*2.3. Data Extraction and Analysis*

Data extraction was performed using the Covidence Extraction 2 tool, based on a custom data extraction template. The template covered general information about the article and where it was published; characteristics of the study setting, population, study objective and design; characteristics of the COVID-19 vaccination indicator; characteristics of the dimensions of inequality; analysis methods; results; and conclusions. After reviewing 10% of studies in tandem to ensure consistency in the interpretation and application of the template, data extraction was performed by one of two reviewers. The reviewers reached consensus on any questions or points of ambiguity that arose with input from a third reviewer.

Using the data extraction outputs, descriptive data analysis was undertaken to assess and tabulate study characteristics. In describing the frequency of dimensions of inequality and inequality trends, we used the PROGRESS-Plus framework as a starting point for grouping dimensions of inequality pertaining to common themes. PROGRESS factors include place of residence, race/ethnicity/culture/language, occupation, gender/sex, religion, education, socioeconomic status and social capital [16]. We also identified the following categories, some of which align with the factors described in the "Plus" component of the above framework: age; disability status; family size or composition; health insurance; housing type or characteristic; marital status; migration status; sexual orientation; subnational region or area; and vulnerability, deprivation or poverty index.

As an extension of our analysis, we assessed the preliminary trends in the findings related to dimensions of inequality that appeared most often in the assessed articles. To this end, we coded and compiled reported findings for age; race, ethnicity, cultural group, language and nationality or country of birth; and sex or gender. For age and sex or gender, where the criteria for measuring the dimension were largely comparable across most studies (as years or male/female, respectively), we coded the main findings according to the directionality of the inequality. For race, ethnicity, cultural group, language and nationality or country of birth, where the criteria for measuring the dimension were heterogeneous, we coded whether inequality related to this dimension was reported as meaningful or not meaningful. Our coding of results as meaningful or not meaningful was based on the conclusions reported in the original studies. Most, but not all, studies defined this as statistically significant in comparisons at *p* < 0.05; however, the nature of statistical comparisons differed by paper and not all papers reported statistical significance.

### **3. Results**
