Sex, drugs, and hypertension: a methodological approach for studying a sensitive subject.

Sexual dysfunction is a recognized side effect of hypertension and antihypertensive medications in men, but is not established as a side effect in women, due to the lack of established methodology. An ambulatory medical record-based, case-control study was designed to study sexual function in treated and untreated hypertensive women and healthy controls. The research was performed in a teaching hospital with satellite clinics in upstate New York. There were 3312 medical records reviewed, 640 premenopausal Caucasian women in heterosexual relationships subjects were eligible for participatic diagnosis of mild hypertension (BP > or =140/90 mmHg and < 160/100 mmHg) for cases; no other significant medical history. A total of 241 women agreed to participate, 224 (35%) completed both a self-administered questionnaire and a telephone interview (112 healthy, 112 hypertensive). There was an initial 74% response rate among those eligible to participate, with 35% completing the entire study. Age and average blood pressure were not significant between 224 participants and 416 nonparticipants by 2-tailed t-test analysis. Seven composite variables were formed from a 47-item sexual response questionnaire. Initial unadjusted chi2 results reported women with hypertension had more difficulty than did healthy controls achieving lubrication and orgasm. Seven questions, each with the highest correlation to its respective composite variable (by Spearman's correlation), formed an abbreviated questionnaire. Quality of female sexual function was quantified in an ambulatory outpatient setting. A method was described to address hypertension, pharmacotherapy, and sexual functioning by employing self-administered questionnaires and telephone interviews. Initial analysis suggested that hypertensive women may have an impaired physiological sexual response. The abbreviated questionnaire generated from questions with the highest correlation to their respective composite variables may be useful in further evaluating this issue.