Family Practice Vol. 20, No. 5, 592-594
© Oxford University Press 2003
Research Methods |
Third mailings in epidemiological studies: are they really necessary?
Department of General Practice and Primary Care, University of Aberdeen, Foresterhill Health Centre, Westburn Road, Aberdeen AB25 2AY, UK.
Correspondence to Dr Alison M Elliott; E-mail: a.m.purves{at}abdn.ac.uk
Elliott AM and Hannaford PC. Third mailings in epidemiological studies: are they really necessary? Family Practice 2003; 20: 592–594.
Received 10 March 2003; Accepted 19 May 2003.
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionReferences |
|---|
Introduction. Whether or not third mailings are appropriate or worthwhile in postal epidemiological studies has not been thoroughly investigated and requires examination.
Methods. A self-completion postal questionnaire of 2184 individuals was conducted in 2000. The socio-demographic and health characteristics of four groups of individuals (first mailing respondents, second mailing respondents, third mailing respondents and non-respondents) were compared.
Results. Some significant differences between the groups were found, however, the inclusion of respondents to the third mailing did not significantly change the overall characteristics of respondents compared to non-respondents.
Discussion. When differences do exist between respondents and non-respondents, our results suggest that a third mailing is unlikely to remove many of these differences. The study supports our preivous suggestion that the effort and resources expended in carrying out a third mailing may not be justified.
Keywords. Data collection, epidemiological methods, health surveys, questionnaires, response bias.
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Three mailings are routinely used in postal epidemiological studies to increase response rates and reduce the possibility of response bias. Whether or not third mailings are appropriate or worthwhile has not been thoroughly investigated. A previous study1 by the authors suggested that the effort and resources expended in carrying out a third mailing in such surveys may not be justified, since the inclusion of respondents to the third mailing did not significantly change the overall characteristics of those responding. This previous work was undertaken in a unique study which may not be generalizable to other population-based surveys. The authors have therefore explored this issue further using data from a postal survey of chronic pain in the community.2
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
In 1996, a postal survey of 4611 individuals from 29 general practices in the north-east of Scotland was conducted to describe the prevalence and distribution of chronic pain in the community.2 All respondents to this survey (n = 3605) who were willing to participate in further research (n = 2422) and were not screened out by their GPs (n = 238) were identified for follow-up. The remaining 2184 individuals were sent a self-completion postal questionnaire in 2000. Up to two reminders were sent to non-respondents at 2 weeks and 4 weeks. Individuals in the follow-up study were categorized into four groups; those responding after the first mailing; those responding after the second mailing; those responding after the third mailing; and non-respondents. The groups were compared on characteristics obtained from the baseline survey as this provided comprehensive data on non-respondents as well as respondents to the follow-up questionnaire. To investigate differences between the groups, chi-squared tests were used. The 5% significance level was used to determine significant differences.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Third mailing respondents were significantly different from other respondents in terms of age group, marital status, employment status, chronic pain presence, physical functioning, mental health and bodily pain, and significantly different from non-respondents in terms of marital status and bodily pain (see Table 1
). Comparing respondents to the first and second mailings with respondents to the third mailing and non-respondents (findings if the third mailing had not been included in the study), we found significant differences for age group, employment status, chronic pain presence, chronic pain grade and mental health status. Comparing respondents to the first, second and third mailings with non-respondents (findings if the third mailing was included in the study), significant differences were found for age group, housing tenure, employment status, chronic pain grade and mental health status.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Significant differences between third mailing respondents and other respondents were found, suggesting that response bias could be an issue. The differences found were broadly in line with previous studies.1,3–5 The inclusion of respondents to the third mailing, however, did not significantly change the overall characteristics of respondents compared with non-respondents, except to decrease the percentage of respondents with chronic pain from 46.4% [95% confidence interval (CI) 43.9–49.0%] to 45.5% (95% CI 43.1–48.0%) and decrease the percentage of non-respondents who owned their home from 71.3% (95% CI 67.3–75.3%) to 68.3% (95% CI 63.2–73.4%).
The main limitation of this study is that it is based on a select group of individuals who had already participated in an earlier study. It has, however, allowed us to examine a large number of socio-demographic, pain and general health variables that are not usually available when assessing the effects of non-response, information that is becoming harder to obtain as data protection and confidentiality rules tighten.
Data collection in epidemiological studies can be costly and slow. A third mailing usually increases the final response rate by only a few percent,3–5 despite being resource intensive. How much time and money should be devoted to obtaining a high response rate in postal epidemiological studies will vary depending on the purpose of the study. When differences do exist between respondents and non-respondents, our results suggest that a third mailing is unlikely to remove many of these differences, partly because the amount of data provided by the third mailing will be small in comparison with the first and second mailings. This study supports our previous suggestion that the effort and resources expended in carrying out a third mailing may not be justified, although further research of this issue is needed.
|
Acknowledgments |
|---|
We thank the general practices and patients for their continued support of this work. We gratefully acknowledge the Scottish Office Home and Health Department for their financial support of the baseline survey, and the Association of Anaesthetists of Great Britain and Ireland for their financial support of the follow-up survey.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
1 Elliott AM, Hannaford PC, Simpson JA, Warskyj M, Ferry S, Owen-Smith J. Should postal epidemiological studies stop doing third mailings: examination of response bias in a health survey of middle aged women living in the UK. Eur J Gen Pract 2001; 7:104–110.
2 Elliott AM, Smith BH, Penny KI, Smith WC, Chambers WA. The epidemiology of chronic pain in the community. Lancet 1999; 354:1248–1252.[CrossRef][ISI][Medline]
3 Emberton M, Black N. Impact of non-response and of late response by patients in a multi-centre surgical outcome audit. Int J Qual Health Care 1995; 7:47–55.
4 Fiset L, Milgrom P, Tarnai J. Dentists’ response to financial incentives in a mail survey of malpractice liability experience. J Public Health Dent 1994; 54:68–72.[ISI][Medline]
5 Paganini-Hill A, Hsu G, Chao A et al. Comparison of early and late respondents to a postal health survey questionnaire. Epidemiology 1993; 4:375–379.[ISI][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  A. D. Steffen, L. N. Kolonel, A. M. Nomura, F. S. Nagamine, K. R. Monroe, and L. R. Wilkens The Effect of Multiple Mailings on Recruitment: The Multiethnic Cohort Cancer Epidemiol. Biomarkers Prev., February 1, 2008; 17(2): 447 - 454. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A Wilson, P Sinfield, S Rodgers, V Hammersley, and T Coleman Drugs to support smoking cessation in UK general practice: are evidence based guidelines being followed? Qual. Saf. Health Care, August 1, 2006; 15(4): 284 - 288. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||