The Impact of COVID-19 Pandemic on Inequity in Routine Childhood Vaccination Coverage: A Systematic Review

Background: Routine childhood vaccination coverage rates fell in many countries during the COVID-19 pandemic, but the impact of inequity on coverage is unknown. Methods: We synthesised evidence on inequities in routine childhood vaccination coverage (PROSPERO, CRD 42021257431). Studies reporting empirical data on routine vaccination coverage in children 0–18 years old during the COVID-19 pandemic by equity stratifiers were systematically reviewed. Nine electronic databases were searched between 1 January 2020 and 18 January 2022. The risk of bias was assessed using the Newcastle-Ottawa Quality Assessment Tool for Cohort Studies. Overall, 91 of 1453 studies were selected for full paper review, and thirteen met the inclusion criteria. Results: The narrative synthesis found moderate evidence for inequity in reducing the vaccination coverage of children during COVID-19 lockdowns and moderately strong evidence for an increase in inequity compared with pre-pandemic months (before March 2020). Two studies reported higher rates of inequity among children aged less than one year, and one showed higher inequity rates in middle- compared with high-income countries. Conclusions: Evidence from a limited number of studies shows the effect of the pandemic on vaccine coverage inequity. Research from more countries is required to assess the global effect on inequity in coverage.


Introduction
The measures taken by governments to suppress the COVID-19 pandemic, including restrictions on movements and lockdowns, have resulted in a reduction in routine childhood vaccination coverage in countries worldwide [1]. It is estimated that from January to December 2020, 30.0 million (27.6-33.1) children missed the third dose of diphtheria, tetanus and pertussis vaccine (DTP3), and 27.2 million (23.4-32.5) children missed the first dose of measles-containing vaccine (MCV1). Although coverage rates have recovered in many countries, millions of children remain unprotected against diseases such as measles [2], and recent improvements in vaccination coverage are in jeopardy [3].

Methods
The search strategy for published and pre-print literature was conducted by a health sciences librarian at the University of Warwick Library (SJ) to identify the published studies that reported data on inequity in routine childhood vaccination coverage during the COVID-19 pandemic. The review protocol was registered with PROSPERO (CRD 42021257431).

Literature Search
Eight online databases were searched, including MEDLINE, EMBASE, Web of Science, Cochrane Central Register of Controlled Trials (CENTRAL), Cochrane CDSR, Sociological Abstracts, ASSIA, and MEDRxiv (initial search only as indexed in PubMed from February 2020), with no language restrictions, initially from 1 January 2020 to 30 April 2021 and updated from 1 May to 18 January 2022. The WHO Global Research on Corona Virus (COVID19) was also consulted (https://www.who.int/emergencies/diseases/novelcoronavirus-2019/global-research-on-novel-coronavirus-2019-ncov accessed initially on 30 April 2021 and updated on 18 January 2022). Full details of the search strategies for each database are available in Supplementary Materials File S1. The following exploded MeSH terms were used in the search: Coronavirus or Coronavirus Infections; Vaccines/or vaccin*.mp. or exp Vaccination or Immunisation Programs/or exp Immunisation/or immunisation.mp.; inequality.mp. or exp Socioeconomic Factors or Poverty/or poverty.mp. or healthcare disparities.mp. or exp Healthcare Disparities or Income/or income.mp. or Social Class/or social status.mp. or social class.mp. or Educational Status/or educat*.mp. or Health Status Disparities/or exp Health Status or Health Services Accessibility; Child/or child *.mp. or infant *.mp. or exp Infant.
Studies were included in the review if they reported data by equity stratifiers on routine vaccination coverage in child populations (0-18 years) within or between HICs and LMICs during the COVID-19 pandemic. Equity stratifiers were defined as individual or area characteristics representing socioeconomic factors, wealth, indigenous status, or racial/ethnic identity that can identify population subgroups at risk of suffering from healthcare disparities [11].
Excluded studies included those that reported on adults in which data on children were not considered separately, inequity in vaccination coverage but not related to the

Study Selection
The records identified by the search were deduplicated. The initial screening of the papers for eligibility criteria and selection of studies for full paper review was conducted in duplicate by four pairs of researchers (NS and WM; RN and NS; GG and NH; MLM and CJ) working independently. Studies selected for full paper review were allocated equally to the same four pairs of researchers who independently assessed whether studies met the inclusion criteria. Disagreements were resolved by consensus and, where necessary, reviewed by another research group member.

Data Extraction and Synthesis
Data from studies selected for full paper review, including study design, population and sample size, and vaccines and the equity stratifiers studied, were extracted in Excel using a pre-developed data extraction template (see Supplementary Table S1). Study authors were approached for further data where necessary.
A meta-analysis was likely to have been unreliable as the studies originated from countries with a wide range of health systems that use different effect measures and equity stratifiers. Consequently, we undertook a narrative review using the Synthesis Without Meta-analysis (SWIM) reporting guidelines [12]. Standard measures of effect (frequencies with 95% confidence interval (CI), unadjusted and adjusted risk and odds ratios (ARR/AOR)) were extracted for each study (where reported). In the presentation and interpretation of the results of the narrative synthesis, the risk of bias assessment and the social, economic and cultural differences between studied populations were considered.
We sought data on the following five outcomes of interest related to routine childhood vaccination coverage during the COVID-19 pandemic: Outcome 1-change in vaccine coverage by equity stratifiers during the pandemic; Outcome 2-change in inequity in vaccine coverage during the pandemic compared with pre-pandemic period; Outcome 3differences in vaccine-specific inequity in coverage; Outcome 4-differences in age-groups' specific inequity in coverage; and Outcome 5-differences in inequity in coverage between HICs and LMICs. The synthesis did not include data by region for which no comparative socioeconomic information was reported. Data on outcomes were sought in all included papers.

Risk of Bias Assessment
Each full-text article that met the criteria for inclusion was independently assessed for risk of bias by two reviewers using the Newcastle-Ottawa Quality Assessment Tool for Cohort Studies [13]. This tool includes the assessment of sample selection (representativeness; exposed and non-exposed from the same population; method of ascertainment of a social group; changes in coverage during a pandemic), compatibility (adjustment for confounding variables) and outcome (method of ascertainment of coverage; adequate period of follow up; adequacy of study population followed up). Two researchers (NH and NS) recorded their assessment independently in Excel. Stars (*) indicate the quality requirements for individual components of the domains have been met, i.e., a higher number of stars = lower risk of bias. Differences were resolved by consensus.

Results
After deduplication, 1453 studies were identified, of which 91 were screened for full paper review, and 13 met the inclusion criteria. Details of the identification and selection of studies are shown in a PRISMA flow diagram ( Figure 1). The study from Karachi [10] was excluded as the reported data were incorporated in the study covering the whole Sindh province [14], of which Karachi is the capital city. Table 1 shows the background characteristics of the included studies, and Table 2 presents the impact of equity stratifiers on childhood vaccination coverage. All the included studies reported on the COVID lockdown in their respective countries during the first wave between March and September 2020. Six studies reported data for a pre-lockdown period in addition to a lockdown period [14,[20][21][22][23][24]. The included studies used several approaches to study inequities in childhood vaccination inequities. Country-level income categories as defined by the World Bank (https: //data.worldbank.org/ accessed on 3 May 2022) to study variations in routine vaccination rates were reported for child populations in two studies for low-income countries (studies from Ethiopia [25,26]), three for lower-middle-income countries (studies from India [20] and Pakistan [14,16]), three for upper-middle-income countries (studies from Brazil [27], China [28] and Colombia [21]), four for high-income countries (three US studies [22][23][24] and UK [15]) and in a study comparing nine middle-income countries with 16 high-income countries [29].
The included studies used seven categories of equity stratifiers, which are summarised in Figure 2. Coverage change during the pandemic was reported for the education level of the responding parent in six studies [14,16,20,25,26,28]. Race/ethnicity was also reported [16,22,23], and employment [29] and wealth were measured at the country level based on World Bank criteria [29]. Further, the household level assessed by assets and building characteristics of the home [27] was used as an equity stratifier; of particular interest was the use of the Hindu system of low assets or low caste with its rigid hierarchical groups based on work and religion [20]. Rural versus urban dwelling was also used in two studies [14,21] and area-level measure of slum dwellings in one [14]. For example, in Pakistan and Colombia, rural areas are poorer than urban areas. In a study from the US, registration with Medicaid was used as an equity stratifier available only to low-income families [24]. included studies reported on the COVID lockdown in their respective countries during the first wave between March and September 2020. Six studies reported data for a prelockdown period in addition to a lockdown period [14,[20][21][22][23][24]. The included studies used several approaches to study inequities in childhood vaccination inequities. Country-level income categories as defined by the World Bank (https://data.worldbank.org/ Accessed on 3 May 2022) to study variations in routine vaccination rates were reported for child populations in two studies for low-income countries (studies from Ethiopia [25,26]), three for lower-middle-income countries (studies from India [20] and Pakistan [14,16]), three for upper-middle-income countries (studies from Brazil [27], China [28] and Colombia [21]), four for high-income countries (three US studies [22][23][24] and UK [15]) and in a study comparing nine middle-income countries with 16 high-income countries [29].
The included studies used seven categories of equity stratifiers, which are summarised in Figure 2. Coverage change during the pandemic was reported for the education level of the responding parent in six studies [14,16,20,25,26,28]. Race/ethnicity was also reported [16,22,23], and employment [29] and wealth were measured at the country level based on World Bank criteria [29]. Further, the household level assessed by assets and building characteristics of the home [27] was used as an equity stratifier; of particular interest was the use of the Hindu system of low assets or low caste with its rigid hierarchical groups based on work and religion [20]. Rural versus urban dwelling was also used in two studies [14,21] and area-level measure of slum dwellings in one [14]. For example, in Pakistan and Colombia, rural areas are poorer than urban areas. In a study from the US, registration with Medicaid was used as an equity stratifier available only to low-income families [24]. All included studies reported data on coverage by one or more equity stratifiers; however, the populations studied, data collection methods employed, and equity stratifiers varied widely, making a meta-analysis unreliable (see Tables 1 and 2). All included studies reported data on coverage by one or more equity stratifiers; however, the populations studied, data collection methods employed, and equity stratifiers varied widely, making a meta-analysis unreliable (see Tables 1 and 2). May 2020 (partially exposed); 796 aged 9 months March to May (heavily exposed); 183 aged 9 months in June-July (postexposed)) children born in or after January 2019 and at least 12 months of age at the time of the survey    lockdown (data only for the heavily exposed group (796) and unexposed group (443)): Analysis 1: Percentage point (PP) differences in fully immunised rates between exposed groups by equity stratifiers (95% CIs) Analysis 2: PP differences in equity stratifier difference (95% CI) Assets: PP difference between exposed groups by education difference = −7.   Table 3 details the risk of bias assessment of the included studies by the Newcastle-Ottawa scale [13]. The risk of bias was lowest for three of the included studies [14,25,27], with maximum stars for all three scale domains (8 stars). Eight studies [15,[20][21][22]24,26,28,29] had a moderate risk of bias (7 or 6 stars) and two studies [16,23] had a high risk of bias (<6 stars).

A Narrative Synthesis of Outcomes
In the following, a narrative synthesis of the five different outcomes of this systematic review is provided.

Narrative Synthesis of Outcome 1: Change in Routine Childhood Vaccination during the COVID-19 Pandemic by Equity Stratifiers
Seven studies [15,16,[25][26][27][28][29] reported coverage by the equity stratifiers during the pandemic period alone. A single study [14] reported separate analyses for the pandemic period alone and a comparison of pre-pandemic with pandemic coverage (see narrative synthesis of outcome 2).

Parental Education
Chandir et al. [14] reported a social gradient for receiving the age-appropriate third dose of Pentavalent vaccine by maternal education. The authors used the lowest maternal education category, 'no years of education', as the reference category. Contrary to the usual convention, an adjusted relative risk (ARR) greater than unity represented an increased likelihood of age-appropriate completion of the third dose of the Pentavalent vaccine. The ARR by the highest maternal education category (12+ years of schooling) was 1.38 (95% CI 1.23-1.55), adjusted for child's sex, place of birth, Penta 2 vaccination by outreach, outreach vaccination history, age at Penta 2 and age at Bacille Calmette-Guérin (BCG) and rural dwelling.
Completion of the vaccine schedule (see Table 2) by maternal education was reported by Miretu et al. [25]. Although AORs for all maternal education categories compared to the reference category (cannot read or write) were greater than 2, confidence intervals in the two highest categories (Grade 9-12 and College/University) crossed unity.
Tegegne et al. [26] also reported a social gradient in incomplete vaccination by maternal education level. An adjustment for marital status, place of delivery, waiting time at health facility and means of transportation to health facility increased the gradient. Compared with the highest maternal education category (diploma, degree or above), the AORs for complete vaccination were greater than 2 in all other education categories.
Rizwan et al. [16] reported a significant association of frequency of missed vaccination during the pandemic lockdown period in Pakistan with fathers' but not mothers' education (p values 0.006 and 0.082, respectively). The authors stated that the chi-square test was employed to estimate the p values but did not specify whether the estimates were for a linear trend.
The study from Wuhan and Shanghai in China [28], based on a survey of households with children aged 3-17, reported delayed vaccination during lockdown. It found a reverse positive social gradient with less educated parents less likely to have delayed vaccination than more educated parents (AOR 0.30 (0.15,0.59)).
Silveira et al. [27] reported a social gradient in the proportion (95% CI) of missed vaccinations based on parental reports and vaccination cards by wealth quintile. According to parental report, the proportions were 22.5% (95% CI 19.3, 26.2) of the children in the poorest quintile compared with 15.0% (11.6, 19.1) of the children from the wealthiest quintile and according to vaccination card record, the percentages were 24.5% (20.6, 28.8) compared with 15.6% for the same respective categories (95% CI 11.3, 21.1).
Low-income households in Rizwan et al.'s study [16] were more likely than higher income households to report missed vaccinations for children <2 years of age (p = 0.023). Bell et al. [15] also reported that the likelihood of overdue scheduled childhood vaccinations among low income and black and ethnic minority households was greater (AOR 1.24 and 2.15, respectively), but confidence intervals crossed unity. Shapiro et al. [29] reported on national panel studies that a polling company conducted in nine middle-income countries and sixteen high-income countries. They found that missed routine childhood vaccinations were significantly more likely in both middle-income (AOR 1.38 (1.14, 1.67)) and highincome (AOR 1.78 (1.26, 2.25)) countries in households with respondents in full-or part-time employment compared with unemployed, retired or students.

Narrative Synthesis of Outcome 2: Change in Inequity in Routine Childhood Vaccination Coverage during the COVID-19 Pandemic Compared with Baseline
Six studies [14,[20][21][22][23][24] included coverage data during the COVID-19 pandemic lockdown compared with a previous period. Chandir et al. [14] studied the percentage difference in enrolment in Sindh Province, Pakistan, during the COVID-19 lockdown (23 March to 9 May 2020) compared with the pre-lockdown period (23 September 2019, to 22 March 2020) by maternal education. They found a statistically significant difference in the percentage of enrolment between mothers with low education and those with higher education ( Table 2). They also reported a percentage reduction in vaccination visits during the lockdown period compared with the baseline by area-based equity stratifiers. Visits reduced in rural areas by 54.9% compared with the 47.5% reduction in urban areas, in slum areas by 53.8% compared with 51.3% in non-slum areas and by 68.1% in Special High-Risk Union Councils (SHRUC), designated as polio-endemic super high-risk sub-districts, compared with 50.6% in non-SHRUCs. No probability measures were reported with these findings.
Jain et al. [20] reported the results of a study conducted in Rajasthan, India. They compared percentage point (PP) differences (95% CI) in fully vaccinated rates in the group reaching nine months of age during the COVID-19 lockdown (March to May 2020) with the group aged >12 months before March 2020. In this comparison, they used three equity stratifiers, i.e., assets, caste, and parental education ( Table 2). There were significant reductions in fully vaccinated rates in both high and low caste groups but only in low assets and low parental education groups. When PP differences between the group exposed to the lockdown period and those not exposed were examined by high v. low levels for each equity stratifier, there were no significant differences.
Moreno-Montoya et al. [21], based on Colombian national administrative vaccination data, reported a more significant reduction in the coverage of individual vaccines in rural versus urban areas in the pandemic lockdown period (March to September 2020) compared with the pre-pandemic period (March to October 2019). Reductions in all scheduled vaccines were significantly higher in rural areas than urban ones among children aged <12 months and those aged 5 years. No significant differences were found among children 12-23 months of age (Table 2).
Bramer et al. [24] reported a reduction in Michigan state, USA, in routine vaccination enrolment in all age cohorts (1, 3, 5, 7, 16, 19 and 24 months) in May 2020 compared with May in the preceding three years 2016-2019 (Table 2). Up-to-date coverage of ageappropriate vaccines decreased in all age cohorts, except the cohort of infants aged less than one month. The coverage for each age cohort assessed in May 2020 was lower for Medicaid enrolled children than non-Medicaid enrolled children. The largest difference was found in the 7-month cohort, with 34.6% of Medicaid recipients compared with 55.0% non-Medicaid recipients up-to-date with age-appropriate vaccines (data for other age cohorts by Medicaid receipt was not reported in the paper). However, a statistical analysis of equity stratifiers was not reported.
Ackerson [23] reported on the percentage reduction in total routine vaccine doses administered to Californian children from 0 to 18 years in the 'stay-at-home' period of the pandemic (13 March to 6 May 2020) compared with the pre-pandemic period (1 January to 12 March 2020). The results show a significant gradient in the percentage reduction by race/ethnicity. The reduction was most significant among Non-Hispanic Blacks, followed by Non-Hispanic Whites and Hispanics, with the smallest reduction among Non-Hispanic Asian Americans (Table 2).
DeSilva et al. [22] reported the results of a study on eight health systems in six US states. They compared the rates of up-to-date scheduled vaccinations in February 2020 (prepandemic) with May 2020 (pandemic period) by race/ethnicity in five age groups (7 months, 18 months, 6 years, 13 years, and 18 years). Up-to-date vaccination rates at 7 months of age reduced significantly in all race/ethnic groups (see Table 2). They reported that at 18 months, non-significant reductions in rates were found in all race/ethnic groups, and, in the older age groups, there were no significant rate reductions. Rates among Black children aged 7 and 18 months were lower than those in other race/ethnic groups during both the pre-pandemic and pandemic periods.

Narrative Synthesis of Outcome 3: Differential Coverage for Specific Vaccines during the COVID-19 Pandemic by Equity Stratifiers
Two studies reported on the coverage of specific vaccines by equity stratifiers [14,21]. Only the study by Moreno-Montoya et al. [21] compared inequity in coverage rates for more than one specific vaccine. Effect sizes of absolute differences in pandemic and pre-pandemic coverage of individual vaccines were expressed as Beta coefficients (95% CI). They found that the difference between rural and urban areas varied in children <12 months of age, with the highest difference for BCG (OR 0.45 (0.22, 0.69)) and the lowest for Polio Vaccine (OPV or IPV) (OR 0.26 (0.06, 0.48)). However, absolute differences in coverage did not reach statistical significance at the 95% level for any of the individual vaccines in the three age groups studied.

Narrative Synthesis of Outcome 4: Differential Coverage in Age Groups during the COVID-19 Pandemic by Equity Stratifiers
DeSilva et al. [22], as reported above, showed inequity by race/ethnicity on up-to-date vaccinations during the pandemic among children at 7 and 18 months of age but not among older age groups (6, 13 and 18 years). Effect sizes of absolute differences during the pandemic compared with pre-pandemic coverage rates for scheduled vaccines by rural versus urban dwellings in Colombia [21] were higher among children <12 months of age than those aged 12-24 months and those aged 5 years.
Bramer et al. [24] reported higher rates of coverage reduction in May 2020 compared with May 2016-19 among Medicaid enrolled compared with non-Medicaid enrolled children in all age cohorts, except for those one month of age, and the difference in the rate reduction was highest in the 7-month cohort. The remaining studies do not report comparative vaccination coverage data by age group. The studies reporting on children 0-2 years of age all report social gradients in vaccination coverage in the same direction.  [29] were more likely than those in HICs to report missed routine childhood vaccinations during the pandemic (7.6% v. 3.0%; p < 0.05). Their study is the only one to directly compare vaccination coverage between HICs and LMICs during the pandemic. A direct comparison of the magnitude of inequity between HICs and LMICs in the remaining included studies cannot be made due to the wide variety of equity stratifiers used. However, inequity in routine vaccination coverage was reported in all four country income groups as defined by the World Bank (low-, lower-middle-, upper-middle-and high-income countries).

Outcome 1: Change in Routine Childhood Vaccination Coverage during the COVID-19 Pandemic by Equity Stratifiers
Of the studies reporting on routine childhood vaccination coverage rates during the pandemic period alone or separately from the pre-pandemic period, six [14][15][16][25][26][27] reported lower rates among children in disadvantaged population groups. Based on the risk of bias assessments (see above and Table 3), the strongest evidence of inequity among the included studies comes from Chandir et al.'s low-risk-of-bias study [14]. Although the remaining five studies [15,16,[25][26][27] reported inequity to be consistent with Chandir et al.'s [14] findings, the strength of the evidence is limited by the higher risk of bias, from moderate [15,25,26] to high [16,23].
Contrary to the findings from the above studies, the studies of Hou et al. [24] and Shapiro [29], both with a moderate risk of bias, reported higher rates of missed or delayed vaccination schedules during the pandemic among children from more advantaged households measured by education [24] and employment status [29].
Studies comparing coverage in pre-pandemic and pandemic periods are considered in the next section; however, all six studies, with low [14,21,23] and moderate [20,22,24] risk of bias, show evidence of inequity in coverage reduction during the pandemic.
In summary, the strength of the evidence for inequity in vaccination coverage during the pandemic lockdown is moderate.

Outcome 2: Change in Inequity in Routine Childhood Vaccination Coverage during the COVID-19 Pandemic Compared with Baseline
Three low-risk-of-bias studies reported evidence of increased inequity during the pandemic lockdown months compared with pre-pandemic months [14,21,23]. Robust evidence of increased inequity between rural compared with urban dwellers in individual vaccine coverage during the pandemic period was reported by Moreno-Montoya et al. [21]. Ackerson et al. [23] also reported robust evidence of increased inequity by race/ethnicity during the pandemic compared with the pre-pandemic period. Chandir et al. [14] showed differences by education and by area-level equity stratifiers in a reduction in vaccine enrolment during the pandemic compared with baseline; however, the evidence was less robust as the analysis was unadjusted for confounding effects.
Two moderate risk-of-bias studies [22,24] provided weak evidence of an increase in inequity in the pandemic compared with the pre-pandemic period and the same month in the three preceding years. An increase in inequity in up-to-date vaccinations in May 2020 compared with the same month in the three preceding years was reported by Bramer et al. [24]; however, percentage reductions were reported for a single age cohort but without probability measures or adjustment for confounding. DeSilva et al. [22] compared up-to-date vaccination rates in May 2020 with those in February 2020 by race/ethnicity. Rates were reduced in all ethnic groups among children who were 7-months old. However, in the absence of a comparative analysis of differences in reduction by ethnic group, there was no clear evidence of an increase in inequity.
Jain et al.'s higher-risk-of-bias study [20] reported statistically significant percentage point differences in full immunisation rates during the lockdown months compared with pre-pandemic months by several equity stratifiers. These included separate high and low equity asset, caste, and education categories. However, percentage point differences between high and low equity stratifier categories during the pandemic lockdown months compared to those prior were not statistically significant. The results suggest that the reductions occurred equally across equity stratifiers, and inequity did not increase.
In summary, although two moderate risk-of-bias studies [20,22] failed to show evidence of increasing inequity during the pandemic, three low-risk-of-bias studies [14,21,23] report moderately strong evidence of increasing inequity, with additional, but weak, supporting evidence from Bramer et al. [24].

Outcome 3: Differential Coverage for Individual Vaccines during the COVID-19 Pandemic by Equity Stratifiers
We identified one low-risk-of-bias study [21], which reported coverage of individual vaccines by equity stratifiers during the pandemic.
In summary, evidence for differential inequity is not supported as differences for individual vaccines by rural versus urban dwellings failed to reach statistical significance.

Outcome 4: Differential Coverage in Age Groups during the COVID-19 Pandemic by Equity Stratifiers
Coverage rates during the pandemic in different age groups by equity stratifiers are reported by one low [21] and two moderate [22,24] risk-of-bias studies. The evidence suggests that inequity during the pandemic was more likely to affect routine vaccination in children of less than 12 months of age.
In summary, the small number of studies reporting inequity by age group and differences in the age groups studied make this evidence inconclusive.

Outcome 5: Inequity in Coverage between HICs and LMICs during the Pandemic
We identified a single moderate risk of bias study [29] reporting empirical evidence on inequity in routine childhood vaccination coverage during the pandemic among families in middle-compared with high-income countries. The study shows higher rates of missed childhood vaccinations in middle-income countries.
In summary, data from a single study represent weak evidence for this outcome.

Discussion
Our systematic review sought evidence of the impact of measures to control the spread of the SARS-CoV-2, such as lockdowns, on routine childhood vaccination coverage inequity. Our literature search from 1 January 2020 to 18 January 2022 identified 13 studies with data on coverage by equity stratifiers. By means of a narrative synthesis, we found moderate evidence for inequity in reducing the vaccination coverage of children and adolescents during lockdowns between March and September 2020 and moderately strong evidence for an increase in inequity compared with pre-pandemic months. While the review found evidence of greater inequity in coverage among children less than 18 months old compared with other age groups and inequity between MICs and HICs, there were too few studies to judge the strength of the evidence, which remains inconclusive.
Inequity is a crucial driver of sub-optimal levels of routine vaccination coverage [4,30]. The small number of studies from nine individual countries and a group of 25 countries published in the first 21 months of the pandemic suggests limited research on inequity's contribution to the reduction in routine vaccination coverage internationally [1,30,31]. WHO [4] estimated the potential for improvement in vaccination coverage by eliminating economic-related inequity in 23 priority study countries (see Figure 3.15, p. 35 in the report). It was concluded that all 23 countries would achieve higher national rates by eliminating economic-related inequity. The most significant potential for improvement was reported in Nigeria, where the national rate of DTP3 vaccination coverage would improve by 40 percentage points by eliminating economic-related inequity. Similarly, Ethiopia and Pakistan, two countries represented in our review, would improve national rates by over 20% by eliminating inequity. In addition, target 3.B.1 of the SDGs, access to affordable essential medicines and vaccines for all by 2030, measured by the indicator "proportion of the target population covered by all vaccines included in their national program" [31], will only be achieved by eliminating inequity, including that resulting from the pandemic.
The reduction in coverage among the most vulnerable children in the pandemic, as reported in the two studies from Pakistan included in the review [14,16], is likely to impede its progress in achieving the targets set by the Global Vaccine Action Plan (GAVP) and the Immunization Agenda 2030 [32]. Pakistan was identified in the Global Burden of Disease study [32] as one of the countries with a high level of 'zero-dose' children, defined as those not receiving any doses of the diphtheria-tetanus-pertussis vaccine. The pandemic is likely to have added to this number among children of mothers with low levels of education and in low-income households. Similar reductions in Ethiopia, reported in the two studies [25,26] included in the review, are likely to affect countrywide vaccination coverage and GAVP targets. Nonetheless, it is difficult to gauge the probable extent of the reduction as the numbers in the lowest education group (cannot read or write) in both studies are very low.
A Brazilian study [27] reported that over 20% of children in the two poorest quintiles missed vaccinations during the pandemic compared with 17%, 17.5% and 15% in higher income quintiles. If this level of inequity occurred throughout the country, it would likely have made the GAVP targets more difficult to achieve. Access to routine vaccination for children in rural areas in LMICs is known to be more difficult than for those in urban areas [4]. The study from Colombia [21] indicates that the pandemic is likely to have increased accessibility problems for rural families, especially among young children.
Pre-pandemic vaccination coverage rates among Black children and those on Medicaid in the USA [33] were low compared with other ethnic groups and non-Medicaid enrolled children. The studies by Ackerson et al. [23] and Bramer et al. [24] suggest that the vaccination coverage in these groups is likely to have further reduced due to the pandemic increasing existing inequities.
The review has the following strengths. The review protocol was registered with PROSPERO, and the search strategy was designed and conducted by a health sciences librarian. We searched nine electronic databases comprehensively, including MedRxiv, the pre-print database, and the WHO Global Research on Corona Virus (COVID19) database. We included all studies without any language restrictions. In addition, the search was up to date till January 2022. PRISMA 2020 guidelines [34] were adhered to when screening abstracts, selecting studies for full paper review and identifying studies for inclusion (see Supplementary Table S2).
Our review has some limitations. We did not undertake a grey literature search and might have missed some data. However, we aimed to include only published data. A meta-analysis was not feasible due to widely differing health systems, outcome and effect size measures and equity stratifiers; however, we followed the standard SWiM guidelines for the narrative synthesis and provided a comprehensive review of the available data [12].
Measures to control the pandemic have disproportionately affected children in poor households and marginalised and isolated communities in LMICs [5] and HICs [6]. They have also disrupted their access to healthcare [35,36]. A UN Development Programme study [37] on the impact of COVID-19 on Sustainable Development Goals found that 44 million people, many of whom will be children, are expected to be pushed into extreme poverty by 2030 due to COVID-19. Applying an equity lens to the study of the impact of the pandemic on routine vaccination coverage is essential to identify children who have missed vaccinations and those with no previous access to vaccination programmes [35]. Our review found moderate evidence of equity-stratifier-specific inequity in routine childhood vaccination coverage during the pandemic and more robust evidence of an increase in equity-stratifier-specific inequity in the pandemic compared with the pre-pandemic period. Evidence of differential inequity by specific vaccines and child age groups was inconclusive. However, the evidence was confined to only thirteen eligible studies from a small number of countries. Before the pandemic, the WHO recommended expanded health-inequality monitoring, especially in low-income countries, to reduce inequity in childhood immunisation and inform equity-oriented programmes aimed at reaching the most disadvantaged population sub-groups [4]. The findings of this review indicate that inequity is likely to have increased due to the pandemic. However, more extensive and robust evidence is needed to accurately identify the extent of the pandemic's short, medium and long-term effects on childhood vaccination inequity. Methodologically robust research covering a wide range of LMICs and HICs using routine vaccination coverage data is available for many countries [1,32]. These include household surveys, such as Demographic and Health Surveys (DHS) [38], Multiple Indicator Cluster Surveys (MICS) [39], and administrative data collection systems. There is an urgent need to synthesise the current evidence and inform interventions to reverse the pandemic's effects on childhood vaccination inequity.