INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In response to the escalating costs of medical care over the past several years, third-party payers have forced hospitals to reduce hospital beds for inpatient care and to downsize or eliminate certain hospital services (1). For patients, hospital restructuring has meant fewer hospital admissions and shorter lengths of stay even once they are admitted (4). For physicians and other caregivers, decreasing availability of hospital beds has led to increasing pressure to discharge patients quickly for financial reasons or to accommodate others, who may be waiting (3). Policy makers have maintained that these changes have improved efficiency without adversely affecting quality of inpatient care or access. Others, however, have argued that this practice has resulted in patients leaving hospitals before they are medically or socially ready to manage themselves at home (5), contributing to worse outcomes for patients (6). Currently, there is a paucity of evidence to support either position.

Chronic obstructive pulmonary disease (COPD) and asthma are leading causes of death in the United States and are associated with a great deal of morbidity (7), accounting for 1.4 million hospital days per year (8). Because elderly patients with COPD or asthma have frequent comorbid conditions and have a high relapse rate (9), these patients may be particularly vulnerable to the effects of premature discharges. To evaluate the impact of very short lengths of hospitalization on the short-term rates of readmission and mortality, we conducted a population-based study among elderly patients with COPD or asthma in Ontario, Canada, who have been recently hospitalized for their disease. In particular, we wanted to address the question: do elderly patients with very short stays in hospital for obstructive airways disease have higher short-term rates of readmission and mortality compared with those with longer stays?

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Population

We searched the Canadian Institute for Health Information (CIHI) hospital discharge database for patients 65 yr of age or older who had at least one hospitalization for COPD or asthma between April 1, 1992 and March 31, 1997 in Ontario, Canada. We combined the diagnosis of COPD and asthma together because of the similarities in clinical presentation and treatment of these two conditions (10). International Classification of Diseases, 9th revision (ICD-9) codes 490, 491, 492, 496 (7, 10), and ICD-9 code 493 (11, 12) were used to identify COPD and asthma patients, respectively. The accuracy and validity of the CIHI database with regard to COPD and asthma diagnoses have been previously shown (13). For patients who had multiple hospitalizations, their incident (first) admission was chosen as the index hospitalization. Index admission and discharge dates were, therefore, defined as the date on which patients were first hospitalized or discharged for COPD or asthma during the study period, respectively. We excluded a priori patients who died during the index hospitalization.

Identification of Recurrent Hospitalization for COPD or Asthma and Mortality

The CIHI database was also used to determine whether or not patients in the cohort were readmitted for the same pulmonary diagnosis within 15 d of the index discharge date using scrambled unique identifiers contained in this database. If patients were readmitted more than once during this interval, only the first readmission was counted. Mortality information was obtained from the Registered Persons Database (RPDB), which contains updated information on vital status of all Ontario residents. It also records the date of death of the decedent (14).

Other Variables and Comorbidity

Outpatient drug information was captured using the Ontario Drug Benefit (ODB) database. Through the ODB program, the Ontario government provides free outpatient medications to its residents 65 yr of age or older. This database allowed us to identify receipt of various airway medications by patients in the study population. These included inhaled ₂-agonists, anticholinergics, and inhaled as well as oral steroids and theophyllines. Information regarding health service utilization was obtained from the Ontario Health Insurance Plan database, which contains billing information on all "fee for service" outpatient visits, including the specialty of physicians and the type of services that were provided for each patient. This database was used to determine the occurrence of outpatient physician and emergency room visits for COPD and asthma exacerbations within 1 yr before the index hospitalization. Information on health care utilization and use of airway medications within 1 yr before the index hospitalization was used as surrogate markers for airway disease severity.

To control for the effects of comorbidity on outcomes, a modified Charlson comorbidity score was calculated from the 15 secondary diagnosis fields for each patient (15). All of these databases were linked together using a unique scrambled health card number for each individual patient. The reliability and validity of these databases have been documented previously (16, 17).

Data Analysis

Because the optimal duration of hospitalizations for obstructive airway disease has not been well defined, we chose a priori to divide the patients into quartile groups, based on the length of stay during their index hospitalization. We then compared the risk of readmission or mortality within 15 d of discharge in various quartile groups using multivariate logistic regression modeling techniques. The reference category was the quartile group with the shortest hospital stay. The basic model included age, sex, and modified Charlson comorbidity score. We also performed a similar subgroup analysis using absolute days of hospitalization in lieu of quartile groups in order to evaluate the effect of very short stays in hospital (i.e., 1- to 2-d hospitalizations). Moreover, to test the robustness of our model, we performed a similar analysis, using 30-day readmission and mortality rates as the response variable. The 30-d results were very similar to the 15-d results (available from the authors). Thus, only the 15-d results are reported. Using similar methods, we also determined 15-d rate of emergency department visits, which did not result in hospitalization. Linear trends were assessed using Mantel-Haenszel ² test. Relative risks are presented with 95% confidence intervals and reported p values are two-tailed. p Values less than 0.05 were considered statistically significant. All analyses were performed with SAS software, release 6.12 (SAS Institute, Cary, NC).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We identified 34,413 elderly residents of Ontario, Canada, with COPD or asthma. Of these, 2,029 were excluded because they died during their index hospitalization. The average age of the remaining 32,384 patients was 75.5 ± 7.0 yr and of these, 49% were women. There were 489 (1.5%) deaths and 1,552 (4.8%) readmissions, which occurred within the first 15 d after discharge. The median length of hospitalization was 6 d (range, 1 to 90).

Patient characteristics divided according to the quartile of hospital duration are shown in Table 1. Patients in the lowest quartile were the youngest, had the lowest modified Charlson comorbidity score, and the lowest rates of drug utilization and outpatient physician visits for COPD/asthma exacerbation before the index hospitalization compared with the rest of the other groups (p < 0.001; comparison between the lowest quartile and each of the other quartile groups, separately), suggesting that these patients had the lowest severity of airway disease.

Overall, patients in the second quartile group had the lowest rates of readmission and mortality, after adjusting for age, sex, and comorbidity (Table 2). Patients in the highest quartile group had a similar 15-d readmission rate compared with those in the lowest quartile group (relative risk [RR], 0.95; 95% confidence interval [CI], 0.83 to 1.09). The 15-d mortality rate, on the other hand, was slightly greater in the highest quartile group compared with the lowest quartile group, which suggests that patients in the highest quartile group had greater severity of disease. However, we cannot rule out an alternate possibility that extremely long stays in the hospital may increase the short-term risk for mortality.

The 15-d risk for mortality as well as readmission was highest in patients who stayed in hospital for less than 2 d (Figures 1 and 2). The shape of the relationship between length of hospitalization and 15-d rates for readmission as well as mortality was U-shaped. Patients with less than 2-d stays in hospital had the highest 15-d readmission rate with the rate falling progressively with increasing length of stay, until Day 6 was reached; for mortality, the lowest rate was observed for patients who stayed in hospital for 7 d. Some patients staying longer than 5 d experienced a slightly elevated risk for readmission and mortality compared with those who stayed 2 to 5 d in hospital. However, compared with those with hospital stay of less than 2 d, the RR for short-term readmission was still below 1, even among patients with hospital stays of  10 d (RR = 0.67; 95% CI, 0.54 to 0.83), suggesting that hospital stay of less than 2 d was an important determinant of morbidity.

View larger version (12K): [in this window] [in a new window]   Figure 1.   RR of readmission within 15 d after discharge according to length of stay during index hospitalization in elderly patients with obstructive airway disease. RR are adjusted for age (continuous), sex, comorbidity.

View larger version (12K): [in this window] [in a new window]   Figure 2.   RR of mortality within 15 d after discharge according to length of stay during index hospitalization in elderly patients with obstructive airway disease. RR are adjusted for age (continuous), sex, comorbidity.

There was also an inverse relationship between length of hospital stay and 15-d rate of visits to emergency departments. The highest rate was observed among those who stayed less than 2 d in hospital (2.6%). The rate progressively decreased until a nadir was reached at Day 5, where the rate was only 1.6%. A slight increase in the rate was observed, beyond Day 8 (1.9%). To determine whether or not discharge to chronic care facilities modified the relationship between readmission and length of stay, we performed a similar analysis excluding patients transferred to these facilities. However, this did not result in significant changes to the overall results (RR = 0.71 [95% CI, 0.59 to 0.86] comparing 4 to 6 d with 3 or less d).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Our data support the hypothesis that very short hospital stays for obstructive airway disease in the elderly population significantly increase the risk for recurrent hospitalization and mortality within the first 2 wk after discharge from hospital. Those patients who stayed in hospital for fewer than 4 d were 39% more likely to be readmitted to hospital and 45% more likely to die in the first 2 wk postdischarge when compared with those who stayed for 4 to 6 d, after controlling for age, sex, and comorbidity. The highest risk was found among patients who stayed in hospital for less than 2 d. These patients had a 69% excess risk for readmission and were 2.08 times more likely to die during the first 15 d postdischarge compared with those who stayed in hospital for 2 d. The risk for readmission and mortality declined progressively until length of stay of greater than 5 d was reached.

Length of stay beyond 5 or 6 d was associated with a slight increase in the RR for short-term readmission and mortality, producing a relationship between length of stay and these outcome variables that was U-shaped. Because length of stay is a known marker for pulmonary disease severity, the slight excess risk in patients who experienced a prolonged stay in hospital is likely to be explained on the basis of disease severity (18, 19). Indeed, patients in the highest quartile of hospital days had the highest frequency of prior physician visits and the greatest utilization of inhaled ₂-agonists, and oral corticosteroids, suggesting a higher burden of airway disease. Thus, the inverse association between length of stay in hospital and the risk for readmission and mortality observed in the present study is likely to be an underestimation of the effect of very short hospital stays on these outcomes.

Previous studies addressing this topic have produced inconsistent results. Our study results are similar to those of Rushworth and Rob, who reported that asthmatic patients with a length of stay of more than 1 d were 60% less likely to experience an early readmission compared with those who stayed 1 d or less (20). We extend these findings by showing that length of stay may also affect early mortality and that the relationship between length of stay and readmission/mortality rates is U-shaped.

Our observations differ from that of Harrison and coworkers (21), who showed that decreasing the length of stay in hospital for asthma or bronchitis was not associated with a significant increase in rates of readmission or physician visits within 30 d postdischarge. However, this study was analyzed at the level of the hospital; thus, it may have lacked sufficient statistical power to detect subtle but important effects of hospital duration on these outcomes. Moreover, their study may not have adequately controlled for differences in disease severity between those with short and long hospital stays. This is important because Mushlin and coworkers have shown that COPD patients with hospital stays  5 d had higher baseline PCO₂, more symptoms and comorbidity, and had greater duration of illness than those with < 5 d of hospital stay (19).

Our observations suggest that the highest risk for readmission and possibly mortality occurs in patients who have less than 2 d of hospitalization. The reasons for this are not clear. However, in a related condition, community-acquired pneumonia, 2 d of in-hospital care was generally required to achieve overall clinical stability (22); a similar time frame may also be required for the management of obstructive airway disease. Thus, very early discharge from hospital may be a marker for suboptimal quality of inpatient care. In one study, Mushlin and coworkers have demonstrated that elderly patients with obstructive airways disease usually required between 5 to 8 d of inpatient care for optimal control of their disease (19). Older patients and those with psychosocial problems may require longer hospital stays to minimize morbidity (23). Thus, it is possible that those who were discharged very early may not have been physically or socially ready to cope with their disease at home.

Weingarten and coworkers (24) have shown that postoperative patients with very short hospital stays required greater utilization of chronic care and rehabilitation facilities after discharge compared with those with intermediate lengths of stay because many of these patients were not ready for discharge. In the present study, those with very short stays in hospital also had higher rates for emergency room visits, suggesting that these patients did not have optimal control of their disease at the time of discharge. Williams and Fitton have suggested that 50 to 60% of unplanned readmissions may be preventable through better inpatient care, discharge planning, and follow-up (25). To accomplish this, however, an extra day or two of inpatient care may be necessary. Indeed, among a group of Medicare patients from Massachusetts, there was a 10% increase in the 14-d readmission rate between 1982 to 1986, which was associated with a 25% reduction in the overall hospital duration (26).

One of the strengths of our study design was that by using administrative databases, we were able to track patients in all hospitals across Ontario for readmissions, which avoided the potential problem of losing patients who were readmitted to a hospital different than the one for the index admission. Moreover, since ours was a population-based study, these results may be more generalizable than those from previous studies (19, 27), which came from a limited number of centers. However, because we did not measure the process of inhospital care, we could not identify what components of care were adversely affected by short stays in hospital. Previous studies have suggested patient education and careful follow-up of medical care may be compromised with premature discharges from hospital (27, 28). Another potential limitation was that we could not distinguish unplanned from planned readmissions to hospital. However, because rates for both short-term emergency visits and mortality were affected in a similar way by very short stays, it is likely that most of these readmissions were unplanned. We also observed a slight increase in the RR for readmission after Day 5; based on our data, we cannot rule out the possibility that prolonged stays in hospital may actually increase the risk for readmission from complications that may arise during the index hospitalization. Moreover, because we could not differentiate accurately those with emphysema from those with chronic bronchitis, we could not determine whether these conditions modified the relationship between length of stay and short-term outcomes.

In the 1980s and 1990s, there has been a movement to increase the efficiency of inpatient care by decreasing hospital stays. Our study results suggest that some elderly patients with very short stays in hospital, especially those with less than two hospital days for obstructive airway disease, may be experiencing worse short-term outcomes compared with those with longer stays. Because early readmissions to hospital are expensive, accounting for 25% of Medicare expenditures for inpatient care (29), it may be more cost-effective, in certain cases, to keep patients in hospital for an additional day or two to optimize their condition and to ensure that adequate follow-up can be provided. Our study was not designed to evaluate the potential causes for worse outcomes in patients with short stays. Future research should focus on ways of optimizing the quality of care delivered to these patients and determining the minimally acceptable length of stay for elderly patients with obstructive airway disease.