Do postage stamps versus pre-paid envelopes increase responses to patient mail surveys? A randomised controlled trial.

BACKGROUND

Studies largely from the market research field suggest that the inclusion of a stamped addressed envelope, rather than a pre-paid business reply, increases the response rate to mail surveys. The evidence that this is also the case regarding patient mail surveys is limited.

METHODS

The aim of this study is to investigate whether stamped addressed envelopes increase response rates to patient mail surveys compared to pre-paid business reply envelopes and compare the relative costs. A sample of 477 initial non-responders to a mail survey of patients attending breast clinics in Greater Manchester between 1/10/2002 - 31/7/2003 were entered into the trial: 239 were randomly allocated to receive a stamped envelope and 238 to receive a pre-paid envelope in with their reminder surveys. Overall cost and per item returned were calculated.

RESULTS

The response to the stamped envelope group was 31.8% (95% CI: 25.9% - 37.7%) compared to 26.9% (21.3% - 32.5%) for the pre-paid group. The difference (4.9% 95% CI: -3.3% - 13.1%) is not significant at alpha = 0.05 (chi2 = 1.39; 2 tailed test, d.f. = 1; P = 0.239). The stamped envelopes were cheaper in terms of cost per returned item (1.20 pounds) than the pre-paid envelopes (1.67 pounds). However if the set up cost for the licence to use the pre-paid service is excluded, the cost of the stamped envelopes is more expensive than pre-paid returns (1.20 pounds versus 0.73 pounds).

CONCLUSION

Compared with pre-paid business replies, stamped envelopes did not produce a statistically significant increase in response rate to this patient survey. However, the response gain of the stamped strategy (4.9%) is similar to that demonstrated in a Cochrane review (5.3%) of strategies to increase response to general mail surveys. Further studies and meta analyses of patient responses to mail surveys via stamped versus pre-paid envelopes are needed with sufficient power to detect response gains of this magnitude in a patient population.