The effect of misclassification errors on case mix measurement

Research

Related Publications

Smolina K, Weymann D, Morgan S, Ross C, Carleton B. Association between regulatory advisories and codeine prescribing to postpartum women. Journal of the American Medical Association. 2015;313(18):1861-2.
Suter E, Misfeldt R, Mallinson S, Wilhelm A, Boakye O, Marchildon G, et al. Comparative Review of the Policy Landscape of Team-based Primary Health Care Service Delivery in Western Canada. Alberta Health Services; 2014.
Laberge M, Pang J, Walker K, Wong ST, Hogg W, Wodchis WP. QUALICOPC (Quality and Costs of Primary Care) Canada: A focus on the aspects of primary care most highly rated by current patients of primary care practices. Ottawa, ON: Canadian Foundation for Healthcare Improvement; 2014.
McGregor MJ, Abu-Laban RB, Ronald L, McGrail KM, Andrusiek D, Baumbusch J, et al. Nursing Home Characteristics Associated with Resident Transfers to Emergency Department. Canadian Journal on Aging. 2012;33(1):38-48.

Publication Topics

The effect of misclassification errors on case mix measurement

Title
Publication TypeJournal Article
Year of Publication2006
AuthorsSutherland JM, Botz CK
JournalHealth policy (Amsterdam, Netherlands)Health Policy
Volume79
Issue2-3
Pages195 - 202
Date Published2006
KeywordsDiagnosis-Related Groups/classification, Forms and Records Control/standards, Humans, National Health Programs, Ontario, Reimbursement Mechanisms/economics/organization & administration, Retrospective Studies
AbstractCase mix systems have been implemented for hospital reimbursement and performance measurement across Europe and North America. Case mix categorizes patients into discrete groups based on clinical information obtained from patient charts in an attempt to identify clinical or cost difference amongst these groups. The diagnosis related group (DRG) case mix system is the most common methodology, with variants adopted in many countries. External validation studies of coding quality have confirmed that widespread variability exists between originally recorded diagnoses and re-abstracted clinical information. DRG assignment errors in hospitals that share patient level cost data for the purpose of establishing cost weights affects cost weight accuracy. The purpose of this study is to estimate bias in cost weights due to measurement error of reported clinical information. DRG assignment error rates are simulated based on recent clinical re-abstraction study results. Our simulation study estimates that 47% of cost weights representing the least severe cases are over weight by 10%, while 32% of cost weights representing the most severe cases are under weight by 10%. Applying the simulated weights to a cross-section of hospitals, we find that teaching hospitals tend to be under weight. Since inaccurate cost weights challenges the ability of case mix systems to accurately reflect patient mix and may lead to potential distortions in hospital funding, bias in hospital case mix measurement highlights the role clinical data quality plays in hospital funding in countries that use DRG-type case mix systems. Quality of clinical information should be carefully considered from hospitals that contribute financial data for establishing cost weights.
Citation Key482