[Billing based on a case-based lump sum for stroke. Did this lead to discharge of patients in a worse clinical condition?]

Abstract

Background: It has been supposed that the introduction of a new inpatient reimbursement system starting in 2004 in Germany using the German diagnosis-related groups (G-DRG) may lead to false incentives with encouragement of premature hospital discharge of patients. Exploring a large database on stroke patients, we addressed the question whether length of stay (LOS) and discharge in more severe condition were associated with the introduction of the G-DRG. We further examined other factors with probable effect on LOS such as variations of patient characteristics and treatment during the observation period.

Patients and methods: All stroke patients treated in 2003-2006 in the German state of Hesse (6,100,000 inhabitants) were assessed with respect to stroke severity, symptoms on admission and discharge, LOS and stroke-related deficits on discharge. We compared LOS and outcome in 2003 (before introduction of the G-DRG) with 2004 when the G-DRG had recently been introduced and with 2006 when the G-DRG was already well established in the clinical routine. The effects of LOS and treatment year on outcome were assessed using a logistic regression model.

Results: During the observation period, we evaluated 37,396 stroke patients. The length of stay was reduced significantly from 12.2 to 10.4 days (p<0.001). Both severity of stroke on admission and outcome on discharge decreased during the observation period. A multivariate analysis revealed a minor but significant association [odds ratio (OR): 1.020 per day of hospital treatment; 95% confidence interval (CI): 1.016-1.024] of LOS on outcome. Treatment in 2006 compared to 2003 led to good outcome with an OR of 1.378 (95% CI: 1.279-1.485). Subgroup analysis limited to patients with severe stroke revealed that LOS was significantly lower in 2006 compared to 2003 also in this patient subgroup; moreover, the proportion of patients discharged with severe outcome was lower in 2006 compared to 2003.

Conclusions: This study reveals a significant reduction of LOS during the years after introduction of the G-DRG. However, reduction of LOS was not associated with more severe outcome on discharge, possibly due to changes in stroke treatment implemented during the observational period. Our results do not support the conjecture that changes in the reimbursement system were associated with compromised patient care.