We then designed a randomized, controlled intervention trial to examine whether testing and treating women identified by the risk factors would affect the subsequent occurrence of pelvic inflammatory disease. We adapted the earlier predictive model to exclude married women, in whom the prevalence of chlamydial infection was very low, and to exclude cervical ectopy as a criterion for screening, since this required an examination. We then developed a brief, self-administered questionnaire that we could use to classify women according to risk status. All women from 18 to 34 years of age who were Group Health Cooperative enrollees as of October 1, 1990, were selected from the computerized enrollment file. After excluding enrollees whose records listed a spouse, we mailed questionnaires to the remaining 36,547 women over a 10-month period. Duplicate surveys
were mailed to those who did not respond. We also telephoned some
of the nonresponders each month to request that they return the questionnaire
or give their responses by telephone. Emphasis was placed
on calling nonresponding women assigned to the intervention group
in order to expedite setting up their clinic appointments for testing.