• Olga C Damman, Janine H Stubbe, Michelle Hendriks, Onyebuchi A Arah, Peter Spreeuwenberg, Diana M J Delnoij, and Peter P Groenewegen.
• Netherlands Institute for Health Services Research, Utrecht, The Netherlands. o.damman@nivel.nl
• Med Care. 2009 Apr 1; 47 (4): 496-503.

BackgroundRatings on the quality of healthcare from the consumer's perspective need to be adjusted for consumer characteristics to ensure fair and accurate comparisons between healthcare providers or health plans. Although multilevel analysis is already considered an appropriate method for analyzing healthcare performance data, it has rarely been used to assess case-mix adjustment of such data. The purpose of this article is to investigate whether multilevel regression analysis is a useful tool to detect case-mix adjusters in consumer assessment of healthcare.MethodsWe used data on 11,539 consumers from 27 Dutch health plans, which were collected using the Dutch Consumer Quality Index health plan instrument. We conducted multilevel regression analyses of consumers' responses nested within health plans to assess the effects of consumer characteristics on consumer experience. We compared our findings to the results of another methodology: the impact factor approach, which combines the predictive effect of each case-mix variable with its heterogeneity across health plans.ResultsBoth multilevel regression and impact factor analyses showed that age and education were the most important case-mix adjusters for consumer experience and ratings of health plans. With the exception of age, case-mix adjustment had little impact on the ranking of health plans.ConclusionsOn both theoretical and practical grounds, multilevel modeling is useful for adequate case-mix adjustment and analysis of performance ratings.

16 keywords...

hide…