Reducing Length of Stay Increases Cardiac Patient Readmissions

Abstract & Commentary

This study was conducted to determine whether utilization reviews that lead to reduced hospital length of stay (LOS) relative to days requested by an attending physician affect the likelihood of readmission among privately insured patients. Data were obtained from a private insurance company on utilization management (UM) decisions from 1989 through 1993. During this five year period, 39,117 inpatient reviews were conducted—4326 (11.1%) of them on patients with cardiovascular disease. All reviews on patients with cardiovascular disease were used in the analysis.

There were 2813 (58%) requests for medical admission and 1513 (42%) requests for surgical/procedural admissions. The five most common surgical reasons for hospital admission were cardiac catherization (n = 456), coronary bypass surgery (n = 257), valve replacement (n = 88), carotid endarterectomy (n = 69), and head/neck vessel replacement (n = 47). The five most common medical reasons for admission were angina pectoris (n = 614), congestive heart failure (n = 416), cerebrovascular accident (n = 414), arrhythmia/conduction disturbance (n = 370), and acute myocardial infarction (n = 313). Data were obtained on the number of days of inpatient treatment requested and approved at time of admission, and the number of days requested and approved for continued stay. LOS reduction was defined as the difference between total days requested and total days approved by UM.

Requests for admission were rarely denied (medical = 1 denial; surgical = 4 denials; total = 4326 requests). LOS was reduced relative to that requested by the treating physician for 17% of medical and 19% of surgical admissions. Cumulative 60-day readmission rates were 9.5% for medical admissions and 12.3% for surgical admissions. There was no relationship between LOS reduction and the likelihood of readmission for medical admissions or for surgical patients whose LOS was reduced by one day (95% CI: 0.72 - 2.80; P = 0.30). However, surgical patients whose LOS was reduced by at least two days were 2.6 times more likely to be readmitted within 60 days compared to patients with no reduction in LOS (95% CI: 1.3-5.1; P < 0.005). Excluding patients admitted for cardiac catherization did not change these results.

The vast majority ( > 87%) of patients readmitted within 60 days of their index admission were readmitted with a cardiovascular medical diagnosis. The most common reasons were angina pectoris, acute myocardial infarction, congestive heart failure, arrhythmias, and stroke. Thirty-day readmission rates were also substantially greater for patients who had their care constrained by two days or more. However, the number of patients admitted after 30 days was small, and the differences did not reach statistical significance. (Lessler DS, et al. Health Serv Res 2000; 34(6):1315-1329.)


The major findings of this study were that UM rarely denied requests for inpatient treatment of cardiovascular illness and that LOS reduction adversely affected clinical outcomes for some patients. UM decisions resulted in denial of hospital admission in a very small minority of cases (5/4326). Consequently, it appears that this aspect of UM review saved no money and contributed substantial costs, considering the time required to process requests and prepare data for review.

In addition to reducing unnecessary hospital admissions, a second UM objective is elimination of unnecessary hospital days. To judge the effect of UM decisions, Lessler and colleagues examined the relationship between LOS reduction and 60-day readmission. The effect of LOS reduction on readmission was "dose dependent." A reduction in LOS of one day did not affect readmission rates, whereas a reduction of two days or more caused a significant increase in the rate of hospital readmission. Lessler et al chose to compare 60-day readmission rates, rather than the more commonly used interval of 30 days. Consequently, it could be argued that the problems that led to readmission were "new" and not the result of a shorter LOS. When 30-day rates were compared, the same trend was seen. However, numbers of patients readmitted at the 30-day interval was small, and the difference did not reach statistical significance.

There are several limitations to this study. Only one UM program was examined. However, the reviews were conducted by a well established UM firm, the same LOS profiles were used for all reviews, and criteria and profiles were updated annually. Lessler et al did not have access to information about the patient’s clinical condition, or access to protocols used by the UM reviewers in conducting preadmisison and continuing stay reviews. Therefore, it was not possible to deduce factors that might have increased risk for readmission.

The study analyzed a broadly representative sample obtained from a large commercial insurance carrier that represented private companies, union trusts, and public organizations in 47 states. However, study data were more than five years old and it is possible that findings would differ if a more recent sample were analyzed. Nonetheless, findings raise concerns. There was no evidence of any benefit from preadmission reviews and evidence of a potential adverse effect of UM on patient outcomes for patients who received LOS reductions. Additional studies are needed to confirm whether findings of this study are broadly applicable. In these studies, it would be helpful to include a comprehensive cost-benefit analysis that includes costs of implementing UM reviews, as well as costs related to hospitalization.