Author + information
- Received March 3, 1997
- Revision received June 10, 1997
- Accepted June 21, 1997
- Published online October 1, 1997.
- Patricia A. Cowper, PhDAB,*,
- Eric D. Peterson, MD, MPHA,
- Elizabeth R. DeLong, PhDBD,
- James G. Jollis, MD, FACCA,
- Lawrence H. Muhlbaier, PhDBC,
- Daniel B. Mark, MD, MPH, FACCA,
- for the Ischemic Heart Disease (IHD) Patient Outcomes Research Team (PORT) Investigators1
- ↵*Dr. Patricia A. Cowper, Box 3865, Duke University Medical Center, Durham, North Carolina 27710. E-mail: cowpe001@mc. duke.edu.
Objectives. This study examined the impact of early hospital discharge on short-term clinical outcomes of elderly patients treated with coronary artery bypass graft surgery (CABG) in the United States in 1992.
Background. Protocols that encourage earlier discharge of patients who have had CABG have been implemented across the country. Although delivery of efficient care benefits both patients and providers, premature discharge can adversely affect clinical outcomes, resulting in increased hospital readmissions and higher long-term costs.
Methods. We examined the prevalence of early discharge (postoperative length of stay ≤5 days) among 83,347 non–health maintenance organization (HMO) Medicare patients who underwent CABG in the United States in 1992. Using logistic regression models, we identified patient characteristics associated with early discharge and obtained risk-adjusted rates of death and readmission or death for postoperative lengths of stay between 4 and 14 days.
Results. In 1992, 6% of Medicare patients undergoing CABG were discharged within 5 days of the operation. The prevalence of early discharge varied considerably among states, ranging from 1% to 21%. Patients discharged early tended to be younger and male and have fewer comorbid illnesses. Risk-adjusted rates of death and death or cardiovascular readmission were lowest among patients discharged early.
Conclusions. As of 1992, early discharge of elderly patients treated with CABG in non-HMO settings was not associated with higher 60-day rates of death or readmission. This suggests that physicians were able to identify low risk candidates for early discharge. Variation across the nation in early discharge rates, along with the percentage of patients without major risk factors for adverse outcomes, suggests that higher rates of early discharge might be safely achieved.
Health care financing has evolved from fee-for-service to case-based payment and more recently to capitated contracts. Current payment methods force health care providers to accept more of the financial risk associated with treating patients. In response, hospital administrators and physicians have initiated various cost reduction strategies, including those designed to decrease hospital length of stay. Between 1982 and 1992 the average length of stay of Medicare patients in acute care hospitals declined 18% .
Early discharge policies for patients who have had coronary artery bypass graft surgery (CABG) have been implemented in many hospitals across the United States [2–7]. If hospital stays are reduced by improving the efficiency of care, both patients and providers benefit. However, if patients are discharged prematurely, clinical outcomes may be adversely affected, resulting in increased readmissions and higher costs over the long term [8–10]. In this study, we examined the impact of early discharge on short-term clinical outcomes of elderly patients who underwent CABG in the United States in 1992. We selected two outcomes, death and a combined end point of cardiovascular readmission or death within 60 days of discharge, as indicators of overall quality of care.
This study included Medicare patients >64 years old who underwent isolated CABG between January 1, 1992 and October 31, 1992, and were discharged to their home within 14 days of the operation. The bypass episode was defined as the bypass admission plus any transfers to acute care facilities. Patients with postoperative stays >14 days were considered to have especially complicated postoperative courses and were not included in the analysis. In addition, patients discharged to nonacute health care facilities were excluded because they were not likely to have been candidates for early discharge. Finally, we excluded members of health maintenance organizations (HMOs) because their follow-up hospital data were incomplete.
1.2 Data sources
The primary data source for this analysis was the 1992 National Claims History File. For each Medicare discharge, the file contains demographic data, up to 10 International Classification of Diseases, 9th revision, Clinical Modification (ICD-9-CM) diagnostic codes and up to six ICD-9-CM procedure codes. The vital status of patients up to 60 days after discharge was obtained from the Medicare enrollment data base.
1.3 Outcomes assessed
The primary outcome in this study was death within 60 days of discharge after the bypass episode. In an effort to capture poor clinical outcomes that did not result in death, we also examined cardiovascular readmission rates within 60 days of discharge. Because readmission is conditional on survival, we defined the secondary outcome as a combined end point of cardiovascular readmission or death within 60 days of discharge. Hospital admissions were classified as cardiovascular if the diagnosis-related group (DRG) was related to heart disease (DRGs 104–108, 110–112, 115–118, 120–129, 130–133, 138–145, 418, 452, 453 and 483). We also conducted a supplemental analysis that included all readmissions, regardless of DRG.
1.4 Statistical analysis
The distribution of postoperative length of stay (PLOS) and the national prevalence of early discharge (PLOS ≤5 days) were examined using descriptive analyses. We then used logistic regression analysis to identify patient characteristics associated with early discharge. Finally, we used logistic regression analysis to model adverse outcomes within 60 days of discharge after the bypass episode across the spectrum of PLOS’s, after adjusting for the patients’ demographic and clinical characteristics.
The logistic model examining early discharge included explanatory variables that represented demographic characteristics (age, gender, race) and illness severity (acute myocardial infarction [MI], comorbid diseases). Age was specified in the model as a five-level variable to allow for nonlinearity in its relation with early discharge. Acute MI was defined as the presence of a principal diagnosis of acute MI (ICD-9-CM 410) or a principal diagnosis of a complication of acute MI with acute MI as a secondary diagnosis. If the episode of care included more than one hospital admission, diagnoses for the final admission had to include acute MI for the patient to be in this category. Comorbid diseases included chronic pulmonary disease, peripheral vascular disease, diabetes mellitus, cerebrovascular disease, congestive heart failure and cancer, as defined previously using ICD-9-CM codes . The large sample allowed us to control for these diseases individually, rather than combine them in a weighted index. Diseases with a prevalence <1% in our sample were not considered in the model.
The logistic models of adverse outcomes included demographic characteristics, descriptors of illness severity and PLOS. Demographic and clinical characteristics were specified as in the early discharge model described earlier. Postoperative length of stay was specified as a multilevel variable, with a level for each length of stay. This specification, which was feasible with the large sample size, was selected to allow maximal flexibility in determining the relation between PLOS and adverse outcomes.
2.1 Study group
A total of 101,812 elderly patients underwent isolated CABG in non-HMO settings between January and October of 1992, 97,432 (95.6%) of whom were discharged alive. Of these, 93,777 (96%) were discharged home. Eleven percent of patients discharged home had postoperative stays >14 days and were excluded from the analysis, leaving a final sample of 83,347 patients. Patients were predominantly white (94%) and male (68%), with an average age of 72 years. Thirteen percent of patients were admitted to the hospital for an acute MI. The most common comorbid illness was diabetes (21%), followed by congestive heart failure (16%), cerebrovascular disease (7%), peripheral vascular disease (6%) and chronic pulmonary disease (4%). Fifty-five percent of patients had no comorbid diseases coded.
2.2 Rates of death and readmission
One percent of patients (n = 710) died within 60 days of discharge home. Eighteen percent (n = 128) of these deaths occurred within 1 week of discharge. Of those surviving, 18% (n = 15,063) were readmitted to an acute care facility within 60 days of discharge. Fifty-one percent (7,682) of readmissions had a DRG related to cardiovascular disease. Another 13% (n = 1,958) of readmissions had a DRG related to respiratory disease, and 3% (n = 452) had a DRG for cerebrovascular disease.
2.3 Early discharge
The mean (±SD) PLOS was 8 ± 2 days, and almost 60% of patients were discharged between 6 and 8 days after the operation (Fig. 1). Six percent of patients (n = 5,296) were discharged early (i.e., within 5 days of the operation). Logistic regression analysis revealed that patients discharged early tended to be younger, male, have fewer comorbid illnesses and not admitted to the hospital for an acute MI (Table 1). In contrast to other comorbid illnesses in the model, cancer was associated with early discharge. At the state level, mean PLOS ranged from 7 to 9 days. The percentage of patients discharged within 5 days of the operation varied widely among states, from <1% to 21% (Table 2).
2.4 Characteristics associated with outcomes
In logistic regression analysis, patient characteristics associated with increased probability of death within 60 days of discharge were age, diabetes, congestive heart failure, peripheral vascular disease and cerebrovascular disease (Table 3). With the exception of cerebrovascular disease, these factors also were positively related to the combined end point—cardiovascular readmission or death within 60 days of discharge. In addition, cardiovascular readmission or death was positively associated with female gender and chronic pulmonary disease. Hospital admission for acute MI was not related to either the postdischarge death rate or cardiovascular readmission or death rate of patients surviving the index hospital stay.
2.5 Postoperative length of stay and outcomes
After adjusting for patient characteristics, 60-day death and cardiovascular readmission or death rates increased with PLOS. The odds ratios for each PLOS, along with 95% confidence intervals, are presented in Fig. 2(death) and Fig. 3(cardiovascular readmission or death). The reference value in each figure is the mean of 8 days. Patients with shorter than average postoperative stays had lower odds of experiencing an adverse outcome relative to patients with average postoperative stays. Patients staying in the hospital longer than average after the operation had higher odds of an adverse outcome.
The risk-adjusted rates of adverse outcomes within 60 days of discharge are shown in Fig. 4for PLOS’s between 4 and 14 days, along with 95% confidence limits. The curves shown are for a prototypical patient (male patient between 70 and 74 years of age with no comorbid illness). Rates of adverse outcomes tended to increase with PLOS. Adjusted cardiovascular readmission or death rates ranged from 5% to 12% across the spectrum of length of stay. Adjusted death rates increased from 0.2% to 1%. When both cardiovascular and noncardiovascular readmissions were included in the combined end point, the positive relation between readmission or death and PLOS persisted, ranging from 11% to 25% for PLOS’s between 4 and 14 days.
Because risk adjustment using claims data is imperfect and the sample includes patients with a variety of risk factors, we also examined the unadjusted rates of death and cardiovascular readmission or death for a group of low risk patients, specifically those <75 years of age with no comorbid illness. The relation between adverse outcome rates and PLOS found using the risk-adjusted model was maintained in this low risk subset, with patients selected for early discharge having the lowest rates of death and readmission or death (Fig. 5).
Length of stay after CABG has declined significantly in recent years. Shorter postoperative stays have been achieved through advances in anesthetic and surgical techniques, as well as through improvements in the efficiency of care [6, 7, 12]. However, there is concern that recent efforts to reduce PLOS after CABG may compromise quality of care, particularly for elderly patients. This national study provides evidence that, as of 1992, the shorter stays of selected elderly patients treated with CABG in non-HMO settings had not come at the expense of worse patient outcomes. Neither death nor cardiovascular readmission within 60 days of the operation was greater for patients discharged early compared with those with longer hospital stays.
3.1 Characteristics associated with early discharge
The prevalence of early discharge among elderly patients treated with CABG was fairly low in 1992, averaging 6% nationally. However, there was considerable variation across the country, with some areas approaching 20%. Patients more likely to be discharged early were younger, had no comorbid conditions and were less likely to have been admitted to the hospital with an acute MI. The association between cancer and early discharge was unexpected and suggests that patients with cancer who undergo CABG may be a select group of survivors. Alternatively, given the low prevalence of disease (3.9%), the finding may be one of chance. Patients without major risk factors for death and readmission were more likely to be discharged early. However, the proportion of patients without major risk factors for death and readmission far exceeded the national prevalence of early discharge in 1992. For example, 23% of patients were <70 years of age and had no comorbid disease, and another 21% were between 70 and 74 years of age with no comorbid disease.
3.2 Association between PLOS and outcomes
We found that patients with shorter PLOS’s had lower rates of readmission and death within 60 days of discharge than patients with longer stays. This relation may arise from several sources. First, patients discharged early may have been less sick before the operation than those with longer postoperative stays. Although we adjusted for baseline risk factors to the extent possible, it is not possible to account fully for illness severity using claims data . We examined rates of readmission and death for the lowest risk subset of patients (younger patients with no comorbid illness) and found similar results to the adjusted analysis. Second, patients with more complicated postoperative courses have extended hospital stays and are more likely to have problems after discharge. Finally, complications caused by poor quality of care may result in longer lengths of stay and worse clinical outcomes.
Our results support the hypothesis that physicians are able to accurately identify low risk candidates for early discharge. Nevertheless, we cannot rule out the possibility that outcomes were affected somewhat by early discharge practices. For example, it is conceivable that the observed rates of adverse outcomes would have been even lower if patients discharged early had stayed in the hospital longer. On the other hand, the adjusted mortality rate at 60 days among patients discharged early was only 0.2%, and it is very unlikely that this mortality rate would be significantly reduced by lengthening postoperative stays.
3.3 Previous research
Our findings are consistent with recent single-site studies of protocols designed to facilitate the early discharge of bypass patients [6, 10]. Patients treated after the adoption of such protocols, which streamlined care and encouraged early extubation, did not have higher rates of readmission or death than those patients treated before the protocol. Early discharge rates among protocol patients ranged from 41% to 47%.
3.4 Study limitations
Several limitations of this study should be noted. 1) The study was restricted to patients at least 65 years of age, and the results may not be generalizable to younger patients. However, because elderly patients are at higher risk for adverse outcomes, one would expect any adverse effects associated with early discharge to be more easily detectable in this population. 2) Early discharge policies have become more aggressive since 1992, and the proportion of bypass patients discharged within 5 days of the operation has increased dramatically. Although recent studies at hospitals with higher early discharge rates among patients of all ages have found no association between early discharge and short-term mortality or morbidity [6, 9, 10], larger studies that include more hospitals are needed. 3) An examination of outpatient, rehabilitation and nursing home costs and quality of life after CABG was beyond the scope of this study. Patients discharged early may require more ancillary care and consume health care resources outside the acute care setting. Although postoperative care in nonacute care settings may be more beneficial for patients and less expensive overall, the financial burden to the family may be greater. In addition, it is possible that patients discharged early experience more difficult recoveries that require more outpatient care, but do not result in hospital admission or death. A more comprehensive study of early discharge that examines morbidity and quality of life would be helpful in resolving these issues.
This national study suggests that, in 1992, physicians in non-HMO settings were able to target successfully low risk elderly patients for early discharge after CABG. Furthermore, variations in the rate of early discharge among states, along with the percentage of patients without major risk factors for death or rehospitalization, suggest that substantially higher rates of early discharge than those seen in 1992 might be safely achieved. Whether medical practice has achieved or exceeded these maximal safe levels will be critical to evaluate as more recent national data become available.
↵fn1 This study was supported by Research Grants HS-06503 and HS-08805-02 from the Agency for Health Care Policy and Research, Rockville, Maryland and was presented in part at the 45th Annual Scientific Session of the American College of Cardiology, Orlando, Florida, March 1996.
- coronary artery bypass graft surgery
- diagnosis-related group
- health maintenance organization
- International Classification of Diseases, 9th revision, Clinical Modification
- myocardial infarction
- postoperative length of stay
- Received March 3, 1997.
- Revision received June 10, 1997.
- Accepted June 21, 1997.
- The American College of Cardiology
- ↵Prospective Payment Assessment Commission. Medicare and the American health care system. Report to the Congress, June 1993:80.
- Barnes RV,
- Lawton L,
- Briggs D
- Nickerson NJ,
- Murphy SF,
- Kouchoukos NT,
- Daily BB,
- Schechtman KB,
- Davila-Roman VG
- Johnson SH,
- Smith LR,
- Eisenstein E,
- Glower DD,
- Lowe JE,
- Smith PK
- Jonsson B,
- Lindgren B