This is the first study estimating socioeconomic inequalities in SARS-CoV-2 infection, COVID-19 hospitalisation and COVID-19 death in urban Italy over the course of the COVID-19 vaccine rollout in 2021. Our findings, based on data from the whole urban population aged 20 years and over, show that census block-level SED was positively associated with SARS-CoV-2 infection and COVID-19-related hospitalisation and death throughout the vaccine rollout in 2021. Although COVID-19-related incidences reduced across all three SED terciles as the vaccination rollout progressed, we found that socioeconomic inequalities in all study outcomes increased between periods of increasing vaccination coverage.

Overall, the positive associations we found between area level socioeconomic deprivation and COVID-19-related outcomes are in concordance with literature exploring inequalities in COVID-19 prior the vaccine rollout [4]. Similar to our findings, a national scale study conducted in Belgium [17] found increasing socioeconomic inequalities in SARS-CoV-2 infection as vaccination coverage increased. Moreover, in a second study conducted in Madrid, Spain, inequalities in infection affecting deprived areas were shown to peak when vaccination coverage reached 70% of the adult population [18].

Conversely, findings from a study in Bavaria, Germany showed that increasing vaccination coverage appeared to balance incidence and mortality rates between the most and least deprived districts [19]. Two of several factors which could explain these temporal pattern differences are inter-study heterogeneity (e.g. in the study design, study population/setting and measure of deprivation) and cross-country contextual discrepancies (e.g. in socioeconomic factors, the vaccination rollout and the pandemic development).

Existing literature indicates that individuals residing in deprived areas are likely to experience increased SARS-CoV-2 exposure/transmission and elevated vulnerability to severe COVID-19 outcomes, attributed to inequalities in housing, working and pre-existing health conditions [6]. We consider that the interaction of these pathways with vaccine rollout characteristics and several other period-specific factors; differentially affecting socioeconomic groups over the course of the vaccination campaign in Italy, may explain the socioeconomic patterns we observed over time.

We found that socioeconomic inequalities were least pronounced in the early vaccine rollout phase (i.e. vaccination coverage: 0–10%) where eligible groups for vaccination included health care and long-term care professionals, persons ≥ 60 years and individuals with chronic comorbidities [20]. Although the latter population group was prioritised for vaccination, findings from Italy show a higher burden of chronic comorbidities in deprived populations [21] and a positive association between SED and vaccine hesitancy [22]. A seemingly reduced vaccine uptake amongst high risk—socioeconomically deprived populations in the early stages of the vaccine rollout—could contribute to the higher COVID-19-related hospitalisation and death rates we found amongst those residing in more deprived areas, whilst also influencing their susceptibility to infection.

Another distinct characteristic of this “baseline” coverage period was the full implementation of a four-tiered restriction system at the regional level [23]. As previously shown in settings where mobility and social distancing restriction measures are deployed, individuals with manual occupations (primarily residing in more deprived areas) are less likely to be able to work remotely and are overall more likely to suffer from poor working and commuting conditions [5]. These factors, alongside household overcrowding, typically encountered in deprived areas in Italy [16], may have resulted in elevated SARS-CoV-2 exposure and transmission in the more deprived census blocks, explaining the higher infection rates we observed in these areas.

During the intermediate coverage period (i.e. vaccination coverage: > 10–60%; as vaccination became available for people aged ≥ 18 years), we observed that incidence rates for SARS-CoV-2 infection and COVID-19-related hospitalisation and death decreased less in the more deprived areas compared to the least deprived ones, translating to slight IRR increases. In this period restriction measures progressively relaxed after May 2021 [24], and the newly introduced Delta variant gradually supplanted Alpha (which had prevailed over the Ancestral variant during the early vaccination rollout period) [25].

Despite the fact that COVID-19 vaccination in Italy has always been free and with a low threshold access through vaccination hubs, several studies from Italy indicate the emergence of socioeconomic inequalities in COVID-19 vaccination to the disadvantage of deprived populations, as vaccines became increasingly available [26, 27]. Such a reality in conjunction with the successive predominance of variants (i.e. Ancestral to Alpha and then Delta) exhibiting increased infectivity and transmissibility, especially amongst non-vaccinated individuals [28, 29], could account for the inequality shifts we observed across all study outcomes.

Furthermore, despite the relaxation of restriction measures during the intermediate vaccination coverage period, individuals with office/desk jobs, traditionally residing in more affluent areas, largely retained their remote working status as per during the early vaccination rollout phase, whilst manual workers increasingly returned to their on-site work activities. This labour mobility context likely resulted in increased relative infection exposure and subsequent workplace and household transmission in the latter group (primarily residing in deprived areas), contributing to the higher SARS-CoV-2 incidence inequalities we found.

In the high coverage period (i.e. vaccination coverage: > 60–74%), we observed that incidence rates for all study outcomes decreased again less in the more deprived areas compared to the least deprived ones, resulting in enhanced inequalities across all study outcomes. Given the high vaccination coverage during this period (> 70% of the population had a complete primary vaccination cycle by early October 2021 [8]) restriction measures were further reduced [24], whilst Delta was the sole predominant variant [25].

Marked socioeconomic inequalities in vaccination shown to exist in other European countries when vaccination coverage was high [30, 31], especially within a context of high primary vaccination cycle coverage (known to further increase protection against infection and severe disease [32]) could be a key factor influencing the significant IRR increases we observed here, particularly evident for COVID-19 hospitalisation. Likely applicable to the Italian context [26, 27], vaccination inequalities per se, or jointly with the sole predominance of the more virulent Delta variant may have increased the impact of chronic health conditions on severe COVID-19 outcomes amongst unvaccinated/partially vaccinated individuals in the most deprived areas [31]; resulting in the heightened relative COVID-19 hospitalisation and COVID-19 death rates we observed.

Nonetheless, several other vaccine centred mechanisms may also be at play. The study conducted by Fabiani and colleagues in Italy, shows that vaccine effectiveness against SARS-CoV-2 infection and severe COVID-19 decreased in the Delta epidemic phase compared to the Alpha phase [33]. This reduction in vaccine induced protection, may have exacerbated the impact of unequal housing conditions, working conditions and comorbidities amongst vaccinated individuals living in the most deprived areas (compared to vaccinated individuals living in affluent areas), contributing to the higher inequalities in SARS-CoV-2 infection, and COVID-19 health outcomes we observed here.

Finally, the autumn season partly overlapping the high vaccination coverage period, may have also influenced SARS-CoV-2 infection incidence inequalities [34]. Specifically, the increased time spent indoors in autumn compared to summer months may have promoted household viral transmission, disproportionately affecting disadvantaged populations owing to crowded housing conditions.

Our study has several limitations. First, as we used population and deprivation index estimates based on 2011 census data, likely differing for 2021, we may have slightly over- or underestimated the true associations between SED and our study outcomes. Second, despite the fact that the deprivation index we used takes into account characteristics pertaining to education, employment status and housing conditions, there may be other important deprivation components not captured by the index. Third, our study did not include periods before the vaccine rollout in order to assess the overall impact of vaccination on socioeconomic inequalities in COVID-19 [30], nor periods with Omicron variant predominance where socioeconomic inequalities in COVID-19 may have evolved differently [35]. Finally, we lacked information on co-morbidities and vaccination data geolocated at the census block level, which could provide further insights into the associations we found.