Measuring the appropriateness of prescribing in primary care: are current measures complete?

Background and objectives: Appropriateness of prescribing is often assessed by standard instruments. We wished to establish whether judgements of appropriateness that included patients' perspectives and contextual factors could lead to different conclusions when compared with commonly used instruments. To explore the predictive accuracy of these instruments. Methods: The design was interviews of patients, audio recordings of the consultation and interviews of the doctors, in varied primary care practices in England. Participants were patients who were likely to discuss a medication issue. The outcome measures were judgements of appropriateness made by the researchers and by two instruments: the Prescribing Appropriateness Index and the Medication Appropriateness Index. Implications for the predictive accuracy of the measures was also investigated. Results: From 35 cases there was agreement between the judges and the instruments in 22 cases, 16 were appropriate and 6 inappropriate. Of 10 cases classified as inappropriate by the instruments the judges thought four were appropriate. Of 18 cases classified as appropriate by the instruments, two were considered inappropriate by the judges. In seven cases the prescribing decisions could not be classified by the instruments because the decision was to not prescribe. Conclusions: Current measures of appropriateness of prescribing depend predominantly on pharmacological criteria, and so do not represent cases that would be judged appropriate when including the patient's views and contextual factors. If most prescribing is appropriate then use of these measures may lead to more false negatives than real negatives. The instruments should be renamed as measures of 'pharmacological appropriateness' and are useful where the incidence of this type of inappropriate prescribing is relatively high.