Prevalence of dementia in users of hormone replacement therapy as defined by prescription data.

Background: Studies of hormone replacement therapy (HRT) and dementia and cognitive impairment show mixed results. This study assessed the prevalence of dementia and cognitive impairment in users and nonusers of HRT defined using computer-stored prescription information.

Methods: The study involved 3924 women 75 years of age and older who were members of the Southern California Kaiser Permanente Medical Care Program in 1998. HRT use was determined based on prescription data for 1992-1998. Cognitive function and dementia were assessed using the Telephone Interview of Cognitive Status supplemented by the Telephone Dementia Questionnaire and medical record review.

Results: Odds ratios (ORs) for cognitive impairment/dementia showed expected associations with age, education, ethnicity, and a history of stroke or Parkinson's disease. After adjustment, the OR for cognitive impairment or dementia in HRT users compared with HRT nonusers was 0.91 (95% confidence interval 0.75-1.10). The adjusted ORs for all dementia and dementia without cause in HRT users compared with nonusers were 0.77 (95% confidence interval 0.59- 1.00) and 0.78 (0.58-1.05), respectively. Twenty percent of women with cognitive impairment or dementia who had been classified as HRT users by prescription (one prescription each year from 1992-1998) denied hormone use compared with 8.8% of women without impairment. Medical record review validated prescription information for the impaired women.

Conclusions: The study identified an important methodologic problem in studies of HRT and cognitive impairment and dementia that depend on recalled information about drug exposure. A protective effect of HRT for cognitive impairment and dementia was neither established nor ruled out based on the prescription data.