According to a new study in the CMAJ (Canadian Medical Association Journal), comparing hospital performance based on readmission rates and early death rates should require caution because those two indicators have significant variation in their values, depending on how they are calculated.
"Hospital-specific readmission rates have been reported as a quality of care indicator but no consensus exists on how these should be calculated. Our results highlight that caution is required when comparing hospital performance based on 30-day or urgent readmissions given their notable variation when methods used in their calculation change," stated Carl van Walraven, M.D., lead author of the study, senior scientist at Ottawa Hospital Research Institute (OHRI) and associate professor at the University of Ottawa (uOttawa).
Due to the use of different methods and variables, readmission rates can result in different outcomes, the authors of the study found. They used hospital-specific 30-day death or readmission rates for adults at all 162 Ontario hospitals between 2005 and 2010 with various methods for confounder adjustment (age-sex v. complete) and different units of analysis (all hospitalizations v. 1 per patient). What they found was hospital-specific rankings varied extensively and changed based on which methodology was used.
According to the authors, readmission rates adjusted for age-sex alone had the greatest variation. "Our results highlight the caution required when comparing hospital performance using rates of death or readmission within 30 days," van Walraven wrote. "To measure quality of care, one would ideally measure the number of avoidable readmissions."
The authors recommend caution in using this measure to compare hospital performance.
"We found that slight changes in the methods used to calculate hospital-specific readmission rates had large influences on their values and the consequent hospital rankings. This should be kept in mind when comparing hospital performance using readmission rates,” van Walraven wrote.
