**4. Discussion**

In this scoping review, we assessed the current state of research pertaining to withincountry inequalities in COVID-19 vaccination coverage. Our findings show that this body of research covers a diversity of populations and settings, suggesting a wide interest in assessing and understanding the patterns of vaccination coverage inequality across populations. The geographical representation of study settings within this literature favoured high-income countries in North America and Europe—for example, half of articles were

conducted in the United States, with only five articles based on populations in the African continent. High-income countries accounted for more than half of the countries represented in this body of literature, whereas low-income countries accounted for less than 10%. This finding was not surprising, as populations in high-income countries tended to have earlier access to vaccines than lower-income countries, and thus implemented vaccine programmes sooner; moreover, timelines, access and incentives for publishing in academic journals may differ between settings. Nevertheless, more research on inequalities in COVID-19 vaccination coverage is warranted in lower-income countries, particularly as vaccines become more widely available in these settings.

We found that demographic factors, including age, race/ethnicity/cultural group/ language/nationality/country of birth and sex/gender, were the most commonly reported dimensions of inequality in vaccine coverage rates in this body of literature. Indeed, early into the COVID-19 pandemic, the scientific community made strong calls to enhance the collection and reporting of data disaggregated by these factors [43–46]. In the cases of age and sex/gender, the application of similar measurement criteria (years and male/female, respectively) allowed us to comment on the general trends in the directionality of inequality reported in these articles. Our preliminary assessment of this literature suggested that vaccination coverage tended to be higher among (relatively) older population groups, across many age ranges. This is in line with WHO guidance [6], suggesting the implementation of vaccine rollout strategies that initially prioritised older age groups. With regards to findings on sex/gender-related inequality, a substantial proportion of articles that reported on this dimension concluded that there were no meaningful differences. Of those articles that did report a difference, the directionality was variable, with vaccination coverage more often reported to be higher in males than in females. It was not feasible to do even preliminary comparisons of findings for the race/ethnicity/cultural group/language/nationality/country of birth dimension of inequality, as the measurement of this dimension is context specific (i.e., not standardised across settings). There were, however, common approaches applied within particular country settings, which could be explored through more narrowly-focused systematic reviews and/or meta-analyses. Indeed, more rigorous meta-analyses, including quality assessments, are warranted to delve into vaccination coverage inequalities by demographic factors.

Among the other dimensions of inequality highlighted in our scoping review were those related to socioeconomic factors (most frequently occupation/employment, education level, economic status and vulnerability, deprivation or poverty indices) and those related to geographical factors (most frequently subnational region/area and place of residence). Characterising patterns of socioeconomic inequality offers deeper understanding into the motivations and barriers experienced by population groups, while geographic patterns of inequality may have immediate and practical implications for program delivery and resource allocation [47]. Other dimensions of inequality, such as sexual orientation, social capital, and migration status, received less attention in the research, although some of these articles provide initial indications that these may be meaningful avenues for future study. For instance, the three articles that reported inequality in COVID-19 vaccination coverage related to a measure of social capital all reported higher vaccination among groups with greater social capital [20,48,49].

The application of multiple disaggregation in one-third of studies permitted exploration of the intersection of different dimensions of inequality. Multiple disaggregation can begin to lend insight into more nuanced patterns of inequality, for example, suggesting how multiple vulnerabilities may put certain groups at heightened risk for lower vaccination coverage [50]. Multiple disaggregation should be incorporated, to the extent possible, in future inequality analyses in this topic [51].

We reported variability in how dimensions of inequality were measured, reflecting diverse study populations, settings and research aims. In some cases, standardised criteria were applied within a country (such as race/ethnicity categories in the United States), enhancing comparability across these studies. For most dimensions of inequality, however, the lack of standardised criteria for measuring dimensions of inequality limits the extent to which direct comparisons of inequality can be made across studies and across settings.

Our scoping review extends on a previous review by Bayati et al. (2022), which had a broader aim of assessing both between-country and within-country factors associated with COVID-19 vaccine distribution [11]. The portion of the review focused on withincountry factors included 19 studies, and concluded that "age, race, ethnic, household income, residency in the deprived areas, employment, poverty, location (urban/rural) and gender were most often mentioned in the literature". Our scoping review, encompassing 167 studies, highlights additional dimensions of inequality that have been explored in the literature, including education level, indices of vulnerability, deprivation or poverty, marital status and family characteristics. Additionally, it provides a more detailed overview of the study settings, data sources, reporting practices and preliminary findings.

The findings of this scoping review are broadly in line with previous reviews exploring inequalities in COVID-19 vaccination intentions and attitudes [52,53]. For instance, the directionality of the age- and sex/gender-related inequality that we reported corresponds with those reported in a meta-analysis on inequalities in COVID-19 vaccination intention, which included 28 nationally representative populations across 13 countries. It reported female sex, younger age and belonging to an ethnic minority group to be consistently associated with lower intention to vaccinate, highlighting "an urgent need to address social inequalities in vaccine hesitancy and promote widespread uptake of vaccines as they become available" [52]. Similarly, a meta-analysis including 63 surveys and more than 30 countries concluded that age, gender and education level were among the factors most often associated with willingness or hesitancy to be vaccinated [53]. We note, however, an important distinction between the body of research pertaining to vaccination coverage from research on vaccination intentions and attitudes. Attitudes towards vaccines have been found to shift over time [9], and do not directly translate into behaviours. A study of vaccination uptake during the 2009 influenza (H1N1) pandemic, for example, found that only a small percentage of those reporting a positive intention to vaccinate followed through on receiving the vaccine after two months [54].

### *Limitations and Further Considerations*

Our findings and their interpretations are subject to a number of limitations and considerations. We acknowledge that there is a bias in this body of literature towards settings where vaccines have been rolled out, studied and reported. Settings that lack reliable data collection about COVID-19 vaccinations are less likely to be represented in published academic literature, and therefore less likely to be included in this scoping review.

Across studies, approaches to defining COVID-19 vaccination coverage were not standardised. For the purpose of our scoping review, we adopted a broad definition for the COVID-19 vaccination coverage indicator and included studies reporting on receipt (or non-receipt) of a single dose, multiple doses and/or booster doses. Nearly one in five of the articles included in our review reported on more than one vaccination coverage indicator that met our inclusion criteria. The application of common definitions for COVID-19 vaccination coverage would facilitate greater cross-study comparability and more nuanced analyses.

We relied on the PROGRESS-Plus framework as a starting point to guide how we grouped and labelled dimensions of inequality. Alternate frameworks may have yielded different conclusions about the most frequently reported dimensions of inequality. We did not report political factors as relevant dimensions of inequality, although we noted that 13 of the 167 articles reported on inequalities based on political views or voting patterns. Of these studies, 11 were conducted in the United States, all of which found lower vaccination among Republican voters and/or higher vaccination among Democrat voters.

Our exploration of inequality trends for selected dimensions of inequality in this scoping review was premised on findings that may be of variable quality. Approaches and thresholds to determine the 'meaningfulness' of inequality were different across studies. More rigorous meta-analyses incorporating quality assessments are required as an extension of our initial findings.

As per the design of our scoping review, we did not account for how countries may have prioritised different populations during phased vaccine rollouts. Initially, COVID-19 vaccine doses were in limited supply and inequality during the early stages of their distribution was inevitable (though COVID-19 vaccination coverage equity remains an end goal for most countries) [10]. Many of the included articles, however, did take this into account in their study design. We did not focus on the reasons underlying vaccination status, such as whether population subgroups remained unvaccinated by choice (low acceptance of the vaccine) or their circumstance (low access to the vaccine). Explorations of the drivers of inequality were outside the scope of this review. We did not differentiate between studies conducted in general populations versus studies that evaluated a specific campaign or programme, which may have been targeted towards certain populations.
