Int J Pharm Pract 1998:6:170-5

Correspondence: Dr Sheridan J.Sheridan@iop.bpmf.ac.uk

National Addiction Centre, 4 Windsor Walk, London, England SE5 8AF
Janie Sheridan, MRPharmS, PhD, research pharmacist
John Strang, FRCPsych, MD, professor of the addictions

Original Papers

Late responders and non-responders to a postal survey questionnaire: analysis of potential further response and non-response bias

JANIE SHERIDAN and JOHN STRANG

The purpose of this study was to evaluate the effect of further mailshots on response rates to a postal survey, and to determine the extent of non-responder bias. A questionnaire on services to drug misusers was mailed to a random sample of one in four community pharmacies in England and Wales (2,654), stratified by family health services authority, followed by three further mailshots to non-responders. The remaining non-responders were contacted by telephone and asked to participate in a brief telephone interview. The addition of the fourth mailshot increased the response rate by only 4.8 per cent overall. Responses to key variables (demographics, participation in service provision) were not significantly different when comparing response to mailshots 1-3 with the fourth mailshot. Responses to the telephone survey succeeded in increasing the overall response rate to 98.4 per cent. These postal non-responders (ie, those responding to the telephone survey) were significantly different from those who responded to the postal survey in the extent of their service involvement, with postal responders being more involved in dispensing Controlled Drugs than telephone responders (50.1 per cent and 39.2 per cent, respectively) and provision of needle exchange (18.9 per cent and 9.6 per cent, respectively).
The use of additional mailshots to improve response rate may not be useful in increasing the generalisability of findings. However, persistent non-responders to postal mailshots were found to differ significantly from responders. If extrapolation to the whole population is deemed valuable, telephone follow-up of non-responders allows for the detection of any non-responder bias, and allows subsequent corrections to be made to postal survey responses.