Effect of correcting outcome data for case mix: an example from stroke medicine

Area of Science:

• Clinical Medicine
• Health Services Research
• Stroke Medicine

Background:

• Evaluating clinical outcomes is essential for healthcare quality assessment.
• Comparative studies are increasingly used to assess healthcare provider performance.
• Acute stroke patient outcomes are influenced by numerous factors, including patient characteristics (case mix).

Purpose of the Study:

• To determine the influence of variations in patient case mix on clinical outcome indicators for acute stroke patients.
• To assess the impact of an organized stroke service on patient outcomes.
• To evaluate the necessity of case mix adjustment in non-randomized comparative studies.

Main Methods:

• A "before and after" cohort study design was employed.
• Prospective, consecutive patient identification and follow-up were conducted.
• Multiple logistic regression analyses were used to adjust for case mix variations.

Main Results:

• Crude outcome data initially suggested improvements in survival and independence after the introduction of a stroke service.
• After adjusting for age and sex, these "improvements" were less pronounced but remained significant.
• Following adjustment for multiple prognostic indicators, all observed outcome differences between the pre- and post-stroke service cohorts became non-significant.

Conclusions:

• Variations in case mix critically influence the interpretation of clinical outcome data.
• Non-randomized comparative studies, common in healthcare, require robust case mix correction for meaningful interpretation.
• The apparent improvements in outcomes may be attributable to case mix differences rather than the intervention itself.