SELF-REPORTED MEDICATION NONCOMPLIANCE IN THE ELDERLY

Objective: To assess self-reported compliance with prescribed medicati ons in a population of elderly patients prior to their hospital admiss ion in an attempt to understand further the factors which influence dr ug taking patterns. Methods: Information which, based on personal clin ical experience and published research, may impact on compliance was c ollected for patients by way of a chart review within 3 days of hospit al admission, a search of patient computerised hospital records and an interview. All crude data were coded and entered into a computerised relational database. Each patient's data were assessed using the Naran jo algorithm and the score was recorded. Chi-square analysis highlight ed those factors which significantly influenced compliance, sub-divide into under-compliance (taking less medicine than prescribed) and over -compliance (taking more medicine than prescribed). Inter-relationship s between variables were investigated using multiple-regression analys is. Results: Overall, 13.7% of the population (n = 512) reported non-c ompliance, with 10.7% reporting under-compliance and 4.3% reporting ov er-compliance. A number of patients reported both under-and over-compl iance. Being prescribed bronchodilators, for example, was found to be associated with under-compliance, while being prescribed analgesics (e xcluding non-steroidal anti-inflammatories) was associated with over-c ompliance using Chi-square analysis. A five-variable non-compliance ri sk model was obtained from logistic regression analysis. This model ha d a specificity of 88.9% and a sensitivity of 33.3%. The factors shown to influence compliance were the type of drug being taken (diuretics, bronchodilators and benzodiazepines), independence when taking medici nes and the number of non-prescription drugs being taken. All other la boratory/test data, diseases/diagnoses, reasons for hospital admission and sociodemographic factors were not significant risk factors for se lf-reported non-compliance in the present model. Conclusions: Although ir is accepted that self-reporting of poor compliance is generally lo wer than actual poor compliance, the present risk model provides furth er insight into the drug-taking habits of elderly patients.