MEASURING THE NEED FOR MEDICAL-CARE IN AN ETHNICALLY DIVERSE POPULATION

Objective. To examine measures of need for health care and their relat ionship to utilization of health services in different racial and ethn ic groups in California. Data Source. Telephone interviews obtained by random-digit dialing and conducted between April 1993 and July 1993 i n California, with 7,264 adults (ages 18-64): 601 African Americans, 2 46 Asians, 917 Latinos interviewed in English; 1,045 Latinos interview ed in Spanish; and 4,437 non-Latino whites. Study Design. A cross-sect ional survey was conducted from a stratified, probability telephone sa mple. Data Collection. Interviews collected self-reported indicators o f need for health care: self-rated health, activity limitation, major chronic conditions, need for ongoing treatment, bed days, and prescrip tion medication. The outcome was self-reported number of physician vis its in the previous three months. Principal Findings. Compared to whit es, one or more of the other ethnic groups varied significantly (P < . 05) on each of the six need-for-care measures after adjustment for hea lth insurance, age, sex, and income. Latinos interviewed in Spanish re ported lower percentages and means on five of the need measures but th e highest percentage with fair or poor health (32 percent versus 7 per cent in whites). Models regressing each need measure on the number of outpatient visits found significant interactions of ethnic group with need compared to whites. After adjustment for insurance and demographi cs, the estimated mean number of visits in those with the indicator of need was consistently lower in Latinos interviewed in Spanish, but th e differences among the other ethnic groups varied depending on the me asure used. Conclusion. No single valid estimate of the relationship b etween need for health care and outpatient visits was found for any of the six indicators across ethnic groups. Applying need adjustment to the use of health care services without regard for ethnic variability may lead to biased conclusions about utilization.