An active surveillance program was developed by the Pharmacovigilance Technical Committee (PhVTC) of the Hospital Clinic of Barcelona. The PhVTC is a multidisciplinary group with representatives from Clinical Pharmacology, Allergology, Pharmacy, Dermatology, Preventive Medicine, Immunology, Hospital Quality Management, and Nursing Professionals. Under the mandate of the PhVTC, the Department of Clinical Pharmacology, with the close collaboration of the Departments of Preventive Medicine and Epidemiology and the Department of Occupational Health Care, designed and implemented an active surveillance program with specific procedures for vaccine safety monitoring and integrated care processes for adverse event clinical management and reporting to the regulatory authorities. Vaccinated persons were informed of the program with a leaflet describing the objectives and methods before vaccination at the vaccination center. The information was explained again when calls were made after each dose. The survey was only delivered if the person agreed. The whole procedure was explained in the protocols assessed and approved by the Research Ethics Committee of the Hospital Clinic (references: HCB/2021/0684 and HCB/2021/0685).

The active surveillance program was designed as a prospective, observational, single-center post-authorization study aimed at characterizing short-term ARs after vaccination. The study was reported following the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines for cross-sectional studies [21] (see eTable 15 the Electronic Supplementary Material (ESM)).

Vaccination Campaign

The vaccination campaign began in the Hospital Clinic of Barcelona, a tertiary care Hospital, on 7 January 2021. The first group to be vaccinated were HCPs. SOTR patients started vaccination on 2 March 2021. The type of vaccine administered was determined by local authority guidance and availability depending on the national supply. Initially, BNT162b2 was the only vaccine available for HCPs. The vaccination plan for HCPs was amended on 22 January with the inclusion of mRNA-1273. Vaccination of SOTRs began on 1 March with mRNA-1273. Two doses were the usual vaccine dose at the time the study was conducted. As of May 2022, three doses in the general population and four doses in immunocompromised patients are recommended by the Spanish Ministry of Health [22].

Active Surveillance Program

The surveillance program was substantiated by a previously agreed pharmacovigilance plan that aimed to cover all vaccinated staff of the hospital and affiliated institutions (Institut d’Investigacions Biomèdiques August Pi i Sunyer (IDIBAPS) and Fundació Clinic per a la Recerca Biomèdica (FCRB)) and SOTR patients from 7 January until 30 April 2021. The primary objective was to make a comparative descriptive assessment of the safety profile and reactogenicity of the two vaccines.

A structured questionnaire was designed containing demographic and medical data and ARs occurring after each vaccine dose.

A predefined list of ARs was agreed on based on the safety profile described in the Summary of Product Characteristics (SmPC) and the European Public Assessment Report (EPAR) of each vaccine [23,24,25,26] (see ESM eTable 13). These were considered solicited ARs. In addition, some ARs of special interest from the list of the ACCESS [3] and SPEAC Projects [5] and the Brighton Collaboration criteria were included in the predefined list of solicited ARs [27]. The following solicited ARs were included (as per Preferred Term (PT) level of the version 24.0 of the Medical Dictionary for Regulatory Activities (MedDRA): Injection site pain (PT: Injection site pain); arm pain (PT: pain in extremity); fatigue, asthenia (PT: fatigue); headache (PT: headache); fever (PT: pyrexia); muscle pain (PT: myalgia); malaise (PT: malaise); joint pain (PT: arthralgia); injection site swelling (PT: application site swelling); injection site redness (PT: application site redness); shoulder pain (PT: musculoskeletal pain); injection site pruritus (PT: vaccination site pruritus); headache (PT: headache); insomnia (PT: insomnia); nausea (PT: nausea); diarrhea (PT: diarrhea); and vomiting (PT: vomiting) (see ESM eTable 13).

A free-text option was provided to collect other adverse events not considered a priori in the predefined list. Reactions collected within the free-text option were considered unsolicited ARs.

Severe ARs were defined by intensity assessed on a Likert scale, with values ranging from 0 to 10, rated using instructions from the interviewer (see ESM eFig. 1).

Delivery of the QuestionnairePreparatory Process

Nine professionals from the Medical Division were trained ad hoc and delivered the questionnaire through phone calls. A guiding document was drafted, which aimed to homogenize its administration. Simulations with ten dummy cases were performed to homogenize the delivery process amongst interviewers.

Implementation

Calls were made after each dose. At least three phone call attempts were made to each interviewee to maximize responsiveness. For HCPs, if we did not obtain a response on at least two separate calendar days, interviewees were then contacted at their institutional email addresses and asked whether: (i) They wanted to be further contacted by phone at a preferred phone number and timeslot; (ii) they preferred to be contacted and answer the questionnaire by email; (iii) they did not want to be contacted further. For interviewees who chose to answer the questionnaire by email, an electronic version was enabled, constructed using Lyme Survey technology (enquesta.clinic.cat).

During the implementation process, there were regular team meetings for doubt resolution and consistency of data recording.

Variables Assessed

The structure of the questionnaire was based on two blocks, which were completed after the first and second doses for all participants.

The first block contained common background variables that were completed by all participants (HCPs and SOTRs) and specific background variables for each vaccinated group. The second block contained vaccine- and AR-related variables, which were completed by all participants.

First Block

Common background variables: Age, sex, medical background; SARS-CoV-2 infection history (positive PCR or antigen test, positive serology, and/or symptomatic infection with or without hospital admission).

Specific background variables: For HCPs, contact with SARS-CoV-2 patients in their daily activity (yes/no) was included in this first block. In case of SOTRs, type of transplant (liver, heart, kidney, kidney-pancreas) was considered.

Second Block

Vaccine and AR related variables: Date of vaccination; type of vaccine administered; AR start date, end date (where available), reported AR, and intensity of AR. Types of ARs were selected from a predefined list in the questionnaire. The interviewer asked a closed-question to assess whether the participant had had the AR (e.g., “Did you have fever”?). The interviewer recorded yes or no. These ARs were considered “solicited ARs.”

A free-text option called “other” was enabled to collect other ARs not considered a priori in the drop-down list. To explore “other ARs” the interviewer asked an open question (e.g., Have you had other types of AR?). Yes or no was recorded accordingly. If yes, the specific type of AR was recorded as a narrative description. These AR were considered “unsolicited ARs.”

Data Collection and Data Management for Clinical Follow-Up

A database was designed and implemented with MACROTM, which provided electronic data capture functionality (EDC) to support electronic data entry. Data completion and integrity of the database were assessed continuously to verify the information consistency and completeness. Hospital electronic records were reviewed to capture potential reactions not collected by the structured interview and to mitigate recall bias.

Solicited and unsolicited ARs were coded using the PT and System Organ Class (SOC) categories of version 24.0 of MedDRA. There was considerable heterogeneity in the way unsolicited ARs were reported and recorded. Therefore, unsolicited ARs were grouped within the closest related SOC and PT.

Statistical AnalysisSample Size Calculation

There was no formal sample size calculation. We included all vaccinated HCPs and SOTRs who did not decline taking part in the survey.

Planned Analyses and Assessed Variables

Three groups were considered for analysis: HCPs vaccinated with BNT162b2, HCPs vaccinated with mRNA-1273, and SOTRs vaccinated with mRNA-1273. Two main comparisons were planned within these three groups: (i) A comparative analysis of the vaccine tolerability and reactogenicity profile between vaccines in HCPs, after the first and second doses; (ii) a comparative analysis of the tolerability and reactogenicity of vaccination between HCPs and SOTRs vaccinated with mRNA-1273, after the first and second vaccine doses. Tolerability was assessed by a Likert scale ranging from 0 to 10. Reactogenicity was assessed by the proportion of ARs in each group.

The key secondary endpoint was to describe solicited and unsolicited ARs recorded in each group, according to reported frequencies and intensity distribution. Comparison of the risk of any severe AR (Likert scale score 7–10) in the three study groups was also considered as an exploratory endpoint.

Statistical Analysis

Categorical variables were described as frequencies and percentages and continuous variables as mean ± standard deviation or median (25–75% interquartile range), as appropriate. Categorical data were compared using the chi-square test and continuous variables using ANOVA with rank-transformed data. To assess baseline homogeneity, we used standardized differences (STDs, differences between groups divided by pooled standard deviation). To compare vaccination in HCPs, the inverse probability of the treatment weights (IPTWs) approach [28] was used to create a pseudo-population in which the two groups were balanced across baseline covariates. The stabilized weights were calculated using propensity scores (PS) [29] aimed to minimize the between-arm standardized differences [30]. Covariate balance was assessed using STDs to achieve absolute values < 0.20 (|< 20%|), which are acceptable values of variability [31, 32]. Variables included in the IPTW construction were comorbidities; previous exposure to SAR-CoV-2; job position and SARS-CoV-2 occupational contact with patients; sex; age (per 1-year strata).

STDs were also calculated to compare HCPs and SOTRs and compare the three study groups. However, IPTW was not applied in the HCP and SOTR comparisons as they were very different populations and the key covariates inherent in each group were not modifiable, and the main outcome was to describe whether the same vaccine was tolerated differently in the two populations. The primary analysis of the comparison of the risk of ARs between vaccines in HCPs (BNT162b2 and mRNA-1273) after the first and second doses was based on IPTW methods. Sensitivity analyses were performed considering raw unadjusted data and covariate-adjusted data. Raw and adjusted regression models were used to estimate odds ratios (ORs) with 95% confidence intervals (CIs) for all model variables. The area under the curve of the receiver operating characteristics (AUC ROC) curve was calculated to assess model performance.

The primary analysis of the comparison of the risk of ARs between HCPs and SOTRs vaccinated with mRNA-1273 was based on covariate-adjusted data using only age, sex, previous SARS-CoV-2 infection, and a history of drug allergy. Adjusted regression models were used to estimate OR with 95% CI for all model variables. The AUC of the ROC curve was calculated to assess the performance of the model.

To analyze the key secondary endpoint, the rate of vaccinated persons with ≥ 1 AR was calculated for each vaccine and dose. The rate of each individual AR (solicited plus unsolicited) was calculated. Intensity was categorized, according to Likert scale values, as low grade or mild (grade 1; score 0–3), moderate grade (grade 2; score 4-6) and severe grade (grade 3; score 7-10). For each type of AR, the rate of vaccinated personnel with grade 3 ARs and median intensity was calculated.

In all analyses, we applied a two-sided type I error of 5%. SAS v9.4 (Cary, NC, USA) software was used throughout.