Detecting selection bias in randomized clinical trials.

Lack of concealment of allocation in randomized clinical trials can invite selection bias, which is the preferential enrollment of specific patients into one treatment group over another. For example, patients more likely to respond may be enrolled only when the next treatment to be assigned is known to be the active treatment, and patients less likely to respond may be enrolled only when the next treatment to be assigned is known to be the control. Despite the fact that selection bias can compromise both the internal and external validity of trials, little methodology has been developed for its detection. An investigator may test the success of the randomization by comparing baseline characteristics across treatment groups, but such test is limited by the potential inability of the measured baseline variables to predict response. A new method for detecting selections bias, based on response data only, is developed for the case in which a small block size, and either unmasking of treatment codes or an open-label design, have compromised the concealment of allocation. This new method complements baseline comparisons, and is sensitive to detect selection bias even in situations in which baseline comparisons are not.     

A survey of the views of palliative care healthcare professionals towards referring cancer patients to participate in randomized controlled trials in palliative care

Goals of work  Clinical trials in palliative care (PC), especially randomised controlled trials (RCTs), are notoriously difficult to complete. One perceived challenge is gatekeeping, the reluctance of some healthcare professionals (HCPs) to refer patients for research studies. This study aimed to identify the extent of gatekeeping from palliative RCTs. Materials and methods  An anonymous questionnaire was sent to 597 HCPs with an interest in PC in Australia and New Zealand to assess their willingness to refer patients for RCTs. Respondents considered key issues that might affect their decision, documented willingness to refer to RCTs of increasing complexity in a hypothetical pain situation and documented the degree of patient inconvenience considered acceptable. Demographic data were collected. Main results  One hundred ninety-eight questionnaires were returned (33%), 122 from doctors and 76 from other HCPs. Very few were willing to refer to complicated studies involving many extra tests and/or hospital visits. Non-medical HCPs were less interested than doctors in studies that involved randomisation, placebo controls or double-blind methodology. The majority would refer patients for non-pharmacological studies, but were less willing to refer for pharmacological studies with possible side effects. Non-medical HCPs were less willing than doctors to refer to trials that involved patient inconvenience. Two factors predicted for greater willingness to refer: previous research experience and male gender. Conclusion  The survey revealed an unwillingness on the part of many HCPs to refer patients for RCTs in PC. It identifies trial-related factors that may encourage or discourage referral. Gatekeeping has the potential block recruitment and introduce a selection bias.     

A recruitment strategy for cluster randomized trials in secondary care settings

Trials of educational or organizational interventions to change clinical practice require cluster randomization, that is, randomization of units such as hospitals or clinical teams rather than individual patients. Cluster randomization is relatively novel in health care settings and raises new methodological challenges, in particular: are units willing to be randomized at an organizational level; and, what procedures should be followed to successfully enrol all of the clinicians in a unit rather than individual clinicians as in conventional multicentre trials. This is particularly problematic for trials of large units such as hospitals. The aim of this study was to develop and partially evaluate a strategy to recruit acute, secondary care NHS hospitals in the UK into cluster randomized trials. Literature search and interviews with senior staff in acute hospitals and relevant national organizations were used to develop a recruitment strategy. The strategy was evaluated by inviting 32 randomly selected clinical directorates to participate in a trial feasibility study. A seven step recruitment strategy was developed: (1) Identify stakeholders and gatekeepers; (2) Inform stakeholders and gatekeepers; (3) Approach gatekeepers; (4) Local negotiation; (5) Conduct the research; (6) Feedback to gatekeepers; (7) Feedback to stakeholders. Key problems were the possibility of multiple gatekeepers and identification of all possible stakeholders in varying organizational structures. The strategy was effective in two respects. First, 32 (100%) of the directorates approached agreed to participate. Second, baseline data collection was successfully achieved in all of the directorates. However, the strategy is costly in terms of time and resources. We conclude that NHS trusts are willing to participate in cluster randomized trials. This recruitment strategy is successful and could be widely adopted, but realistic time and financial cost estimates are required at the planning stage.     

Delta inflation: a bias in the design of randomized controlled trials in critical care medicine

Introduction  

Mortality is the most widely accepted outcome measure in randomized controlled trials of therapies for critically ill adults, but most of these trials fail to show a statistically significant mortality benefit. The reasons for this are unknown.     

Impact of baseline covariate imbalance on bias in treatment effect estimation in cluster randomized trials: Race as an example

Individual-level baseline covariate imbalance could happen more frequently in cluster randomized trials, and may influence the observed treatment effect. Using computer and real-data simulations, this paper quantifies the extent and impact of covariate imbalance on the estimated treatment effect for both continuous and binary outcomes, and relates it to the degree of imbalance for different numbers of clusters, cluster sizes, and covariate intraclass correlation coefficients. We focused on the impact of race as a covariate, given the emphasis of regulatory and funding bodies on understanding the influence of demographic characteristics on treatment effectiveness. We found that bias in the treatment effect is proportional to both the degree of baseline covariate imbalance and the covariate effect size. Larger numbers of clusters result in lower covariate imbalance, and increasing cluster size is less effective in reducing imbalance compared to increasing the number of clusters. Models adjusted for important baseline confounders are superior to unadjusted models for minimizing bias in both model-based simulations and an innovative simulation based on real clinical trial data. Higher outcome intraclass correlation coefficients did not affect bias but resulted in greater variance in treatment estimates.     

Bias in recruitment to cluster randomized trials: a review of recent publications

Objectives To assess recruitment bias and the techniques employed to counter this problem in a recent selection of published cluster randomized trials. Design Review of 24 cluster trials published in 2008 in four leading medical journals. Data extraction Studies were assessed by four reviewers to identify if an alternative design could have been employed using individual randomization. Data were also extracted on the randomization procedure and the likelihood of this introducing bias to the selection of participants into the study. Results Of the 24 trials, eight could have used individual randomization as an alternative to cluster allocation. Seven studies could have recruited participants prior to cluster randomization but did not. In eight studies where recruitment bias was possible, more than half (five) demonstrated some evidence of differential recruitment rates. Conclusions Many cluster trials published in leading medical journals are not clear in their justification for the design. We also found significant proportions of cluster trials used suboptimal designs that increase their risk of introducing selection bias. Better design of cluster trials is possible and should be adopted.     

Design and analysis of stepped wedge cluster randomized trials

Cluster randomized trials (CRT) are often used to evaluate therapies or interventions in situations where individual randomization is not possible or not desirable for logistic, financial or ethical reasons. While a significant and rapidly growing body of literature exists on CRTs utilizing a "parallel" design (i.e. I clusters randomized to each treatment), only a few examples of CRTs using crossover designs have been described. In this article we discuss the design and analysis of a particular type of crossover CRT - the stepped wedge - and provide an example of its use.     

Randomized controlled trials in palliative care: overcoming the obstacles

Problems associated with conducting randomized controlled trials (RCTs) in palliative care are well documented. Palliative care researchers commonly report difficulties with recruitment, attrition, and randomization. Despite these difficulties, there is a crucial need to enhance the evidence base of palliative care and RCTs are a recognized method for achieving this. Few publications to date have provided clear strategies for researchers considering RCTs in palliative care. This article presents an overview of the features of the RCT and reasons for using this research method. Problems with conducting RCTs are considered with discussion focusing on an approach for managing these hurdles. Examples are provided from an ongoing RCT investigating a support and information programme for lay caregivers of people receiving palliative care at home. The aim of the article is to provide a guide to conducting RCTs and to promote their use in palliative care when researchers are evaluating new interventions.     

Adjustment for baseline measurement error in randomized controlled trials induces bias

When estimating the treatment effect in a randomized controlled trial, it is common to have a continuous outcome which is also observed at baseline. These observations are often prone to measurement error, for example due to within-patient variability. Controversy exists in the literature about whether baseline measurement error should be adjusted for in this context. Computer simulations were used to compare the biases in the estimated treatment effect, with and without adjusting for measurement error, and for different levels of observed baseline imbalance. The impacts of sample size (30 per group and 300 per group) and reliability coefficient (0.6, 0.8 and 1) were also assessed. The results show that in randomized controlled trials, the ordinary least squares (OLS) estimator without adjusting for measurement error is unbiased. On the contrary, adjusting for measurement error leads to bias, especially when sample sizes are small and/or measurement error is large. The treatment effect adjusting for measurement error is on average overestimated when the baseline mean of the control group is larger than that of the treated group. It is underestimated when the control group has a smaller baseline mean.     

Control group selection in critical care randomized controlled trials evaluating interventional strategies: An ethical assessment

BACKGROUND: Ethical concern has been raised with critical care randomized controlled trials in which the standard of care reflects a broad range of clinical practices. Commentators have argued that trials without an unrestricted control group, in which standard practices are implemented at the discretion of the attending physician, lack the ability to redefine the standard of care and might expose subjects to excessive harms due to an inability to stop early. OBJECTIVE: To develop a framework for analyzing control group selection for critical care trials. METHOD: Ethical analysis. RESULTS: A key ethical variable in trial design is the extent with which the control group adequately reflects standard care practices. Such a control group might incorporate either the "unrestricted" practices of physicians or a protocol that specifies and restricts the parameters of standard practices. Control group selection should be determined with respect to the following ethical objectives of trial design: 1) clinical value, 2) scientific validity, 3) efficiency and feasibility, and 4) protection of human subjects. Because these objectives may conflict, control group selection will involve trade-offs and compromises. Trials using a protocolized rather than an unrestricted standard care control group will likely have enhanced validity. However, if the protocolized control group lacks representativeness to standard care practices, then trials that use such groups will offer less clinical value and could provide less assurance of protecting subjects compared with trials that use unrestricted control groups. For trials evaluating contrasting strategies that do not adequately represent standard practices, use of a third group that is more representative of standard practices will enhance clinical value and increase the ability to stop early if needed to protect subjects. These advantages might come at the expense of efficiency and feasibility. CONCLUSION: Weighing and balancing the competing ethical objectives of trial design should be done for each trial.     

Improving the effectiveness of interventions in palliative care: the potential role of qualitative research in enhancing evidence from randomized controlled trials

Evaluating interventions in palliative care using randomized controlled trials (RCTs) has helped advance the specialty and create an evidence base for the delivery of care. RCTs, however, are notoriously difficult to conduct in palliative care, raising a variety of practical, ethical and moral dilemmas. Mixed-methods research, which combines qualitative research and RCTs, offers a potential solution to these problems. This paper begins by examining the theoretical basis for combining the two approaches, before reviewing the specific role qualitative research could play in planning, conducting and implementing trials. The paper then goes on to explore how palliative care research currently uses the mixed-methods approach, by searching the trials included in six Cochrane Systematic Reviews (n = 146) on the incorporation of qualitative research. Only one trial undertook qualitative research. These findings reflect some of the challenges facing mixed-methods research, which include lack of experience in a research team, the problems of obtaining funding and difficulties in publishing. The paper concludes that while combining qualitative and quantitative research is not a panacea for methodological problems in palliative care research, with careful planning and integration, the approach may enhance the clinical and ethical utility of trial findings, which in turn will improve patient care.     

Truth may hurt but deceit hurts more: communication in palliative care

Healthcare professionals often censor their information giving to patients in an attempt to protect them from potentially hurtful, sad or bad news. There is a commonly expressed belief that what people do not know does not harm them. Analysis of doctor and nurse/patient interactions reveals that this well-intentioned but misguided assumption about human behaviour is present at all stages of cancer care. Less than honest disclosure is seen from the moment that a patient reports symptoms, to the confirmation of diagnosis, during discussions about the therapeutic benefits of treatment, at relapse and terminal illness. This desire to shield patients from the reality of their situation usually creates even greater difficulties for patients, their relatives and friends and other members of the healthcare team. Although the motivation behind economy with the truth is often well meant, a conspiracy of silence usually results in a heightened state of fear, anxiety and confusion--not one of calm and equanimity. Ambiguous or deliberately misleading information may afford short-term benefits while things continue to go well, but denies individuals and their families opportunities to reorganize and adapt their lives towards the attainment of more achievable goals, realistic hopes and aspirations. In this paper, some examples and consequences of accidental, deliberate, if well-meaning, attempts to disguise the truth from patients, taken verbatim from interviews, are given, together with cases of unintentional deception or misunderstandings created by the use of ambiguous language. We also provide evidence from research studies showing that although truth hurts, deceit may well hurt more. 'I think the best physician is the one who has the providence to tell to the patients according to his knowledge the present situation, what has happened before, and what is going to happen in the future' (Hippocrates).