Inequalities in the mental health of adults before and during the COVID-19 pandemic: technical notes

On this page

• Canadian Community Health Survey (CCHS)
• Survey on COVID-19 and Mental Health (SCMH)
• Data limitations
• Analysis

Canadian Community Health Survey (CCHS)

Statistics Canada collected data for the 2019 CCHS between January 2, 2019 and December 24, 2019.

The CCHS includes people aged 12 and older in every Canadian province and territory. However, territorial data were not released for the 2019 CCHS. This is because 2 years of CCHS data collection are required before territorial data are representative.

We also excluded 2019 CCHS data for youth (12 to 17) because the 2020/2021 Survey on COVID-19 and Mental Health (SCMH) only recruited adults.

The CCHS does not collect data from people who:

• are full-time Canadian Forces members
• live in institutions
• live in foster care
• live on reserves or other Indigenous settlements
• live in the health regions of Nunavik and Terres-Cries-de-la-Baie-James

As a result, less than 3% of the Canadian population is excluded.

The response rate of adults in the 2019 CCHS was 55%, with a sample size of 57,034. Respondents completed the voluntary 2019 CCHS by computer-assisted telephone interview or personal interview.

Survey on COVID-19 and Mental Health (SCMH)

Statistics Canada collected data for the 2020 SCMH between September 11, 2020 and December 4, 2020. They collected data for the 2021 SCMH between February 1, 2021 and May 7, 2021. For simplicity, these time periods are referred to as late 2020 and early 2021.

The SCMH includes people aged 18 years and older in every Canadian province and in the capital cities of the territories (Whitehorse, Yellowknife, and Iqaluit).

Because territorial data was unavailable in the 2019 CCHS, we excluded 2020/2021 SCMH data from respondents living in the territorial capitals.

In addition to people under 18 and people living in the territories but outside capital cities, the 2020/2021 SCMH did not collect data from people who live:

• in institutions
• on reserves
• in collective, vacant, inactive or unmailable dwellings

The response rate was 53% for the 2020 SCMH and 49% for the 2021 SCMH.

We combined data from the 2020 SCMH and 2021 SCMH to maximize statistical power to detect:

• significant differences between sociodemographic groups during the pandemic and
• significant changes in the magnitude of these inequalities before versus during the pandemic

After excluding respondents in the territorial capitals and respondents who did not agree to share their data with the Public Health Agency of Canada, the combined sample size of the 2020 and 2021 SCMH was 17,066.

Respondents completed the voluntary 2020/2021 SCMH by computer-assisted telephone interview or electronic questionnaire.

Data limitations

The combined SCMH sample was still much smaller than the 2019 CCHS sample. This could partially account for some non-significant results.

Another limitation is the potential for seasonal effects to confound results due to the different data collection periods of the 2019 CCHS and the 2020/2021 SCMH.

The different data collection methods used in the CCHS versus the SCMH might also confound the results.

Analysis

We used SAS Enterprise Guide 7.1 to analyze the data.

We used:

• sampling weights to obtain representative results and account for the survey design
• bootstrap weights to estimate 95% confidence intervals and coefficients of variation

We obtained pre-pandemic estimates from the 2019 CCHS data, and pandemic estimates from the combined 2020/2021 SCMH data. The only exception was for marital status, which only used 2021 SCMH data because marital status was not assessed in the 2020 SCMH.

The weights for the 2020/2021 SCMH were multiplied by 0.5 to follow Statistics Canada guidelines for combining the 2 cycles of the survey.

For the categorical mental health outcomes, we examined absolute inequality before the pandemic in 2019, and during the pandemic in late 2020/early 2021. We did this by comparing the percentage difference for a mental health outcome (recent suicide ideation, high self-rated mental health, or high community belonging) in one sociodemographic group to that of another group (for example, female vs. male). When the 95% confidence interval (CI) of the percentage difference did not include 0, the absolute inequality between groups on that mental health outcome was determined to be statistically significant at that point in time.

To examine whether the magnitude of absolute inequality changed during the pandemic for each sociodemographic comparison, we compared the 95% CIs of the percentage differences in 2019 and in late 2020/early 2021. If the CIs did not overlap, we concluded that there was a statistically significant difference in the magnitude of absolute inequality for that sociodemographic comparison.

We also examined relative inequality for the categorical mental health outcomes at each point in time. We did this by obtaining the percentage ratio for a mental health outcome (recent suicide ideation, high self-rated mental health, or high community belonging) in one sociodemographic group over another group. When the 95% CI of the percentage ratio did not include 1.00, the relative inequality between groups was determined to be statistically significant at that point in time.

To determine whether the magnitude of relative inequality changed during the pandemic for each sociodemographic comparison, we compared the 95% CIs of the percentage ratio in 2019 to that in late 2020/early 2021. If the CIs did not overlap, we concluded that there was a statistically significant difference in the magnitude of relative inequality for that sociodemographic comparison.

For the numerical outcome of average life satisfaction, we examined absolute inequality at each point in time by comparing the mean difference for life satisfaction in one sociodemographic group compared to another. When the 95% CI of the mean difference did not include 0, the absolute inequality between groups on life satisfaction was determined to be statistically significant at that point in time.

To determine whether the magnitude of absolute inequality in life satisfaction changed during the pandemic for each sociodemographic comparison, we compared the 95% CIs of the mean difference in 2019 and in late 2020/early 2021. If the CIs did not overlap, we concluded that there was a statistically significant difference in the magnitude of absolute inequality for that sociodemographic comparison.

If the mean difference in life satisfaction between groups was statistically significant at either point in time, we also reported the percentage increase/decrease in average life satisfaction that the reference group would experience if it had the same mean life satisfaction as the comparison group (that is, if the inequality did not exist).