Recruitment to Trials

1
Recruitment to Trials
2
Background

• Recruitment of participants is a VERY important
  issue.
• The general consensus is that most trials under
  recuit.

3
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.

4
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.

5
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 2002Unpublished
6
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.

7
Results

• We emailed 196 authors of individually randomised
  trials.
• 33 were bounced back from invalid email
  addresses.
• We received 79 valid responses (48).

8
Recruitment Problems
9
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.

10
Authors comments

• Facilitate
• Secondary care
• Use previously successful methods
• Few exclusion criteria large sampling frame
• Pilot study

• Hinder
• Competing with other trials.
• Ethics.
• Inaccurate incidence.
• Clinician resistance.
• Narrowly defined population.

11
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.

12
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.

13
DAMASK recruitment
14
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.

15
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.

16
What did we do?

• Increased eligibility criteria.
• Mailed out to more GPs
• Publicity
• Enrolled a 6th Centre.

17
Hip Protector Trial
18
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.

19
Other problems

• Calcium and D trial was a multicentred study.
• One centre was late in starting but eventually
  DID recruit its target.

20
Calcium and D trial
21
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.

22
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?

23
GP Recruitment
24
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.

25
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.

26
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.

27
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 Zelens method does show
  increased recruitment rates.

28
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 2002325766-770
29
Percent of eligible men recruited
30
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

31
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.

32
Natural changes in recruitment rates.
33
DAMASK monthly recruitment
NB no qualitative research intervention.
34
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
  centres performance.

35
Summary

• Recruitment is an important issue.
• Most trials under-recruit.
• Careful attention needs to be paid to recruitment
  issues from the start.