Choosing Wisely: Things We Do For No Reason

RECOMMENDATIONS

CONCLUSION

RECOMMENDATIONS

CONCLUSION

RECOMMENDATIONS

CONCLUSION

The Things We Do for No Reason (TWDFNR) series reviews practices which have become common parts of hospital care but which may provide little value to our patients. Practices reviewed in the TWDFNR series do not represent black and white conclusions or clinical practice standards, but are meant as a starting place for research and active discussions among hospitalists and patients. We invite you to be part of that discussion. https://www.choosingwisely.org/

Inherited thrombophilia refers to a genetic condition that predisposes to an increased risk of venous thromboembolism (VTE). This disorder is prevalent in approximately 7% of the population and includes mutations such as factor V Leiden, prothrombin 20210, protein C deficiency, protein S deficiency, antithrombin deficiency, and methylene tetrahydrofolate reductase. The relative risk of VTE is 3‐ to 20‐fold greater in patients with inherited thrombophilia compared with the general population. Is testing for inherited thrombophilia recommended? The available evidence suggests that testing for inherited thrombophilia is not recommended in most clinical settings. In patients without a personal history of VTE, thrombophilia results do not change management, as there is no evidence to support thromboprophylaxis in this setting. In patients with a personal history of provoked or unprovoked VTE, inpatient testing is not indicated, as results do not influence management, testing is not cost‐effective, and a positive test result may lead to unnecessary patient anxiety or may result in unnecessary involvement of consultants. Testing in hospitalized patients has even more limitations because many thrombophilia tests are inaccurate in the setting of acute VTE and/or anticoagulation.

CASE PRESENTATION

A 23‐year‐old man presents to the emergency room with pleuritic chest pain and new oxygen requirement of 2 L nasal cannula. He has a history of unprovoked lower extremity deep venous thrombosis (DVT) diagnosed at age 20 and completed 3 months of systemic anticoagulation without complications. He reports no family history of clotting disorders or venous thromboembolism (VTE) and no reversible risk factors for VTE such as prolonged immobility, recent surgery, or high‐risk medications. A computed tomogram pulmonary embolism protocol shows multiple right lower lobe, segmental pulmonary emboli. Anticoagulation is initiated, and the patient is admitted to the hospital. Will inpatient inherited thrombophilia testing impact management for this case?

WHY MAY INHERITED THROMBOPHILIA TESTING PROVE HELPFUL?

The annual incidence rate of a first VTE event is estimated as 117 per 100,000 individuals per year.[1] The most common presentations are symptomatic DVT of the leg (annual incidence approximately 48 per 100,000 people), or a pulmonary embolism (annual incidence approximately 69 per 100,000 people).[1] Pulmonary embolism results in death in up to 30% of untreated patients and 2.5% of patients who receive systemic anticoagulation.[2] Principal in the pathogenesis of VTE are factors described by Virchow's triad: venous stasis, endothelial injury, and systemic hypercoagulability. By identifying a mutation in 1 or more of the factors in the clotting pathway, an evaluation for inherited thrombophilia theoretically may unearth factors that drive systemic hypercoagulability and inform decision making so as to prevent future events.

Inherited thrombophilia refers to a genetic condition that predisposes to an increased risk of VTE.[3] Approximately 7% of the general population has inherited thrombophilia, which includes factor V Leiden (FVL) mutation, prothrombin 20210 mutation (PT20210), protein C deficiency, protein S deficiency, antithrombin III (ATIII) deficiency, and methylene tetrahydrofolate reductase mutation (MTHFR).[4] Of note, the definition does not include acquired etiologies, such as antiphospholipid antibody syndrome. Depending on the underlying condition and expression of the genetic abnormality, the relative risk of VTE in patients with inherited thrombophilia is 3‐ to 20‐fold greater than that of the general population.[5] Therefore, it is logical to consider that testing for inherited thrombophilia might be clinically useful. However, the evidence for doing so is very limited.

DOES INHERITED THROMBOPHILIA TESTING CHANGE MANAGEMENT?

An inherited thrombophilia evaluation is unlikely to affect management in most clinical settings. There is no current evidence to support primary prophylaxis[6] nor is there evidence that management of patients with recurrent VTE should be altered in the setting of inherited thrombophilia.

To date, no prospective trials have evaluated the efficacy of anticoagulant use for primary prevention of VTE in patients with inherited thrombophilia.[6] Given the limited evidence for thromboprophylaxis and risks of anticoagulation, primary prevention for patients with inherited thrombophilia that remain asymptomatic is not recommended by the current American College of Chest Physicians guidelines.[7, 8]

Similarly, in patients with a first VTE or recurrent VTE, diagnosis of inherited thrombophilia is often not associated with recurrent events, which suggests that other nongenetic factors may be just as important, if not more important, in determining the risk of recurrence.[9] Although no randomized controlled or controlled clinical trials have evaluated the effects of testing for inherited thrombophilia on recurrent VTE,[10, 11] several prospective studies have assessed risk factors for recurrence. Data from these studies suggest that recurrence rates after unprovoked VTE are only weakly correlated with inherited thrombophilia status.[12, 13] Rather, it is postulated that patients with recurrent VTE may exhibit a prothrombotic tendency regardless of underlying genetic predisposition. In this case, decisions regarding anticoagulation do not vary by thrombophilia status. Instead, thrombophilia testing may divert attention away from the management of more prevalent, potentially modifiable risk factors such as immobility, oral contraceptive use, or malignancy, all of which are associated with recurrent VTE.[14] These provoking factors are the most important determinants of the chance of VTE recurrence as well as the most significant factors to take into account when deciding duration of anticoagulation.

Christiansen et al. performed a prospective study evaluating the association between recurrent VTE and thrombophilia status. After following 474 patients with confirmed first episode VTE for a mean of 7.3 years, no statistically significant risk of VTE was found for patients with FVL (hazard ratio [HR]: 1.2, 95% confidence interval [CI]: 0.7‐1.9), PT20210 (HR: 0.7, 95% CI: 0.3‐2.0), or an anticoagulant (protein C, protein S or ATIII) deficiency (HR: 1.8, 95% CI: 0.9‐3.7).[15] Although unexplained VTE was statistically associated with VTE recurrence, heritable thrombophilia status was not.

In a systematic review and meta‐analysis investigating the association of FVL and PT20210 with recurrent VTE, Ho and colleagues found a statistically significant risk of recurrent VTE in patients with inherited thrombophilia due to FVL (odds ratio [OR]: 1.41, 95% CI: 1.14‐1.75) and PT20210 (OR: 1.72, 95% CI: 1.27‐2.31), and reported that at most, only up to 1 in 6 recurrent VTEs may be attributable to these mutations.[16] Based on this relatively modest effect, the authors question the utility of testing for inherited thrombophilia, as thrombophilia status is unlikely to warrant a change in type or duration of treatment.

Regardless of whether an underlying inherited thrombophilia is identified, patients with history of recurrent VTE are often candidates for long‐term anticoagulation. Testing for inherited thrombophilia in patients with prior VTE events will therefore not influence decisions regarding clinical management. Additionally, such testing may be confounded by ongoing disease or treatment (Table 1). For example, protein C, protein S antigen, and ATIII levels are low in the setting of acute VTE.[17, 18] Likewise, protein C and S (vitamin Kdependent proteins) will be low in the setting of anticoagulation with warfarin.[19] Moreover, ATIII activity and antigen levels are low in the setting of heparin use.[20] Lack of provider awareness regarding these interactions may have important negative consequences, including a spurious diagnosis of thrombophilia,[21, 22] unnecessary hematology consultation, and psychological distress to patients in the form of ongoing unwarranted testing or apprehension regarding recurrence.[23]

Additionally, this expensive evaluation has estimated direct costs of $1100 to $2400 per thrombophilia panel based on estimation of charges billed by a large commercial laboratory.[24, 25] In 2014, over 280,000 claims were submitted under Medicare Part B across all care settings for a thrombophilia analysis including FVL, PT20210, and MTHFR gene mutations,[24] which would equate to between $300 million to $672 million.[26] Unfortunately, there have been no large‐scale trials to assess cost‐effectiveness. However, the Evaluation of Genomic Applications in Practice and Prevention (EGAPP) Working Group stated that cost‐effectiveness modeling studies in this area require updating with current VTE risk estimates but are suggestive that routine FVL/PT20210 testing is not cost‐effective.[27]

ARE THERE CIRCUMSTANCES IN WHICH INPATIENT INHERITED THROMBOPHILIA TESTING PROVES BENEFICIAL?

The evidence for when to test for inherited thrombophilia is very limited and is often based on individualized risk. The current EGAPP guidelines acknowledge this limitation, specifically noting that there is a paucity of data evaluating management or prophylaxis of patients with homozygous or compound heterozygous FVL or P20210 mutation, and a lack of data surrounding whether or not knowledge of thrombophilia mutation should affect anticoagulation treatment.[27] This is why an individualized approach is deemed necessary. For example, the decision to prescribe hormone replacement therapy in women with a family history of inherited thrombophilia may be better informed by testing prior to treatment. Similarly, pregnant women with a family history or personal history of VTE may also benefit from inherited thrombophilia testing, as this may influence antepartum or postpartum management.[28, 29] The National Institute for Health and Clinical Excellence (NICE) guidelines recommend consideration of testing for hereditary thrombophilia in patients with unprovoked VTE and a first‐degree relative with VTE, if stopping anticoagulation treatment is planned; however, these recommendations are based solely on Guideline Development Group's experience and opinion.[30] Regardless, testing for inherited thrombophilia has significant potential consequences. Patients at risk should meet with an outpatient hematologist and/or a genetic counselor, if available, to determine the risks and benefits of testing.

WHAT DO GUIDELINES SAY ABOUT INHERITED THROMBOPHILIA TESTING?

The most recent NICE guidelines recommend against offering inherited thrombophilia testing to patients presenting with a provoked VTE in any clinical setting.[30] In patients diagnosed with unprovoked VTE, testing should not be considered unless a first degree relative with a history of VTE exists.[30] The NICE guidelines also recommend against routinely offering thrombophilia testing to asymptomatic first‐degree relatives of patients with a history of VTE or known inherited thrombophilia. This recommendation is reflected in the American Society of Hematology's Choosing Wisely recommendations since 2013.[31] Further, The American College of Medical Genetics and Genomics' Choosing Wisely recommendations from 2015 state that MTHFR mutations should never be included in any thrombophilia workup, as recent meta‐analyses have disproven an association between the presence of these variants and venous thromboembolism.[32]

The EGAPP Working Group recommends against routine testing for FVL or PT20210 in patients who present with an idiopathic VTE, as longer‐term anticoagulation offers similar benefits to patients with or without these mutations.[27] EGAPP also recommends against testing asymptomatic adult family members of patients with VTE and/or an FVL or PT20210 mutation for the purpose of considering primary prophylactic anticoagulation. In these circumstances, it is felt that the potential risks of thrombophilia testing outweigh any potential benefits.

HOW SHOULD HOSPITALISTS APPROACH TESTING OF INHERITED THROMBOPHILIA?

The providers in our case presentation are challenged with determining whether inpatient thrombophilia evaluation will add value to the evaluation of patients with unprovoked VTE. The available evidence suggests that clinicians should avoid ordering thrombophilia testing for hospitalized patients with unprovoked VTE because (1) many thrombophilia tests are inaccurate in the setting of acute VTE and/or anticoagulation, (2) results of testing often do not influence management, (3) testing is not cost‐effective, (4) a positive test result may lead to unnecessary patient anxiety, and (5) testing may result in inappropriately prolonged anticoagulation courses or unnecessary involvement of inpatient consultants. For these reasons, the patient in our case presentation should not be tested for inherited thrombophilia. In patients with personal or family histories of recurrent thromboembolism, modifiable clinical risk factors should be addressed, as these are more likely to influence treatment decisions compared to genetic testing. Finally, patients may be referred to an outpatient hematologist or geneticist for individualized discussions of risks and benefits of testing for inherited thrombophilia.

CONCLUSION

Inpatient evaluation for inherited thrombophilia for VTE is not clinically useful, cost‐effective, or reliable in the setting of VTE. The result of such testing does not affect management of acute primary or recurrent VTE. Testing should only be considered using an individualized approach in the outpatient setting with appropriate genetic counseling.

Disclosure: Christopher M. Petrilli, MD, and Lauren Heidemann, MD, contributed equally to this work. The authors report no conflicts of interest.

Do you think this is a low‐value practice? Is this truly a Thing We Do for No Reason? Share what you do in your practice and join in the conversation online by retweeting it on Twitter (#TWDFNR) and liking it on Facebook. We invite you to propose ideas for other Things We Do for No Reason topics by emailing TWDFNR@hospitalmedicine.org.

The Things We Do for No Reason (TWDFNR) series reviews practices which have become common parts of hospital care but which may provide little value to our patients. Practices reviewed in the TWDFNR series do not represent black and white conclusions or clinical practice standards, but are meant as a starting place for research and active discussions among hospitalists and patients. We invite you to be part of that discussion. https://www.choosingwisely.org/

Inherited thrombophilia refers to a genetic condition that predisposes to an increased risk of venous thromboembolism (VTE). This disorder is prevalent in approximately 7% of the population and includes mutations such as factor V Leiden, prothrombin 20210, protein C deficiency, protein S deficiency, antithrombin deficiency, and methylene tetrahydrofolate reductase. The relative risk of VTE is 3‐ to 20‐fold greater in patients with inherited thrombophilia compared with the general population. Is testing for inherited thrombophilia recommended? The available evidence suggests that testing for inherited thrombophilia is not recommended in most clinical settings. In patients without a personal history of VTE, thrombophilia results do not change management, as there is no evidence to support thromboprophylaxis in this setting. In patients with a personal history of provoked or unprovoked VTE, inpatient testing is not indicated, as results do not influence management, testing is not cost‐effective, and a positive test result may lead to unnecessary patient anxiety or may result in unnecessary involvement of consultants. Testing in hospitalized patients has even more limitations because many thrombophilia tests are inaccurate in the setting of acute VTE and/or anticoagulation.

CASE PRESENTATION

A 23‐year‐old man presents to the emergency room with pleuritic chest pain and new oxygen requirement of 2 L nasal cannula. He has a history of unprovoked lower extremity deep venous thrombosis (DVT) diagnosed at age 20 and completed 3 months of systemic anticoagulation without complications. He reports no family history of clotting disorders or venous thromboembolism (VTE) and no reversible risk factors for VTE such as prolonged immobility, recent surgery, or high‐risk medications. A computed tomogram pulmonary embolism protocol shows multiple right lower lobe, segmental pulmonary emboli. Anticoagulation is initiated, and the patient is admitted to the hospital. Will inpatient inherited thrombophilia testing impact management for this case?

WHY MAY INHERITED THROMBOPHILIA TESTING PROVE HELPFUL?

The annual incidence rate of a first VTE event is estimated as 117 per 100,000 individuals per year.[1] The most common presentations are symptomatic DVT of the leg (annual incidence approximately 48 per 100,000 people), or a pulmonary embolism (annual incidence approximately 69 per 100,000 people).[1] Pulmonary embolism results in death in up to 30% of untreated patients and 2.5% of patients who receive systemic anticoagulation.[2] Principal in the pathogenesis of VTE are factors described by Virchow's triad: venous stasis, endothelial injury, and systemic hypercoagulability. By identifying a mutation in 1 or more of the factors in the clotting pathway, an evaluation for inherited thrombophilia theoretically may unearth factors that drive systemic hypercoagulability and inform decision making so as to prevent future events.

Inherited thrombophilia refers to a genetic condition that predisposes to an increased risk of VTE.[3] Approximately 7% of the general population has inherited thrombophilia, which includes factor V Leiden (FVL) mutation, prothrombin 20210 mutation (PT20210), protein C deficiency, protein S deficiency, antithrombin III (ATIII) deficiency, and methylene tetrahydrofolate reductase mutation (MTHFR).[4] Of note, the definition does not include acquired etiologies, such as antiphospholipid antibody syndrome. Depending on the underlying condition and expression of the genetic abnormality, the relative risk of VTE in patients with inherited thrombophilia is 3‐ to 20‐fold greater than that of the general population.[5] Therefore, it is logical to consider that testing for inherited thrombophilia might be clinically useful. However, the evidence for doing so is very limited.

DOES INHERITED THROMBOPHILIA TESTING CHANGE MANAGEMENT?

An inherited thrombophilia evaluation is unlikely to affect management in most clinical settings. There is no current evidence to support primary prophylaxis[6] nor is there evidence that management of patients with recurrent VTE should be altered in the setting of inherited thrombophilia.

To date, no prospective trials have evaluated the efficacy of anticoagulant use for primary prevention of VTE in patients with inherited thrombophilia.[6] Given the limited evidence for thromboprophylaxis and risks of anticoagulation, primary prevention for patients with inherited thrombophilia that remain asymptomatic is not recommended by the current American College of Chest Physicians guidelines.[7, 8]

Similarly, in patients with a first VTE or recurrent VTE, diagnosis of inherited thrombophilia is often not associated with recurrent events, which suggests that other nongenetic factors may be just as important, if not more important, in determining the risk of recurrence.[9] Although no randomized controlled or controlled clinical trials have evaluated the effects of testing for inherited thrombophilia on recurrent VTE,[10, 11] several prospective studies have assessed risk factors for recurrence. Data from these studies suggest that recurrence rates after unprovoked VTE are only weakly correlated with inherited thrombophilia status.[12, 13] Rather, it is postulated that patients with recurrent VTE may exhibit a prothrombotic tendency regardless of underlying genetic predisposition. In this case, decisions regarding anticoagulation do not vary by thrombophilia status. Instead, thrombophilia testing may divert attention away from the management of more prevalent, potentially modifiable risk factors such as immobility, oral contraceptive use, or malignancy, all of which are associated with recurrent VTE.[14] These provoking factors are the most important determinants of the chance of VTE recurrence as well as the most significant factors to take into account when deciding duration of anticoagulation.

Christiansen et al. performed a prospective study evaluating the association between recurrent VTE and thrombophilia status. After following 474 patients with confirmed first episode VTE for a mean of 7.3 years, no statistically significant risk of VTE was found for patients with FVL (hazard ratio [HR]: 1.2, 95% confidence interval [CI]: 0.7‐1.9), PT20210 (HR: 0.7, 95% CI: 0.3‐2.0), or an anticoagulant (protein C, protein S or ATIII) deficiency (HR: 1.8, 95% CI: 0.9‐3.7).[15] Although unexplained VTE was statistically associated with VTE recurrence, heritable thrombophilia status was not.

In a systematic review and meta‐analysis investigating the association of FVL and PT20210 with recurrent VTE, Ho and colleagues found a statistically significant risk of recurrent VTE in patients with inherited thrombophilia due to FVL (odds ratio [OR]: 1.41, 95% CI: 1.14‐1.75) and PT20210 (OR: 1.72, 95% CI: 1.27‐2.31), and reported that at most, only up to 1 in 6 recurrent VTEs may be attributable to these mutations.[16] Based on this relatively modest effect, the authors question the utility of testing for inherited thrombophilia, as thrombophilia status is unlikely to warrant a change in type or duration of treatment.

Regardless of whether an underlying inherited thrombophilia is identified, patients with history of recurrent VTE are often candidates for long‐term anticoagulation. Testing for inherited thrombophilia in patients with prior VTE events will therefore not influence decisions regarding clinical management. Additionally, such testing may be confounded by ongoing disease or treatment (Table 1). For example, protein C, protein S antigen, and ATIII levels are low in the setting of acute VTE.[17, 18] Likewise, protein C and S (vitamin Kdependent proteins) will be low in the setting of anticoagulation with warfarin.[19] Moreover, ATIII activity and antigen levels are low in the setting of heparin use.[20] Lack of provider awareness regarding these interactions may have important negative consequences, including a spurious diagnosis of thrombophilia,[21, 22] unnecessary hematology consultation, and psychological distress to patients in the form of ongoing unwarranted testing or apprehension regarding recurrence.[23]

Additionally, this expensive evaluation has estimated direct costs of $1100 to $2400 per thrombophilia panel based on estimation of charges billed by a large commercial laboratory.[24, 25] In 2014, over 280,000 claims were submitted under Medicare Part B across all care settings for a thrombophilia analysis including FVL, PT20210, and MTHFR gene mutations,[24] which would equate to between $300 million to $672 million.[26] Unfortunately, there have been no large‐scale trials to assess cost‐effectiveness. However, the Evaluation of Genomic Applications in Practice and Prevention (EGAPP) Working Group stated that cost‐effectiveness modeling studies in this area require updating with current VTE risk estimates but are suggestive that routine FVL/PT20210 testing is not cost‐effective.[27]

ARE THERE CIRCUMSTANCES IN WHICH INPATIENT INHERITED THROMBOPHILIA TESTING PROVES BENEFICIAL?

The evidence for when to test for inherited thrombophilia is very limited and is often based on individualized risk. The current EGAPP guidelines acknowledge this limitation, specifically noting that there is a paucity of data evaluating management or prophylaxis of patients with homozygous or compound heterozygous FVL or P20210 mutation, and a lack of data surrounding whether or not knowledge of thrombophilia mutation should affect anticoagulation treatment.[27] This is why an individualized approach is deemed necessary. For example, the decision to prescribe hormone replacement therapy in women with a family history of inherited thrombophilia may be better informed by testing prior to treatment. Similarly, pregnant women with a family history or personal history of VTE may also benefit from inherited thrombophilia testing, as this may influence antepartum or postpartum management.[28, 29] The National Institute for Health and Clinical Excellence (NICE) guidelines recommend consideration of testing for hereditary thrombophilia in patients with unprovoked VTE and a first‐degree relative with VTE, if stopping anticoagulation treatment is planned; however, these recommendations are based solely on Guideline Development Group's experience and opinion.[30] Regardless, testing for inherited thrombophilia has significant potential consequences. Patients at risk should meet with an outpatient hematologist and/or a genetic counselor, if available, to determine the risks and benefits of testing.

WHAT DO GUIDELINES SAY ABOUT INHERITED THROMBOPHILIA TESTING?

The most recent NICE guidelines recommend against offering inherited thrombophilia testing to patients presenting with a provoked VTE in any clinical setting.[30] In patients diagnosed with unprovoked VTE, testing should not be considered unless a first degree relative with a history of VTE exists.[30] The NICE guidelines also recommend against routinely offering thrombophilia testing to asymptomatic first‐degree relatives of patients with a history of VTE or known inherited thrombophilia. This recommendation is reflected in the American Society of Hematology's Choosing Wisely recommendations since 2013.[31] Further, The American College of Medical Genetics and Genomics' Choosing Wisely recommendations from 2015 state that MTHFR mutations should never be included in any thrombophilia workup, as recent meta‐analyses have disproven an association between the presence of these variants and venous thromboembolism.[32]

The EGAPP Working Group recommends against routine testing for FVL or PT20210 in patients who present with an idiopathic VTE, as longer‐term anticoagulation offers similar benefits to patients with or without these mutations.[27] EGAPP also recommends against testing asymptomatic adult family members of patients with VTE and/or an FVL or PT20210 mutation for the purpose of considering primary prophylactic anticoagulation. In these circumstances, it is felt that the potential risks of thrombophilia testing outweigh any potential benefits.

HOW SHOULD HOSPITALISTS APPROACH TESTING OF INHERITED THROMBOPHILIA?

The providers in our case presentation are challenged with determining whether inpatient thrombophilia evaluation will add value to the evaluation of patients with unprovoked VTE. The available evidence suggests that clinicians should avoid ordering thrombophilia testing for hospitalized patients with unprovoked VTE because (1) many thrombophilia tests are inaccurate in the setting of acute VTE and/or anticoagulation, (2) results of testing often do not influence management, (3) testing is not cost‐effective, (4) a positive test result may lead to unnecessary patient anxiety, and (5) testing may result in inappropriately prolonged anticoagulation courses or unnecessary involvement of inpatient consultants. For these reasons, the patient in our case presentation should not be tested for inherited thrombophilia. In patients with personal or family histories of recurrent thromboembolism, modifiable clinical risk factors should be addressed, as these are more likely to influence treatment decisions compared to genetic testing. Finally, patients may be referred to an outpatient hematologist or geneticist for individualized discussions of risks and benefits of testing for inherited thrombophilia.

CONCLUSION

Inpatient evaluation for inherited thrombophilia for VTE is not clinically useful, cost‐effective, or reliable in the setting of VTE. The result of such testing does not affect management of acute primary or recurrent VTE. Testing should only be considered using an individualized approach in the outpatient setting with appropriate genetic counseling.

Disclosure: Christopher M. Petrilli, MD, and Lauren Heidemann, MD, contributed equally to this work. The authors report no conflicts of interest.

Do you think this is a low‐value practice? Is this truly a Thing We Do for No Reason? Share what you do in your practice and join in the conversation online by retweeting it on Twitter (#TWDFNR) and liking it on Facebook. We invite you to propose ideas for other Things We Do for No Reason topics by emailing TWDFNR@hospitalmedicine.org.

The Things We Do for No Reason (TWDFNR) series reviews practices which have become common parts of hospital care but which may provide little value to our patients. Practices reviewed in the TWDFNR series do not represent black and white conclusions or clinical practice standards, but are meant as a starting place for research and active discussions among hospitalists and patients. We invite you to be part of that discussion. https://www.choosingwisely.org/

Inherited thrombophilia refers to a genetic condition that predisposes to an increased risk of venous thromboembolism (VTE). This disorder is prevalent in approximately 7% of the population and includes mutations such as factor V Leiden, prothrombin 20210, protein C deficiency, protein S deficiency, antithrombin deficiency, and methylene tetrahydrofolate reductase. The relative risk of VTE is 3‐ to 20‐fold greater in patients with inherited thrombophilia compared with the general population. Is testing for inherited thrombophilia recommended? The available evidence suggests that testing for inherited thrombophilia is not recommended in most clinical settings. In patients without a personal history of VTE, thrombophilia results do not change management, as there is no evidence to support thromboprophylaxis in this setting. In patients with a personal history of provoked or unprovoked VTE, inpatient testing is not indicated, as results do not influence management, testing is not cost‐effective, and a positive test result may lead to unnecessary patient anxiety or may result in unnecessary involvement of consultants. Testing in hospitalized patients has even more limitations because many thrombophilia tests are inaccurate in the setting of acute VTE and/or anticoagulation.

CASE PRESENTATION

A 23‐year‐old man presents to the emergency room with pleuritic chest pain and new oxygen requirement of 2 L nasal cannula. He has a history of unprovoked lower extremity deep venous thrombosis (DVT) diagnosed at age 20 and completed 3 months of systemic anticoagulation without complications. He reports no family history of clotting disorders or venous thromboembolism (VTE) and no reversible risk factors for VTE such as prolonged immobility, recent surgery, or high‐risk medications. A computed tomogram pulmonary embolism protocol shows multiple right lower lobe, segmental pulmonary emboli. Anticoagulation is initiated, and the patient is admitted to the hospital. Will inpatient inherited thrombophilia testing impact management for this case?

WHY MAY INHERITED THROMBOPHILIA TESTING PROVE HELPFUL?

The annual incidence rate of a first VTE event is estimated as 117 per 100,000 individuals per year.[1] The most common presentations are symptomatic DVT of the leg (annual incidence approximately 48 per 100,000 people), or a pulmonary embolism (annual incidence approximately 69 per 100,000 people).[1] Pulmonary embolism results in death in up to 30% of untreated patients and 2.5% of patients who receive systemic anticoagulation.[2] Principal in the pathogenesis of VTE are factors described by Virchow's triad: venous stasis, endothelial injury, and systemic hypercoagulability. By identifying a mutation in 1 or more of the factors in the clotting pathway, an evaluation for inherited thrombophilia theoretically may unearth factors that drive systemic hypercoagulability and inform decision making so as to prevent future events.

Inherited thrombophilia refers to a genetic condition that predisposes to an increased risk of VTE.[3] Approximately 7% of the general population has inherited thrombophilia, which includes factor V Leiden (FVL) mutation, prothrombin 20210 mutation (PT20210), protein C deficiency, protein S deficiency, antithrombin III (ATIII) deficiency, and methylene tetrahydrofolate reductase mutation (MTHFR).[4] Of note, the definition does not include acquired etiologies, such as antiphospholipid antibody syndrome. Depending on the underlying condition and expression of the genetic abnormality, the relative risk of VTE in patients with inherited thrombophilia is 3‐ to 20‐fold greater than that of the general population.[5] Therefore, it is logical to consider that testing for inherited thrombophilia might be clinically useful. However, the evidence for doing so is very limited.

DOES INHERITED THROMBOPHILIA TESTING CHANGE MANAGEMENT?

An inherited thrombophilia evaluation is unlikely to affect management in most clinical settings. There is no current evidence to support primary prophylaxis[6] nor is there evidence that management of patients with recurrent VTE should be altered in the setting of inherited thrombophilia.

To date, no prospective trials have evaluated the efficacy of anticoagulant use for primary prevention of VTE in patients with inherited thrombophilia.[6] Given the limited evidence for thromboprophylaxis and risks of anticoagulation, primary prevention for patients with inherited thrombophilia that remain asymptomatic is not recommended by the current American College of Chest Physicians guidelines.[7, 8]

Similarly, in patients with a first VTE or recurrent VTE, diagnosis of inherited thrombophilia is often not associated with recurrent events, which suggests that other nongenetic factors may be just as important, if not more important, in determining the risk of recurrence.[9] Although no randomized controlled or controlled clinical trials have evaluated the effects of testing for inherited thrombophilia on recurrent VTE,[10, 11] several prospective studies have assessed risk factors for recurrence. Data from these studies suggest that recurrence rates after unprovoked VTE are only weakly correlated with inherited thrombophilia status.[12, 13] Rather, it is postulated that patients with recurrent VTE may exhibit a prothrombotic tendency regardless of underlying genetic predisposition. In this case, decisions regarding anticoagulation do not vary by thrombophilia status. Instead, thrombophilia testing may divert attention away from the management of more prevalent, potentially modifiable risk factors such as immobility, oral contraceptive use, or malignancy, all of which are associated with recurrent VTE.[14] These provoking factors are the most important determinants of the chance of VTE recurrence as well as the most significant factors to take into account when deciding duration of anticoagulation.

Christiansen et al. performed a prospective study evaluating the association between recurrent VTE and thrombophilia status. After following 474 patients with confirmed first episode VTE for a mean of 7.3 years, no statistically significant risk of VTE was found for patients with FVL (hazard ratio [HR]: 1.2, 95% confidence interval [CI]: 0.7‐1.9), PT20210 (HR: 0.7, 95% CI: 0.3‐2.0), or an anticoagulant (protein C, protein S or ATIII) deficiency (HR: 1.8, 95% CI: 0.9‐3.7).[15] Although unexplained VTE was statistically associated with VTE recurrence, heritable thrombophilia status was not.

In a systematic review and meta‐analysis investigating the association of FVL and PT20210 with recurrent VTE, Ho and colleagues found a statistically significant risk of recurrent VTE in patients with inherited thrombophilia due to FVL (odds ratio [OR]: 1.41, 95% CI: 1.14‐1.75) and PT20210 (OR: 1.72, 95% CI: 1.27‐2.31), and reported that at most, only up to 1 in 6 recurrent VTEs may be attributable to these mutations.[16] Based on this relatively modest effect, the authors question the utility of testing for inherited thrombophilia, as thrombophilia status is unlikely to warrant a change in type or duration of treatment.

Regardless of whether an underlying inherited thrombophilia is identified, patients with history of recurrent VTE are often candidates for long‐term anticoagulation. Testing for inherited thrombophilia in patients with prior VTE events will therefore not influence decisions regarding clinical management. Additionally, such testing may be confounded by ongoing disease or treatment (Table 1). For example, protein C, protein S antigen, and ATIII levels are low in the setting of acute VTE.[17, 18] Likewise, protein C and S (vitamin Kdependent proteins) will be low in the setting of anticoagulation with warfarin.[19] Moreover, ATIII activity and antigen levels are low in the setting of heparin use.[20] Lack of provider awareness regarding these interactions may have important negative consequences, including a spurious diagnosis of thrombophilia,[21, 22] unnecessary hematology consultation, and psychological distress to patients in the form of ongoing unwarranted testing or apprehension regarding recurrence.[23]

Additionally, this expensive evaluation has estimated direct costs of $1100 to $2400 per thrombophilia panel based on estimation of charges billed by a large commercial laboratory.[24, 25] In 2014, over 280,000 claims were submitted under Medicare Part B across all care settings for a thrombophilia analysis including FVL, PT20210, and MTHFR gene mutations,[24] which would equate to between $300 million to $672 million.[26] Unfortunately, there have been no large‐scale trials to assess cost‐effectiveness. However, the Evaluation of Genomic Applications in Practice and Prevention (EGAPP) Working Group stated that cost‐effectiveness modeling studies in this area require updating with current VTE risk estimates but are suggestive that routine FVL/PT20210 testing is not cost‐effective.[27]

ARE THERE CIRCUMSTANCES IN WHICH INPATIENT INHERITED THROMBOPHILIA TESTING PROVES BENEFICIAL?

The evidence for when to test for inherited thrombophilia is very limited and is often based on individualized risk. The current EGAPP guidelines acknowledge this limitation, specifically noting that there is a paucity of data evaluating management or prophylaxis of patients with homozygous or compound heterozygous FVL or P20210 mutation, and a lack of data surrounding whether or not knowledge of thrombophilia mutation should affect anticoagulation treatment.[27] This is why an individualized approach is deemed necessary. For example, the decision to prescribe hormone replacement therapy in women with a family history of inherited thrombophilia may be better informed by testing prior to treatment. Similarly, pregnant women with a family history or personal history of VTE may also benefit from inherited thrombophilia testing, as this may influence antepartum or postpartum management.[28, 29] The National Institute for Health and Clinical Excellence (NICE) guidelines recommend consideration of testing for hereditary thrombophilia in patients with unprovoked VTE and a first‐degree relative with VTE, if stopping anticoagulation treatment is planned; however, these recommendations are based solely on Guideline Development Group's experience and opinion.[30] Regardless, testing for inherited thrombophilia has significant potential consequences. Patients at risk should meet with an outpatient hematologist and/or a genetic counselor, if available, to determine the risks and benefits of testing.

WHAT DO GUIDELINES SAY ABOUT INHERITED THROMBOPHILIA TESTING?

The most recent NICE guidelines recommend against offering inherited thrombophilia testing to patients presenting with a provoked VTE in any clinical setting.[30] In patients diagnosed with unprovoked VTE, testing should not be considered unless a first degree relative with a history of VTE exists.[30] The NICE guidelines also recommend against routinely offering thrombophilia testing to asymptomatic first‐degree relatives of patients with a history of VTE or known inherited thrombophilia. This recommendation is reflected in the American Society of Hematology's Choosing Wisely recommendations since 2013.[31] Further, The American College of Medical Genetics and Genomics' Choosing Wisely recommendations from 2015 state that MTHFR mutations should never be included in any thrombophilia workup, as recent meta‐analyses have disproven an association between the presence of these variants and venous thromboembolism.[32]

The EGAPP Working Group recommends against routine testing for FVL or PT20210 in patients who present with an idiopathic VTE, as longer‐term anticoagulation offers similar benefits to patients with or without these mutations.[27] EGAPP also recommends against testing asymptomatic adult family members of patients with VTE and/or an FVL or PT20210 mutation for the purpose of considering primary prophylactic anticoagulation. In these circumstances, it is felt that the potential risks of thrombophilia testing outweigh any potential benefits.

HOW SHOULD HOSPITALISTS APPROACH TESTING OF INHERITED THROMBOPHILIA?

The providers in our case presentation are challenged with determining whether inpatient thrombophilia evaluation will add value to the evaluation of patients with unprovoked VTE. The available evidence suggests that clinicians should avoid ordering thrombophilia testing for hospitalized patients with unprovoked VTE because (1) many thrombophilia tests are inaccurate in the setting of acute VTE and/or anticoagulation, (2) results of testing often do not influence management, (3) testing is not cost‐effective, (4) a positive test result may lead to unnecessary patient anxiety, and (5) testing may result in inappropriately prolonged anticoagulation courses or unnecessary involvement of inpatient consultants. For these reasons, the patient in our case presentation should not be tested for inherited thrombophilia. In patients with personal or family histories of recurrent thromboembolism, modifiable clinical risk factors should be addressed, as these are more likely to influence treatment decisions compared to genetic testing. Finally, patients may be referred to an outpatient hematologist or geneticist for individualized discussions of risks and benefits of testing for inherited thrombophilia.

CONCLUSION

Inpatient evaluation for inherited thrombophilia for VTE is not clinically useful, cost‐effective, or reliable in the setting of VTE. The result of such testing does not affect management of acute primary or recurrent VTE. Testing should only be considered using an individualized approach in the outpatient setting with appropriate genetic counseling.

Disclosure: Christopher M. Petrilli, MD, and Lauren Heidemann, MD, contributed equally to this work. The authors report no conflicts of interest.

Do you think this is a low‐value practice? Is this truly a Thing We Do for No Reason? Share what you do in your practice and join in the conversation online by retweeting it on Twitter (#TWDFNR) and liking it on Facebook. We invite you to propose ideas for other Things We Do for No Reason topics by emailing TWDFNR@hospitalmedicine.org.

The Things We Do for No Reason (TWDFNR) series reviews practices which have become common parts of hospital care but which may provide little value to our patients. Practices reviewed in the TWDFNR series do not represent black and white conclusions or clinical practice standards, but are meant as a starting place for research and active discussions among hospitalists and patients. We invite you to be part of that discussion. https://www.choosingwisely.org/

A 65‐year‐old immunocompetent man with a history of obesity, diabetes, and chronic lower extremity edema presents to the emergency room with a 1‐day history of right lower extremity pain and increased swelling. He reports no antecedent trauma and states he just noticed the symptoms that morning. On examination, he appears generally well. His temperature is 100F, pulse 92 beats per minute, blood pressure 120/60 mm Hg, and respiratory rate 16 breaths per minute. The rest of the exam is notable for right lower extremity erythema and swelling extending from his right shin to his right medial thigh without associated fluctuance or drainage. Labs reveal a mildly elevated white blood cell count of 13,000/L and normal serum creatinine. Are broad‐spectrum antibiotics like vancomycin and piperacillin/tazobactam the preferred regimen?

BACKGROUND

The term skin and soft tissue infection (SSTI) includes a heterogeneous group of infections including cellulitis, cutaneous abscess, diabetic foot infections, surgical site infections, and necrotizing soft tissue infections. As a group, SSTIs are the second most common type of infection in hospitalized adults in the United States behind pneumonia and result in more than 600,000 admissions per year.[1] The current guideline on SSTIs by the Infectious Disease Society of America (IDSA) makes the distinction between purulent and nonpurulent soft tissue infections based on the presence or absence of purulent drainage or abscess and between mild, moderate, and severe infections based on the presence and severity of systemic signs of infection.[2] Figure 1 provides an overview of the IDSA recommendations.

Figure 1

THE PROBLEM: OVERUSE OF BROAD‐SPECTRUM ANTIBIOTICS

Studies over the past decade have shown that the majority of patients hospitalized with SSTI receive broad‐spectrum antibiotics, usually with combinations of antibiotics active against gram‐positive (including methicillin‐resistant Staphylococcus aureus [MRSA]), gram‐negative (often including Pseudomonas aeruginosa), and anaerobic organisms. Broad‐spectrum treatment occurs despite guidelines from the IDSA, which state that the most common pathogens for nonpurulent cellulitis are ‐hemolytic streptococci, which remain susceptible to penicillin.[2, 3] One multicenter study of hospitalized adults with nonpurulent cellulitis, for example, reported that 85% of patients received therapy effective against MRSA (primarily vancomycin), 61% received broad gram‐negative coverage (primarily ‐lactam with ‐lactamase inhibitor), and 74% received anaerobic coverage.[4] Another multicenter study reported that the most common antibiotics given for cellulitis (excluding cases associated with cutaneous abscess) were vancomycin (60%), ‐lactam/‐lactamase combinations (32%), and clindamycin (19%). Only 13% of patients with cellulitis were treated with cefazolin, and only 1.1% of patients were treated with nafcillin or oxacillin.[5] According to the Centers for Disease Control and Prevention, unnecessary antibiotic use is associated with increased cost, development of antibiotic resistance, and increased rates of Clostridium difficile.[6]

The current use of broad‐spectrum antibiotics for nonpurulent cellulitis is likely due to several factors, including the emergence of community‐associated (CA)‐MRSA, confusion due to the heterogeneity of SSTI, and the limited data regarding the microbiology of nonpurulent cellulitis. The resulting uncertainty about cellulitis has been termed an existential crisis for the treating physician and is likely the single biggest factor behind the out‐of‐control prescribing.[7]

THE SOLUTION: NARROW‐SPECTRUM ANTIBIOTICS FOR MOST

Although studies of the microbiology of cellulitis remain inconclusive, several recent clinical trials have indicated that treatment with antimicrobials limited to ‐hemolytic streptococci and methicillin‐susceptible S aureus (MSSA) are as effective as antimicrobials against MRSA. A prospective study from 2010 of consecutive hospitalized adults with nonpurulent cellulitis found that 73% had serologic evidence for streptococcal infection, and overall 95.8% responded to cefazolin monotherapy.[12] More recently, a study of emergency room patients with nonpurulent cellulitis randomized patients to cephalexin alone or cephalexin plus trimethoprim‐sulfamethoxazole. These authors found no difference in response rates and concluded that the addition of anti‐MRSA therapy (trimethoprim‐sulfamethoxazole, in this study) for uncomplicated cellulitis was unnecessary.[13] This later study is the only randomized controlled study to assess the need for MRSA coverage for cellulitis, and the answer for outpatients, at least, is that MRSA coverage is unnecessary. Both of these studies are cited by the IDSA guideline from 2014, which recommends antibiotics for mild‐moderate cellulitis to be limited to antimicrobials effective against ‐hemolytic streptococci and MSSA. The guideline specifically does not recommend routinely treating for MRSA, gram‐negative, or anaerobic organisms citing lack of benefit as well as risks of antibiotic resistance and C difficile infection. A recent study from the University of Utah reported the development of a cellulitis order set, which included a pathway for nonpurulent cellulitis based on the use of cefazolin. These authors reported that the use of the pathway was associated with a 59% decrease in the use of broad‐spectrum antibiotics, a 23% decrease in pharmacy costs, a 13% decrease in total facility cost, with no change in hospital length of stay or readmission rate.[14] One important caveat to the use of clinical pathways is that they are often underused. In the study from the University of Utah, for example, only 55% of eligible patients had the clinical pathway ordered.

WHEN BROAD‐SPECTRUM ANTIBIOTICS ARE RECOMMENDED

The IDSA does recommend empiric broad‐spectrum antibiotics with combination gram‐positive and gram‐negative coverage in several situations, including severe infections in which necrotizing soft tissue infection is suspected, animal bites, immersion injuries, as well as for severely immunocompromised patients or those who have failed limited spectrum antibiotics. Additionally, the IDSA recommends antimicrobials effective against MRSA for purulent infections with systemic signs of inflammation as well as severe nonpurulent infections or those associated with penetrating trauma, injection drug use, and nasal colonization with MRSA (Figure 1).

RECOMMENDATIONS

Our patient has no associated purulence and no abscess and therefore has nonpurulent cellulitis. Based on his mild tachycardia and leukocytosis but intact immune system and lack of suspicion for necrotizing soft tissue infection, he would be classified as moderate‐severity cellulitis by the IDSA. In patients hospitalized with nonpurulent cellulitis who are not severely immunocompromised or severely ill and for whom necrotizing soft tissue infection is not suspected:

1. Antibiotics should be directed at ‐hemolytic streptococci and MSSA, with 1 of the suggested antibiotics by the IDSA including penicillin, ceftriaxone, cefazolin, or clindamycin.
2. Antibiotics effective against MRSA should be limited to situations described by the IDSA.
3. If the cellulitis has not improved within 48 hours, then consider broader‐spectrum antibiotics.
4. Hospitals should strongly consider implementation of a cellulitis pathway based on the IDSA recommendations to improve antibiotic stewardship as well as costs.

The Things We Do for No Reason (TWDFNR) series reviews practices which have become common parts of hospital care but which may provide little value to our patients. Practices reviewed in the TWDFNR series do not represent black and white conclusions or clinical practice standards, but are meant as a starting place for research and active discussions among hospitalists and patients. We invite you to be part of that discussion. https://www.choosingwisely.org/

A 65‐year‐old immunocompetent man with a history of obesity, diabetes, and chronic lower extremity edema presents to the emergency room with a 1‐day history of right lower extremity pain and increased swelling. He reports no antecedent trauma and states he just noticed the symptoms that morning. On examination, he appears generally well. His temperature is 100F, pulse 92 beats per minute, blood pressure 120/60 mm Hg, and respiratory rate 16 breaths per minute. The rest of the exam is notable for right lower extremity erythema and swelling extending from his right shin to his right medial thigh without associated fluctuance or drainage. Labs reveal a mildly elevated white blood cell count of 13,000/L and normal serum creatinine. Are broad‐spectrum antibiotics like vancomycin and piperacillin/tazobactam the preferred regimen?

BACKGROUND

The term skin and soft tissue infection (SSTI) includes a heterogeneous group of infections including cellulitis, cutaneous abscess, diabetic foot infections, surgical site infections, and necrotizing soft tissue infections. As a group, SSTIs are the second most common type of infection in hospitalized adults in the United States behind pneumonia and result in more than 600,000 admissions per year.[1] The current guideline on SSTIs by the Infectious Disease Society of America (IDSA) makes the distinction between purulent and nonpurulent soft tissue infections based on the presence or absence of purulent drainage or abscess and between mild, moderate, and severe infections based on the presence and severity of systemic signs of infection.[2] Figure 1 provides an overview of the IDSA recommendations.

Figure 1

THE PROBLEM: OVERUSE OF BROAD‐SPECTRUM ANTIBIOTICS

Studies over the past decade have shown that the majority of patients hospitalized with SSTI receive broad‐spectrum antibiotics, usually with combinations of antibiotics active against gram‐positive (including methicillin‐resistant Staphylococcus aureus [MRSA]), gram‐negative (often including Pseudomonas aeruginosa), and anaerobic organisms. Broad‐spectrum treatment occurs despite guidelines from the IDSA, which state that the most common pathogens for nonpurulent cellulitis are ‐hemolytic streptococci, which remain susceptible to penicillin.[2, 3] One multicenter study of hospitalized adults with nonpurulent cellulitis, for example, reported that 85% of patients received therapy effective against MRSA (primarily vancomycin), 61% received broad gram‐negative coverage (primarily ‐lactam with ‐lactamase inhibitor), and 74% received anaerobic coverage.[4] Another multicenter study reported that the most common antibiotics given for cellulitis (excluding cases associated with cutaneous abscess) were vancomycin (60%), ‐lactam/‐lactamase combinations (32%), and clindamycin (19%). Only 13% of patients with cellulitis were treated with cefazolin, and only 1.1% of patients were treated with nafcillin or oxacillin.[5] According to the Centers for Disease Control and Prevention, unnecessary antibiotic use is associated with increased cost, development of antibiotic resistance, and increased rates of Clostridium difficile.[6]

The current use of broad‐spectrum antibiotics for nonpurulent cellulitis is likely due to several factors, including the emergence of community‐associated (CA)‐MRSA, confusion due to the heterogeneity of SSTI, and the limited data regarding the microbiology of nonpurulent cellulitis. The resulting uncertainty about cellulitis has been termed an existential crisis for the treating physician and is likely the single biggest factor behind the out‐of‐control prescribing.[7]

THE SOLUTION: NARROW‐SPECTRUM ANTIBIOTICS FOR MOST

Although studies of the microbiology of cellulitis remain inconclusive, several recent clinical trials have indicated that treatment with antimicrobials limited to ‐hemolytic streptococci and methicillin‐susceptible S aureus (MSSA) are as effective as antimicrobials against MRSA. A prospective study from 2010 of consecutive hospitalized adults with nonpurulent cellulitis found that 73% had serologic evidence for streptococcal infection, and overall 95.8% responded to cefazolin monotherapy.[12] More recently, a study of emergency room patients with nonpurulent cellulitis randomized patients to cephalexin alone or cephalexin plus trimethoprim‐sulfamethoxazole. These authors found no difference in response rates and concluded that the addition of anti‐MRSA therapy (trimethoprim‐sulfamethoxazole, in this study) for uncomplicated cellulitis was unnecessary.[13] This later study is the only randomized controlled study to assess the need for MRSA coverage for cellulitis, and the answer for outpatients, at least, is that MRSA coverage is unnecessary. Both of these studies are cited by the IDSA guideline from 2014, which recommends antibiotics for mild‐moderate cellulitis to be limited to antimicrobials effective against ‐hemolytic streptococci and MSSA. The guideline specifically does not recommend routinely treating for MRSA, gram‐negative, or anaerobic organisms citing lack of benefit as well as risks of antibiotic resistance and C difficile infection. A recent study from the University of Utah reported the development of a cellulitis order set, which included a pathway for nonpurulent cellulitis based on the use of cefazolin. These authors reported that the use of the pathway was associated with a 59% decrease in the use of broad‐spectrum antibiotics, a 23% decrease in pharmacy costs, a 13% decrease in total facility cost, with no change in hospital length of stay or readmission rate.[14] One important caveat to the use of clinical pathways is that they are often underused. In the study from the University of Utah, for example, only 55% of eligible patients had the clinical pathway ordered.

WHEN BROAD‐SPECTRUM ANTIBIOTICS ARE RECOMMENDED

The IDSA does recommend empiric broad‐spectrum antibiotics with combination gram‐positive and gram‐negative coverage in several situations, including severe infections in which necrotizing soft tissue infection is suspected, animal bites, immersion injuries, as well as for severely immunocompromised patients or those who have failed limited spectrum antibiotics. Additionally, the IDSA recommends antimicrobials effective against MRSA for purulent infections with systemic signs of inflammation as well as severe nonpurulent infections or those associated with penetrating trauma, injection drug use, and nasal colonization with MRSA (Figure 1).

RECOMMENDATIONS

Our patient has no associated purulence and no abscess and therefore has nonpurulent cellulitis. Based on his mild tachycardia and leukocytosis but intact immune system and lack of suspicion for necrotizing soft tissue infection, he would be classified as moderate‐severity cellulitis by the IDSA. In patients hospitalized with nonpurulent cellulitis who are not severely immunocompromised or severely ill and for whom necrotizing soft tissue infection is not suspected:

1. Antibiotics should be directed at ‐hemolytic streptococci and MSSA, with 1 of the suggested antibiotics by the IDSA including penicillin, ceftriaxone, cefazolin, or clindamycin.
2. Antibiotics effective against MRSA should be limited to situations described by the IDSA.
3. If the cellulitis has not improved within 48 hours, then consider broader‐spectrum antibiotics.
4. Hospitals should strongly consider implementation of a cellulitis pathway based on the IDSA recommendations to improve antibiotic stewardship as well as costs.

The Things We Do for No Reason (TWDFNR) series reviews practices which have become common parts of hospital care but which may provide little value to our patients. Practices reviewed in the TWDFNR series do not represent black and white conclusions or clinical practice standards, but are meant as a starting place for research and active discussions among hospitalists and patients. We invite you to be part of that discussion. https://www.choosingwisely.org/

A 65‐year‐old immunocompetent man with a history of obesity, diabetes, and chronic lower extremity edema presents to the emergency room with a 1‐day history of right lower extremity pain and increased swelling. He reports no antecedent trauma and states he just noticed the symptoms that morning. On examination, he appears generally well. His temperature is 100F, pulse 92 beats per minute, blood pressure 120/60 mm Hg, and respiratory rate 16 breaths per minute. The rest of the exam is notable for right lower extremity erythema and swelling extending from his right shin to his right medial thigh without associated fluctuance or drainage. Labs reveal a mildly elevated white blood cell count of 13,000/L and normal serum creatinine. Are broad‐spectrum antibiotics like vancomycin and piperacillin/tazobactam the preferred regimen?

BACKGROUND

The term skin and soft tissue infection (SSTI) includes a heterogeneous group of infections including cellulitis, cutaneous abscess, diabetic foot infections, surgical site infections, and necrotizing soft tissue infections. As a group, SSTIs are the second most common type of infection in hospitalized adults in the United States behind pneumonia and result in more than 600,000 admissions per year.[1] The current guideline on SSTIs by the Infectious Disease Society of America (IDSA) makes the distinction between purulent and nonpurulent soft tissue infections based on the presence or absence of purulent drainage or abscess and between mild, moderate, and severe infections based on the presence and severity of systemic signs of infection.[2] Figure 1 provides an overview of the IDSA recommendations.

Figure 1

THE PROBLEM: OVERUSE OF BROAD‐SPECTRUM ANTIBIOTICS

Studies over the past decade have shown that the majority of patients hospitalized with SSTI receive broad‐spectrum antibiotics, usually with combinations of antibiotics active against gram‐positive (including methicillin‐resistant Staphylococcus aureus [MRSA]), gram‐negative (often including Pseudomonas aeruginosa), and anaerobic organisms. Broad‐spectrum treatment occurs despite guidelines from the IDSA, which state that the most common pathogens for nonpurulent cellulitis are ‐hemolytic streptococci, which remain susceptible to penicillin.[2, 3] One multicenter study of hospitalized adults with nonpurulent cellulitis, for example, reported that 85% of patients received therapy effective against MRSA (primarily vancomycin), 61% received broad gram‐negative coverage (primarily ‐lactam with ‐lactamase inhibitor), and 74% received anaerobic coverage.[4] Another multicenter study reported that the most common antibiotics given for cellulitis (excluding cases associated with cutaneous abscess) were vancomycin (60%), ‐lactam/‐lactamase combinations (32%), and clindamycin (19%). Only 13% of patients with cellulitis were treated with cefazolin, and only 1.1% of patients were treated with nafcillin or oxacillin.[5] According to the Centers for Disease Control and Prevention, unnecessary antibiotic use is associated with increased cost, development of antibiotic resistance, and increased rates of Clostridium difficile.[6]

The current use of broad‐spectrum antibiotics for nonpurulent cellulitis is likely due to several factors, including the emergence of community‐associated (CA)‐MRSA, confusion due to the heterogeneity of SSTI, and the limited data regarding the microbiology of nonpurulent cellulitis. The resulting uncertainty about cellulitis has been termed an existential crisis for the treating physician and is likely the single biggest factor behind the out‐of‐control prescribing.[7]

THE SOLUTION: NARROW‐SPECTRUM ANTIBIOTICS FOR MOST

Although studies of the microbiology of cellulitis remain inconclusive, several recent clinical trials have indicated that treatment with antimicrobials limited to ‐hemolytic streptococci and methicillin‐susceptible S aureus (MSSA) are as effective as antimicrobials against MRSA. A prospective study from 2010 of consecutive hospitalized adults with nonpurulent cellulitis found that 73% had serologic evidence for streptococcal infection, and overall 95.8% responded to cefazolin monotherapy.[12] More recently, a study of emergency room patients with nonpurulent cellulitis randomized patients to cephalexin alone or cephalexin plus trimethoprim‐sulfamethoxazole. These authors found no difference in response rates and concluded that the addition of anti‐MRSA therapy (trimethoprim‐sulfamethoxazole, in this study) for uncomplicated cellulitis was unnecessary.[13] This later study is the only randomized controlled study to assess the need for MRSA coverage for cellulitis, and the answer for outpatients, at least, is that MRSA coverage is unnecessary. Both of these studies are cited by the IDSA guideline from 2014, which recommends antibiotics for mild‐moderate cellulitis to be limited to antimicrobials effective against ‐hemolytic streptococci and MSSA. The guideline specifically does not recommend routinely treating for MRSA, gram‐negative, or anaerobic organisms citing lack of benefit as well as risks of antibiotic resistance and C difficile infection. A recent study from the University of Utah reported the development of a cellulitis order set, which included a pathway for nonpurulent cellulitis based on the use of cefazolin. These authors reported that the use of the pathway was associated with a 59% decrease in the use of broad‐spectrum antibiotics, a 23% decrease in pharmacy costs, a 13% decrease in total facility cost, with no change in hospital length of stay or readmission rate.[14] One important caveat to the use of clinical pathways is that they are often underused. In the study from the University of Utah, for example, only 55% of eligible patients had the clinical pathway ordered.

WHEN BROAD‐SPECTRUM ANTIBIOTICS ARE RECOMMENDED

The IDSA does recommend empiric broad‐spectrum antibiotics with combination gram‐positive and gram‐negative coverage in several situations, including severe infections in which necrotizing soft tissue infection is suspected, animal bites, immersion injuries, as well as for severely immunocompromised patients or those who have failed limited spectrum antibiotics. Additionally, the IDSA recommends antimicrobials effective against MRSA for purulent infections with systemic signs of inflammation as well as severe nonpurulent infections or those associated with penetrating trauma, injection drug use, and nasal colonization with MRSA (Figure 1).

RECOMMENDATIONS

Our patient has no associated purulence and no abscess and therefore has nonpurulent cellulitis. Based on his mild tachycardia and leukocytosis but intact immune system and lack of suspicion for necrotizing soft tissue infection, he would be classified as moderate‐severity cellulitis by the IDSA. In patients hospitalized with nonpurulent cellulitis who are not severely immunocompromised or severely ill and for whom necrotizing soft tissue infection is not suspected:

1. Antibiotics should be directed at ‐hemolytic streptococci and MSSA, with 1 of the suggested antibiotics by the IDSA including penicillin, ceftriaxone, cefazolin, or clindamycin.
2. Antibiotics effective against MRSA should be limited to situations described by the IDSA.
3. If the cellulitis has not improved within 48 hours, then consider broader‐spectrum antibiotics.
4. Hospitals should strongly consider implementation of a cellulitis pathway based on the IDSA recommendations to improve antibiotic stewardship as well as costs.

The Things We Do for No Reason (TWDFNR) series reviews practices which have become common parts of hospital care but which may provide little value to our patients. Practices reviewed in the TWDFNR series do not represent black and white conclusions or clinical practice standards, but are meant as a starting place for research and active discussions among hospitalists and patients. We invite you to be part of that discussion. https://www.choosingwisely.org/

A previously healthy 6‐year‐old boy presented to the emergency room with 3 days of right lower leg pain and fevers up to 102F. The leg pain had progressed until he refused to walk. The patient and family did not recall any trauma to the leg. In the emergency department, he had a blood culture drawn. Because he had elevated inflammatory markers and a negative x‐ray of his right leg, a magnetic resonance imaging scan of the right leg was obtained that revealed right tibial osteomyelitis. He was taken to the operating room for debridement. After obtaining blood and bone cultures, he was started on intravenous (IV) vancomycin. His blood and surgical cultures grew methicillin‐resistant Staphylococcus aureus, sensitive to clindamycin. Subsequent blood cultures were negative, and his inflammatory markers trended down shortly after starting therapy. As he clinically improved, a peripherally inserted central catheter (PICC) was placed, and he was discharged home to complete a 6‐week course of IV vancomycin.

BACKGROUND

Osteoarticular infections (osteomyelitis and septic arthritis) are common problems in the pediatric population, affecting 1/2000 children annually and accounting for approximately 1% of all pediatric hospitalizations.[1, 2] Osteomyelitis can occur in children of all ages and usually requires hospitalization for diagnosis and initial management. The most common mechanism of infection in children is hematogenous inoculation of the bone during an episode of bacteremia (acute hematogenous osteomyelitis), particularly in young children, due to the highly vascular nature of the developing bone. Long bones, such as the femur, tibia, and humerus, are most commonly involved. Treatment of acute osteomyelitis requires prolonged administration of antimicrobial agents. Inadequately treated osteomyelitis can result in progression to chronic infection and loss of function of the affected bone.[3]

WHY YOU MIGHT THINK PARENTERAL ANTIBIOTICS AT DISCHARGE IS SUPERIOR TO ENTERAL THERAPY

In the United States, a large proportion of children with hematogenous osteomyelitis are discharged from the hospital with long‐term parenteral intravenous antibiotics through a PICC line.[3] The medical community historically favored parenteral therapy for young children with serious bacterial infections given concerns regarding impaired enteral absorption. As a result, children with osteomyelitis were initially stabilized in the hospital and discharged with parenteral therapy through a PICC line to continue or complete care, even when the organism was susceptible to a viable oral alternative such as clindamycin or cephalexin. Recommendations regarding the safety and timing to transition to oral antibiotics have been lacking. There is also extreme variation in practice in route of administration (oral vs prolonged IV therapy) in patients being discharged from the hospital with osteomyelitis.[3, 4] The most recent Infectious Diseases Society of America (IDSA) guidelines do not clearly state when transition to oral antibiotics may be safe. Specifically, they state that if patients are stable and without ongoing bacteremia, they can transition to oral therapy to complete a 4‐ to 6‐week course.[5]

WHY LONG‐TERM PARENTERAL ANTIBIOTICS MAY NOT BE SUPERIOR

The use of PICC lines has increased substantially in recent years. This has led to an increasing awareness of complications associated with PICC lines. As a result, guidelines for the appropriate use of PICC lines have been established in adults by collaborators at the University of Michigan.[6] Mounting evidence has called into question whether longer parenteral therapy is truly a more conservative or safer approach for the treatment of osteomyelitis.[3, 4, 7, 8, 9] Providing antibiotics via a PICC line in both the inpatient and outpatient settings may not be as benign as once accepted and may not improve outcomes in osteomyelitis as expected.

WHEN TO CONSIDER PROLONGED PARENTERAL ANTIBIOTICS

The studies indicating the safe transition to oral antibiotics discussed above all excluded children with certain comorbid conditions. Although this varied from study to study, exclusions were as general in some as not previously healthy, and others were as specific as hematologic malignancies, immunocompromised states, sickle cell disease, malabsorption, and penetrating injuries. Also, although we know blood cultures obtained in children with osteomyelitis are positive in only approximately half of the patients,[19] the studies cited do not contain information for their study populations regarding the duration of bacteremia or endovascular complications, such as septic thrombophlebitis, which are well described in the literature.[20, 21] There are limited data on optimal treatment of children with prolonged bacteremia and endovascular complications. Because studies generally involved previously healthy children and do not specifically address these potential complications, the safety of early oral transition in complicated cases is not clear. The current IDSA and Red Book Committee on Infectious Diseases recommend intravenous therapy for bacteremia and endovascular infections with methicillin‐resistant S aureus.[5, 22] Clinical judgement should be used when treating children with comorbid illnesses who experience persistent bacteremia >48 hours, or who have endovascular complications.

WHAT YOU SHOULD DO INSTEAD

For children with acute hematogenous osteomyelitis who are either culture negative and improve on empiric therapy, or who have culture results (blood or tissue) that are susceptible to a reasonable oral antibiotic agent and who have clinical improvement on initial IV antibiotic therapy, a growing body of evidence indicates that the benefit of early transition to oral antibiotics outweighs the risks of continuing with parenteral therapy. Discharging children on oral antibiotics does not increase their risk of treatment failure but seems to decrease the risk of therapy‐associated complications, including increased healthcare utilization with return visits to the emergency department or the hospital. The possible exceptions to early transition to enteral antibiotics are prolonged bacteremia or endovascular infection, though there are insufficient data in the literature indicating benefits or risks of one administration route over the other.

RECOMMENDATIONS

1. Previously healthy children with acute hematogenous osteomyelitis, without endovascular complications, should be transitioned to enteral antibiotics when they are showing signs of clinical improvement, as defined by: resolution of fever, improving physical exam, ability to take oral medications, and decreasing C‐reactive protein.
2. The choice of oral antibiotics should be based on the organism's antibiotic susceptibility. If cultures are negative and the child has improved on empiric IV therapy, transition to an oral regimen with similar spectrum is acceptable.
3. Patients with acute osteomyelitis should have close follow‐up after discharge from the hospital, within 1 to 2 weeks, to ensure continued improvement on therapy.

CONCLUSION

Early transition to oral antibiotics should be used in children with acute, uncomplicated osteomyelitis. A growing body of evidence shows that early transition to oral antibiotics does not increase the risk of treatment failure and can obviate the need for an outpatient PICC line. Oral antibiotics do not carry the risk of potential complications and complexity that are inherent in outpatient parenteral therapy. The transition to oral therapy should occur prior to discharge from the hospital after clinical improvement. Close follow‐up is essential to ensure successful treatment in children with acute osteomyelitis.

Disclosure: Nothing to report.

Do you think this is a low‐value practice? Is this truly a Thing We Do for No Reason? Share what you do in your practice and join in the conversation online by retweeting it on Twitter (#TWDFNR) and liking it on Facebook. We invite you to propose ideas for other Things We Do for No Reason topics by emailing TWDFNR@hospitalmedicine.org.

The Things We Do for No Reason (TWDFNR) series reviews practices which have become common parts of hospital care but which may provide little value to our patients. Practices reviewed in the TWDFNR series do not represent black and white conclusions or clinical practice standards, but are meant as a starting place for research and active discussions among hospitalists and patients. We invite you to be part of that discussion. https://www.choosingwisely.org/

A previously healthy 6‐year‐old boy presented to the emergency room with 3 days of right lower leg pain and fevers up to 102F. The leg pain had progressed until he refused to walk. The patient and family did not recall any trauma to the leg. In the emergency department, he had a blood culture drawn. Because he had elevated inflammatory markers and a negative x‐ray of his right leg, a magnetic resonance imaging scan of the right leg was obtained that revealed right tibial osteomyelitis. He was taken to the operating room for debridement. After obtaining blood and bone cultures, he was started on intravenous (IV) vancomycin. His blood and surgical cultures grew methicillin‐resistant Staphylococcus aureus, sensitive to clindamycin. Subsequent blood cultures were negative, and his inflammatory markers trended down shortly after starting therapy. As he clinically improved, a peripherally inserted central catheter (PICC) was placed, and he was discharged home to complete a 6‐week course of IV vancomycin.

BACKGROUND

Osteoarticular infections (osteomyelitis and septic arthritis) are common problems in the pediatric population, affecting 1/2000 children annually and accounting for approximately 1% of all pediatric hospitalizations.[1, 2] Osteomyelitis can occur in children of all ages and usually requires hospitalization for diagnosis and initial management. The most common mechanism of infection in children is hematogenous inoculation of the bone during an episode of bacteremia (acute hematogenous osteomyelitis), particularly in young children, due to the highly vascular nature of the developing bone. Long bones, such as the femur, tibia, and humerus, are most commonly involved. Treatment of acute osteomyelitis requires prolonged administration of antimicrobial agents. Inadequately treated osteomyelitis can result in progression to chronic infection and loss of function of the affected bone.[3]

WHY YOU MIGHT THINK PARENTERAL ANTIBIOTICS AT DISCHARGE IS SUPERIOR TO ENTERAL THERAPY

In the United States, a large proportion of children with hematogenous osteomyelitis are discharged from the hospital with long‐term parenteral intravenous antibiotics through a PICC line.[3] The medical community historically favored parenteral therapy for young children with serious bacterial infections given concerns regarding impaired enteral absorption. As a result, children with osteomyelitis were initially stabilized in the hospital and discharged with parenteral therapy through a PICC line to continue or complete care, even when the organism was susceptible to a viable oral alternative such as clindamycin or cephalexin. Recommendations regarding the safety and timing to transition to oral antibiotics have been lacking. There is also extreme variation in practice in route of administration (oral vs prolonged IV therapy) in patients being discharged from the hospital with osteomyelitis.[3, 4] The most recent Infectious Diseases Society of America (IDSA) guidelines do not clearly state when transition to oral antibiotics may be safe. Specifically, they state that if patients are stable and without ongoing bacteremia, they can transition to oral therapy to complete a 4‐ to 6‐week course.[5]

WHY LONG‐TERM PARENTERAL ANTIBIOTICS MAY NOT BE SUPERIOR

The use of PICC lines has increased substantially in recent years. This has led to an increasing awareness of complications associated with PICC lines. As a result, guidelines for the appropriate use of PICC lines have been established in adults by collaborators at the University of Michigan.[6] Mounting evidence has called into question whether longer parenteral therapy is truly a more conservative or safer approach for the treatment of osteomyelitis.[3, 4, 7, 8, 9] Providing antibiotics via a PICC line in both the inpatient and outpatient settings may not be as benign as once accepted and may not improve outcomes in osteomyelitis as expected.

WHEN TO CONSIDER PROLONGED PARENTERAL ANTIBIOTICS

The studies indicating the safe transition to oral antibiotics discussed above all excluded children with certain comorbid conditions. Although this varied from study to study, exclusions were as general in some as not previously healthy, and others were as specific as hematologic malignancies, immunocompromised states, sickle cell disease, malabsorption, and penetrating injuries. Also, although we know blood cultures obtained in children with osteomyelitis are positive in only approximately half of the patients,[19] the studies cited do not contain information for their study populations regarding the duration of bacteremia or endovascular complications, such as septic thrombophlebitis, which are well described in the literature.[20, 21] There are limited data on optimal treatment of children with prolonged bacteremia and endovascular complications. Because studies generally involved previously healthy children and do not specifically address these potential complications, the safety of early oral transition in complicated cases is not clear. The current IDSA and Red Book Committee on Infectious Diseases recommend intravenous therapy for bacteremia and endovascular infections with methicillin‐resistant S aureus.[5, 22] Clinical judgement should be used when treating children with comorbid illnesses who experience persistent bacteremia >48 hours, or who have endovascular complications.

WHAT YOU SHOULD DO INSTEAD

For children with acute hematogenous osteomyelitis who are either culture negative and improve on empiric therapy, or who have culture results (blood or tissue) that are susceptible to a reasonable oral antibiotic agent and who have clinical improvement on initial IV antibiotic therapy, a growing body of evidence indicates that the benefit of early transition to oral antibiotics outweighs the risks of continuing with parenteral therapy. Discharging children on oral antibiotics does not increase their risk of treatment failure but seems to decrease the risk of therapy‐associated complications, including increased healthcare utilization with return visits to the emergency department or the hospital. The possible exceptions to early transition to enteral antibiotics are prolonged bacteremia or endovascular infection, though there are insufficient data in the literature indicating benefits or risks of one administration route over the other.

RECOMMENDATIONS

1. Previously healthy children with acute hematogenous osteomyelitis, without endovascular complications, should be transitioned to enteral antibiotics when they are showing signs of clinical improvement, as defined by: resolution of fever, improving physical exam, ability to take oral medications, and decreasing C‐reactive protein.
2. The choice of oral antibiotics should be based on the organism's antibiotic susceptibility. If cultures are negative and the child has improved on empiric IV therapy, transition to an oral regimen with similar spectrum is acceptable.
3. Patients with acute osteomyelitis should have close follow‐up after discharge from the hospital, within 1 to 2 weeks, to ensure continued improvement on therapy.

CONCLUSION

Early transition to oral antibiotics should be used in children with acute, uncomplicated osteomyelitis. A growing body of evidence shows that early transition to oral antibiotics does not increase the risk of treatment failure and can obviate the need for an outpatient PICC line. Oral antibiotics do not carry the risk of potential complications and complexity that are inherent in outpatient parenteral therapy. The transition to oral therapy should occur prior to discharge from the hospital after clinical improvement. Close follow‐up is essential to ensure successful treatment in children with acute osteomyelitis.

Disclosure: Nothing to report.

Do you think this is a low‐value practice? Is this truly a Thing We Do for No Reason? Share what you do in your practice and join in the conversation online by retweeting it on Twitter (#TWDFNR) and liking it on Facebook. We invite you to propose ideas for other Things We Do for No Reason topics by emailing TWDFNR@hospitalmedicine.org.

The Things We Do for No Reason (TWDFNR) series reviews practices which have become common parts of hospital care but which may provide little value to our patients. Practices reviewed in the TWDFNR series do not represent black and white conclusions or clinical practice standards, but are meant as a starting place for research and active discussions among hospitalists and patients. We invite you to be part of that discussion. https://www.choosingwisely.org/

A previously healthy 6‐year‐old boy presented to the emergency room with 3 days of right lower leg pain and fevers up to 102F. The leg pain had progressed until he refused to walk. The patient and family did not recall any trauma to the leg. In the emergency department, he had a blood culture drawn. Because he had elevated inflammatory markers and a negative x‐ray of his right leg, a magnetic resonance imaging scan of the right leg was obtained that revealed right tibial osteomyelitis. He was taken to the operating room for debridement. After obtaining blood and bone cultures, he was started on intravenous (IV) vancomycin. His blood and surgical cultures grew methicillin‐resistant Staphylococcus aureus, sensitive to clindamycin. Subsequent blood cultures were negative, and his inflammatory markers trended down shortly after starting therapy. As he clinically improved, a peripherally inserted central catheter (PICC) was placed, and he was discharged home to complete a 6‐week course of IV vancomycin.

BACKGROUND

Osteoarticular infections (osteomyelitis and septic arthritis) are common problems in the pediatric population, affecting 1/2000 children annually and accounting for approximately 1% of all pediatric hospitalizations.[1, 2] Osteomyelitis can occur in children of all ages and usually requires hospitalization for diagnosis and initial management. The most common mechanism of infection in children is hematogenous inoculation of the bone during an episode of bacteremia (acute hematogenous osteomyelitis), particularly in young children, due to the highly vascular nature of the developing bone. Long bones, such as the femur, tibia, and humerus, are most commonly involved. Treatment of acute osteomyelitis requires prolonged administration of antimicrobial agents. Inadequately treated osteomyelitis can result in progression to chronic infection and loss of function of the affected bone.[3]

WHY YOU MIGHT THINK PARENTERAL ANTIBIOTICS AT DISCHARGE IS SUPERIOR TO ENTERAL THERAPY

In the United States, a large proportion of children with hematogenous osteomyelitis are discharged from the hospital with long‐term parenteral intravenous antibiotics through a PICC line.[3] The medical community historically favored parenteral therapy for young children with serious bacterial infections given concerns regarding impaired enteral absorption. As a result, children with osteomyelitis were initially stabilized in the hospital and discharged with parenteral therapy through a PICC line to continue or complete care, even when the organism was susceptible to a viable oral alternative such as clindamycin or cephalexin. Recommendations regarding the safety and timing to transition to oral antibiotics have been lacking. There is also extreme variation in practice in route of administration (oral vs prolonged IV therapy) in patients being discharged from the hospital with osteomyelitis.[3, 4] The most recent Infectious Diseases Society of America (IDSA) guidelines do not clearly state when transition to oral antibiotics may be safe. Specifically, they state that if patients are stable and without ongoing bacteremia, they can transition to oral therapy to complete a 4‐ to 6‐week course.[5]

WHY LONG‐TERM PARENTERAL ANTIBIOTICS MAY NOT BE SUPERIOR

The use of PICC lines has increased substantially in recent years. This has led to an increasing awareness of complications associated with PICC lines. As a result, guidelines for the appropriate use of PICC lines have been established in adults by collaborators at the University of Michigan.[6] Mounting evidence has called into question whether longer parenteral therapy is truly a more conservative or safer approach for the treatment of osteomyelitis.[3, 4, 7, 8, 9] Providing antibiotics via a PICC line in both the inpatient and outpatient settings may not be as benign as once accepted and may not improve outcomes in osteomyelitis as expected.

WHEN TO CONSIDER PROLONGED PARENTERAL ANTIBIOTICS

The studies indicating the safe transition to oral antibiotics discussed above all excluded children with certain comorbid conditions. Although this varied from study to study, exclusions were as general in some as not previously healthy, and others were as specific as hematologic malignancies, immunocompromised states, sickle cell disease, malabsorption, and penetrating injuries. Also, although we know blood cultures obtained in children with osteomyelitis are positive in only approximately half of the patients,[19] the studies cited do not contain information for their study populations regarding the duration of bacteremia or endovascular complications, such as septic thrombophlebitis, which are well described in the literature.[20, 21] There are limited data on optimal treatment of children with prolonged bacteremia and endovascular complications. Because studies generally involved previously healthy children and do not specifically address these potential complications, the safety of early oral transition in complicated cases is not clear. The current IDSA and Red Book Committee on Infectious Diseases recommend intravenous therapy for bacteremia and endovascular infections with methicillin‐resistant S aureus.[5, 22] Clinical judgement should be used when treating children with comorbid illnesses who experience persistent bacteremia >48 hours, or who have endovascular complications.

WHAT YOU SHOULD DO INSTEAD

For children with acute hematogenous osteomyelitis who are either culture negative and improve on empiric therapy, or who have culture results (blood or tissue) that are susceptible to a reasonable oral antibiotic agent and who have clinical improvement on initial IV antibiotic therapy, a growing body of evidence indicates that the benefit of early transition to oral antibiotics outweighs the risks of continuing with parenteral therapy. Discharging children on oral antibiotics does not increase their risk of treatment failure but seems to decrease the risk of therapy‐associated complications, including increased healthcare utilization with return visits to the emergency department or the hospital. The possible exceptions to early transition to enteral antibiotics are prolonged bacteremia or endovascular infection, though there are insufficient data in the literature indicating benefits or risks of one administration route over the other.

RECOMMENDATIONS

1. Previously healthy children with acute hematogenous osteomyelitis, without endovascular complications, should be transitioned to enteral antibiotics when they are showing signs of clinical improvement, as defined by: resolution of fever, improving physical exam, ability to take oral medications, and decreasing C‐reactive protein.
2. The choice of oral antibiotics should be based on the organism's antibiotic susceptibility. If cultures are negative and the child has improved on empiric IV therapy, transition to an oral regimen with similar spectrum is acceptable.
3. Patients with acute osteomyelitis should have close follow‐up after discharge from the hospital, within 1 to 2 weeks, to ensure continued improvement on therapy.

CONCLUSION

Early transition to oral antibiotics should be used in children with acute, uncomplicated osteomyelitis. A growing body of evidence shows that early transition to oral antibiotics does not increase the risk of treatment failure and can obviate the need for an outpatient PICC line. Oral antibiotics do not carry the risk of potential complications and complexity that are inherent in outpatient parenteral therapy. The transition to oral therapy should occur prior to discharge from the hospital after clinical improvement. Close follow‐up is essential to ensure successful treatment in children with acute osteomyelitis.

Disclosure: Nothing to report.

Do you think this is a low‐value practice? Is this truly a Thing We Do for No Reason? Share what you do in your practice and join in the conversation online by retweeting it on Twitter (#TWDFNR) and liking it on Facebook. We invite you to propose ideas for other Things We Do for No Reason topics by emailing TWDFNR@hospitalmedicine.org.

The Things We Do for No Reason (TWDFNR) series reviews practices which have become common parts of hospital care but which may provide little value to our patients. Practices reviewed in the TWDFNR series do not represent black and white conclusions or clinical practice standards, but are meant as a starting place for research and active discussions among hospitalists and patients. We invite you to be part of that discussion. https://www.choosingwisely.org/

A 55‐year‐old patient with a history of chronic obstructive pulmonary disease and diabetes mellitus presented to the emergency room with acute shortness of breath and right leg swelling that began 1 week after lumbar disk surgery. The emergency department team decides against ordering a chest CT scan with contrast to evaluate for a possible pulmonary embolism after noting that the patient's allergies include shellfish, which cause urticaria and facial edema. A ventilation‐perfusion scan reveals heterogeneous perfusion defects consistent with an intermediate probability (20%80%) for pulmonary embolism. The treating physicians consider starting the patient on a steroid regimen to prepare him for a CT scan with IV contrast, while presumptively anticoagulating the patient for 24 hours in order for the steroids to provide maximal protective effect before obtaining the scan. Should a history of shellfish allergy affect decision making regarding whether to administer IV contrast?

WHY YOU MIGHT THINK ASKING ABOUT SHELLFISH ALLERGIES BEFORE PERFORMING CONTRAST‐ENHANCED CT SCANS IS HELPFUL

Fish and shellfish contain iodine, and allergic reactions to seafood are quite common, with a prevalence ranging anywhere between 2% and 6% of the population.[1] As a result, patients with suspected shellfish allergies are often told by providers that they are allergic to iodine. In 1 study, nearly 92% of patients presenting to a pediatrics clinic with a suspected seafood or shellfish allergy cited iodine as the culprit.[2] As contrast‐enhanced CT scans utilize a variety of iodine‐based agents, patients are often told to avoid CT scans with iodinated contrast agents or receive corticosteroid/antihistamine premedications prior to undergoing CT scans to mitigate potentially life‐threatening allergic reactions. A survey of radiologists and interventional cardiologists revealed that 65.3% and 88.9%, respectively, asked about seafood or shellfish allergies prior to administering contrast enhanced CT scans, and 34.7% and 50.0%, respectively, stated that they would withhold contrast media or recommend premedication with corticosteroid/antihistamines for patients with seafood or shellfish allergy.[2]

WHY ASKING ABOUT SHELLFISH ALLERGIES BEFORE IV CONTRAST CT SCANS DOES NOT REDUCE THE RISK OF CONTRAST REACTIONS

WHAT YOU SHOULD DO INSTEAD: ASK ABOUT PRIOR CONTRAST REACTIONS BEFORE ADMINISTERING CONTRAST

When a patient presents for a contrast‐enhanced CT scan, patients should be asked if they have experienced reactions to contrast and the severity and type of the associated reactions. Providers and support staff should not ask specifically about shellfish allergies, as they have not been found to be associated with an elevated risk of contrast reactions compared with other allergens. Although all allergies seem to increase the likelihood of having a reaction to contrast, only a history of previous contrast reactions will prompt a change in management. Asking specifically about seafood allergies before performing an IV contrast CT scan is a Thing We Do for No Reason.

RECOMMENDATIONS

1. Before performing contrast‐enhanced CT scans, patients should be asked if they have experienced reactions to IV contrast. There is no reason for providers and support staff to specifically inquire about seafood allergies.
2. Patients with seafood and other food allergies do not require premedication prior to CT scans. Seafood and other food allergies do not represent contraindications to obtaining contrast‐enhanced CT scans and should not prompt a change in management.

The Things We Do for No Reason (TWDFNR) series reviews practices which have become common parts of hospital care but which may provide little value to our patients. Practices reviewed in the TWDFNR series do not represent black and white conclusions or clinical practice standards, but are meant as a starting place for research and active discussions among hospitalists and patients. We invite you to be part of that discussion. https://www.choosingwisely.org/

A 55‐year‐old patient with a history of chronic obstructive pulmonary disease and diabetes mellitus presented to the emergency room with acute shortness of breath and right leg swelling that began 1 week after lumbar disk surgery. The emergency department team decides against ordering a chest CT scan with contrast to evaluate for a possible pulmonary embolism after noting that the patient's allergies include shellfish, which cause urticaria and facial edema. A ventilation‐perfusion scan reveals heterogeneous perfusion defects consistent with an intermediate probability (20%80%) for pulmonary embolism. The treating physicians consider starting the patient on a steroid regimen to prepare him for a CT scan with IV contrast, while presumptively anticoagulating the patient for 24 hours in order for the steroids to provide maximal protective effect before obtaining the scan. Should a history of shellfish allergy affect decision making regarding whether to administer IV contrast?

WHY YOU MIGHT THINK ASKING ABOUT SHELLFISH ALLERGIES BEFORE PERFORMING CONTRAST‐ENHANCED CT SCANS IS HELPFUL

Fish and shellfish contain iodine, and allergic reactions to seafood are quite common, with a prevalence ranging anywhere between 2% and 6% of the population.[1] As a result, patients with suspected shellfish allergies are often told by providers that they are allergic to iodine. In 1 study, nearly 92% of patients presenting to a pediatrics clinic with a suspected seafood or shellfish allergy cited iodine as the culprit.[2] As contrast‐enhanced CT scans utilize a variety of iodine‐based agents, patients are often told to avoid CT scans with iodinated contrast agents or receive corticosteroid/antihistamine premedications prior to undergoing CT scans to mitigate potentially life‐threatening allergic reactions. A survey of radiologists and interventional cardiologists revealed that 65.3% and 88.9%, respectively, asked about seafood or shellfish allergies prior to administering contrast enhanced CT scans, and 34.7% and 50.0%, respectively, stated that they would withhold contrast media or recommend premedication with corticosteroid/antihistamines for patients with seafood or shellfish allergy.[2]

WHY ASKING ABOUT SHELLFISH ALLERGIES BEFORE IV CONTRAST CT SCANS DOES NOT REDUCE THE RISK OF CONTRAST REACTIONS

WHAT YOU SHOULD DO INSTEAD: ASK ABOUT PRIOR CONTRAST REACTIONS BEFORE ADMINISTERING CONTRAST

When a patient presents for a contrast‐enhanced CT scan, patients should be asked if they have experienced reactions to contrast and the severity and type of the associated reactions. Providers and support staff should not ask specifically about shellfish allergies, as they have not been found to be associated with an elevated risk of contrast reactions compared with other allergens. Although all allergies seem to increase the likelihood of having a reaction to contrast, only a history of previous contrast reactions will prompt a change in management. Asking specifically about seafood allergies before performing an IV contrast CT scan is a Thing We Do for No Reason.

RECOMMENDATIONS

1. Before performing contrast‐enhanced CT scans, patients should be asked if they have experienced reactions to IV contrast. There is no reason for providers and support staff to specifically inquire about seafood allergies.
2. Patients with seafood and other food allergies do not require premedication prior to CT scans. Seafood and other food allergies do not represent contraindications to obtaining contrast‐enhanced CT scans and should not prompt a change in management.

The Things We Do for No Reason (TWDFNR) series reviews practices which have become common parts of hospital care but which may provide little value to our patients. Practices reviewed in the TWDFNR series do not represent black and white conclusions or clinical practice standards, but are meant as a starting place for research and active discussions among hospitalists and patients. We invite you to be part of that discussion. https://www.choosingwisely.org/

A 55‐year‐old patient with a history of chronic obstructive pulmonary disease and diabetes mellitus presented to the emergency room with acute shortness of breath and right leg swelling that began 1 week after lumbar disk surgery. The emergency department team decides against ordering a chest CT scan with contrast to evaluate for a possible pulmonary embolism after noting that the patient's allergies include shellfish, which cause urticaria and facial edema. A ventilation‐perfusion scan reveals heterogeneous perfusion defects consistent with an intermediate probability (20%80%) for pulmonary embolism. The treating physicians consider starting the patient on a steroid regimen to prepare him for a CT scan with IV contrast, while presumptively anticoagulating the patient for 24 hours in order for the steroids to provide maximal protective effect before obtaining the scan. Should a history of shellfish allergy affect decision making regarding whether to administer IV contrast?

WHY YOU MIGHT THINK ASKING ABOUT SHELLFISH ALLERGIES BEFORE PERFORMING CONTRAST‐ENHANCED CT SCANS IS HELPFUL

Fish and shellfish contain iodine, and allergic reactions to seafood are quite common, with a prevalence ranging anywhere between 2% and 6% of the population.[1] As a result, patients with suspected shellfish allergies are often told by providers that they are allergic to iodine. In 1 study, nearly 92% of patients presenting to a pediatrics clinic with a suspected seafood or shellfish allergy cited iodine as the culprit.[2] As contrast‐enhanced CT scans utilize a variety of iodine‐based agents, patients are often told to avoid CT scans with iodinated contrast agents or receive corticosteroid/antihistamine premedications prior to undergoing CT scans to mitigate potentially life‐threatening allergic reactions. A survey of radiologists and interventional cardiologists revealed that 65.3% and 88.9%, respectively, asked about seafood or shellfish allergies prior to administering contrast enhanced CT scans, and 34.7% and 50.0%, respectively, stated that they would withhold contrast media or recommend premedication with corticosteroid/antihistamines for patients with seafood or shellfish allergy.[2]

WHY ASKING ABOUT SHELLFISH ALLERGIES BEFORE IV CONTRAST CT SCANS DOES NOT REDUCE THE RISK OF CONTRAST REACTIONS

WHAT YOU SHOULD DO INSTEAD: ASK ABOUT PRIOR CONTRAST REACTIONS BEFORE ADMINISTERING CONTRAST

When a patient presents for a contrast‐enhanced CT scan, patients should be asked if they have experienced reactions to contrast and the severity and type of the associated reactions. Providers and support staff should not ask specifically about shellfish allergies, as they have not been found to be associated with an elevated risk of contrast reactions compared with other allergens. Although all allergies seem to increase the likelihood of having a reaction to contrast, only a history of previous contrast reactions will prompt a change in management. Asking specifically about seafood allergies before performing an IV contrast CT scan is a Thing We Do for No Reason.

RECOMMENDATIONS

1. Before performing contrast‐enhanced CT scans, patients should be asked if they have experienced reactions to IV contrast. There is no reason for providers and support staff to specifically inquire about seafood allergies.
2. Patients with seafood and other food allergies do not require premedication prior to CT scans. Seafood and other food allergies do not represent contraindications to obtaining contrast‐enhanced CT scans and should not prompt a change in management.

The Things We Do for No Reason (TWDFNR) series reviews practices which have become common parts of hospital care but which may provide little value to our patients. Practices reviewed in the TWDFNR series do not represent black and white conclusions or clinical practice standards, but are meant as a starting place for research and active discussions among hospitalists and patients. We invite you to be part of that discussion. https://www.choosingwisely.org/

A 37‐year‐old man presents to the emergency department complaining of acute onset abdominal pain associated with nausea and vomiting. The pain is constant and achy in nature. It is located in the upper abdomen and radiates to the back. The patient reports binge alcohol consumption the day prior to the onset of his pain.

His physical examination is remarkable for fever, with a temperature of 100.6F and epigastric tenderness to palpation without rebound or guarding. He is not hypotensive, and there is no evidence of the Cullen sign or Grey‐Turner sign.

In this patient presenting with acute abdominal pain, is ordering amylase alone, lipase alone, or amylase and lipase together the most high‐value method to evaluate him for acute pancreatitis?

WHY YOU MIGHT THINK AMYLASE TESTING IS HELPFUL

Amylase was one of the earliest, easily measurable laboratory tests that provided a relatively high degree of sensitivity and specificity for identifying patients with acute pancreatitis among those presenting with acute abdominal pain.[1] Since the introduction of amylase, additional tests, including lipase, have been introduced into clinical practice, which offer superior sensitivity and specificity compared to amylase for the diagnosis of acute pancreatitis.[2] However, amylase testing is routinely ordered at many healthcare institutions, with co‐ordering of these tests occurring greater than 90% of the time in some cases.[3, 4] Amylase testing may remain in clinical practice for several reasons including: (1) greater experience with amylase given its earlier introduction into clinical practice, (2) the belief that co‐ordering amylase and lipase provides greater accuracy than either test alone, (3) the notion that pancreatic enzymes provide prognostic information or allow monitoring of clinical progress, and (4) coupling of amylase and lipase in electronic order sets or including the tests together as part of routine lab panels used for the evaluation of abdominal pain.

WHY AMYLASE TESTING OFFERS NO ADDITIONAL VALUE TO LIPASE TESTING

WHAT YOU SHOULD DO INSTEAD: ORDER LIPASE ALONE

In all cases where a patient presents with abdominal pain concerning for acute pancreatitis, we recommend ordering lipase alone rather than either amylase alone or co‐ordering amylase and lipase. In addition, we suggest that healthcare providers do not perform daily measurement of pancreatic enzymes.

When lipase testing is available, amylase testing provides no clinical value in assisting with the diagnosis or management of acute pancreatitis, but comes at a significant cost to patients. At an average charge of $35 per test, amylase testing represents at least $19 million in annual charges to Medicare alone, according to 2013 payment data.[20] Given that the average charge for a lipase test is $41, it is also difficult to argue that amylase testing is significantly less costly than lipase testing. In addition to the direct costs, amylase tests in these settings can result in diagnostic delays or misdiagnosis, imposing additional costs on patients and the health system. For instance, a patient who presents with symptoms of pancreatitis, a positive lipase, and a negative amylase could receive an unnecessary CT scan to further assess for acute pancreatitis, despite already meeting diagnostic criteria based on the symptoms and lipase test alone.

At institutions where amylase and lipase are listed together in common order sets, removing amylase from these order sets may be a simple, durable intervention to reduce amylase testing.[3] Educational interventions aimed at alleviating some of the cognitive factors associated with amylase and lipase co‐ordering described above should also be considered.

RECOMMENDATIONS

1. In patients suspected of having acute pancreatitis, lipase should be ordered alone rather than ordering either amylase alone or amylase and lipase together.
2. Pancreatic enzymes should not be repeated after making the diagnosis of acute pancreatitis, as this practice does not provide additional information that is of clinical utility.

CONCLUSION

In the evaluation of acute pancreatitis, the majority of the evidence suggests that amylase, when compared to lipase, has inferior sensitivity and specificity, adds no additional diagnostic information when co‐ordered, and does not provide additional prognostic information (Table 1). In this setting, many guidelines and thought leaders recommend ordering lipase alone rather than either amylase alone or co‐ordering amylase and lipase. In addition, daily monitoring of pancreatic enzymes is not recommended because it does not help assess clinical progress or severity of illness. As a result, we believe that amylase testing should no longer be ordered for patients with suspected acute pancreatitis. Ordering amylase for the evaluation of abdominal pain is a thing we do for no reason.

Disclosure: Nothing to report.

Do you think this is a low‐value practice? Is this truly a Thing We Do for No Reason? Share what you do in your practice and join in the conversation online by retweeting it on Twitter (#TWDFNR) and liking it on Facebook. We invite you to propose ideas for other Things We Do for No Reason topics by emailing TWDFNR@hospitalmedicine.org.

The Things We Do for No Reason (TWDFNR) series reviews practices which have become common parts of hospital care but which may provide little value to our patients. Practices reviewed in the TWDFNR series do not represent black and white conclusions or clinical practice standards, but are meant as a starting place for research and active discussions among hospitalists and patients. We invite you to be part of that discussion. https://www.choosingwisely.org/

A 37‐year‐old man presents to the emergency department complaining of acute onset abdominal pain associated with nausea and vomiting. The pain is constant and achy in nature. It is located in the upper abdomen and radiates to the back. The patient reports binge alcohol consumption the day prior to the onset of his pain.

His physical examination is remarkable for fever, with a temperature of 100.6F and epigastric tenderness to palpation without rebound or guarding. He is not hypotensive, and there is no evidence of the Cullen sign or Grey‐Turner sign.

In this patient presenting with acute abdominal pain, is ordering amylase alone, lipase alone, or amylase and lipase together the most high‐value method to evaluate him for acute pancreatitis?

WHY YOU MIGHT THINK AMYLASE TESTING IS HELPFUL

Amylase was one of the earliest, easily measurable laboratory tests that provided a relatively high degree of sensitivity and specificity for identifying patients with acute pancreatitis among those presenting with acute abdominal pain.[1] Since the introduction of amylase, additional tests, including lipase, have been introduced into clinical practice, which offer superior sensitivity and specificity compared to amylase for the diagnosis of acute pancreatitis.[2] However, amylase testing is routinely ordered at many healthcare institutions, with co‐ordering of these tests occurring greater than 90% of the time in some cases.[3, 4] Amylase testing may remain in clinical practice for several reasons including: (1) greater experience with amylase given its earlier introduction into clinical practice, (2) the belief that co‐ordering amylase and lipase provides greater accuracy than either test alone, (3) the notion that pancreatic enzymes provide prognostic information or allow monitoring of clinical progress, and (4) coupling of amylase and lipase in electronic order sets or including the tests together as part of routine lab panels used for the evaluation of abdominal pain.

WHY AMYLASE TESTING OFFERS NO ADDITIONAL VALUE TO LIPASE TESTING

WHAT YOU SHOULD DO INSTEAD: ORDER LIPASE ALONE

In all cases where a patient presents with abdominal pain concerning for acute pancreatitis, we recommend ordering lipase alone rather than either amylase alone or co‐ordering amylase and lipase. In addition, we suggest that healthcare providers do not perform daily measurement of pancreatic enzymes.

When lipase testing is available, amylase testing provides no clinical value in assisting with the diagnosis or management of acute pancreatitis, but comes at a significant cost to patients. At an average charge of $35 per test, amylase testing represents at least $19 million in annual charges to Medicare alone, according to 2013 payment data.[20] Given that the average charge for a lipase test is $41, it is also difficult to argue that amylase testing is significantly less costly than lipase testing. In addition to the direct costs, amylase tests in these settings can result in diagnostic delays or misdiagnosis, imposing additional costs on patients and the health system. For instance, a patient who presents with symptoms of pancreatitis, a positive lipase, and a negative amylase could receive an unnecessary CT scan to further assess for acute pancreatitis, despite already meeting diagnostic criteria based on the symptoms and lipase test alone.

At institutions where amylase and lipase are listed together in common order sets, removing amylase from these order sets may be a simple, durable intervention to reduce amylase testing.[3] Educational interventions aimed at alleviating some of the cognitive factors associated with amylase and lipase co‐ordering described above should also be considered.

RECOMMENDATIONS

1. In patients suspected of having acute pancreatitis, lipase should be ordered alone rather than ordering either amylase alone or amylase and lipase together.
2. Pancreatic enzymes should not be repeated after making the diagnosis of acute pancreatitis, as this practice does not provide additional information that is of clinical utility.

CONCLUSION

In the evaluation of acute pancreatitis, the majority of the evidence suggests that amylase, when compared to lipase, has inferior sensitivity and specificity, adds no additional diagnostic information when co‐ordered, and does not provide additional prognostic information (Table 1). In this setting, many guidelines and thought leaders recommend ordering lipase alone rather than either amylase alone or co‐ordering amylase and lipase. In addition, daily monitoring of pancreatic enzymes is not recommended because it does not help assess clinical progress or severity of illness. As a result, we believe that amylase testing should no longer be ordered for patients with suspected acute pancreatitis. Ordering amylase for the evaluation of abdominal pain is a thing we do for no reason.

Disclosure: Nothing to report.

Do you think this is a low‐value practice? Is this truly a Thing We Do for No Reason? Share what you do in your practice and join in the conversation online by retweeting it on Twitter (#TWDFNR) and liking it on Facebook. We invite you to propose ideas for other Things We Do for No Reason topics by emailing TWDFNR@hospitalmedicine.org.

The Things We Do for No Reason (TWDFNR) series reviews practices which have become common parts of hospital care but which may provide little value to our patients. Practices reviewed in the TWDFNR series do not represent black and white conclusions or clinical practice standards, but are meant as a starting place for research and active discussions among hospitalists and patients. We invite you to be part of that discussion. https://www.choosingwisely.org/

A 37‐year‐old man presents to the emergency department complaining of acute onset abdominal pain associated with nausea and vomiting. The pain is constant and achy in nature. It is located in the upper abdomen and radiates to the back. The patient reports binge alcohol consumption the day prior to the onset of his pain.

His physical examination is remarkable for fever, with a temperature of 100.6F and epigastric tenderness to palpation without rebound or guarding. He is not hypotensive, and there is no evidence of the Cullen sign or Grey‐Turner sign.

In this patient presenting with acute abdominal pain, is ordering amylase alone, lipase alone, or amylase and lipase together the most high‐value method to evaluate him for acute pancreatitis?

WHY YOU MIGHT THINK AMYLASE TESTING IS HELPFUL

Amylase was one of the earliest, easily measurable laboratory tests that provided a relatively high degree of sensitivity and specificity for identifying patients with acute pancreatitis among those presenting with acute abdominal pain.[1] Since the introduction of amylase, additional tests, including lipase, have been introduced into clinical practice, which offer superior sensitivity and specificity compared to amylase for the diagnosis of acute pancreatitis.[2] However, amylase testing is routinely ordered at many healthcare institutions, with co‐ordering of these tests occurring greater than 90% of the time in some cases.[3, 4] Amylase testing may remain in clinical practice for several reasons including: (1) greater experience with amylase given its earlier introduction into clinical practice, (2) the belief that co‐ordering amylase and lipase provides greater accuracy than either test alone, (3) the notion that pancreatic enzymes provide prognostic information or allow monitoring of clinical progress, and (4) coupling of amylase and lipase in electronic order sets or including the tests together as part of routine lab panels used for the evaluation of abdominal pain.

WHY AMYLASE TESTING OFFERS NO ADDITIONAL VALUE TO LIPASE TESTING

WHAT YOU SHOULD DO INSTEAD: ORDER LIPASE ALONE

In all cases where a patient presents with abdominal pain concerning for acute pancreatitis, we recommend ordering lipase alone rather than either amylase alone or co‐ordering amylase and lipase. In addition, we suggest that healthcare providers do not perform daily measurement of pancreatic enzymes.

When lipase testing is available, amylase testing provides no clinical value in assisting with the diagnosis or management of acute pancreatitis, but comes at a significant cost to patients. At an average charge of $35 per test, amylase testing represents at least $19 million in annual charges to Medicare alone, according to 2013 payment data.[20] Given that the average charge for a lipase test is $41, it is also difficult to argue that amylase testing is significantly less costly than lipase testing. In addition to the direct costs, amylase tests in these settings can result in diagnostic delays or misdiagnosis, imposing additional costs on patients and the health system. For instance, a patient who presents with symptoms of pancreatitis, a positive lipase, and a negative amylase could receive an unnecessary CT scan to further assess for acute pancreatitis, despite already meeting diagnostic criteria based on the symptoms and lipase test alone.

At institutions where amylase and lipase are listed together in common order sets, removing amylase from these order sets may be a simple, durable intervention to reduce amylase testing.[3] Educational interventions aimed at alleviating some of the cognitive factors associated with amylase and lipase co‐ordering described above should also be considered.

RECOMMENDATIONS

1. In patients suspected of having acute pancreatitis, lipase should be ordered alone rather than ordering either amylase alone or amylase and lipase together.
2. Pancreatic enzymes should not be repeated after making the diagnosis of acute pancreatitis, as this practice does not provide additional information that is of clinical utility.

CONCLUSION

In the evaluation of acute pancreatitis, the majority of the evidence suggests that amylase, when compared to lipase, has inferior sensitivity and specificity, adds no additional diagnostic information when co‐ordered, and does not provide additional prognostic information (Table 1). In this setting, many guidelines and thought leaders recommend ordering lipase alone rather than either amylase alone or co‐ordering amylase and lipase. In addition, daily monitoring of pancreatic enzymes is not recommended because it does not help assess clinical progress or severity of illness. As a result, we believe that amylase testing should no longer be ordered for patients with suspected acute pancreatitis. Ordering amylase for the evaluation of abdominal pain is a thing we do for no reason.

Disclosure: Nothing to report.

Do you think this is a low‐value practice? Is this truly a Thing We Do for No Reason? Share what you do in your practice and join in the conversation online by retweeting it on Twitter (#TWDFNR) and liking it on Facebook. We invite you to propose ideas for other Things We Do for No Reason topics by emailing TWDFNR@hospitalmedicine.org.

CASE PRESENTATION

A 66‐year‐old man with a history of hypertension is hospitalized for a transient loss of consciousness while shopping at a farmers market with his wife on a hot summer day. He recalls feeling lightheaded seconds before he lost consciousness. He had no chest pain, diaphoresis, dyspnea, shaking movements, slurred speech, or head trauma. He felt mildly fatigued following the episode, but has since returned to his baseline. Physical examination, including a thorough cardiac and neurological examination, is normal. The hospitalist ponders whether to order a carotid artery ultrasound as part of a syncope evaluation.

BRIEF OVERVIEW

Syncope is defined as a rapid onset loss of consciousness of short duration as a result of global cerebral hypoperfusion with loss of postural tone, which is followed by spontaneous and complete recovery.[1] This definition describes syncope as a symptom rather than a disease. The challenge for providers is to determine the etiology of the syncope along with its attendant risk of morbidity and mortality. Given the wide variety of etiologies and concern over potentially missing an important etiology, diagnostic testing can become elaborate, expensive, and frequently low yield.

In the adult population, it is believed that approximately 35% of individuals will experience a syncopal episode in their lifetime.[2] As a result, syncope accounts for 1% to 3% of all emergency department visits and 1% to 6% of hospital admissions from emergency departments in the United States.[3, 4] The incidence and rate of hospitalization increase with age, as does the risk of mortality.[5, 6] There are 3 main types of syncope: cardiac, neurocardiogenic (vasovagal), and orthostatic. The presence of associated signs or symptoms with the syncope helps to differentiate the type and complexity of the syncope, while helping guide the diagnostic evaluation. Simple syncope is defined as the absence of focal neurological deficits or other signs or symptoms suggestive of transient ischemic attack (TIA) or cerebrovascular accident (CVA).[7] A differential diagnosis for a transient loss of consciousness that includes TIA and CVA will prompt a very different evaluation.

WHY YOU MIGHT THINK ORDERING CAROTID ARTERY ULTRASOUNDS FOR SYNCOPE EVALUATION ARE HELPFUL

Carotid artery ultrasounds are used to assess the extracranial carotid arteries for the presence of stenosis and to determine the direction of blood flow. The use of carotid artery ultrasound as a diagnostic tool in the evaluation of syncope can be traced to multiple articles from the 1980s. These articles noted the utility of screening patients with dizziness, lightheadedness, or syncope using carotid artery ultrasound due to possible decreased flow in the carotid artery circulation affecting cerebral perfusion.[8, 9] An association was noted between these symptoms and the presence of carotid artery stenosis. Further, a 1997 position paper from the American College of Physicians recommended that carotid artery or transcranial ultrasonography be reserved for syncope patients with carotid artery bruits or a history of neurovascular signs or symptoms.[10] More recent studies reveal carotid artery ultrasounds are still being performed regularly in syncope patients. In 2 studies evaluating syncope in the elderly, approximately 13% to 16% of syncope patients had a carotid artery ultrasound performed in an effort to identify an etiology.[7, 11]

Additionally, practitioners sometimes choose to perform carotid artery ultrasound in the evaluation of carotid sinus hypersensitivity. The carotid artery ultrasound can assess for the presence of stenosis or atheroma prior to performing carotid sinus massage, although the rate of persistent neurological complications from carotid sinus massage is estimated to occur in 1:1000 patients.[12]

WHY THERE IS NO REASON TO ORDER CAROTID ARTERY ULTRASOUNDS FOR THE EVALUATION OF SIMPLE SYNCOPE

Carotid artery ultrasounds are unlikely to determine the etiology of the syncope. We should expect a high‐value test to reveal an etiology for the syncope episodes at a significant rate. In the 2009 study by Mendu et al. at YaleNew Haven Hospital, 267 ultrasounds were performed on 2106 syncope admissions of high‐risk elderly patients (1920 total patients).[11] Of the 267 ultrasounds, only 2 of the tests (0.8%) helped to determine an etiology. Although 46% of the ultrasounds had abnormal findings, the measuring stick for these studies should be whether they uncover the etiology for syncope, not whether they find other unrelated vascular disease. In contrast, performing postural blood pressures helped to determine an etiology 15% to 21% of the time, depending on the criteria used to define an abnormal drop in postural blood pressures.

Similarly, in the 2005 study by Schnipper et al. at Massachusetts General Hospital, only 140 of 4199 adult patients (3.3%) who presented as either inpatients or outpatients for syncope or presyncope were referred for neurovascular testing.[13] Carotid artery ultrasound was performed in 109 of these patients, and the study neurologist could invoke cerebrovascular lesions as potential factors for syncope in only 2 patients, both of whom had syncope and focal neurologic symptoms. Moreover, both of the patients had severe cardiovascular disease (severe ischemic cardiomyopathy with complete heart block following coronary bypass surgery in 1 and aortic stenosis with decreased left ventricular ejection fraction in the other). It is quite possible that the ultrasounds did not find the etiology for any of the 140 high‐risk patients with syncope in the study.

In addition, the 2014 study by Scott et al. at Brigham and Women's Hospital analyzed carotid artery duplex ultrasounds performed on 313 inpatients and outpatients with syncope over a 5‐year period, excluding those with focal neurological deficits or carotid bruits.[7] Although 48 of the 313 patients (15.4%) were diagnosed with carotid stenosis of greater than 50%, the carotid artery ultrasound did not reveal a causal diagnosis in any patients. On the other hand, 7 patients had a change in medical management, and 1 patient underwent carotid endarterectomy following the carotid artery ultrasound, which was incidental to what prompted the evaluation.

Mendu et al. calculated the cost per test affecting the diagnosis or management of syncope (although diagnosis is the only important parameter). The cost per test was calculated as the charge per test multiplied by the cost‐to‐charge ratio of 0.34, based on the 2007 YaleNew Haven Hospital cost‐to‐charge ratio.[11] For carotid artery ultrasound, the cost per test was $19,580 to affect diagnosis or management as compared to $23 to $33 for postural blood pressures. Combining these findings with the results from the Schnipper et al. and Scott et al. articles, where carotid artery ultrasounds may not have found the cause of syncope in any of the patients, it seems clear that obtaining a carotid artery ultrasound in the evaluation of simple syncope is a low‐value proposition.

Many low‐value tests, like carotid artery ultrasounds, suffer from both upfront costs, as calculated in the Mendu et al. study, and downstream costs triggered by the testing itself. Performing carotid artery ultrasounds in elderly high‐risk syncope populations is likely to reveal asymptomatic carotid artery vascular disease, which may lead to more unwarranted testing or treatments in light of the initial indication for the test. In the Mendu et al. article, 122 (46%) of the 267 carotid artery ultrasounds performed on elderly patients admitted with syncope were abnormal. Abnormal findings were defined as any abnormality that was not seen on prior testing as written in the test reports. Similarly, Schnipper et al. found that 40% of the 140 highly selected patients had mild‐to‐severe carotid vascular disease.

National guideline recommendations are aligned with these findings. The National Institute for Health and Clinical Excellence Guideline for the Management of Transient Loss of Consciousness does not include carotid artery ultrasound in the summary of clinical recommendations.[14] Furthermore, the American Academy of Neurology Choosing Wisely campaign's recommendation 2 is: Do not perform imaging of the carotid arteries for simple syncope without other neurologic symptoms.[15]

WHAT YOU SHOULD DO INSTEAD: CHECK POSTURAL BLOOD PRESSURES

As is true for most of medicine, greater focus should be paid to the history and physical examination during the initial evaluation of the patient with syncope. Take great care to determine which patients have a history or symptoms concerning for neurologic or cardiac etiologies. Use this information to guide further diagnostic testing. Additionally, orthostatic testing is too often overlooked as an important diagnostic study. As described in the Mendu et al. study, orthostatic testing is inexpensive and effective, helping to determine an etiology 15% to 21% of the time. Carotid artery ultrasounds should be reserved for those patients with transient or permanent focal neurological symptoms.

RECOMMENDATIONS

1. In patients suspected of syncope in the absence of other neurologic symptoms, carotid artery ultrasound should not be included in the diagnostic evaluation.
2. Utilize postural blood pressures in the initial evaluation of syncope as an inexpensive and high‐value component of the physical examination.
3. For patients with acute neurological findings in the setting of possible syncope, evaluate the patient for stroke.
4. Use the history and physical examination to guide further evaluation.

CONCLUSION

Carotid artery ultrasounds should not be used to evaluate the cause of syncope in an effort to find incident symptomatic carotid vascular disease. Carotid artery ultrasounds rarely help determine the etiology of the syncopal episode and are more likely to find asymptomatic carotid vascular disease in the elderly population. The identification of carotid vascular disease can lead to further inappropriate testing and treatments unrelated to the indication for testing.

CASE PRESENTATION

A 66‐year‐old man with a history of hypertension is hospitalized for a transient loss of consciousness while shopping at a farmers market with his wife on a hot summer day. He recalls feeling lightheaded seconds before he lost consciousness. He had no chest pain, diaphoresis, dyspnea, shaking movements, slurred speech, or head trauma. He felt mildly fatigued following the episode, but has since returned to his baseline. Physical examination, including a thorough cardiac and neurological examination, is normal. The hospitalist ponders whether to order a carotid artery ultrasound as part of a syncope evaluation.

BRIEF OVERVIEW

Syncope is defined as a rapid onset loss of consciousness of short duration as a result of global cerebral hypoperfusion with loss of postural tone, which is followed by spontaneous and complete recovery.[1] This definition describes syncope as a symptom rather than a disease. The challenge for providers is to determine the etiology of the syncope along with its attendant risk of morbidity and mortality. Given the wide variety of etiologies and concern over potentially missing an important etiology, diagnostic testing can become elaborate, expensive, and frequently low yield.

In the adult population, it is believed that approximately 35% of individuals will experience a syncopal episode in their lifetime.[2] As a result, syncope accounts for 1% to 3% of all emergency department visits and 1% to 6% of hospital admissions from emergency departments in the United States.[3, 4] The incidence and rate of hospitalization increase with age, as does the risk of mortality.[5, 6] There are 3 main types of syncope: cardiac, neurocardiogenic (vasovagal), and orthostatic. The presence of associated signs or symptoms with the syncope helps to differentiate the type and complexity of the syncope, while helping guide the diagnostic evaluation. Simple syncope is defined as the absence of focal neurological deficits or other signs or symptoms suggestive of transient ischemic attack (TIA) or cerebrovascular accident (CVA).[7] A differential diagnosis for a transient loss of consciousness that includes TIA and CVA will prompt a very different evaluation.

WHY YOU MIGHT THINK ORDERING CAROTID ARTERY ULTRASOUNDS FOR SYNCOPE EVALUATION ARE HELPFUL

Carotid artery ultrasounds are used to assess the extracranial carotid arteries for the presence of stenosis and to determine the direction of blood flow. The use of carotid artery ultrasound as a diagnostic tool in the evaluation of syncope can be traced to multiple articles from the 1980s. These articles noted the utility of screening patients with dizziness, lightheadedness, or syncope using carotid artery ultrasound due to possible decreased flow in the carotid artery circulation affecting cerebral perfusion.[8, 9] An association was noted between these symptoms and the presence of carotid artery stenosis. Further, a 1997 position paper from the American College of Physicians recommended that carotid artery or transcranial ultrasonography be reserved for syncope patients with carotid artery bruits or a history of neurovascular signs or symptoms.[10] More recent studies reveal carotid artery ultrasounds are still being performed regularly in syncope patients. In 2 studies evaluating syncope in the elderly, approximately 13% to 16% of syncope patients had a carotid artery ultrasound performed in an effort to identify an etiology.[7, 11]

Additionally, practitioners sometimes choose to perform carotid artery ultrasound in the evaluation of carotid sinus hypersensitivity. The carotid artery ultrasound can assess for the presence of stenosis or atheroma prior to performing carotid sinus massage, although the rate of persistent neurological complications from carotid sinus massage is estimated to occur in 1:1000 patients.[12]

WHY THERE IS NO REASON TO ORDER CAROTID ARTERY ULTRASOUNDS FOR THE EVALUATION OF SIMPLE SYNCOPE

Carotid artery ultrasounds are unlikely to determine the etiology of the syncope. We should expect a high‐value test to reveal an etiology for the syncope episodes at a significant rate. In the 2009 study by Mendu et al. at YaleNew Haven Hospital, 267 ultrasounds were performed on 2106 syncope admissions of high‐risk elderly patients (1920 total patients).[11] Of the 267 ultrasounds, only 2 of the tests (0.8%) helped to determine an etiology. Although 46% of the ultrasounds had abnormal findings, the measuring stick for these studies should be whether they uncover the etiology for syncope, not whether they find other unrelated vascular disease. In contrast, performing postural blood pressures helped to determine an etiology 15% to 21% of the time, depending on the criteria used to define an abnormal drop in postural blood pressures.

Similarly, in the 2005 study by Schnipper et al. at Massachusetts General Hospital, only 140 of 4199 adult patients (3.3%) who presented as either inpatients or outpatients for syncope or presyncope were referred for neurovascular testing.[13] Carotid artery ultrasound was performed in 109 of these patients, and the study neurologist could invoke cerebrovascular lesions as potential factors for syncope in only 2 patients, both of whom had syncope and focal neurologic symptoms. Moreover, both of the patients had severe cardiovascular disease (severe ischemic cardiomyopathy with complete heart block following coronary bypass surgery in 1 and aortic stenosis with decreased left ventricular ejection fraction in the other). It is quite possible that the ultrasounds did not find the etiology for any of the 140 high‐risk patients with syncope in the study.

In addition, the 2014 study by Scott et al. at Brigham and Women's Hospital analyzed carotid artery duplex ultrasounds performed on 313 inpatients and outpatients with syncope over a 5‐year period, excluding those with focal neurological deficits or carotid bruits.[7] Although 48 of the 313 patients (15.4%) were diagnosed with carotid stenosis of greater than 50%, the carotid artery ultrasound did not reveal a causal diagnosis in any patients. On the other hand, 7 patients had a change in medical management, and 1 patient underwent carotid endarterectomy following the carotid artery ultrasound, which was incidental to what prompted the evaluation.

Mendu et al. calculated the cost per test affecting the diagnosis or management of syncope (although diagnosis is the only important parameter). The cost per test was calculated as the charge per test multiplied by the cost‐to‐charge ratio of 0.34, based on the 2007 YaleNew Haven Hospital cost‐to‐charge ratio.[11] For carotid artery ultrasound, the cost per test was $19,580 to affect diagnosis or management as compared to $23 to $33 for postural blood pressures. Combining these findings with the results from the Schnipper et al. and Scott et al. articles, where carotid artery ultrasounds may not have found the cause of syncope in any of the patients, it seems clear that obtaining a carotid artery ultrasound in the evaluation of simple syncope is a low‐value proposition.

Many low‐value tests, like carotid artery ultrasounds, suffer from both upfront costs, as calculated in the Mendu et al. study, and downstream costs triggered by the testing itself. Performing carotid artery ultrasounds in elderly high‐risk syncope populations is likely to reveal asymptomatic carotid artery vascular disease, which may lead to more unwarranted testing or treatments in light of the initial indication for the test. In the Mendu et al. article, 122 (46%) of the 267 carotid artery ultrasounds performed on elderly patients admitted with syncope were abnormal. Abnormal findings were defined as any abnormality that was not seen on prior testing as written in the test reports. Similarly, Schnipper et al. found that 40% of the 140 highly selected patients had mild‐to‐severe carotid vascular disease.

National guideline recommendations are aligned with these findings. The National Institute for Health and Clinical Excellence Guideline for the Management of Transient Loss of Consciousness does not include carotid artery ultrasound in the summary of clinical recommendations.[14] Furthermore, the American Academy of Neurology Choosing Wisely campaign's recommendation 2 is: Do not perform imaging of the carotid arteries for simple syncope without other neurologic symptoms.[15]

WHAT YOU SHOULD DO INSTEAD: CHECK POSTURAL BLOOD PRESSURES

As is true for most of medicine, greater focus should be paid to the history and physical examination during the initial evaluation of the patient with syncope. Take great care to determine which patients have a history or symptoms concerning for neurologic or cardiac etiologies. Use this information to guide further diagnostic testing. Additionally, orthostatic testing is too often overlooked as an important diagnostic study. As described in the Mendu et al. study, orthostatic testing is inexpensive and effective, helping to determine an etiology 15% to 21% of the time. Carotid artery ultrasounds should be reserved for those patients with transient or permanent focal neurological symptoms.

RECOMMENDATIONS

1. In patients suspected of syncope in the absence of other neurologic symptoms, carotid artery ultrasound should not be included in the diagnostic evaluation.
2. Utilize postural blood pressures in the initial evaluation of syncope as an inexpensive and high‐value component of the physical examination.
3. For patients with acute neurological findings in the setting of possible syncope, evaluate the patient for stroke.
4. Use the history and physical examination to guide further evaluation.

CONCLUSION

Carotid artery ultrasounds should not be used to evaluate the cause of syncope in an effort to find incident symptomatic carotid vascular disease. Carotid artery ultrasounds rarely help determine the etiology of the syncopal episode and are more likely to find asymptomatic carotid vascular disease in the elderly population. The identification of carotid vascular disease can lead to further inappropriate testing and treatments unrelated to the indication for testing.

CASE PRESENTATION

A 66‐year‐old man with a history of hypertension is hospitalized for a transient loss of consciousness while shopping at a farmers market with his wife on a hot summer day. He recalls feeling lightheaded seconds before he lost consciousness. He had no chest pain, diaphoresis, dyspnea, shaking movements, slurred speech, or head trauma. He felt mildly fatigued following the episode, but has since returned to his baseline. Physical examination, including a thorough cardiac and neurological examination, is normal. The hospitalist ponders whether to order a carotid artery ultrasound as part of a syncope evaluation.

BRIEF OVERVIEW

Syncope is defined as a rapid onset loss of consciousness of short duration as a result of global cerebral hypoperfusion with loss of postural tone, which is followed by spontaneous and complete recovery.[1] This definition describes syncope as a symptom rather than a disease. The challenge for providers is to determine the etiology of the syncope along with its attendant risk of morbidity and mortality. Given the wide variety of etiologies and concern over potentially missing an important etiology, diagnostic testing can become elaborate, expensive, and frequently low yield.

In the adult population, it is believed that approximately 35% of individuals will experience a syncopal episode in their lifetime.[2] As a result, syncope accounts for 1% to 3% of all emergency department visits and 1% to 6% of hospital admissions from emergency departments in the United States.[3, 4] The incidence and rate of hospitalization increase with age, as does the risk of mortality.[5, 6] There are 3 main types of syncope: cardiac, neurocardiogenic (vasovagal), and orthostatic. The presence of associated signs or symptoms with the syncope helps to differentiate the type and complexity of the syncope, while helping guide the diagnostic evaluation. Simple syncope is defined as the absence of focal neurological deficits or other signs or symptoms suggestive of transient ischemic attack (TIA) or cerebrovascular accident (CVA).[7] A differential diagnosis for a transient loss of consciousness that includes TIA and CVA will prompt a very different evaluation.

WHY YOU MIGHT THINK ORDERING CAROTID ARTERY ULTRASOUNDS FOR SYNCOPE EVALUATION ARE HELPFUL

Carotid artery ultrasounds are used to assess the extracranial carotid arteries for the presence of stenosis and to determine the direction of blood flow. The use of carotid artery ultrasound as a diagnostic tool in the evaluation of syncope can be traced to multiple articles from the 1980s. These articles noted the utility of screening patients with dizziness, lightheadedness, or syncope using carotid artery ultrasound due to possible decreased flow in the carotid artery circulation affecting cerebral perfusion.[8, 9] An association was noted between these symptoms and the presence of carotid artery stenosis. Further, a 1997 position paper from the American College of Physicians recommended that carotid artery or transcranial ultrasonography be reserved for syncope patients with carotid artery bruits or a history of neurovascular signs or symptoms.[10] More recent studies reveal carotid artery ultrasounds are still being performed regularly in syncope patients. In 2 studies evaluating syncope in the elderly, approximately 13% to 16% of syncope patients had a carotid artery ultrasound performed in an effort to identify an etiology.[7, 11]

Additionally, practitioners sometimes choose to perform carotid artery ultrasound in the evaluation of carotid sinus hypersensitivity. The carotid artery ultrasound can assess for the presence of stenosis or atheroma prior to performing carotid sinus massage, although the rate of persistent neurological complications from carotid sinus massage is estimated to occur in 1:1000 patients.[12]

WHY THERE IS NO REASON TO ORDER CAROTID ARTERY ULTRASOUNDS FOR THE EVALUATION OF SIMPLE SYNCOPE

Carotid artery ultrasounds are unlikely to determine the etiology of the syncope. We should expect a high‐value test to reveal an etiology for the syncope episodes at a significant rate. In the 2009 study by Mendu et al. at YaleNew Haven Hospital, 267 ultrasounds were performed on 2106 syncope admissions of high‐risk elderly patients (1920 total patients).[11] Of the 267 ultrasounds, only 2 of the tests (0.8%) helped to determine an etiology. Although 46% of the ultrasounds had abnormal findings, the measuring stick for these studies should be whether they uncover the etiology for syncope, not whether they find other unrelated vascular disease. In contrast, performing postural blood pressures helped to determine an etiology 15% to 21% of the time, depending on the criteria used to define an abnormal drop in postural blood pressures.

Similarly, in the 2005 study by Schnipper et al. at Massachusetts General Hospital, only 140 of 4199 adult patients (3.3%) who presented as either inpatients or outpatients for syncope or presyncope were referred for neurovascular testing.[13] Carotid artery ultrasound was performed in 109 of these patients, and the study neurologist could invoke cerebrovascular lesions as potential factors for syncope in only 2 patients, both of whom had syncope and focal neurologic symptoms. Moreover, both of the patients had severe cardiovascular disease (severe ischemic cardiomyopathy with complete heart block following coronary bypass surgery in 1 and aortic stenosis with decreased left ventricular ejection fraction in the other). It is quite possible that the ultrasounds did not find the etiology for any of the 140 high‐risk patients with syncope in the study.

In addition, the 2014 study by Scott et al. at Brigham and Women's Hospital analyzed carotid artery duplex ultrasounds performed on 313 inpatients and outpatients with syncope over a 5‐year period, excluding those with focal neurological deficits or carotid bruits.[7] Although 48 of the 313 patients (15.4%) were diagnosed with carotid stenosis of greater than 50%, the carotid artery ultrasound did not reveal a causal diagnosis in any patients. On the other hand, 7 patients had a change in medical management, and 1 patient underwent carotid endarterectomy following the carotid artery ultrasound, which was incidental to what prompted the evaluation.

Mendu et al. calculated the cost per test affecting the diagnosis or management of syncope (although diagnosis is the only important parameter). The cost per test was calculated as the charge per test multiplied by the cost‐to‐charge ratio of 0.34, based on the 2007 YaleNew Haven Hospital cost‐to‐charge ratio.[11] For carotid artery ultrasound, the cost per test was $19,580 to affect diagnosis or management as compared to $23 to $33 for postural blood pressures. Combining these findings with the results from the Schnipper et al. and Scott et al. articles, where carotid artery ultrasounds may not have found the cause of syncope in any of the patients, it seems clear that obtaining a carotid artery ultrasound in the evaluation of simple syncope is a low‐value proposition.

Many low‐value tests, like carotid artery ultrasounds, suffer from both upfront costs, as calculated in the Mendu et al. study, and downstream costs triggered by the testing itself. Performing carotid artery ultrasounds in elderly high‐risk syncope populations is likely to reveal asymptomatic carotid artery vascular disease, which may lead to more unwarranted testing or treatments in light of the initial indication for the test. In the Mendu et al. article, 122 (46%) of the 267 carotid artery ultrasounds performed on elderly patients admitted with syncope were abnormal. Abnormal findings were defined as any abnormality that was not seen on prior testing as written in the test reports. Similarly, Schnipper et al. found that 40% of the 140 highly selected patients had mild‐to‐severe carotid vascular disease.

National guideline recommendations are aligned with these findings. The National Institute for Health and Clinical Excellence Guideline for the Management of Transient Loss of Consciousness does not include carotid artery ultrasound in the summary of clinical recommendations.[14] Furthermore, the American Academy of Neurology Choosing Wisely campaign's recommendation 2 is: Do not perform imaging of the carotid arteries for simple syncope without other neurologic symptoms.[15]

WHAT YOU SHOULD DO INSTEAD: CHECK POSTURAL BLOOD PRESSURES

As is true for most of medicine, greater focus should be paid to the history and physical examination during the initial evaluation of the patient with syncope. Take great care to determine which patients have a history or symptoms concerning for neurologic or cardiac etiologies. Use this information to guide further diagnostic testing. Additionally, orthostatic testing is too often overlooked as an important diagnostic study. As described in the Mendu et al. study, orthostatic testing is inexpensive and effective, helping to determine an etiology 15% to 21% of the time. Carotid artery ultrasounds should be reserved for those patients with transient or permanent focal neurological symptoms.

RECOMMENDATIONS

1. In patients suspected of syncope in the absence of other neurologic symptoms, carotid artery ultrasound should not be included in the diagnostic evaluation.
2. Utilize postural blood pressures in the initial evaluation of syncope as an inexpensive and high‐value component of the physical examination.
3. For patients with acute neurological findings in the setting of possible syncope, evaluate the patient for stroke.
4. Use the history and physical examination to guide further evaluation.

CONCLUSION

Carotid artery ultrasounds should not be used to evaluate the cause of syncope in an effort to find incident symptomatic carotid vascular disease. Carotid artery ultrasounds rarely help determine the etiology of the syncopal episode and are more likely to find asymptomatic carotid vascular disease in the elderly population. The identification of carotid vascular disease can lead to further inappropriate testing and treatments unrelated to the indication for testing.

The Things We Do for No Reason (TWDFNR) series reviews practices which have become common parts of hospital care but which may provide little value to our patients. Practices reviewed in the TWDFNR series do not represent black and white conclusions or clinical practice standards, but are meant as a starting place for research and active discussions among hospitalists and patients. We invite you to be part of that discussion. https://www.choosingwisely.org/

A 70‐year‐old woman with a history of diabetes mellitus type 2 and hypertension was admitted with abdominal pain following 2 days of nausea and diarrhea. Initial laboratory studies revealed blood urea nitrogen (BUN) 25 mg/dL and serum creatinine 1.3 mg/dL. Computed tomography of the abdomen and pelvis with nonionic, low osmolar intravenous and oral contrast demonstrated acute diverticulitis with an associated small abscess. She was administered intravenous 0.9% sodium chloride solution and antibiotics. Blood pressure on admission was 92/55 mm Hg, and 24 hours later, her BUN and serum creatinine increased to 33 mg/dL and 1.9 mg/dL, respectively. Her urine output during the preceding 24 hours was 500 mL.

In the evaluation of acute kidney injury (AKI), is the measurement of fractional excretion of sodium (FeNa) and fractional excretion of urea (FeUr) of value?

WHY YOU MIGHT THINK ORDERING FeNa AND/OR FeUr IN THE EVALUATION OF AKI IS HELPFUL

The proper maintenance of sodium balance is paramount to regulating the size of body fluid compartments. Through the interaction of multiple physiologic processes, the kidney regulates tubular reabsorption (or lack thereof) of sodium chloride to match excretion to intake. In normal health, FeNa is typically 1%, although it may vary depending on the dietary sodium intake. The corollary is that 99% of filtered sodium is reabsorbed. Acute tubular injury (ATI) that impairs the tubular resorptive capacity for sodium may increase FeNa to >3%. In addition, during states of water conservation, urea is reabsorbed from the medullary collecting duct, explaining the discrepant rise in BUN relative to creatinine in prerenal azotemia. FeUr falls progressively as water is reabsorbed and urine flow declines, and FeUr less than 35% to 40% may result during prerenal azotemia versus >50% in health or ATI. Theoretically, FeUr is largely unaffected by diuretics, whereas FeNa is increased by diuretics.

In 1976, Espinel reported on the use of FeNa in 17 oliguric patients to discriminate prerenal azotemia from ATI.[1] Establishing what are now familiar indices, FeNa <1% was deemed consistent with prerenal physiology versus >3% indicating ATI. Notably, the study excluded patients who had received diuretics or in whom chronic kidney disease (CKD), glomerulonephritis, or urinary obstruction was suspected.

Given the limitations of FeNa in the context of diuretic use, many physicians instead use FeUr to distinguish prerenal versus ATI causes of AKI. Carvounis et al. reported FeUr and FeNa in 50 patients with prerenal azotemia, 27 with prerenal azotemia receiving diuretics and 25 patients with ATI.[2] Patients with interstitial nephritis, glomerulonephritis, and obstruction were excluded. In the entire cohort, the authors reported sensitivity of 90% and specificity of 96% for FeUr <35% in identifying prerenal azotemia (Table 1). FeNa <1% was slightly less sensitive for prerenal azotemia in the entire cohort at 77%, and this fell to 48% in the presence of diuretics as compared to 89% for FeUr. Naturally, the specificity of FeNa for ATI will fall with the use of diuretics. As shown in Table 1, FeUr <35% has an excellent positive likelihood ratio (LR+) of 22 for prerenal azotemia and a moderate LR+ of 9 for FeUr 35% being consistent with ATI, regardless of the presence of diuretics. This contrasts with FeNa, which if 1% in the presence of diuretics, lacked utility in the diagnosis of ATI. Of note, diuretic use was reported only in the prerenal azotemia group and not specifically in the ATI group. Thus, these comparisons assume diuretics have no effect on test characteristics in ATI. This assumption, however, may not be valid.

WHY THERE IS LITTLE REASON TO ROUTINELY ORDER FeNa AND FeUr IN PATIENTS WITH AKI

The argument against FeNa and FeUr is not primarily financial. FeNa and FeUr testing on all Medicare patients discharged with AKI in 2013 would have cost US$6 million.[3] Although a tiny fraction of annual healthcare expenditure, it would nevertheless be wasteful spending, and its true harm lays in the application of flawed diagnostic reasoning.

That flaw in our conceptual approach to AKI is the broad categorization of patients into either a prerenal or intrinsic etiology of AKI. In reality, renal injury is often multifactorial, and significant prerenal injury may progress to or coexist with intrinsic disease that is commonly ATI. Measurement of a urinary index at a single point in time will often fail to capture this spectrum of causes for AKI. Unfortunately, accurately assessing volume status through physical examination is difficult.[4] Considering FeNa and FeUr may be low in both hemorrhage as well as congestive heart failure, the measurement of these variables adds little to body volume assessment.

It cannot be overemphasized that application of FeNa and FeUr is predicated on the provider already knowing the diagnosis is either prerenal azotemia or ATI. Studies have generally excluded patients >65 years old and those with CKD or notable comorbid renal processes apart from prerenal azotemia or ATI. It is important to recall that a third of kidney biopsies may yield a diagnosis different than the prebiopsy clinical diagnosis, and the gold standard for ATI in studies of FeNa and FeUr was simply a failure of kidney function to improve promptly.[5] Why send a test that is predicated on largely already knowing the answer?

WHAT YOU SHOULD DO INSTEAD: DECIDE IF VOLUME MANIPULATION IS APPROPRIATE

The gold standard for diagnosis, as in many of the above studies, is the prompt improvement of prerenal azotemia with correction of renal hypoperfusion. Ultimately, the decision to administer intravenous fluids or diuretics in the management of AKI will often be independent of both FeNa and FeUr. In considering, for example, the case described above, it is not possible to realistically dichotomize the patient into either a prerenal or ATI category; both are quite likely present. If the clinical assessment supports a component of prerenal azotemia, a low FeNa and/or FeUr will not change the intervention. An elevated FeNa and/or FeUr, however, has at best moderate and potentially no impact on the likelihood for ATI. A patient, moreover, may still require volume manipulation in the context of established ATI. As such, these indices should not alter therapeutic decisions. There may be value in utilizing and identifying new approaches to determining a priori which patients will be fluid responsive, such as inferior vena cava ultrasound.[16] Lastly, evaluation of the urine sediment is an underutilized tool that may prove more useful in discriminating prerenal azotemia from ATI. It also helps to exclude other etiologies of AKI, such as glomerulonephritis and acute interstitial nephritis, which are typical exclusion criteria in studies of FeNa and FeUr.[17, 18]

WHEN IS FeNa AND/OR FeUr USEFUL IN DEFINING THE ETIOLOGY OF AKI?

FeNa and FeUr at best only support a clinical impression of prerenal azotemia or ATI in oliguric AKI, and the accuracy of these metrics is questionable in the setting of CKD, older age, and a variety of comorbidities. There is, however, a setting in which FeNa may be helpful. In practice, FeNa is useful in the evaluation of hepatorenal syndrome, a disorder characterized by oliguria and intense renal sodium reabsorption with resultant spot urine sodium <10 mEq/L and FeNa <1%.[19]

CONCLUSION

The evidence base supporting the use of FeNa and FeUr is limited and often not generalizable to many patients with AKI. The small sample sizes of the studies do not permit adequate capture of diverse mechanisms for renal injury, and these studies are of patients referred for Nephrology consultation and may not be representative of the larger population of patients with less severe AKI. Ultimately, the true etiology will be proven by time and response to therapy. Apart from a supportive role in the diagnosis of hepatorenal syndrome, there is little practical utility to FeNa and FeUr measurement, and these indices should not alter therapeutic decisions when inconsistent with the clinical impression. The evaluation of AKI requires thoughtful clinical assessment, and the gold standard still remains the judicious decision of when to manipulate the intravascular volume status of a patient. In regard to the presented case, urine chemistries are unhelpful due to the combined vasoconstrictive and tubulotoxic effects of the administered intravenous contrast. The ongoing hypotension further contributes to both pre‐renal as well as ischemic tubular injury.

RECOMMENDATIONS

1. FeNa can aid in the diagnosis of hepatorenal syndrome. Otherwise, the routine use of FeNa and FeUr in the diagnosis and management of AKI should be avoided.
2. In pre‐renal azotemia, therapeutic intervention is guided by etiology of the disorder (e.g., intravenous crystalloid support based on a history of hypovolemia and ongoing hypoperfusion, diuresis and/or inotropic support in setting of decompensated heart failure, etc.), without regard to baseline FeNa and FeUr.
3. In ATI, fluid administration is appropriate if hypovolemia is present. FeNa and FeUr cannot diagnose hypovolemia.