a b s t r a c t

Background: To assess the effectiveness of Messenger RNA vaccines against severe acute respiratory syndrome coronavirus 2 (SARS-COV-2) in preventing emergency department (ED) presentations for acute respiratory ill- ness.

Basic procedures: We conducted a retrospective study assessing adult presentations (age >= 18) to the University of Miami Hospital’s ED from January 1st through August 25th, 2021, with a SARS-COV-2 PCR test and acute respi- ratory infection symptoms. Vaccine effectiveness was calculated using a test-negative design. Both univariable and multivariable (adjusted for age, gender, race, insurance status, imputed body mass index [BMI], vaccine type, week of presentation) regression analyses were conducted for the full cohort and subgroups.

Main findings: The cohort consisted of 13,203 ED presentations–3134 (23.7%) fully vaccinated and SARS-COV-2 negative, 108 (0.8%) fully vaccinated and SARS-COV-2 positive, 8817 (66.8%) unvaccinated and SARS-COV-2 neg- ative, and 1144 (8.7%) unvaccinated and SARS-COV-2 positive. Unadjusted vaccination effectiveness was 73.4% (95% confidence interval: 67.5%,78.3%) and, after adjustment, 73.8% (66.2%,79.7%). The Moderna vaccine’s effec- tiveness was numerically higher (unadjusted: 78.2% [68.8%, 84.7%]; adjusted: 78.0% [68.1%, 84.9%]) than the Pfizer vaccine’s (unadjusted: 70.8% [62.9%, 76.9%]; adjusted: 73.9% [66.3%,79.8%]). We found a significant differ- ence in adjusted vaccine effectiveness across categories was BMI (p < 0.001)–BMI <25: 66.3% (45.3%,79.2%); BMI 25-29: 71.3% (56.1%, 81.2%); BMI 30-34: 84.5% (71.7%, 91.5%); and BMI >=35: 72.7% (50.5%, 84.9%).

Principal conclusions: We demonstrated excellent real-world effectiveness of mRNA vaccines in preventing ED presentation for SARS-COV-2 in a diverse U.S. cohort. Notably, vaccine effectiveness improved with increasing BMI (until class 2 obesity).

(C) 2022

1. Introduction

Two messenger RNA vaccines against severe acute respiratory syn- drome coronavirus 2 (SARS-COV-2) were released to the general popu- lation at the turn of 2020. These vaccines were shown to be 94.1 and 95% efficacious at preventing symptomatic SARS-CoV-2 infection [1,2]. While numerous trials have evaluated the effectiveness of severe SARS-COV-2 vaccinations in Healthcare workers [3-5] and inpatients [6,7], there remains a paucity of data in the general U.S. population pre- senting to the Emergency Department (ED). Moreover, focus has been

* Corresponding author at: University of Miami, 1951 NW 7th Avenue, Miami, FL 33136, USA.

E-mail address: cmallow@miami.edu (C. Mallow).

primarily on the prevention of severe disease; less is known about sig- nificant disease which may not require hospitalization. In this study, we assessed vaccine effectiveness in preventing ED presentations for acute respiratory illness.

1. Methods

We conducted a retrospective study assessing adult presentations (age >= 18) to the University of Miami Hospital’s ED from January 1st through August 25th, 2021, with a SARS-COV-2 PCR test and acute re- spiratory infection symptoms, identified using ICD-10 codes. Institu- tional review board approval was obtained from the University of Miami (20200739). Presentations were classified into three groups:

(1) Fully vaccinated: 14 days after 2nd mRNA vaccine dose; (2) Partially

https://doi.org/10.1016/j.ajem.2022.01.066

0735-6757/(C) 2022

Fig. 1. Forest Plots demonstrating Vaccine Effectiveness. Panel A displays the unadjusted vaccine effectiveness. Panel B shows the adjusted vaccine effectiveness^a.

^aTo assess the statistical significance of differences in vaccine effectiveness by each subgrouping, the full cohort model was rerun with an interaction term for vaccination status (fully vs unvaccinated) and subgroup of interest. P-values for Wald testing of the interaction terms are: gender–unadjusted 0.74, adjusted 0.34; age–unadjusted <0.001, adjusted 0.36; race–unadjusted <0.001, adjusted 0.48; ethnicity–unadjusted 0.55, adjusted 0.87; and payor–unadjusted <0.001, adjusted 0.09; body mass index–unadjusted <0.0001, adjusted

<0.001. Interaction terms were not considered for vaccine type as vaccine type is a subgroup of the exposure, vaccination status.

vaccinated: received at least one dose but not fully vaccinated; (3) Un- vaccinated. Single dose adenovirus vaccination was excluded from our analysis due to the low frequency of receipt in our population. Vaccina- tion status was obtained at the time of ED presentation through the electronic medical record (EMR) which is synced with the state registry, Florida Shots. Manual extraction of vaccination data from Florida Shots for 50 randomly selected presentations (25 fully vaccinated, 25 unvacci- nated in the EMR) was performed to assess EMR accuracy.

Vaccine effectiveness was calculated using a test-negative design, previously described [6,8] (Fig. 1). Primary analyses compared fully vac- cinated to unvaccinated presentations (partially vaccinated presenta- tions excluded); we performed a sensitivity analysis including presentations with partial vaccination as a separate exposure group. Both univariable and multivariable (adjusted for age, gender, race, in- surance status, imputed body mass index [BMI], vaccine type) regres- sion analyses were conducted for the full cohort and subgroups (with assessment of subgroup-related differences by inclusion of interaction terms in full cohort model). Because of the temporal variability in SARS-COV-2 case positivity and vaccination rates, we also adjusted for week of ED visit (as restricted cubic splines with 4 knots). All statistics were performed in STATA 16 (College Station, TX).

1. Results

The cohort consisted of 13,203 ED presentations–3134 (23.7%) fully vaccinated and SARS-COV-2 negative, 108 (0.8%) fully vaccinated and SARS-COV-2 positive, 8817 (66.8%) unvaccinated and SARS-COV-2 neg- ative, and 1144 (8.7%) unvaccinated and SARS-COV-2 positive. Unvacci- nated patients tended to be younger and more commonly insured by Medicaid (Table 1). Random sampling manual review, via our state da- tabase, demonstrated perfect positive and negative predictive value (100%) of true vaccination status by EMR documentation.

Unadjusted full vaccination effectiveness was 73.4% (95% CI 67.5%,78.3%) and, after adjustment, 73.8% (66.2%,79.7%). The Moderna vaccine’s effectiveness was numerically higher (unadjusted: 78.2% [68.8%, 84.7%]; adjusted: 78.0% [68.1%, 84.9%]) than the Pfizer vaccine’s

(unadjusted: 70.8% [62.9%, 76.9%]; adjusted: 73.9% [66.3%,79.8%];

Fig. 2). While differences were noted in the unadjusted vaccine effec- tiveness by age, race, payor, and BMI (p < 0.001), the only subgrouping with a significant difference in adjusted vaccine effectiveness was BMI (p < 0.001)–BMI <25: 66.3% (45.3%,79.2%); BMI 25-29: 71.3% (56.1%,

81.2%); BMI 30-34: 84.5% (71.7%, 91.5%); and BMI >=35: 72.7% (50.5%,

84.9%).

Sensitivity analysis was performed with 14,173 presentations, 970 of which were partially vaccinated. The effectiveness of partial

vaccination was 36.3% (18.7%,50.2%) unadjusted and 19.6% (-4.7%,38.2%) adjusted; in this cohort, the adjusted effectiveness of full vaccination was 64.0% (55.1%, 71.2%).

1. Discussion

In this study, we demonstrate excellent real-world effectiveness of mRNA vaccines in preventing ED presentation for SARS-COV-2 in a di- verse U.S. cohort. Notably, vaccine effectiveness improved with increas- ing BMI (until class 2 obesity), with the highest effectiveness among those with a BMI of 30-34. There was no statistically significant differ- ence in adjusted vaccine effectiveness among patients of different gen- ders, ages, races, ethnicities or payors.

To our knowledge, this is the first study to evaluate vaccine effective- ness in preventing, specifically, ED presentations. A recent publication demonstrated a higher mRNA vaccine effectiveness (91% vs our 73.8%) at preventing ED or urgent care visits [6]. Three major differences exist between this study and ours. First, Thompson et al. [6] included ur- gent care visits; however, this inclusion of a presumably less acutely ill population would be likely bias towards a reduced vaccine effective- ness. A second important difference was the time period of study, nota- bly our inclusion of presentations during the summer of 2021. During this time the delta variant became predominant in the U.S. [9], and vac- cines have been found to be less effective in preventing delta-related symptomatic SARS-CoV-2 infection [10]. Notably, we test and genotype all our ED presentations with acute upper respiratory symptoms, and the delta variant accounted for nearly 100% of the SARS-COV-2 infec- tions as of July 2021. Finally, the two cohorts studied differed substan- tially–ours was younger (58.8% vs 34% <65 years-old) and more racially/ethnically diverse (69.3% vs 82% White and 42.6% vs 84% non- Hispanic). While adjustments for these measured differences were made, it is unclear whether and to what extent coincident unmeasured differences may have contributed to our disparate results.

Our study is novel in focusing on ED presentations for SARS-CoV-2, a more expansive marker of significant disease than hospitalization alone, in a diverse U.S. cohort. There are, however, limitations. First, we only had 108 cases of vaccinated SARS-COV-2 and < 10 in some subgroups; even a single additional case would meaningfully impact our calcula- tions for these. Second, we did not verify all presentations’ vaccination status through registries. Our manual review suggests that vaccination misclassification is rare to non-existent, though. Third, our cohort in- cluded more vaccinations with Pfizer than Moderna; whether this disproportionality impacted our results is unknown. Last, as our study is retrospective in nature, the impact of residual confounding despite multivariable adjustment cannot be fully excluded.

Table 1

Cohort characteristics.

Characteristic	Overalla	Fully vaccinated &	Fully vaccinated &	Not vaccinated &	Not vaccinated &
		COVID-	COVID+	COVID-	COVID+
# of ED Presentations, # (row %)	13,203 (100%)	3134 (23.7%)	108 (0.8%)	8817 (66.8%)	1144 (8.7%)
Vaccination typeb
mRNA, Moderna	1162 (8.8%)	1130 (36%0.1)	32 (29.6%)	n/a	n/a
mRNA, Pfizer	2080 (15.8%)	2004 (63.9%)	76 (70.4%)	n/a	n/a
Agec (years), median (IQR)	61 (47,73)	68 (57,77)	69 (51,78)	59 (44,70)	55 (40,68)
Male Sex	6406 (48.5%)	1652 (52.7%)	60 (55.6%)	4654 (52.8%)	613 (54.5%)
Race
White	9.147 (69.3%)	2525 (80.6%)	86 (79.6%)	5865 (66.5%)	671 (58.7%)
Black	3349 (25.4%)	450 (14.4%)	17 (15.7%)	2484 (28.2%)	398 (34.8%)
Other/Unknown	707 (5.4%)	159 (5.1%)	5 (4.6%)	468 (5.3%)	75 (6.6%)
Ethnicity
Non-Hispanic	5621 (42.6%)	1311 (41.8%)	40 (37.0%)	3795 (43.0%)	475 (41.5%)
Hispanic	7312 (55.4%)	1766 (56.3%)	66 (61.1%)	4831 (54.8%)	649 (56.7%)
Other/Unknown	270 (2%)	57 (1.8%)	2 (1.9%)	191 (2.2%)	20 (1.7%)
Elixhauser Comorbiditiesd
Congestive heart failure	2391 (18.1%)	604 (19.2%)	14 (13.0%)	1674 (19.0%)	99 (8.7%)
Valvular disease	2446 (18.5%)	729 (23.3%)	22 (20.4%)	1606 (18.2%)	89 (7.8%)
Peripheral vascular disorders	2397 (18.2%)	696 (22.2%)	14 (13.0%)	1608 (18.2%)	79 (6.9%)
Hypertension	8799 (66.6%)	2351 (75.0%)	70 (64.8%)	5779 (65.5%)	599 (52.4%)
neurologic disorders	3010 (22.8%)	776 (24.8%)	15 (13.9%)	2104 (23.9%)	115 (10.1%)
Chronic pulmonary disease	3520 (26.7%)	940 (30.0%)	20 (18.5%)	2351 (26.7%)	209 (18.3%)
Diabetes mellitus
Uncomplicated	3780 (28.6%)	975 (31.1%)	40 (37.0%)	2485 (28.2%)	280 (18.2%)
Complicated	3450 (26.1%)	882 (28.1%)	29 (26.9%)	2310 (26.2%)	229 (20.0%)
Hypothyroidism	1936 (14.7%)	587 (18.7%)	15 (13.9%)	1245 (14.1%)	89 (7.8%)
Renal Failure	2591 (19.6%)	692 (22.1%)	27 (25.0%)	1745 (19.8%)	127 (11.1%)
Liver failure	2179 (16.5%)	583 (18.6%)	9 (8.3%)	1488 (16.9%)	99 (8.7%)
Metastatic cancer	2380 (18.0%)	649 (20.7%)	7 (6.5%)	1672 (19.0%)	52 (4.5%)
Solid tumor (no metastases)	3558 (26.9%)	1017 (32.5%)	16 (14.8%)	2433 (27.6%)	92 (8.0%)
Coagulopathy	2361 (17.9%)	567 (18.1%)	12 (11.1%)	1671 (19.0%)	111 (9.7%)
Obesity	4465 (33.8%)	1131 (36.1%)	34 (31.5%)	2959 (33.6%)	341 (29.8%)
Weight loss	2657 (20.1%)	642 (20.5%)	12 (11.1%)	1926 (21.8%)	77 (6.7%)
Fluid & Electrolyte disorders	6401 (48.5%)	1480 (47.2%)	36 (33.3%)	4485 (50.9%)	400 (35.0%)
Deficiency anemia	5836 (44.2%)	1338 (42.7%)	32 (29.6%)	4175 (47.4%)	291 (25.4%)
Depression	2966 (22.5%)	753 (24.0%)	18 (16.7%)	2092 (23.7%)	103 (9.0%)
Body mass indexe
<25	4803 (36.4%)	1112 (35.5%)	27 (25.0%)	3406 (38.6%)	258 (22.6%)
25-30	4074 (31.9%)	1024 (32.7%)	37 (34.3%)	2676 (30.4%)	337 (29.5%)
30-35	2292 (17.4%)	567 (19.0%)	20 (18.5%)	1432 (16.2%)	273 (23.9%)
>35	1871 (14.2%)	419 (13.4%)	21 (19.4%)	1199 (13.6%)	232 (20.3%)
Insurance
Commercial	4597 (34.8%)	1036 (33.1%)	33 (30.6%)	3107 (35.2%)	421 (36.8%)
Medicaid	1880 (14.2%)	243 (7.8%)	5 (4.6%)	1482 (16.8%)	150 (13.1%)
Medicare	5594 (42.4%)	1761 (56.2%)	55 (50.9%)	3434 (38.9%)	344 (30.1%)
Other	1132 (8.6%)	94 (3.0%)	15 (13.9%)	794 (9.0%)	229 (20%)
Maximum SOFA score within 24 h of ED presentationf, median (IQR)	1 (0,2)	1 (0,2)	1 (0,3)	1 (0,2)	1 (0,2)
Outcomes
Admitted to hospital from ED	9894 (75%)	2569 (26.0%)	41 (38.0%)	6768 (76.8%)	516 (45.1%)
Final ED or hospital dispositiong Death	614 (4.7%)	120 (4.8%)	5 (4.6%)h	431 (4.9%)	58 (5.1%)
Facility	896 (6.8%)	225 (7.2%)	4 (3.7%)	601 (6.8%)	66 (5.8%)
Home	11,671 (88.4%)	2786 (88.9%)	99 (91.7%)	7771 (88.1%)	1015 (88.7%)

ED: emergency department; IQR: interquartile range; SOFA: sequential organ failure assessment.

^a Inclusive of fully vaccinated and unvaccinated.

^b No vaccinations with adenovirus-based vaccines were included in the cohort.

^c 1 (0.008%) missing data on age.

^d All other Elixhauser comorbidities were < 10% of the full cohort and included: pulmonary circulation disorders, 5.3%; paralysis, 5.7%; Peptic ulcer disease, 5.2%; AIDS, 2.0%; lymphoma, 6.0%; rheumatoid arthritis/collagen vascular disease, 5.9%; blood loss anemia, 6.3%; alcohol abuse, 4.6%; drug abuse, 6.0%; and psychoses, 6.3%.

^e 163 (1.23%) missing data on body mass index.

^f 259 (1.96%) missing data on SOFA score.

^g 22 (0.17%) remained as hospital admissions at the time of data collection.

^h All five vaccinated patients who died were in high risk categories: three were >= 85 years old; one was 65 years old with pulmonary fibrosis on mycophenolate; and, one was 74 years old with pulmonary fibrosis.

Fig. 2. Forest Plots demonstrating Vaccine Effectiveness. Panel A displays the unadjusted vaccine effectiveness. Panel B shows the adjusted vaccine effectiveness.

This study demonstrates excellent real-world effectiveness of SARS- COV-2 mRNA vaccines in preventing significant disease in a diverse U.S. population.

Presentations

None.

Author contributions

CM, TF, BS, PW, MS, DP and HG all contributed to the study design. CM contributed to the analysis and interpretation of the data, drafting of the manuscript, and critical revision of the manuscript. HG contrib- uted to the analysis and interpretation of the data, and critical revision of the manuscript. PW contributed to data acquisition, analysis of the data and revision of the manuscript. TF, BS, MS, and DP contributed to the interpretation of the data and revision of the manuscript.

Disclosures

Mr. Warde and Dr.’s Shukla, Sosa, and Gershengorn received funding from University of Miami Hospital and Clinics data analytics research team.

CRediT authorship contribution statement

Christopher Mallow: Conceptualization, Investigation, Methodol- ogy, Writing – original draft, Writing – review & editing. Tanira Ferreira: Writing – review & editing, Methodology, Investigation, Conceptualization. Bhavarth Shukla: Writing – review & editing, Methodology, Investigation, Conceptualization. Prem Warde: Writing – review & editing, Formal analysis, Data curation, Conceptualization. Marie Anne Sosa: Writing – review & editing, Methodology, Investigation,

Conceptualization. Dipen J. Parekh: Writing – review & editing, Method- ology, Investigation, Conceptualization. Hayley B. Gershengorn: Writing – review & editing, Methodology, Investigation, Formal analysis, Data curation, Conceptualization.

Declaration of Competing Interest

Dr’s Shukla, Sosa, and Gershengorn received funding from University of Miami Hospital and Clinics data analytics research team.

References

1. Baden LR, El Sahly HM, Essink B, et al. Efficacy and safety of the mRNA-1273 SARS- CoV-2 vaccine. N Engl J Med. 2021;384:403-16.
2. Polack FP, Thomas SJ, Kitchin N, et al. Safety and efficacy of the BNT162b2 mRNA Covid-19 vaccine. N Engl J Med. 2020;383:2603-15.
3. Fabiani M, Ramigni M, Gobbetto V, Mateo-Urdiales A, Pezzotti P, Piovesan C. Effec- tiveness of the Comirnaty (BNT162b2, BioNTech/Pfizer) vaccine in preventing SARS-CoV-2 infection among healthcare workers, Treviso province, Veneto region, Italy, 27 December 2020 to 24 March 2021. Eurosurveillance. 2021:26.
4. Hall VJ, Foulkes S, Saei A, et al. COVID-19 vaccine coverage in health-care workers in England and effectiveness of BNT162b2 mRNA vaccine against infection (SIREN): a prospective, multicentre, cohort study. Lancet. 2021;397:1725-35.
5. Swift MD, Breeher LE, Tande AJ, et al. Effectiveness of mRNA COVID-19 vaccines against SARS-CoV-2 infection in a cohort of Healthcare personnel. Clin Infect Dis. 2021;73(6):e1376-9. https://doi.org/10.1093/cid/ciab361.
6. Thompson MG, Stenehjem E, Grannis S, et al. Effectiveness of Covid-19 vaccines in ambulatory and inpatient care settings. N Engl J Med. 2021;385(15):1355-71. https://doi.org/10.1056/NEJMoa2110362.
7. Tenforde MW, Patel MM, Ginde AA, et al. Effectiveness of SARS-CoV-2 mRNA vac- cines for preventing COVID-19 hospitalizations in the United States. Clin Infect Dis. 2021;ciab687. https://doi.org/10.1093/cid/ciab687.
8. Jackson ML, Nelson JC. The test-negative design for estimating Influenza vaccine ef- fectiveness. Vaccine. 2013;31:2165-8.
9. COVID Data Tracker. Center for Disease Control and Prevention; 2021. (Accessed September 16th, 2021, 2021).
10. Lopez Bernal J, Andrews N, Gower C, et al. Effectiveness of Covid-19 vaccines against the B.1.617.2 (Delta) variant. N Engl J Med. 2021;385:585-94.