Lessons learnt during a complex, multicentre cluster randomised controlled trial: the ProAct65+ trial
1 Research Department of Primary Care and Population Health, UCL Medical School, Royal Free Hospital, London, NW3 2PF, UK
2 School of Community Health Sciences, Division of Primary Care, Room 1309, Tower Building, University Park, University of Nottingham, Nottingham, NG7 2RD, UK
3 Nottingham University Hospitals NHS Trust and University of Derby, Hucknall Road, Nottingham, NG5 1PB, UK
4 School of Health & Life Sciences, Glasgow Caledonian University, A230 Govan Mbeki Building, Glasgow Caledonian University, Cowcaddens Road, Glasgow, G4 0BA, Scotland
Trials 2013, 14:192 doi:10.1186/1745-6215-14-192Published: 1 July 2013
Failure to recruit to target or schedule is common in randomized controlled trials (RCTs). Innovative interventions are not always fully developed before being tested, and maintenance of fidelity to the intervention during trials can be problematic. Missing data can compromise analyses, and inaccurate capture of risks to participants can influence reporting of intervention harms and benefits.
In this paper we describe how challenges of recruitment and retention of participants, standardisation and quality control of interventions and capture of adverse events were overcome in the ProAct65+ cluster RCT. This trial compared class-based and home-based exercise with usual care in people aged 65 years and over, recruited through general practice. The home-based exercise participants were supported by Peer Mentors.
(1) Organisational factors, including room availability in general practices, slowed participant recruitment so the recruitment period was extended and the number invited to participate increased. (2) Telephone pre-screening was introduced to exclude potential participants who were already very active and those who were frequent fallers. (3) Recruitment of volunteer peer mentors was difficult and time consuming and their acceptable case load less than expected. Lowering the age limit for peer mentors and reducing their contact schedule with participants did not improve recruitment. (4) Fidelity to the group intervention was optimised by introducing quality assurance observation of classes by experienced exercise instructors. (5) Diaries were used to capture data on falls, service use and other exercise-related costs, but completion was variable so their frequency was reduced. (6) Classification of adverse events differed between research sites so all events were assessed by both sites and discrepancies discussed.
Recruitment rates for trials in general practice may be limited by organisational factors and longer recruitment periods should be allowed for. Exercise studies may be attractive to those who least need them; additional screening measures can be employed to avoid assessment of ineligible participants. Enrolment of peer mentors for intervention support is challenging and needs to be separately tested for feasibility. Standardisation of exercise interventions is problematic when exercise programmes are tailored to participants’ capabilities; quality assurance observations may assure fidelity of the intervention. Data collection by diaries can be burdensome to participants, resulting in variable and incomplete data capture; compromises in completion frequency may reduce missing data. Risk assessments are essential in exercise promotion studies, but categorisation of risks can vary between assessors; methods for their standardisation can be developed.