Stepped wedge designs could reduce the required sample size in cluster randomized trials

Woertman, Willem; Hoop, Esther de; Moerbeek, Mirjam; Zuidema, Sytse U.; Gerritsen, Debby L.; Teerenstra, Steven

doi:10.1016/j.jclinepi.2013.01.009

Cited by 261 publications

(329 citation statements)

References 18 publications

Supporting

Mentioning

319

Contrasting

Order By: Relevance

“…This approach can inform the choice between any two candidate designs: for example, that between a stepped‐wedge and parallel design 14, 16, 17, 18, 19, 20, 21, 22. From (4) the stepped‐wedge is the more efficient only if R > r 0 where the threshold r 0 satisfies

a_{S W} - b_{S W} r_{0} = a_{P D} - b_{P D} r_{0} .

…”

Section: The Precision Of the Effect Estimate Under A Linear Mixed Efmentioning

confidence: 99%

See 1 more Smart Citation

Statistical efficiency and optimal design for stepped cluster studies under linear mixed effects models

Girling

Hemming

2016

Statistics in Medicine

113

266

View full text Add to dashboard Cite

In stepped cluster designs the intervention is introduced into some (or all) clusters at different times and persists until the end of the study. Instances include traditional parallel cluster designs and the more recent stepped‐wedge designs. We consider the precision offered by such designs under mixed‐effects models with fixed time and random subject and cluster effects (including interactions with time), and explore the optimal choice of uptake times. The results apply both to cross‐sectional studies where new subjects are observed at each time‐point, and longitudinal studies with repeat observations on the same subjects.The efficiency of the design is expressed in terms of a ‘cluster‐mean correlation’ which carries information about the dependency‐structure of the data, and two design coefficients which reflect the pattern of uptake‐times. In cross‐sectional studies the cluster‐mean correlation combines information about the cluster‐size and the intra‐cluster correlation coefficient. A formula is given for the ‘design effect’ in both cross‐sectional and longitudinal studies.An algorithm for optimising the choice of uptake times is described and specific results obtained for the best balanced stepped designs. In large studies we show that the best design is a hybrid mixture of parallel and stepped‐wedge components, with the proportion of stepped wedge clusters equal to the cluster‐mean correlation. The impact of prior uncertainty in the cluster‐mean correlation is considered by simulation. Some specific hybrid designs are proposed for consideration when the cluster‐mean correlation cannot be reliably estimated, using a minimax principle to ensure acceptable performance across the whole range of unknown values. © 2016 The Authors. Statistics in Medicine published by John Wiley & Sons Ltd.

show abstract

a_{S W} - b_{S W} r_{0} = a_{P D} - b_{P D} r_{0} .

…”

Section: The Precision Of the Effect Estimate Under A Linear Mixed Efmentioning

confidence: 99%

“…Even without these additions, the Hussey and Hughes model has been found useful for the design of stepped studies 13, 14. It is directly applicable when the outcome is continuous, and approximately for binary outcomes when m is large 11.…”

Section: Introductionmentioning

confidence: 99%

Statistical efficiency and optimal design for stepped cluster studies under linear mixed effects models

Girling

Hemming

2016

Statistics in Medicine

113

266

View full text Add to dashboard Cite

show abstract

“…The care-as-usual group treated 53% of the patients of which 27% were outpatients, as compared to 47 and 38%, respectively, for the intervention group. The balance of a cluster-randomized trial can be improved by including crossovers in the study design, such as stepped wedge designs and cluster-randomized before-after studies, where each cluster has a time period with and without the intervention (Table 4, item 6) [49][50][51][52]. Each cluster can then also serve as its own control, enhancing the balance between study groups.…”

Section: Choosing a Study Designmentioning

confidence: 99%

Evaluating the impact of prediction models: lessons learned, challenges, and recommendations

et al. 2018

View full text Add to dashboard Cite

An important aim of clinical prediction models is to positively impact clinical decision making and subsequent patient outcomes. The impact on clinical decision making and patient outcome can be quantified in prospective comparative-ideally cluster-randomized-studies, known as 'impact studies'. However, such impact studies often require a lot of time and resources, especially when they are (cluster-)randomized studies. Before envisioning such large-scale randomized impact study, it is important to ensure a reasonable chance that the use of the prediction model by the targeted healthcare professionals and patients will indeed have a positive effect on both decision making and subsequent outcomes. We recently performed two differently designed, prospective impact studies on a clinical prediction model to be used in surgical patients. Both studies taught us new valuable lessons on several aspects of prediction model impact studies, and which considerations may guide researchers in their decision to conduct a prospective comparative impact study. We provide considerations on how to prepare a prediction model for implementation in practice, how to present the model predictions, and how to choose the proper design for a prediction model impact study.

show abstract

“…Rapid response systems are applied at a system level and individual patient randomization risks contamination between control and intervention participants in the same system. Priestly and colleagues used a stepped wedge design where each hospital ward was randomly allocated consecutive time points to implement the RRS such that at the end of the trial all wards had implemented the intervention [8,10]. This methodology is attractive when end users believe an intervention is beneficial because all participating sites end up with the intervention and only the timing of implementation varies.…”

mentioning

confidence: 99%

“…This methodology is attractive when end users believe an intervention is beneficial because all participating sites end up with the intervention and only the timing of implementation varies. Unfortunately, this study did not record measurements at each time interval so we cannot address the influence of time spent receiving the intervention on its effectiveness [10]. The study reported 7450 eligible patients, but only 2903 (39 %) were randomly allocated at a ward level [8].…”

mentioning

confidence: 99%

Rapid response teams improve outcomes: no

Maharaj

Stelfox

2016

Intensive Care Med

View full text Add to dashboard Cite

Rapid response systems (RRS) have been widely endorsed and adopted by hospital systems in industrialized countries to respond to deteriorating patients outside the intensive care unit (ICU) [1]. There is face validity in the idea that a system that promotes early recognition of deterioration and the implementation of time-critical interventions would save lives. Recent systematic reviews point to the potential effectiveness of RRS, but important caveats need to be considered [2][3][4].The vast majority of studies included in these analyses were before-after studies with no contemporaneous control (Table 1) [2,3]. Whilst this methodology is simple to implement and understand, the absence of a control group limits causal inference. The observed effects may be due to changes in case-mix, referral patterns, or overall improvements of ward-based care over time independent of an RRS. It is often unclear from the reported studies whether there were other interventions occurring at the same time.Bristow and colleagues performed a controlled beforeafter study comparing an intervention hospital to two control hospitals contemporaneously [5]. This methodology can account for secular trends, but does not completely account for baseline differences. Hospitals may experience changes not attributable to the intervention and not accounted for by regression techniques. The unadjusted analysis suggested benefit from implementation of the RRS, but the adjusted analysis yielded conflicting results [5]. After adjustment one control hospital performed worse while the other performed better than the RRS hospital. This suggests that the observed treatment effect of the RRS may be significantly attributable to observed and unobserved baseline imbalances.The interrupted time series (ITS) conducted by Howell and colleagues is a more robust methodology [6]. Their study of an RRS in an academic center found a reduction in unexpected mortality [patients without a do not attempt resuscitation (DNAR) order], but no reduction in overall hospital mortality [6]. The reasons for this observation are unclear. About 8 % of RRS activations led to a DNAR order, but it is unlikely that increased utilization of DNAR orders would completely account for the observed difference in expected and overall hospital mortality. The ITS design does not account for time-varying confounders and other unmeasured events unrelated to the study intervention. For example, the introduction of the RRS may have coincided with a general increased utilization of DNAR orders independent of the RRS. This study utilized the patient's usual care team and mandated communication with the patient's treating physician within 1 h of activation using explicit criteria [6]. This raises the possibility that the RRS acted as a substitute for timely senior clinical input from the usual care team [7].Priestley and Hillman performed experimental evaluations of RRS [8,9]. Rapid response systems are applied at a system level and individual patient randomization risks contamination between cont...

show abstract

Stepped wedge designs could reduce the required sample size in cluster randomized trials

Abstract: For CRTs, the stepped wedge design is far more efficient than the parallel group and ANCOVA design in terms of sample size.

Cited by 261 publications

References 18 publications

Statistical efficiency and optimal design for stepped cluster studies under linear mixed effects models

Statistical efficiency and optimal design for stepped cluster studies under linear mixed effects models

Evaluating the impact of prediction models: lessons learned, challenges, and recommendations

Rapid response teams improve outcomes: no

Contact Info

Product

Resources

About