Background Healthcare variations are a worrisome and well-documented problem. Such variations occur at different levels: between health care systems, geographic areas, organizational settings, patients, and doctors. Research is increasingly focused on the contribution of doctors’ decision making to the generation or amplification of healthcare disparities. Clinical decision making studies continue to evolve through generations of work that focus on different types of influence: First generation studies identified patient-level attributes (e.g., gender, age, race/ethnicity, SES, health insurance status); Second generation studies on variations associated with physician characteristics (e.g., age, gender and education, specialty and type of remuneration); Third generation studies on health system and organizational influences (e.g., size, profit status, location and population served, and most recently, organizational culture).
Methods We used a factorial experimental design in which the subjects, primary care doctors (n=192), viewed clinically authentic vignettes of “patients” presenting with identical signs and symptoms suggesting diabetes. They were stratified according to gender and level of experience. During an in-person interview, they were asked how they would diagnosis and manage the vignette ‘patient’.
Results After controlling for the first two levels of influence (patient and provider), each of which contributed 4.4 and 2% respectively, organizational culture significantly contributed to their behavior, accounting for 14.3% of the variation in clinical decision-making for diabetes. Considering nine different dimensions of practice culture, organizational trust and business emphasis contributed most to the variance in treatment for diabetic foot neuropathy.
Conclusion Attempts to reduce health care variations continues to focus on the levels of patient attributes and physician characteristics (e.g. improved educational efforts). Findings from this experiment suggest a need to further investigate the contribution of of organizational factors (third generation) and suggest appropriate interventions at that level. As important however, is the consideration of a fourth generation: cognitive aspects of physician decisions. Qualitative methods (i.e. “think aloud”) and newly developing methods to objectively measure unconscious bias (e.g., the Implicit Association Test) can provide more robust assessment of what goes on “inside the doctors’ head,” which eventually produce healthcare disparities.