Factors distinguishing general practitioners who more readily participated in a large randomized trial were identified

What works to recruit general practices to trials? A rapid review

Background: Recruitment challenges are a barrier to the conduct of trials in general practice, yet little is known about which recruitment strategies work best to recruit practices for randomised controlled trials (RCTs). We aimed to describe the types of strategies used to recruit general practices for trials and synthesize any available evidence of effectiveness. Methods: We conducted a rapid evidence review in line with guidance from Tricco et al. Eligible studies reported or evaluated any strategy to improve practice recruitment to participate in clinical or implementation RCTs. PubMed, Embase, and Cochrane Central Library were searched from inception to June 22 nd, 2021. Reference lists of included studies were screened. Data were synthesized narratively. Results: Over 9,162 articles were identified, and 19 studies included. Most (n=13, 66.7%) used a single recruitment strategy. The most common strategies were: in-person practice meetings/visits by the research team (n=12, 63.2%); phone calls (n=10, 52.6%); financial incentives (n=9, 47.4%); personalised emails (n=7, 36.8%) or letters (n=6, 52.6%) (as opposed to email 'blasts' or generic letters); targeting practices that participated in previous studies or with which the team had existing links (n=6, 31.6%) or targeting of practices within an existing practice or research network (n=6, 31.6%).  Three studies reporting recruitment rates >80%, used strategies such as invitation letters with a follow-up phone call to non-responders, presentations by the principal investigator and study coordinator, or in-person meetings with practices with an existing affiliation with the University or research team.  Conclusions: Few studies directly compared recruitment approaches making it difficult to draw conclusions about their comparative effectiveness. However, the role of more personalised letter/email, in-person, or phone contact, and capitalising on existing relationships appears important. Further work is needed to standardise how recruitment methods are reported and to directly compare different recruitment strategies within one study . PROSPERO registration: CRD42021268140 (15/08/2021).

Developing and evaluating packages to support implementation of quality indicators in general practice: the ASPIRE research programme, including two cluster RCTs

Background

Dissemination of clinical guidelines is necessary but seldom sufficient by itself to ensure the reliable uptake of evidence-based practice. There are further challenges in implementing multiple clinical guidelines and clinical practice recommendations in the pressurised environment of general practice.

Objectives

We aimed to develop and evaluate an implementation package that could be adapted to support the uptake of a range of clinical guideline recommendations and be sustainably integrated within general practice systems and resources. Over five linked work packages, we developed ‘high-impact’ quality indicators to show where a measurable change in clinical practice can improve patient outcomes (work package 1), analysed adherence to selected indicators (work package 2), developed an adaptable implementation package (work package 3), evaluated the effects and cost-effectiveness of adapted implementation packages targeting four indicators (work package 4) and examined intervention fidelity and mechanisms of action (work package 5).

Setting and participants

Health-care professionals and patients from general practices in West Yorkshire, UK.

Design

We reviewed recommendations from existing National Institute for Health and Care Excellence clinical guidance and used a multistage consensus process, including 11 professionals and patients, to derive a set of ‘high-impact’ evidence-based indicators that could be measured using routinely collected data (work package 1). In 89 general practices that shared data, we found marked variations and scope for improvement in adherence to several indicators (work package 2). Interviews with 60 general practitioners, practice nurses and practice managers explored perceived determinants of adherence to selected indicators and suggested the feasibility of adapting an implementation package to target different indicators (work package 3). We worked with professional and patient panels to develop four adapted implementation packages. These targeted risky prescribing involving non-steroidal anti-inflammatory and antiplatelet drugs, type 2 diabetes control, blood pressure control and anticoagulation for atrial fibrillation. The implementation packages embedded behaviour change techniques within audit and feedback, educational outreach and (for risky prescribing) computerised prompts. We randomised 178 practices to implementation packages targeting either diabetes control or risky prescribing (trial 1), or blood pressure control or anticoagulation (trial 2), or to a further control (non-intervention) group, and undertook economic modelling (work package 4). In trials 1 and 2, practices randomised to the implementation package for one indicator acted as control practices for the other package, and vice versa. A parallel process evaluation included a further eight practices (work package 5).

Main outcome measures

Trial primary end points at 11 months comprised achievement of all recommended levels of glycated haemoglobin, blood pressure and cholesterol; risky prescribing levels; achievement of recommended blood pressure; and anticoagulation prescribing.

Results

We recruited 178 (73%) out of 243 eligible general practices. We randomised 80 practices to trial 1 (40 per arm) and 64 to trial 2 (32 per arm), with 34 non-intervention controls. The risky prescribing implementation package reduced risky prescribing (odds ratio 0.82, 97.5% confidence interval 0.67 to 0.99; p = 0.017) with an incremental cost-effectiveness ratio of £2337 per quality-adjusted life-year. The other three packages had no effect on primary end points. The process evaluation suggested that trial outcomes were influenced by losses in fidelity throughout intervention delivery and enactment, and by the nature of the targeted clinical and patient behaviours.

Limitations

Our programme was conducted in one geographical area; however, practice and patient population characteristics are otherwise likely to be sufficiently diverse and typical to enhance generalisability to the UK. We used an ‘opt-out’ approach to recruit general practices to the randomised trials. Subsequently, our trial practices may have engaged with the implementation package less than if they had actively volunteered. However, this approach increases confidence in the wider applicability of trial findings as it replicates guideline implementation activities under standard conditions.

Conclusions

This pragmatic, rigorous evaluation indicates the value of an implementation package targeting risky prescribing. In broad terms, an adapted ‘one-size-fits-all’ approach did not consistently work, with no improvement for other targeted indicators.

Future work

There are challenges in designing ‘one-size-fits-all’ implementation strategies that are sufficiently robust to bring about change in the face of difficult clinical contexts and fidelity losses. We recommend maximising feasibility and ‘stress testing’ prior to rolling out interventions within a definitive evaluation. Our programme has led on to other work, adapting audit and feedback for other priorities and evaluating different ways of delivering feedback to improve patient care.

Trial registration

Current Controlled Trials ISRCTN91989345.

Funding

This project was funded by the National Institute for Health Research (NIHR) Programme Grants for Applied Research programme and will be published in full in Programme Grants for Applied Research; Vol. 8, No. 4. See the NIHR Journals Library website for further project information.

A prospective cohort and extended comprehensive-cohort design provided insights about the generalizability of a pragmatic trial: the ProtecT prostate cancer trial

OBJECTIVES

Randomized controlled trials (RCTs) deliver robust internally valid evidence but generalizability is often neglected. Design features built into the Prostate testing for cancer and Treatment (ProtecT) RCT of treatments for localized prostate cancer (PCa) provided insights into its generalizability.

STUDY DESIGN AND SETTING

Population-based cluster randomization created a prospective study of prostate-specific antigen (PSA) testing and a comprehensive-cohort study including groups choosing treatment or excluded from the RCT, as well as those randomized. Baseline information assessed selection and response during RCT conduct.

RESULTS

The prospective study (82,430 PSA-tested men) represented healthy men likely to respond to a screening invitation. The extended comprehensive cohort comprised 1,643 randomized, 997 choosing treatment, and 557 excluded with advanced cancer/comorbidities. Men choosing treatment were very similar to randomized men except for having more professional/managerial occupations. Excluded men were similar to the randomized socio-demographically but different clinically, representing less healthy men with more advanced PCa.

CONCLUSION

The design features of the ProtecT RCT provided data to assess the representativeness of the prospective cohort and generalizability of the findings of the RCT. Greater attention to collecting data at the design stage of pragmatic trials would better support later judgments by clinicians/policy-makers about the generalizability of RCT findings in clinical practice.

Variations in achievement of evidence-based, high-impact quality indicators in general practice: An observational study

BACKGROUND

There are widely recognised variations in the delivery and outcomes of healthcare but an incomplete understanding of their causes. There is a growing interest in using routinely collected 'big data' in the evaluation of healthcare. We developed a set of evidence-based 'high impact' quality indicators (QIs) for primary care and examined variations in achievement of these indicators using routinely collected data in the United Kingdom (UK).

METHODS

Cross-sectional analysis of routinely collected, electronic primary care data from a sample of general practices in West Yorkshire, UK (n = 89). The QIs covered aspects of care (including processes and intermediate clinical outcomes) in relation to diabetes, hypertension, atrial fibrillation, myocardial infarction, chronic kidney disease (CKD) and 'risky' prescribing combinations. Regression models explored the impact of practice and patient characteristics. Clustering within practice was accounted for by including a random intercept for practice.

RESULTS

Median practice achievement of the QIs ranged from 43.2% (diabetes control) to 72.2% (blood pressure control in CKD). Considerable between-practice variation existed for all indicators: the difference between the highest and lowest performing practices was 26.3 percentage points for risky prescribing and 100 percentage points for anticoagulation in atrial fibrillation. Odds ratios associated with the random effects for practices emphasised this; there was a greater than ten-fold difference in the likelihood of achieving the hypertension indicator between the lowest and highest performing practices. Patient characteristics, in particular age, gender and comorbidity, were consistently but modestly associated with indicator achievement. Statistically significant practice characteristics were identified less frequently in adjusted models.

CONCLUSIONS

Despite various policy and improvement initiatives, there are enduring inappropriate variations in the delivery of evidence-based care. Much of this variation is not explained by routinely collected patient or practice variables, and is likely to be attributable to differences in clinical and organisational behaviour.

Impact of financial incentives on alcohol intervention delivery in primary care: a mixed-methods study

BACKGROUND

Local and national financial incentives were introduced in England between 2008 and 2015 to encourage screening and brief alcohol intervention delivery in primary care. We used routine Read Code data and interviews with General Practitioners (GPs) to assess their impact.

METHODS

A sequential explanatory mixed-methods study was conducted in 16 general practices representing 106,700 patients and 99 GPs across two areas in Northern England. Data were extracted on screening and brief alcohol intervention delivery for 2010-11 and rates were calculated by practice incentive status. Semi-structured interviews with 14 GPs explored which factors influence intervention delivery and recording in routine consultations.

RESULTS

Screening and brief alcohol intervention rates were higher in financially incentivised compared to non-incentivised practices. However absolute rates were low across all practices. Rates of short screening test administration ranged from 0.05% (95% CI: 0.03-0.08) in non-incentivised practices to 3.92% (95% CI: 3.70-4.14) in nationally incentivised practices. For the full AUDIT, rates were also highest in nationally incentivised practices (3.68%, 95% CI: 3.47-3.90) and lowest in non-incentivised practices (0.17%, 95% CI: 0.13-0.22). Delivery of alcohol interventions was highest in practices signed up to the national incentive scheme (9.23%, 95% CI: 8.91-9.57) and lowest in non-incentivised practices (4.73%, 95% CI: 4.50-4.96). GP Interviews highlighted a range of influences on alcohol intervention delivery and subsequent recording including: the hierarchy of different financial incentive schemes; mixed belief in the efficacy of alcohol interventions; the difficulty of codifying complex conditions; and GPs' beliefs about patient-centred practice.

CONCLUSIONS

Financial incentives have had some success in encouraging screening and brief alcohol interventions in England, but levels of recorded activity remain low. To improve performance, future policies must prioritise alcohol prevention work within the quality and outcomes framework, and address the values, attitudes and beliefs that shape how GPs' provide care.

Exceeding the recruitment target in a primary care paediatric trial: an evaluation of the Choice of Moisturiser for Eczema Treatment (COMET) feasibility randomised controlled trial

Background

Recruiting to target in randomised controlled trials of investigational medicinal products (CTIMPs) in primary care and paediatric populations is notoriously difficult. More evidence is needed for effective recruitment strategies in these settings. We report on the impact of different recruitment strategies used in the Choice of Moisturiser in Eczema Treatment (COMET) study – a feasibility trial comparing the effectiveness of four emollients for the treatment of childhood eczema – recruiting via general practitioner (GP) surgeries.

Methods

Initially, 16 GP practices invited potentially eligible children to take part in the trial by sending an invitation letter (self-referral pathway) or by consenting and randomising them into the study during a visit to the practice (in-consultation referral). Measures implemented during the study to maximise accrual included signing up six additional GP practices, increasing the upper age limit eligibility criterion from 3 to 5 years, and permitting healthcare professionals other than doctors to confirm participant eligibility. We used descriptive statistics and univariate linear regression models to explore associations with practice recruitment rates.

Results

A total of 197 participants were recruited, exceeding the target of 160. Of these, 107 children entered via self-referral and 90 by in-consultation pathways. Of the recruited population, 12.6 % were aged between 3 and 5 years (the raised upper age limit). The six additional practices contributed 37.4 % (40 of 107) of participants recruited by self-referral. Only almost one-third (18 of 56 [32.1 %]) of potential recruiting clinicians recruited one or more participants in-consultation, which was a more problematic pathway because of data verification issues. Three research nurses and a pharmacist from four practices recruited 48.9 % (44 of 90) of participants via this pathway. Univariate linear regression models showed no evidence of association between the number of children recruited via the self-referral pathway by practice and practice list size (p = 0.092) or practice deprivation decile (p = 0.270), but practice deprivation was associated with a higher number of children recruited in-consultation (p = 0.020) by practice.

Conclusions

Self-referral and in-consultation recruitment yielded similar numbers, but the in-consultation pathway was more problematic. Future trials of this type should consider the condition, normal care pathway and number of potentially eligible children and be prepared to use multiple recruitment strategies to achieve recruitment targets.

Trial registration

ISRCTN21828118. Registered on 1 May 2014.

EudraCT2013-003001-26. Registered on 23 Dec 2013.

Optimizing primary care research participation: a comparison of three recruitment methods in data-sharing studies

BACKGROUND

Recruitment of representative samples in primary care research is essential to ensure high-quality, generalizable results. This is particularly important for research using routinely recorded patient data to examine the delivery of care. Yet little is known about how different recruitment strategies influence the characteristics of the practices included in research.

OBJECTIVE

We describe three approaches for recruiting practices to data-sharing studies, examining differences in recruitment levels and practice representativeness.

METHODS

We examined three studies that included varying populations of practices from West Yorkshire, UK. All used anonymized patient data to explore aspects of clinical practice. Recruitment strategies were 'opt-in', 'mixed opt-in and opt-out' and 'opt-out'. We compared aggregated practice data between recruited and not-recruited practices for practice list size, deprivation, chronic disease management, patient experience and rates of unplanned hospital admission.

RESULTS

The opt-out strategy had the highest recruitment (80%), followed by mixed (70%) and opt-in (58%). Practices opting-in were larger (median 7153 versus 4722 patients, P = 0.03) than practices that declined to opt-in. Practices recruited by mixed approach were larger (median 7091 versus 5857 patients, P = 0.04) and had differences in the clinical quality measure (58.4% versus 53.9% of diabetic patients with HbA1c ≤ 59 mmol/mol, P < 0.01). We found no differences between practices recruited and not recruited using the opt-out strategy for any demographic or quality of care measures.

CONCLUSION

Opt-out recruitment appears to be a relatively efficient approach to ensuring participation of typical general practices. Researchers should, with appropriate ethical safeguards, consider opt-out recruitment of practices for studies involving anonymized patient data sharing.

Preventing and Lessening Exacerbations of Asthma in School-aged children Associated with a New Term (PLEASANT): Recruiting Primary Care Research Sites-the PLEASANT experience

BACKGROUND

Recruitment of general practices and their patients into research studies is frequently reported as a challenge. The Preventing and Lessening Exacerbations of Asthma in School-aged children Associated with a New Term (PLEASANT) trial recruited 142 general practices, across England and Wales and delivered the study intervention to time and target.

AIMS

To describe the process of recruitment used within the cluster randomised PLEASANT trial and present results on factors that influenced recruitment.

METHODS

Data were collected on the number of and types of contact used to gain expression of interest and subsequent randomisation into the PLEASANT trial. Practice size and previous research experience were also collected.

RESULTS

The mean number of contacts required to gain expression of interest were m=3.01 (s.d. 1.6) and total number of contacts from initial invitation to randomisation m=6.8 (s.d. 3.5). Previous randomised controlled trial involvement (hazard ratio (HR)=1.81 (confidence interval (CI) 95%, 1.55-2.11) P<0.001) and number of studies a practice had previously engaged in (odds ratio (OR) 1.91 (CI 95%, (1.52-2.42)) P<0.001), significantly influenced whether a practice would participate in PLEASANT. Practice size was not a significant deciding factor (OR=1.04 (95% CI 0.99-1.08) P=0.137).

CONCLUSIONS

Recruitment to time and target can be achieved in general practice. The amount of resource required for site recruitment should not, however, be underestimated and multiple strategies for contacting practices should be considered. General practitioners with more research experience are more likely to participate in studies.

Design and preliminary recruitment results of the Cluster randomised triAl of PSA testing for Prostate cancer (CAP)

BACKGROUND

Screening for prostate cancer continues to generate controversy because of concerns about over-diagnosis and unnecessary treatment. We describe the rationale, design and recruitment of the Cluster randomised triAl of PSA testing for Prostate cancer (CAP) trial, a UK-wide cluster randomised controlled trial investigating the effectiveness and cost-effectiveness of prostate-specific antigen (PSA) testing.

METHODS

Seven hundred and eighty-five general practitioner (GP) practices in England and Wales were randomised to a population-based PSA testing or standard care and then approached for consent to participate. In the intervention arm, men aged 50-69 years were invited to undergo PSA testing, and those diagnosed with localised prostate cancer were invited into a treatment trial. Control arm practices undertook standard UK management. All men were flagged with the Health and Social Care Information Centre for deaths and cancer registrations. The primary outcome is prostate cancer mortality at a median 10-year-follow-up.

RESULTS

Among randomised practices, 271 (68%) in the intervention arm (198,114 men) and 302 (78%) in the control arm (221,929 men) consented to participate, meeting pre-specified power requirements. There was little evidence of differences between trial arms in measured baseline characteristics of the consenting GP practices (or men within those practices).

CONCLUSIONS

The CAP trial successfully met its recruitment targets and will make an important contribution to international understanding of PSA-based prostate cancer screening.

Asthma and chronic obstructive pulmonary disease guideline implementation: lessons learned on recruitment of primary care physicians to a knowledge translation study

BACKGROUND

Implementation of current clinical practice guidelines in asthma and chronic obstructive pulmonary disease (COPD) is suboptimal. New implementation strategies should be developed and evaluated.

METHODS

The authors report the rationale and planned methods of a project that sought to use a multifaceted knowledge translation intervention consisting of interactive education, mentorship through quality circles and practice-based tools in primary care to address key asthma and COPD care gaps. The present study was aborted due to inadequate primary care physician recruitment. Accordingly, the authors provide a critical review of their recruitment strategies and discuss alternative approaches and examples based on previous literature.

DISCUSSION

These practical lessons and discussion seek to inform researchers involved in designing and recruiting for future knowledge translation studies addressing chronic disease management in primary care.

Radical prostatectomy versus watchful waiting for prostate cancer

BACKGROUND

The lack of evidence regarding the effectiveness of treatment options for clinically localised prostate cancer continues to impact on clinical decision-making. Two such options are radical prostatectomy (RP) and watchful waiting (WW). WW involves providing no initial treatment and monitoring the patient with the intention of providing palliative treatment if there is evidence of disease progression.

OBJECTIVES

To compare the beneficial and harmful effects of RP versus WW for the treatment of localised prostate cancer.

SEARCH STRATEGY

MEDLINE, EMBASE, The Cochrane Library, ISI Science Citation Index, DARE and LILACS were searched through 30 July 2010.

SELECTION CRITERIA

Randomised or quasi-randomised controlled trials comparing the effects of RP versus WW for clinically localised prostate cancer.

DATA COLLECTION AND ANALYSIS

Data extraction and quality assessment were carried out independently by two authors.

MAIN RESULTS

Two trials met the inclusion criteria. Both trials commenced prior to the widespread availability of prostate-specific antigen (PSA) screening; hence the results may not be applicable to men with PSA-detected disease.One trial (N = 142), conducted in the US, was judged to be of poor quality. All cause (overall) mortality was not significantly different between RP and WW groups after fifteen years of follow up (Hazard Ratio (HR) 0.9 (95% Confidence Interval (CI) 0.56 to 1.43).The second trial (N = 695), conducted in Scandinavia, was judged to be of good quality. After 12 years of follow up, the trial results were compatible with a beneficial effect of RP on the risks of overall mortality, prostate cancer mortality and distant metastases compared with WW but the precise magnitude of the effect is uncertain as indicated by the width of the confidence intervals for all estimates (risk difference (RD) -7.1% (95% CI -14.7 to 0.5); RD -5.4% (95% CI -11.1 to 0.2); RD -6.7% (95% CI -13.2 to -0.2), respectively).        Compared to WW, RP increased the absolute risks of erectile dysfunction (RD 35% (95% CI 25 to 45)) and urinary leakage (RD 27% (95% CI 17 to 37)). These estimates must be interpreted cautiously as they are derived from data obtained from a self-administered questionnaire survey of a sample of the trial participants (N = 326), no baseline quality of life data were obtained and nerve-sparing surgery was not routinely performed on trial participants undergoing RP.

AUTHORS' CONCLUSIONS

The existing trials provide insufficient evidence to allow confident statements to be made about the relative beneficial and harmful effects of RP and WW for patients with localised prostate cancer. The results of ongoing trials should help to inform treatment decisions for men with screen-detected localised prostate cancer.