Relative impact of patient and clinic factors on adherence to primary carepreventive service guidelines - An exploratory study

BACKGROUND. Preventive care service use is commonly compared across health plans, clinics, or individual providers, yet little is known about the infl uence of the clinic versus patient factors on utilization of these services . OBJECTIVES. To measure the relative influence of the facility (clinic) vers us patient factors (demographic, behavioral and functional characteristics) on patients' utilization of mammography, Pap smears, cholesterol screening , and retinal exams for those with diabetes. RESEARCH DESIGN. Retrospective analysis, using administrative and patient s urvey data SUBJECTS. Enrollees in 2 University-based clinics and a county hospital-bas ed clinic serving a predominantly low-income population with limited access to health care. Eligibility for cervical cancer screening, screening mammo graphy, cholesterol screening, or annual retinal exam (diabetes) was define d by age, sex, and diagnosis. MEASURES. Multivariate models, one using readily available administrative d ata, and another using detailed health status and behavior data gathered fr om a clinics-wide survey. RESULTS. Unadjusted screening rates for three of four procedures were signi ficantly and substantially lower at the county hospital based clinic than t he two University-based clinics. After adjusting for patient characteristic s, utilization of three screening services at the county hospital remained significantly below the University-based clinics (Odds Ratios [95% CI]: mam mogram 0.15 [0.06-0.35]; Pap smear 0.32 [0.21-0.50]; cholesterol 0.19 [0.09 -0.38]; diabetes retinal exam10.68 [0.93-3.01]). The models with detailed s urvey data performed only marginally better than the models using only admi nistrative data. CONCLUSIONS. Patient characteristics were much less important than the clin ic for predicting whether patients received primary care preventive service s. Our results suggest that case mix adjustment is unlikely to explain away discrepancies in performance between clinics or provider groups.