A cohort examination to establish reporting of the remit and function of Trial Steering Committees in randomised controlled trials

Background The DAMOCLES project established a widely used Data Monitoring Committee (DMC) Charter for randomised controlled trials (RCTs). Typically, within the UK, the DMC is advisory and recommends to another executive body; the Trial Steering Committee (TSC). Despite the executive role of the TSC, the CONSORT Statement does not explicitly require reporting of TSC activity, although is included as an example of good reporting. A lack of guidance on TSC reporting can impact transparency of trial oversight, ultimately leading to a misunderstanding regarding role and, subsequently, further variation in practice. This review aimed to establish reporting practice of TSC involvement in RCTs, and thus make recommendations for reporting. Methods A cohort examination identifying reporting practice was undertaken. The cohort comprised RCTs published in three leading medical journals (the British Medical Journal, The Lancet and the New England Journal of Medicine) within 6 months in 2012 and the full NIHR HTA Monograph series. Details of TSC constitution and impact were extracted from main publications and published supplements. Results Of 415 publications, 264 were eligible. These were typical in terms of trial design. Variations in reporting between journals and monographs was notable. TSC presence was identified in approximately half of trials (n = 144), of which 109 worked alongside a DMC. No publications justified not convening a TSC. When reported, the role of the committee and examples of impact in design, conduct and analysis were summarised. Conclusions We present the first review of reporting TSC activity in the published academic literature. An absence of reporting standards with regards to TSC constitution, activity and impact on trial conduct was identified which can influence transparency of reporting trial oversight. Consistent reporting is vital for the benefits and impact of the TSC role to be understood to support adoption of this oversight structure and reduce global variations in practice. Electronic supplementary material The online version of this article (doi:10.1186/s13063-017-2300-1) contains supplementary material, which is available to authorized users.


Background
The DAMOCLES project [1] established a Data Monitoring Committee Charter [2] which has been widely used for randomised controlled trials since 2005. As established within DAMOCLES, the Data Monitoring Committee (DMC) is typically advisory and makes recommendations to another executive body; considered the Trial Steering Committee (TSC). Currently however, no evidence-based Charter exists for TSCs to establish their role and functionality in RCTs.
The MRC Guidelines for Good Clinical Practice (1998) defined a three-committee oversight structure: the day-to-day Trial Management Group, the DMC and the executive TSC [3]. This document provided guidance and a suggested terms of reference for TSCs which has been widely adopted within the UK [4]. However, a need for redevelopment and expansion of these guidelines has been identified [4,5]. This may have led to these guidelines recently being withdrawn, and subsequently revised, with limited changes mainly concerning responsibilities to funder. The need for the development of expanded universal guidelines remains [3,6].
Both DAMOCLES [1] and the Consolidated Standards of Reporting Trials (CONSORT) Statement [7], an evidence-based minimum set of recommendations for reporting randomised trials, recommend that trial reports should include information about interim analyses and the data monitoring process. The purpose of CONSORT is to facilitate the complete and transparent reporting of trials and, as such, has been widely adopted by journals to aid critical appraisal and interpretation. However, there is an absence of content focussing on the clinical trial oversight structure and responsibilities for decision-making. While reporting of DMC activity is included within CONSORT as part of reporting interim analyses, and DAMOCLES suggests reporting DMC membership, the reporting of TSC activities is not covered.
The objective of this cohort examination was to establish current practice of reporting of TSC involvement in RCTs, to recommend reporting standards for TSC activity and to identify impact on trial conduct.

Search strategy
EJC searched publications within a 6-month period (1 July 2012 to 31 December 2012) from four sources. Three top general medical journals (the British Medical Journal (BMJ), The Lancet (Lancet) and the New England Journal of Medicine (NEJM)) and within the full UK National Institute for Health Research Health Technology Assessment (NIHR HTA) Monograph series. Journals were selected that are known for endorsing high standards of reporting when publishing RCTs. The NIHR HTA Monograph series (HTA), a peer-reviewed, open-access journal that publishes full details of a single study funded by the NIHR HTA funding stream. NIHR HTA is a major UK funding body which supports policy-makers such as the National Institute for Health and Care Excellence, the National Screening Committee and the Department of Health [8]. The full NIHR HTA Monograph series was searched as opposed to those published within the set timeframe because this series has a suggested word count of 50,000 and so enables more details of the work to be included when compared to a typical peer-reviewed journal and so were considered more likely to provide a comprehensive description of TSC remit and function. Therefore, when summarising examples of reporting, results from journals and monographs are reported separately.
Published RCTs were identified by searching of titles, abstracts and keywords of primary research papers published within the timeframe using the search term random*. When eligibility was unclear a second reviewer (CG) was consulted.

Inclusion criteria
We included all RCTs publishing main trial results. Articles presenting results of: secondary analyses; preliminary analyses; and additional reports of published RCTs; for example, results of long-term follow-up, were excluded.

Data extraction and analysis
A Data Extraction Form was designed and piloted by EJC and CG. Data was extracted from all published materials (main trial report and supplementary material when applicable). EJC extracted data on trial design, trial stopping and oversight committee reporting (see Table 1) and entered into an MS Access database. BA independently extracted data from a random 10%, stratified by TSC reporting and source.
Quantitative items were analysed using descriptive statistics. Standard statistical software was used throughout (Statistical Analysis Software (SAS 9.1.3; SAS Institute Inc., Cary, NC, USA)). Text extracts from articles were examined using NVivo qualitative data analysis software (QSR International Pty Ltd. Version 10, 2012). EJC and CG identified themes within text items which were used to contextualise and illuminate quantitative results. Extracts are denoted by (…) and (words) denoted the addition of words or replaced words to aid understanding. Each paper was mapped to a unique project identification number; details of this mapping are given in Additional file 1: Table S1. Due to differences in focus of paper between journal manuscripts and the monograph series, results are presented split by source and, where applicable, split into sections appropriate for the content.

Variations in published material
Due to variations in publishing requirements, differences were anticipated between monographs and journals. However, there were clear differences between sources, Key journals were restricted by word count though supplements were often published. While all three journals encouraged protocol and supplementary appendices, this appeared to be endorsed by the NEJM only. Further details are provided in Table 3. ' A data monitoring committee for efficacy is not required for this study. Data safety monitoring will be conducted on an ongoing basis as detailed in the Safety Review Plan'. (NEJM44) HTA publications justifying not having a DMC (n = 3) gave reasons: the trial examined routine therapies (HTA67); no interim analyses were planned (HTA83)  Of the 120 trials not explicitly reporting use of a TSC, none justified its absence.

Trial oversight committees
Aside from trials with cellular or gene therapy interventions of which there were few, TSCs were consistently reported regardless of intervention type, from 48% of psychological or behavioural intervention trials (14/ 29) to 79% physical intervention trials (11/14). DMCs were reported less frequently, from 23% in resources and infrastructure trials (5/22) to 73% in pharmaceutical intervention trials (97/132) or medical device trials (19/ 26). Trials with neither committee most commonly involved a complimentary intervention (3/6, 50%) or psychological and behavioural (13/29, 45%). Further details are provided in Tables 5 and 6.

Membership
Four fifths indicated the number of committee members, varying from half (BMJ, 3/6) to 85% (HTA, 71/84). The number of members ranged from 2 to 52 with a median of 7. Details are provided in Table 8.
One hundred and ten publications listed members of which 84 specified a Chair. Other roles or fields of expertise were detailed in almost half of trials with a TSC (66/144). Common representation were clinical (n = 46), statistical (n = 29) and patient and/or public (PPI) (n = 39).
Reporting was unclear with regards to membership and independence.
Journal manuscripts One publication specified that the funder (NIHR) appointed the TSC, stating that members were 'researchers independent of the study funders, although several have served on their advisory or funding committees'. (BMJ12) One publication indicated voting members when listing members (Lancet25).
Ten NEJM publications gave details of TSC membership beyond listing members and affiliations. Half of these described sponsor representation, with one specifying this representation as none voting (NEJM20).
Thirteen NEJM protocols discussed committee membership. Protocol reporting consisted listing members by role and/or field of expertise (n = 10) or by name (n = 6), of which three reported both.
Monographs Forty-six gave details of TSC membership beyond listing members and affiliations.
Independence was discussed in 32 publications (32/ 46, 69.6%), of which 30 specified an independent Chair. Two reported funder input in selecting the Chair.

Meetings
Journal manuscripts One publication discussed meeting frequency (bimonthly meetings (BMJ15)). One stated yearly reports were circulated (Lancet3) possibly indicating yearly meetings.
Seven supplementary protocols provided TSC meeting information. Four gave the frequency of these, ranging from monthly to annually. Two were unclear stating meetings were held periodically. Others reported that meetings held by teleconference (n = 2) and gave insight to report contents (n = 1).
Monographs Twenty-five publications (25/144) discussed meeting frequency or timing. Timings, when specified, were biyearly (n = 12); yearly (n = 7); quarterly (n = 2) and as required (n = 1). Three gave meeting dates and another three gave the total number of meetings. One specified the length of TSC meetings as 80-100 min. Four indicated timing of Other units of randomisation: Clinic (n = 1); Family (n = 2); Hospital (n = 1); Household (n = 1); Partner (n = 1); School (n = 1); Village (n = 1); Year group (n = 1); Paediatric diabetes services (n = 1) i Justification for no blinding: Not possible or practical due to nature of intervention or trial design (n = 30); Not possible/practical as in practice caused difficulties for patients (n = 1); Not possible/practical as shown by other similar trials (n = 2); Not possible/practical so cluster randomisation approach used (n = 1); Not possibleno additional justification given (n = 4); Attempted to blind although were not successful (n = 1); Large sample size means that results are not compromise (n = 1); Not blinding reflects real practice (n = 2); Test for impact of not blinding post trial (n = 1)   first meeting as prior to (n = 2) or at (n = 1) trial commencement and before recruitment (n = 1)

Role
Journal manuscripts Eleven Lancet and 17 NEJM publications indicated role or responsibility. No BMJ articles discussed role. One supplementary appendix reporting TSCs (1/23 NEJM) gave insight into TSC role.

Design and oversight
Four Lancet publications reported involvement in trial design, specifically the committee designing the study with sponsor (2/4), under surveillance of the DMC (1/4), or supervising the design (1/4). Eleven NEJM publications reported involvement in design, wherein the role encompassed overseeing (2/11), being responsible for (3/11) and/ or involvement in (7/11) trial design, with one specifying that this was independent of the sponsor.
Three Lancet publications reported the TSC oversee the trial, of which one stated that the TSC supervised operations. Of the nine NEJM publications discussing oversight and conduct, five stated that the TSC oversees the trial (5/9), of which two required that they oversee conduct specifically (2/9). A   further four stated that the TSC was responsible for study conduct (4/9). One supplementary appendix reporting TSCs (1/32, 1/9 Lancet) indicated operational oversight role, specifically: 'The Steering Committee was responsible for overseeing the scientific and operational aspect of the study'. (Lancet49) Thirteen protocols discussed a role in oversight and trial design: the role encompassed monitoring (n = 5) and oversight of various aspects of the study (n = 5). The TSC had a responsibility or participated in trial conduct (n = 8) and design (n = 5). One stated that the TSC was independent.

Decision-making
Two Lancet publications reported a decision-making role, stating that the TSC could decide on study continuation. While no NEJM main papers reported the decisionmaking role, nine protocols did. Five explicitly defined the TSC as the decision-making body, the remaining four opted to provide examples of TSC decisions, such as altering sample size (n = 2), patient withdrawals (n = 1) and trial stopping (n = 1).
Contribution to trial documentation, data and analysis Seven Lancet publications involvement in trial documentation, stating that the TSC wrote or contributed to the final report (3/7), made the decision to publish (2/7), or both (1/7). The TSC coordinated and resolved doubts in interpretation in the protocol in another. Analysis and data involvement was reported in five studies, three of these stated the TSC had full access to data (3/5), one supervised the analysis (1/5) and three interpreted the results (one stating that this was done independently). Another stated: 'The (TSC) vouched for the completion and accuracy of the data gathering and analysis'. (Lancet6) Seven NEJM publications reported TSC involvement in trial documentation, in contributing to writing the manuscript (4/7) and developing the protocol (3/7), of which one developed this with the sponsor. Four reported TSC involvement in trial data, specifically the TSC had full access to data (1/4) and was involved with data collection (1/4); interpretation (1/4) and analysis (1/4). In seven, the committee made the decision to publish, one stated that the TSC vouches for integrity and completeness of the data and six stated: 'The (TSC) vouches for the accuracy and completeness of the data and the analysis and the fidelity of the study to the protocol'.
One supplementary appendix reporting TSCs (1/32, 1/23 NEJM) gave insight into role. Fifteen protocols discussed TSC input into trial documentation, data and analysis. The TSC was reported to have an input in publications (n = 13), the protocol (n = 7) and side studies (n = 4).

Communication
No publications specified communication between TSC and other committees. Eight protocols discussed the TSC communicating with the DMC (n = 7) and a Critical Event Committee (n = 1).
Monograph Forty monographs described role.

Design and oversight
Twenty-seven monographs described an oversight role, generally (11/27) or, more specifically, overseeing progress towards interim and overall objectives (8/27) or study progress as a whole (8/27). Three stated independent oversight and two that his was on behalf of the sponsor.
Generic definitions of TSC monitoring were provided in nine monographs. This was done in accordance with the MRC guidelines (n = 6), Good Clinical Practice (n = 1) and in two: 'The (TSC) ensured that the rights, safety and wellbeing of the trial participants were the most important considerations and prevailed over the interests of science and society'.

Decision-making
Role in decision-making was reported in seven monographs. Examples were about how the study is run (n = 2); premature closing (n = 5); and how pilot data will inform the main trial (n = 2).

Contribution to trial documentation, data and analysis
Twelve had the TSC review trial specific documentation. Specifically, the statistical analysis plan (n = 8) and protocol (n = 5).
Ten stated that the TSC review external information and five had authority over the publication strategy. One had the TSC approve further analysis and one approved additional studies.

Communication
The committee received recommendations from the DMC in six monographs, informed funders on trial progress in three, advised funders in one and liaised between the DMC and the Trial Management Group (TMG) in another. One stated that TSC responsibility in resolving disputes between PIs and another had the DMC as a subgroup of the TSC.

Activities
Journal manuscripts TSC activity having impact on trial design, conduct and analysis was reported in 12 publications (3 BMJ; 7 Lancet; 2 NEJM) with a total 14 examples reported (4 BMJ; 8 Lancet; 2 NEJM). No activities were reported in supplementary protocols.
The BMJ reported TSC involvement changing the sample size (n = 2) and primary outcome (n = 1): 'Before the start of recruitment and data collection, we changed the primary outcome to the reported quit attempt measure, which is predictive of eventual cessation. This followed expert advice from the Trial Steering Committee on the basis of smoking cessation research and approval from the Data Monitoring Committee'. (BMJ15) All seven Lancet decisions regarded early stopping. In five, the TSC decided to stop recruitment, of which three reported that this decision was based on DMC recommendations. Others reported the TSC deciding to close recruitment to a trial arm (n = 1) and make the decision to continue recruitment following an interim analysis (n = 1): 'Without revealing any results, the DSMB recommended to the Executive Committee and sponsor that the trial continue to the original pre-planned sample size. The basis for this recommendation was that, because of the rapid enrolment at the time of the interim analysis, there was insufficient 90-day data to assess the secondary endpoints, although there were no safety concerns. The Executive Committee and sponsor accepted the DSMB recommendation to continue enrolment, but remained masked to all study results'. (Lancet36) Within the NEJM publications reporting TSC activity relating to trial design (n = 2), the committee established when patients could have their dose tapered (NEJM42) and when crossover could be permitted: 'The independent Data and Safety Monitoring Committee and Study Steering Committee concluded that both progression-free survival and overall survival were significantly longer in the trametinib group than in the chemotherapy group and that immediate crossover to trametinib should be permitted'. (NEJM23) One supplementary appendix reported TSCs (1/32) publishing the letter of recommendation from the DMC to the TSC requesting one arm be closed due to accruing safety data (NEJM38), this was consistent with the main publication wherein the trial was prematurely stopped.

Conduct
No publication reporting TSC activity related to trial conduct.
One described the TSC deciding to write a new SAP (BMJ3) Another reported the TSC determining exclusions from analysis and imputations for deaths or drop outs (Lancet10).
Monographs Thirty-seven monographs reported 60 examples of TSC activity.

Design
Thirty-four examples of TSC activity impacting trial design was reported in 23 monographs.
Activities included the TSC changing the entry criteria (n = 7); endpoints or outcomes (n = 7); sample size (n = 2); and randomisation ratio (n = 1). Another reported the TSC closing a treatment arm (n = 1). Others reported TSC impact in defining the study design (n = 6) and recruitment period (n = 2).
Most significantly, the TSC made the decision to close the trial in five studies, as in HTA238: 'In a meeting of the Trial Steering Committee, it was accepted that it would not be viable to proceed with the trial and the formal procedure for closure

Conduct
Fifteen examples of TSC activity impacting trial conduct was given in 14 monographs.
Examples of conduct were where the TSC: had input in determining the data collection process (4/14); considered consent issues (1/14) and input into safety issues; for example, such as reviewing death data (1/14) and determining serious adverse event (SAE) data requirements (3/14).
In three (3/14) the TSCs made the decision to changing treatment regimens; for example: 'The project Steering Group determined that (a prescription) was inappropriate to a pragmatic study of this kind. It was agreed that the outcome would be more likely to represent the likely outcome of introducing TUVP if staff were to manage patients according to existing norms'. (HTA30) Others (3/14) changed the recruitment procedure, for example: 'On reflection and discussion of these issues, the research team and the Trial Steering Committee members felt that some of these issues could have been addressed (…). They concluded that many of the problems encountered were a direct consequence of the changes in research governance and ethical procedures that prevent members of the research team approaching patients directly, but instead place the burden of recruiting patients on busy primary care professionals'. (HTA75)

Analysis
Ten examples of TSC activity impacting analysis was given in ten monographs.
The TSC impacted the analysis plan in seven monographs (7/10). Examples were: determining variables for regression model (2/7), defining equivalence limits (1/7), removing previously planned subgroup analyses (1/7), deciding on the analysis approach to be used, e.g. intention-to-treat (2/7) or suggesting additional analysis: 'Finally, at the request of the TSC, a further exploratory analysis to examine the interactive effect of age on the effectiveness of MRI compared with no MRI was conducted'. (HTA111) Other activities were the TSC determining protocol violations (1/10) and making the decision to unblind the trial team (2/10), for example:

Discussion
The extent of the adoption of a TSC, the committee with a majority of independent members to whom the DMC make their recommendations, for trial oversight outside of the UK is unknown. Within the UK the establishment of a TSC is required by a number of major public funders yet, despite this, there is an absence of reporting standards regarding their constitution, activities and impact on trial conduct. This paper aimed to provide the first review of reporting of TSC activity by reviewing published academic