Optimal Measures for Primary Care Physician Encounters after Stroke and Association with Survival: A Data Linkage Study

Abstract

Background and Purpose: Primary care physicians (PCPs) provide ongoing management after stroke. However, little is known about how best to measure physician encounters with reference to longer term outcomes. We aimed to compare methods for measuring regularity and continuity of PCP encounters, based on survival following stroke using linked healthcare data. Methods: Data from the Australian Stroke Clinical Registry (2010–2014) were linked with Australian Medicare claims from 2009 to 2016. Physician encounters were ascertained within 18 months of discharge for stroke. We calculated three separate measures of continuity of encounters (consistency of visits with primary physician) and three for regularity of encounters (distribution of service utilization over time). Indices were compared based on 1-year survival using multivariable Cox regression models. The best performing measures of regularity and continuity, based on model fit, were combined into a composite “optimal care” variable. Results: Among 10,728 registrants (43% female, 69% aged ≥65 years), the median number of encounters was 17. The measures most associated with survival (hazard ratio [95% confidence interval], Akaike information criterion [AIC], and Bayesian information criterion [BIC]) were the Continuity of Care Index (COCI, as a measure of continuity; 0.88 [0.76–1.02], p = 0.099, AIC = 13,746, BIC = 13,855) and our persistence measure of regularity (encounter at least every 6 months; 0.80 [0.67–0.95], p = 0.011, AIC = 13,742, BIC = 13,852). Our composite measure, persistent plus COCI ≥80% (24% of registrants; 0.80 [0.68–0.94], p = 0.008, AIC = 13,742, BIC = 13,851), performed marginally better than our persistence measure alone. Conclusions: Our persistence measure of regularity or composite measure may be useful when measuring physician encounters following stroke.