Physiciansʼ patient load per DRG, the consumption of hospital resources, and the incentives of the DRG prospective payment system

Publisher Website
Google Scholar
Abstract
From 1985 through 1987, the authors assessed the relationship between two main variables at a large academic medical center: (1) the numbers (high or low) of patients per diagnosis-related group (DRG) treated by individual physicians and (2) hospital resource consumption of the patients. The patients were classified according to their routes of admission (emergency or non-emergency); the physicians with eight or more patients per DRG were labeled “high-patient-load physicians” (hereafter called “high-load physicians”), and those with five or fewer patients were labeled “low-patient-load physicians,” (“low-load physicians”). The resource variables studied were length of stay (LOS) and total hospital cost. For the non-emergency admissions, the low-load physicians' patients had an average LOS that was 56.2% greater and an average hospital cost that was 58.3% greater than were the LOS and cost of the patients of the high-load physicians. (Both LOS and cost per patient were adjusted for DRG weight index.) This was due in part to a greater severity of illness for the patients (as measured by total ICD-9-CM codes per patient) of the low-load physicians. For the emergency admissions, the low-load physicians' patients had an average LOS that was 9.5% greater and an average hospital cost that was 10.5% greater than the LOS and cost of the patients of the high-load physicians. As in the non-emergency admissions group, severity of illness for the patients of the low-load physicians was greater. The only category of patients that was economically profitable to the hospital was the non-emergency admissions group treated by the high-load physicians. Even though these findings were the result of many factors (discussed in the text), they suggest a relationship between hospital resource consumption and the physicians' patient load per DRG.

This publication has 0 references indexed in Scilit: