We used Qualtrics XM (Qualtrics) to obtain consent, collect demographic information, and administer patient-reported outcome measures (PROMs). PROMs allow us to collect quantitative data on subjective psychosocial constructs [21]. Participants were given the link to the Qualtrics survey and asked to complete the survey within 7 days; an interview was then scheduled. We conducted semistructured interviews via teleconference using Zoom (Zoom Video Communications) to collect qualitative data on participants’ experiences with COVID-19 restrictions. Prior to being used with study participants, the survey was pilot-tested with research team members and one participant who had experienced a stroke. In order to increase accessibility to the survey, if the person was unable to complete the survey online, it was administered over the telephone.

Prior to the first interview, we collected data on participants’ age, city of residence, country of birth, sex, gender, first language, living situation, educational level, income at baseline, employment status, chronic illnesses, and status regarding COVID-19 (ie, whether they have tested positive or been exposed to someone who had tested positive).

Following the completion of the surveys, participants took part in semistructured interviews on the University of British Columbia’s (UBC’s) secure Zoom platform to maintain physical distance. Three interview guides were created (Multimedia Appendix 1): a guide for the first interview at T1, a guide for the interviews at T2 and T3, and a guide for the final interview at T4. Participants were asked about their experiences with daily activities, changes in activity, their feelings about these changes, and strategies they used to cope with the impact of COVID-19 restrictions. Each interview guide was tested in three pilot interviews prior to being used with participants. Each participant was assigned the same interviewer for each of their interviews to facilitate rapport and relevance of follow-up questions based on previous interviews. At the start of the first interview, the interviewer confirmed any changes to the demographic information collected by the survey at each subsequent interview (eg, living arrangements and employment status).

Participants were also invited to take pictures during the course of the study that represent their daily and social experiences during the period of COVID-19 restrictions. The photos acted as a catalyst to enrich the sharing of information, as interviewers prompted participants to share the meaning behind the photos. Participants were invited to send photos to the research coordinator at each time point, which were then combined into a collage by the research team. The interviewer shared the collage with participants in the final interview and asked them to choose two to three photos from the collage and describe what the photos meant to them in the context of the COVID-19 restrictions.

The interviews were recorded on a password-protected Zoom account as well as on a voice recorder as a backup, to ensure accuracy of the transcription, and were then transcribed verbatim. Participants were anonymized with a unique participant ID; the key matching the ID with the name and contact information for each participant is kept in a password-protected and encrypted Microsoft Excel file on a secure UBC server. When transcribing the interviews, each participant’s ID was used to refer to them and other proper nouns were replaced with pseudonyms. Participants received a Can $30 (US $23.78) honorarium after each interview.

Before conducting the quantitative analyses, data will be imported from Qualtrics to SPSS (IBM Corp). First, univariate statistics will be used to account for the number and percent of missing data in each measure. The patterns of missing values for each measure will then be visualized. Missing value patterns will be evaluated carefully to determine missing data mechanisms (ie, missing completely at random, at random, or not at random) [33]. In addition, to determine whether the observed missing value patterns are related to other variables in the study, for each measure that contains missing values, an indicator variable will be developed. The indicator variable will separate the participants who provided complete responses to that measure from the participants who did not complete that measure. Logistic regression will be used to determine whether other variables in the study, including demographic variables, can statistically predict the indicator variable. If none of the variables in the logistic regression analyses predicts the indicator variable, it can be concluded that the missing pattern mechanisms are missing completely at random or missing at random. If the values are missing completely at random or missing at random, and the percentage of the missing values is less than 30%, we will impute the missing values using a multiple imputation technique. Multiple imputation will result in multiple sets of plausible values for each missing value. In these analyses, the multiple imputation will be used to compute five plausible values for each missing value. The missing values will then be replaced by the mean of the five plausible values. Multiple imputation analyses will be run for each measure that contains missing values in each group separately. After imputing the missing values, to test our hypotheses and research questions, the following statistical analyses will be used. First, univariate analyses (eg, mean, sum, standard deviation, variance, range, and frequency) will describe the sample. In cross-sectional analyses, multivariate analyses of variance will be used to test whether there is a statistically significant difference between groups on the outcome variables. Correlational analyses, including the Pearson correlation coefficient (r), will be used to test the strength of the associations between different variables. Regression analyses (eg, linear regression and logistic regression) will be used to estimate the relationship between dependent variables and the outcome variables.

To analyze the longitudinal data, hierarchical linear modeling (HLM) growth curve analysis will be used to investigate the changes in the participants over time [34]. HLM growth curve analysis has several benefits over repeated-measures analysis of variance. First, HLM can be used when the interval between the time points is not equal and when data contain missing values. In addition, in contrast to the repeated-measures analysis of variance, HLM focuses on individual differences over time by considering each participant’s initial intercept and slope score, while repeated-measures analysis of variance focuses on the group differences [35].

Transcripts will be analyzed using content analysis to develop a qualitative description [18] of events and experiences of participants within and across the four time points. Analyses will be conducted separately for each of the participant groups—families, people with SCI, people with stroke, people with disabilities, and older adults—and each time point; they will then be combined, if appropriate, to examine trends over time or themes across groups [36]. A subteam of researchers will analyze each group. Because of the large research team, the blend of novice and experienced researchers, and the five participant groups, a consistent approach to coding will be facilitated by developing coding manuals with notations to explain coding decisions [37]. We anticipate creating multiple codebooks based on participant status, data collection time point, and omnibus analyses. These coding guides remain flexible to the possibility of adding new codes and recoding previous interviews. For each coding team, two primary coders will read and reread transcripts, code the first two to three interviews, compare codes, and discuss any discrepancies with the coding team. This coding guide will then be applied to subsequent interviews and revised in consultation with the qualitative research team for each participant group. During initial coding, the primary analysts will assign tentative codes to each idea reflective in the text, which are recorded in a codebook. The codebook and sample quotes supporting the codes will be shared with senior researchers. After integrating the comments from the senior researchers, the coders will code the next three to four interviews separately and once again compare their codes to ensure consistency in applying the codebook, updating as needed to reflect new or revised codes. The codebook developed at T1 will be used to code T2 interviews, updated with new codes developed at T2, which will be applied and updated at T3 and then again at T4; that is, codebooks will be updated as interviews are conducted throughout the longitudinal study. The coding procedure will be applied to each participant group separately, requiring four iterations of five participant group codebooks. Codes will then be organized into categories representing and interpreting the main topics shared by participants. The final qualitative description is expected to be presented as a set of themes and their interrelationships, supported by illustrative quotes from participants to explain their experiences of pandemic restrictions during the study period.

To integrate quantitative and qualitative data, there are three potential models of integration that may be used. Our primary approach will be to analyze both data sources separately. We will also explore the possibility of sequential analysis in which quantitative or qualitative analysis is used to inform a subsequent analysis using the alternative type of data [37] (eg, we may identify different types of experiences based on the qualitative data and compare scores on quantitative measures among participants who have similar experiences, or vice versa).

With regard to trustworthiness [38], an audit trail includes reflexive memos to document research discussions, possible biases, and analytical decisions (ie, dependability and confirmability). Comparing qualitative themes with numerical results from the PROMs will be used as a form of data triangulation and will generate questions to consider in refining qualitative themes. Researcher triangulation will occur within each subteam, in their analysis of a specific participant group, and in the larger team, where all researchers will be involved across all participant groups. Where available, member checking will further enhance analytical rigor.