COVID-19 Vaccine Effectiveness in a Diverse Urban Health Care Worker Population.

To the Editor:

Whereas there is emerging, real-world research investigating COVID-19 vaccine effectiveness (VE) on health care workers (HCWs), , specific VE data from hospital settings with diverse urban employee populations are lacking. In addition, most studies have not controlled for demographic characteristics, including race and background community COVID-19 incidence, which are risk factors among HCWs.

We investigated the COVID-19 VE among employees in our ethnically diverse community health care system in Massachusetts (44% of our HCWs are non-White) during its initial immunization campaign. The HCWs of the system were retrospectively included from the beginning of a COVID-19 vaccination program (December 16, 2020) until March 31, 2021. Those with a prior COVID-19 infection before December 15 were excluded. The Occupational Health department of the system ran a COVID-19 screening and testing referral program for workers, consistently throughout the study period. A master database comprising the demographic characteristics, COVID-19 polymerase chain reaction assays, and vaccinations of each HCW had been established (described previously) and updated. The database was deidentified, and the study was exempted by the Cambridge Health Alliance Institutional Review Board (4/29/202-003).

The Pfizer and Moderna vaccines were made available to HCWs starting on December 16 and December 23, 2020, respectively, and opened to all employees on December 29. Participation was voluntary at conveniently located hospital-based vaccination sites; no appointment was required. After Emergency Use Authorization in February 2021, a limited number of J&J/Janssen vaccine doses were available. Fully vaccinated employees with breakthrough infections were telephonically interviewed by our Occupational Health medical staff following the screening/referral protocol. We built an Andersen-Gill extension of the Cox proportional hazards models to account for correlated data and further adjusted for potential confounders: age, sex, race, and the Massachusetts statewide 7-day average of new cases on the date of the first vaccine dose. The VE was calculated as 100% × (1 − hazard ratio).

Among the 4317 eligible HCWs, 3249 (75%) received any vaccination during the study period. Vaccinated HCWs were older (45.7±13.5 years vs 41.3±12.8 years; P<.001) and more likely to be non-Hispanic Whites. In addition, medical providers were more likely to be vaccinated compared with other HCWs (89% vs 73%; P<.001). After adjusting for potential confounders, we observed a VE of 80.2% (95% CI, 57.5% to 90.8%) for 14 or more days after the first dose of Pfizer/Moderna and 95.5% (95% CI, 88.2% to 98.3%) among those fully vaccinated (ie, ≧14 days after the second dose of Pfizer/Moderna or the single dose of J&J/Janssen; Table ). During the study period, there were 6 breakthrough infections, all paucisymptomatic or asymptomatic, with no hospitalizations or death. No variants of concern were discovered among the genotyped samples.

Table

Rate of Infection During the Study Period Across the 5 Vaccination Categories (Separating Period With First Dose Only to <14 and 14+ Days and Excluding 318 people Infected Before December 15, 2020)

Status	Person-days	No. of infections	Rate per 10,000 person-days	Unadjusted vaccine effectiveness, % (95% CI)	Adjusted vaccine effectiveness, % (95% CI)a
Unvaccinated	172,845	133	7.69	Not applicable	Not applicable
First dose (<14 days)	40,344	28	6.94	26.9 (−17.6 to 54.5)	28.1 (−15.9 to 55.4)
First dose (14+ days)	40,577	8	1.97	80.1 (57.8-90.6)	80.2 (57.5-90.8)
Second dose	41,817	2	0.48	95.4 (80.8-98.9)	95.2 (80.0-98.8)
Fully vaccinated	148,475	4	0.27	97.2 (92.5-99.0)	95.5 (88.2-98.3)

Vaccine effectiveness (95% CI) derived from the Andersen-Gill extension of the Cox proportional hazards model.

Adjusted for age, sex, race, and the Massachusetts statewide 7-day average of new cases at the date for the first vaccine dose. Those with the race of “American Indian or Alaska Native,” “Hawaiian or Pacific Islander,” or “Two or More” were pooled into 1 level, “other race.”

Our findings show that COVID-19 vaccines are promising, and these data in concert with culturally appropriate outreach may decrease vaccine hesitancy. The study has strengths, including that other than age and sex, we adjusted for race/ethnicity and 7-day incidence in Massachusetts at the time of vaccination to account for the background rate. Our population is multiethnic, allowing us to draw better conclusions about populations underrepresented in clinical trials.