Covid-19 vaccination among migrants in Rome, Italy

Migrants may be susceptible to vaccine barriers and hesitancy. We evaluated the association between migrant status, as measured by the citizenship from a High Migratory Pressure Country (HMPC), and COVID-19 vaccination uptake in the resident population in Rome, Italy. We also investigated sex differences. We followed participants for vaccination against COVID-19 in 2021. We calculated crude- and adjusted-vaccination rates and Cox hazard ratios of vaccination for migrants compared to Italians. Among migrants from HMPCs, we estimated HRs for females compared to males, stratifying by geographical area of origin. Models were adjusted for age and deprivation index and stratified by infection history. In 2021, among 1,731,832 18–64-year-olds, migrants were 55% less likely to uptake at least one COVID-19 vaccine dose than their Italian counterpart. Past SARS-CoV-2 infection reduced the difference between migrants and Italians to 27%. Among migrants from HMPCs, we observed a slight excess of vaccination uptake among females compared to males. Focusing on geographical areas, we observed that only females from central-western Asia were 9% less likely to uptake vaccination than males. Health communication strategies oriented to migrants and considering their different languages, cultures, and health literacy should be adopted for prevention before emergencies.


Descriptive data
Table 1 describes the main characteristics of the study population by migrant status (the primary exposure variable).People from HMPCs represented 16% of our population and were the youngest (mean age 41.6 years); the prevalence of females was higher among HMPCs (53%) and High Developed Countries (HDCs, 61%) compared to Italy (51%).The deprivation index (DI) indicated a higher percentage of disadvantaged adults among HMPCs (23%) compared to Italy (18%) and HDCs (12%).People from HDCs had the highest percentage of SARS-CoV-2 infection (48%).
Participants totalized 916,213 years of follow-up.Table 2 summarises follow-up times (days) by citizenship.On average, people were followed for 193 days.Median follow-up times among non-Italians (228 days for HDCs and 222 for HMPCs) were higher than among Italians (164 days).

Outcome data
Figure 1 shows the outcome data and reasons for censoring by citizenship.Out of 1,731,595 individuals analysed, Italians had the highest percentages of people vaccinated with at least one vaccine dose (88.3%), while non-Italians coming from HDCs showed the lowest percentages of vaccination uptake (51.7%).Reasons for censoring were due either to death for 1880 individuals (0.11%) or loss to follow-up for 17,043 individuals (0.98%), while 256,178 (14.8%) completed the follow-up until the end of the study.
Time to vaccination distributions were slightly right-skewed and showed a higher median time among people from HMPCs (181 days) compared to Italians (159 days) and people from HDCs (153 days).

Migrant status
We performed Cox regression models among people from HMPCs and Italians.The association between migrant status and vaccination (dependent variable) showed a lower vaccination hazard among migrants than Italians (HR = 0.440; 95%CI 0.438-0.442).Age and DI adjustment did not change the observed associations (HR = 0.451; 95%CI 0.448-0.453).
Table 5 shows unadjusted, age-, and age-DI-adjusted HRs stratified by past SARS-CoV-2 infection.Past infection was an effect modifier: among participants without past infection, migrants showed a lower hazard of being vaccinated compared to Italians (HR = 0.432; 95%CI 0.429-0.434)that got closer to the hazard for Italians among participants with past infection (HR = 0.733; 95%CI 0.719-0.747).The interaction term tested in the model was statistically significant (p < 0.0001).We did not find a relevant interaction between either sex or age and migrant status.www.nature.com/scientificreports/

Sex
We performed Cox regression models only among people from HMPCs.Overall, the association between sex and vaccination (dependent variable) showed a higher risk of being vaccinated among women than men (HR = 1.110; 95%CI 1.100-1.121).Age and DI were not relevant confounders of the observed associations (HR = 1.049; 95%CI 1.039-1.060).When stratified by past infection, such differences between females and males disappeared among those with a history of past infection (Table 6).The interaction term tested in the model was statistically significant (p < 0.01).
The focus on each geographical area of origin showed a different pattern among people of central-western Asia (where 86% of people came from Bangladesh, India, Sri Lanka, and Pakistan), with women having a lower age-DI-adjusted hazard of being vaccinated compared to men (HR = 0.911; 95%CI 0.890-0.932).We did not observe differences according to past infections.

Additional analysis
When we selected the participants, excluding those who had an infection before 27 December 2020 (n = 1,658,686), and we truncated the follow-up time at the date of infection for those who got infected in 2021 (n = 72,881), we found comparable results with those presented above: the migrant status was inversely associated to the COVID-19 vaccination (HR = 0.436; 95%CI 0.434-0.439)and women had a higher probability than men of getting the vaccine (HR = 1.051; 95%CI 1.040-1.062).The results were comparable for central-western Asian women (HR = 0.911; 95%CI 0.890-0.933).

Discussion
In 2021, among 1,731,832 residents in Rome aged 18-64, migrants from HMPCs were 55% less likely to uptake at least one COVID-19 vaccine dose than their Italian counterpart, independently of age and area deprivation index.Past SARS-CoV-2 infection reduced the difference between migrants and Italians to 27%, explained by an increase in vaccination uptake after the infection among migrants and a decrease among Italians.Among migrants from HMPCs, we observed a slight excess of vaccination uptake among females compared to males; while, focusing on geographical areas of origin we did observe that only females from central-western Asia were 9% less likely to uptake vaccination than males.The additional analysis showed comparable results.Some limitations should be considered.Suppose there were different patterns of vaccination uptake, inside or outside the region, between migrants and Italians.In that case, the observed associations might be biased as we accessed only data from Lazio (the region of Rome).Mobility among migrants may be higher than among non-migrants, which would cause an underestimate of vaccination coverage and bias the observed hazard ratios toward the lower bound.Nevertheless, we studied the resident population, and resident migrants are likely less prone to mobility than non-resident migrants.In addition, due to pandemic restrictions, it is likely that during 2021 mobility was reduced.Another limitation may be related to the record-linkage procedures that could be less efficient among the migrant population than Italians.This would again yield an underestimation of vaccination uptake among migrants.Concerning the adjustment for the deprivation index, based on 2011 Census data, any change in the social tissue that occurred over ten years might imply misclassification.In relation to other sources of confounding, we could adjust only for major factors (age, DI), while other factors, like www.nature.com/scientificreports/comorbidities, occupation, or marital status, might also play a role.Finally, contextual factors changing over time, such as fluctuating mobility restrictions, risk of infection, or policies oriented to mandatory vaccination, might directly or indirectly affect the associations.However, the interpretation of models that include calendar time, with cut-offs appropriate for each factor, would be complex.As such, we decided not to adjust or stratify for calendar time in the present study.
The results indicate the vulnerability of migrants residing in Rome concerning COVID-19 vaccination access in 2021 and suggest inequalities in health.Lower vaccination uptake has also been observed in an Italian study conducted in Lazio, the administrative region of Rome, which found that foreign residents have a triple probability of Italians not accessing the vaccine 12 , and in a study conducted in a Local Health Authority in Rome 13 .Furthermore, a lower vaccination coverage among foreigners has also been observed in a study conducted in Brescia Province, one of the most dramatically hit by COVID-19 at the beginning of the pandemic 15 .To interpret the findings, we should consider that COVID cases among migrants might have different characteristics affecting the risk of infection and the need to be vaccinated.Among these, later diagnoses and poorer outcomes in COVID cases among foreigners compared to Italians were reported 10,11 .In the study conducted by a Local Health Authority in Rome, HMPC citizens were younger than Italians, less likely to be frail and more likely to receive the less effective brand of vaccine (Janssen) 13 .However, we do not have striking evidence of different risks of infection among immigrants in Italy, though it is possible to argue that the lack of findings is the consequence of the lower access to diagnostic tests 16,17 .Concerning the young age structure of migrants, in our study, we selected 18-64-year-olds, reducing the age gap between Italians and immigrants (mean age 43.8 vs 41.6, Table 1) and, as such, differences in frailty.At the international level, several studies analysed the association between ethnicity and vaccination uptake, using different study designs and measures from our own.For example, in a study conducted over 24 million adults in England, the first dose of COVID-19 vaccination was lower among all ethnic minority groups compared with white British adults 18 .Another study conducted in Denmark over 4.9 million individuals aged 12 years or more in 2021 found that non-vaccination was most pronounced among migrants or descendants 19 .In contrast, a study conducted in Switzerland did not find an association between Swiss-born and foreign-born individuals 20 .We also observed that women from central-western Asia showed lower vaccination coverage than men.In this area, the most prevalent origin countries of subjects living in Rome are Bangladesh, India, Sri Lanka, and Pakistan, and most subjects are males.Most central-western Asian women typically come to Italy for family reunification.They are often unemployed and have few social contacts 21 .All these aspects may partly explain the lower vaccination coverage among women in this subgroup of migrants.
Various factors in the literature have been identified as possible explanations for low vaccine uptake and hesitancy, for example, the "delay in acceptance or refusal of vaccines despite availability of vaccination services" 22 among migrants.A systematic review exploring barriers and facilitators of vaccine uptake has identified language, literacy, communication, practical, legal, and service barriers in the uptake of vaccines 1 .In our country, access barriers to health services have been identified for migrants 23 .These barriers also played a role in the vaccination uptake, especially at the beginning of the vaccination campaign.Later, on 15 October 2021, the vaccine became mandatory for all people over 50 years and for occupied people.Then, the possession of a green pass, a document certifying the vaccination, was imposed at work to demonstrate full vaccination coverage 24 .
According to the 3C model on vaccine hesitancy developed by the SAGE Working Group, three main factors influence vaccine uptake: confidence, complacency, and convenience barriers 22 .In a recent systematic review performed to synthetise qualitative studies on the reasons for vaccine hesitancy among migrants, the Authors found the "confidence" dimension of the 3C model, that is, "…people are vaccine hesitant because they have low confidence in the vaccine's effectiveness and safety and distrust scientists, policymakers and health profession-als…" 22 , represents a disproportionately large barrier to vaccine uptake in ethnic minority groups 25 .We argue that the confidence dimension may explain vaccine hesitancy among the migrants in Italy because communication during the vaccination campaign was challenging due to linguistic barriers and the different health literacy of migrants compared to Italians, despite some communication strategies adopted in the country 26 .Health literacy may be associated with vaccination, although evidence is scarce 27 .In addition, the "convenience" dimension, that is "…people are vaccine hesitant because there are a number of barriers (physical, logistical or economical) that hinder them from getting a vaccine…" 22 , may have represented another important explanation for the vaccine hesitancy among migrants in Italy.In fact, it is already documented that migrants encounter, in Italy 23,28 as in other European countries 29 , various barriers in accessing health services that during the pandemic may have represented a critical issue for COVID-19 prevention 30,31 .In addition, we argue that since migrants in Italy are often employed in temporary and precarious jobs 32,33 , their intention to be vaccinated may be undermined by the fear of possible vaccine collateral effects limiting their chance to work.The ECDC suggests various approaches to strengthening vaccine uptake in migrants.Some of them may be particularly relevant in Italy and adopted, such as the provision of "…simple, accurate culturally-relevant resources about the COVID-19 vaccine in a range of languages, literacy levels and formats…" and the provision of cultural mediators in primary care 34 .In addition, the availability of data stratified by origin country and other relevant factors, such as gender and socioeconomic status, is of paramount relevance as it allows the calculation of immunisation indicators across subgroups of the population 35 and highlights the unmet prevention needs of vulnerable groups.For the generalizability of results, although our evaluation covers the population living in a single city only, the findings can, in part, be indicative of differences in vaccination coverage between Italians and migrants in our country, as the study was based on a vast city (2.7 million inhabitants), and a whole population, and access to vaccination was offered to all residents, both Italians and non-Italians, without restrictions in all the Italian regions.
In conclusion, migrants residing in Rome, Italy, showed a lower uptake of COVID-19 vaccination over the first year of the vaccination campaign, independently of socioeconomic factors.Vaccination uptake was lower among migrant women from central-western Asia than migrant males.Health communication strategies oriented to migrants and considering their different languages, cultures, and health literacy, as well as the possible www.nature.com/scientificreports/(national legislative decree on privacy policy no.196/30 June 2003).The Department of Epidemiology of the Lazio Regional Health Service is the regional referral center for epidemiological research, and it has full access to anonymized health information systems.We declare that all methods were carried out in accordance with relevant guidelines and regulations.

Table 1 .
Characteristics of the resident population by citizenship.Rome, Italy, December 2020.

Table 2 .
Follow-up time of study participants by citizenship.

Table 3 .
Crude and adjusted vaccination rates (per person-year) by citizenship and past SARS-CoV-2 infection.

Table 4 .
Crude and adjusted vaccination rates (per person-year) for people of HMPCs by sex and past SARS-CoV-2 infection.

Table 6 .
Association between sex and vaccination by past SARS-CoV-2 infection among people of HMPCs: results from Cox regression models.