Recruitment to Clinical Trials

Recruitment to Trials Background  Recruitment of participants is a VERY important issue. The general consensus is that most trials under recuit.  Poor Recruitment  Poor or slow recruitment to trials leads to the following problems:    Increased risk of Type II error (concluding, erroneously there is no difference); Delay in implementing research findings; Research commissioners may seek other INFERIOR but quicker evaluative methods of research. How common is the problem? There is little quantitative data to support the qualitative view that many trials under recruit.  One study in USA noted that of 41 trials 34% failed to achieve sample size only another 34% got sample size on time.  Survey  To ascertain whether poor trial recruitment was as poor as believed we undertook a survey of corresponding authors of a sample of trials published in 2000 and 2001. Puffer & Torgerson 2002;Unpublished Method We identified all, individually, randomised trials published in the BMJ and Lancet years 2000 to 2001.  Corresponding authors were emailed a brief questionnaire asking about recruitment problems.  One email reminder was sent.  Results We emailed 196 authors of individually randomised trials.  33 were bounced back from invalid email addresses.  We received 79 valid responses (48%).  Recruitment Problems 60 50 40 30 20 10 0 Problems Extension Both Funds Recruitment    Multicentred trials had significantly more problems than single centred studies (51% vs 23%; p = 0.02). Primary and Secondary trials had similar problems 43% and 39% primary vs secondary. No significant age difference 48 years vs 52 years for good recruiters vs poor recruiters. Authors‟ comments  Facilitate    Hinder    Secondary care Use previously successful methods Few exclusion criteria large sampling frame Pilot study     Competing with other trials. Ethics. Inaccurate incidence. Clinician resistance. Narrowly defined population. Summary of Comments Recruitment is a problem in the MAJORITY of trials.  Worse in multicentred trials.  Need to use pilot studies.  Need to use tried and tested strategies.  Need to use large sampling frames.  Examples of poor recruitment York backpain trial needed 300 attained 180.  MRC backpain needed 1300 (achieved target but needed extra funds and extended recruitment).  Vein graft trial needed 1200 got 100 (trial collapsed).  SAPPHIRE trial is under-recruiting.  DAMASK recruitment Damask Overall Recruitment - Oct 2003 600 500 400 300 200 100 0 Nov-02 Dec-02 Jan-03 Feb-03 Mar-03 Apr-03 May- Jun-03 Jul-03 Aug- Sep-03 Oct-03 Nov-03 Dec-03 Jan-04 Feb-04 Mar-04 03 03 Nov-02 Dec-02 Jan-03 Feb-03 Mar-03 Apr-03 Target Actual 0 2 0 4 0 5 0 12 0 24 42 39 MayAugJun-03 Jul-03 Sep-03 Oct-03 Nov-03 Dec-03 Jan-04 Feb-04 Mar-04 03 03 84 57 126 85 168 107 210 135 252 165 294 183 336 378 420 462 504 Hip protector trial Aimed to recruit 4500 women at risk of hip fracture to wear hip protectors or act as controls.  Used a combination of GP practices and publicity.  Expected 10% pick up. Pilot showed 2.5%.  Other problems Hip protector trial was a multicentred study (orginally 5 centres).  Two centres did not start on time, 1 was abandoned, the other started late and only got 50% of expected recruits.  What did we do? Increased eligibility criteria.  Mailed out to more GPs  Publicity  Enrolled a 6th Centre.  Hip Protector Trial Hip Protector Trial Recruitment 700 600 4000 3500 3000 2500 2000 300 200 100 0 1500 1000 500 0 Recruitment per Month Total Recruited 500 400 Recruitment per Month Total Recruited 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Time (months) Calcium and D recruitment     Again we recruited women with 1+ risk factors for hip fracture over 70 years. Again overestimated recruitment rates at 10%. Pilot showed a recruitment rate of 5%. We doubled the number of GPs to be included in study. Other problems Calcium and D trial was a multicentred study.  One centre was late in starting but eventually DID recruit its target.  Calcium and D trial Calcium and Vitamin D Trial Recruitment 800 3500 700 3000 600 2500 Recruitment per Month 2000 400 1500 300 1000 200 Total Recruited 500 Recruitment per Month Total Recruited 100 500 0 1 2 3 4 5 6 7 8 9 10 11 12 Time (months) 0 Recruiting Doctors Often recruting trial participants is only part of the problem need to recruit doctors as well.  Primary care physicians, unlike secondary care doctors, do not have a career incentive to become involved in research.  GP recruitment Rate We paid both sets of GPs £50 to mail out prepaid envelopes.  For calcium and D we also paid GPs £30 per patient randomised to active treatment.  Required practice nurse to see patient for 20 minutes.  Which trial recruited most GPs?  GP Recruitment Number of women aged 70+ Participant Recruitment 80000 60000 40000 20000 0 48987 3452 Calcium and Vitamin D 66103 3525 Hip Protector s Women contacted Women agreed to participate Trial MRC RECORD Trial MRC RECORD trial recruited people from fracture clinics.  Under recruited due to over optimistic recruitment predictions.  BUT could not increase number of centres easily due to budget restraints.  What was done? Payment to trial centres was made conditional on recruitment rates. Initial contracts for 6 months of a research nurse.  Poor performing centres were closed or finance was reduced.  Trial extension sought and was given.  RECORD result  RECORD trial will be late and not achieved initial sample size. Event rate was higher than expected so loss of power will not occur. Evidence based recruitment?  Not many RCTs of different methods of recruitment.       Cooper et al, showed using a patient preference trial had not + or – effect on recruitment rates. RCT of nurses vs consultants for prostate cancer trial – no difference. RCT of trial co-ordinator visits to centres vs mailing recruitment packs in French cancer trial – no difference. Two open RCTs showed increase in recruitment. Education of GPs shows increase in recruitment Case control data of Zelen‟s method does show increased recruitment rates. Qualitative methods & recruitment  Donovan and colleagues introduced a „rolling‟ qualitative research process into patient recruitment for a trial on prostate cancer. Donovan et al. BMJ 2002;325:766-770 Percent of eligible men recruited 70 60 50 40 30 20 10 0 0 1 2 3 4 Authors‟ conclusions  “Embedding the controversial ProtecT randomised trial within qualitative research allowed detailed investigation of the presentation of study information by recruiters and its interpretation by participants” “Changes to the content and delivery of study information increased recruitment rates from 40% to 70%”  Oh Dear    As we ALL now know before and after studies CANNOT infer causality. Interesting data BUT we NEED an RCT to be sure that this intervention does improve recruitment rates. There is a natural tendency for recruitment rates to be poor at the start of a study anyway. „Natural‟ changes in recruitment rates. Hip Protector Trial Recruitment 700 600 4000 3500 3000 2500 2000 300 200 100 0 1500 1000 500 0 Recruitment per Month Total Recruited 500 400 Recruitment per Month Total Recruited 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Time (months) DAMASK monthly recruitment 30 25 20 15 10 5 0 N D J F M A M J J A S O NB – no qualitative research intervention. Recruitment Solutions? Assume from day 1 recruitment WILL be difficult and delayed.  Find solutions from the beginning (e.g. extra ethics permission, loosening inclusion criteria).  In multicentred trials tailor finance to each centre‟s performance.  Summary Recruitment is an important issue.  Most trials under-recruit.  Careful attention needs to be paid to recruitment issues from the start. 