A total of 14,003 individuals (≥ 18 years of age) who had a SARS-CoV-2 test between March 2020 – May 2022 in the Meuhedet HMO were included in the final study population (See Fig. 1 for the flow diagram of study participants). Of these, 1273 were diagnosed with schizophrenia and these were matched by age and sex to 12,730 individuals without schizophrenia. Sociodemographic characteristics, smoking status, and BMI of the study population are presented in Table 1. The average age was 48.05 years (± 14.92) and 2.2% of the participants were aged 80 years and above. Males comprised 58.8% of the study population. Approximately 26.4% of individuals diagnosed with schizophrenia belonged to the Ultra-Orthodox Jewish subpopulation, compared to 21.6% of the control group (p-value < 0.001), 9.1% of people with schizophrenia and 16.5% of the controls belonged to the Arab minority population (p-value < 0.001). Among patients with SSD, 30.3% were smokers, compared to only 17.4% of controls (p-value < 0.001). BMI for patients with SSD was 29 kg/m2 ± 6.2, compared to 27.3 kg/m2 ± 5.3 in the control group. A higher percentage of individuals with schizophrenia (42.1%) were from a low socioeconomic position, compared to 36.1% of the control group.

Flowchart of 14,003 individuals included in this study

Comorbidities were more frequently observed in individuals with schizophrenia, the most common being depression (56.4% vs. 13/%, p = value < 0.001), type 2 diabetes mellitus (DM) (19.6% vs. 9.7%, p-value < 0.001), hypothyroidism (13.4% vs. 7.5%, p-value < 0.001), COPD (12.5% vs. 5%, p-value < 0.001), hypertension (10.1% vs.7.5%, p-value < 0.001). Among individuals with schizophrenia, 26.9% had no comorbidities, 37.6% had one, 18.1% had two, and 17.4% had three or more comorbidities. Additional comparisons are presented in Table 2. The association between individual medical comorbidities with COVID-19-related hospitalizations and mortality is presented in Table S1 and Table S2.

A total of 11,680 (83.4%) of the study population were vaccinated within the first round of vaccination, among these, 1015 (80.0%) were patients with SSD, a significantly lower percentage than among the controls (84%, p < 0.001). However, there was no statistically significantly difference for the second and third rounds (p = 0.188 and p-value = 0.569, respectively) (Table 2).

Table 3 illustrates SARS-CoV-2 infection sequelae. Compared to controls, patients with SSD were less likely to have a positive test for SARS-CoV-2 (37.5% vs. 48.7%, p-value < 0.001). Among patients with a positive SARS-CoV-2 test hospitalization, length of hospitalization, and all-cause 30-day and one-year mortality were more frequently observed among schizophrenia patients compared to healthy controls (all p < 0.001). Among patients with SSD who tested positive, 230 (48.2%) were hospitalized, and 54 (11.3%) were hospitalized for two or more days. Figure 2 illustrates that the median length of hospitalization for patients with SSD was more than two days compared to less than two days in the control group. Among patients with SSD following COVID-19, one-year mortality was 7.8%, compared to 1.5% in the comparison population (p-value < 0.001). Only three (0.6%) schizophrenia patients were admitted to an ICU (p = 0.307, Table 3), and hence, we did not further analyze this outcome further in multivariable models.

Box Plot showing the length of hospitalization of people with (blue box) and without schizophrenia (red box)

Schizophrenia was significantly associated with hospitalization (OR = 5.18, 95%CI = 3.90–6.89, p < 0.001) (Table 4). In a multivariable-adjusted logistic regression model controlling for sex, age, sector, SES, BMI, smoking, and the number of medical comorbidities, schizophrenia was independently associated with increased hospitalization (OR = 4.49, 95%CI = 3.29–6.13, p < 0.001) (Table 4). Other factors related to hospitalization included multiple comorbidities, advanced age, smoking, and SES. The association between schizophrenia and COVID-19 hospitalization remained significant but was attenuated after adding vaccination to the model (OR = 3.80, 95%CI = 2.70–5.34, p < 0.001) (Table S6), even though vaccination had a dose-responsive protective effect.

Schizophrenia patients experienced a longer length-of-stay after being admitted to the hospital following SARS-CoV-2 positive test (β = 1.24, p = 0.003) (Table 4). After multivariable adjustment for sex, age, sector, SES, BMI, smoking, and number of medical comorbidities, a multiple linear regression analysis showed a statistically significant association between schizophrenia and log-transformed number of hospital days (β = 1.2, p < 0.001) (Table 4). This association remained significant when performing a Poisson regression. After further adjustment for vaccination, the association between schizophrenia and length of hospitalization maintained its statistical significance (β = 1.18, p < 0.001) (Table 4). Furthermore, vaccination had a protective effect on the length of hospitalization (Table S6).

Individuals with schizophrenia had significantly higher odds of 30-day mortality compared to those without schizophrenia (OR = 7.54, 95%CI: 3.15–18.07, p < 0.001) (Table 4). The association was strengthened (OR = 9.07, 95%CI: 3.11–26.44, p < 0.001) after adjustment for sex, sector, SES, age, BMI, smoking, and the number of comorbidities (Table 4). The association between schizophrenia and 30-day mortality was attenuated after further including vaccination (OR = 4.54, 95%CI: 1.54–13.38, p = 0.006) (Table 4). Advanced age, multiple comorbidities and smoking, as well as sector and SES were other factors related to 30-day mortality (Table S5).

Schizophrenia was significantly associated with increased one-year mortality (OR = 7.78, 95%CI: 4.00–15.15, p < 0.001) (Table 4). This finding remained statistically significant in a multivariate logistic regression, adjusting for sex, sector, SES, age, BMI, smoking, and number of comorbidities (OR = 6.27, 95%CI: 2.73–14.39, p < 0.001) (Table 4). The association was maintained after the effect of vaccination was taken into consideration (OR = 4.98, 95%CI: 2.23–11.15, p < 0.001) (Table 4). Vaccination had a protective effect (Table S6).

Formal testing using multivariate logistic regression confirmed the presence of a statistically significant interaction between vaccination status and SSD (P < 0.001).

In stratified age-adjusted models, among individuals who had received vaccination, the effect of schizophrenia on 30-day mortality was attenuated (OR = 4.79, 95%CI = 0.82–28.13, p = 0.082), whereas, in unvaccinated individuals, schizophrenia had a statistically significant effect on 30-day mortality (OR = 7.53, 95%CI = 2.19–25.95, p = 0.001) (Table 5). However, vaccination did not appear to attenuate the link between schizophrenia and one-year mortality. The association of schizophrenia on one-year mortality was evident both in vaccinated (OR = 10.24, 95%CI = 3.25–32.31, p < 0.001) and unvaccinated individuals (OR = 5.76, 95%CI = 2.13–15.60, p = 0.001) (Table 5).