Objective : To examine past trend and present status of polypharmacy, multiple drug therapy, which can be a cause of serious adverse events induced by drug interactions.
Design : Drug utilization survey using a database of the Kagoshima University Hospital Information System.
Methods : Data of the prescriptions issued during every May between 1985 and 1994 were extracted from the database. Monthly prescriptions were classified by the number of drugs they included.Furthermore, the data were sorted by patients' ID to obtain the age distribution of patients observed in each year and the distribution of the number of drugs per prescription was analyzed in each age group.
Results : The average number of medicines per prescription was increased from 3.07±2.16 (average±S.D.) in 1985 to 3.46±2.61 in 1994. During 10 years between 1985 and 1994, the proportion of patients receiving ten drugs or more also increased from 1.50%to 3.90%. There was a clear tendency that a large number of drugs were given to old patients and comparatively few drugs to young patients.Proportion of the old patients over 59 years of age increased linearly from 25.71%in 1985 to 35.37% in 1994.
Conclusion : The average number of medicines per prescription was increased during 10 years. This trend is considered to be attributable in part to the increase in the proportion of the old patients. The concomitant use of so many drugs is a problem and a rule for diminishing returns recently adopted as a reimbursement policy of health insurance system has been applied to prescriptions with ten drugs or more. Further studies are needed to examine possible adverse outcomes and other characteristic features of multiple therapy.

Background :There have been only a few comparative observational studies on the safety and effectiveness of drugs in Japan. Comparative observational studies would provide important information to address these issues and thus we need to establish a means to facilitate such studies. In comparative studies, it is important to prevent the distortion of results due to selection bias. Though we do not yet have a claims database for use in pharmacoepidemiological studies, recently many hospitals and pharmacies have computerized prescription data which may be used to minimize selection bias. Good standardized procedures for the identification of patients prescribed one of two or more drugs to compare in a study using computerized prescription data would serve as a basis for a variety of pharmacoepidemiological studies in Japan.
Methods :We carried out a questionnaire survey in 2753 hospitals and 909 community pharmacies to estimate the fraction of hospitals where computerized data can be used to identify all eligible patients who used a specific drug.
Results :Questionnaires were returned by 1942 (71%) of 2753 hospitals and 632 (70%) of 909 pharmacies. From among those which responded, patients were identified, the patient list was printed, and the electronic file of the patient list was generated in 75%, 64% and 36% of the 1942 hospitals and in 100%, 93% and 49% of the 632 pharmacies respectively.
Conclusion :With procedures using computerized prescription data, the cohort for observational comparative studies may be identified with a minimal selection bias in a majority of hospitals and pharmacies.