Family Practice Advance Access originally published online on May 29, 2007
Family Practice 2007 24(4):388-394; doi:10.1093/fampra/cmm019
An analysis of factors that predict patient consent to take part in a randomized controlled trial
Department of Primary Care and General Practice, The University of Birmingham, Edgbaston, Birmingham B15 2TT, UK
Correspondence to: Kate Fletcher, Department of Primary Care and General Practice, The University of Birmingham, Edgbaston, Birmingham B15 2TT, UK; Email: k.fletcher{at}bham.ac.uk
Received 26 May 2006; Revised 21 March 2007; Accepted 21 March 2007.
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionDeclarationReferences |
|---|
Background. Recruitment targets of patients to multi-centre primary care-based randomized controlled trials (RCT) are often not met. A critical step in the pathway is whether an eligible patient will give consent.
Objective. To assess whether patient, practice or practitioner characteristics are associated with a patient's likelihood of giving consent to participation in a large primary care-based RCT.
Methods. A cross-sectional study of patients from 260 practices in England and Wales who met the eligibility criteria for an RCT of aspirin versus warfarin for stroke prevention and attended an appointment with their GP to consider trial participation. Logistic regression analysis was used to determine which patient and practitioner factors independently predicted whether or not a patient would give consent to take part in the trial.
Results. Of the 1740 patients, 973 (55.9%) gave consent. On multivariable analysis, patient factors associated with increased likelihood of giving consent were younger age, current use of warfarin and year of recruitment to the trial. Patients with a history of transient ischaemic attack, angina or valve disease were less likely to give consent. Practice/practitioner factors that were associated with increased likelihood of consent were smaller practice size (practices with greater than eight GPs as compared with those with one to two GPs, odds ratio 0.40, 95% confidence interval 0.21–0.75) and older GPs.
Conclusions. The strong association of consent with year of recruitment may be due to changes in trial procedures and investigator training. If so, this has important implications for the conduct of future trials.
Keywords. Consent, primary care, RCT, recruitment.
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionDeclarationReferences |
|---|
Randomized controlled trials (RCTs) are widely accepted as the best way of evaluating treatment effects.1 A key success criterion for RCTs is that a reasonable proportion of eligible patients are included. This ensures both that sufficient participants are recruited in a reasonable timescale, allowing the trial to meet the study power requirement, and that the recruited cohort is representative of the study population.
Recruitment of patients to RCTs is often difficult, with many studies struggling to reach their targets.2,3 In multi-centre trials, local clinicians are relied on to recruit patients, leaving the study centre with limited control over recruitment.4 If clinicians do not recruit enough patients, then studies become more expensive, take longer and fail to achieve their recruitment targets. Primary care-based RCTs often rely on a large number of research sites to recruit patients, and these trials also suffer from problems with patient recruitment. However, reasons for low patient accrual are poorly understood and few methods employed by research teams to improve recruitment are evidence based.5
Many reasons are given for poor recruitment, including overestimation of yield from particular sources, inadequate planning and inability to amend recruitment strategies rapidly if recruitment is slow.6 Extensive literature exists that cites problems with the consent process as being a major barrier to recruitment,2 but these relate mainly to secondary care cancer trials that are likely to have specific problems that do not relate to primary care-based trials looking at other diseases. The literature often focuses on the need to determine patient levels of understanding to ensure that patients have given fully informed consent, but less evidence is available on whether patients have also given fully informed refusal.7 This is an important issue if researchers are to achieve their targets while ensuring that the right patients both enter and decline trial participation. More understanding of the factors that influence consent is needed so that these can be taken into account when planning future trials.
This analysis is based upon a large primary care-based RCT that aimed to recruit patients aged 75 and over with a diagnosis of atrial fibrillation (AF) to a trial of warfarin versus aspirin for prevention of stroke.8 Although our recruitment target was achieved, patient entry was slower than anticipated and required a re-structuring of our recruitment strategy to achieve study power within a reasonable timescale. To identify reasons for this initial delay and to help inform the design and conduct of future primary care trials, we explored what patient, practice and/or practitioner characteristics predicted whether an individual would consent to take part in the trial.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionDeclarationReferences |
|---|
The methods of the trial are described elsewhere.8 In brief, potentially eligible patients were identified through computer searches for AF supplemented by opportunistic screening of the pulse. An electrocardiograph (ECG) showing AF was required to confirm eligibility for the study. One doctor per practice acted as the local investigator for the study. The research team trained investigators in study procedures and Good Clinical Practice over the 4-year period (2001–2004) that patients were recruited. During this time, changes were made both to study procedures and to investigator training (Table 1). Trial procedures were simplified and payments to practices were altered.
|
Once trained, investigators screened their patients to determine eligibility. The medical records of patients aged over 74 who were found to have ECG confirmed AF were examined for presence of any study exclusion criteria (rheumatic heart disease, history of intracranial haemorrhage, major gastrointestinal haemorrhage in last 5 years, active peptic ulcer, allergy to study medications). If the primary care physician (GP) was in equipoise as to whether or not they should be treated with aspirin or warfarin after review of relevant risk factors for stroke and haemorrhage, the patient was invited to attend a study clinic.8 Eligible patients were sent a patient information sheet and a letter inviting them to see their primary care physician to discuss trial participation. During this appointment, the patient made the decision whether or not to take part in the study.
This analysis focuses on the 1763 patients from 260 practices in England and Wales who demonstrated that they were willing to consider study participation by attending the study clinic. Information on past medical history and medications was collected from the medical records. Disability score was completed by the patients during the study clinic appointment. Practice and practitioner characteristics were collected in a questionnaire completed by each practice on recruitment to the study. Data on the year that the practice and the patient were recruited to the study and the number of patients attending a study clinic appointment from each practice were available in the study database.
The effect of patient, practice and practitioner factors on patient consent was explored using logistic regression, univariable for factors considered one at a time and multivariable for factors considered in combination. Further multivariable analysis was conducted using a multi-level mixed effects model.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionDeclarationReferences |
|---|
In total, 1763 people attended the study appointment. Twenty-three (1.3%) had absolute exclusion criteria and so were excluded from this analysis. Of the 1740 included, 51% were male, 94% white, 32% already taking warfarin, 44% already taking aspirin, 9% had a history of transient ischaemic attack (TIA), 6% had a history of stroke and 70% were already known to have AF. The mean age of attendees was 81.9 years (range 74–96). In total, 973 (55.9%) patients consented to be randomized, although there was a wide variation between local investigators in the proportion of eligible patients who gave their consent (0–100%) (Fig. 1).
|
Patient factors
On univariable analysis, people already on warfarin were more likely to give consent. Patients offered trial entry in 2004 were almost twice as likely to participate as those considered in 2001. Patient socio-economic status also predicted whether or not a patient would consent, although there was not a simple linear association between the quartiles. Patients on aspirin and those with an increased risk of stroke (history of TIA, angina or non-rheumatic valve disease) were less likely to give their consent, as were people aged over 85. There was no significant difference, however, between the two younger age groups (Table 2).
|
On multivariable analysis taking both patient and practice factors into account, patient deprivation and use of aspirin were no longer significant. All other factors that were significant in univariable analysis remained so with the logistic regression model (Table 2). Patient deprivation was eliminated, as it was strongly associated with the number of practitioners in a practice, the number of patients attending a study clinic appointment and the date that the patient entered the study. Use of aspirin was no longer significant because there was a strong inverse association with use of warfarin. With the multi-level model, the pattern of effects was very similar to that seen with the logistic regression, with the regression coefficients from the two analyses showing a correlation of 0.95. However, as expected, effects were generally less significant. Previous TIA was no longer significant and the significance of valve disease was marginal (P = 0.06). Other factors remained significant.
Practice factors
On univariable analysis all practice factors, with the exception of primary care physician sex, had a significant effect on a patient's likelihood of giving consent. Patients registered with practices who attended study training during 2004 were almost twice as likely to consent as patients in practices who trained in 2001. Similarly, patients in smaller practices were most likely to consent. There was a clear linear association with practice size, with people being less likely to consent as the number of practitioners within their practice increased. Patients in practices where fewer people attended a study clinic were also more likely to consent. Patients of longer practising primary care physicians [those registered with the General Medical Council (GMC) prior to 1975] were also more likely to give their consent.
On multivariable analysis, the year that the practice was recruited to the study was no longer significant, due to its strong association with the year of patient recruitment. All other factors that were significant in univariable analysis remained so (Table 2). With the multi-level model, year of full GMC registration was no longer significant.
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionDeclarationReferences |
|---|
This is the largest study to look at factors influencing patient consent to a multi-centre primary care-based RCT. We found both patient and practice factors independently predicted whether a patient would consent to take part.
Patient factors
The patient factors may reflect general attitudes towards trials, with understanding and acceptance of issues such as randomization, uncertainty and acceptability of treatment arms playing a part in a patient's decision on whether to participate.9–11 Many of the patient factors could also be specific to the interventions (warfarin and aspirin) being tested,3 with patients already on warfarin being more likely to give consent. This presumably reflects the inconvenience and unpopularity of the treatment.12 Patients with a history of valve disease, angina or TIA were less likely to give consent. This could be due to the impact of perceived illness severity on trial participation,10 or it may be that these patients had a strong preference for warfarin as they felt that they had more to benefit from this therapy.11,13 The reduced likelihood of older patients (in this case over the age of 85) giving consent has been reported in other studies.14 The year that the patient was invited to enter the study was the most important patient factor, with a patient approached in 2004 being twice as likely to take part as one approached in 2001.
There are several possible explanations as to why there was such a strong association between when a patient was recruited and whether or not they gave consent. It is possible that changes to investigator training methods had an impact on the way in which investigators conducted the consent process (Table 1). Many doctors encounter problems obtaining consent satisfactorily,15 and changes in training may have helped to alleviate some of these. An individual patient meta-analysis published at the end of 200216 was incorporated into the training sessions in 2003. This helped demonstrate the uncertainty as to whether the benefits of warfarin outweighed potential harm as compared to aspirin in the study age group. Investigators trained after this date may, therefore, have been more confident that equipoise existed for individual patients,4 possibly resulting in an increased likelihood that a patient would consent to take part. There were several changes to trial procedures (see Table 1) which might have had an effect, though most of these changes will have affected the trial ‘upstream’ of the consent process. These amendments potentially had a two-fold effect. Simplification of study procedures resulted in a reduced workload for investigators, minimizing the negative impact that a time-consuming protocol has on a physician's ability to engage with the research,3 thus making the project more attractive to them.17 Changes to both payment methods and study procedures reduced the time between investigator training and commencement of patient recruitment (from approximately 4 months to approximately 2 months) so investigators were possibly more familiar with the trial rationale and protocol than they would be after a lengthy delay. Haidich and Ioannidis18 demonstrated in a secondary care setting that the longer the gap between study commencement and enrolment of the first patient, the less likely the site is to recruit well. The broadening of the inclusion criteria to allow people on warfarin to be considered may be a partial explanation of the temporal trend observed in patient consent, though it is not clear why there should have been a 2-year lag before it had an effect. It may be that there was a secular change in patient attitude, either to trial participation in general or value of warfarin in particular over the study period. Warfarin use has become more common in older people over time,19 and this may have influenced willingness to participate.
Practice and practitioner factors
In this study, older physicians were more likely to obtain consent than younger physicians, although there was not a simple linear relationship with age, and a higher proportion of patients from smaller practices gave consent than from larger practices. Practitioners can be uncomfortable describing personal equipoise and the randomization process20 and will often emphasize aspects of the trial that they expect patients to understand more easily.21 These difficulties may be mitigated in smaller practices where continuity of care is possible, and physicians can get to know their patients better, giving patients a feeling of security22 and the primary care physician a better understanding of23 and increased feeling of responsibility towards their patient.24 This enables them to be both more selective about which patients they invite, and more confident in their relationship.
The effect of age may reflect a number of factors, including variations in consulting style or differing attitudes to warfarin. Older doctors are seen to be more willing to listen and more reassuring than younger doctors, and are viewed more positively by older patients.25
Strengths and limitations
The patients in this analysis were all considered eligible for the study by their primary care physician and all attended a clinic appointment, so were not averse to trial participation per se. Therefore, the factors considered here impact directly on the process of obtaining consent, rather than on other parts of the patient pathway leading to trial participation, such as willingness to attend, or physician decisions about eligibility. Another strength is the large size, and the high degree of data completeness, with over 90% data completeness (worst data item was prior aspirin use, with 7.8% missing data).
There are some weaknesses to this analysis. Although it demonstrates that a wide variety of factors influence recruitment, it does not allow us to understand how or why these characteristics have such a significant effect. It is based around a single trial, so may be difficult to generalize to other primary care trials. However, there are findings here that are of general interest. The effect of practice size and the impact of year of recruitment are unlikely to have been primarily related to the nature of the trial interventions. Furthermore, Donovan et al.26 demonstrated that the way that study treatments are described to a patient influences whether or not they are willing to take part, so it is possible that even study-specific factors such as prior use of warfarin were influenced by how the investigator conducted the consent process.
Implications for future primary care-based RCTs
The two-fold difference in the likelihood of whether or not a patient would give consent over the time course of the study having adjusted for influence of other patient and practice factors suggests that the conduct of a trial can have a dramatic effect on patient recruitment. It is possible that changes made to investigator training, study procedures and payment methods account for this finding. As the analysis only considered patients who were eligible for the study and who went through the informed consent procedure, the changes must have affected either the type of patients attending this appointment, or what was said to them during the consent process. The evidence base for how to train investigators and conduct trials in primary care is poor.5 This analysis suggests that research into this area could yield important benefits in terms of enhancing recruitment to future studies.
|
Declaration |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionDeclarationReferences |
|---|
Funding: The study was funded by The Medical Research Council.
Ethical approval: Approval for the main trial was obtained from West Midlands Multi-Centre Research Ethics Committee in 2000 and from all appropriate Local Research Ethics Committees and Primary Care Trusts.
Conflicts of interest: No authors have any conflict of interest.
|
Notes |
|---|
Fletcher K, Mant J, Holder R, Fitzmaurice D, Lip GYH and Hobbs FDR. An analysis of factors that predict patient consent to take part in a randomized controlled trial. Family Practice 2007; 24: 388–394.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionDeclarationReferences |
|---|
1 Altman DG. Better reporting of randomised controlled trials: the CONSORT statement. Br Med J (1996) 313:570–571.
2 Ross S, Grant A, Counsell C, Gillespie W, Russell I, Prescott R. Barriers to participation in randomised controlled trials: a systematic review. J Clin Epidemiol (1999) 52:1143–1156.[CrossRef][ISI][Medline]
3 Prescott RJ, Counsell CE, Gillespire WJ, et al. Factors that limit the quality, number and progress of randomised controlled trials. Health Technol Assess (1999) 3:1–143.[Medline]
4 Gates S, Brocklehurst P, Campbell M, Elbourne D. Recruitment to multicentre trials. Br J Obstet Gynaecol (2004) 111:3–5.
5 Foy R, Parry J, Duggan A, et al. How evidence based are recruitment strategies to randomised controlled trials in primary care? Experience from seven studies. Fam Pract (2003) 20:83–92.
6 Lovato L, Kristin M, Hertert S, et al. Recruitment for controlled clinical trials: literature summary and annotated bibliography. Control Clin Trials (1997) 18:328–357.[CrossRef][ISI][Medline]
7 Robinson EJ, Kerr CEP, Stevens AJ, et al. Lay public's understanding of equipoise and randomisation in randomised controlled trials. Health Technol Assess (2005) 9:1–192.[ISI][Medline]
8 Mant JWF, Richards SH, Hobbs FD, et al. Protocol for Birmingham Atrial Fibrillation Treatment of the Aged study (BAFTA): a randomised controlled trial of warfarin versus aspirin for stroke prevention in the management of atrial fibrillation in an elderly primary care population [ISRCTN89345269]. BMC Cardiovasc Disord (2003) 3:9.[CrossRef][Medline]
9 Featherstone K, Donovan J. Random allocation or allocation at random? Patients' perspectives of participation in a randomised controlled trial. Br Med J (1998) 317:1177–1180.
10 Schaeffer M, Krantz D, Wichman A, Masur H, Reed E, Vinicky J. The impact of disease severity on the informed consent process in clinical research. Am J Med. 100:261–268.
11 Mills N, Donovan J, Smith M, Jacoby A, Neal D, Hamdy F. Perceptions of equipoise are crucial to trial participation: a qualitative study of men in the ProtecT study. Control Clin Trials (2003) 24:272–282.[CrossRef][ISI][Medline]
12 Fuller R, Dudley N, Blacktop J. Avoidance hierarchies and preferences for anticoagulation—semi-qualitative analysis of older patients' views about stroke prevention and the use of warfarin. Age Ageing (2004) 33:608–611.
13 Jack WJ, Chetty U, Rodger A. Recruitment to a prospective breast conservation trial: why are so few patients randomised? Br Med J (1990) 301:83–85.[ISI][Medline]
14 Petty DR, Zermansky AG, Raynor DK, et al. ‘No thank you’: why elderly patients declined to participate in a research study. Pharm World Sci (2001) 23:22–27.[CrossRef][ISI][Medline]
15 Mason SA, Allmark PJ. Obtaining informed consent to neonatal randomised controlled trials: interviews with parents and clinicians in the Euricon study. Lancet (2000) 356:2045–2051.[CrossRef][ISI][Medline]
16 Van Walraven C, Hart RG, Singer DE, et al. Oral anticoagulants vs aspirin in nonvalvular atrial fibrillation: an individual patient meta-analysis. J Am Med Assoc (2002) 288:2441–2448.
17 Hetzel MR, Lee T, Prescott RJ, et al. Multi-centre clinical respiratory research: a new approach. J R Coll Physicians Lond (1998) 32:412–416.[ISI][Medline]
18 Haidich A, Ioannidis J. Late-starter sites in randomised controlled trials. J Clin Epidemiol (2003) 56:408–415.[CrossRef][ISI][Medline]
19 Garcia D, Regan S, Crowther M, Hughes R, Hylek E. Warfarin maintenance dosing patterns in clinical practice: implications for safer anticoagulation in the elderly population. Chest (2005) 127:2049–2056.[CrossRef][ISI][Medline]
20 Jenkins VA, Fallowfield LJ, Souhami A, Satwell M. How do doctors explain randomised clinical trials to their patients? Eur J Cancer (1999) 35:1187–1193.[CrossRef][ISI][Medline]
21 Verheggen FW, Jonkers R, Kok G. Patients' perceptions on informed consent and the quality of information disclosure in clinical trials. Patient Educ Couns (1996) 29:137–153.[CrossRef][ISI][Medline]
22 Von Bultzingslowen I, Eliasson G, Sarvimaki A, Mattsson B, Hjortdahl P. Patients' views on interpersonal continuity in primary care: a sense of security based on four core foundations. Fam Pract (2006) 23:210–219.
23 Kearley K, Freeman G, Heath A. An exploration of the value of the personal doctor-patient relationship in general practice. Br J Gen Pract (2001) 712–718.
24 Hjortdahl P. Continuity of care: general practitioners' knowledge about and sense of responsibility toward their patients. Fam Pract (1992) 9:3–8.
25 Mckinstry B, Yang S. Do patients care about the age of their general practitioner? A questionnaire survey in five practices. Br J Gen Pract (1994) 44:349–351.[ISI][Medline]
26 Donovan J, Mills N, Smith M, et al. Improving design and conduct of randomised trials by embedding them in qualitative research: ProtecT (prostate testing for cancer and treatment) study. Br Med J (2002) 325:766–770.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||