Statistical Issues and Recommendations for Clinical Trials Conducted During the COVID-19 Pandemic

Figures & data

Fig. 1 Key dimensions of assessment, mitigations and documentation to address the COVID-19 impact.

Table 1 COVID-19-related factors and examples of their potential impacts on clinical trials.

Factor	Example of impact/risk
Quarantines, travel limitations, participant unable/unwilling to travel to site due to personal pandemic-related reasons, site closures or reduced availability of site staff	Missed or delayed visits and assessments Inability to access study treatment Loss to follow-up Longer query response time Different investigators/different measurement modalities Delayed site monitoring Delayed patient enrolment
Interruptions to supply chain of experimental drug and/or other medications	Missed dosing of study drugs Changes in non-COVID-19 concomitant medications
Alternative administration of drug	Increased risk of dosing errors Lack of equivalence of methods of administration
Alternative collection of specimens	Challenges in reconciliation and verification
Alternative data collection	Lack of exchangeability of methods
COVID-19 infection/treatment	Temporary/permanent interruption of study treatment and/or study participation Potential effect on efficacy endpoints/estimands/safety Interactions of COVID-19 concomitant medications with study drugs

Table 2 Attributes of pandemic-related intercurrent events.

Participant’s adherence to study treatment	Study treatment permanently discontinued, and no new treatment started for disease under study; Study treatment permanently discontinued and switched to an alternative therapy for disease under study; Study treatment temporarily interrupted, or compliance significantly reduced without changes in concomitant therapy for disease under study; Study treatment temporarily interrupted, or compliance significantly reduced with changes in the concomitant therapy for disease under study, for example, start of rescue medications.
Study treatment accessibility	Study/region-wise drug supply interruption; Site unavailable to administer/dispense study treatment; Study treatment available at site but participant is unable/unwilling to get study treatment due to personal pandemic-related reasons.
Participant’s COVID-19 infection condition	Participant positive for COVID-19 and alive; Participant deceased due to COVID-19^a ; Participant with a suspected COVID-19 infection (may be distinguished further by presence or absence of symptoms); Participant with an exacerbation of underlying health issues due to reduced healthcare access.
Participant’s COVID-19 concomitant treatment(s)^a	Participants treated for COVID-19 (pharmacologically, oxygen, etc.); Hospitalized, not in intensive care; Admitted to intensive care.

^a COVID-19 related deaths and initiation of treatment for COVID-19 infections may also be considered as ICEs if they occur after the completion of study treatment or after other nonpandemic-related ICEs and before the time point associated with the endpoint of interest.

Table 3 Summary of analysis considerations.

Review of planned analyses	Review all planned main and sensitivity analyses to ensure alignment with the revised estimand(s). Review/amend methods for handling of missing data, or censoring rules, to accommodate pandemic-related missingness.
Summaries of pandemic impact	Summarize the occurrence of pandemic-related ICEs and protocol deviations. Summarize the number of missed or unusable assessments for all key endpoints. Summarize the number of assessments performed using alternative modalities. Summarize study population characteristics before and after pandemic onset.
Additional sensitivity and supportive analyses	Plan additional analyses for sensitivity to pandemic-related missingness. Consider the need for additional, alternative summary measures of treatment effect. Consider exploring inclusion of additional auxiliary variables, interaction effects, and time-varying exogenous covariates in the analysis methods. Consider subgroup analyses based on subgroups defined by pandemic impact, for example, primary endpoint visits before or after pandemic onset. Consider the need for evaluation of potential impact of alternative data collection modalities. Consider sources of data external to the trial, for example, to justify use of alternative modalities. Plan for additional safety analyses.

Table 4 Attributes of pandemic-related missing data.

Missing endpoint measurement	Assessment missing due to a participant’s premature discontinuation from the study overall for pandemic-related reasons; Assessment missing due to missed study visits/procedures while a participant remains in the study (intermittent missing data); Assessment delayed (out-of-window) and deemed unusable for an analysis; A composite score (e.g., ACR20 in rheumatoid arthritis) cannot be calculated because some components are missing; Assessment deemed to be influenced by pandemic-related factors and deemed unusable for a particular analysis because the interpretability of the results may be impacted (e.g., in assessments of quality of life, activity/functional scales, healthcare utilization, etc.); Recorded data cannot be properly verified or adjudicated due to COVID-19-related factors and deemed to be unreliable for analysis; Assessment performed after an intercurrent event intended to be handled with a hypothetical strategy and collected data are deemed unusable for this estimand.
Assessment accessibility	Site (facilities or staff) unavailable to perform study-related assessments; Site/assessment procedure available but participant is unable/unwilling to get assessment done due to personal pandemic-related reasons.
Participant’s COVID-19 infection condition	Participant positive for COVID-19 and alive; Participant deceased due to COVID-19; Participant without a known COVID-19 infection.
Participant’s COVID-19 concomitant treatment(s)	Participants treated for COVID-19 (pharmacologically, oxygen, etc.); Hospitalized, not in intensive care; Admitted to intensive care.

NOTE: Row 1 summarizes reasons of missing data, rows 2–4 summarize related conditions that contribute to those reasons.