Risk adjustment using administrative data: impact of a diagnosis-type indicator

Area of Science:

• Health Services Research
• Clinical Informatics
• Cardiovascular Surgery Outcomes

Background:

• Administrative datasets are crucial for health outcomes research.
• Accurate coding of clinical variables as baseline comorbidities versus complications is essential for reliable risk adjustment.
• The impact of a 'diagnosis-type indicator' on risk adjustment for coronary artery bypass graft surgery (CABG) outcomes is not well-established.

Purpose of the Study:

• To assess the frequency of clinical variables being coded as complications versus baseline comorbidities.
• To compare risk-adjusted outcome analyses for CABG surgery when a "diagnosis-type indicator" is used versus ignored.

Main Methods:

• Analysis of existing administrative data from 23 Canadian hospitals.
• Inclusion of 50,357 coronary artery bypass graft surgery cases.
• Comparison of risk adjustment models with and without the utilization of a "diagnosis-type indicator".

Main Results:

• Seven clinical variables were frequently coded as complications, including severe liver disease (35%) and hemiplegia (34%).
• Risk adjustment analyses for in-hospital mortality showed significant differences when the "diagnosis-type indicator" was used compared to when it was ignored.
• Hospital mortality rates and rankings changed substantially, impacting 19 out of 23 hospitals.

Conclusions:

• Analyses of Canadian administrative data for CABG surgery yield considerably different results when the "diagnosis-type indicator" is utilized versus ignored.
• The implementation of a "diagnosis-type indicator" is recommended for other countries to prevent potentially misleading risk-adjustment analyses.
• Accurate differentiation between complications and comorbidities is vital for reliable healthcare quality assessment.