Scientific Papers

Good Statistical Practice—development of tailored Good Clinical Practice training for statisticians | Trials

Description of Image

Review of current GCP training

An initial scoping exercise with UKCRC CTU Statisticians highlighted that, although interesting and research related, the GCP training statisticians received was felt to be unrelated to their statistical role. At a bi-annual meeting attended by at least 1 senior statistician from each of the 45 UKCRC CTUs registered at that time, all but one attendee felt there was a need for more role specific training. A follow-on survey of 45 statisticians representing the 45 UKCRC CTUs, completed at the next bi-annual meeting, confirmed the need and clear desire for the development of a dedicated GCP training for statisticians (Table 1). The majority of responders, 34 (76%), were senior statisticians responsible for designing trials and supervising analyses; 11 (24%) were statisticians responsible for analysing trial data; 22 (49%) had worked in clinical trials for more than 10 years; 10 (22%) 5–10 years and 39 (87%) worked on CTIMP trials.

Table 1 Survey responses

One responder had not received any GCP training, but of those that had, 21 (48%) reported they had received certified GCP training and 19 (43%) received in-house training. Not exclusively, 15 (34%) attended NIHR-certified face-to-face GCP training, 11 (25%) NIHR-certified online GCP training and 5 (11%) attended Institute of Clinical Research (ICR)-certified online training. NIHR GCP training is designed for individuals involved in the conduct of studies at research sites, and NIHR acknowledge their training will not prepare those who have responsibility for other elements of a study.

Crucially, of 44 recipients, only five (11%) considered the GCP training they had received was highly relevant to their role, and only seven (16%) thought it helped them understand GCP requirements related to their role. The development of a dedicated GCP training course for statisticians was supported by 30 (68%) statisticians; only 4 (9%) thought there was no need for a dedicated GCP training course.

Respondents were asked to choose their preferred form(s) of GCP training with 25 (56%) stating online, 20 (44%) face-to-face and 6 (13%) reading/workbook based; 4 (9%) reported no preference. Collective team training within statistical teams were preferred by 31 (69%); 13 (29%) preferred multi-professional CTU restricted teams; 4 (9%) multi-professional and un-restricted.

Design and development of GSP training material

A comprehensive set of training material has been developed to provide training for the translation of GCP into practice for statisticians involved in the conduct and analysis of clinical research in the UK. The training material has been developed by an experienced team of statisticians with knowledge of UK regulators and funders and in consultation with NIHR Learn and MHRA. It provides a high-level overview of GCP requirements and recommendations for best statistical practice.

The training material was developed through a version-controlled process to a final set of 114 slides, separated by learning objectives into five subject-specific modules, accompanied with detailed notes, specifically: (i) Module 1: Core GCP and Regulations; (ii) Module 2: Record Keeping and Documentation; (iii) Module 3: Trial Design; (iv) Module 4: Data Management and (v) Module 5: Statistical Analysis and Reporting. The five modules (Fig. 3, Table 2) focus on GCP requirements or recommendations directly related to statistical activities, or activities which would usually require some statistical involvement, as well as additional topics applicable to staff working in research more generally, but of which statistical staff should have an awareness.

Fig. 3
figure 3

Modular training structure

Table 2 Training topics and modules

Feedback from the UKCRC CTU Statistics group activities indicated a desire for training to be stand-alone, which resulted in the additional generic module on core GCP principles (Module 1) to be developed. The modular approach follows the logical order of the progressive stages of a clinical trial, from trial design through to data analysis and reporting. The final module (Module 5) contains content most relevant to statistical programming and analysis incorporating recommendations from other UKCRC statistical practice guidance. References are provided to point to more in-depth guidance.

The modular format allows flexibility regarding delivery (face to face or e-learning) either as stand-alone training or to supplement usual local GCP training practices. Face-to-face training is intended for delivery within statistics teams where the lead training provider is an experienced researcher with a good understanding of local processes. Group activities in a face-to-face small group teaching environment provide an opportunity to consider how GCP principles can be implemented in line with local statistical practice and standard operating procedures, including consideration of risk proportionate approaches given the variability in trials’ portfolios across CTUs. Translation of the training material to e-learning is crucial for accessibility. E-learning is essential for statisticians working in research teams but isolated from other statisticians, allows immediate access for new statisticians and allows accessible continued professional development. Training certificates document specific modules completed.

Critical review and piloting

During the development phase, updates were provided, and feedback was received at 6-monthly meetings of the UKCRC CTU Statistics Group. This feedback provided direction to both the content and format for presentation. A draft version of the complete training package was piloted through small group face to face training in five CTUs (Oxford, Royal Marsden and Imperial Cancer Research, Edinburgh, Leeds, Newcastle) and took approximately 2 h to deliver, up to 3 h including exercises and discussion. A senior statistician within each CTU delivered draft versions of the slide content and provided collective feedback regarding understanding, relevance and practical implementation. It received overwhelmingly positive feedback: “this is a really valuable tool to add to our training”; “something the stats community definitely needs and pleased that this is being taken forward”; “all-in-all that was a very positive experience”; “the general feeling was that it was a lot more useful than an afternoon spent at standard GCP training”; “people were engaged… thinking if any [local] practices could be improved.” The face-to-face engagement was particularly highlighted: “the face-to-face aspects are particularly useful as this enabled us to discuss the various aspects in relation to our CTU SOPs, processes and documentation etc.” as was the relevance to new starters: “I wish I’d had this when I first started out.” The feedback was extensive and detailed including suggested amendments to content, presentation and language in order to clarify ambiguities. Following feedback, the training materials were extended to incorporate (i) training certificates, (ii) core regulatory GCP principles to save having to complete two courses and (iii) a recommendation for frequency every 2 years.

Additionally, a standalone, day-long, small group teaching session, based on updated draft versions of the slide content and accompanied detailed notes, was delivered by members of the development team, to NIHR staff in an external unit working on non-clinical trial NIHR funded research, including statisticians and data management staff. Fourteen of the attendees provided formal written feedback (Table 3): all would recommend the session to a fellow researcher, four scored the session 5/5 (excellent), the remainder scoring 4/5; 13 respondents “strongly agreed” or “agreed” the training material was relevant to their role, and all “strongly agreed” or “agreed” that the training increased their learning and understanding of GCP requirements in relation to their role. Useful, supportive comments included “far more relevant to ‘real life’ than basic GCP training,” “essential info for trials researchers,” “informative, relevant, good structure,” “comprehensive” and “this course should be made available to anyone using/collecting data rather than just statisticians. All of the people who work with statisticians should work as a team and therefore be offered similar training opportunities where roles/activities overlap.”

Table 3 Quantitative feedback from pilot work

The training was summarised at an invited parallel workshop at the NIHR Statistics Group annual meeting to ensure applicability outside of the UKCRC Registered CTU network. Structured, anonymous feedback from eleven roundtable groups (Table 3) demonstrated that every group either “strongly agreed” or “agreed” that the training material was relevant to their role. The majority (82%) “strongly agreed” or “agreed” that the training increased their learning and understanding of GCP requirements in relation to their role. A final draft was circulated to the MHRA for comment.


The Good Statistical Practice training materials are freely available and accessible through a variety of portals:

  1. 1.

    To CTUs via the UKCRC online platform

  2. 2.

    To NIHR researchers via the NIHR Learn platform

  3. 3.

    Worldwide via the University of Leeds Institute of Clinical Trials Research

The material lends itself to statistical team training, pre-conference training at relevant statistical and/or clinical trials conferences as well as individual learning.

An oral presentation at the International Clinical Trials Methodology Conference [18] shared the need, development and pilot work and outlined content and modules, initiating wider dissemination activities and global awareness. The Good Statistical Practice training was launched in an accessible UKCRC led Trial Methodology Research Partnership webinar [19].

As a UKCRC CTU initiated project, the network will be able to update the material as necessary, monitor access metrics and measure longer-term impact.

Description of Image

Source link