The importance of using a valid method to assess medication compliance in psychosis.

Sir,

Rao et al.[1] examined aspects of medication compliance and noncompliance in patients with psychosis. We appreciate their study of a challenging behavior that impacts treatment outcomes. However, we take issue with their use of casefile data to operationalize definitions of compliance and noncompliance. Given that data recorded in outpatient casefiles are unlikely to be either sufficiently detailed or sufficiently accurate for research purposes, assessing compliance through a chart review is likely to be valueless. So, even if the rest of the study and the data analysis were perfect, the study would fail on this single criterion. Expressed otherwise, compliance was assessed using a method that would not meet any definition of validity.

Compliance can be prospectively evaluated in several ways; the simplest is to ask the patient or caregiver. However, poor caregiver supervision, cheeking of caregiver-administered medication where noncompliance is deliberate rather than due to forgetting, nondisclosure of noncompliance by the patient, faulty reporting due to cognitive problems in the patient or caregiver, and other factors could compromise the validity of a simple enquiry.

For research purposes, more objective assessments are employed. These include ticking taken doses in a patient diary, counting pills that remain at follow up visits, using electronic medication bottles that register each time the bottle is opened, using special pills that contain microchips that generate a signal when swallowed, and monitoring serum drug levels.[23] Such methods also have drawbacks but are more objective than a simple enquiry. Of note, questionnaires such as the Morrisky-Green-Levine Scale and the Medication Adherence Rating Scale can also be used in psychiatric populations to assess treatment adherence.[45]

Ideally, a study on compliance should assess compliance using more than one method, and operationalization of compliance on continuous rather than categorical measures should be preferred.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.