One of the problems with carrying out assessments is getting enough readers, and a representative cross-section of readers, to respond. Another is to make sure that the information they provide is accurate.

Reader feedback poses other potential problems including recall bias, because of people’s selective memory, and non-representative results, because people who bother to fill in a questionnaire and send it back may have different opinions from those who do not reply. In addition, the quality of the data may be poor due to incomplete responses to surveys, inconsistent interview technique or a poorly designed survey. The likelihood of poor quality data can be minimised through validating questions and using standardised interview techniques. The references on survey design listed at the end of this chapter provide additional information.

Assessments of readers’ opinions measure how people think the bulletin has influenced their behaviour, not actual behavioural changes. Despite these limitations, canvassing the views of readers through surveys, discussions or interviews provides invaluable feedback and should be a regular part of the activity of all drug bulletins. Letters to the Editor and other feedback (e.g. letters to other publications) can also give an indication about what readers are thinking.

When designing a readership survey, it is useful to include questions relating to the bulletin’s impact. For example, a survey of doctors might ask whether articles have influenced prescribing practices, helped them to recognise an adverse drug reaction or interaction in a patient, or changed the advice given to patients. These types of questions cannot measure true behavioural change. People’s reports of changes in their own behaviour are often much more optimistic than reality. However, they do provide an indication of the value readers place on the information in the bulletin. If they are suspicious or critical of the bulletin’s advice, they are less likely to report a behavioural change.