Table 1 Characteristics of included studies which adopted a stepped wedge randomised controlled trial design
First author | Study start date (publication date) | Study duration | Country | Intervention | Primary outcome | Cluster definition | Why did investigators choose stepped wedge trial design? |
|---|---|---|---|---|---|---|---|
Presentation of trial results - research articles | |||||||
Bacchieri et al. [24] | 2006 (2010) | 20 months | Brazil | Education intervention to prevent traffic accidents among cyclists | Traffic accidents and near accidents | 40 sectors within 5 neighbourhoods | Ethical - no equipoise; phased implementation - cannot implement in many clusters at same time |
Bashour et al. [25] | 2008 (2013) | 10 months | Syria | Training resident doctors in interpersonal and communication skills | Women’s satisfaction with interpersonal and/or communication skills of doctors working in labour and delivery rooms | 4 teaching public maternity hospitals | Ethical and practical |
Durovni et al. [26] | 2005 (2013) | 42 months | Brazil | Implementation of widespread isoniazid preventive therapy for HIV-positive patients | Incidence of active tuberculosis | 29 HIV clinics | Ethical - no equipoise (intervention recommended, but not implemented); phased implementation - cannot implement in many clusters at same time |
Fuller et al. [27] | 2006 (2012) | 38 months | United Kingdom | Feedback intervention to improve hand hygiene compliance in UK healthcare workers | Hand hygiene compliance measured by observers blinded to the hospital allocation | 16 hospitals | Ethical - no equipoise; phased implementation - cannot implement in many clusters at same time; prevent contamination and disappointment effects in control hospitals; clusters act as own controls so higher statistical power; extended duration allows assessment of sustainability |
Gruber et al. [28] | 2009 (2013) | 15 months | Mexico | Ultraviolet-disinfection system designed to treat household drinking water. | Proportion of households with contaminated drinking water and 7-day prevalence of diarrhoea (co-primary) | 24 rural communities | Phased implementation - cannot implement in many clusters at same time |
Horner et al. [29] | 2006 (2012) | 28 months | United Kingdom | Staff training and education on the topic of infection prevention and effective hand hygiene | Prevalence of MRSA infection | 65 care homes | Allow measurement of prevalence before the intervention, directly after the intervention and further follow-up in two of the three study groups; participating residents and staff in each group of homes acted as controls for each other |
Mhurchu et al. [30] | 2010 (2013) | 11 months | New Zealand | Free daily before-school breakfast programme | The proportion of students achieving a school attendance of 95 % or higher | 16 schools | None given |
2011 (2013/2014) | 12 months | Australia | A multifaceted intervention incorporating a malnutrition screening tool, nutritional supplements and red trays | Rate of change in body mass index over weekly periods from admission to discharge | 25 hospital wards | Political - intervention is to be rolled out to all clusters eventually; ethical - no equipoise; phased implementation - cannot implement in many clusters at same time; improvements can be made to the intervention; temporal changes in effectiveness can be modelled; clusters act as own controls so higher statistical power | |
Roy et al. [33] | 2009 (2013) | 7 months | United Kingdom | Universal offer of testing without detailed pre-test discussion; training of clinic staff; and the provision of tailor-made information material for patients and healthcare workers | HIV test acceptance amongst those offered a test | 24 tuberculosis clinics | Political - intervention to be rolled out to all clusters eventually |
Stern et al. [34] | 2010 (2014) | 17 Months | Canada | Educating staff on the prevention and treatment of pressure ulcers; use of Enhanced Multi-Disciplinary Team (EMDT) | Rate of reduction in pressure ulcer surface area | 12 long-term care facilities | Desire to have benefits of randomization; ethical - no equipoise; phased implementation - cannot implement in many clusters at same time |
Conference abstracts | |||||||
Fearon et al. [35] | 2013 (2013) | 15 months | United Kingdom | Telephone hotline to link GPs directly with stroke patients’ specialists for: immediate discussion, treatment advice, prioritisation of investigations | Reduction in the time from referral to specialist stroke team input | 72 GP practices | None given |
Trial protocol/design articles | |||||||
Bennett et al. [36] | 2013 (2013) | 12 months | Australia | Accredited exercise physiologist coordinated program on physical function | Objective physical function measured using the 30-second sit to stand test. | 15 haemodialysis clinics | Ethical - no equipoise; phased implementation - cannot implement in many clusters at same time |
Bernabe-Ortiz et al. [37] | 2012 (2014) | 7 months | Peru | Population-level social marketing campaign to introduce a low-sodium, high potassium salt substitute | Blood pressure and use of salt | 6 villages | Phased implementation - cannot implement in many clusters at same time |
Brimblecombe et al. [38] | 2012 (2013) | 12 months | Australia | Price intervention: 20 % discount on food in store. Combined intervention: price discount and in-store nutrition education strategy | Per capita daily weight of combined fruit and vegetables purchased through the community store. | 20 communities | Phased implementation - cannot implement in many clusters at same time |
Dainty et al. [39] | 2010 (2011) | 24 months | Canada | Multi-faceted knowledge translation strategy designed to increase the utilisation rate of induced hypothermia in survivors of cardiac arrest | Proportion of survivors of cardiac arrest presenting to the emergency department that achieve the target temperature within six hours of ED arrival. | 37 hospitals | Ethical - no equipoise (intervention recommended, but not implemented); phased implementation - cannot implement in many clusters at same time; temporal changes in effectiveness can be modelled |
Dreischulte et al. [40] | 2011 (2012) | 96 months | United Kingdom | Data-Driven Quality Improvement in Primary Care (DQIP) with three components: education, informatics and financial incentive | Composite score of prescribing outcomes | 40 GP practices | Phased implementation - cannot implement in many clusters at same time; prevents control clusters dropping out; higher statistical power |
Gerritsen et al. [41] | 2009 (2011) | 24 months | Netherlands | Act In Case of Depression: multidisciplinary care program to improve the management of depression in nursing home residents | Frequency of depression and quality of life | 32 somatic and dementia special care units | Higher statistical power; all clusters receive the intervention - expected to increase motivation of clusters to participate in the study |
Gucciardi et al. [42] | 2012 (2012) | 24 months | Canada | Mobile diabetes education team (MDET) intervention to support primary care providers by offering a diabetes education team | Change in HbA1c (an index of diabetes control) | 12 primary care sites | All participating physicians want the intervention; all clusters receive intervention - gives additional data on effectiveness |
Keriel-Gascou et al. [43] | 2012 (2013) | 18 months | France | Interactive program that encouraged patients to report adverse drug events in primary care | Reporting of adverse drug events by antihypertensive-treated patients to their GPs | 8 clusters of GP practices | Ethical - no equipoise; phased implementation - cannot implement in many clusters at same time; clusters act as own controls so higher statistical power; temporal changes in effectiveness can be modelled |
Kjeken et al. [44] | 2011 (2014) | 10 months | Norway | New rehabilitation program PRAISE versus current rehabilitation program | Goal attainment and health-related quality of life | 6 rehabilitation centres | Ethical - no equipoise; phased implementation - cannot implement in many clusters at same time |
Marshall et al. [45] | 2012 (2012) | 18 months in first area and 12 months in second area | United Kingdom | Targeted case finding of patients at high risk of CVD versus opportunistic assessment | Number of high-risk patients started on at least one preventive treatment: an antihypertensive drug or a statin | 32 GP practices in two areas | Phased implementation - cannot implement in many clusters at same time; evaluate effects of the case finding programme before and after implementation of intervention |
Mouchoux et al. [46] | 2011 (2011) | 24 months | France | Multifaceted prevention program involving structured geriatric consultation, training sessions and practice analysing medical records | Post-operative delirium rate within 7 days after surgery | Surgical wards within 3 districts | Ethical - no equipoise; phased implementation - cannot implement in many clusters at same time; clusters act as own controls so higher statistical power; temporal changes in effectiveness can be modelled |
Poldervaart et al. [47] | 2013 (2013) | 14 months | Netherlands | Use of the HEART score, a clinical prediction rule, to provide a simple, early and reliable predictor of cardiac risk | Occurrence of major adverse cardiac events | 10 hospitals | Within-hospital comparison less confounded by case-mix differences than between hospitals; all hospitals receive intervention so provide data about implementation problems; gradual intervention implementation provides data about the process; all clusters receive the intervention - expected to increase motivation of clusters to participate in the study |
Praveen et al. [48] | 2013 (2013) | 24 months | India | Clinical decision support system to assist health workers in making decisions to lower patients’ cardiovascular disease (CVD) risks | Difference in proportion of high risk individuals (with or without CVD) who are achieving optimal blood pressure levels (systolic <140 mmHg) | 18 primary health care centres | Ensure all receive intervention |
Rasmussen et al. [49] | 2013 (2013) | 15 months | Denmark | Multifaceted worksite intervention consisting of participatory economic, physical exercise and cognitive behavioural training for lower back pain. | Lower back pain is measured by days with and intensity of pain each month throughout the data collection period | 21 clusters each consisting of one team, unless small teams in similar location | Phased implementation - cannot implement in many clusters at same time; all clusters receive the intervention - expected to increase motivation of clusters to participate in the study |
Ratanawongsa et al. [50] | 2009 (2012) | 24 months | USA | Automated Telephone Support Management intervention to promote care manager efficiency | Physical and mental functional status and the number of days spent in bed due to illness | 8 clusters of participants | Ethical - no equipoise; phased implementation - cannot implement in many clusters at same time; temporal changes in effectiveness can be modelled |
Solomon et al. [51] | 2011 (2012) | 23 months | United Kingdom | Devon Active Villages intervention to improve participation in physical activity | Proportion of adults meeting recommended daily guidelines for the minimum level of physical activity | 128 villages | Ethical - no equipoise; phased implementation - cannot implement in many clusters at same time |
Stringer et al. [52] | 2011 (2013) | 48 months | Zambia | Implementation of clinical protocols, forms and systems by Quality Improvement (QI) teams; engagement of community health workers. | Community level all-cause mortality among those aged <60 years | 42 primary healthcare facilities and their catchment areas | Ethical - no equipoise; phased implementation - cannot implement in many clusters at same time |
Tirlea et al. [53] | 2001 (2013) | 9 months | Australia | Girls on the Go! Program aimed at increasing self-esteem and self-efficacy | The Rosenberg Self-Esteem Scale and the Eating Disorders Assessment | 12 schools | None given |
Turner et al. [54] | 2011 (2011) | 4 months | Australia | Brief tailored psychosocial intervention in cancer care with focused training and clinical supervision | Change in depression as measured by Hospital Anxiety and Depression Score | 5 hospitals | Able to account for systematic differences between sites and times during the trial, and also for case-mix differences between patients |
Van de Steeg et al. [55] | 2011 (2012) | 11 months | Netherlands | E-learning course about delirium aimed at nursing staff. | Percentage of patients screened for risk of delirium; sample size based on screening for delirium risk and the effect on knowledge | 18 hospitals | Ethical - no equipoise; all clusters receive the intervention - expected to increase motivation of clusters to participate in the study; reduce contamination bias as each hospital acts as their own control; take into account the effect of time on outcomes measures |
van Holland et al. [56] | 2012 (2012) | 32 months | Netherlands | Employees offered health surveillance programs to reduce sickness absence | Work ability, productivity and absenteeism | 5 meat processing companies | Clusters act as own controls so higher statistical power and fewer confounding factors |
Trial registrations | |||||||
Craine [57] | 2012 (2011) | 12 months | United Kingdom | Dried blood spit testing (DBS) for blood borne viral infections versus standard venepuncture-based testing | Change in blood-borne viral diagnostic testing rate in prisons with introduction of DBS | 5 prisons | None given |
Everingham [58] | 2014 (2014) | 21 months | United Kingdom | Quality improvement project to help staff deliver highest standard of care for emergency laparotomy patients | All-cause mortality at 90 days following surgery | 90 hospitals | Control adoption bias; adjust for time-based changes in the background level; can offer to every site |
Grande [59] | 2014 (2012) | 24 months | United Kingdom | Formalised, comprehensive procedure for carer support needs assessment, prioritisation and follow-up | Quality of life | 6 hospice home care services | None given |
Koeberlein-Neu [60] | 2013 (2013) | 17 months | Germany | An inter-professional medication therapy management | Change in the Medication Appropriateness Index Scores measured every three months | 14 GP surgeries | Phased implementation - cannot implement in many clusters at same time |
Williams [62] | 2012 (2012) | 24 months | United Kingdom | Physiotherapists trained in clinical reasoning skills via a clinical mentoring program | Function measured by The Patient Specific Functional Scale. | 12 physiotherapists | None given |