In-Hospital Observation After Antibiotic Switch in Pneumonia: A National Evaluation

The American Journal of Medicine (2006) 119, 512-518 CLINICAL RESEARCH STUDY In-Hospital Observation After Antibiotic Switch in Pneumonia: A National Evaluation Ramesh V. Nathan, MD, a David C. Rhew, MD, a,b,c,d Cynthia Murray, PhD, e Dale W. Bratzler, DO, MPH, e Peter M. Houck, MD, f Scott R. Weingarten, MD, MPH b,c,d a Division of Infectious Diseases, Department of Medicine, Veterans Affairs Greater Los Angeles Healthcare System, Los Angeles, Calif; b Zynx Health Incorporated, Los Angeles, Calif; c Department of Health Services Research, Cedars-Sinai Medical Center, Los Angeles, Calif; d The David Geffen School of Medicine at UCLA, Los Angeles, Calif; e Oklahoma Foundation for Medical Quality, Oklahoma City, Okla; and f Centers for Medicare and Medicaid Services, Seattle, Wash. ABSTRACT PURPOSE: To evaluate the clinical benefit of in-hospital observation after the switch from intravenous (IV) to oral antibiotics in a large Medicare population. Retrospective studies of relatively small size indicate that the practice of in-hospital observation after the switch from IV to oral antibiotics for patients hospitalized with community-acquired pneumonia (CAP) is unnecessary. METHODS: We performed a retrospective examination of the US Medicare National Pneumonia Project database. Eligible patients were discharged with an ICD-9-CM diagnosis consistent with communityacquired pneumonia and divided into 2 groups: 1) a not observed cohort, in which patients were discharged on the same day as the switch from IV to oral antibiotics and 2) an observed for 1 day cohort, in which patients remained hospitalized for 1 day after the switch from IV to oral antibiotics. We compared clinical outcomes between these 2 cohorts. RESULTS: A total of 39,242 cases were sampled, representing 4341 hospitals in all 50 states and the District of Columbia. There were 5248 elderly patients who fulfilled eligibility criteria involving a length of stay of no more than 7 hospital days (2536 not observed and 2712 observed for 1 day patients). Mean length of stay was 3.8 days for the not observed cohort and 4.5 days for the observed for 1 day cohort (P.0001). There was no significant difference in 14-day hospital readmission rate (7.8% in the not observed cohort vs 7.2% observed for 1 day cohort, odds ratio 0.91; 95% confidence interval [CI] 0.74-1.12; P.367) and 30-day mortality rate (5.1% not observed cohort vs 4.4% in the observed for 1 day cohort, odds ratio 0.86; 95% CI, 0.67-1.11; P.258) between the not observed and observed for 1 day cohorts. CONCLUSIONS: Our analysis of the US Medicare Pneumonia Project database provides further evidence that the routine practice of in-hospital observation after the switch from IV to oral antibiotics for patients with CAP may be avoided in patients who are clinically stable although these findings should be verified in a large randomized controlled trial. 2006 Elsevier Inc. All rights reserved. KEYWORDS: Pneumonia; Antibiotics; Hospital; Health services research Community-acquired pneumonia (CAP) is the sixth leading cause of death in the United States and accounts for approximately 10 million physician visits, 500,000 hospitalizations, and 45,000 deaths annually. 1 It has been estimated Requests for reprints should be addressed to David C. Rhew, MD, Zynx Health Inc., a subsidiary of the Hearst Corporation, 10880 Wilshire Blvd., Suite # 1450, Los Angeles, CA 90024. E-mail address: drhew@zynx.com that the total annual cost of treating CAP in the United States is $9.7 billion. 2 The initial management of patients hospitalized for the treatment of CAP typically includes the use of intravenous (IV) antibiotics. 2 Although clinical improvement is often seen during the initial 3 days of IV antibiotic treatment, many patients with CAP are treated in-hospital for 7 days. 3 Several clinical trials have demonstrated that hospital length-of-stay (LOS) may be safely reduced for patients with CAP by applying strategies that 0002-9343/$ -see front matter 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.amjmed.2005.09.012

Nathan et al In-Hospital Observation after Antibiotic Switch in Pneumonia 512.e2 promote the early switch from IV to oral antibiotics followed by discharge from the hospital. 1,4 7 In some of these early switch studies, it has been recommended that patients receive 1 day of in-hospital observation after the switch from IV to oral antibiotics. 6,7 However, further analyses have demonstrated that there does not appear to be a clinical benefit to in-hospital observation after switching from IV to oral antibiotics and that discharging patients on the day of the switch from IV to oral antibiotics is safe. 6,8 10 Small sample size and lack of postdischarge follow up are 2 important limitations of the previous studies. For example, all of the studies addressing the topic of inhospital observation in CAP enrolled 200 or fewer patients per study and were performed at a single institution. 6,8 10 Also, only 2 of these studies 6,8 performed postdischarge follow-up to determine whether or not patients were readmitted to the hospital or died within 30 days after discharge. In our current study, we analyze both inpatient and postdischarge follow-up data from a large, nationally representative database of Medicare patients hospitalized with CAP. CLINICAL SIGNIFICANCE METHODS Subjects and Sample Selection The National Pneumonia Project used Medicare fee-forservice hospital claims to identify cases of potential pneumonia hospitalization. A case was defined by a claim with a principal diagnosis of pneumonia (International Classification of Disease, Ninth Edition, Clinical Modification [ICD-9-CM] codes 480.0 through 483.8, 485 through 486 or 487.0) or a principal diagnosis of septicemia or respiratory failure (ICD-9-CM codes 038.XX or 518.81), and a secondary diagnosis of pneumonia. For Medicare programmatic reasons, claims in each state were sampled during one of two 6-month periods: July 1 through December 31, 1998 and September 1, 1998 through March 31, 1999. Nationally, there were 346,105 cases during these time periods. A systematic random sample of up to 850 cases was selected from each state, resulting in an original database with 39,242 cases. Data Collection Hospitals sent photocopies of medical records to 1 of 2 clinical data abstraction centers (CDACs). Abstractors used computerized tools with explicit predefined entry criteria to record data elements that included patient demographics and clinical characteristics, antibiotic selection, and timing of antibiotic administration. The definition of days of observation after switch was based on calendar days and not There does not appear to be a significant clinical benefit to in-hospital observation of patients with communityacquired pneumonia after switching them from intravenous to oral antibiotics. Earlier hospital discharge of patients with community-acquired pneumonia would greatly reduce treatment costs. hours. Data quality was monitored monthly through inter- CDAC reliability testing on a sample of records. Overall, inter-cdac reliability averaged 92%. Inter-CDAC agreement on administration of antibiotics within 4 hours of arrival was 91% with a kappa coefficient of 0.80. We used Medicare enrollment data to detect deaths and Medicare Part A claims to identify 14-day readmission data and patient destination at discharge. Patient Enrollment After collecting data from medical records, we applied multiple exclusion criteria to produce an analytic database representing 5248 hospitalizations (Figure). Medical record abstraction was terminated if the patient had no working diagnosis of pneumonia at the time of admission, no radiographic evidence of pneumonia, was aged younger than 65 years or age was not recorded in the chart, or had an immunocompromised condition (receiving corticosteroids or antineoplastic therapy, or history of organ transplantation, leukemia, lymphoma, or HIV infection). Patients also were excluded if they were transferred or discharged to another hospital, admitted for comfort/palliative care only, hospitalized within the past 14 days, died or left against medical advice, received no antibiotics during the first 36 hours at the hospital, or were hospitalized in the Virgin Islands. Multiple hospitalizations were excluded when the 30-day mortality rates were calculated. Patients who had complicated pneumonia were excluded from the study. These patients included those who received 2 or more courses of antibiotics during the hospitalization (implying the presence of another infection or a difficultto-treat initial infection) and those whose last dose of oral antibiotic during hospitalization was received more than 2 days before discharge. Patients who remained in the hospital more than 1 day after the switch from IV to oral antibiotics were not evaluated, because we assumed that these patients were more likely to remain hospitalized due to a medical or social reason rather than for the purpose of in-hospital observation after the switch from IV to oral antibiotics. Additional exclusion criteria were applied to identify patients who were not eligible for the switch from IV to oral antibiotics. These criteria included having the last IV antibiotic discontinued after the last oral antibiotic was initiated (due to the overlap of IV and oral antibiotics), not receiving an oral antibiotic either during hospitalization or upon discharge, not receiving an IV antibiotic during hospitalization, receiving one or more antibiotics with unknown route or with missing dates of administration, having the last IV antibiotic discontinued more than 1 day before the initiation of the last oral antibiotic (ie, gap between stopping IV and

512.e3 The American Journal of Medicine, Vol 119, No 6, June 2006 Figure Patient enrollment

Nathan et al In-Hospital Observation after Antibiotic Switch in Pneumonia 512.e4 Table 1 Patient Demographics Observed for 1 day (n 2712) Not observed (n 2536) P value Age.585 Mean 79.8 79.7 Median 80 80 Range 65-110 65-105 Race.108 White 2401 (88.5%) 2280 (89.9%) African-American 142 (5.2%) 124 (4.9%) Other 165 (6.1%) 132 (5.2%) Unable to determine 4 (0.1%) 0 (0.0%) Sex.084 Male 1302 (48.0%) 1278 (50.4%) Female 1410 (52.0%) 1258 (49.6%) Pneumonia Severity Index Class.780 II 238 (8.8%) 222 (8.8%) III 708 (26.1%) 682 (26.9%) IV 1286 (47.4%) 1168 (46.1%) V 480 (17.7%) 464 (18.3%) Quality Indicators First antibiotics within 4 hours of arrival 1595 (58.8%) 1546 (61.0%).112 Empiric antibiotics within 24 hours of arrival 2261 (83.4%) 2075 (81.8%).139 Pre-arrival setting Home 1648 (60.8%) 1532 (60.4%) Post-ambulatory surgery 0 (0.0%) 1 (0.0%) Non-institutional setting 69 (2.5%) 49 (1.9%) Transfer from another ER 21 (0.8%) 14 (0.6%) SNF/ECF/ICF 513 (18.9%) 546 (21.5%) Outpatient setting 417 (15.4%) 344 (13.6%) Secured facility 0 (0.0%) 0 (0.0%) Homeless/shelter 1 (0.0%) 1 (0.0%) Rehabilitation hospital 6 (0.2%) 4 (0.2%) Psychiatric hospital 3 (0.1%) 1 (0.0%) Transitional or stepdown care 1 (0.0%) 1 (0.0%) Unable to determine 33 (1.2%) 43 (1.7%) SNF Skilled nursing facility; ECF extended care facility; ICF intermediate care facility. starting oral antibiotics), and patients who had received no antibiotic during hospitalization. Furthermore, patients with total length-of-stay (LOS) of 1 day were excluded because we assumed that many of these patients could potentially have been treated with oral medications in the outpatient setting, 11 and there likely would have been no perceived need for in-hospital observation after the switch from IV to oral antibiotics for these patients. Patients were then divided into 2 groups: an observed for 1 day cohort that was switched from IV to oral antibiotics and monitored in the hospital for 1 day, and a not observed cohort that was switched from IV to oral antibiotics and immediately discharged from the hospital or discharged with instructions to start oral antibiotics after leaving the hospital. Finally, we attempted to exclude patients who were unstable on the day of the switch from IV to oral antibiotics (ie, the day of discharge for patients in the not observed group) by excluding patients with total LOS 7 days. We assumed that patients with LOS 7 days were more likely to have an unstable condition at the time of IV to oral antibiotic switch as compared with those having LOS 7 days. 12,13 Statistical Analyses We abstracted data on all the patients with LOS 2-7 days and analyzed the 2-6-day patients in the not observed groups and the 3-7-day patients in the observed for a day group. Statistical analysis was performed using SAS software program (SAS version 8.2, SAS Institute, Cary, NC). The 2-tailed t test was used to compare means between 2 groups (age, length of stay). Chi-square test or a Fisher s exact test was used to compare proportions between 2 groups (race, sex, severity score, readmission rates, and mortality rates). Logistic regression was used as a multivariate method to compute adjusted odds ratios with 95% confidence intervals for readmission rates and mortality rates adjusted for pneumonia severity index (PSI) scores. 14 RESULTS Patient Demographics Table 1 shows the demographics of the 2 study populations. There was no statistical difference between the observed

512.e5 The American Journal of Medicine, Vol 119, No 6, June 2006 Table 2 Primary Study Endpoints Observed for 1 day (n 2712) Not observed (n 2536) P value Length of Stay (days) Mean 4.5 (95% CI 4.4-4.5) 3.8 (95% CI 3.7-3.8).0001 Median 4 4 Range 3-7 2-6 Number of days receiving IV antibiotics Mean 3.28 (95% CI, 3.23-3.33) 3.58 (95% CI, 3.52-3.62).0001 Median 3 3 Range 1-6 1-6 14-day readmission rate 195 (7.2%) 199 (7.8%) OR 0.91 (95% CI 0.74-1.12); adj OR 0.91 (95% CI 0.74-1.12)*; P.367 30-day mortality rate 120 (4.4%) 129 (5.1%) OR 0.86 (95% CI 0.67-1.11); adj OR 0.86 (95% CI 0.67-1.11)*; P.258 Discharge Destination Home 1851 (68.3%) 1654 (65.2%) Non-institutional setting 52 (1.9%) 39 (1.5%) SNF/ECF/ICF 709 (26.1%) 739 (29.1%) Outpatient setting 3 (0.1%) 0 (0.0%) Secured facility 1 (0.0%) 0 (0.0%) Homeless/shelter 1 (0.0%) 0 (0.0%) Rehabilitation hospital 32 (1.2%) 20 (0.8%) Psychiatric hospital 2 (0.1%) 1 (0.0%) Transitional or stepdown care 40 (1.5%) 51 (2.0%) Unable to determine 21 (0.8%) 32 (1.3%) *Odds ratio adjusted for Pneumonia Severity Score. 14 for one day and not observed groups with regard to age, ethnicity, PSI risk class, or receipt of antibiotics within 4 hours of arrival. Outcomes Table 2 shows the primary endpoints assessed in this study and demonstrates that there were no significant differences between the observed for 1 day and not observed groups with respect to readmission or mortality rates. The 14-day readmission rate was not statistically significant for the observed for one day cohort, as compared with the not observed cohort (7.2% vs 7.8%, odds ratio [OR] 0.91; 95% CI, 0.74-1.12; P.367). The odds ratios were similar after adjusting for PSI scores. The 30-day mortality rate was not significantly lower for the observed for 1 day cohort, as compared with the not observed cohort ( observed for 1 day vs not observed: 4.4% vs 5.1%, OR 0.86; 95% CI, 0.67-1.11; P.258). After adjustment for PSI scores, the OR for 30-day mortality remained unchanged. Readmission rates at 14 days and mortality rates at 30 days did not differ between the 2 cohorts when data were additionally analyzed to include only patients who had received antibiotics within 4 hours (and an empiric antibiotic within 24 hours) of hospital arrival (data not shown). DISCUSSION Our retrospective review of the Medicare National Pneumonia Project database represents the largest evaluation of in-hospital observation after the switch from IV to oral antibiotics in patients with CAP. The study demonstrates that mortality and readmission rates for patients with CAP who do not receive in-hospital observation after the switch from IV to oral antibiotics (with LOS 2-6 days) are not significantly different, as compared with outcomes in patients who receive 1 day of in-hospital observation (with LOS 3-7 days). However, the observed non-significant trend towards lower 30-day mortality in the cohort receiving in-hospital observation indicates that a clinical benefit with in-hospital observation after a switch from IV to oral antibiotics cannot be definitively ruled out with our study. Nonetheless, these findings are consistent with those from smaller retrospective studies 6,8 10 in that there does not appear to be a clinical benefit with in-hospital observation after the switch from IV to oral antibiotics. The economic implications of early discharge are substantial. One study projected a savings of $680 per patient with a 1-day reduction in hospital stay. 15 In our current study, 2712 patients (6.9% of the sample size) were observed for 1 day. Assuming that all of these patients were to be discharged without in-hospital observation, the potential cost savings would total over $1.8 million. Our study evaluated the potential opportunity to avoid in-hospital observation after the switch from IV to oral antibiotics in a random sample of 39,242 Medicare recipients between April 1998 and March 1999. If this strategy were to be implemented nationally to all Medicare patients, we project an annual savings of $27.1 million.

Nathan et al In-Hospital Observation after Antibiotic Switch in Pneumonia 512.e6 The large sample size in both groups ensures that power should not be a concern. The rates for mortality and readmission between both groups are not statistically significant; the power for the difference in our readmission rates between the 2 groups, using continuity, is 14.8%. For 80% power, using the readmission rate of 7.8% for the notobserved group, the observed group would have needed a readmission rate of approximately 5.9% to detect the difference as being significant. The power for the difference in our mortality rates, using continuity correction, is 20.6%. Our study had several limitations. First, because our definition of days of observation after switch was based on calendar days and not hours, it was possible that some patients who remained in the hospital for the same number of hours after the switch may have been classified in different groups (eg, no observation, observation for 1 day, observation for 1 day ). However, generally speaking, because most patients are usually switched from intravenous to oral antibiotics in the morning or during the day, the number of instances in which inconsistent classifications occurred was minimized. Moreover, basing the classification of in-hospital observation on calendar days rather than on hours provided a closer correlation to the actual cost of care. Second, similar to previous studies, the current study was limited by the fact that it was not a prospective, randomized controlled trial. Therefore, we were unable to determine if the physician s decision to observe or not observe the patient in-hospital was based on undocumented factors. For instance, we could not precisely determine which patients required continued hospitalization at the time of switch. We attempted to address this limitation by restricting the analysis to patients who were not observed versus those observed for 1 day. Furthermore, we assumed that patients who remained hospitalized for more than 7 days likely had medical conditions other than CAP contributing to the need for ongoing hospitalization. Therefore, we only evaluated data from patients with a LOS 7 days. It is possible that we may have excluded some patients with LOS 7 days who had no reasons for continued hospitalization after switch, although the total number of these patients was likely not large. Also, we were unable to determine if patients were clinically stable at the time of discharge. In order to address this, we assumed that if a patient was to be discharged, then the patient would be clinically stable at the time of discharge. It is possible that some patients may have been prematurely discharged from the hospital, which would have increased the risk of disease recurrence and death. 16,17 However, it is uncertain whether patients who were discharged without in-hospital observation experienced a greater incidence of clinical instability at the time of discharge, because the decision to observe versus not observe was always based on the clinician s decision and not influenced by a study protocol or guideline. Another limitation to our study was that we applied different lengths of stay for the 2 comparison arms. The not observed cohort had a length of stay of 2-6 days, whereas the observed for one day cohort had a length of stay of 3-7 days. We chose not to analyze cohorts with the same upperbound length of stay (eg, 2-6 days for the not observed cohort vs 3-6 days for the observed for 1 day cohort), because this would potentially have excluded from the observed for 1 day cohort those patients who had a length of stay that was 1 day longer than the upper bound. Because patients with longer lengths of stay are generally sicker than those with shorter lengths of stay, excluding patients with a length of stay that was 1 day longer than the upper bound from the observed for 1 day cohort would have made the observed for 1 day cohort appear less ill and would have biased the results in favor of this cohort. Similarly, we chose not to perform analysis on cohorts with the same lowerbound length of stay (eg, 2-6 days for the not observed cohort vs 2-7 days for the observed for 1 day cohort), because this would have potentially excluded from the not observed cohort those patients who had a length of stay that was 1 day less than the lower bound. Excluding patients with a length of stay 1 day less than the lower bound from the not observed cohort would have made the not observed cohort appear sicker and would again have biased the results in favor of the observed for 1 day cohort. Patients in our study essentially fell into 3 groups: 1) those who were not observed; 2) those who were observed for 1 day for the purpose of observation; and 3) those who were observed for 1 day to address active medical or social reasons. The purpose of our analysis was to compare Group 1 versus Group 2 patients. However, it is likely, despite our extensive efforts to exclude these patients, that some Group 3 patients were included in the final analysis and that their data were combined with Group 2 patients to form the observed for 1 day group. It remains uncertain as to what percentage of the observed for 1 day group came from Group 3 and whether Group 3 patients actually had worsened outcomes (because of greater severity of illness) or had similar or improved outcomes (because they received inhospital observation), as compared with Group 1 patients. Consequently, our inability to identify and exclude all Group 3 patients from the large Medicare database may have confounded our results. On the other hand, patients in the observed for 1 day group received fewer days of IV antibiotics, as compared with patients in the not observed group. These data indicate that patients in the observed for 1 day group were not sicker than patients in the not observed group, because if they were sicker, then we would have expected that the number of days of IV antibiotics would be greater in the observed for 1 day group, as compared with the not observed group. Consequently, there would be even less reason to observe these patients in-hospital after the switch from IV to oral antibiotics. Moreover, these data may explain why the observed for 1 day patients apparently remained in the hospital for 0.7 days, rather than 1 day, longer than patients in the not

512.e7 The American Journal of Medicine, Vol 119, No 6, June 2006 observed group. A potential trade-off may have occurred, in which patients who were to be discharged without in-hospital observation received a longer course of IV antibiotics as compared with patients who were to receive in-hospital observation. Finally, although our study was performed in an elderly population (mean age 80 years), our findings should not necessarily be limited to this older cohort. Age remains the most significant risk factor for mortality, 14 and younger patients are less likely to die from pneumonia than their older counterparts. Consequently, if we had performed the current study in a younger cohort of patients, it would have been even less likely than in the current study (in which no statistically significant difference was seen) to observe a significant difference in mortality between the Not observed and Observed for 1 day groups. Despite these limitations, our study is noteworthy in that it addresses the issue of in-hospital observation in a diverse population of patients with community-acquired pneumonia and varying degrees of severity of illness. In fact, nearly 18% of patients in the current study had a Pneumonia Severity Index class V (ie, expected 30-day mortality of 27%-31%). 14 Furthermore, the implications of this study are profound in that many more patients could potentially be discharged from the hospital 1 day earlier than current practice. The potential benefits of treating patients at home with oral antibiotics versus in-hospital may involve a lower risk for nosocomial infections 18 and improved patient satisfaction. 19 In conclusion, the current study represents the largest assessment of in-hospital observation for patients hospitalized with CAP. Our findings are consistent with the findings of smaller studies in that there does not appear to be a significant clinical benefit to in-hospital observation after the switch from IV to oral antibiotics for patients with CAP. However, due to the observed non-significant trend towards lower 30-day mortality in the cohort receiving in-hospital observation, a clinical benefit with in-hospital observation after the switch from IV to oral antibiotics cannot definitively be ruled out with our study. The findings in our study should be verified in a large randomized controlled trial. ACKNOWLEDGMENTS The analyses upon which this publication is based were performed under Contract Number 500-02-OK-03, funded by the Centers for Medicare & Medicaid Services, an agency of the US Department of Health and Human Services. The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products or organizations imply endorsement by the US Government. The author assumes full responsibility for the accuracy and completeness of the ideas presented. 4-PNEJA-OK-1-05. References 1. Rhew DC, Tu GS, Ofman J, Henning JM, Richards MS, Weingarten SR. Early switch and early discharge strategies in patients with community-acquired pneumonia: a meta-analysis. Arch Intern Med. 2001; 161:722-727. 2. Lave JR, Lin CJ, Fine MJ, Hughes-Cromwick P. The cost of treating patients with community-acquired pneumonia. Semin Respir Crit Care Med. 1999;20:189-197. 3. Fine MJ, Pratt HM, Obrosky DS, et al. Relation between length of hospital stay and costs of care for patients with community-acquired pneumonia. Am J Med. 2000;109:378-385. 4. Marrie TJ, Lau CY, Wheeler SL, Wong CJ, Vandervoort MK, Feagan BG. A controlled trial of a critical pathway for treatment of community-acquired pneumonia. CAPITAL Study Investigators. Community- Acquired Pneumonia Intervention Trial Assessing Levofloxacin. JAMA. 2000;283:749-755. 5. Ramirez JA, Vargas S, Ritter GW, et al. Early switch from intravenous to oral antibiotics and early hospital discharge: a prospective observational study of 200 consecutive patients with community-acquired pneumonia. Arch Intern Med. 1999;159:2449-2454. 6. Ramirez JA, Srinath L, Ahkee S, Huang A, Raff MJ. Early switch from intravenous to oral cephalosporins in the treatment of hospitalized patients with community-acquired pneumonia. Arch Intern Med. 1995; 155:1273-1276. 7. Weingarten SR, Riedinger MS, Hobson P, et al. Evaluation of a pneumonia practice guideline in an interventional trial. Am J Respir Crit Care Med. 1996;153:1110-1115. 8. Rhew DC, Hackner D, Henderson L, Ellrodt AG, Weingarten SR. The clinical benefit of in-hospital observation in low-risk pneumonia patients after conversion from parenteral to oral antimicrobial therapy. Chest. 1998;113:142-146. 9. Dunn AS, Peterson KL, Schecter CB, Rabito P, Gotlin AS, Smith LG. The utility of an in-hospital observation period after discontinuing intravenous antibiotics. Am J Med. 1999;106:6-10. 10. Beumont M, Schuster MG. Is an observation period necessary after intravenous antibiotics are changed to oral administration? Am J Med. 1999;106:114-116. 11. Atlas SJ, Benzer TI, Borowsky LH, et al. Safely increasing the proportion of patients with community-acquired pneumonia treated as outpatients: an interventional trial. Arch Intern Med. 1998;158:1350-1356. 12. Halm EA, Fine MJ, Marrie TJ, et al. Time to clinical stability in patients hospitalized with community- acquired pneumonia: implications for practice guidelines. JAMA. 1998;279:1452-1457. 13. Angus DC, Marrie TJ, Obrosky DS, et al. Severe community-acquired pneumonia: use of intensive care services and evaluation of American and British Thoracic Society diagnostic criteria. Am J Respir Crit Care Med. 2002;166:717-723. 14. Fine MJ, Auble TE, Yealy DM, et al. A prediction rule to identify low-risk patients with community-acquired pneumonia. N Engl J Med. 1997;336:243-250. 15. Fine MJ, Pratt HM, Obrosky DS, et al. Relation between length of hospital stay and costs of care for patients with community-acquired pneumonia. Am J Med. 2000;109:378-385. 16. Halm EA, Fine MJ, Kapoor WN, Singer DE, Marrie TJ, Siu AL. Instability on hospital discharge and the risk of adverse outcomes in patients with pneumonia. Arch Intern Med. 2002;162:1278-1284. 17. Kosecoff J, Kahn KL, Rogers WH, et al. Prospective payment system and impairment at discharge. The quicker-and-sicker story revisited. JAMA. 1990;264:1980-1983. 18. Gomez J, Esquinas A, Agudo MD, et al. Retrospective analysis of risk factors and prognosis in non-ventilated patients with nosocomial pneumonia. Eur J Clin Microbiol Infect Dis. 1995;14:176-181. 19. Coley CM, Li YH, Medsger AR, et al. Preferences for home versus hospital care among low-risk patients with community-acquired pneumonia. Arch Intern Med. 1996;156:1565-1571.