Comparison of patient questionnaires, medical records, and plasma assays in assessing exposure to benzodiazepines in elderly subjects.

OBJECTIVE: Exposure in pharmacoepidemiologic studies can rely on various sources such as medical records, patient questionnaires, or plasma samples, which do not always concur. This study endeavored to compare sources of information on current exposure to benzodiazepines in elderly subjects.
METHODS: In a study in a hospital admissions department, 1136 elderly subjects included in a case-control study each completed a structured questionnaire. In addition, an inspection of the medical records of each subject was performed, as well as screening of a plasma sample (high-pressure liquid chromatography--diode array detector) for current exposure to benzodiazepines.
RESULTS: Benzodiazepines were found in the plasma of 33% of 1013 patients, in the records of 31% of patients, and in the questionnaires of 36% of 797 respondents. With use of the plasma results as a standard, questionnaires had 11% false positives and 28% false negatives; medical records had 14% false positives and 23% false negatives. The kappa for concordance between questionnaires and records was 0.63. Most of the errors were related to the unexpected presence in plasma of clorazepate, commonly used as a hypnotic agent.
CONCLUSIONS: Patient recall and medical records are not reliable measures of current exposure to benzodiazepines in elderly persons, although this unreliability may be more marked with certain drugs used as hypnotic agents.