COVID-19 Prevalence and Mortality Among Schizophrenia Patients

A Large-Scale Retrospective Cohort Study

Dana Tzur Bitan; Israel Krieger; Khalaf Kridin; Doron Komantscher; Yochai Scheinman; Orly Weinstein; Arnon Dov Cohen; Assi Albert Cicurel; Daniel Feingold


Schizophr Bull. 2021;47(5):1211-1217. 

In This Article


Data Source

The study utilized the databases of the Clalit Health Organization (CHS), the largest operating healthcare organization in Israel.[11] The CHS is 1 of 4 operating healthcare organizations to provide healthcare for all citizens of Israel and covers nearly 5 million citizens, over 50% of the country's population,[12] through 1427 primary healthcare units and additional medical facilities. Clalit Health Services regularly updates its databases with real-time information derived from medical care facilities and hospitals, pharmacies, and administrative medical operating systems. The data mining of the current database was based on the registries of the CHS, which undergo periodic validation processes such as direct comparisons between various diagnostic sources and random evaluations of medical files. Diagnoses and medical fields were obtained through the chronic diseases registry, which has been previously validated by the developers of its algorithms,[13] as well as by many other authors utilizing the database.[14–17] The establishment of the computerized systems was initiated in 2000 and is continuously updated with clinical information. The extraction of the current database, containing schizophrenia patients and their matched controlled cases, was originally mined at the end of 2017.[18] Since the beginning of the pandemic, patients have been able to be tested for COVID-19 through their primary care clinics, in mass drive-in facilities initiated by the Ministry of Health and local city municipalities, through first-aid mobile staffs, and in general and psychiatric hospitals. COVID-19 information is automatically synchronized with the CHS, the Ministry of Health, and primary care databases through the registered, licensed laboratories in Israel authorized to analyze polymerase chain reaction (PCR) panels from all sources performing the tests. For the purposes of the current study, the database was updated for COVID-19 and demographic fields in October 2020. This update included mortality rates; therefore, individuals who were deceased prior to the onset of the pandemic were removed from the database along with their matched controls.

Definition of Study Variables

Diagnosis of schizophrenia was based on a community senior psychiatrist's medical registration or if listed on a psychiatric hospital's discharge letter. To qualify for a schizophrenia diagnosis, the file had to contain one of the ICD-9 codes for schizophrenia (code 295) or one of the ICD-10 codes for schizophrenia (code F20).[19,20] Validation of accuracy of schizophrenia diagnosis had been previously performed by an expert senior psychiatrist who reviewed the clinical files of 10% of the sample and found a 94% accuracy rate.[18] Control group participants comprised random cases from the population of CHS-insured citizens, which were matched by age and gender to the schizophrenia group. These included individuals with no diagnosis of schizophrenia who were insured by the CHS at the time of data mining and randomly sampled at a 1:1 ratio. Age was matched with a maximum allowable difference of 1 year, and sex was matched at a 1:1 ratio.

COVID-19 characteristics included results of COVID-19 tests, number of tests taken, hospitalizations, and deaths. In Israel, multiple diagnostic polymerase chain reaction (rRT-PCR) panels, which were approved by the Ministry of Health, are utilized, such as Xpert Xpress SARS-CoV-2 (Cepheid) and the real-time fluorescent RT-PCR kit for detecting 2019-nCoV (BGI Genomics). Test results were available and retrieved from March 2020 to October 2020. Referrals for PCR tests were initially given if patients reported COVID-19 symptoms or were in close contact with COVID-19 positive patients. After the first lockdown in April 2020, alongside the expansion of the national capacity to perform PCR testing for COVID-19, indications for testing were expanded and included early testing of asymptomatic direct contacts of COVID cases, periodic screening of health workers working with vulnerable populations, screening of patients before elective surgery, and screening of patients before admission for long-term psychiatric hospitalization. Hospitalization data were reported by general hospitals to the Ministry of Health and synchronized with the CHS databases, and death reports were derived from hospital discharge letters and were also verified by cross-examination with the Interior Ministry databases.

The following clinical conditions were examined as potential risk factors for COVID-19 hospitalization and death: smoking (ICD-9 code 3051, V1582), which refers to a chronic diagnosis as registered by a community physician or as listed in the diagnoses of discharge letters, obesity (ICD-9 codes 278, 75981), diabetes (ICD-9 codes 250, 3620), hypertension (ICD-9 codes 401–405), hyperlipidemia (ICD-9 codes 2720), chronic obstructive pulmonary disease (COPD, ICD-9 codes of 4912, 4920, 4928, 496 and 515) and ischemic heart disease (IHD, ICD-9 codes of 410–414, 429, V458, Z360-Z261). ICD codes were also verified using textual identification and lab test results. Socioeconomic status was obtained using the CHS index, calculated as an index score combining information from social services, as well as sociodemographic variables such as district and current address. Sector was used to denote the main sub-populations in Israel (general, Ultraorthodox, and Arab). The study was reviewed and approved by the CHS institutional review board (IRB), where informed consent was waived due to the nature of data extraction.

Statistical Analysis

Differences in prevalence of sociodemographic characteristics, medical conditions, and COVID-19 parameters between the schizophrenia and control group were examined using univariate logistic regressions for categorical variables and t-tests for continuous variables. Univariate and multivariate logistic regressions were employed to examine the odds for COVID-19 testing, COVID-19 positive cases, and COVID-19 hospitalization and mortality, among individuals with schizophrenia compared to controls, while reporting the base-adjusted model (adjusted for age and sex), the model accounting for sociodemographic and socioeconomic factors (marital status, sector, and socioeconomic status), and the model accounting for clinical risk factors (obesity, smoking, diabetes, hyperlipidemia, COPD, and IHD). Odds ratios and 95% confidence intervals were reported, as well as significance levels as set to 5%. All statistical analysis was performed using SPSS software, version 25 (SPSS).