Hospitalists commonly encounter the challenges of infectious diseases in their hospitalized patients. Choosing the correct antibiotic, interpreting blood cultures, working up causes of fever, treating patients with an allergy to penicillin, and caring for patients with human immunodeficiency virus (HIV) commonly confront the hospitalist. This article presents evidence‐based pearls which will help hospitalists avoid common infectious disease pitfalls and guide their decision about when to consult an infectious diseases specialist.

1. Avoid Spiraling Empiricism and Understand Common Fallacies in Prescribing Empiric Antimicrobial Therapy

The term spiraling empiricism describes the inappropriate treatment, or the unjustifiable escalation of treatment, of suspected but undocumented infectious diseases.1 Initiation of carefully considered empiric broad‐spectrum antibiotic therapy for an acutely ill patient is an entirely appropriate and reasonable strategy. But all too often, practitioners are confronted with clinical dilemmas such as persistent fever or lack of response to therapy. In these circumstances, clinicians are faced with deciding whether to add or change antibiotics to broaden coverage. Changes in empiric therapy should be made sparingly, and only when there is new information or symptoms to justify an addition or change. In order to make an accurate assessment of response, steady‐state levels should be achieved and usually 3 to 5 days should be allowed to pass. Lack of response to broad‐spectrum therapy should trigger further investigation for occult infection or consideration of noninfectious etiologies and not simply the addition of a new antimicrobial agent. If a microbial pathogen is isolated from a blood culture(s) or other relevant source, antimicrobials should be tailored to the narrowest spectrum and least toxic therapy based on the sensitivities of that organism. For critically ill patients or patients who do not appear to be improving, an infectious diseases consultation may be warranted.

2. Know the Important Drug‐Drug Interactions Between Antimicrobials and Commonly‐used Inpatient Medications, Particularly With Those Involving Warfarin

Most antimicrobials (especially antifungals, quinolones, metronidazole, and sulfonamides) can cause unpredictable elevations in the international normalized ratio (INR) concurrent with warfarin administration, either through inhibition of warfarin metabolism or alterations in vitamin Kproducing gut flora. When using antimicrobials in patients on warfarin, the patient's INR should be carefully monitored and adjustment of the warfarin dose may be necessary. Antimicrobials that are inhibitors of cytochrome P‐450 enzymes include ciprofloxacin, levofloxacin, isoniazid, fluconazole, and clarithromycin. In contrast, rifampin is a potent inducer of most known cytochrome P‐450 enzymes and increases the metabolism of many drugs used in patients in the hospital setting, including anticonvulsants, beta‐blockers, calcium channel blockers, and other antibiotics like fluoroquinolones, and sulfonylureas. Moreover, the concurrent oral intake of tablets or solutions (including tube feeds) with a high concentration of trivalent and divalent cations (such as aluminum, magnesium, and, to a lesser extent, calcium, iron, and zinc) impairs gastrointestinal absorption of fluoroquinolones and should be avoided or spaced apart in time. Since fluoroquinolones can potentially prolong the QT interval, careful monitoring is necessary when a patient is prescribed other QT prolonging agents. Finally, many antimicrobials reduce the effectiveness of oral or other systemic hormonal contraceptives and patients should be routinely advised to use nonhormonal methods of birth control during therapy.

3. Positive Blood Cultures for Bacteria or Fungus Should be Repeated Serially Every 24 to 48 Hours Until the Cultures Are Negative

An important step in the management of a positive blood culture for bacteria or yeast is to check follow‐up blood cultures every 24 to 48 hours until the bacteremia or fungemia has cleared. This is particularly true of bacteremia caused by Staphylococcus aureus (S. aureus), Enterococcus species, and fungemia caused by Candida species. The duration of bacteremia or fungemia has a significant impact on the predictive values of further testing for endovascular or deep‐seated sources of infection as well as treatment duration. This is particularly true for the treatment of candidemia in nonneutropenic adults and for bacterial endocarditis, in which the recommended duration of treatment starts from the day of the last positive blood culture.2, 3 In addition to repeat blood cultures, a blood culture positive for S. aureus should always prompt an aggressive workup for a source (including strong consideration of a transesophageal echocardiogram to evaluate for endocarditis). S. aureus bacteremia should never be disregarded as a contaminant, and should prompt strong consideration of removal of all indwelling intravenous lines.4

4. Removal of Indwelling Intravascular Catheters Is Essential in the Management of Patients with Candidemia. In These Patients, Retention of Central Lines Is Significantly Related to Poor Outcomes

In patients with culture‐proven Candida fungemia, all intravascular catheters must be removed if at all possible. In a study by Nguyen et al.,5 the mortality rate for patients with a catheter‐related candidemia in whom catheters were retained was significantly higher than that of patients in whom the catheters were removed (41% vs. 21%, P 0.001). Likewise, in a separate study, Luzzati et al.6 noted that central line removal independently reduced the high mortality of the disease. This recommendation applies to all Candida species.

5. Although Candida Species Are Frequently Noted to Colonize Sputum and Urine Cultures, Their Recovery From Multiple Sites May Be an Indicator of Occult Candidemia in an Acutely Ill Patient

Candida species uncommonly cause pneumonia or urinary tract infection, so their isolation from cultures of the respiratory and genitourinary tract often represents colonization. However, the presence of Candida species at multiple sites may be an indicator of occult candidemia in a patient with multiple risk factors for candidemia, including intensive care unit (ICU) admission, immunosuppression (particularly neutropenia and recent receipt of corticosteroids), central venous catheterization, total parenteral nutrition, recent broad‐spectrum antibiotics, and recent abdominal or gastrointestinal surgery.7

6. Patients with Asymptomatic Bacteriuria, With or Without Pyuria, Should Not Be Treated with Antibiotics. Pregnant Women and Patients Undergoing a Genitourinary Procedure Are the Exception and Should Be Treated With Antibiotics

Asymptomatic bacteriuria is commonly encountered in the hospital setting, but is usually benign. Bacteriuria is defined as a voided urine specimen with 1 bacterial species isolated in a quantitative count of 10⁵ cfu/mL. Treatment of asymptomatic bacteriuria is only recommended for pregnant women or prior to invasive genitourinary procedures, including transurethral resection of the prostate. Patients with structural or functional abnormalities of the urinary tract may have a high prevalence of bacteriuria. Despite its prevalence, asymptomatic bacteriuria is seldom associated with adverse outcomes. Studies have noted that antimicrobial treatment of asymptomatic bacteriuria does not decrease recurrence. Negative outcomes with antimicrobial treatment do occur, including adverse drug reactions and reinfection with organisms of increasing resistance. Clinical trials in spinal‐cord injury patients, diabetic women, elderly patients living in the community or nursing home, and patients with indwelling urethral catheters have consistently found no benefit with treatment of asymptomatic bacteriuria.8, 9 The presence or absence of pyuria does not differentiate symptomatic from asymptomatic urinary infection. Patients with symptomatic urinary tract infection (fever and/or dysuria) should be treated after urine cultures are obtained. Other causes of pyuria in the absence of an acute urinary tract infection include urethritis, tuberculosis, prostatitis, nephrolithiasis, and malignancy.

7. Evaluate All Patients Who Have a History of Penicillin Allergy and Consider Desensitization for Patients With a History Consistent With Immunoglobulin Emediated Allergy Who Require Treatment With a Beta‐Lactam Antibiotic

Patients commonly claim to have an allergy to penicillin. True penicillin allergy is very serious and can be life‐threatening. Because of this, patients labeled as penicillin allergic are typically not treated with beta‐lactam antibiotics. Instead, they may be prescribed medications which are typically less effective, more toxic, have a broader spectrum, or are more expensive.10, 11 Many patients are inappropriately labeled as having a penicillin allergy. A history of penicillin allergy is reported in approximately 10% of hospitalized patients, but only approximately 10% of those who report a history of penicillin allergy actually have an allergic reaction when treated with penicillin. Exanthems are frequently associated with beta‐lactam use during an episode of infectious mononucleosis but these are not considered an allergic reaction. Such patients are generally able to tolerate beta‐lactams subsequent to this episode. Nonpruritic maculopapular rashes are also reported in 3% to 7% of children taking amoxicillin and are not a contraindication for future beta‐lactam or cephalosporin use.12 All patients who describe an allergy should be questioned in detail about the type of penicillin received, as well as the type, severity, and timing of the reaction. Typical immunoglobulin E (IgE)‐mediated severe reactions to penicillin include urticaria, pruritus, angioedema, bronchospasm, and hypotension. These patients should not be given other agents that share the same beta‐lactam ring, including cephalosporins (risk of cross‐reactivity is greatest with first‐generation and second‐generation cephalosporins). Carbapenems have minimal cross‐reactivity, particularly meropenem.13 Monobactams (eg, aztreonam) do not cross‐react. While skin testing to penicillin can be considered in patients with a history of a severe reaction to penicillin, neither the major nor minor determinants are commercially available at this time. In patients with a history of a possible IgE‐mediated reaction and when there is no suitable alternative antibiotic (usually determined from infectious diseases consultation), desensitization to beta‐lactams or carbapenems can be considered. Desensitization should be reserved only for clinicians experienced with these techniques, preferably in consultation with a specialist in allergy and immunology. Patients who report a non‐IgE‐mediated reaction may be prescribed a cephalosporin if necessary (preferably a third‐generation or fourth‐generation).14

8. An Abrupt Increase in Leukocytosis In a Hospitalized Patient Should Prompt Consideration of Clostridium difficile Infection

In recent years, there has been a marked increase in the incidence and severity of Clostridium difficile (C. difficile) infection (CDI). A new hypervirulent strain, NAP1/BI/027, has emerged and is becoming endemic in the United States, Canada, and Europe. Typically C. difficile causes diarrhea, abdominal pain, and fever. Often patients have received antibiotics in the recent past, placing them at higher risk, but cases can occur sporadically (even in the community setting) or be transmitted nosocomially. Early detection appears to be essential in reducing the serious morbidity and mortality associated with this disease. Observational studies suggested that C. difficile infection is a common cause of unexplained leukocytosis or a sudden worsening of preexisting leukocytosis.15, 16 In a prospective study evaluating 60 patients with unexplained leukocytosis (white blood cell count 15,000/mm³), 58% of patients with leukocytosis in the absence of localizing symptoms and signs of infection were subsequently diagnosed with CDI. The authors believe that the percent may have been as high as 73% when they included patients with a negative toxin assay who rapidly responded to metronidazole therapy.17 White blood cell counts can range from 10,000 to 20,000/mm³ in moderate disease. Counts as high as 40,000/mm³ can occur, especially in patients with severe disease. Although the use of clindamycin and cephalosporins have been classically associated with the subsequent development of CDI, the current widespread use of fluoroquinolones has led to significant fluoroquinolone resistance among strains of C. difficile, especially the hypervirulent NAP1/BI/027 strain.18 The judicious use of antibiotics, especially fluoroquinolones, remains the cornerstone in preventing CDI. Remember that hand washing with soap and water is essential as alcohol‐based hand sanitizers do not eradicate the C. difficile spores. The drug of choice for initial treatment of mild to moderate CDI remains oral metronidazole, and it may be used for a first recurrence of CDI. Increasing data support the use of oral vancomycin for moderately severe to severe CDI or for multiple recurrences.19 Intravenous metronidazole is often added to oral vancomycin in patients with ileus, but it is not reliably effective alone for CDI.

9. Fever Is Common in the First 48 Hours After a Major Surgical Procedure, and Is a Poor Indicator of Infection. The use of Antibiotics in Response to Fever in the Absence of Other Localizing Signs and Symptoms of Infection Should Be Avoided

Early postoperative fever is relatively common but most fevers that develop within the first 48 hours after surgery do not have an infectious etiology.2023 However, fever that begins or persists beyond the fifth postoperative day is much more likely to represent a clinically significant infection. The continued use of antibiotics outside the window for wound prophylaxis (>24 hours) does not decrease the risk of postoperative infection but it does increase the risk of acquiring resistant bacteria and adverse drug reactions, including CDI.

10. Facts All Clinicians Should Know About Patients with HIV Infection

The 2 most common laboratory abnormalities routinely associated with antiretroviral therapy for HIV infection are unconjugated hyperbilirubinemia associated with atazanavir and an elevated mean corpuscular volume (MCV) associated with zidovudine (and, to a lesser extent, stavudine). Immune reconstitution inflammatory syndrome (IRIS) is a condition seen in patients with advanced acquired immune deficiency syndrome (AIDS) who have recently started antiretroviral therapy. As the immune system begins to recover, it may respond to a previously acquired opportunistic infection with an overwhelming inflammatory response that paradoxically makes the symptoms of infection worse. IRIS is associated with a pathological inflammatory response that can have substantial morbidity and mortality.24 For this reason, when considering whether to start or stop continuous or highly active antiretroviral therapy (also known as HAART), an infectious diseases consult is recommended. Pneumocystis jiroveci (PCP) remains a cause of pneumonia in patients with advanced AIDS' though in the era of HAART, its presentation may be more subtle. Finally, the principle of parsimony (Occam's razor) often does not hold in the diagnosis of opportunistic infections in patients with advanced AIDS, as these patients can often present with multiple infections simultaneously.25, 26

Conclusion

Infectious diseases are commonly encountered by physicians who care for hospitalized patients. Early recognition, evaluation, and appropriate treatment and/or referral to an infectious diseases specialist are necessary to moderate the significant morbidity and mortality that are often associated with infectious diseases.

Hospitalists commonly encounter the challenges of infectious diseases in their hospitalized patients. Choosing the correct antibiotic, interpreting blood cultures, working up causes of fever, treating patients with an allergy to penicillin, and caring for patients with human immunodeficiency virus (HIV) commonly confront the hospitalist. This article presents evidence‐based pearls which will help hospitalists avoid common infectious disease pitfalls and guide their decision about when to consult an infectious diseases specialist.

1. Avoid Spiraling Empiricism and Understand Common Fallacies in Prescribing Empiric Antimicrobial Therapy

The term spiraling empiricism describes the inappropriate treatment, or the unjustifiable escalation of treatment, of suspected but undocumented infectious diseases.1 Initiation of carefully considered empiric broad‐spectrum antibiotic therapy for an acutely ill patient is an entirely appropriate and reasonable strategy. But all too often, practitioners are confronted with clinical dilemmas such as persistent fever or lack of response to therapy. In these circumstances, clinicians are faced with deciding whether to add or change antibiotics to broaden coverage. Changes in empiric therapy should be made sparingly, and only when there is new information or symptoms to justify an addition or change. In order to make an accurate assessment of response, steady‐state levels should be achieved and usually 3 to 5 days should be allowed to pass. Lack of response to broad‐spectrum therapy should trigger further investigation for occult infection or consideration of noninfectious etiologies and not simply the addition of a new antimicrobial agent. If a microbial pathogen is isolated from a blood culture(s) or other relevant source, antimicrobials should be tailored to the narrowest spectrum and least toxic therapy based on the sensitivities of that organism. For critically ill patients or patients who do not appear to be improving, an infectious diseases consultation may be warranted.

2. Know the Important Drug‐Drug Interactions Between Antimicrobials and Commonly‐used Inpatient Medications, Particularly With Those Involving Warfarin

Most antimicrobials (especially antifungals, quinolones, metronidazole, and sulfonamides) can cause unpredictable elevations in the international normalized ratio (INR) concurrent with warfarin administration, either through inhibition of warfarin metabolism or alterations in vitamin Kproducing gut flora. When using antimicrobials in patients on warfarin, the patient's INR should be carefully monitored and adjustment of the warfarin dose may be necessary. Antimicrobials that are inhibitors of cytochrome P‐450 enzymes include ciprofloxacin, levofloxacin, isoniazid, fluconazole, and clarithromycin. In contrast, rifampin is a potent inducer of most known cytochrome P‐450 enzymes and increases the metabolism of many drugs used in patients in the hospital setting, including anticonvulsants, beta‐blockers, calcium channel blockers, and other antibiotics like fluoroquinolones, and sulfonylureas. Moreover, the concurrent oral intake of tablets or solutions (including tube feeds) with a high concentration of trivalent and divalent cations (such as aluminum, magnesium, and, to a lesser extent, calcium, iron, and zinc) impairs gastrointestinal absorption of fluoroquinolones and should be avoided or spaced apart in time. Since fluoroquinolones can potentially prolong the QT interval, careful monitoring is necessary when a patient is prescribed other QT prolonging agents. Finally, many antimicrobials reduce the effectiveness of oral or other systemic hormonal contraceptives and patients should be routinely advised to use nonhormonal methods of birth control during therapy.

3. Positive Blood Cultures for Bacteria or Fungus Should be Repeated Serially Every 24 to 48 Hours Until the Cultures Are Negative

An important step in the management of a positive blood culture for bacteria or yeast is to check follow‐up blood cultures every 24 to 48 hours until the bacteremia or fungemia has cleared. This is particularly true of bacteremia caused by Staphylococcus aureus (S. aureus), Enterococcus species, and fungemia caused by Candida species. The duration of bacteremia or fungemia has a significant impact on the predictive values of further testing for endovascular or deep‐seated sources of infection as well as treatment duration. This is particularly true for the treatment of candidemia in nonneutropenic adults and for bacterial endocarditis, in which the recommended duration of treatment starts from the day of the last positive blood culture.2, 3 In addition to repeat blood cultures, a blood culture positive for S. aureus should always prompt an aggressive workup for a source (including strong consideration of a transesophageal echocardiogram to evaluate for endocarditis). S. aureus bacteremia should never be disregarded as a contaminant, and should prompt strong consideration of removal of all indwelling intravenous lines.4

4. Removal of Indwelling Intravascular Catheters Is Essential in the Management of Patients with Candidemia. In These Patients, Retention of Central Lines Is Significantly Related to Poor Outcomes

In patients with culture‐proven Candida fungemia, all intravascular catheters must be removed if at all possible. In a study by Nguyen et al.,5 the mortality rate for patients with a catheter‐related candidemia in whom catheters were retained was significantly higher than that of patients in whom the catheters were removed (41% vs. 21%, P 0.001). Likewise, in a separate study, Luzzati et al.6 noted that central line removal independently reduced the high mortality of the disease. This recommendation applies to all Candida species.

5. Although Candida Species Are Frequently Noted to Colonize Sputum and Urine Cultures, Their Recovery From Multiple Sites May Be an Indicator of Occult Candidemia in an Acutely Ill Patient

Candida species uncommonly cause pneumonia or urinary tract infection, so their isolation from cultures of the respiratory and genitourinary tract often represents colonization. However, the presence of Candida species at multiple sites may be an indicator of occult candidemia in a patient with multiple risk factors for candidemia, including intensive care unit (ICU) admission, immunosuppression (particularly neutropenia and recent receipt of corticosteroids), central venous catheterization, total parenteral nutrition, recent broad‐spectrum antibiotics, and recent abdominal or gastrointestinal surgery.7

6. Patients with Asymptomatic Bacteriuria, With or Without Pyuria, Should Not Be Treated with Antibiotics. Pregnant Women and Patients Undergoing a Genitourinary Procedure Are the Exception and Should Be Treated With Antibiotics

Asymptomatic bacteriuria is commonly encountered in the hospital setting, but is usually benign. Bacteriuria is defined as a voided urine specimen with 1 bacterial species isolated in a quantitative count of 10⁵ cfu/mL. Treatment of asymptomatic bacteriuria is only recommended for pregnant women or prior to invasive genitourinary procedures, including transurethral resection of the prostate. Patients with structural or functional abnormalities of the urinary tract may have a high prevalence of bacteriuria. Despite its prevalence, asymptomatic bacteriuria is seldom associated with adverse outcomes. Studies have noted that antimicrobial treatment of asymptomatic bacteriuria does not decrease recurrence. Negative outcomes with antimicrobial treatment do occur, including adverse drug reactions and reinfection with organisms of increasing resistance. Clinical trials in spinal‐cord injury patients, diabetic women, elderly patients living in the community or nursing home, and patients with indwelling urethral catheters have consistently found no benefit with treatment of asymptomatic bacteriuria.8, 9 The presence or absence of pyuria does not differentiate symptomatic from asymptomatic urinary infection. Patients with symptomatic urinary tract infection (fever and/or dysuria) should be treated after urine cultures are obtained. Other causes of pyuria in the absence of an acute urinary tract infection include urethritis, tuberculosis, prostatitis, nephrolithiasis, and malignancy.

7. Evaluate All Patients Who Have a History of Penicillin Allergy and Consider Desensitization for Patients With a History Consistent With Immunoglobulin Emediated Allergy Who Require Treatment With a Beta‐Lactam Antibiotic

Patients commonly claim to have an allergy to penicillin. True penicillin allergy is very serious and can be life‐threatening. Because of this, patients labeled as penicillin allergic are typically not treated with beta‐lactam antibiotics. Instead, they may be prescribed medications which are typically less effective, more toxic, have a broader spectrum, or are more expensive.10, 11 Many patients are inappropriately labeled as having a penicillin allergy. A history of penicillin allergy is reported in approximately 10% of hospitalized patients, but only approximately 10% of those who report a history of penicillin allergy actually have an allergic reaction when treated with penicillin. Exanthems are frequently associated with beta‐lactam use during an episode of infectious mononucleosis but these are not considered an allergic reaction. Such patients are generally able to tolerate beta‐lactams subsequent to this episode. Nonpruritic maculopapular rashes are also reported in 3% to 7% of children taking amoxicillin and are not a contraindication for future beta‐lactam or cephalosporin use.12 All patients who describe an allergy should be questioned in detail about the type of penicillin received, as well as the type, severity, and timing of the reaction. Typical immunoglobulin E (IgE)‐mediated severe reactions to penicillin include urticaria, pruritus, angioedema, bronchospasm, and hypotension. These patients should not be given other agents that share the same beta‐lactam ring, including cephalosporins (risk of cross‐reactivity is greatest with first‐generation and second‐generation cephalosporins). Carbapenems have minimal cross‐reactivity, particularly meropenem.13 Monobactams (eg, aztreonam) do not cross‐react. While skin testing to penicillin can be considered in patients with a history of a severe reaction to penicillin, neither the major nor minor determinants are commercially available at this time. In patients with a history of a possible IgE‐mediated reaction and when there is no suitable alternative antibiotic (usually determined from infectious diseases consultation), desensitization to beta‐lactams or carbapenems can be considered. Desensitization should be reserved only for clinicians experienced with these techniques, preferably in consultation with a specialist in allergy and immunology. Patients who report a non‐IgE‐mediated reaction may be prescribed a cephalosporin if necessary (preferably a third‐generation or fourth‐generation).14

8. An Abrupt Increase in Leukocytosis In a Hospitalized Patient Should Prompt Consideration of Clostridium difficile Infection

In recent years, there has been a marked increase in the incidence and severity of Clostridium difficile (C. difficile) infection (CDI). A new hypervirulent strain, NAP1/BI/027, has emerged and is becoming endemic in the United States, Canada, and Europe. Typically C. difficile causes diarrhea, abdominal pain, and fever. Often patients have received antibiotics in the recent past, placing them at higher risk, but cases can occur sporadically (even in the community setting) or be transmitted nosocomially. Early detection appears to be essential in reducing the serious morbidity and mortality associated with this disease. Observational studies suggested that C. difficile infection is a common cause of unexplained leukocytosis or a sudden worsening of preexisting leukocytosis.15, 16 In a prospective study evaluating 60 patients with unexplained leukocytosis (white blood cell count 15,000/mm³), 58% of patients with leukocytosis in the absence of localizing symptoms and signs of infection were subsequently diagnosed with CDI. The authors believe that the percent may have been as high as 73% when they included patients with a negative toxin assay who rapidly responded to metronidazole therapy.17 White blood cell counts can range from 10,000 to 20,000/mm³ in moderate disease. Counts as high as 40,000/mm³ can occur, especially in patients with severe disease. Although the use of clindamycin and cephalosporins have been classically associated with the subsequent development of CDI, the current widespread use of fluoroquinolones has led to significant fluoroquinolone resistance among strains of C. difficile, especially the hypervirulent NAP1/BI/027 strain.18 The judicious use of antibiotics, especially fluoroquinolones, remains the cornerstone in preventing CDI. Remember that hand washing with soap and water is essential as alcohol‐based hand sanitizers do not eradicate the C. difficile spores. The drug of choice for initial treatment of mild to moderate CDI remains oral metronidazole, and it may be used for a first recurrence of CDI. Increasing data support the use of oral vancomycin for moderately severe to severe CDI or for multiple recurrences.19 Intravenous metronidazole is often added to oral vancomycin in patients with ileus, but it is not reliably effective alone for CDI.

9. Fever Is Common in the First 48 Hours After a Major Surgical Procedure, and Is a Poor Indicator of Infection. The use of Antibiotics in Response to Fever in the Absence of Other Localizing Signs and Symptoms of Infection Should Be Avoided

Early postoperative fever is relatively common but most fevers that develop within the first 48 hours after surgery do not have an infectious etiology.2023 However, fever that begins or persists beyond the fifth postoperative day is much more likely to represent a clinically significant infection. The continued use of antibiotics outside the window for wound prophylaxis (>24 hours) does not decrease the risk of postoperative infection but it does increase the risk of acquiring resistant bacteria and adverse drug reactions, including CDI.

10. Facts All Clinicians Should Know About Patients with HIV Infection

The 2 most common laboratory abnormalities routinely associated with antiretroviral therapy for HIV infection are unconjugated hyperbilirubinemia associated with atazanavir and an elevated mean corpuscular volume (MCV) associated with zidovudine (and, to a lesser extent, stavudine). Immune reconstitution inflammatory syndrome (IRIS) is a condition seen in patients with advanced acquired immune deficiency syndrome (AIDS) who have recently started antiretroviral therapy. As the immune system begins to recover, it may respond to a previously acquired opportunistic infection with an overwhelming inflammatory response that paradoxically makes the symptoms of infection worse. IRIS is associated with a pathological inflammatory response that can have substantial morbidity and mortality.24 For this reason, when considering whether to start or stop continuous or highly active antiretroviral therapy (also known as HAART), an infectious diseases consult is recommended. Pneumocystis jiroveci (PCP) remains a cause of pneumonia in patients with advanced AIDS' though in the era of HAART, its presentation may be more subtle. Finally, the principle of parsimony (Occam's razor) often does not hold in the diagnosis of opportunistic infections in patients with advanced AIDS, as these patients can often present with multiple infections simultaneously.25, 26

Conclusion

Infectious diseases are commonly encountered by physicians who care for hospitalized patients. Early recognition, evaluation, and appropriate treatment and/or referral to an infectious diseases specialist are necessary to moderate the significant morbidity and mortality that are often associated with infectious diseases.

Hospitalists commonly encounter the challenges of infectious diseases in their hospitalized patients. Choosing the correct antibiotic, interpreting blood cultures, working up causes of fever, treating patients with an allergy to penicillin, and caring for patients with human immunodeficiency virus (HIV) commonly confront the hospitalist. This article presents evidence‐based pearls which will help hospitalists avoid common infectious disease pitfalls and guide their decision about when to consult an infectious diseases specialist.

1. Avoid Spiraling Empiricism and Understand Common Fallacies in Prescribing Empiric Antimicrobial Therapy

The term spiraling empiricism describes the inappropriate treatment, or the unjustifiable escalation of treatment, of suspected but undocumented infectious diseases.1 Initiation of carefully considered empiric broad‐spectrum antibiotic therapy for an acutely ill patient is an entirely appropriate and reasonable strategy. But all too often, practitioners are confronted with clinical dilemmas such as persistent fever or lack of response to therapy. In these circumstances, clinicians are faced with deciding whether to add or change antibiotics to broaden coverage. Changes in empiric therapy should be made sparingly, and only when there is new information or symptoms to justify an addition or change. In order to make an accurate assessment of response, steady‐state levels should be achieved and usually 3 to 5 days should be allowed to pass. Lack of response to broad‐spectrum therapy should trigger further investigation for occult infection or consideration of noninfectious etiologies and not simply the addition of a new antimicrobial agent. If a microbial pathogen is isolated from a blood culture(s) or other relevant source, antimicrobials should be tailored to the narrowest spectrum and least toxic therapy based on the sensitivities of that organism. For critically ill patients or patients who do not appear to be improving, an infectious diseases consultation may be warranted.

2. Know the Important Drug‐Drug Interactions Between Antimicrobials and Commonly‐used Inpatient Medications, Particularly With Those Involving Warfarin

Most antimicrobials (especially antifungals, quinolones, metronidazole, and sulfonamides) can cause unpredictable elevations in the international normalized ratio (INR) concurrent with warfarin administration, either through inhibition of warfarin metabolism or alterations in vitamin Kproducing gut flora. When using antimicrobials in patients on warfarin, the patient's INR should be carefully monitored and adjustment of the warfarin dose may be necessary. Antimicrobials that are inhibitors of cytochrome P‐450 enzymes include ciprofloxacin, levofloxacin, isoniazid, fluconazole, and clarithromycin. In contrast, rifampin is a potent inducer of most known cytochrome P‐450 enzymes and increases the metabolism of many drugs used in patients in the hospital setting, including anticonvulsants, beta‐blockers, calcium channel blockers, and other antibiotics like fluoroquinolones, and sulfonylureas. Moreover, the concurrent oral intake of tablets or solutions (including tube feeds) with a high concentration of trivalent and divalent cations (such as aluminum, magnesium, and, to a lesser extent, calcium, iron, and zinc) impairs gastrointestinal absorption of fluoroquinolones and should be avoided or spaced apart in time. Since fluoroquinolones can potentially prolong the QT interval, careful monitoring is necessary when a patient is prescribed other QT prolonging agents. Finally, many antimicrobials reduce the effectiveness of oral or other systemic hormonal contraceptives and patients should be routinely advised to use nonhormonal methods of birth control during therapy.

3. Positive Blood Cultures for Bacteria or Fungus Should be Repeated Serially Every 24 to 48 Hours Until the Cultures Are Negative

An important step in the management of a positive blood culture for bacteria or yeast is to check follow‐up blood cultures every 24 to 48 hours until the bacteremia or fungemia has cleared. This is particularly true of bacteremia caused by Staphylococcus aureus (S. aureus), Enterococcus species, and fungemia caused by Candida species. The duration of bacteremia or fungemia has a significant impact on the predictive values of further testing for endovascular or deep‐seated sources of infection as well as treatment duration. This is particularly true for the treatment of candidemia in nonneutropenic adults and for bacterial endocarditis, in which the recommended duration of treatment starts from the day of the last positive blood culture.2, 3 In addition to repeat blood cultures, a blood culture positive for S. aureus should always prompt an aggressive workup for a source (including strong consideration of a transesophageal echocardiogram to evaluate for endocarditis). S. aureus bacteremia should never be disregarded as a contaminant, and should prompt strong consideration of removal of all indwelling intravenous lines.4

4. Removal of Indwelling Intravascular Catheters Is Essential in the Management of Patients with Candidemia. In These Patients, Retention of Central Lines Is Significantly Related to Poor Outcomes

In patients with culture‐proven Candida fungemia, all intravascular catheters must be removed if at all possible. In a study by Nguyen et al.,5 the mortality rate for patients with a catheter‐related candidemia in whom catheters were retained was significantly higher than that of patients in whom the catheters were removed (41% vs. 21%, P 0.001). Likewise, in a separate study, Luzzati et al.6 noted that central line removal independently reduced the high mortality of the disease. This recommendation applies to all Candida species.

5. Although Candida Species Are Frequently Noted to Colonize Sputum and Urine Cultures, Their Recovery From Multiple Sites May Be an Indicator of Occult Candidemia in an Acutely Ill Patient

Candida species uncommonly cause pneumonia or urinary tract infection, so their isolation from cultures of the respiratory and genitourinary tract often represents colonization. However, the presence of Candida species at multiple sites may be an indicator of occult candidemia in a patient with multiple risk factors for candidemia, including intensive care unit (ICU) admission, immunosuppression (particularly neutropenia and recent receipt of corticosteroids), central venous catheterization, total parenteral nutrition, recent broad‐spectrum antibiotics, and recent abdominal or gastrointestinal surgery.7

6. Patients with Asymptomatic Bacteriuria, With or Without Pyuria, Should Not Be Treated with Antibiotics. Pregnant Women and Patients Undergoing a Genitourinary Procedure Are the Exception and Should Be Treated With Antibiotics

Asymptomatic bacteriuria is commonly encountered in the hospital setting, but is usually benign. Bacteriuria is defined as a voided urine specimen with 1 bacterial species isolated in a quantitative count of 10⁵ cfu/mL. Treatment of asymptomatic bacteriuria is only recommended for pregnant women or prior to invasive genitourinary procedures, including transurethral resection of the prostate. Patients with structural or functional abnormalities of the urinary tract may have a high prevalence of bacteriuria. Despite its prevalence, asymptomatic bacteriuria is seldom associated with adverse outcomes. Studies have noted that antimicrobial treatment of asymptomatic bacteriuria does not decrease recurrence. Negative outcomes with antimicrobial treatment do occur, including adverse drug reactions and reinfection with organisms of increasing resistance. Clinical trials in spinal‐cord injury patients, diabetic women, elderly patients living in the community or nursing home, and patients with indwelling urethral catheters have consistently found no benefit with treatment of asymptomatic bacteriuria.8, 9 The presence or absence of pyuria does not differentiate symptomatic from asymptomatic urinary infection. Patients with symptomatic urinary tract infection (fever and/or dysuria) should be treated after urine cultures are obtained. Other causes of pyuria in the absence of an acute urinary tract infection include urethritis, tuberculosis, prostatitis, nephrolithiasis, and malignancy.

7. Evaluate All Patients Who Have a History of Penicillin Allergy and Consider Desensitization for Patients With a History Consistent With Immunoglobulin Emediated Allergy Who Require Treatment With a Beta‐Lactam Antibiotic

Patients commonly claim to have an allergy to penicillin. True penicillin allergy is very serious and can be life‐threatening. Because of this, patients labeled as penicillin allergic are typically not treated with beta‐lactam antibiotics. Instead, they may be prescribed medications which are typically less effective, more toxic, have a broader spectrum, or are more expensive.10, 11 Many patients are inappropriately labeled as having a penicillin allergy. A history of penicillin allergy is reported in approximately 10% of hospitalized patients, but only approximately 10% of those who report a history of penicillin allergy actually have an allergic reaction when treated with penicillin. Exanthems are frequently associated with beta‐lactam use during an episode of infectious mononucleosis but these are not considered an allergic reaction. Such patients are generally able to tolerate beta‐lactams subsequent to this episode. Nonpruritic maculopapular rashes are also reported in 3% to 7% of children taking amoxicillin and are not a contraindication for future beta‐lactam or cephalosporin use.12 All patients who describe an allergy should be questioned in detail about the type of penicillin received, as well as the type, severity, and timing of the reaction. Typical immunoglobulin E (IgE)‐mediated severe reactions to penicillin include urticaria, pruritus, angioedema, bronchospasm, and hypotension. These patients should not be given other agents that share the same beta‐lactam ring, including cephalosporins (risk of cross‐reactivity is greatest with first‐generation and second‐generation cephalosporins). Carbapenems have minimal cross‐reactivity, particularly meropenem.13 Monobactams (eg, aztreonam) do not cross‐react. While skin testing to penicillin can be considered in patients with a history of a severe reaction to penicillin, neither the major nor minor determinants are commercially available at this time. In patients with a history of a possible IgE‐mediated reaction and when there is no suitable alternative antibiotic (usually determined from infectious diseases consultation), desensitization to beta‐lactams or carbapenems can be considered. Desensitization should be reserved only for clinicians experienced with these techniques, preferably in consultation with a specialist in allergy and immunology. Patients who report a non‐IgE‐mediated reaction may be prescribed a cephalosporin if necessary (preferably a third‐generation or fourth‐generation).14

8. An Abrupt Increase in Leukocytosis In a Hospitalized Patient Should Prompt Consideration of Clostridium difficile Infection

In recent years, there has been a marked increase in the incidence and severity of Clostridium difficile (C. difficile) infection (CDI). A new hypervirulent strain, NAP1/BI/027, has emerged and is becoming endemic in the United States, Canada, and Europe. Typically C. difficile causes diarrhea, abdominal pain, and fever. Often patients have received antibiotics in the recent past, placing them at higher risk, but cases can occur sporadically (even in the community setting) or be transmitted nosocomially. Early detection appears to be essential in reducing the serious morbidity and mortality associated with this disease. Observational studies suggested that C. difficile infection is a common cause of unexplained leukocytosis or a sudden worsening of preexisting leukocytosis.15, 16 In a prospective study evaluating 60 patients with unexplained leukocytosis (white blood cell count 15,000/mm³), 58% of patients with leukocytosis in the absence of localizing symptoms and signs of infection were subsequently diagnosed with CDI. The authors believe that the percent may have been as high as 73% when they included patients with a negative toxin assay who rapidly responded to metronidazole therapy.17 White blood cell counts can range from 10,000 to 20,000/mm³ in moderate disease. Counts as high as 40,000/mm³ can occur, especially in patients with severe disease. Although the use of clindamycin and cephalosporins have been classically associated with the subsequent development of CDI, the current widespread use of fluoroquinolones has led to significant fluoroquinolone resistance among strains of C. difficile, especially the hypervirulent NAP1/BI/027 strain.18 The judicious use of antibiotics, especially fluoroquinolones, remains the cornerstone in preventing CDI. Remember that hand washing with soap and water is essential as alcohol‐based hand sanitizers do not eradicate the C. difficile spores. The drug of choice for initial treatment of mild to moderate CDI remains oral metronidazole, and it may be used for a first recurrence of CDI. Increasing data support the use of oral vancomycin for moderately severe to severe CDI or for multiple recurrences.19 Intravenous metronidazole is often added to oral vancomycin in patients with ileus, but it is not reliably effective alone for CDI.

9. Fever Is Common in the First 48 Hours After a Major Surgical Procedure, and Is a Poor Indicator of Infection. The use of Antibiotics in Response to Fever in the Absence of Other Localizing Signs and Symptoms of Infection Should Be Avoided

Early postoperative fever is relatively common but most fevers that develop within the first 48 hours after surgery do not have an infectious etiology.2023 However, fever that begins or persists beyond the fifth postoperative day is much more likely to represent a clinically significant infection. The continued use of antibiotics outside the window for wound prophylaxis (>24 hours) does not decrease the risk of postoperative infection but it does increase the risk of acquiring resistant bacteria and adverse drug reactions, including CDI.

10. Facts All Clinicians Should Know About Patients with HIV Infection

The 2 most common laboratory abnormalities routinely associated with antiretroviral therapy for HIV infection are unconjugated hyperbilirubinemia associated with atazanavir and an elevated mean corpuscular volume (MCV) associated with zidovudine (and, to a lesser extent, stavudine). Immune reconstitution inflammatory syndrome (IRIS) is a condition seen in patients with advanced acquired immune deficiency syndrome (AIDS) who have recently started antiretroviral therapy. As the immune system begins to recover, it may respond to a previously acquired opportunistic infection with an overwhelming inflammatory response that paradoxically makes the symptoms of infection worse. IRIS is associated with a pathological inflammatory response that can have substantial morbidity and mortality.24 For this reason, when considering whether to start or stop continuous or highly active antiretroviral therapy (also known as HAART), an infectious diseases consult is recommended. Pneumocystis jiroveci (PCP) remains a cause of pneumonia in patients with advanced AIDS' though in the era of HAART, its presentation may be more subtle. Finally, the principle of parsimony (Occam's razor) often does not hold in the diagnosis of opportunistic infections in patients with advanced AIDS, as these patients can often present with multiple infections simultaneously.25, 26

Conclusion

Infectious diseases are commonly encountered by physicians who care for hospitalized patients. Early recognition, evaluation, and appropriate treatment and/or referral to an infectious diseases specialist are necessary to moderate the significant morbidity and mortality that are often associated with infectious diseases.

A looming gap in the supply of intensivists prompted the American College of Chest Physicians (ACCP), the American Thoracic Society (ATS), and the Society of Critical Care Medicine (SCCM) to publish a report in 2000 by the Committee on Manpower for Pulmonary and Critical Care Societies (COMPACCS). This study predicted that beginning in 2007 a shortfall would become apparent and steadily increase to 22% by 2020 and to 35% by 2030. Subsequent reports have reiterated those projections, including a report to Congress in 2006 by the U.S. Department of Health and Human Services/Health Resources and Services Administration.14

The concern regarding the shortage of intensivists has been increased by the growing evidence that supports improved critical care outcomesespecially decreased intensive care unit (ICU) and hospital mortalitywith intensivist staffing of ICUs.5, 6 Based on this data and on recommendations from the Society of Critical Care Medicine, the Leapfrog Group made onsite, high‐intensity ICU staffing with intensivists 1 of their 4 leaps.7 A paper by Pronovost et al.8 published in 2001, however, noted that in order for all ICUs in the United States to meet the Leapfrog ICU Physician Staffing (IPS) standard, the number of intensivists would need to increase by a factor of 2.6. Interestingly, a retrospective study published in the Annals of Internal Medicine in June of 2008 by Levy et al.9 suggested that mortality rates may actually be higher in intensivist‐staffed ICUs. An accompanying editorial raised concerns about limitations of the study design, but endorsed Levy's recommendation that more carefully designed, prospective studies were needed; (ie, we still are not certain as to optimal physician staffing for the care of patients requiring the sophisticated treatment available only in an ICU.)10

The health policy challenge, however, remains clear: while there is basic consensus that care of critically ill patients by intensivists improves outcomes, the reality is that the shortage of intensivists in the United States as predicted by the COMPACCS report will only increase, leading some to refer to this as a healthcare crisis. Two major task forces attempted to address this situation, resulting in the publication of the 2004 Framing Options for Critical Care in the United States (FOCCUS) report, The Critical Care Medicine Crisis: A Call for Federal ActionA White Paper from the Critical Care Professional Societies; and the 2007 Prioritizing the Organization and Management of Intensive Care Services in the Unites States (PrOMIS) Conference Report.11, 12 Both reports made specific recommendations including, for example, development of uniform standards for accreditation of institutional critical care capacity, identification and endorsement of core competencies in critical care, investment in health services research, the use of uniform protocols for ICU care, leverage of information technology to promote standardization and improve efficiency, and the development of incentives to attract healthcare professionals to critical care medicine.

A Possible Solution: The Role of Hospitalists

Multiple important efforts are already underway to increase the competency of professionals providing critical care services including the Society of Critical Care's Fundamentals in Critical Care Support (FCCS) program. Additionally, physician assistants and nurse practitioners are playing an increasingly important role as members of critical care services. As another component of this collaborative effort, the PrOMIS Report noted the potential impact of hospitalists in addressing this crisis.

As early as 1999, surveys revealed that as many as 35% of hospitalists were providing critical care services.13 According to the 2005/2006 Society of Hospital Medicine (SHM) National Survey, that number has increased to 75% with a low of 66% in the eastern United States and a high of 84% in the western United States. In community hospitals, 87% of hospitalists care for patients in the ICU, and 30% provide critical care services in academic medical centers.13 While there is some research14, 15 and many anecdotal reports that suggest hospitalists perform well in the ICU, there is, unfortunately, little data addressing outcomes for patients cared for by hospitalists. The results from a prospective, severity‐adjusted study from the Emory University Section of Hospital Medicine and the Division of Pulmonary/Critical Care Medicine examining outcomes for critical care patients cared for by hospitalists with criteria for Pulmonary/Intensivist consults vs. patients cared for by the Pulmonary/Critical Care Medical ICU Service await peer‐review publication.

Despite the lack of outcome data regarding adult hospitalists, it is clear that by default they are already providing a significant proportion of critical care services across the healthcare system, including in tertiary care centers. The two primary models of care include: (1) hospitalists serving as the primary provider without critical care consultant services and (2) comanagement of patients where intensivists and hospitalists collaborate. These collaborative models involve hospitalists actively co‐managing critical care patients along with intensivists or hospitalists managing less critically ill patients with intensivist consultation when indicated. In hospitals lacking intensivists, hospitalists often manage critically ill patients either with intensivist phone consultation, or with the intent to stabilize and transfer. Electronic ICUs are another expanding model of care that provide intensivist support to hospitalists and other primary care providersdecreasing ICU length of stay and severity‐adjusted ICU mortality.16 There are now 40 electronic ICU programs in the United States, and that number continues to grow.

In 2003, there were approximately 10,000 hospitalists in the United States,17 and recent data from an American Hospital Association survey indicates that the number has grown to about 28,000 in 2009. Recent research also documents that hospitalists are soon likely to care for the majority of elderly hospitalized patients in America.18 Aware that the number of intensivists is unlikely to change significantly over the next 25 years the question is no longer if hospitalists should be in the ICU; rather, the question is how to assure quality and improved clinical outcomes through enhanced collaboration between Hospital Medicine and Critical Care Medicine.

Recommendations

There are 3 steps that should be taken urgently to meet this challenge:

• 1

  Per the recommendation of the FOCCUS Report and the PrOMIS Conference Report, uniform protocols for intensive care treatmentmany of which already exist but are not used consistentlyshould be identified and implemented across all ICUs regardless of the level or certification of the provider.

• 2

  Also per the PrOMIS Report, a process for certification of physicians providing critical care services should be established by the appropriate governing bodies, including the Society for Critical Care Medicine, the Society of Hospital Medicine, and the American Thoracic Society, among others. While the PrOMIS Report called for cross‐training of hospital‐based providers to provide intensive care services in lower tier hospitals, a more realistic recommendation given current involvement of hospitalists in the provision of critical care services in secondary and tertiary centers is a competency‐assurance process that includes hospitalists practicing at all levels. This would not be equivalent to board certification, but would be based on a rigorous, comprehensive education and skills training process leading to recognition that would distinguish the recipient as having competencies beyond those obtained in internal medicine residency training. Models for certification could include 4‐month onsite training or a distance learning curriculum with regular blocks of onsite training. Another strategy might be for appropriate governing bodies to establish basic criteria for competency that would then be provided by individual institutions. Emory University, for example, has developed a pilot program incorporating significant components from the European Society for Critical Care Medicine's Syllabus for Competency Based Training in Intensive Care Medicine in Europe.19 Other institutions are also exploring the creation of certification/competency programs. Minimally, and prior to any decision about establishing formal criteria, institutions could identify designated hospitalists within groups who have particular interest and ability in the critical care setting. These providers, based on models already in place at sites across the United States, could, as an example, be required to spend a minimum of 50% of their clinical time in the ICU and to complete 10 to 20 hours of critical care continuing medical education (CME) per year. One strategy to address this issue and develop clear consensus and guidelines would be to convene the often discussed PrOMIS II working group.

• 3

  Per both the FOCCUS Report and the PrOMIS Report as well as a number of other publications,19 health services research in ICU care should be identified, funded, and implemented. A major focus of this effort should be the evaluation of clinical outcomes for ICU patients cared for by hospitalists. This research is needed for at least 2 reasons:

• As noted, there is little research that has assessed hospitalists' impact on outcomes of ICU patients. Hospitalists are already caring for patients in ICUs across the United States and given the research that has identified the outcomes benefit provided by intensivists, it is important to know objectively if hospitalists have similar levels of performance.

• An increasing number of hospitals and healthcare systems are now committed to achieving the Leapfrog IPS standard‐a challenge for many because of the difficulty with recruiting intensivists. If new research reveals that hospitalists with board certification in Internal Medicine, and more specifically with additional competency training in critical care, also improve outcomes in the ICU then it may be possible for Leapfrog to revise the criteria for meeting the IPS standard.

Summary

As discussed in a number of publications,20 including an article from the Mayo Clinic in the April 2009 edition of Chest entitled, Physicians Staffing Models and Patient Safety in the ICU,21 along with an accompanying editorial, Should Intensive Care Medicine Itself Be on the Critical List,22 creative and realistic solutions are urgently needed to address the crisis in critical care in the U.S. Collaborative efforts between Critical Care Medicine and Hospital Medicine to meet this challenge benefit all involved:

• Intensivists will continue to direct tertiary care units and/or co‐manage patients in tertiary and secondary care centers with Hospitalists.

• Hospitalists will benefit by having the opportunity to secure critical care competency training and by having their appropriate role in the ICU defined.

• All secondary and tertiary care institutions will have a realistic opportunity to meet Leapfrog IPS criteria and therefore benefit from the potential decreased length of stay (LOS), decreased mortality, and improved quality.

• Patients benefit by receiving uniform, evidence‐based, protocol‐driven care.

There is now a need and an opportunity for ACCP, SCCM, ATS, and the American Association of Critical Care Nurses (ACCN), to expand the important work they have already begun through the Critical Care Workforce Partnership. The Partnership should join with the SHM to take the lead in supporting and promoting this collaborative relationship between intensivists and hospitalists: aware that in the final analysis, it is the patients we serve who will benefit the most.

A looming gap in the supply of intensivists prompted the American College of Chest Physicians (ACCP), the American Thoracic Society (ATS), and the Society of Critical Care Medicine (SCCM) to publish a report in 2000 by the Committee on Manpower for Pulmonary and Critical Care Societies (COMPACCS). This study predicted that beginning in 2007 a shortfall would become apparent and steadily increase to 22% by 2020 and to 35% by 2030. Subsequent reports have reiterated those projections, including a report to Congress in 2006 by the U.S. Department of Health and Human Services/Health Resources and Services Administration.14

The concern regarding the shortage of intensivists has been increased by the growing evidence that supports improved critical care outcomesespecially decreased intensive care unit (ICU) and hospital mortalitywith intensivist staffing of ICUs.5, 6 Based on this data and on recommendations from the Society of Critical Care Medicine, the Leapfrog Group made onsite, high‐intensity ICU staffing with intensivists 1 of their 4 leaps.7 A paper by Pronovost et al.8 published in 2001, however, noted that in order for all ICUs in the United States to meet the Leapfrog ICU Physician Staffing (IPS) standard, the number of intensivists would need to increase by a factor of 2.6. Interestingly, a retrospective study published in the Annals of Internal Medicine in June of 2008 by Levy et al.9 suggested that mortality rates may actually be higher in intensivist‐staffed ICUs. An accompanying editorial raised concerns about limitations of the study design, but endorsed Levy's recommendation that more carefully designed, prospective studies were needed; (ie, we still are not certain as to optimal physician staffing for the care of patients requiring the sophisticated treatment available only in an ICU.)10

The health policy challenge, however, remains clear: while there is basic consensus that care of critically ill patients by intensivists improves outcomes, the reality is that the shortage of intensivists in the United States as predicted by the COMPACCS report will only increase, leading some to refer to this as a healthcare crisis. Two major task forces attempted to address this situation, resulting in the publication of the 2004 Framing Options for Critical Care in the United States (FOCCUS) report, The Critical Care Medicine Crisis: A Call for Federal ActionA White Paper from the Critical Care Professional Societies; and the 2007 Prioritizing the Organization and Management of Intensive Care Services in the Unites States (PrOMIS) Conference Report.11, 12 Both reports made specific recommendations including, for example, development of uniform standards for accreditation of institutional critical care capacity, identification and endorsement of core competencies in critical care, investment in health services research, the use of uniform protocols for ICU care, leverage of information technology to promote standardization and improve efficiency, and the development of incentives to attract healthcare professionals to critical care medicine.

A Possible Solution: The Role of Hospitalists

Multiple important efforts are already underway to increase the competency of professionals providing critical care services including the Society of Critical Care's Fundamentals in Critical Care Support (FCCS) program. Additionally, physician assistants and nurse practitioners are playing an increasingly important role as members of critical care services. As another component of this collaborative effort, the PrOMIS Report noted the potential impact of hospitalists in addressing this crisis.

As early as 1999, surveys revealed that as many as 35% of hospitalists were providing critical care services.13 According to the 2005/2006 Society of Hospital Medicine (SHM) National Survey, that number has increased to 75% with a low of 66% in the eastern United States and a high of 84% in the western United States. In community hospitals, 87% of hospitalists care for patients in the ICU, and 30% provide critical care services in academic medical centers.13 While there is some research14, 15 and many anecdotal reports that suggest hospitalists perform well in the ICU, there is, unfortunately, little data addressing outcomes for patients cared for by hospitalists. The results from a prospective, severity‐adjusted study from the Emory University Section of Hospital Medicine and the Division of Pulmonary/Critical Care Medicine examining outcomes for critical care patients cared for by hospitalists with criteria for Pulmonary/Intensivist consults vs. patients cared for by the Pulmonary/Critical Care Medical ICU Service await peer‐review publication.

Despite the lack of outcome data regarding adult hospitalists, it is clear that by default they are already providing a significant proportion of critical care services across the healthcare system, including in tertiary care centers. The two primary models of care include: (1) hospitalists serving as the primary provider without critical care consultant services and (2) comanagement of patients where intensivists and hospitalists collaborate. These collaborative models involve hospitalists actively co‐managing critical care patients along with intensivists or hospitalists managing less critically ill patients with intensivist consultation when indicated. In hospitals lacking intensivists, hospitalists often manage critically ill patients either with intensivist phone consultation, or with the intent to stabilize and transfer. Electronic ICUs are another expanding model of care that provide intensivist support to hospitalists and other primary care providersdecreasing ICU length of stay and severity‐adjusted ICU mortality.16 There are now 40 electronic ICU programs in the United States, and that number continues to grow.

In 2003, there were approximately 10,000 hospitalists in the United States,17 and recent data from an American Hospital Association survey indicates that the number has grown to about 28,000 in 2009. Recent research also documents that hospitalists are soon likely to care for the majority of elderly hospitalized patients in America.18 Aware that the number of intensivists is unlikely to change significantly over the next 25 years the question is no longer if hospitalists should be in the ICU; rather, the question is how to assure quality and improved clinical outcomes through enhanced collaboration between Hospital Medicine and Critical Care Medicine.

Recommendations

There are 3 steps that should be taken urgently to meet this challenge:

• 1

  Per the recommendation of the FOCCUS Report and the PrOMIS Conference Report, uniform protocols for intensive care treatmentmany of which already exist but are not used consistentlyshould be identified and implemented across all ICUs regardless of the level or certification of the provider.

• 2

  Also per the PrOMIS Report, a process for certification of physicians providing critical care services should be established by the appropriate governing bodies, including the Society for Critical Care Medicine, the Society of Hospital Medicine, and the American Thoracic Society, among others. While the PrOMIS Report called for cross‐training of hospital‐based providers to provide intensive care services in lower tier hospitals, a more realistic recommendation given current involvement of hospitalists in the provision of critical care services in secondary and tertiary centers is a competency‐assurance process that includes hospitalists practicing at all levels. This would not be equivalent to board certification, but would be based on a rigorous, comprehensive education and skills training process leading to recognition that would distinguish the recipient as having competencies beyond those obtained in internal medicine residency training. Models for certification could include 4‐month onsite training or a distance learning curriculum with regular blocks of onsite training. Another strategy might be for appropriate governing bodies to establish basic criteria for competency that would then be provided by individual institutions. Emory University, for example, has developed a pilot program incorporating significant components from the European Society for Critical Care Medicine's Syllabus for Competency Based Training in Intensive Care Medicine in Europe.19 Other institutions are also exploring the creation of certification/competency programs. Minimally, and prior to any decision about establishing formal criteria, institutions could identify designated hospitalists within groups who have particular interest and ability in the critical care setting. These providers, based on models already in place at sites across the United States, could, as an example, be required to spend a minimum of 50% of their clinical time in the ICU and to complete 10 to 20 hours of critical care continuing medical education (CME) per year. One strategy to address this issue and develop clear consensus and guidelines would be to convene the often discussed PrOMIS II working group.

• 3

  Per both the FOCCUS Report and the PrOMIS Report as well as a number of other publications,19 health services research in ICU care should be identified, funded, and implemented. A major focus of this effort should be the evaluation of clinical outcomes for ICU patients cared for by hospitalists. This research is needed for at least 2 reasons:

• As noted, there is little research that has assessed hospitalists' impact on outcomes of ICU patients. Hospitalists are already caring for patients in ICUs across the United States and given the research that has identified the outcomes benefit provided by intensivists, it is important to know objectively if hospitalists have similar levels of performance.

• An increasing number of hospitals and healthcare systems are now committed to achieving the Leapfrog IPS standard‐a challenge for many because of the difficulty with recruiting intensivists. If new research reveals that hospitalists with board certification in Internal Medicine, and more specifically with additional competency training in critical care, also improve outcomes in the ICU then it may be possible for Leapfrog to revise the criteria for meeting the IPS standard.

Summary

As discussed in a number of publications,20 including an article from the Mayo Clinic in the April 2009 edition of Chest entitled, Physicians Staffing Models and Patient Safety in the ICU,21 along with an accompanying editorial, Should Intensive Care Medicine Itself Be on the Critical List,22 creative and realistic solutions are urgently needed to address the crisis in critical care in the U.S. Collaborative efforts between Critical Care Medicine and Hospital Medicine to meet this challenge benefit all involved:

• Intensivists will continue to direct tertiary care units and/or co‐manage patients in tertiary and secondary care centers with Hospitalists.

• Hospitalists will benefit by having the opportunity to secure critical care competency training and by having their appropriate role in the ICU defined.

• All secondary and tertiary care institutions will have a realistic opportunity to meet Leapfrog IPS criteria and therefore benefit from the potential decreased length of stay (LOS), decreased mortality, and improved quality.

• Patients benefit by receiving uniform, evidence‐based, protocol‐driven care.

There is now a need and an opportunity for ACCP, SCCM, ATS, and the American Association of Critical Care Nurses (ACCN), to expand the important work they have already begun through the Critical Care Workforce Partnership. The Partnership should join with the SHM to take the lead in supporting and promoting this collaborative relationship between intensivists and hospitalists: aware that in the final analysis, it is the patients we serve who will benefit the most.

A looming gap in the supply of intensivists prompted the American College of Chest Physicians (ACCP), the American Thoracic Society (ATS), and the Society of Critical Care Medicine (SCCM) to publish a report in 2000 by the Committee on Manpower for Pulmonary and Critical Care Societies (COMPACCS). This study predicted that beginning in 2007 a shortfall would become apparent and steadily increase to 22% by 2020 and to 35% by 2030. Subsequent reports have reiterated those projections, including a report to Congress in 2006 by the U.S. Department of Health and Human Services/Health Resources and Services Administration.14

The concern regarding the shortage of intensivists has been increased by the growing evidence that supports improved critical care outcomesespecially decreased intensive care unit (ICU) and hospital mortalitywith intensivist staffing of ICUs.5, 6 Based on this data and on recommendations from the Society of Critical Care Medicine, the Leapfrog Group made onsite, high‐intensity ICU staffing with intensivists 1 of their 4 leaps.7 A paper by Pronovost et al.8 published in 2001, however, noted that in order for all ICUs in the United States to meet the Leapfrog ICU Physician Staffing (IPS) standard, the number of intensivists would need to increase by a factor of 2.6. Interestingly, a retrospective study published in the Annals of Internal Medicine in June of 2008 by Levy et al.9 suggested that mortality rates may actually be higher in intensivist‐staffed ICUs. An accompanying editorial raised concerns about limitations of the study design, but endorsed Levy's recommendation that more carefully designed, prospective studies were needed; (ie, we still are not certain as to optimal physician staffing for the care of patients requiring the sophisticated treatment available only in an ICU.)10

The health policy challenge, however, remains clear: while there is basic consensus that care of critically ill patients by intensivists improves outcomes, the reality is that the shortage of intensivists in the United States as predicted by the COMPACCS report will only increase, leading some to refer to this as a healthcare crisis. Two major task forces attempted to address this situation, resulting in the publication of the 2004 Framing Options for Critical Care in the United States (FOCCUS) report, The Critical Care Medicine Crisis: A Call for Federal ActionA White Paper from the Critical Care Professional Societies; and the 2007 Prioritizing the Organization and Management of Intensive Care Services in the Unites States (PrOMIS) Conference Report.11, 12 Both reports made specific recommendations including, for example, development of uniform standards for accreditation of institutional critical care capacity, identification and endorsement of core competencies in critical care, investment in health services research, the use of uniform protocols for ICU care, leverage of information technology to promote standardization and improve efficiency, and the development of incentives to attract healthcare professionals to critical care medicine.

A Possible Solution: The Role of Hospitalists

Multiple important efforts are already underway to increase the competency of professionals providing critical care services including the Society of Critical Care's Fundamentals in Critical Care Support (FCCS) program. Additionally, physician assistants and nurse practitioners are playing an increasingly important role as members of critical care services. As another component of this collaborative effort, the PrOMIS Report noted the potential impact of hospitalists in addressing this crisis.

As early as 1999, surveys revealed that as many as 35% of hospitalists were providing critical care services.13 According to the 2005/2006 Society of Hospital Medicine (SHM) National Survey, that number has increased to 75% with a low of 66% in the eastern United States and a high of 84% in the western United States. In community hospitals, 87% of hospitalists care for patients in the ICU, and 30% provide critical care services in academic medical centers.13 While there is some research14, 15 and many anecdotal reports that suggest hospitalists perform well in the ICU, there is, unfortunately, little data addressing outcomes for patients cared for by hospitalists. The results from a prospective, severity‐adjusted study from the Emory University Section of Hospital Medicine and the Division of Pulmonary/Critical Care Medicine examining outcomes for critical care patients cared for by hospitalists with criteria for Pulmonary/Intensivist consults vs. patients cared for by the Pulmonary/Critical Care Medical ICU Service await peer‐review publication.

Despite the lack of outcome data regarding adult hospitalists, it is clear that by default they are already providing a significant proportion of critical care services across the healthcare system, including in tertiary care centers. The two primary models of care include: (1) hospitalists serving as the primary provider without critical care consultant services and (2) comanagement of patients where intensivists and hospitalists collaborate. These collaborative models involve hospitalists actively co‐managing critical care patients along with intensivists or hospitalists managing less critically ill patients with intensivist consultation when indicated. In hospitals lacking intensivists, hospitalists often manage critically ill patients either with intensivist phone consultation, or with the intent to stabilize and transfer. Electronic ICUs are another expanding model of care that provide intensivist support to hospitalists and other primary care providersdecreasing ICU length of stay and severity‐adjusted ICU mortality.16 There are now 40 electronic ICU programs in the United States, and that number continues to grow.

In 2003, there were approximately 10,000 hospitalists in the United States,17 and recent data from an American Hospital Association survey indicates that the number has grown to about 28,000 in 2009. Recent research also documents that hospitalists are soon likely to care for the majority of elderly hospitalized patients in America.18 Aware that the number of intensivists is unlikely to change significantly over the next 25 years the question is no longer if hospitalists should be in the ICU; rather, the question is how to assure quality and improved clinical outcomes through enhanced collaboration between Hospital Medicine and Critical Care Medicine.

Recommendations

There are 3 steps that should be taken urgently to meet this challenge:

• 1

  Per the recommendation of the FOCCUS Report and the PrOMIS Conference Report, uniform protocols for intensive care treatmentmany of which already exist but are not used consistentlyshould be identified and implemented across all ICUs regardless of the level or certification of the provider.

• 2

  Also per the PrOMIS Report, a process for certification of physicians providing critical care services should be established by the appropriate governing bodies, including the Society for Critical Care Medicine, the Society of Hospital Medicine, and the American Thoracic Society, among others. While the PrOMIS Report called for cross‐training of hospital‐based providers to provide intensive care services in lower tier hospitals, a more realistic recommendation given current involvement of hospitalists in the provision of critical care services in secondary and tertiary centers is a competency‐assurance process that includes hospitalists practicing at all levels. This would not be equivalent to board certification, but would be based on a rigorous, comprehensive education and skills training process leading to recognition that would distinguish the recipient as having competencies beyond those obtained in internal medicine residency training. Models for certification could include 4‐month onsite training or a distance learning curriculum with regular blocks of onsite training. Another strategy might be for appropriate governing bodies to establish basic criteria for competency that would then be provided by individual institutions. Emory University, for example, has developed a pilot program incorporating significant components from the European Society for Critical Care Medicine's Syllabus for Competency Based Training in Intensive Care Medicine in Europe.19 Other institutions are also exploring the creation of certification/competency programs. Minimally, and prior to any decision about establishing formal criteria, institutions could identify designated hospitalists within groups who have particular interest and ability in the critical care setting. These providers, based on models already in place at sites across the United States, could, as an example, be required to spend a minimum of 50% of their clinical time in the ICU and to complete 10 to 20 hours of critical care continuing medical education (CME) per year. One strategy to address this issue and develop clear consensus and guidelines would be to convene the often discussed PrOMIS II working group.

• 3

  Per both the FOCCUS Report and the PrOMIS Report as well as a number of other publications,19 health services research in ICU care should be identified, funded, and implemented. A major focus of this effort should be the evaluation of clinical outcomes for ICU patients cared for by hospitalists. This research is needed for at least 2 reasons:

• As noted, there is little research that has assessed hospitalists' impact on outcomes of ICU patients. Hospitalists are already caring for patients in ICUs across the United States and given the research that has identified the outcomes benefit provided by intensivists, it is important to know objectively if hospitalists have similar levels of performance.

• An increasing number of hospitals and healthcare systems are now committed to achieving the Leapfrog IPS standard‐a challenge for many because of the difficulty with recruiting intensivists. If new research reveals that hospitalists with board certification in Internal Medicine, and more specifically with additional competency training in critical care, also improve outcomes in the ICU then it may be possible for Leapfrog to revise the criteria for meeting the IPS standard.

Summary

As discussed in a number of publications,20 including an article from the Mayo Clinic in the April 2009 edition of Chest entitled, Physicians Staffing Models and Patient Safety in the ICU,21 along with an accompanying editorial, Should Intensive Care Medicine Itself Be on the Critical List,22 creative and realistic solutions are urgently needed to address the crisis in critical care in the U.S. Collaborative efforts between Critical Care Medicine and Hospital Medicine to meet this challenge benefit all involved:

• Intensivists will continue to direct tertiary care units and/or co‐manage patients in tertiary and secondary care centers with Hospitalists.

• Hospitalists will benefit by having the opportunity to secure critical care competency training and by having their appropriate role in the ICU defined.

• All secondary and tertiary care institutions will have a realistic opportunity to meet Leapfrog IPS criteria and therefore benefit from the potential decreased length of stay (LOS), decreased mortality, and improved quality.

• Patients benefit by receiving uniform, evidence‐based, protocol‐driven care.

There is now a need and an opportunity for ACCP, SCCM, ATS, and the American Association of Critical Care Nurses (ACCN), to expand the important work they have already begun through the Critical Care Workforce Partnership. The Partnership should join with the SHM to take the lead in supporting and promoting this collaborative relationship between intensivists and hospitalists: aware that in the final analysis, it is the patients we serve who will benefit the most.

An 83‐year‐old female reported increased swelling of her legs over the past 2 years. She noted temperature intolerance, low energy, and constipation, but denied any hair loss or nail changes. On exam, she had marked bilateral lower extremity edema that was predominately nonpitting. Overlying the edema there were thickened, well‐defined plaques with a peau d'orange appearance surrounded by brown, thin plaques on the pretibial areas sparing the dorsum of the feet (Figure 1). A punch biopsy was obtained and demonstrated increased deposition of mucin throughout the dermis along with fragmentation and increased numbers of elastic fibers consistent with a diagnosis of pretibial myxedema. Measured thyrotropin level was elevated at 3.9 U/mL (normal, 0.3‐3.8 U/mL), consistent with hypothyroidism. This is a severe example of pretibial myxedema, or infiltrative dermopathy, which can occur, more commonly, in the setting of Graves' disease or, in rare circumstances, hypothyroidism.1 Myxedema results from the accumulation of hyaluronic acid and chondroitin sulfate in the dermis.2 Treatment is difficult and includes topical glucocorticoids under occlusion and, if indicated, thyroid corrective therapy.3

Figure 1

An 83‐year‐old female reported increased swelling of her legs over the past 2 years. She noted temperature intolerance, low energy, and constipation, but denied any hair loss or nail changes. On exam, she had marked bilateral lower extremity edema that was predominately nonpitting. Overlying the edema there were thickened, well‐defined plaques with a peau d'orange appearance surrounded by brown, thin plaques on the pretibial areas sparing the dorsum of the feet (Figure 1). A punch biopsy was obtained and demonstrated increased deposition of mucin throughout the dermis along with fragmentation and increased numbers of elastic fibers consistent with a diagnosis of pretibial myxedema. Measured thyrotropin level was elevated at 3.9 U/mL (normal, 0.3‐3.8 U/mL), consistent with hypothyroidism. This is a severe example of pretibial myxedema, or infiltrative dermopathy, which can occur, more commonly, in the setting of Graves' disease or, in rare circumstances, hypothyroidism.1 Myxedema results from the accumulation of hyaluronic acid and chondroitin sulfate in the dermis.2 Treatment is difficult and includes topical glucocorticoids under occlusion and, if indicated, thyroid corrective therapy.3

Figure 1

An 83‐year‐old female reported increased swelling of her legs over the past 2 years. She noted temperature intolerance, low energy, and constipation, but denied any hair loss or nail changes. On exam, she had marked bilateral lower extremity edema that was predominately nonpitting. Overlying the edema there were thickened, well‐defined plaques with a peau d'orange appearance surrounded by brown, thin plaques on the pretibial areas sparing the dorsum of the feet (Figure 1). A punch biopsy was obtained and demonstrated increased deposition of mucin throughout the dermis along with fragmentation and increased numbers of elastic fibers consistent with a diagnosis of pretibial myxedema. Measured thyrotropin level was elevated at 3.9 U/mL (normal, 0.3‐3.8 U/mL), consistent with hypothyroidism. This is a severe example of pretibial myxedema, or infiltrative dermopathy, which can occur, more commonly, in the setting of Graves' disease or, in rare circumstances, hypothyroidism.1 Myxedema results from the accumulation of hyaluronic acid and chondroitin sulfate in the dermis.2 Treatment is difficult and includes topical glucocorticoids under occlusion and, if indicated, thyroid corrective therapy.3

Figure 1

Severe sepsis is responsible for significant morbidity and mortality. In the United States, approximately 750,000 cases occur each year with an estimated mortality of 30% to 50%.1 Early goal‐directed therapy has been shown to decrease mortality in patients with severe sepsis and septic shock.2, 3 As a result, efforts have been focused toward providing early and aggressive intervention once sepsis has been established. In many cases this has been accomplished through the implementation of a protocol with guidelines for fluid management, antibiotic and vasopressor administration, and other interventions.410 Prior studies have demonstrated that care of hospitalized patients before intensive care unit (ICU) admission is often suboptimal,1113 and have suggested that patients with clear indicators of acute deterioration may go unrecognized on the ward. We previously reported the effects of implementing a hospital‐wide protocol for the management of severe sepsis,14 finding that although there was a significant reduction in overall mortality there was no difference for patients who developed severe sepsis on the hospital ward. This finding also suggests that the initial care of patients with severe sepsis on hospital wards may differ in intensity compared to emergency departments and ICUs. Failure on the part of the clinician to recognize the harbingers of impending sepsis before the onset of organ dysfunction or hypotension may contribute to a delay in aggressive therapy.

Previous efforts at early recognition of sepsis have relied on diagnostic studies or specific biomarkers to screen at‐risk patients. These have included such studies as messenger RNA (mRNA) expression,15 C‐reactive protein,16 procalcitonin in newborns,17 immunocompetence measures in burn patients,18 protein C concentration in neutropenic patients,19 and several immune markers (eg, tumor necrosis factor‐alpha, interleukin [IL]‐1 beta, IL‐6, IL‐8, and IL‐10).20 However, these biomarkers have been studied only in specific patient populations, require suspicion on the part of the clinician and the measurement of diagnostic or laboratory values that would otherwise not have been obtained. The ideal tool for predicting the onset of sepsis would be applicable to a broad patient population, not require specific suspicion on the part of the clinician, and use only routinely obtained clinical measurements and laboratory values.

Prediction models and scoring systems that use routine hemodynamic and laboratory values for several endpoints related to sepsis and septic shock have been developed. Many such tools are used to define severity of illness and predict outcome, while others have been developed to predict such events as bacteremia in patients presenting with fever,21 the probability of infection in the critically ill,22 and end‐organ dysfunction in severe sepsis.23 Little work has been done to develop such a model capable of predicting the onset of sepsis,24 and there have been no attempts to deploy a model as a large‐scale screening tool.

Our objective was to develop a simple algorithm that can be used in an automated fashion to screen hospitalized patients for impending septic shock. Such a model would be derived from routine hemodynamic and laboratory values, and take advantage of a computerized medical record system for data collection.

Patients and Methods

Results

Predictors of Sepsis

RPART analysis of the 2005 patient data set demonstrated that the most significant predictors of sepsis in the 24 hours preceding transfer to the medical ICU were the partial pressure of arterial oxygen (PaO2), systolic blood pressure, absolute neutrophil count, blood urea nitrogen (BUN), pH, bicarbonate, chloride, and albumin. This resulted in a simple algorithm with nine classification splits (Figure 1), which was then prospectively applied to the 2006 and 2007 patient data sets. These results are summarized in Table 2.

Figure 1

Table 2.Results of 2005‐Derived Model Using All Clinical Data Applied to the 2005, 2006, and 2007 Patient Data Sets

	Total Number	Number Correctly Classified (%)	Case Identification Time Before ICU Admission (minutes)	PPV (%)	NPV (%)	MCR (%)
NOTE: Values shown represent mean SD. Abbreviations: ICU, intensive care unit; MCR, misclassification rate; NPV, negative predictive value; PPV, positive predictive value.
2005				27.9	98.1	7.8
Cases	562	320 (56.9)
Controls	13,223	12,394 (93.7)
2006			179 230	28.7	97.7	8.4
Cases	635	347 (54.7)
Controls	13,102	12,241 (93.4)
2007			192 210	28.3	97.6	8.8
Cases	667	367 (55.0)
Controls	13,270	12,341 (93.0)

The resulting classification model had a low total misclassification rate for the 2005 data. Of the 562 septic patients, 320 (56.9%) were correctly classified, and 12,394 (93.7%) of the control patients were appropriately identified. The number of septic and control patients misclassified was 242 and 829, respectively, yielding a total misclassification rate of 7.8%. When applied to the 2006 patient data set, 347 (54.7%) of the 635 septic shock patients were correctly identified, while 12,241 (93.4%) of the 13,102 control patients were correctly classified. The total misclassification rate for the 2006 patient set was 8.4%. For the 2007 patient data, 367 (55.0%) of the 667 case patients were correctly identified, and 12,341 (93.0%) of the 13,270 control patients were correctly identified. This resulted in a total misclassification rate of 8.8%.

The 2006 and 2007 case patients were identified 179 230 minutes and 192 210 minutes before ICU transfer, respectively (Figure 2). The algorithm demonstrated positive and negative predictive values of 28.7% and 97.7% for the 2006 patient set, respectively, and 28.3% and 97.6% for the 2007 patient set, respectively.

Figure 2

Although the prediction algorithm shown in Figure 1 identified the majority of the case patients with ample time for clinical intervention prior to ICU transfer, the analysis used to derive this model included values for the arterial blood gas (ABG). As this is not a routinely obtained study for hospitalized patients outside of an ICU, it is possible that the performance of this model can in part be attributed to clinical acumen rather than changes in patient physiology. The ABG would likely only be obtained in patients with a more concerning or deteriorating clinical course, and thus more likely to develop shock. To address this possibility, a second analysis was performed that did not include the values for the ABG. The result was an algorithm with 13 classification splits, as shown in Figure 3.

Figure 3

The most predictive clinical variables in this analysis included the shock index (heart rate divided by systolic blood pressure), mean arterial pressure, total bilirubin, international normalized ratio (INR), total white blood cell count, absolute neutrophil count, albumin, hemoglobin, and sodium. This model was again applied to the 2006 and 2007 patient data sets (Table 3).

Table 3.Results of 2005‐Derived Model, Excluding Arterial Blood Gas Data, Applied to the 2005, 2006, and 2007 Patient Data Sets

	Total Number	Number Correctly Classified (%)	Case Identification Time Before ICU Admission (minutes)	PPV (%)	NPV (%)	MCR (%)
NOTE: Values shown represent mean SD. Abbreviations: MCR, misclassification rate; NPV, negative predictive value; PPV, positive predictive value.
2005				20.5	96.7	6.7
Cases	562	126 (22.4)
Controls	13,223	12,735 (96.3)
2006			508 536	21.4	96.1	7.0
Cases	635	121 (19.1)
Controls	13,102	12,657 (96.6)
2007			496 512	19.5	95.8	7.1
Cases	667	102 (15.3)
Controls	13,270	12,850 (96.8)

The overall misclassification rates for 2006 and 2007 were 7.0% and 7.1%, respectively. The model correctly identified 121 (19.1%) of the 635 cases and 12,657 (96.6%) of the 13,102 control patients from 2006, and 102 (15.3%) of the 667 cases and 12,850 (96.8%) of the 13,270 control patients from 2007. The respective positive and negative predictive values were 21.4% and 96.1% for 2006, respectively, and 19.5% and 95.8% for 2007, respectively.

Although the overall performance of the model derived without the ABG data was not as good, the identification times prior to ICU transfer were significantly improved. For the 2006 data, patients were identified 508 536 minutes before transfer (Figure 4), compared to 179 230 minutes for the model that included the ABG data (P < 0.01). For the 2007 data, patients were identified 496 512 minutes prior to ICU admission (Figure 4), compared to 192 210 minutes for the previous model (P < 0.01).

Figure 4

Discussion

We have demonstrated a simple method for generating an algorithm derived from routine laboratory and hemodynamic values that is capable of predicting the onset of sepsis in a significant proportion of non‐ICU patients. Two prediction models were generated, 1 with and 1 without ABG data included in the analysis. In the 2006 and 2007 validation cohorts, the model including these data correctly classified 54.7% and 55.0% of the patients who developed septic shock and 93.4% and 93.0% of control patients, respectively. The second model identified 19.1% and 15.3% of the septic shock patients and 96.6% and 96.8% of the control patients for 2006 and 2007, respectively. The methods used in generating this model are relatively simple and can be executed with the use of an electronic medical record system.

Early, goal‐directed cardiovascular resuscitation and adequate initial antibiotic therapy have been shown to decrease mortality in patients with severe sepsis and septic shock.2, 26 Prior studies employing early, targeted resuscitation strategies have demonstrated decreased use of vasopressors10 and decreased mortality.510 In addition, we previously demonstrated that a standardized order set for the management of severe sepsis in the emergency department that focused on early and aggressive intervention was associated with decreased 28‐day mortality.1 These studies suggest that early, aggressive management of septic shock can improve outcomes. Identification of patients prior to overt clinical deterioration may allow for early intervention aimed at preventing shock or improving its outcome.

The purpose of this method is to develop a model capable of recognizing patterns in clinical data that herald a patient's otherwise unidentified clinical deterioration. It is not intended to replace existing outcome prediction tools or severity of illness scoring systems, where a high degree of accuracy would be required. Rather, it would be best implemented as an automated screening tool incorporated into an electronic medical record system. When a hospitalized patient is identified as a possible septic shock patient by the classification tree, a notification is then issued to the clinicians caring for the patient. The primary goal of this method is to notify clinicians of potential clinical deterioration. Any action taken as a result of this notification is at the discretion of the clinician. This method could be employed for any population of hospitalized patients, though because of variations in clinical practice and patient physiology, different models would need to be generated for differing patient populations.

This method has limitations, the foremost of which is the possible instability of the resulting classification model. This type of analysis results in an algorithm that depends on binary splits to classify patients. In generating the algorithm, the recursive partitioning analysis selects the variables and cutoff values that result in the strongest decision tree with the most pure classifications at the end nodes. These variables and cutoff values may not immediately seem logical from a clinical standpoint, and may vary with changes in practice and even possibly between divisions within a hospital. As a result, the algorithm would likely require intermittent updating to remain effective and a model derived from 1 hospital or patient population would not necessarily be applicable to patients at another institution or from a different population. However, once the method has been developed at an institution, the process of revising the algorithm could be essentially automated and uses few resources.

Another shortcoming of this method is the relatively low sensitivity of the resulting algorithm. In a role as an automated alert system, a low false‐positive rate is particularly desirable to avoid unnecessary frequent distraction of clinicians. The sensitivity of the model can be improved through manipulation of how the analysis is performed, but this would be at the expense of a higher false‐positive rate, which is not acceptable. Finally, prior studies examining treatment for sepsis have demonstrated an advantage to early and aggressive therapy. It is not clear, however, if identifying these patients prior to the onset of clinically evident sepsis would result in improved outcomes. Further work is required to determine if this is the case. We are currently conducting a prospective study that employs the method described here in conjunction with an automated alert system to ascertain if it impacts outcomes on patients admitted to the medicine wards of Barnes‐Jewish Hospital.

In conclusion, the method presented here represents a technique that consumes few resources and is capable of identifying some patients before septic shock becomes clinically evident. When applied in an automated fashion with the capability to alert clinicians caring for a patient, the method demonstrated here may allow for earlier diagnosis and possibly intervention for septic shock patients.

Severe sepsis is responsible for significant morbidity and mortality. In the United States, approximately 750,000 cases occur each year with an estimated mortality of 30% to 50%.1 Early goal‐directed therapy has been shown to decrease mortality in patients with severe sepsis and septic shock.2, 3 As a result, efforts have been focused toward providing early and aggressive intervention once sepsis has been established. In many cases this has been accomplished through the implementation of a protocol with guidelines for fluid management, antibiotic and vasopressor administration, and other interventions.410 Prior studies have demonstrated that care of hospitalized patients before intensive care unit (ICU) admission is often suboptimal,1113 and have suggested that patients with clear indicators of acute deterioration may go unrecognized on the ward. We previously reported the effects of implementing a hospital‐wide protocol for the management of severe sepsis,14 finding that although there was a significant reduction in overall mortality there was no difference for patients who developed severe sepsis on the hospital ward. This finding also suggests that the initial care of patients with severe sepsis on hospital wards may differ in intensity compared to emergency departments and ICUs. Failure on the part of the clinician to recognize the harbingers of impending sepsis before the onset of organ dysfunction or hypotension may contribute to a delay in aggressive therapy.

Previous efforts at early recognition of sepsis have relied on diagnostic studies or specific biomarkers to screen at‐risk patients. These have included such studies as messenger RNA (mRNA) expression,15 C‐reactive protein,16 procalcitonin in newborns,17 immunocompetence measures in burn patients,18 protein C concentration in neutropenic patients,19 and several immune markers (eg, tumor necrosis factor‐alpha, interleukin [IL]‐1 beta, IL‐6, IL‐8, and IL‐10).20 However, these biomarkers have been studied only in specific patient populations, require suspicion on the part of the clinician and the measurement of diagnostic or laboratory values that would otherwise not have been obtained. The ideal tool for predicting the onset of sepsis would be applicable to a broad patient population, not require specific suspicion on the part of the clinician, and use only routinely obtained clinical measurements and laboratory values.

Prediction models and scoring systems that use routine hemodynamic and laboratory values for several endpoints related to sepsis and septic shock have been developed. Many such tools are used to define severity of illness and predict outcome, while others have been developed to predict such events as bacteremia in patients presenting with fever,21 the probability of infection in the critically ill,22 and end‐organ dysfunction in severe sepsis.23 Little work has been done to develop such a model capable of predicting the onset of sepsis,24 and there have been no attempts to deploy a model as a large‐scale screening tool.

Our objective was to develop a simple algorithm that can be used in an automated fashion to screen hospitalized patients for impending septic shock. Such a model would be derived from routine hemodynamic and laboratory values, and take advantage of a computerized medical record system for data collection.

Patients and Methods

Results

Predictors of Sepsis

RPART analysis of the 2005 patient data set demonstrated that the most significant predictors of sepsis in the 24 hours preceding transfer to the medical ICU were the partial pressure of arterial oxygen (PaO2), systolic blood pressure, absolute neutrophil count, blood urea nitrogen (BUN), pH, bicarbonate, chloride, and albumin. This resulted in a simple algorithm with nine classification splits (Figure 1), which was then prospectively applied to the 2006 and 2007 patient data sets. These results are summarized in Table 2.

Figure 1

Table 2.Results of 2005‐Derived Model Using All Clinical Data Applied to the 2005, 2006, and 2007 Patient Data Sets

	Total Number	Number Correctly Classified (%)	Case Identification Time Before ICU Admission (minutes)	PPV (%)	NPV (%)	MCR (%)
NOTE: Values shown represent mean SD. Abbreviations: ICU, intensive care unit; MCR, misclassification rate; NPV, negative predictive value; PPV, positive predictive value.
2005				27.9	98.1	7.8
Cases	562	320 (56.9)
Controls	13,223	12,394 (93.7)
2006			179 230	28.7	97.7	8.4
Cases	635	347 (54.7)
Controls	13,102	12,241 (93.4)
2007			192 210	28.3	97.6	8.8
Cases	667	367 (55.0)
Controls	13,270	12,341 (93.0)

The resulting classification model had a low total misclassification rate for the 2005 data. Of the 562 septic patients, 320 (56.9%) were correctly classified, and 12,394 (93.7%) of the control patients were appropriately identified. The number of septic and control patients misclassified was 242 and 829, respectively, yielding a total misclassification rate of 7.8%. When applied to the 2006 patient data set, 347 (54.7%) of the 635 septic shock patients were correctly identified, while 12,241 (93.4%) of the 13,102 control patients were correctly classified. The total misclassification rate for the 2006 patient set was 8.4%. For the 2007 patient data, 367 (55.0%) of the 667 case patients were correctly identified, and 12,341 (93.0%) of the 13,270 control patients were correctly identified. This resulted in a total misclassification rate of 8.8%.

The 2006 and 2007 case patients were identified 179 230 minutes and 192 210 minutes before ICU transfer, respectively (Figure 2). The algorithm demonstrated positive and negative predictive values of 28.7% and 97.7% for the 2006 patient set, respectively, and 28.3% and 97.6% for the 2007 patient set, respectively.

Figure 2

Although the prediction algorithm shown in Figure 1 identified the majority of the case patients with ample time for clinical intervention prior to ICU transfer, the analysis used to derive this model included values for the arterial blood gas (ABG). As this is not a routinely obtained study for hospitalized patients outside of an ICU, it is possible that the performance of this model can in part be attributed to clinical acumen rather than changes in patient physiology. The ABG would likely only be obtained in patients with a more concerning or deteriorating clinical course, and thus more likely to develop shock. To address this possibility, a second analysis was performed that did not include the values for the ABG. The result was an algorithm with 13 classification splits, as shown in Figure 3.

Figure 3

The most predictive clinical variables in this analysis included the shock index (heart rate divided by systolic blood pressure), mean arterial pressure, total bilirubin, international normalized ratio (INR), total white blood cell count, absolute neutrophil count, albumin, hemoglobin, and sodium. This model was again applied to the 2006 and 2007 patient data sets (Table 3).

Table 3.Results of 2005‐Derived Model, Excluding Arterial Blood Gas Data, Applied to the 2005, 2006, and 2007 Patient Data Sets

	Total Number	Number Correctly Classified (%)	Case Identification Time Before ICU Admission (minutes)	PPV (%)	NPV (%)	MCR (%)
NOTE: Values shown represent mean SD. Abbreviations: MCR, misclassification rate; NPV, negative predictive value; PPV, positive predictive value.
2005				20.5	96.7	6.7
Cases	562	126 (22.4)
Controls	13,223	12,735 (96.3)
2006			508 536	21.4	96.1	7.0
Cases	635	121 (19.1)
Controls	13,102	12,657 (96.6)
2007			496 512	19.5	95.8	7.1
Cases	667	102 (15.3)
Controls	13,270	12,850 (96.8)

The overall misclassification rates for 2006 and 2007 were 7.0% and 7.1%, respectively. The model correctly identified 121 (19.1%) of the 635 cases and 12,657 (96.6%) of the 13,102 control patients from 2006, and 102 (15.3%) of the 667 cases and 12,850 (96.8%) of the 13,270 control patients from 2007. The respective positive and negative predictive values were 21.4% and 96.1% for 2006, respectively, and 19.5% and 95.8% for 2007, respectively.

Although the overall performance of the model derived without the ABG data was not as good, the identification times prior to ICU transfer were significantly improved. For the 2006 data, patients were identified 508 536 minutes before transfer (Figure 4), compared to 179 230 minutes for the model that included the ABG data (P < 0.01). For the 2007 data, patients were identified 496 512 minutes prior to ICU admission (Figure 4), compared to 192 210 minutes for the previous model (P < 0.01).

Figure 4

Discussion

We have demonstrated a simple method for generating an algorithm derived from routine laboratory and hemodynamic values that is capable of predicting the onset of sepsis in a significant proportion of non‐ICU patients. Two prediction models were generated, 1 with and 1 without ABG data included in the analysis. In the 2006 and 2007 validation cohorts, the model including these data correctly classified 54.7% and 55.0% of the patients who developed septic shock and 93.4% and 93.0% of control patients, respectively. The second model identified 19.1% and 15.3% of the septic shock patients and 96.6% and 96.8% of the control patients for 2006 and 2007, respectively. The methods used in generating this model are relatively simple and can be executed with the use of an electronic medical record system.

Early, goal‐directed cardiovascular resuscitation and adequate initial antibiotic therapy have been shown to decrease mortality in patients with severe sepsis and septic shock.2, 26 Prior studies employing early, targeted resuscitation strategies have demonstrated decreased use of vasopressors10 and decreased mortality.510 In addition, we previously demonstrated that a standardized order set for the management of severe sepsis in the emergency department that focused on early and aggressive intervention was associated with decreased 28‐day mortality.1 These studies suggest that early, aggressive management of septic shock can improve outcomes. Identification of patients prior to overt clinical deterioration may allow for early intervention aimed at preventing shock or improving its outcome.

The purpose of this method is to develop a model capable of recognizing patterns in clinical data that herald a patient's otherwise unidentified clinical deterioration. It is not intended to replace existing outcome prediction tools or severity of illness scoring systems, where a high degree of accuracy would be required. Rather, it would be best implemented as an automated screening tool incorporated into an electronic medical record system. When a hospitalized patient is identified as a possible septic shock patient by the classification tree, a notification is then issued to the clinicians caring for the patient. The primary goal of this method is to notify clinicians of potential clinical deterioration. Any action taken as a result of this notification is at the discretion of the clinician. This method could be employed for any population of hospitalized patients, though because of variations in clinical practice and patient physiology, different models would need to be generated for differing patient populations.

This method has limitations, the foremost of which is the possible instability of the resulting classification model. This type of analysis results in an algorithm that depends on binary splits to classify patients. In generating the algorithm, the recursive partitioning analysis selects the variables and cutoff values that result in the strongest decision tree with the most pure classifications at the end nodes. These variables and cutoff values may not immediately seem logical from a clinical standpoint, and may vary with changes in practice and even possibly between divisions within a hospital. As a result, the algorithm would likely require intermittent updating to remain effective and a model derived from 1 hospital or patient population would not necessarily be applicable to patients at another institution or from a different population. However, once the method has been developed at an institution, the process of revising the algorithm could be essentially automated and uses few resources.

Another shortcoming of this method is the relatively low sensitivity of the resulting algorithm. In a role as an automated alert system, a low false‐positive rate is particularly desirable to avoid unnecessary frequent distraction of clinicians. The sensitivity of the model can be improved through manipulation of how the analysis is performed, but this would be at the expense of a higher false‐positive rate, which is not acceptable. Finally, prior studies examining treatment for sepsis have demonstrated an advantage to early and aggressive therapy. It is not clear, however, if identifying these patients prior to the onset of clinically evident sepsis would result in improved outcomes. Further work is required to determine if this is the case. We are currently conducting a prospective study that employs the method described here in conjunction with an automated alert system to ascertain if it impacts outcomes on patients admitted to the medicine wards of Barnes‐Jewish Hospital.

In conclusion, the method presented here represents a technique that consumes few resources and is capable of identifying some patients before septic shock becomes clinically evident. When applied in an automated fashion with the capability to alert clinicians caring for a patient, the method demonstrated here may allow for earlier diagnosis and possibly intervention for septic shock patients.

Severe sepsis is responsible for significant morbidity and mortality. In the United States, approximately 750,000 cases occur each year with an estimated mortality of 30% to 50%.1 Early goal‐directed therapy has been shown to decrease mortality in patients with severe sepsis and septic shock.2, 3 As a result, efforts have been focused toward providing early and aggressive intervention once sepsis has been established. In many cases this has been accomplished through the implementation of a protocol with guidelines for fluid management, antibiotic and vasopressor administration, and other interventions.410 Prior studies have demonstrated that care of hospitalized patients before intensive care unit (ICU) admission is often suboptimal,1113 and have suggested that patients with clear indicators of acute deterioration may go unrecognized on the ward. We previously reported the effects of implementing a hospital‐wide protocol for the management of severe sepsis,14 finding that although there was a significant reduction in overall mortality there was no difference for patients who developed severe sepsis on the hospital ward. This finding also suggests that the initial care of patients with severe sepsis on hospital wards may differ in intensity compared to emergency departments and ICUs. Failure on the part of the clinician to recognize the harbingers of impending sepsis before the onset of organ dysfunction or hypotension may contribute to a delay in aggressive therapy.

Previous efforts at early recognition of sepsis have relied on diagnostic studies or specific biomarkers to screen at‐risk patients. These have included such studies as messenger RNA (mRNA) expression,15 C‐reactive protein,16 procalcitonin in newborns,17 immunocompetence measures in burn patients,18 protein C concentration in neutropenic patients,19 and several immune markers (eg, tumor necrosis factor‐alpha, interleukin [IL]‐1 beta, IL‐6, IL‐8, and IL‐10).20 However, these biomarkers have been studied only in specific patient populations, require suspicion on the part of the clinician and the measurement of diagnostic or laboratory values that would otherwise not have been obtained. The ideal tool for predicting the onset of sepsis would be applicable to a broad patient population, not require specific suspicion on the part of the clinician, and use only routinely obtained clinical measurements and laboratory values.

Prediction models and scoring systems that use routine hemodynamic and laboratory values for several endpoints related to sepsis and septic shock have been developed. Many such tools are used to define severity of illness and predict outcome, while others have been developed to predict such events as bacteremia in patients presenting with fever,21 the probability of infection in the critically ill,22 and end‐organ dysfunction in severe sepsis.23 Little work has been done to develop such a model capable of predicting the onset of sepsis,24 and there have been no attempts to deploy a model as a large‐scale screening tool.

Our objective was to develop a simple algorithm that can be used in an automated fashion to screen hospitalized patients for impending septic shock. Such a model would be derived from routine hemodynamic and laboratory values, and take advantage of a computerized medical record system for data collection.

Patients and Methods

Results

Predictors of Sepsis

RPART analysis of the 2005 patient data set demonstrated that the most significant predictors of sepsis in the 24 hours preceding transfer to the medical ICU were the partial pressure of arterial oxygen (PaO2), systolic blood pressure, absolute neutrophil count, blood urea nitrogen (BUN), pH, bicarbonate, chloride, and albumin. This resulted in a simple algorithm with nine classification splits (Figure 1), which was then prospectively applied to the 2006 and 2007 patient data sets. These results are summarized in Table 2.

Figure 1

Table 2.Results of 2005‐Derived Model Using All Clinical Data Applied to the 2005, 2006, and 2007 Patient Data Sets

	Total Number	Number Correctly Classified (%)	Case Identification Time Before ICU Admission (minutes)	PPV (%)	NPV (%)	MCR (%)
NOTE: Values shown represent mean SD. Abbreviations: ICU, intensive care unit; MCR, misclassification rate; NPV, negative predictive value; PPV, positive predictive value.
2005				27.9	98.1	7.8
Cases	562	320 (56.9)
Controls	13,223	12,394 (93.7)
2006			179 230	28.7	97.7	8.4
Cases	635	347 (54.7)
Controls	13,102	12,241 (93.4)
2007			192 210	28.3	97.6	8.8
Cases	667	367 (55.0)
Controls	13,270	12,341 (93.0)

The resulting classification model had a low total misclassification rate for the 2005 data. Of the 562 septic patients, 320 (56.9%) were correctly classified, and 12,394 (93.7%) of the control patients were appropriately identified. The number of septic and control patients misclassified was 242 and 829, respectively, yielding a total misclassification rate of 7.8%. When applied to the 2006 patient data set, 347 (54.7%) of the 635 septic shock patients were correctly identified, while 12,241 (93.4%) of the 13,102 control patients were correctly classified. The total misclassification rate for the 2006 patient set was 8.4%. For the 2007 patient data, 367 (55.0%) of the 667 case patients were correctly identified, and 12,341 (93.0%) of the 13,270 control patients were correctly identified. This resulted in a total misclassification rate of 8.8%.

The 2006 and 2007 case patients were identified 179 230 minutes and 192 210 minutes before ICU transfer, respectively (Figure 2). The algorithm demonstrated positive and negative predictive values of 28.7% and 97.7% for the 2006 patient set, respectively, and 28.3% and 97.6% for the 2007 patient set, respectively.

Figure 2

Although the prediction algorithm shown in Figure 1 identified the majority of the case patients with ample time for clinical intervention prior to ICU transfer, the analysis used to derive this model included values for the arterial blood gas (ABG). As this is not a routinely obtained study for hospitalized patients outside of an ICU, it is possible that the performance of this model can in part be attributed to clinical acumen rather than changes in patient physiology. The ABG would likely only be obtained in patients with a more concerning or deteriorating clinical course, and thus more likely to develop shock. To address this possibility, a second analysis was performed that did not include the values for the ABG. The result was an algorithm with 13 classification splits, as shown in Figure 3.

Figure 3

The most predictive clinical variables in this analysis included the shock index (heart rate divided by systolic blood pressure), mean arterial pressure, total bilirubin, international normalized ratio (INR), total white blood cell count, absolute neutrophil count, albumin, hemoglobin, and sodium. This model was again applied to the 2006 and 2007 patient data sets (Table 3).

Table 3.Results of 2005‐Derived Model, Excluding Arterial Blood Gas Data, Applied to the 2005, 2006, and 2007 Patient Data Sets

	Total Number	Number Correctly Classified (%)	Case Identification Time Before ICU Admission (minutes)	PPV (%)	NPV (%)	MCR (%)
NOTE: Values shown represent mean SD. Abbreviations: MCR, misclassification rate; NPV, negative predictive value; PPV, positive predictive value.
2005				20.5	96.7	6.7
Cases	562	126 (22.4)
Controls	13,223	12,735 (96.3)
2006			508 536	21.4	96.1	7.0
Cases	635	121 (19.1)
Controls	13,102	12,657 (96.6)
2007			496 512	19.5	95.8	7.1
Cases	667	102 (15.3)
Controls	13,270	12,850 (96.8)

The overall misclassification rates for 2006 and 2007 were 7.0% and 7.1%, respectively. The model correctly identified 121 (19.1%) of the 635 cases and 12,657 (96.6%) of the 13,102 control patients from 2006, and 102 (15.3%) of the 667 cases and 12,850 (96.8%) of the 13,270 control patients from 2007. The respective positive and negative predictive values were 21.4% and 96.1% for 2006, respectively, and 19.5% and 95.8% for 2007, respectively.

Although the overall performance of the model derived without the ABG data was not as good, the identification times prior to ICU transfer were significantly improved. For the 2006 data, patients were identified 508 536 minutes before transfer (Figure 4), compared to 179 230 minutes for the model that included the ABG data (P < 0.01). For the 2007 data, patients were identified 496 512 minutes prior to ICU admission (Figure 4), compared to 192 210 minutes for the previous model (P < 0.01).

Figure 4

Discussion

We have demonstrated a simple method for generating an algorithm derived from routine laboratory and hemodynamic values that is capable of predicting the onset of sepsis in a significant proportion of non‐ICU patients. Two prediction models were generated, 1 with and 1 without ABG data included in the analysis. In the 2006 and 2007 validation cohorts, the model including these data correctly classified 54.7% and 55.0% of the patients who developed septic shock and 93.4% and 93.0% of control patients, respectively. The second model identified 19.1% and 15.3% of the septic shock patients and 96.6% and 96.8% of the control patients for 2006 and 2007, respectively. The methods used in generating this model are relatively simple and can be executed with the use of an electronic medical record system.

Early, goal‐directed cardiovascular resuscitation and adequate initial antibiotic therapy have been shown to decrease mortality in patients with severe sepsis and septic shock.2, 26 Prior studies employing early, targeted resuscitation strategies have demonstrated decreased use of vasopressors10 and decreased mortality.510 In addition, we previously demonstrated that a standardized order set for the management of severe sepsis in the emergency department that focused on early and aggressive intervention was associated with decreased 28‐day mortality.1 These studies suggest that early, aggressive management of septic shock can improve outcomes. Identification of patients prior to overt clinical deterioration may allow for early intervention aimed at preventing shock or improving its outcome.

The purpose of this method is to develop a model capable of recognizing patterns in clinical data that herald a patient's otherwise unidentified clinical deterioration. It is not intended to replace existing outcome prediction tools or severity of illness scoring systems, where a high degree of accuracy would be required. Rather, it would be best implemented as an automated screening tool incorporated into an electronic medical record system. When a hospitalized patient is identified as a possible septic shock patient by the classification tree, a notification is then issued to the clinicians caring for the patient. The primary goal of this method is to notify clinicians of potential clinical deterioration. Any action taken as a result of this notification is at the discretion of the clinician. This method could be employed for any population of hospitalized patients, though because of variations in clinical practice and patient physiology, different models would need to be generated for differing patient populations.

This method has limitations, the foremost of which is the possible instability of the resulting classification model. This type of analysis results in an algorithm that depends on binary splits to classify patients. In generating the algorithm, the recursive partitioning analysis selects the variables and cutoff values that result in the strongest decision tree with the most pure classifications at the end nodes. These variables and cutoff values may not immediately seem logical from a clinical standpoint, and may vary with changes in practice and even possibly between divisions within a hospital. As a result, the algorithm would likely require intermittent updating to remain effective and a model derived from 1 hospital or patient population would not necessarily be applicable to patients at another institution or from a different population. However, once the method has been developed at an institution, the process of revising the algorithm could be essentially automated and uses few resources.

Another shortcoming of this method is the relatively low sensitivity of the resulting algorithm. In a role as an automated alert system, a low false‐positive rate is particularly desirable to avoid unnecessary frequent distraction of clinicians. The sensitivity of the model can be improved through manipulation of how the analysis is performed, but this would be at the expense of a higher false‐positive rate, which is not acceptable. Finally, prior studies examining treatment for sepsis have demonstrated an advantage to early and aggressive therapy. It is not clear, however, if identifying these patients prior to the onset of clinically evident sepsis would result in improved outcomes. Further work is required to determine if this is the case. We are currently conducting a prospective study that employs the method described here in conjunction with an automated alert system to ascertain if it impacts outcomes on patients admitted to the medicine wards of Barnes‐Jewish Hospital.

In conclusion, the method presented here represents a technique that consumes few resources and is capable of identifying some patients before septic shock becomes clinically evident. When applied in an automated fashion with the capability to alert clinicians caring for a patient, the method demonstrated here may allow for earlier diagnosis and possibly intervention for septic shock patients.

If clinician‐quality improvers are to gain traction as academicians,1 their first objective should be to bring quality improvement (QI) sandly into the world of scientific method. We believe that Dr. Chakraborti's 2 pointsthat the reasons for afferent limb failure need to be more closely investigated, and that lessons learned from 1 hospital's rapid response system (RRS) may not generalize to other hospitalsreflect the immaturity of QI as a science. In clinical science, 3 well‐defined testing phases bring 1 homogeneous, rigorously tested product to market that is monitored in a fourth phase. While Dr. Chakraborti urges us to examine our afferent limb failures more closely, the monitoring and reporting strategies used in the Josie King Patient Safety Program2 resonate with the postmarketing surveillance of Phase IV trials.

Although necessary and valid, we believe that the majority of the QI conundrum of RRS lies in the lack of premarket, stepwise testing of QI products. QI initiatives are often promulgated before an appropriate evidence base has been established. This lack of scientific rigor has resulted in RRS with calling criteria that have poor operating characteristics,3 undetermined methods for achieving afferent success,4 and efferent response arms of varying sizes and compositions.5 Consequently, a heterogeneous group of RRS have produced equivocal outcomes6 and diminished the applicability of lessons learned across institutions.

Indeed, while it is important to ask, What do we do now?, it may be more informative to answer the question, How did we get here?

If clinician‐quality improvers are to gain traction as academicians,1 their first objective should be to bring quality improvement (QI) sandly into the world of scientific method. We believe that Dr. Chakraborti's 2 pointsthat the reasons for afferent limb failure need to be more closely investigated, and that lessons learned from 1 hospital's rapid response system (RRS) may not generalize to other hospitalsreflect the immaturity of QI as a science. In clinical science, 3 well‐defined testing phases bring 1 homogeneous, rigorously tested product to market that is monitored in a fourth phase. While Dr. Chakraborti urges us to examine our afferent limb failures more closely, the monitoring and reporting strategies used in the Josie King Patient Safety Program2 resonate with the postmarketing surveillance of Phase IV trials.

Although necessary and valid, we believe that the majority of the QI conundrum of RRS lies in the lack of premarket, stepwise testing of QI products. QI initiatives are often promulgated before an appropriate evidence base has been established. This lack of scientific rigor has resulted in RRS with calling criteria that have poor operating characteristics,3 undetermined methods for achieving afferent success,4 and efferent response arms of varying sizes and compositions.5 Consequently, a heterogeneous group of RRS have produced equivocal outcomes6 and diminished the applicability of lessons learned across institutions.

Indeed, while it is important to ask, What do we do now?, it may be more informative to answer the question, How did we get here?

If clinician‐quality improvers are to gain traction as academicians,1 their first objective should be to bring quality improvement (QI) sandly into the world of scientific method. We believe that Dr. Chakraborti's 2 pointsthat the reasons for afferent limb failure need to be more closely investigated, and that lessons learned from 1 hospital's rapid response system (RRS) may not generalize to other hospitalsreflect the immaturity of QI as a science. In clinical science, 3 well‐defined testing phases bring 1 homogeneous, rigorously tested product to market that is monitored in a fourth phase. While Dr. Chakraborti urges us to examine our afferent limb failures more closely, the monitoring and reporting strategies used in the Josie King Patient Safety Program2 resonate with the postmarketing surveillance of Phase IV trials.

Although necessary and valid, we believe that the majority of the QI conundrum of RRS lies in the lack of premarket, stepwise testing of QI products. QI initiatives are often promulgated before an appropriate evidence base has been established. This lack of scientific rigor has resulted in RRS with calling criteria that have poor operating characteristics,3 undetermined methods for achieving afferent success,4 and efferent response arms of varying sizes and compositions.5 Consequently, a heterogeneous group of RRS have produced equivocal outcomes6 and diminished the applicability of lessons learned across institutions.

Indeed, while it is important to ask, What do we do now?, it may be more informative to answer the question, How did we get here?

Tears well up in my eyes. Not from pain, but frustration.

I pleaded with my 3‐year‐olds to come to me, so I could help them wash and dress for the day. Ordinarily, I would have come into their room, leaned over their beds, and whispered good morning into their ears.

Four days after surgery, I wasn't able to do that. Not without the crutches. Even with the crutches, I was moving slowly. I scooted up, put a few pillows behind my back, and carefully lifted up my leg. Putting on the knee immobilizer would take too long; I would only be crossing the hall.

Weighing less than 800 g, the immobilizer is an adjustable aluminum frame attached to foam rubber and 4 wide Velcro straps. It is a device of torment. I can never seem to find the proper fit. If it is too tight, my leg hurts and starts turning blue. When it is too loose, it pulls on my incision, multiplying the pain. You put it on while sitting, but you only discover whether it is too tight or too loose when you stand up.

With one foot on the floor at the edge of the bed, I took one crutch in hand, shifted to the left, and grabbed for the other crutch. I squeezed the handgrips of the crutches tightly and pulled myself up.

Put your weight on your hands, NOT your armpits, the instructions said. Easier said than done, because my armpits were now sore too.

Keep your crutches even with each other. I tried to remember to make an equilateral triangle with my good foot and 2 crutches, but the instructions only seemed to account for movement in a straight line.

Keep your elbows slightly bent and close to your sides to help keep the crutches under your arm. I am trying that too.

Lock your elbows. This instruction contradicts the previous one. Which should I follow?

Place your crutches 2 to 3 inches outside of each foot. How do I do this and keep my elbows close to my sides?

Swing your injured leg through first. But not too much, or I'll pole vault across the room.

After 3 steps, my leg started to throb, and my quadriceps went into a spasm. I needed to sit down, immediately.

Non‐weight‐bearing for next 4 weeks, was written on the doctor's note. Four weeks feels like an eternity. If my heel or toe even touched the ground, I felt an immediate throb of pain in my knee.

How do my orthopedic patients do it: those with broken hips and bad or broken knees, with hip replacements, or knee replacements? I was sensitive to their pain and could optimize control. Postoperatively, I could support gastrointestinal and other organ systems. I made sure that the basic weight‐bearing order was correct. I carefully followed the physical and/or occupational therapy recommendation for home, rehabilitation, or a nursing home. I spoke with families. Yet I did not dwell on activities of daily living nor how my patients felt to be dependent on others for simple things they needed. Some patients were non‐weight‐bearing for weeks, some for months, others never walked again. How can one deeply understand it if one has never experienced it?

According to Centers for Disease Control and Prevention (CDC) statistics,1 unintentional injuries are the leading cause of death between the ages of 1 and 44 years. Accidents are again the leading cause of death after age 75 years. More startling is the fact that these numbers have not been significantly reduced since these data were first collected. Nor do these numbers reveal the number of those debilitated, but still living. In my case, I was not playing basketball, water skiing, or rock climbing: I slipped and fell while stepping into a wading pool with my children, severing ligaments, tearing meniscus, and creating a hairline fracture of my tibia.

Expect to move slowly with the crutches. Yes, I am moving very slowly, only 1 to 2 feet with each stride. If I take a longer stride, I lose my balance.

Learn to sit down with the crutches. Learn to stand up. Learn to get into a car.

Learn to go up the stairs. Learn to go down the stairs.

Avoid wet surfaces. Otherwise you'll start skating and reinjure that knee.

It was effortless to think of patients and colleagues of mine. Mr. S., how do you do it? At age 30 years, he was quadriplegic from falling from a tree as a teenager. Mr. D, how about you, in a wheelchair, after being hit by a car and multiple postoperative leg infections and amputations: living in hotels, with estranged family and no social support. How did you do it, while we nagged you about controlling your glucose. Dr. J? The sadness fills my heart. At only 60 years old, my wonderful professor now made rounds in an electric wheelchair, the victim of amyotrophic lateral sclerosis (ALS). My fate, in comparison, is fortunate; my immobility only temporary. I can still think. And talk. And use a phone. And eat, as well as write this essay. And, best of all, read stories to my children. Be careful about this leg, I remind them. Don't come too close!

Maybe what life is trying to tell you is, slow down. In fact, it's forcing you to do that, a colleague and friend said. I had prided myself in thorough and compassionate doctoring. Now, having been a patient on crutches, I have a greater understanding of how limits to mobility can impact daily living on my patients after I discharge them. And the mundane subject of accident prevention has gained a new urgency.

Tears well up in my eyes. Not from pain, but frustration.

I pleaded with my 3‐year‐olds to come to me, so I could help them wash and dress for the day. Ordinarily, I would have come into their room, leaned over their beds, and whispered good morning into their ears.

Four days after surgery, I wasn't able to do that. Not without the crutches. Even with the crutches, I was moving slowly. I scooted up, put a few pillows behind my back, and carefully lifted up my leg. Putting on the knee immobilizer would take too long; I would only be crossing the hall.

Weighing less than 800 g, the immobilizer is an adjustable aluminum frame attached to foam rubber and 4 wide Velcro straps. It is a device of torment. I can never seem to find the proper fit. If it is too tight, my leg hurts and starts turning blue. When it is too loose, it pulls on my incision, multiplying the pain. You put it on while sitting, but you only discover whether it is too tight or too loose when you stand up.

With one foot on the floor at the edge of the bed, I took one crutch in hand, shifted to the left, and grabbed for the other crutch. I squeezed the handgrips of the crutches tightly and pulled myself up.

Put your weight on your hands, NOT your armpits, the instructions said. Easier said than done, because my armpits were now sore too.

Keep your crutches even with each other. I tried to remember to make an equilateral triangle with my good foot and 2 crutches, but the instructions only seemed to account for movement in a straight line.

Keep your elbows slightly bent and close to your sides to help keep the crutches under your arm. I am trying that too.

Lock your elbows. This instruction contradicts the previous one. Which should I follow?

Place your crutches 2 to 3 inches outside of each foot. How do I do this and keep my elbows close to my sides?

Swing your injured leg through first. But not too much, or I'll pole vault across the room.

After 3 steps, my leg started to throb, and my quadriceps went into a spasm. I needed to sit down, immediately.

Non‐weight‐bearing for next 4 weeks, was written on the doctor's note. Four weeks feels like an eternity. If my heel or toe even touched the ground, I felt an immediate throb of pain in my knee.

How do my orthopedic patients do it: those with broken hips and bad or broken knees, with hip replacements, or knee replacements? I was sensitive to their pain and could optimize control. Postoperatively, I could support gastrointestinal and other organ systems. I made sure that the basic weight‐bearing order was correct. I carefully followed the physical and/or occupational therapy recommendation for home, rehabilitation, or a nursing home. I spoke with families. Yet I did not dwell on activities of daily living nor how my patients felt to be dependent on others for simple things they needed. Some patients were non‐weight‐bearing for weeks, some for months, others never walked again. How can one deeply understand it if one has never experienced it?

According to Centers for Disease Control and Prevention (CDC) statistics,1 unintentional injuries are the leading cause of death between the ages of 1 and 44 years. Accidents are again the leading cause of death after age 75 years. More startling is the fact that these numbers have not been significantly reduced since these data were first collected. Nor do these numbers reveal the number of those debilitated, but still living. In my case, I was not playing basketball, water skiing, or rock climbing: I slipped and fell while stepping into a wading pool with my children, severing ligaments, tearing meniscus, and creating a hairline fracture of my tibia.

Expect to move slowly with the crutches. Yes, I am moving very slowly, only 1 to 2 feet with each stride. If I take a longer stride, I lose my balance.

Learn to sit down with the crutches. Learn to stand up. Learn to get into a car.

Learn to go up the stairs. Learn to go down the stairs.

Avoid wet surfaces. Otherwise you'll start skating and reinjure that knee.

It was effortless to think of patients and colleagues of mine. Mr. S., how do you do it? At age 30 years, he was quadriplegic from falling from a tree as a teenager. Mr. D, how about you, in a wheelchair, after being hit by a car and multiple postoperative leg infections and amputations: living in hotels, with estranged family and no social support. How did you do it, while we nagged you about controlling your glucose. Dr. J? The sadness fills my heart. At only 60 years old, my wonderful professor now made rounds in an electric wheelchair, the victim of amyotrophic lateral sclerosis (ALS). My fate, in comparison, is fortunate; my immobility only temporary. I can still think. And talk. And use a phone. And eat, as well as write this essay. And, best of all, read stories to my children. Be careful about this leg, I remind them. Don't come too close!

Maybe what life is trying to tell you is, slow down. In fact, it's forcing you to do that, a colleague and friend said. I had prided myself in thorough and compassionate doctoring. Now, having been a patient on crutches, I have a greater understanding of how limits to mobility can impact daily living on my patients after I discharge them. And the mundane subject of accident prevention has gained a new urgency.

Tears well up in my eyes. Not from pain, but frustration.

I pleaded with my 3‐year‐olds to come to me, so I could help them wash and dress for the day. Ordinarily, I would have come into their room, leaned over their beds, and whispered good morning into their ears.

Four days after surgery, I wasn't able to do that. Not without the crutches. Even with the crutches, I was moving slowly. I scooted up, put a few pillows behind my back, and carefully lifted up my leg. Putting on the knee immobilizer would take too long; I would only be crossing the hall.

Weighing less than 800 g, the immobilizer is an adjustable aluminum frame attached to foam rubber and 4 wide Velcro straps. It is a device of torment. I can never seem to find the proper fit. If it is too tight, my leg hurts and starts turning blue. When it is too loose, it pulls on my incision, multiplying the pain. You put it on while sitting, but you only discover whether it is too tight or too loose when you stand up.

With one foot on the floor at the edge of the bed, I took one crutch in hand, shifted to the left, and grabbed for the other crutch. I squeezed the handgrips of the crutches tightly and pulled myself up.

Put your weight on your hands, NOT your armpits, the instructions said. Easier said than done, because my armpits were now sore too.

Keep your crutches even with each other. I tried to remember to make an equilateral triangle with my good foot and 2 crutches, but the instructions only seemed to account for movement in a straight line.

Keep your elbows slightly bent and close to your sides to help keep the crutches under your arm. I am trying that too.

Lock your elbows. This instruction contradicts the previous one. Which should I follow?

Place your crutches 2 to 3 inches outside of each foot. How do I do this and keep my elbows close to my sides?

Swing your injured leg through first. But not too much, or I'll pole vault across the room.

After 3 steps, my leg started to throb, and my quadriceps went into a spasm. I needed to sit down, immediately.

Non‐weight‐bearing for next 4 weeks, was written on the doctor's note. Four weeks feels like an eternity. If my heel or toe even touched the ground, I felt an immediate throb of pain in my knee.

How do my orthopedic patients do it: those with broken hips and bad or broken knees, with hip replacements, or knee replacements? I was sensitive to their pain and could optimize control. Postoperatively, I could support gastrointestinal and other organ systems. I made sure that the basic weight‐bearing order was correct. I carefully followed the physical and/or occupational therapy recommendation for home, rehabilitation, or a nursing home. I spoke with families. Yet I did not dwell on activities of daily living nor how my patients felt to be dependent on others for simple things they needed. Some patients were non‐weight‐bearing for weeks, some for months, others never walked again. How can one deeply understand it if one has never experienced it?

According to Centers for Disease Control and Prevention (CDC) statistics,1 unintentional injuries are the leading cause of death between the ages of 1 and 44 years. Accidents are again the leading cause of death after age 75 years. More startling is the fact that these numbers have not been significantly reduced since these data were first collected. Nor do these numbers reveal the number of those debilitated, but still living. In my case, I was not playing basketball, water skiing, or rock climbing: I slipped and fell while stepping into a wading pool with my children, severing ligaments, tearing meniscus, and creating a hairline fracture of my tibia.

Expect to move slowly with the crutches. Yes, I am moving very slowly, only 1 to 2 feet with each stride. If I take a longer stride, I lose my balance.

Learn to sit down with the crutches. Learn to stand up. Learn to get into a car.

Learn to go up the stairs. Learn to go down the stairs.

Avoid wet surfaces. Otherwise you'll start skating and reinjure that knee.

It was effortless to think of patients and colleagues of mine. Mr. S., how do you do it? At age 30 years, he was quadriplegic from falling from a tree as a teenager. Mr. D, how about you, in a wheelchair, after being hit by a car and multiple postoperative leg infections and amputations: living in hotels, with estranged family and no social support. How did you do it, while we nagged you about controlling your glucose. Dr. J? The sadness fills my heart. At only 60 years old, my wonderful professor now made rounds in an electric wheelchair, the victim of amyotrophic lateral sclerosis (ALS). My fate, in comparison, is fortunate; my immobility only temporary. I can still think. And talk. And use a phone. And eat, as well as write this essay. And, best of all, read stories to my children. Be careful about this leg, I remind them. Don't come too close!

Maybe what life is trying to tell you is, slow down. In fact, it's forcing you to do that, a colleague and friend said. I had prided myself in thorough and compassionate doctoring. Now, having been a patient on crutches, I have a greater understanding of how limits to mobility can impact daily living on my patients after I discharge them. And the mundane subject of accident prevention has gained a new urgency.

Clostridium difficile colitis (CDC) is the most common cause of hospital‐acquired diarrhea.1 The incidence of CDC has sharply increased over the past decade despite increasing awareness among health care professionals.24 C. difficile pathogenic strains induce diarrhea through the elaboration and secretion of 2 exotoxins: toxin A and toxin B. Toxin A is an inflammatory toxin, leading to fluid secretion, increased mucosal permeability, and marked enteritis and colitis.5 Toxin B is cytotoxic, leading to cell injury and apoptosis.5 Combined, toxin A and toxin B can cause a wide spectrum of clinical presentations, ranging from mild diarrhea that resolves with the discontinuation of antibiotics to a fulminant colitis requiring surgical intervention.

The severity of clinical manifestations has been shown to be inversely proportional to the host anti‐toxin A antibody level in response to toxin exposure. Kyne et al.6 demonstrated that asymptomatic C. difficile carriers produced significantly higher anti‐toxin A immunoglobulin G (IgG) levels compared to symptomatic patients. Among the latter group, patients with mild disease had a higher antibody level compared to those with severe colitis.6, 7 Additional risk factors predisposing to severe colitis are advanced age, severe underlying illness,8 and immunocompromised state.9

Recently, a new C. difficile strain (BI/NAP1) with a mutated toxin A and toxin B promoter silencer, a binary toxin gene, and fluoroquinolone resistance has been described in Canada and the United States.3, 10 This strain has been associated with an increased incidence of CDC among hospitalized patients, especially the incidence of severe disease requiring colectomy. At the same time, several reports describing metronidazole treatment failure have been published.1114 These recent findings emphasize the importance of finding alternative treatments for CDC.

Intravenous immunoglobulin (IVIG) was used to treat CDC for the first time in 1991.15 Since then, 12 case reports and small case series, along with 1 case‐control study have been published documenting IVIG treatment outcomes.1527 However, only 5 reports to date examined patients with severe CDC.2125 In the present study, we report the largest series of patients with severe CDC treated with IVIG in the literature to our knowledge.

Patients and Methods

Results

Discussion

To our knowledge, the present study is the largest series published in the literature to date on the use of IVIG for severe CDC. It is also the first study to report a high mortality rate compared to the 5 previous smaller studies on this topic. In the first report on IVIG use for CDC, Leung et al.15 used IVIG to treat 5 pediatric patients suffering from chronic relapsing CDC. It was not until 7 years later, in 1997, that the first IVIG use for severe CDC was reported.22 Since then, a total of 13 works have been published on IVIG for CDC treatment, and only in 5 of these was IVIG administered for severe CDC treatment:2125 3 case reports, 1 case series, and 1 case‐control study. Although the 4 uncontrolled reports concluded that IVIG is beneficial for severe CDC, the only controlled study reported no significant difference between cases and controls for all‐cause mortality, length of stay, and colectomy rate.25

The definition of severe CDC varied between reports, making comparison difficult. McPherson et al.21 and Hassoun and Ibrahim24 defined severe disease as one causing pancolitis on CT scan either with or without megacolon. In the study by Juang et al.,25 disease severity was assessed using the modified criteria of Rubin et al.9 Salcedo et al.22 defined severe CDC as one causing pancolitis in one patient and thumbprinting on CT scan in another, whereas Chandrasekar et al.23 defined it as one causing shock requiring inotropic support and presence of pseudomembranes on colonoscopy.

The present report is unique in that it provided 2 scales to characterize disease in each patient. The first scale measured colonic involvement anatomically and physiologically using a combination of computerized axial tomography (CAT) scan findings, presence or absence of ileus or referral for possible colectomy. This is not a prognostic scale, however, since CAT scan findings have been previously shown to be poor predictors of treatment outcome.29 The second scale measured the severity of systemic involvement using a well‐validated and standardized scale, the APACHE II score. We have shown in this report that it is associated with prognosis in the context of IVIG use.

Our study reports a higher mortality rate than previously described, and suggests that risk stratification and patient selection are important before IVIG administration, since not all patients seem to benefit from this treatment as previously suggested by smaller case series. Previously, several physical findings and laboratory values were found to be associated with worse outcome in CDC. These were increasing age, immunosuppression, shock requiring vasopressors, peak white blood cell count, peak serum lactate level, hypoalbuminemia, a fall in serum albumin level of >1.1 g/dL at the onset of CDC symptoms, use of 3 or more antibiotics, comorbid disease, previous history of CDC, acute renal failure and hypotension, underlying altered or depressed mental status, abdominal pain or distention, white blood cell count over 20,000/mm³ or <1500/mm³ and/or a >10% band forms on the white blood cell differential count, and ascites or pneumatosis coli by abdominal imaging.9, 3033 Using the APACHE II scale for the same purpose has the advantage of utilizing a well‐validated and objective scale that is expected to measure the degree of systemic involvement more reliably compared to the clinical and laboratory values above.

Timing of IVIG infusion remains controversial. Due to the lack of randomized controlled trials, the current practice is guided by expert opinion, leading to wide variations between reports. Since the APACHE II score was positively associated with mortality in the setting of IVIG treatment, the same scale could be used to guide decisions regarding timing of IVIG infusion. Our results suggest that IVIG should be preferentially used while the APACHE II score is still relatively low. This association and the specific APACHE II score at which to initiate or hold treatment need to be validated in the setting of a randomized controlled study before being used in clinical practice.

Although the current study was not designed to test this theory, IVIG could be associated conceptually with treatment success for patients with severe disease that is still restricted to the colon (without other organ dysfunction or at least at an early stage of extracolonic organ failure and thus associated with a low APACHE II score) but not for severe colonic disease with secondary multiple organ failure (high APACHE II score). This may be because colonic disease is toxin‐mediated whereas secondary systemic involvement is mediated through toxin‐induced inflammatory mediators (interleukin‐8, macrophage‐inflammatory protein‐2, substance P, tumor necrosis factor‐alpha) released locally in the colon,3436 triggering a SIRS and hematogenous translocation of colonic bacteria,37 both of which are poorly responsive to immunoglobulin infusion. Along the same lines, waiting for failure of conventional therapy before IVIG use might result in IVIG treatment failure because of disease progression and secondary sepsis, at which point no treatment may be effective. No study thus far has addressed this issue specifically.

Overall, the combined cohort of patients with severe CDC treated with IVIG in the literature includes 51 patients (Table 4). The current report contributes 41% of these patients. The patients' average age was 68 years, with a 2 to 1 female‐male ratio. The dose of IVIG used varied largely also, with 400 mg/kg being the mode (range, 75400 mg/kg from 1 to 5 doses). This dose is significantly below the doses used in the treatment of other diseases, like Guillain‐Barr syndrome, myasthenia gravis, Kawasaki disease, autoimmune hemolytic anemia, agammaglobulinemia, and hypogammaglobulinemia, where the usual dose is 400 mg/kg for 5 days. The resolution of diarrhea in these cases occurred after an average of 9 (range, 142) days. The index hospitalization survival rate varied from 43% to 100%. Patients received standard treatment for an average of 13 (range, 065) days before IVIG infusion. Thirty‐two of 51 patients survived their illness (63%). Neither total IgG nor anti‐toxin A IgG levels were measured in any of the reports. Of the 32 patients who had clinical resolution, 3 (10%) experienced symptoms recurrence in a follow‐up period of 1 to 13 months. This number is most probably an underestimation of the true recurrence rate resulting from an incomplete reporting because there was no uniform or active recurrence ascertainment mechanism in any of the studies. The recurrences were at 10, 14, and 30 days posttreatment. Since standard treatment was not discontinued in any of the reports once IVIG was given, the relative contribution and the ideal timing for IVIG infusion are still unclear.

The mechanism of action of IVIG is passive immunization (with anti‐toxin A and anti‐toxin B antibodies present in the pooled immunoglobulin) of a host who is usually unable to mount an adequate protective immune response.15, 22 IVIG is formed from pooling immunoglobulin from several random donors. It has been shown that many such donors express high anti‐toxin A and anti‐toxin B antibody serum titers.38, 39 In addition, high levels of anti‐toxin A and anti‐toxin B antibodies were present in the IVIG preparations and the recipients after infusion.1517, 22 Although constituting only a small fraction of the total IVIG administered, these antitoxin antibodies are believed to neutralize toxin A and B and help the host recover from the disease. In fact, Babcock et al.40 used an experimental hamster model of CDC to demonstrate a mortality reduction from 100% to 55% postinfusion of combined anti‐toxin A and anti‐toxin B antibodies. While some early reports indicated that anti‐toxin B antibodies were the major determinants of protection against colitis,41 later reports correlated disease severity pathologically42 and clinically43, 44 with anti‐toxin A levels. Anti‐toxin B antibodies were later shown to play an adjunctive role in conferring immunity against CDC40, 45, 46 when added to anti‐toxin A antibodies, but not to have any significant role on their own.

However, IVIG has been shown to contain IgG, but not IgA, anti‐toxin A and anti‐toxin B antibodies while it is only the IgA class of anti‐toxin A antibodies, and not the IgG class, that could neutralize toxin A in vitro and in vivo.47, 48 Babcock et al.40 solved this apparent dilemma by showing that a combination of 3 different monoclonal IgG anti‐toxin A antibodies could neutralize toxin A activity in vitro and prevent disease in the hamster model in vivo. Each of the 3 antibodies recognized a different toxin A domain: the first neutralized toxin A enzymatic activity, while the second prevented toxin A binding to its receptor on enterocytes, and the third prevented toxin internalization after binding to the receptor.

Thus, the mechanism of action of IVIG is most likely through the transfer of IgG anti‐toxin A antibodies that gain access to the intestinal lumen presumably secondary to inflammation‐induced mucosal damage and neutralize toxin A. Transfer of yet undetected IgA anti‐toxin A antibodies that prevent toxin A from binding to its receptor is much less likely, although possible.

The present study has limitations. As in all retrospective studies, selection bias was unavoidable. In addition, the decision to initiate IVIG administration was dependent on the attending physician, who also decided the dose, leading to heterogeneity in the total dose of IVIG infused. Such heterogeneity, however, is primarily the result of a lack of a standard dose for IVIG infusion for CDC in the published literature, as reported above. In addition, since IVIG is not yet approved by the U.S. Food and Drug Administration (FDA) for the treatment of severe CDC, standard treatment was not discontinued in any of the reports to date, including ours.

Choosing appropriate controls for patients suffering from severe CDC is challenging. This patient population is usually frail, with severe and multiple underlying diseases. The deteriorating clinical condition (and subsequently the need for multiple antibiotics) may be either the result of or the cause of CDC. Furthermore, IVIG has been in short supply for several years and therefore it has been expensive, making its administration to the number of patients needed to design adequately‐powered controlled studies difficult. These are mainly the reasons no randomized, multicenter, placebo‐controlled trial on IVIG use in severe CDC has been conducted to date.

Conclusions

In the present study, we report the results of IVIG use for the treatment of 21 patients with severe CDC. This is the largest cohort to our knowledge in the literature. Unlike previous studies on the subject, the present report provided 2 scales for disease assessment: the first based on the extent of colonic involvement and the second measuring the severity of systemic involvement using the APACHE II score. The latter was positively associated with mortality in this context. Of the 21 study patients treated with IVIG, only 9 patients (43%) survived their illness. This is the highest reported mortality rate among all studies on this subject so far. Further studies on the ideal timing of IVIG infusion, dose, and patient selection are needed before accepting IVIG as a standard of care for severe CDC treatment. The role of APACHE II score in the decision to use IVIG is promising and should be validated in randomized controlled trials.

Clostridium difficile colitis (CDC) is the most common cause of hospital‐acquired diarrhea.1 The incidence of CDC has sharply increased over the past decade despite increasing awareness among health care professionals.24 C. difficile pathogenic strains induce diarrhea through the elaboration and secretion of 2 exotoxins: toxin A and toxin B. Toxin A is an inflammatory toxin, leading to fluid secretion, increased mucosal permeability, and marked enteritis and colitis.5 Toxin B is cytotoxic, leading to cell injury and apoptosis.5 Combined, toxin A and toxin B can cause a wide spectrum of clinical presentations, ranging from mild diarrhea that resolves with the discontinuation of antibiotics to a fulminant colitis requiring surgical intervention.

The severity of clinical manifestations has been shown to be inversely proportional to the host anti‐toxin A antibody level in response to toxin exposure. Kyne et al.6 demonstrated that asymptomatic C. difficile carriers produced significantly higher anti‐toxin A immunoglobulin G (IgG) levels compared to symptomatic patients. Among the latter group, patients with mild disease had a higher antibody level compared to those with severe colitis.6, 7 Additional risk factors predisposing to severe colitis are advanced age, severe underlying illness,8 and immunocompromised state.9

Recently, a new C. difficile strain (BI/NAP1) with a mutated toxin A and toxin B promoter silencer, a binary toxin gene, and fluoroquinolone resistance has been described in Canada and the United States.3, 10 This strain has been associated with an increased incidence of CDC among hospitalized patients, especially the incidence of severe disease requiring colectomy. At the same time, several reports describing metronidazole treatment failure have been published.1114 These recent findings emphasize the importance of finding alternative treatments for CDC.

Intravenous immunoglobulin (IVIG) was used to treat CDC for the first time in 1991.15 Since then, 12 case reports and small case series, along with 1 case‐control study have been published documenting IVIG treatment outcomes.1527 However, only 5 reports to date examined patients with severe CDC.2125 In the present study, we report the largest series of patients with severe CDC treated with IVIG in the literature to our knowledge.

Patients and Methods

Results