Reporting a study disrupted by an external event
A trial that runs through a pandemic, a war, or any other event that closes sites and stops people travelling produces a clinical study report that has to explain the interruption as well as the results: what the disruption did to the study, where the contingency measures, the deviations and the affected participants are reported, and how much of it to include. The emergency guidance the EMA and the FDA issued during the COVID-19 pandemic of 2020 set these requirements out.
What the disruption did to trials
An estimated more than 80% of the clinical trials running in 2020 were disrupted by the pandemic in one way or another. Between about March and May 2020 thousands of trials, around 80% of non-COVID-19 trials, reported a disruption, most of them early-phase trials not yet enrolling. The disruptions reported were delayed or suspended enrolment, delayed start-up or site initiation, delayed dosing, enrolment terminated at particular sites, early termination, and delayed completion or availability of data.
The disruptions ranged from simple ones to the study-ending. Recruitment was delayed and visits were cancelled. More seriously, manufacture and delivery of investigational product were delayed, samples, tests and data points went missing, and, because the disease was new, new adverse events and new concomitant medications appeared. At worst, participants dropped out, sites closed and trials were terminated. Entries in ClinicalTrials.gov from the period show, for example, recruitment suspended in one trial and a site closed in another.
The most affected therapeutic areas included those enrolling people with advanced or life-threatening conditions such as cardiovascular disease and cancer, who are the patients most at risk from COVID-19 and often have few other treatment options. Oncology treatments are typically given intravenously, so patients have to come to the site for infusions, which added logistical hurdles. In many cases, trials nearly or fully enrolled with patients who had life-threatening conditions continued with modifications to address the challenges of running a trial in an already burdened health care system.
How far a disruption affected a study depended on where the study stood in its timeline when the disruption arrived: in startup, in conduct or in closeout.
The emergency guidance and the principles beneath it
The emergency guidance set out to meet trials' need for continuity and contingency plans, and it brought both additional and new requirements. The EMA and the FDA both reacted quickly with emergency guidance and have kept updating it. Sources disagree on when the FDA's guidance first appeared. A webinar given in March 2021 dates the EMA guidance to March and the FDA guidance to April. An AMWA Journal article of Summer 2021 dates the FDA guidance to March 2020. The EMA split its guidance in two: one document on clinical trial management, and a second, less widely known one on methodological aspects, which covers data collection, data validation and statistical analysis. The FDA put the equivalent material into a single document. The guidance is still evolving, so the version to check is the most recent one.
Under good clinical practice, each disruption is judged by its effect on participant safety and on the data integrity of the trial. The EMA's guiding principles for the risk-benefit assessment put participant safety first and data validity second, and add a third consideration: the strain on healthcare professionals. Investigators and site staff carried the heavy workload of treating patients with COVID-19, and regulatory authorities and ethics committees were overloaded with work on clinical trials. A decision should also not overload the sites and the competent authorities with workload and paperwork. Each decision depends on the trial, and no one way of reaching it fits every trial.
The EU recommendations and the EMA guidance
In the EU, the European Commission, the EMA and the national heads of medicines agencies published new recommendations for sponsors on 20 March 2020. The pandemic's impact on European health systems and on society would make it necessary for sponsors to adjust how they managed clinical trials and the people taking part in them. The recommendations gave sponsors concrete information on the changes and protocol deviations that extraordinary situations might need, such as participants in self-isolation or quarantine, limited access to public places including hospitals, and healthcare professionals being reallocated. They set out harmonised recommendations to keep participants safe across the EU while preserving the quality of the data, and advised on how to communicate the changes to authorities. Trials in the EU are authorised and supervised nationally, and sponsors were advised to check for national legislation and guidance that might complement the recommendations or take priority over them.
The sections of the EMA guidance fall into three groups. The first is risk-benefit assessment and safety reporting. Clinical researchers, sites, sponsors and CROs need to assess risk and benefit and to collect and report adverse events continuously through the disruption, and adverse-event reporting in particular should not stop. On that continuous assessment they can decide whether a trial already under way enrols new participants. The second group is communication: clear communication with the competent authorities and ethics committees, clear communication with the sites, with the documentation properly tracked, and communication with the participants about their trial and also about the disruption and how it affects them. The third and largest covers changes to the conduct of ongoing trials, to distribution of the investigational product, and to monitoring and audits.
In the conduct of trials, activity moved from physical to remote wherever possible, some of it was postponed or even cancelled, some participants were moved to other sites, and trials generally ran longer. Monitoring and audits also moved from face to face to remote, and whatever could be postponed had to be. Distribution of the investigational product changed too and used much new technology, but the aim throughout was the participants' safety and well-being: that they received their life-saving medicines, and that accountability for the investigational product was maintained. Consent matters too: technology brought many alternative consent methods for when consent cannot be taken face to face, and their documentation is again stressed. The EMA guidance also has a section on reimbursement and expenses during the disruption, which is of little relevance to the medical writer.
The FDA guidance
The FDA guidance on conducting trials during the pandemic aims to keep participants safe, maintain compliance with good clinical practice and minimise risks to trial integrity, and it has been updated regularly. Among other topics, it covers:
- electronic and other remote consent
- alternative visit formats, such as phone or virtual visits, alternative locations, local laboratories or imaging centres, and home nursing
- alternative formats for patient-, clinician- and observer-reported outcome assessments, with protection of data privacy
- supply-chain matters, such as direct-to-patient shipping, administration by a home nurse and disposal of investigational product
- remote site monitoring, including prioritising monitoring activities where sites closed, infection-control restrictions applied or local travel was constrained
- serious adverse event reporting, including COVID-19 cases in non-COVID-19 trials and cross-reporting across multiple INDs
Its key point is to collect information on when and how each mitigation was implemented, so that the pandemic's impact on the trial objectives can be reported later.
Under the FDA guidance, screening that forms part of a country's standard healthcare is not a protocol deviation, so a participant's COVID-19 test is not one, and pausing enrolment does not require an amendment. One experienced writer reads both points as the FDA's way of saying not to over-report: to report the important things, not everything. Policies and procedures should also follow national policy for COVID-19 management and control. This matters especially for a global trial running in many countries, which has to follow the national guidelines. An update to the FDA guidance also states that COVID-19 vaccines authorised for emergency use are not to be considered investigational but standard of care.
The documents the disruption reaches
The disruption reached study conduct, data collection, analysis and reporting, and the essential trial documents a medical writer routinely handles. Asked which documents the emergency guidance directly affected, with several answers allowed, respondents named the protocol and its amendments (97%), the clinical study report (93%), the informed consent form (84%), the statistical analysis plan (81%) and, much less often, the investigator's brochure (33%). One experienced writer agrees that the amendments, the consent form, the analysis plan and the report are the documents most affected, and the investigator's brochure less so. The documents fall at different points of the study's timeline: the protocol is developed during startup, and during conduct the focus is on the amendments and the statistical analysis plan. Writers are fully responsible for some of these documents and only contribute to or review others.
Case report forms, coding and integrated or separate reporting
MedDRA, normally updated each spring and autumn with the larger update in spring, received an additional update to include COVID-19 terms, in the first half of the year, around when the lockdowns began. The new terms cover not only adverse events and symptoms but also immunisation and testing, which have to be captured and coded in a standardised manner too.
The disruption raises decisions such as whether to update the whole case report form or add pages specific to the disruption, which varies with the type of trial, and, where coding is already finished, whether adverse events are recoded with the new terms. The main decision is whether the disruption-related data are integrated into the normal outputs and the normal report or kept separate. The FDA leaves the choice open, asking for the disruption-related data to be described in the clinical study report or in a separate document.
Polled on the practice followed, 49% of respondents had written no report covering the disruption, 20% had written at least one, 18% were working on one and 12% were about to start. In a second poll, 56% reported the disruption data integrated into one report, 29% were still in the process of developing one and only 3% reported the data in a separate document, so practice seemed to be moving towards the integrated report.
Affected data domains and their analysis
The data affected include participant disposition, study compliance, protocol deviations, exposure and dosing, concomitant medications, adverse events, laboratory data and efficacy, and possibly others. For every affected domain the guidance asks for the same three things to be examined: validation, treatment effects and sources of bias. For example, where a measurement is taken at home with a different device, validation shows whether the measurements from the two devices are comparable. The trial's sample size has to be revisited too. The FDA guidance has similar requirements on handling missing data and on analysing disruption-related deviations.
Resources beyond the regulators' guidance
Resources beyond the two regulators' guidance documents are more specific to the writer's needs. PHUSE, a non-profit whose members are data scientists, biostatisticians and statistical programmers, published a set of freely downloadable white papers on the disruption's effect on adverse events, laboratory data, efficacy, protocol deviations and the other affected domains, and some disruption-specific deviation categories come from them. CDISC guidance covers producing the corresponding data sets.
TransCelerate, a consortium of member companies working to standardise key clinical trial activities, released a clinical study report template for describing COVID-19-related trial impact. Because information changed regularly, effects varied by region and global health agencies kept issuing updates, any such tool is a work in progress rather than a static guideline, and writers need to keep up with its updates.
One freelance regulatory writer who worked on nothing directly related to COVID-19 still had to adapt to the ever-changing guidance on addressing the pandemic's impact on trials in clinical study reports, and to teach clients about it. That writer highly recommends the FDA guidance and the clinical study report template wording TransCelerate suggests.
Contingency measures and deviations in the report
The FDA asks for the contingency measures taken during the disruption to be described in the clinical study report, together with their impact on the safety and efficacy results. TransCelerate's long document CSR Considerations for Studies Disrupted by the Pandemic says that the report has to tell the story by summarising the impact, be forthcoming, describe the disruptions and the deviations, establish emphasis by making clear how substantial the impact was, and allow for subsequent learning, perhaps for a future pandemic.
The contingency measures are reported in Section 9.8 of the ICH E3 structure, which covers changes in the conduct of the study and in the planned analyses. Each measure is described and justified there. Changes in study conduct include visits moved from face-to-face to home visits, site transfers, laboratory changes and possibly a data monitoring committee set up at short notice. Changes in the planned analyses include the modifications to the case report forms, any change in how data were captured, the analyses and the additional analyses the EMA and the FDA require to determine the impact of the disruption. Section 9.8 is likely to run somewhat longer than usual.
What effect the mitigations had on the continuation of non-COVID-19 trials during the pandemic is unclear. Their success has to be analysed and reported study by study, according to what each company judged necessary when the pandemic reached the trial, and any publication based on the trial is footnoted with that context. Because the measures could be implemented differently across regions, some may have succeeded in some regions but not others, or succeeded overall while other planned mitigations failed because of local circumstances. Putting the results of the mitigations in context and reporting them is time-consuming, and needs complete source documentation and follow-up with CRO and site partners.
Deviations arising from the disruption are reported in the sections on protocol deviations, supported by a listing of them. Not every effect of the disruption on protocol conduct necessarily makes it into a formal protocol amendment, and those that do not are non-conformances. In one writer's experience, non-conformances are reported less in Section 9.8 than earlier, in the methodology sections.
Participant listings and comparisons
The FDA specifically asks for a listing of every participant whose participation was altered by the disruption, documenting the nature of the impact for each. Such a listing can be structured by study, group and participant number, with the impact described and a yes-or-no field for whether there was an impact at all.
Participant disposition can carry a single added line for discontinuation due to the disrupting event where the impact was small, or a fuller breakdown where it was severe. Product accountability and compliance show that participants actually received their medication, and modified visits have their own listing. Missed visits are recorded with whether the reason was related to the disruption, in terms specific to it, such as quarantine, travel restriction, a site recommendation, symptoms or a diagnosis, rather than in the standard categories. Deviation categories divide the same way, between the standard ones on eligibility, dosing and safety follow-up and disruption-specific ones such as movement restriction, site action and interruption of investigational product shipment. Adverse-event listings can carry, beside the start and end of each event, an event-of-interest field and whether the event is related to the disrupting condition. Mortality rates and causes of death are documented in a further table, with each death classed as related or unrelated to the disrupting condition.
Two comparisons are expected. The first is between the periods before, during and, where the study reaches it, after the disruption, to show its effect on the treatment effect and the potential sources of bias. The second is between treatment arms: whether the arms were affected differently, and whether one was in effect more protected than the other. The handling of dropouts and missing data, and the examination of subgroups, such as participants who were infected and those who were not, are worked out with the statistician.
TransCelerate supplies boilerplate text for describing the disruption in the report that works almost as a template, with the passages to fill in or modify marked. It recommends opening with a chronology: when the trial started, when it completed, and where the disruption falls within that span, followed by a description of the impact on the trial and on the results. It also gives examples of outputs showing how the data set is affected by the disruption. Further example text covers the before-and-during comparison and the evaluation of imbalance between treatment arms.
Confounders
Confounders have to be examined as well. The study population may have been more susceptible to the disrupting condition: an elderly or immunosuppressed population may differ from healthy volunteers in that respect. The disruption to dosing and the delay to treatment can be considered confounders too, with their impact on the treatment effect and on safety. The disrupting circumstances may also have made the disease itself worse, as in mental-health trials run through lockdown, where the lockdown affected the condition being studied. Any other intercurrent event during the disruption is a possible confounder too.
How much to report
Reporting a disrupted study calls for balance. Not everything belongs in the report, and common sense and sound judgement decide what to report and how. Reporting of the disruption was not standardised, the requirements were evolving, and the data could be reported in many different ways. The report should be forthcoming but avoid over-reporting.
The EMA asks for the many protocol deviations the disruption brought to be documented, handled and escalated properly under standard operating procedures, and documented in the clinical study report, but proportionately and without over-reporting, because reporting every small deviation would overwhelm the writer with documentation.