Other Literature of Interest

1. Carratala J, FernandezSabe N, Ortega L, et al. Outpatient care compared with hospitalization for community-acquired pneumonia: a randomized trial in low-risk patients. Ann Intern Med. 2005;142: 165-72.

The appropriate triage and management of patients with community-acquired pneumonia (CAP) has important implications for patient outcomes and the allocation of health care resources. Despite the availability of validated risk stratification tools significant variability in clinical practice which results in hospitalization rates that are often inconsistent with the severity of illness. In this unblinded, randomized controlled trial, 224 patients with CAP and a low-risk pneumonia severity index (PSI) score between 51 and 90 (class II and III) were randomized to outpatient oral levofloxacin therapy versus inpatient sequential intravenous and oral levofloxacin therapy. Exclusion criteria included quinolone allergy or use within the previous 3 months, PaO₂ < 60 mm Hg, complicated pleural effusion, lung abscess, metastatic infection, inability to maintain oral intake, and severe psychosocial problems precluding outpatient therapy. In an intention-to-treat analysis, the primary endpoints, of cure of pneumonia (resolution of signs, symptoms, and radiographic changes at 30 days), absence of adverse drug reactions, medical complications, or need for hospitalization at 30 days were achieved in 83.6% of outpatients and in 80.7% of hospitalized patients. For the secondary endpoint of patient satisfaction, 91.2% of outpatients versus 79.1% of hospitalized patients (p=.03) were satisfied, but there were no differences between groups with respect to the secondary endpoint of health-related quality of life. Mortality was similar between the 2 groups, and although the study was not sufficiently powered to address this outcome, and interestingly there was trend toward increased medical complications in the hospitalized patients.

Limitations of this study include lack of blinding by investigators and questions about whether the results can be generalized given the geographic variation in microbial susceptibility to quinolone antibiotics. As the authors suggest, this study also highlights limitations in the PSI scoring system, given that patients with clinical findings and comorbidities who would never be treated in the outpatient setting may in fact fall into low-risk PSI categories. These concerns notwithstanding, this study adds to our ability to identify an additional subset of patients with CAP who can be safely managed as outpatients.

2. Choudhry NK, Fletcher RH, Soumerai SB. Systematic review: the relationship between clinical experience and quality of health care.Ann Intern Med. 2005;142:260-73.

Early in the hospital medicine movement, when it was clear that hospitalists provided more efficient care than their colleagues, experience was cited as a reason for this difference. If, for example, a hospitalist cares for patients with community-acquired pneumonia daily, he or she is more likely to make the transition to oral antibiotics sooner, resulting in a shorter length of stay. Everyone recognized the hospitalists were younger, but is it plausible their “inexperience” explained the difference in care?

Choudhry and colleagues explored the available data surrounding clinical experience and quality of care delivered by physicians. They found few studies that specifically evaluated the effects of experience on quality of care. They did find articles that looked at quality of care and included experience or age as part of the physician characteristics

that possibly explained the differences. They reviewed 59 articles, available on MEDLINE, published since 1966. Forty-five studies found an inverse relationship between increasing experience and performance. For example, physicians more recently out of training programs were more familiar with evidence-based therapies for myocardial infarction and more familiar with NIH recommendations for treatment of breast cancer. Experienced physicians were less likely to screen for hypertension and more likely to prescribe inappropriate medications for elderly patients. This led them to the unexpected conclusion that experienced physicians may be at risk for providing lower-quality care and may need improvement interventions. An accompanying editorial by Drs. Weinberger, Duffy, and Cassel of the American Board of Internal Medicine stated, “The profession cannot ignore this striking finding and its implications: Practice does not make perfect, but it must be accompanied by ongoing active effort to maintain competence and quality of care.” They urged all physicians to “embrace the concepts behind maintenance of (board) certification.”

The image of Marcus Welby, MD, would lead one to believe that experience promotes higher quality care. But don’t ask a hospitalist: Many aren’t old enough to remember seeing him on television.

3. Kucher N, Koo S, Quiroz R, et al. Electronic alerts to prevent venous thromboembolism among hospitalized patients. N Engl J Med. 2005;352:969-77.

March was DVT (deep vein thrombosis) Awareness Month. Despite the availability of numerous guidelines, providers fail to consistently prescribe prophylactic measures against venous thromboembolism (VTE) for their hospitalized patients who meet criteria for prophylaxis.

Kucher and colleagues tested an innovative approach to remind providers to undertake such measures for their patients. They designed a computer program to identify hospitalized patients at increased risk for VTE who were not presently receiving VTE prophylaxis. The program reviewed the records of inpatients on the medical and surgical services and assigned a VTE risk score for each patient based on their history (i.e., history of cancer, hypercoagulability, etc.) and their present medical treatment (i.e., hormone therapy, prescribed bed rest, etc.). For patients considered “high risk” for VTE, the computer reviewed orders to identify ongoing use of VTE prophylactic measures. High-risk patients not receiving prophylactic therapies were randomized into 2 groups. The responsible physician in the intervention group received an electronic alert about the risk of VTE in their patient. No alerts were sent to the physicians in the control group. Physicians who received the alerts were forced to acknowledge the alert by either actively withholding prophylaxis or ordering prophylaxis (mechanical or pharmacologic measures). Patients were followed for 90 days with a primary endpoint of clinically diagnosed, objectively confirmed deep vein thrombosis (DVT) or pulmonary embolism (PE). The primary endpoint occurred in 8.2% of the control group versus 4.9% in the invention group (p<.001). The alert reduced the risk of DVT or PE at 90 days by 41% (p=.001).

The results of the study are interesting. The authors acknowledged that many physicians had patients in both groups. So receiving 1 alert may have affected their use of prophylaxis in both groups. They also could not eliminate the possibility of diagnostic bias. Prophylaxis was not blinded and VTE testing was not routinely performed. Would physicians be more likely to order an imaging study for symptomatic patients on no prophylaxis than patients on prophylaxis? Nevertheless, for hospitals that have sufficient computer resources, implementation of such alerts can elevate physician awareness about VTE and other clinical conditions.

4. Lau DT, Kasper JD, Pofer DE, et al. Hospitalization and death associated with potentially inappropriate medication prescriptions among elderly nursing home residents. Arch Intern Med. 2005;165: 68-74.

Lau and colleagues studied the impact of potentially inappropriate medications among residents of longtermcare facilities. They used information from a 1996 national survey of home residents. The sample included 3372 residents, 65 years and older, who lived in a nursing home for 3 months or longer. Over half of the residents were older than 85 years old and 75% were female. Only 10% were black. Nearly two thirds had dementia or other mental disorders. The study used the Beers Criteria to define potentially inappropriate medications. The potential errors in medications were categorized as 1 of 3 types:

1. inappropriate choice of medication
2. excessive medication dosage
3. drug–disease interactions

Residents were considered to have a potentially inappropriate medication if their medication administration records revealed any of the above findings.

A univariate analysis showed that the risk of hospitalization was almost 30% higher among residents who received potentially inappropriate medications in the preceding month and 33% higher among residents who received potentially inappropriate medications for 2 consecutive months, compared with residents with no inappropriate medication exposure. The odds of death in any month were 21% higher among residents who had inappropriate medication exposure during the month of death or the preceeding month, compared with those with no inappropriate medication exposure.

These findings can be generalized to the inpatient setting, where hospitalists have the opportunity to influence and modify prescribing practices in the elderly population.

5. Lessnau KD. Is chest radiography necessary after uncomplicated insertion of a triplelumen catheter in the right internal jugular vein, using the anterior approach? Chest. 2005;127:220-3.

The routine use of chest radiography to confirm proper triplelumen catheter (TLC) placement may be an unnecessary and costly intervention. Lessnau conducted a prospective observational study of 100 consecutive patients over a 4-month period who required non-urgent TLC placement. The primary operators of the procedure included 18 medical residents, 3 pulmonary fellows, and a pulmonary attending with supervision provided for more junior clinicians. Operators followed a standardized approach to TLC placement utilizing the anterior approach to the right internal jugular vein. Complicated procedures were predefined as any procedure that required more than 3 needle passes, resulted in hemorrhage or hematoma formation (where there was concern for pneumothorax), or an absence of blood return in any of the TLC’s lumens. All subjects underwent routine post-procedure chest radiography to determine proper placement of the catheter and to exclude pneumothorax. A blinded radiologist reviewed these images.

Ninety-eight of the 100 catheters were in proper position. One malpositioned catheter was 7 cm above the right atrium in a patient who was 215 cm (>7 feet) tall. The second was noted to be in an S-shaped position on chest radiography. This procedure had required 20 needle passes and 5 slides of the catheter; additionally, blood return was inadequate in 2 lumens of the catheter. An operator reported a possible complication in 10 other procedures, but the only clinical finding in these cases was the development of a local hematoma in 1 patient. Eighty-eight patients had uncomplicated insertions and had normal chest radiographs. There were no pneumothoraces.

This study demonstrates that in carefully controlled and supervised situations, as described in the study, routine chest radiography may be omitted if the insertion goes smoothly. It is important to note that these results are specific to the technique described in the study (using the anterior approach to the right internal jugular, using a short finder needle to initially locate the vein) and cannot be extrapolated to other methods of TLC insertion. Important limitations of the study include the sample size of only 100 patients and the use of only a single anatomic approach to TLC insertion. These findings, although an important first step, will need to be reproduced on a larger scale before we can recommend the cessation of routine chest radiography after TLC placement on a more widespread basis.

6. Safdar N, Fine JP, Maki DG. Metaanalysis: methods for diagnosing intravascular devicerelated bloodstream infection. Ann Intern Med. 2005;142:451-66.

Intravascular device (IVD)–related blood stream infections are a frequent cause of morbidity and mortality, and yet there is lack of a clear consensus on the most accurate method to make this diagnosis.

In this metaanalysis, Safdar et al. reviewed 185 studies, including 8 different diagnostic tests, for the detection of IVD-related bloodstream infections, of which 51 studies met the inclusion criteria. Tests were divided into IVD-sparing and those requiring IVD removal. Pooled sensitivity and specificity, summary measures of accuracy, and the mean log odds ratio were determined. The most accurate IVD-sparing test was paired quantitative blood cultures (simultaneous blood cultures from the IVD and a peripheral site, with a positive result defined as an IVD-site microorganism concentration 3–5 times greater than peripheral site) with a sensitivity of 0.87 (95% CI: 0.83–0.91) and specificity of 0.98 (95% CI: 0.97–0.99). This was followed by quantitative IVD-drawn blood cultures alone (positive result defined as growth of ≥100 CFU), with a sensitivity of 0.77 (95% CI: 0.69–0.85) and a specificity of 0.90 (95% CI: 0.88–0.92). IVD-drawn qualitative blood cultures had a sensitivity of 0.87 (95% CI: 0.80–0.94) and a specificity of 0.83 (95% CI: 0.78–0.88), and IVD- and peripheral-drawn qualitative blood cultures with differential time to positivity had a sensitivity of 0.85 (95% CI: 0.78–0.92) and specificity of 0.81 (95% CI: 0.81–0.97).

The most accurate test requiring IVD removal was quantitative catheter segment culture (segment of catheter is flushed or sonicated and plated, positive if ≥1000 CFU), with sensitivity of 0.83 (95% CI: 0.78–0.88) and specificity of 0.87 (95% CI: 0.85–0.89), followed by semi-quantitative catheter segment culture (5cm segment plated, positive if ≥ 15 CFU) with sensitivity of 0.82 (95% CI: 0.81–0.89) and specificity of 0.82 (95% CI: 0.80–0.84). The least accurate was qualitative catheter segment culture (positive if any growth) with a sensitivity of 0.90 (95% CI: 0.83–0.97) and specificity of 0.72 (95% CI: 0.66–0.78).

The limitations of this study include heterogeneity of study design, including limited data on the use of antibiotics before culture data was obtained and the baseline prevalence of bacteremia in the study populations. In addition, all data was obtained prior to the widespread use of antibiotic-coated catheters. While these results support the catheter-tip quantitative culture techniques that are already widely in use, they are less applicable to blood culture testing techniques, because quantitative assays are rarely used. Fortunately, all of these assays have a high negative predictive value, and false-positive results can be minimized by reserving testing for patients in whom there is moderate-to-high pretest probability of IVD related bloodstream infection.

7. Sopena N, Sabria M, Neunos 2000 Study Group. Multicenter study of hospital-acquired pneumonia in non-ICU patients. Chest. 2005;127:213-9.

A growing body of literature exists on hospital-acquired pneumonia (HAP) in the ICU setting. Sopena and colleagues extend the HAP literature to the non-ICU setting in a multicenter cross-sectional study. Cases of HAP were identified if clinical or radiographic evidence of pneumonia developed 72 hours after admission or within 10 days of a previous discharge. Patients who developed pneumonia in the ICU were excluded from analysis.

During an 18-month study period, 165 cases were identified with complete clinical and microbiologic data. The incidence of HAP was 3.1 ± 1.4 per 1000 hospital admissions. Ninety-eight (59.4%) patients diagnosed with HAP had severe underlying diseases that were classified as fatal (<1 year) or ultimately fatal (in 5 years). Extrinsic risk factors observed in patients with HAP included concurrent steroid use (29%), antibiotic therapy (53.3%), use of H2 blockers (37%), and hospitalization greater than 5 days (76%). Microbiologic data were positive in 60 (36.4%) cases. Streptococcus pneumoniae was diagnosed in 16 cases (9.7%), enterobacteriaceae in 8 (4.8%), Legionella pneumophila in 7 (4.2%), Aspergillus sp in 7 (4.2%), Pseudomonas aeruginosa in 7 (4.2%). Four cases of Staphylococcus aureus were diagnosed (3%), only one of which was methicillin resistant.

Complications of HAP occurred in 52.1% of cases and included respiratory failure (34.5%), pleural effusion (20.6%), septic shock (9.6%), renal failure (4.8%), and empyema (2.4%). Forty-three (26%) patients died during the hospitalization; 23 of these cases were directly attributed to HAP.

A limitation of the study is that the incidence of HAP was somewhat lower than reported in the literature and thus might represent an unintended sampling bias. Moreover, the study demonstrated underlying factors seen in patients with HAP, but these are not necessarily causative. Results useful to hospitalists include a higher than expected rate of Legionella and Aspergillus sp causing HAP in this population. A Legionella outbreak was not the explanation, as these cases were diagnosed in 5 different hospitals. The high frequency of adverse outcomes associated with HAP should alert hospitalists to the risk of nosocomial pneumonia in the non-ICU setting.

1. Carratala J, FernandezSabe N, Ortega L, et al. Outpatient care compared with hospitalization for community-acquired pneumonia: a randomized trial in low-risk patients. Ann Intern Med. 2005;142: 165-72.

The appropriate triage and management of patients with community-acquired pneumonia (CAP) has important implications for patient outcomes and the allocation of health care resources. Despite the availability of validated risk stratification tools significant variability in clinical practice which results in hospitalization rates that are often inconsistent with the severity of illness. In this unblinded, randomized controlled trial, 224 patients with CAP and a low-risk pneumonia severity index (PSI) score between 51 and 90 (class II and III) were randomized to outpatient oral levofloxacin therapy versus inpatient sequential intravenous and oral levofloxacin therapy. Exclusion criteria included quinolone allergy or use within the previous 3 months, PaO₂ < 60 mm Hg, complicated pleural effusion, lung abscess, metastatic infection, inability to maintain oral intake, and severe psychosocial problems precluding outpatient therapy. In an intention-to-treat analysis, the primary endpoints, of cure of pneumonia (resolution of signs, symptoms, and radiographic changes at 30 days), absence of adverse drug reactions, medical complications, or need for hospitalization at 30 days were achieved in 83.6% of outpatients and in 80.7% of hospitalized patients. For the secondary endpoint of patient satisfaction, 91.2% of outpatients versus 79.1% of hospitalized patients (p=.03) were satisfied, but there were no differences between groups with respect to the secondary endpoint of health-related quality of life. Mortality was similar between the 2 groups, and although the study was not sufficiently powered to address this outcome, and interestingly there was trend toward increased medical complications in the hospitalized patients.

Limitations of this study include lack of blinding by investigators and questions about whether the results can be generalized given the geographic variation in microbial susceptibility to quinolone antibiotics. As the authors suggest, this study also highlights limitations in the PSI scoring system, given that patients with clinical findings and comorbidities who would never be treated in the outpatient setting may in fact fall into low-risk PSI categories. These concerns notwithstanding, this study adds to our ability to identify an additional subset of patients with CAP who can be safely managed as outpatients.

2. Choudhry NK, Fletcher RH, Soumerai SB. Systematic review: the relationship between clinical experience and quality of health care.Ann Intern Med. 2005;142:260-73.

Early in the hospital medicine movement, when it was clear that hospitalists provided more efficient care than their colleagues, experience was cited as a reason for this difference. If, for example, a hospitalist cares for patients with community-acquired pneumonia daily, he or she is more likely to make the transition to oral antibiotics sooner, resulting in a shorter length of stay. Everyone recognized the hospitalists were younger, but is it plausible their “inexperience” explained the difference in care?

Choudhry and colleagues explored the available data surrounding clinical experience and quality of care delivered by physicians. They found few studies that specifically evaluated the effects of experience on quality of care. They did find articles that looked at quality of care and included experience or age as part of the physician characteristics

that possibly explained the differences. They reviewed 59 articles, available on MEDLINE, published since 1966. Forty-five studies found an inverse relationship between increasing experience and performance. For example, physicians more recently out of training programs were more familiar with evidence-based therapies for myocardial infarction and more familiar with NIH recommendations for treatment of breast cancer. Experienced physicians were less likely to screen for hypertension and more likely to prescribe inappropriate medications for elderly patients. This led them to the unexpected conclusion that experienced physicians may be at risk for providing lower-quality care and may need improvement interventions. An accompanying editorial by Drs. Weinberger, Duffy, and Cassel of the American Board of Internal Medicine stated, “The profession cannot ignore this striking finding and its implications: Practice does not make perfect, but it must be accompanied by ongoing active effort to maintain competence and quality of care.” They urged all physicians to “embrace the concepts behind maintenance of (board) certification.”

The image of Marcus Welby, MD, would lead one to believe that experience promotes higher quality care. But don’t ask a hospitalist: Many aren’t old enough to remember seeing him on television.

3. Kucher N, Koo S, Quiroz R, et al. Electronic alerts to prevent venous thromboembolism among hospitalized patients. N Engl J Med. 2005;352:969-77.

March was DVT (deep vein thrombosis) Awareness Month. Despite the availability of numerous guidelines, providers fail to consistently prescribe prophylactic measures against venous thromboembolism (VTE) for their hospitalized patients who meet criteria for prophylaxis.

Kucher and colleagues tested an innovative approach to remind providers to undertake such measures for their patients. They designed a computer program to identify hospitalized patients at increased risk for VTE who were not presently receiving VTE prophylaxis. The program reviewed the records of inpatients on the medical and surgical services and assigned a VTE risk score for each patient based on their history (i.e., history of cancer, hypercoagulability, etc.) and their present medical treatment (i.e., hormone therapy, prescribed bed rest, etc.). For patients considered “high risk” for VTE, the computer reviewed orders to identify ongoing use of VTE prophylactic measures. High-risk patients not receiving prophylactic therapies were randomized into 2 groups. The responsible physician in the intervention group received an electronic alert about the risk of VTE in their patient. No alerts were sent to the physicians in the control group. Physicians who received the alerts were forced to acknowledge the alert by either actively withholding prophylaxis or ordering prophylaxis (mechanical or pharmacologic measures). Patients were followed for 90 days with a primary endpoint of clinically diagnosed, objectively confirmed deep vein thrombosis (DVT) or pulmonary embolism (PE). The primary endpoint occurred in 8.2% of the control group versus 4.9% in the invention group (p<.001). The alert reduced the risk of DVT or PE at 90 days by 41% (p=.001).

The results of the study are interesting. The authors acknowledged that many physicians had patients in both groups. So receiving 1 alert may have affected their use of prophylaxis in both groups. They also could not eliminate the possibility of diagnostic bias. Prophylaxis was not blinded and VTE testing was not routinely performed. Would physicians be more likely to order an imaging study for symptomatic patients on no prophylaxis than patients on prophylaxis? Nevertheless, for hospitals that have sufficient computer resources, implementation of such alerts can elevate physician awareness about VTE and other clinical conditions.

4. Lau DT, Kasper JD, Pofer DE, et al. Hospitalization and death associated with potentially inappropriate medication prescriptions among elderly nursing home residents. Arch Intern Med. 2005;165: 68-74.

Lau and colleagues studied the impact of potentially inappropriate medications among residents of longtermcare facilities. They used information from a 1996 national survey of home residents. The sample included 3372 residents, 65 years and older, who lived in a nursing home for 3 months or longer. Over half of the residents were older than 85 years old and 75% were female. Only 10% were black. Nearly two thirds had dementia or other mental disorders. The study used the Beers Criteria to define potentially inappropriate medications. The potential errors in medications were categorized as 1 of 3 types:

1. inappropriate choice of medication
2. excessive medication dosage
3. drug–disease interactions

Residents were considered to have a potentially inappropriate medication if their medication administration records revealed any of the above findings.

A univariate analysis showed that the risk of hospitalization was almost 30% higher among residents who received potentially inappropriate medications in the preceding month and 33% higher among residents who received potentially inappropriate medications for 2 consecutive months, compared with residents with no inappropriate medication exposure. The odds of death in any month were 21% higher among residents who had inappropriate medication exposure during the month of death or the preceeding month, compared with those with no inappropriate medication exposure.

These findings can be generalized to the inpatient setting, where hospitalists have the opportunity to influence and modify prescribing practices in the elderly population.

5. Lessnau KD. Is chest radiography necessary after uncomplicated insertion of a triplelumen catheter in the right internal jugular vein, using the anterior approach? Chest. 2005;127:220-3.

The routine use of chest radiography to confirm proper triplelumen catheter (TLC) placement may be an unnecessary and costly intervention. Lessnau conducted a prospective observational study of 100 consecutive patients over a 4-month period who required non-urgent TLC placement. The primary operators of the procedure included 18 medical residents, 3 pulmonary fellows, and a pulmonary attending with supervision provided for more junior clinicians. Operators followed a standardized approach to TLC placement utilizing the anterior approach to the right internal jugular vein. Complicated procedures were predefined as any procedure that required more than 3 needle passes, resulted in hemorrhage or hematoma formation (where there was concern for pneumothorax), or an absence of blood return in any of the TLC’s lumens. All subjects underwent routine post-procedure chest radiography to determine proper placement of the catheter and to exclude pneumothorax. A blinded radiologist reviewed these images.

Ninety-eight of the 100 catheters were in proper position. One malpositioned catheter was 7 cm above the right atrium in a patient who was 215 cm (>7 feet) tall. The second was noted to be in an S-shaped position on chest radiography. This procedure had required 20 needle passes and 5 slides of the catheter; additionally, blood return was inadequate in 2 lumens of the catheter. An operator reported a possible complication in 10 other procedures, but the only clinical finding in these cases was the development of a local hematoma in 1 patient. Eighty-eight patients had uncomplicated insertions and had normal chest radiographs. There were no pneumothoraces.

This study demonstrates that in carefully controlled and supervised situations, as described in the study, routine chest radiography may be omitted if the insertion goes smoothly. It is important to note that these results are specific to the technique described in the study (using the anterior approach to the right internal jugular, using a short finder needle to initially locate the vein) and cannot be extrapolated to other methods of TLC insertion. Important limitations of the study include the sample size of only 100 patients and the use of only a single anatomic approach to TLC insertion. These findings, although an important first step, will need to be reproduced on a larger scale before we can recommend the cessation of routine chest radiography after TLC placement on a more widespread basis.

6. Safdar N, Fine JP, Maki DG. Metaanalysis: methods for diagnosing intravascular devicerelated bloodstream infection. Ann Intern Med. 2005;142:451-66.

Intravascular device (IVD)–related blood stream infections are a frequent cause of morbidity and mortality, and yet there is lack of a clear consensus on the most accurate method to make this diagnosis.

In this metaanalysis, Safdar et al. reviewed 185 studies, including 8 different diagnostic tests, for the detection of IVD-related bloodstream infections, of which 51 studies met the inclusion criteria. Tests were divided into IVD-sparing and those requiring IVD removal. Pooled sensitivity and specificity, summary measures of accuracy, and the mean log odds ratio were determined. The most accurate IVD-sparing test was paired quantitative blood cultures (simultaneous blood cultures from the IVD and a peripheral site, with a positive result defined as an IVD-site microorganism concentration 3–5 times greater than peripheral site) with a sensitivity of 0.87 (95% CI: 0.83–0.91) and specificity of 0.98 (95% CI: 0.97–0.99). This was followed by quantitative IVD-drawn blood cultures alone (positive result defined as growth of ≥100 CFU), with a sensitivity of 0.77 (95% CI: 0.69–0.85) and a specificity of 0.90 (95% CI: 0.88–0.92). IVD-drawn qualitative blood cultures had a sensitivity of 0.87 (95% CI: 0.80–0.94) and a specificity of 0.83 (95% CI: 0.78–0.88), and IVD- and peripheral-drawn qualitative blood cultures with differential time to positivity had a sensitivity of 0.85 (95% CI: 0.78–0.92) and specificity of 0.81 (95% CI: 0.81–0.97).

The most accurate test requiring IVD removal was quantitative catheter segment culture (segment of catheter is flushed or sonicated and plated, positive if ≥1000 CFU), with sensitivity of 0.83 (95% CI: 0.78–0.88) and specificity of 0.87 (95% CI: 0.85–0.89), followed by semi-quantitative catheter segment culture (5cm segment plated, positive if ≥ 15 CFU) with sensitivity of 0.82 (95% CI: 0.81–0.89) and specificity of 0.82 (95% CI: 0.80–0.84). The least accurate was qualitative catheter segment culture (positive if any growth) with a sensitivity of 0.90 (95% CI: 0.83–0.97) and specificity of 0.72 (95% CI: 0.66–0.78).

The limitations of this study include heterogeneity of study design, including limited data on the use of antibiotics before culture data was obtained and the baseline prevalence of bacteremia in the study populations. In addition, all data was obtained prior to the widespread use of antibiotic-coated catheters. While these results support the catheter-tip quantitative culture techniques that are already widely in use, they are less applicable to blood culture testing techniques, because quantitative assays are rarely used. Fortunately, all of these assays have a high negative predictive value, and false-positive results can be minimized by reserving testing for patients in whom there is moderate-to-high pretest probability of IVD related bloodstream infection.

7. Sopena N, Sabria M, Neunos 2000 Study Group. Multicenter study of hospital-acquired pneumonia in non-ICU patients. Chest. 2005;127:213-9.

A growing body of literature exists on hospital-acquired pneumonia (HAP) in the ICU setting. Sopena and colleagues extend the HAP literature to the non-ICU setting in a multicenter cross-sectional study. Cases of HAP were identified if clinical or radiographic evidence of pneumonia developed 72 hours after admission or within 10 days of a previous discharge. Patients who developed pneumonia in the ICU were excluded from analysis.

During an 18-month study period, 165 cases were identified with complete clinical and microbiologic data. The incidence of HAP was 3.1 ± 1.4 per 1000 hospital admissions. Ninety-eight (59.4%) patients diagnosed with HAP had severe underlying diseases that were classified as fatal (<1 year) or ultimately fatal (in 5 years). Extrinsic risk factors observed in patients with HAP included concurrent steroid use (29%), antibiotic therapy (53.3%), use of H2 blockers (37%), and hospitalization greater than 5 days (76%). Microbiologic data were positive in 60 (36.4%) cases. Streptococcus pneumoniae was diagnosed in 16 cases (9.7%), enterobacteriaceae in 8 (4.8%), Legionella pneumophila in 7 (4.2%), Aspergillus sp in 7 (4.2%), Pseudomonas aeruginosa in 7 (4.2%). Four cases of Staphylococcus aureus were diagnosed (3%), only one of which was methicillin resistant.

Complications of HAP occurred in 52.1% of cases and included respiratory failure (34.5%), pleural effusion (20.6%), septic shock (9.6%), renal failure (4.8%), and empyema (2.4%). Forty-three (26%) patients died during the hospitalization; 23 of these cases were directly attributed to HAP.

A limitation of the study is that the incidence of HAP was somewhat lower than reported in the literature and thus might represent an unintended sampling bias. Moreover, the study demonstrated underlying factors seen in patients with HAP, but these are not necessarily causative. Results useful to hospitalists include a higher than expected rate of Legionella and Aspergillus sp causing HAP in this population. A Legionella outbreak was not the explanation, as these cases were diagnosed in 5 different hospitals. The high frequency of adverse outcomes associated with HAP should alert hospitalists to the risk of nosocomial pneumonia in the non-ICU setting.

1. Carratala J, FernandezSabe N, Ortega L, et al. Outpatient care compared with hospitalization for community-acquired pneumonia: a randomized trial in low-risk patients. Ann Intern Med. 2005;142: 165-72.

The appropriate triage and management of patients with community-acquired pneumonia (CAP) has important implications for patient outcomes and the allocation of health care resources. Despite the availability of validated risk stratification tools significant variability in clinical practice which results in hospitalization rates that are often inconsistent with the severity of illness. In this unblinded, randomized controlled trial, 224 patients with CAP and a low-risk pneumonia severity index (PSI) score between 51 and 90 (class II and III) were randomized to outpatient oral levofloxacin therapy versus inpatient sequential intravenous and oral levofloxacin therapy. Exclusion criteria included quinolone allergy or use within the previous 3 months, PaO₂ < 60 mm Hg, complicated pleural effusion, lung abscess, metastatic infection, inability to maintain oral intake, and severe psychosocial problems precluding outpatient therapy. In an intention-to-treat analysis, the primary endpoints, of cure of pneumonia (resolution of signs, symptoms, and radiographic changes at 30 days), absence of adverse drug reactions, medical complications, or need for hospitalization at 30 days were achieved in 83.6% of outpatients and in 80.7% of hospitalized patients. For the secondary endpoint of patient satisfaction, 91.2% of outpatients versus 79.1% of hospitalized patients (p=.03) were satisfied, but there were no differences between groups with respect to the secondary endpoint of health-related quality of life. Mortality was similar between the 2 groups, and although the study was not sufficiently powered to address this outcome, and interestingly there was trend toward increased medical complications in the hospitalized patients.

Limitations of this study include lack of blinding by investigators and questions about whether the results can be generalized given the geographic variation in microbial susceptibility to quinolone antibiotics. As the authors suggest, this study also highlights limitations in the PSI scoring system, given that patients with clinical findings and comorbidities who would never be treated in the outpatient setting may in fact fall into low-risk PSI categories. These concerns notwithstanding, this study adds to our ability to identify an additional subset of patients with CAP who can be safely managed as outpatients.

2. Choudhry NK, Fletcher RH, Soumerai SB. Systematic review: the relationship between clinical experience and quality of health care.Ann Intern Med. 2005;142:260-73.

Early in the hospital medicine movement, when it was clear that hospitalists provided more efficient care than their colleagues, experience was cited as a reason for this difference. If, for example, a hospitalist cares for patients with community-acquired pneumonia daily, he or she is more likely to make the transition to oral antibiotics sooner, resulting in a shorter length of stay. Everyone recognized the hospitalists were younger, but is it plausible their “inexperience” explained the difference in care?

Choudhry and colleagues explored the available data surrounding clinical experience and quality of care delivered by physicians. They found few studies that specifically evaluated the effects of experience on quality of care. They did find articles that looked at quality of care and included experience or age as part of the physician characteristics

that possibly explained the differences. They reviewed 59 articles, available on MEDLINE, published since 1966. Forty-five studies found an inverse relationship between increasing experience and performance. For example, physicians more recently out of training programs were more familiar with evidence-based therapies for myocardial infarction and more familiar with NIH recommendations for treatment of breast cancer. Experienced physicians were less likely to screen for hypertension and more likely to prescribe inappropriate medications for elderly patients. This led them to the unexpected conclusion that experienced physicians may be at risk for providing lower-quality care and may need improvement interventions. An accompanying editorial by Drs. Weinberger, Duffy, and Cassel of the American Board of Internal Medicine stated, “The profession cannot ignore this striking finding and its implications: Practice does not make perfect, but it must be accompanied by ongoing active effort to maintain competence and quality of care.” They urged all physicians to “embrace the concepts behind maintenance of (board) certification.”

The image of Marcus Welby, MD, would lead one to believe that experience promotes higher quality care. But don’t ask a hospitalist: Many aren’t old enough to remember seeing him on television.

3. Kucher N, Koo S, Quiroz R, et al. Electronic alerts to prevent venous thromboembolism among hospitalized patients. N Engl J Med. 2005;352:969-77.

March was DVT (deep vein thrombosis) Awareness Month. Despite the availability of numerous guidelines, providers fail to consistently prescribe prophylactic measures against venous thromboembolism (VTE) for their hospitalized patients who meet criteria for prophylaxis.

Kucher and colleagues tested an innovative approach to remind providers to undertake such measures for their patients. They designed a computer program to identify hospitalized patients at increased risk for VTE who were not presently receiving VTE prophylaxis. The program reviewed the records of inpatients on the medical and surgical services and assigned a VTE risk score for each patient based on their history (i.e., history of cancer, hypercoagulability, etc.) and their present medical treatment (i.e., hormone therapy, prescribed bed rest, etc.). For patients considered “high risk” for VTE, the computer reviewed orders to identify ongoing use of VTE prophylactic measures. High-risk patients not receiving prophylactic therapies were randomized into 2 groups. The responsible physician in the intervention group received an electronic alert about the risk of VTE in their patient. No alerts were sent to the physicians in the control group. Physicians who received the alerts were forced to acknowledge the alert by either actively withholding prophylaxis or ordering prophylaxis (mechanical or pharmacologic measures). Patients were followed for 90 days with a primary endpoint of clinically diagnosed, objectively confirmed deep vein thrombosis (DVT) or pulmonary embolism (PE). The primary endpoint occurred in 8.2% of the control group versus 4.9% in the invention group (p<.001). The alert reduced the risk of DVT or PE at 90 days by 41% (p=.001).

The results of the study are interesting. The authors acknowledged that many physicians had patients in both groups. So receiving 1 alert may have affected their use of prophylaxis in both groups. They also could not eliminate the possibility of diagnostic bias. Prophylaxis was not blinded and VTE testing was not routinely performed. Would physicians be more likely to order an imaging study for symptomatic patients on no prophylaxis than patients on prophylaxis? Nevertheless, for hospitals that have sufficient computer resources, implementation of such alerts can elevate physician awareness about VTE and other clinical conditions.

4. Lau DT, Kasper JD, Pofer DE, et al. Hospitalization and death associated with potentially inappropriate medication prescriptions among elderly nursing home residents. Arch Intern Med. 2005;165: 68-74.

Lau and colleagues studied the impact of potentially inappropriate medications among residents of longtermcare facilities. They used information from a 1996 national survey of home residents. The sample included 3372 residents, 65 years and older, who lived in a nursing home for 3 months or longer. Over half of the residents were older than 85 years old and 75% were female. Only 10% were black. Nearly two thirds had dementia or other mental disorders. The study used the Beers Criteria to define potentially inappropriate medications. The potential errors in medications were categorized as 1 of 3 types:

1. inappropriate choice of medication
2. excessive medication dosage
3. drug–disease interactions

Residents were considered to have a potentially inappropriate medication if their medication administration records revealed any of the above findings.

A univariate analysis showed that the risk of hospitalization was almost 30% higher among residents who received potentially inappropriate medications in the preceding month and 33% higher among residents who received potentially inappropriate medications for 2 consecutive months, compared with residents with no inappropriate medication exposure. The odds of death in any month were 21% higher among residents who had inappropriate medication exposure during the month of death or the preceeding month, compared with those with no inappropriate medication exposure.

These findings can be generalized to the inpatient setting, where hospitalists have the opportunity to influence and modify prescribing practices in the elderly population.

5. Lessnau KD. Is chest radiography necessary after uncomplicated insertion of a triplelumen catheter in the right internal jugular vein, using the anterior approach? Chest. 2005;127:220-3.

The routine use of chest radiography to confirm proper triplelumen catheter (TLC) placement may be an unnecessary and costly intervention. Lessnau conducted a prospective observational study of 100 consecutive patients over a 4-month period who required non-urgent TLC placement. The primary operators of the procedure included 18 medical residents, 3 pulmonary fellows, and a pulmonary attending with supervision provided for more junior clinicians. Operators followed a standardized approach to TLC placement utilizing the anterior approach to the right internal jugular vein. Complicated procedures were predefined as any procedure that required more than 3 needle passes, resulted in hemorrhage or hematoma formation (where there was concern for pneumothorax), or an absence of blood return in any of the TLC’s lumens. All subjects underwent routine post-procedure chest radiography to determine proper placement of the catheter and to exclude pneumothorax. A blinded radiologist reviewed these images.

Ninety-eight of the 100 catheters were in proper position. One malpositioned catheter was 7 cm above the right atrium in a patient who was 215 cm (>7 feet) tall. The second was noted to be in an S-shaped position on chest radiography. This procedure had required 20 needle passes and 5 slides of the catheter; additionally, blood return was inadequate in 2 lumens of the catheter. An operator reported a possible complication in 10 other procedures, but the only clinical finding in these cases was the development of a local hematoma in 1 patient. Eighty-eight patients had uncomplicated insertions and had normal chest radiographs. There were no pneumothoraces.

This study demonstrates that in carefully controlled and supervised situations, as described in the study, routine chest radiography may be omitted if the insertion goes smoothly. It is important to note that these results are specific to the technique described in the study (using the anterior approach to the right internal jugular, using a short finder needle to initially locate the vein) and cannot be extrapolated to other methods of TLC insertion. Important limitations of the study include the sample size of only 100 patients and the use of only a single anatomic approach to TLC insertion. These findings, although an important first step, will need to be reproduced on a larger scale before we can recommend the cessation of routine chest radiography after TLC placement on a more widespread basis.

6. Safdar N, Fine JP, Maki DG. Metaanalysis: methods for diagnosing intravascular devicerelated bloodstream infection. Ann Intern Med. 2005;142:451-66.

Intravascular device (IVD)–related blood stream infections are a frequent cause of morbidity and mortality, and yet there is lack of a clear consensus on the most accurate method to make this diagnosis.

In this metaanalysis, Safdar et al. reviewed 185 studies, including 8 different diagnostic tests, for the detection of IVD-related bloodstream infections, of which 51 studies met the inclusion criteria. Tests were divided into IVD-sparing and those requiring IVD removal. Pooled sensitivity and specificity, summary measures of accuracy, and the mean log odds ratio were determined. The most accurate IVD-sparing test was paired quantitative blood cultures (simultaneous blood cultures from the IVD and a peripheral site, with a positive result defined as an IVD-site microorganism concentration 3–5 times greater than peripheral site) with a sensitivity of 0.87 (95% CI: 0.83–0.91) and specificity of 0.98 (95% CI: 0.97–0.99). This was followed by quantitative IVD-drawn blood cultures alone (positive result defined as growth of ≥100 CFU), with a sensitivity of 0.77 (95% CI: 0.69–0.85) and a specificity of 0.90 (95% CI: 0.88–0.92). IVD-drawn qualitative blood cultures had a sensitivity of 0.87 (95% CI: 0.80–0.94) and a specificity of 0.83 (95% CI: 0.78–0.88), and IVD- and peripheral-drawn qualitative blood cultures with differential time to positivity had a sensitivity of 0.85 (95% CI: 0.78–0.92) and specificity of 0.81 (95% CI: 0.81–0.97).

The most accurate test requiring IVD removal was quantitative catheter segment culture (segment of catheter is flushed or sonicated and plated, positive if ≥1000 CFU), with sensitivity of 0.83 (95% CI: 0.78–0.88) and specificity of 0.87 (95% CI: 0.85–0.89), followed by semi-quantitative catheter segment culture (5cm segment plated, positive if ≥ 15 CFU) with sensitivity of 0.82 (95% CI: 0.81–0.89) and specificity of 0.82 (95% CI: 0.80–0.84). The least accurate was qualitative catheter segment culture (positive if any growth) with a sensitivity of 0.90 (95% CI: 0.83–0.97) and specificity of 0.72 (95% CI: 0.66–0.78).

The limitations of this study include heterogeneity of study design, including limited data on the use of antibiotics before culture data was obtained and the baseline prevalence of bacteremia in the study populations. In addition, all data was obtained prior to the widespread use of antibiotic-coated catheters. While these results support the catheter-tip quantitative culture techniques that are already widely in use, they are less applicable to blood culture testing techniques, because quantitative assays are rarely used. Fortunately, all of these assays have a high negative predictive value, and false-positive results can be minimized by reserving testing for patients in whom there is moderate-to-high pretest probability of IVD related bloodstream infection.

7. Sopena N, Sabria M, Neunos 2000 Study Group. Multicenter study of hospital-acquired pneumonia in non-ICU patients. Chest. 2005;127:213-9.

A growing body of literature exists on hospital-acquired pneumonia (HAP) in the ICU setting. Sopena and colleagues extend the HAP literature to the non-ICU setting in a multicenter cross-sectional study. Cases of HAP were identified if clinical or radiographic evidence of pneumonia developed 72 hours after admission or within 10 days of a previous discharge. Patients who developed pneumonia in the ICU were excluded from analysis.

During an 18-month study period, 165 cases were identified with complete clinical and microbiologic data. The incidence of HAP was 3.1 ± 1.4 per 1000 hospital admissions. Ninety-eight (59.4%) patients diagnosed with HAP had severe underlying diseases that were classified as fatal (<1 year) or ultimately fatal (in 5 years). Extrinsic risk factors observed in patients with HAP included concurrent steroid use (29%), antibiotic therapy (53.3%), use of H2 blockers (37%), and hospitalization greater than 5 days (76%). Microbiologic data were positive in 60 (36.4%) cases. Streptococcus pneumoniae was diagnosed in 16 cases (9.7%), enterobacteriaceae in 8 (4.8%), Legionella pneumophila in 7 (4.2%), Aspergillus sp in 7 (4.2%), Pseudomonas aeruginosa in 7 (4.2%). Four cases of Staphylococcus aureus were diagnosed (3%), only one of which was methicillin resistant.

Complications of HAP occurred in 52.1% of cases and included respiratory failure (34.5%), pleural effusion (20.6%), septic shock (9.6%), renal failure (4.8%), and empyema (2.4%). Forty-three (26%) patients died during the hospitalization; 23 of these cases were directly attributed to HAP.

A limitation of the study is that the incidence of HAP was somewhat lower than reported in the literature and thus might represent an unintended sampling bias. Moreover, the study demonstrated underlying factors seen in patients with HAP, but these are not necessarily causative. Results useful to hospitalists include a higher than expected rate of Legionella and Aspergillus sp causing HAP in this population. A Legionella outbreak was not the explanation, as these cases were diagnosed in 5 different hospitals. The high frequency of adverse outcomes associated with HAP should alert hospitalists to the risk of nosocomial pneumonia in the non-ICU setting.