Trends in disparities in COVID hospitalizations among community‐dwelling residents of two counties in Connecticut, before and after vaccine introduction, March 2020–September 2021

Abstract Background Prior to the introduction of vaccines, COVID‐19 hospitalizations of non‐institutionalized persons in Connecticut disproportionately affected communities of color and individuals of low socioeconomic status (SES). Whether the magnitude of these disparities changed 7–9 months after vaccine rollout during the Delta wave is not well documented. Methods All initially hospitalized patients with laboratory‐confirmed COVID‐19 during July–September 2021 were obtained from the Connecticut COVID‐19‐Associated Hospitalization Surveillance Network database, including patients' geocoded residential addresses. Census tract measures of poverty and crowding were determined by linking geocoded residential addresses to the 2014–2018 American Community Survey. Age‐adjusted incidence and relative rates of COVID‐19 hospitalization were calculated and compared with those from July to December 2020. Vaccination levels by age and race/ethnicity at the beginning and end of the study period were obtained from Connecticut's COVID vaccine registry, and age‐adjusted average values were determined. Results There were 708 COVID‐19 hospitalizations among community residents of the two counties, July–September 2021. Age‐adjusted incidence was the highest among non‐Hispanic Blacks and Hispanic/Latinx compared with non‐Hispanic Whites (RR 4.10 [95% CI 3.41–4.94] and 3.47 [95% CI 2.89–4.16]). Although RR decreased significantly among Hispanic/Latinx and among the lowest SES groups, it increased among non‐Hispanic Blacks (from RR 3.2 [95% CI 2.83–3.32] to RR 4.10). Average age‐adjusted vaccination rates among those ≥12 years were the lowest among non‐Hispanic Blacks compared with Hispanic/Latinx and non‐Hispanic Whites (50.6% vs. 64.7% and 66.6%). Conclusions Although racial/ethnic and SES disparities in COVID‐19 hospitalization have mostly decreased over time, disparities among non‐Hispanic Blacks increased, possibly due to differences in vaccination rates.

dently associated with incidence of hospitalization regardless of poverty or crowding status. 5 Previous studies of influenza hospitalization data in New Haven County had found that census tract poverty and crowding levels were more strongly associated with hospitalization than race/ethnicity. 7,8 The researchers hypothesized that the epidemiology of SARS-CoV-2 hospitalization was greatly impacted by the "stay at home" orders, which protected the majority of the population from exposure, with the exception of "essential workers" and their household contacts. 5 Essential workers with decreased access to personal protective equipment and social support disproportionately included people of color, accounting for the racial/ethnic disparities in hospitalization incidence. 5 Hospitalization data from July to December 2020 also revealed that race and ethnicity were more strongly associated with higher incidence of hospitalization compared with census tract levels of poverty and crowding. 6 However, the racial/ethnic disparities (as well as SES disparities) in hospitalization were smaller from July to December 2020 than those found during the initial "lockdown" time period. The researchers attributed this reduction to increased accessibility of personal protective equipment and the lifting of restrictions on nonessential businesses, gatherings, and activities outside the home which led to exposures across a wider diversity of ages, SES, and racial/ethnic groups. 6 The first objective of this study was to determine the populationbased epidemiology of COVID-19 hospitalization in July-September 2021 in the same study areas, Middlesex and New Haven counties, during the Delta variant wave and beginning after all persons 12 years and older had the opportunity to be vaccinated against COVID-19 for at least 2-7 months. [9][10][11] The second was to determine trends in racial/ethnic and SES hospitalization incidence disparities by comparing these results with the two previous pre-vaccination time periods.
By September 30, 2021, an estimated 75.5% of Connecticut's population age ≥ 12 had received two vaccination doses and 81.6% had received at least one dose. 12

| Study population
The selected study population included non-institutionalized residents of New Haven and Middlesex Counties who were hospitalized for the first time with laboratory-confirmed COVID-19 during the time period of July to September 2021. Non-institutionalized, or communitydwelling individuals, were defined as people living at a residential address at the time of their hospitalization, which excluded people admitted to the hospital from a skilled nursing facility, correctional facility, rehabilitation facility, homeless shelter, and so on. The population included individuals of any age.

| Census data
We used the 2010 census tracts and the American Community Survey crowding was determined by percent of residents living with more than one person per room: 0-0.009% was categorized as very low levels of crowding, 0.01-0.25% as low levels of crowding, 0.026-0.05% as moderate levels of crowding, and 0.05-4% as high levels of crowding.

| Statistical analysis
Community-dwelling cases were compared by demographic variables with those excluded because of institutional residence using odds ratios. Crude and age-adjusted incidence rates by demographic variables were calculated for community-dwelling cases. Age-adjusted rates were calculated using the 2000 US standard population proportions. 14 Chi-square tests were used to compare hospitalization incidence between demographic and SES groups. Chi-square for trend was used to determine whether there was a significant trend association between increasing poverty and crowding levels with ageadjusted incidence. Chi-square tests and 95% confidence intervals (CI) were used to compare the case characteristics of communitydwelling patients hospitalized due to COVID-19 for this study and the study conducted from July to December 2021. Average age-adjusted rates of vaccination by race/ethnicity were calculated using age and race/ethnicity-specific data made available online from the Connecticut Department of Public Health using data from the weeks ending July 7 and September 29, 2021, and averaging the results. 13 All statistical analyses were performed using SAS version 9.4 and Epi Info.

| RESULTS
There were 751 cases of initial hospitalization with laboratory-confirmed SARS CoV-2 during July-September 2021. Of these 751 cases, 708 were hospitalized from a residential address and 43 from an institution, with the majority of the latter from long-term care facilities.
For age-adjusted relative risks of hospitalized cases in community-dwelling individuals, non-Hispanic Black individuals were 4.10 times more likely and Hispanic/Latinx individuals 3.47 times more likely to be hospitalized than White individuals (  (Table 4). However, the high versus very low relative risks in both levels of poverty and crowding remained relatively stable between the July-December 2021 and July-September 2020 time periods.
Vaccination completion rates at both the beginning and end of the study period varied by age group and race/ethnicity (Table 5).
Within each race/ethnic group, vaccination rates increased with age. * P-value < .05 of the percentage of total July-September 2021 compared with July-December 2020. ** P-value < .01 of the percentage of total July-September 2021 compared with July-December 2020. *** P-value < .001 of the percentage of total July-September 2021 compared with July-December 2020. rates in each age group and in the overall age-adjusted average, which was 14-16% lower than for the non-Hispanic White and Hispanic/ Latinx groups. Although White individuals had higher age-specific vaccination rates in age groups <45 years, Hispanic/Latinx individuals had higher vaccination rates for age groups 45 and older.  16 Additionally, disinvestment due to racialized practices and policies has left many communities of color without adequate access to medical care and health-promoting resources, including vaccines. 15 Whether due to mistrust, lack of access, or both, at the end of September 2021, Connecticut's vaccine eligible, non-Hispanic Black population was much less likely to be fully vaccinated than either the Hispanic/Latinx and non-Hispanic White populations. Notably, the gap in vaccination rates between Connecticut's non-Hispanic White population and Hispanic/Latinx population is less pronounced than the disparity between the non-Hispanic White and non-Hispanic Black populations, and Hispanic/Latinx vaccination rates are higher than the non-Hispanic White rates among those 45 years and older. Because increased age is associated with a higher likelihood of COVID-19 hospitalization, the relatively higher rates of vaccination in Hispanic/Latinx persons could explain why the age-adjusted relative risk for Hispanic/Latinx individuals compared with non-Hispanic White individuals decreased from 4.73 in the immediate pre-vaccine era to 3.47 during July-September 2021.

| DISCUSSION
T A B L E 4 Time period comparisons of age-adjusted relative risks by race/ethnicity and socioeconomic status of residents of New Haven and Middlesex Counties, CT, in March-May 2020, July-December 2020, and July-September 2021 Note: Fully vaccinated = two doses of Moderna or Pfizer vaccines or one dose of Johnson and Johnson vaccine. Vaccination rates are underestimates as denominators are full denominators for each group based on the 2020 census, but 5.0-4.5% of vaccinated persons had unknown race/ethnicity. a Age-adjusted by the four age groups using 2020 census for CT. Average of age-adjusted rates for July 7 and September 29, 2021.