Risk of COVID-19 Hospitalization and Protection Associated With mRNA Vaccination Among US Adults With Psychiatric Disorders

Abstract

Background: Although psychiatric disorders have been associated with reduced immune responses to other vaccines, it remains unknown whether they influence COVID-19 vaccine effectiveness (VE). This study evaluated risk of COVID-19 hospitalization and estimated mRNA VE stratified by psychiatric disorder status.

Methods: In a retrospective cohort analysis of the VISION Network in four US states, the rate of laboratory-confirmed COVID-19-associated hospitalization between December 2021 and August 2022 was compared across psychiatric diagnoses and by monovalent mRNA COVID-19 vaccination status using Cox proportional hazards regression.

Results: Among 2,436,999 adults, 22.1% had ≥1 psychiatric disorder. The incidence of COVID-19-associated hospitalization was higher among patients with any versus no psychiatric disorder (394 vs. 156 per 100,000 person-years, p < 0.001). Any psychiatric disorder (adjusted hazard ratio [aHR], 1.27; 95% CI, 1.18-1.37) and mood (aHR, 1.25; 95% CI, 1.15-1.36), anxiety (aHR, 1.33, 95% CI, 1.22-1.45), and psychotic (aHR, 1.41; 95% CI, 1.14-1.74) disorders were each significant independent predictors of hospitalization. Among patients with any psychiatric disorder, aHRs for the association between vaccination and hospitalization were 0.35 (95% CI, 0.25-0.49) after a recent second dose, 0.08 (95% CI, 0.06-0.11) after a recent third dose, and 0.33 (95% CI, 0.17-0.66) after a recent fourth dose, compared to unvaccinated patients. Corresponding VE estimates were 65%, 92%, and 67%, respectively, and were similar among patients with no psychiatric disorder (68%, 92%, and 79%).

Conclusion: Psychiatric disorders were associated with increased risk of COVID-19-associated hospitalization. However, mRNA vaccination provided similar protection regardless of psychiatric disorder status, highlighting its benefit for individuals with psychiatric disorders.

Keywords: COVID-19; anxiety disorders; electronic health records; epidemiology; mental disorders; mood disorders; psychotic disorders; vaccination.

FIGURE 1

Associations between psychiatric disorders and COVID‐19‐associated hospitalization, stratified by age group. A hazard ratio (HR) > 1.0 indicates that the respective psychiatric disorder was associated with a higher risk of COVID‐19‐associated hospitalization. Each HR was obtained from a separate model comparing patients with the respective psychiatric disorder (any, mood, anxiety, trauma‐stressor‐related, or psychotic) to patients with no psychiatric disorder. HRs were adjusted for site, age (natural spline with four knots), sex (male, female, unknown), race and ethnicity (Asian, Black, Hispanic, White, other, unknown), Medicaid coverage (yes, no, unknown), underlying respiratory condition (yes, no), underlying non‐respiratory condition (yes, no), number of underlying medical conditions (square‐root transformed), number of SARS‐CoV‐2 test records documented in the patient's electronic medical record prior to the start of the study period (0, 1, 2–4, ≥5), and time‐varying mRNA COVID‐19 vaccination status (unvaccinated, two doses 14–149 days earlier, two doses ≥150 days earlier, three doses 7–119 days earlier, three doses ≥120 days earlier, four doses 7–59 days earlier, four doses ≥60 days earlier). CI, confidence interval.

FIGURE 2

Kaplan–Meier survival curve of time to COVID‐19‐associated hospitalization, stratified by psychiatric disorder status. Time 0 is December 16, 2021, which was the earliest date a patient could start contributing eligible follow‐up. Sites had staggered entries from December 16 to 26, 2021 based on the date on which the SARS‐CoV‐2 Omicron variant first accounted for ≥50% of all sequenced specimens at each site. Individual patients could also enter the cohort at a later date if they became eligible based on a new COVID‐19 vaccination status. The large majority of patients (2,322,169; 95.3%) contributed follow‐up starting from their site‐specific start date in December 2021 through August 30, 2022. Smaller proportions entered the analytic cohort mid‐study (59,473; 2.4%) and/or were censored (53,164; 2.2%) either due to departure from the health network (33,403; 1.4%), exclusionary vaccine doses (8919; 0.4%), or death (10,842; 0.4%). Periods of estimated ≥50% BA.1 sublineage predominance (as early as December 16–26, 2021), ≥50% BA.2/BA.2.12.1 sublineage predominance (as early as March 19–24, 2022), and ≥50% BA.4/BA.5 sublineage predominance (as early as June 19–29, 2022) are displayed. The shaded areas indicate 95% confidence intervals (CIs). The unadjusted hazard ratio (HR), 95% CI, and log‐rank p‐value that are shown were obtained from comparing patients with any psychiatric disorder to patients with no psychiatric disorder (reference group).

FIGURE 3

Associations between any psychiatric disorder and COVID‐19‐associated hospitalization, stratified by vaccination status. A hazard ratio (HR) > 1.0 indicates that any psychiatric disorder was associated with a higher risk of COVID‐19‐associated hospitalization. HRs were adjusted for site, age (natural spline with four knots), sex (male, female, unknown), race and ethnicity (Asian, Black, Hispanic, White, other, unknown), Medicaid coverage (yes, no, unknown), underlying respiratory condition (yes, no), underlying non‐respiratory condition (yes, no), number of underlying medical conditions (square‐root transformed), and number of SARS‐CoV‐2 test records documented in the patient's electronic medical record prior to the start of the study period (0, 1, 2–4, ≥5). All HRs except for those for four doses were obtained from the same model with the exposure variable defined by patients' vaccination status (time‐varying) and psychiatric disorder status (not time‐varying). The analysis for four doses 7–59 days earlier was limited to person‐time after April 5, 2022 among patients aged ≥50 years. The analysis for four doses ≥60 days earlier was limited to person‐time after May 28, 2022 among patients aged ≥50 years. CI, confidence interval.

FIGURE 4

Associations between vaccination status and COVID‐19‐associated hospitalization, stratified by age group and psychiatric disorder status. A hazard ratio (HR) < 1.0 indicates that being vaccinated versus unvaccinated was associated with a lower risk of COVID‐19‐associated hospitalization. HRs were adjusted for site, age (natural spline with four knots), sex (male, female, unknown), race and ethnicity (Asian, Black, Hispanic, White, other, unknown), Medicaid coverage (yes, no, unknown), underlying respiratory condition (yes, no), underlying non‐respiratory condition (yes, no), number of underlying medical conditions (square‐root transformed), and number of SARS‐CoV‐2 test records documented in the patient's electronic medical record prior to the start of the study period (0, 1, 2–4, ≥5). For four doses 7–59 days earlier, only person‐time after April 5, 2022, among patients aged ≥50 years was analyzed. For four doses ≥60 days earlier, only person‐time after May 28, 2022, among patients aged ≥50 years was analyzed. HRs were not calculated (NC) for four doses 7–59 days earlier or four doses ≥60 days earlier in ages 18–64 due to the limited number of hospitalizations among patients aged 50–64 years in those categories. Vaccine effectiveness (VE) for prevention of COVID‐19‐associated hospitalization was estimated from HRs using the equation: VE = (1 − HR) × 100%. Vaccination status was defined as a time‐varying variable. Throughout follow‐up, 903,142 patients (37.1%) transitioned from one vaccination status to another at least once, including 289,477 (11.9%) with a new dose and 889,113 (36.5%) surpassing the cutoff of 150, 120, or 60 days since second, third, or fourth dose, respectively. At the end of follow‐up, 1,006,142 (41.3%) remained unvaccinated (patients with no psychiatric disorder, 43.8%; patients with any psychiatric disorder, 32.4%), 687,426 (28.2%) had received two doses (patients with no psychiatric disorder, 26.9%; patients with any psychiatric disorder, 32.8%) (median days since second dose, 482; IQR, 397–518), 646,315 (26.5%) had received three doses (patients with no psychiatric disorder, 25.4%; patients with any psychiatric disorder, 30.6%) (median days since third dose, 270; IQR: 243–300), and 97,113 (4.0%) had received four doses (patients with no psychiatric disorder, 3.9%; patients with any psychiatric disorder, 4.3%) (median days since fourth dose, 110; IQR: 73–134). CI, confidence interval.