90) than those cared for by generalists alone but simultaneously noted that patients admitted to generalists only had more comorbidities than either the collaborative care or neurologist‐only patient groups. If sicker patients were triaged to generalist admission, as occurs in confounding by indication (also known as channeling bias), then incomplete adjustment for comorbid disease may bias outcomes in favor of neurologist involvement. The GT analysis we employed is specifically designed to overcome this exact type of selection bias.

In our study, patients admitted to neurologists received tPA significantly more often than those admitted to generalists. GT analysis also found that hospitals admitting a higher proportion of strokes to neurologists treated more patients with tPA. This result is consistent with a prior study demonstrating that academic institutions employing a vascular neurologist had significantly higher odds of administering tPA.21 Since tPA must be administered within 3 hours of symptom onset,27 it is commonly delivered in the emergency department prior to admission. Thus, patients may be preferentially selected for admission to neurologic services because of their receipt of tPA, rather than that this association reflects an actual increased use of tPA by neurologists over generalists. Alternatively, institutions with a higher rate of stroke admissions to neurology may simply be more familiar with tPA protocols. Importantly, the poor sensitivity of our data for actual tPA administration may affect the analysis of its use by physician specialty; however, the failure to administratively code tPA use is unlikely to be differentially biased based on physician specialty. Thus, undercoding of tPA use would be expected to bias these analyses toward the null.

The potential advantage and efficacy of stroke centers, stroke units, stroke services, and other institutional processes of care are not addressed by our data. Previously, among academic hospitals, we found that acute ischemic stroke mortality was lower at hospitals employing a vascular neurologist and at those whose guidelines allowed only neurologists to administer tPA.21 A later analysis evaluated the impact of all elements of stroke center care supported by the original Brain Attack Coalition consensus28 and found that no single element improved mortality.29 However, recent studies have found significant mortality benefit associated with stroke units30, 31 and stroke services.32 Clearly, the debate continues over these important questions.

Our study had several limitations. First, generalizability may be lessened because only academic medical centers contributed data and only admissions through the ED were included. However, limiting the study population to academic centers provided a homogenous study population and greatly reduced the potential for confounding at the institutional level. Although the selection of ED cases mitigated the effects of referral bias and the use of only academic hospitals minimized interinstitutional differences, institutions whose rate of admissions to neurology was above the 50th percentile differed from those whose rate admissions to neurology was below the 50th percentile. However, this difference did not consistently result in patients with worse pretreatment prognostic factors being cared for at hospitals with higher rates of admission to neurology. Second, there are important limitations to using administrative data. In our study, patients were selected based on diagnostic coding of records analysts at discharge, and the diagnostic accuracy of such coding for stroke is imperfect.33 Furthermore, missing or incomplete information could have impaired adjustments for patient differences. Third, details of patient treatment were limited. The lack of information about formal and informal consultations may have obscured a true difference in outcomes among specialties.7 Additionally, academic institutions may use systematized care plans more often than do community hospitals, potentially minimizing differences between specialties. Fourth, at the time of our study, tPA had been recently introduced into stroke care. Current rates of tPA use among neurologists and generalists may be more similar. Fifth, the ability of in‐hospital mortality to adequately assess quality of care is limited, and longer‐term and functional outcomes would be better measures and more clinically relevant.

After controlling for selection bias using GT analysis, we found stroke outcomes to be similar regardless of whether a neurologist or a generalist was the admitting physician. This result contrasts with the findings of several previous studies that suggested admitting stroke patients to a neurologist resulted in better clinical outcomes.710 Because only 1 neurologist is employed for approximately every 19.8 generalists in the United States34 and 40% of acute strokes were cared for by generalists, even in this sample entirely restricted to university hospitals, such findings would suggest that many U.S. stroke patients receive inferior care. Because the role of the neurologist as consultant rather than as attending physician is significantly more feasible in most practice settings, the demonstration of equivalent outcomes by both types of physicians is reassuring and certainly reinforces the important role that unmeasured confounders may play in observational studies.

However, these results do imply that it is vital that generalists remain fully trained in the current best practices of acute stroke management in order to maintain the equivalence of care suggested here. Given how common acute stroke is, any proposed future hospitalist training, certification, and recertification programs should include a focus on acute stroke management.

Many academic medical centers (AMCs) have developed nonhousestaff services to provide clinical care once provided by physicians‐in‐training. These services, often staffed by hospitalists and/or midlevel providers, have experienced tremendous growth in the past few years, yet very little exists in the literature about their development, structure, efficacy, or impact on hospitals, patients, and hospital medicine programs. The primary forces driving this growth include Accreditation Council for Graduate Medical Education (ACGME) resident duty hour restrictions,1 growth of the hospitalist movement,2 and the emphasis on simultaneously improving financial performance and quality of care in AMCs.3

Resident Duty Hour Restrictions

In 2003, the ACGME mandated restrictions on resident work hours, limiting trainees to 80 hours per week.1 Many training programs struggled with how to provide important clinical services while complying with the new restrictionscreating numerous models that bridged care between different shifts of residents.45 Implementation of day floats (a dedicated resident who rounds with the postcall team), night floats (a dedicated overnight resident who admits and cross‐covers patients), or some variation of both was common.6 No guidelines accompanied the ACGME mandate, leaving institutions to independently structure their programs without a known best practice.

Subsequent literature carefully addressed how the duty hour restrictions affect residents' lives and education but failed to discuss models for providing care.711 Training programs began to institute necessary changes but in doing so, created greater patient discontinuity and increased handoffs between residents, elevating the potential for adverse patient outcomes.12 Recent large‐scale studies indicate that inpatient care is the same or improved since adoption of the duty hour restrictions,1316 but controversy continues, with several editorials debating the issue.1719

Because increasing the volume of patients on housestaff services was not a viable option,20 many AMCs created nonhousestaff services and hired midlevel providers (nurse practitioners and physician assistants) to offset resident workloads and comply with the new restrictions. However, this strategy represented a very expensive alternative.21 Moreover, the current 80‐hour work limits may be revised downward, particularly given the lower restrictions in other countries,22 and this will further drive the demand for nonhousestaff services. Hospitalists, with their documented impact on efficiency and return on investment,23 represent a solution to fill these needs and have quickly become the predominant approach at AMCs.

The Hospitalist Movement

Since the term hospitalist was first coined in 1996,24 the remarkable growth of the number of practicing hospitalists emphasizes how first community hospitals and now AMCs have embraced this approach.25 With more than 20,000 nationwide and projections that the field will grow to 30,000 by 2010,26 hospitalists are becoming the primary providers for in‐patients.2 This growth was further catalyzed when widely expressed concerns about safety and quality became public,2728 and hospitalists incorporated patient safety and quality improvement activities into their efforts.3 The confluence of these factors also prompted emergence of hospital medicine programs at AMCs, a growth that came with anticipated dangers.29 Reflecting the recognition that hospital medicine is becoming a separate specialty30 and is integral to the functioning of an AMC, institutions now operate dedicated divisions of hospital medicine.

AMCs and Hospital Performance

AMCs operate 3 related enterprises: a medical school that trains future physicians, a research arena that promotes basic and clinical investigation, and health care services that often encompass both hospitals and clinics. The financial viability of AMCs has always been a topic of debate, largely because of the different missions they pursue and the financial means by which they survive.3133 Over the past decade, cuts in Medicare reimbursement, challenges in balancing bed availability with occupancy rates, and a growing emphasis on cost reduction have created a more competitive health care environment, but without the predicted demise of AMCs.34 Because education and research generally fail to bolster the bottom line, AMCs have focused on optimizing clinical services to promote financial viability.

Hospitalists are uniquely positioned to help this bottom line, just as they do at community hospitals. Their involvement in patient care may produce reductions in length of stay, greater efficiency in discharge planning, and significant cost savings.3537 Hospitalists may also improve throughput in emergency departments and decrease wait times, leading to more efficient bed utilization.38 This leads to a potential for greater hospital revenue by increasing both the number of admissions, particularly surgical cases, and staffed inpatient beds, the latter a premium, as AMCs continue to expand their bed capacity almost annually. Finally, hospitalists may serve as change agents in improving the quality and safety of care delivered, an increasingly important metric given the desire for and expansion of publicly reported measures.

From a financial standpoint, Medicare support to AMCs for training residents now subsidizes fewer clinical care hours. Hospitalist‐driven nonhousestaff services will continue to fulfill a need created by this marked change in residency training. The tension of who pays for nonhousestaff servicesincreased federal support, financial backing from AMCs, or academic department fundsposes an ongoing struggle. In fact, this may be the most important issue currently debated among hospital administrators and department chairs. Regardless, AMCs continue to view hospitalists as a mechanism (or even solution) to maintaining their financial bottom line through improving care delivery systems, adhering to resident work hour restrictions, leading quality and safety improvement initiatives, and improving clinical patient outcomes.

MODELS FOR NONHOUSESTAFF MEDICAL SERVICES

For AMCs developing nonhousestaff services, the process begins by addressing a series of important questions (Table 1). How these questions are answered is often driven by local factors such as the vision of local leadership and the availability of important resources. Nonetheless, it is important for hospitals to share their experiences because best practices remain unclear. Table 2 provides a tabular snapshot of nonhousestaff medicine services at 5 AMCs to highlight similarities and differences. Data in the table were compiled by having a representative from each AMC report the different attributes, which reflects each program as of July 2007. Table 2 provides no data on the quality or efficiency of housestaff versus nonhousestaff services, though this type of investigation is underway and will be critical in future planning.3940