This study was developed using the JH-CROWN: COVID Precision Medicine Analytics Platform Registry, which includes inpatient and outpatient records for patients who tested positive for SARS-CoV-2 in the Johns Hopkins Health System (JHHS). The JHHS comprises six hospitals and over forty outpatient facilities located in Maryland, Florida and Washington, DC. The research was approved by the Johns Hopkins Institutional Review Board. Patients hospitalized within the JHHS for COVID-19 infection during the period of March 1, 2020 until November 30, 2021 were eligible for inclusion. A query identified patients who were previously treated with B-cell depletion therapy on or after September 1, 2019 and who were subsequently hospitalized in the Johns Hopkins Health System with a laboratory-confirmed SARS-CoV-2 infection. No common platform for SARS-CoV-2 testing was employed during the period of study. B-cell depletion therapy was defined as rituximab, ocrelizumab, obinutuzumab, ibritumomab tiuxetan, and ofatumumab. A full list of medication search terms is included in the Appendix. Individual charts were reviewed to ensure that only patients who previously received B-cell depletion therapy on or after September 1, 2019 and prior to hospitalization with SARS-CoV-2 infection were included in the analysis. September 1, 2019 was chosen as it is six months prior to the start of COVID-19 hospitalizations among JHHS patients and would reflect the biologic effects of prior B-cell therapy. JHHS instituted mandatory screening of hospitalized patients for SARS-CoV-2 during the study period; patients with an incidentally positive SARS-CoV-2 test were eligible for analysis. The need to obtain informed consent was waived by the Johns Hopkins University School of Medicine Institutional Review Board (IRB00247569).

Outcomes

The primary outcome was time to death from hospital admission. The secondary outcome was time to a composite outcome of severe illness or death from hospital admission. An additional secondary outcome was time to clinical improvement from the date of hospital admission, defined as a 2-point decrease in World Health Organization (WHO) severity score or discharge alive from the hospital within 30 days. The WHO Severity score is a 10-point scale ranging from 0 (uninfected) to 10 (deceased)¹⁴.

Disease severity at admission was defined by the highest WHO severity score in the first 12 h of the patient’s admission. Severe illness was defined as score ≥ 6. Failure of clinical improvement was censored at the last day of follow-up or 30 days, whichever came first. Patients who were discharged alive were censored at 30 days.

Statistical analysis

A two-step, hybrid approach was used to account for the differences between the groups with and without B-cell therapy. In the first step, patients hospitalized with COVID-19 who received prior B-cell treatment were identified. Then a comparison group from the registry of patients who had not received B-cell therapy but matched on characteristics with the treated patient was selected. Matching criteria included age at time of hospitalization (± 5 years), timing of hospitalization (± 28 days), sex at birth, race, WHO COVID-19 disease severity, COVID-19 specific treatments (remdesivir, dexamethasone), and vaccination status before admission. All the patients who were matched to the b-cell cohort were included in the comparison group.

The second step employed overlap weighting, a propensity score (PS) approach, to further address differences between the B-cell treatment population and controls¹⁵. The PS score, indicating the likelihood of receiving B-cell depletion treatment, was derived from age, race, sex, body mass index (BMI), COVID-19 severity on the WHO scale, smoking status, existing comorbidities, specific COVID-19 treatments, and vaccination status at the time of admission. Medical comorbidities included hypertension, chronic obstructive lung disease, congestive heart failure, chronic kidney disease, end stage renal disease, human immunodeficiency virus, cirrhosis and cancer. Comorbidities were determined from relevant ICD codes present in the medical record. Use of overlap weighting based on the propensity score estimated from a logistic model leads to an exact balance for all covariates between different groups^16,17.

Finally, inverse probability weighted Cox regression was used to estimate the association between B-cell depletion treatment and outcomes of interests. Age, BMI, COVID-19 specific medications and vaccination information were included as covariates in Cox regression models based on clinical interest and knowledge. The proportional hazards assumption was tested for violation using individual covariates and each outcome. A prespecified subgroup analysis was conducted to examine effects in patients with B-cell depletion treatment less than 90 days prior to COVID-19 hospitalization. For the sensitivity analysis we used the original control group and reperformed overlap weighting for the subgroup. All data analyses were executed using R software version 4.0.2 (R Foundation for Statistical Computing, Vienna, Austria)¹⁸.