Associations Between Length of Stay and Readmission

John Goethe, MD, director of the Burlingame Center for Psychiatric Research and Education at the Institute of Living, Hartford, CT, presented data at the Annual Meeting of the American Psychiatric Association (APA) in San Diego demonstrating associations between length of stay (LOS) and readmission (RA).

In describing the motivation for the study, Goethe said, “These widely used metrics in healthcare have been associated in psychiatric studies with cost of care and clinical outcome. However, most previous studies have investigated these associations using only a single time metric, such as RA within 1 month or 1 year after discharge.”

Goethe continued, “Because such associations may be found at one time interval but not another, we assessed associations between LOS and time to RA at 4 different intervals. Similarly, because these associations can vary by diagnosis, we also examined associations for each of 4 diagnoses as well as for the sample as a whole.” These diagnoses comprised major depressive disorder (MDD), schizophrenia (SZ), schizoaffective disorder (SA), and bipolar disorder (BD).

The sample used in the study was more than 7500 inpatients aged 18 to 64 years with 1 of the 4 diagnoses who were discharged from the Burlingame Center over a 5-year period (2010 to 2015). Over half had MDD, about one-fourth had BD, with roughly 15% having SA and 10% having SZ.

Associations between the risk of RA within 15, 30, 90, and 180 days and LOS were each determined by logistic regression analysis. The lower quartile of LOS (4 days) was also compared with both the upper quartile (10 days) as well as the middle 2 quartiles (5 to 9 days).

The mean LOS was 8 ± 7.5 days in the sample overall, but diagnosis-specific LOS varied significantly, with the mean LOS for SZ significantly greater than for SA (12.14 days vs 10.94 days), BP (7.91 days), or MDD (6.96 days). An LOS of 4 days compared with a 10-day LOS was associated with increased risk of RA for all intervals in SZ.

However, there was a decreased RA risk in MDD in the similar comparison. LOS was also associated with RA in SA, but only for the 180-day interval. There were no significant associations between LOS and RA in BD.

Goethe said, “Comparisons of an LOS of four days with LOS of 5 to 9 days gave results similar to those of SZ where an increased risk of RA was seen for all intervals. However, this association was only significant in MDD at 180 days and was not significant in SA or BD within any time interval.”

Regarding the design of the study, he said, “Few studies have considered that variations in time-based metrics may alter the results in research about the associations between LOS and RA.”

In concluding his remarks, Goethe said, “The highest RA risk in SZ was associated with an LOS of just 4 days, suggesting that longer a LOS may provide greater stability to the patient. Conversely, the fact that an LOS of 4 days in BD was associated with a 40% lower RA risk suggests that there may be a subset of MDD patients for whom very brief hospitalization is sufficient, perhaps due to lower illness severity or lower risk of recurrence. Unexpected was that there were no similar associations found in SA and SZ or that there were no significant LOS-RA associations seen in BP. Further study will be needed to clarify the situation.”