Table 3 Considerations for measurement and analysis of COVID-19 impacts by PICOTS domain

Patient population

∙ Measure changes to participant demographic and clinical characteristics

∙ Measure changes to access/attitudes (i.e., Coronavirus Impact Scale)

∙ Evaluate COVID-related methods for electronic phenotyping

∙ Compare characteristics over time periods (i.e., pre and post March 2020)

∙ Compare characteristics over time periods defined by local COVID status or impacts on research conduct

Intervention of interest

∙ Document modifications to ensure the safety of patients and providers

∙ Document/measure changes to the methods to communicate and deliver intervention

∙ Document/measure changes to any components or mechanisms associated with intervention

∙ Compare engagement in intervention components over time periods

∙ Consider modification to primary analysis to adjust (stratify) for COVID phases

∙ Consider sensitivity analyses that restrict to periods of time based on COVID status

∙ Consider subgroup (interaction) analysis of heterogeneity of treatment effect (HTE) across COVID time periods

∙ Consider time-specific mediation analysis to evaluate putative mechanisms across COVID time periods

Comparison intervention

∙ Document modifications to ensure the safety of patients and providers

∙ Measure trends in concomitant care over time

∙ Compare engagement in comparison (usual care) over time periods

∙ Clarify the estimand of interest—typically the average treatment effect that averages over explicit patient and time characteristics while potentially controlling certain factors

∙ Consider subgroup (interaction) analysis of HTE across COVID time periods

∙ Report the consequence of COVID impacts that are specific to the study design (i.e., parallel randomized trials will be impacted differently from crossover or stepped-wedge designs)

Outcome(s)

∙ Document/measure changes in the way that clinical assessments and patient-reported outcomes are obtained

∙ Measure the impact of COVID-related distress on patient-reported outcomes

∙ Report the psychometric properties of any modified measures

∙ Consider analysis of subgroups based on patient-reported characteristics at baseline that are predictive of outcomes or COVID susceptibility

∙ Describe temporal trends in patient-reported outcomes

∙ Evaluate patterns and correlates of missing data over time to assess differential non-response

Timing

∙ Document COVID status at the regional level over time

∙ Document COVID status at the clinic level over time

∙ Measure participant-level impact of COVID over time

∙ Consider evaluation of participant outcomes at longer (shorter) lags to determine potential impacts on response trajectories

∙ Consider effect modification with proximal measures of COVID impact (patient level, clinic level)

∙ Describe longitudinal trajectories in patient outcomes across treatment groups and across COVID time periods

Setting

∙ Document the research setting (modality) for both intervention delivery and participant assessment

∙ Describe differences in intervention, comparison, and concomitant care across specific research settings

∙ Describe differences in outcomes across research settings