A longitudinal study of use of health care services among older women.

This research tested the hypothesis that the percent of variance explained in use of health care services by the health care services utilization model could be significantly increased by including measures of past use of health care services and of past health status. Data from older women who participated in the Social Security Administration's Longitudinal Retirement History Survey (N = 1894) were analyzed by means of regression analysis. The results revealed that measures of previous use of health care services were more strongly related to current use of health care services in 1979 than were measures of previous health status. Inclusion of previous use and previous health status variables almost doubled the amount of variance explained by current predictors in number of physician visits, and more than doubled the explained variance in having to put off health care while the amount of variance explained in number of hospital episodes and in number of hospitalized nights was increased by approximately one-third.