Predictors of patient preference toward primary health care

Background: Access to care is a national health problem. As a response to the lack of broad federal health care reform, communities have developed local solutions to address healthcare needs of uninsured populations. To support planning efforts for an insurance vehicle to cover uninsured patients, the Sedgwick County Health Access Project, a public-private coalition, explored the perceptions of residents toward healthcare insurance and services. Methods: An existing dataset of a 2009 convenience sample on-line and paper/pencil survey of adults ages 18-64 (N = 385) was collected to explore attitudes toward affordability and availability of healthcare insurance and the perceived priority of six types of healthcare services. Respondents rank ordered, by perceived need, six healthcare services including: primary care, hospitalization, emergency care, specialty care, tests (laboratory and radiological), or other services (chiropractic, physical therapy, home health, etc.). Because only uninsured respondents ranked the healthcare services (N = 101) and primary care was ranked first by 71.4% of these respondents, multivariate logistic regression analysis was performed to identify predictors of a first choice ranking of primary care. The model included gender, age, education level, marital status, race, income, employment status and health status. Results: Respondents were predominantly female (74.5%), aged 40 years and younger (66.7%), white (67.7%), single (52.1%), employed (62.1%), had some college education or technical school training (46.8%), reported an annual income of less than $20,000 (43.5%) and reported the health status was excellent or very good (40.4%). The final model retained all eight variables and explained 70 percent of the variability in a first choice ranking of primary care. Of these variables two were significant. Females were 4.23 (CI 1.13-16.37, p < .033) times more likely to rank primary care as their first choice. Respondents aged 51-64 years had an inverse OR (OR 0.90, CI 0.008-0.967), meaning that the odds of ranking primary care as their first choice was 90% less likely than younger respondents. Although not significant, the final six variables did improve the explanatory strength of the model: marital status, minority status, health status, education level, and annual income. Discussion: Logistic regression analysis revealed additional predictors of healthcare service preferences that were not evident using a simple, weighted rank ordering of preferences. Results will further inform planning efforts in Sedgwick County, Kansas. Conclusions: Multivariate analysis can reveal underlying factors that may be useful to community planners in crafting local solutions to access to care issues.