The Texas CARES population and study design are described in detail elsewhere [9,10]; the program began enrolling a convenience sample of adults and children from across Texas in October 2020. Participants provided serum samples at approximately three-month intervals, which were tested for anti-SARS-CoV-2 spike and nucleocapsid proteins using the Roche Elecsys® Anti-SARS-CoV-2 quantitative S test and qualitative N test assays, respectively. Participants completed questionnaires at each test timepoint to capture demographic and clinical characteristics, including COVID-19 infection history, vaccination status, and medical history. The program includes innovative return of results processes to inform and engage study participants. As of February 2023, Texas CARES has enrolled 90,011 participants, of whom 40,572 have completed at least four antibody testing timepoints.

Beginning in late November of 2021, certain participants contacted program staff with concerns about their Anti-SARS-CoV-2 spike protein results, having noticed a precipitous decrease in reported antibody concentration between consecutive test dates, inconsistent with expected antibody decay. In response to participant concerns, the program conducted a quality audit, in which 40 participants’ samples collected from November 2021 –February 2022, were retested according to the original Roche instructions for use (IFU) [11]. All retested samples produced antibody levels consistent with the original result, excluding laboratory error as the source of the anomalous results. Investigators considered and excluded mitigating factors from natural or artificial immunosuppression through dialogues with affected participants. The program found that affected participants frequently reported natural infections and/or immunization boosters and had previously reported high antibody results, triggering the consideration of a (at the time) previously uncharacterized prozone effect.

To evaluate possible hook effect in 3 specimens that reproducibly tested in the analytical measurable range (AMR) of the assay (0.4–250 U/mL with onboard 10X automated dilution, clinical reportable range (CRR): 0.4–2500 U/mL), the laboratory re-evaluated with initial predilution of 1:5 and 1:50 prior to testing in accordance with the IFU. Final endpoint results of samples pre-diluted at 1:50 were above the clinical reportable maximum of 2500 U/mL, supporting the prozone effect theory. In August 2022, the program conducted an additional confirmation study, identifying participants who had recently received low antibody titers (0.8–250 U/mL reported result) immediately following a high antibody titer (>2500 U/mL reported result). Study staff then requested new serum samples from eleven of these participants for repeat testing using the 1:50 pre-dilution protocol.

Beginning October 12, 2022, all study samples with initial neat results of 0.8–250 U/mL were subjected to a dilution algorithm with initial retest at 1:10 and reviewed for possible prozone effect (Fig 1). Initial (neat) results greater than or equal to 0.8 U/mL and less than 250 U/mL underwent 1:10 dilution. The dilution-adjusted values were compared to the initial neat values, and if the two values were within 20% of each other, the initial neat result was reported. If the dilution-adjusted value exhibited a greater than 20% difference from the initial neat value, the sample was classified as a prozone case. Among these prozone cases, if the dilution-adjusted value was greater than or equal to 2500 U/mL, a value of >2500 U/mL was reported to participants; if the dilution-adjusted value was less than 2500 U/mL, an additional 1:50 dilution was performed to rule out persistent prozone effect.

Laboratory values were reported as the median and interquartile range (IQR) for continuous variables. All analyses were performed on Stata MP version 17.0 (StataCorp LLC, College Station, TX, USA).

All protocols were reviewed and approved by the UTHealth Committee for Protection of Human Subjects, Institutional Review Board: # HSC-SPH-20-0825, but also were deemed public health practice by the Texas Department of State Health Services IRB. All adults consented electronically to be in the study, and children under 18 either consented or assented to be in the study as appropriate for their age. The catchment area for the cohort study was the entire state of Texas. Recruitment efforts were taken to enroll rural and urban participants spanning the state. More details about the study are publicly available on the Texas CARES dashboard [12].