Non-attendance and effective equity of access at four public specialist outpatient centers in Hong Kong.

This study tests whether socio-economic status (SES), at either the individual or ecologic levels, exerts a direct impact on non-attendance or an indirect impact on attendance through longer waiting time for appointments and/or doctor-shopping behavior at four public specialist outpatient centers in Hong Kong. We collected information through three main sources, namely patients' referral letters, telephone interviews with both open- and closed-ended questions (e.g. doctor-shopping data) and hospital administrative databases from a total of 6495 attenders and non-attenders enrolled from July 2000 through October 2001. Individual-level SES was measured by education, occupation and monthly household income. Tertiary planning unit (TPU)-level SES data consisted of proportion unemployed, proportion with tertiary education, median income and Gini coefficient. Direct effects of SES on non-attendance were examined by logistic regression. Indirect contributions mediated through waiting time and doctor-shopping were analyzed by structural equation modeling. We found that SES, at the individual or ecologic level, did not exert a direct effect on non-attendance. Instead, TPU-level SES contributed positively to waiting time (beta=0.06+/-0.03, p=0.048), i.e. worse-off neighborhoods (and those with greater income inequality) had a shorter waiting time. Individual-level SES was also directly associated with the likelihood of doctor-shopping (beta=0.16+/-0.02, p<0.001), i.e. the poor were less likely to doctor-shop. Both waiting time (beta=0.12+/-0.02, p<0.001) and doctor-shopping (beta=0.37+/-0.02, p<0.001) were significantly related to non-attendance. Our findings suggest a highly equitable specialist ambulatory care public system in Hong Kong. Health care resources are appropriately targeted at the socially indigent, and the poor are not discriminated against and pushed to seek alternative sources of care by the system. These results should be confirmed using a prospective design.