Using postal randomization to replace telephone randomization had no significant effect on recruitment of patients

Stephen D. Brealey, Christine Atwell, Stirling Bryan, Simon Coulton, Helen Cox, Ben Cross, Fiona Fylan, Andrew Garratt, Fiona J. Gilbert, Maureen G. C. Gillan, Maggie Hendry, Kerenza Hood, Helen Houston, David King, Veronica Morton, Jo Orchard, Michael Robling, Ian T. Russell, David Torgerson, Valerie WadsworthClare Wilkinson

Research output: Contribution to journalArticlepeer-review


Objective: To test the effect of postal randomization on recruitment of patients into a randomized trial in primary care.

Study Design and Setting: General practices used a telephone service to randomize patients in our trial. Delays in the start of recruitment at some sites led us to modify the randomization procedure. When new practices took part patients completed and posted baseline materials to the Trial Secretary in York who performed the randomization and informed those concerned of the allocation.

Results: Of the 647 practices who were invited to take part, 130 (45%) of 288 agreed to participate using telephone randomization and 155 (43%) of 359 using the postal method. These practices recruited 553 patients from November 2002 to October 2004 across 11 sites in the United Kingdom. The postal method reduced the number of patients recruited by a factor of 0.86 (95% confidence interval = 0.62-1.20), or 14%. The number of general practitioners working in a practice significantly increased patient recruitment by a factor of 1.12 (1.05-1.20), whereas practice distance from hospital significantly decreased recruitment by a factor of 0.98 (0.97-0.99).

Conclusion: Postal randomization had no significant effect on recruitment of patients into our trial. (C) 2007 Elsevier Inc. All rights reserved.

Original languageEnglish
Pages (from-to)1046-1051
Number of pages6
JournalJournal of Clinical Epidemiology
Issue number10
Publication statusPublished - Oct 2007


  • patient recruitment
  • family practice
  • randomized controlled trial
  • magnetic resonance imaging
  • internal derangement of the knee
  • negative binomial regression
  • KNEE

Cite this