Figure. Multivariate-adjusted odds ratios and 95% confidence intervals for always having knee pain as compared with never/occasionally/often having knee pain were plotted using a logarithmic scale, after adjusting for age, sex, marital status, education, medical consultation, current/past history of major disease, and smoking. Participants who died during follow-up were excluded. Additionally, participants who were institutionalized during follow-up were excluded from the estimation of odds ratios of LTC eligibility and basic ADL dependence on the Katz ADL.
