Objectives and study design
The primary objective of this study was to estimate the VE of the CoronaVac vaccine against COVID-19 in pregnant women. We did that by conducting a test-negative design (TND) in all pregnant women in Brazil who had an RT-PCR test for COVID-19. We estimated the effectiveness of the vaccine against symptomatic COVID by comparing women’s vaccine status in symptomatic women with a positive test to vaccine status in those with a negative test and the effectiveness against severe COVID-19 (by vaccine status in those with severe, hospitalised or fatal COVID-19 with the vaccine status of those with a negative test).
All data used was abstracted from 3 routinely collected sources: the national surveillance system for RT-PCR test for COVID-19 (e-SUS Notifica), the information system for severe acute respiratory illness (SIVEP-Gripe) and the national immunisation system (SI-PNI).
This database contains information on all suspected cases of COVID-19 recorded in the country. It includes all positive and negative RT-PCR test results and information on residence, demographic and clinical data of individuals, such as the presence of co-morbidities and pregnancy status (so we can identify women registered during pregnancy) and presence of symptoms, with acute respiratory diseases defined as the presence of at least two of the following signs and symptoms: fever (even if referred), chills, sore throat, headache, cough, runny nose and loss or change to a sense of smell or taste . Asymptomatic individuals with an RT-PCR test were not included in this study, independent of the test result.
SIVEP-Gripe is the national register for severe acute respiratory syndrome (SARS) in Brazil, created after the influenza pandemic of 2009. In 2020, it was expanded to include COVID-19. All COVID-19 hospitalisations and deaths are meant to be registered in this system . In SIVEP-Gripe, severe acute respiratory illness is defined as an individual with acute respiratory disease who presents dyspnea/respiratory discomfort, persistent pressure or pain in the chest and oxygen saturation less than 95% without oxygen or cyanosis of the lips or face . Individuals who died with severe acute respiratory illness independent of hospitalisation are also registered. By linking these data with e-SUS Notifica, we identified which pregnant women in e-SUSNotify with a positive RT-PCR test progressed to severe disease.
SI-PNI contains data on all vaccines administered in Brazil. COVID-19 vaccines are administered by health services and recorded in point-of-care applications . From SI-PNI, we extracted information on which COVID-19 vaccine was received with first and second doses dates. By linking these data with the data on pregnant women in the other files, we were able to determine (i) which pregnant women who tested negative for COVID-19 had been vaccinated, (ii) which pregnant women with confirmed symptomatic COVID-19 infections had been vaccinated and (iii) which pregnant women with severe COVID-19-associated severe case had been vaccinated. We assumed that pregnant women whose records did not link to a SI-PNI vaccination record were not vaccinated.
All data were extracted on October 05, 2021, and made available by the Brazilian Ministry of Health. The information technology bureau of the Brazilian Ministry of Health provided pseudo-anonymised data with a common unique identifier that were used to link individual-level records from the three databases (more details about linkage procedures are available at https://vigivac.fiocruz.br/).
All pregnant women with symptoms suggesting COVID-19, aged between 18 and 49 years in Brazil with a record of an RT-PCR test between March 15, 2021, and October 03, 2021, registered in e-SUS Notifica. Variants of concern have played an important role in the SARS-CoV-2 pandemic in Brazil. From March to July 2021, the gamma variant accounted for most of the SARS-CoV-2 isolates genotyped in Brazil; from August to October, the delta variant was predominant .
Testing for COVID-19 in Brazil is accessible to anyone through the universal public health system (SUS). Subjects who received any other COVID-19 vaccine were excluded: ChAdOx1 nCoV-19 or Ad26.COV2.S (Janssen/Johnson & Johnson) because these are not indicated for pregnant women in Brazil, and BNT162b2 (Pfizer) because numbers of women with complete regimen were too small to allow evaluation given they were only included in the Brazilian program more recently, and the long interval between doses. As a result, the study is restricted to evaluating CoronaVac vaccine effectiveness. The population consisted of symptomatic pregnant women who were tested with RT-PCR for COVID-19 classified into 3 groups: RT-PCR test negative, RT-PCR test positive with COVID-19 symptoms and RT-PCR test positive with severe COVID-19. The study population in the TND included all symptomatic women with an RT-PCR irrespective of the test result.
Definition of outcome, cases and controls
The primary outcome was a positive RT-PCR test in a symptomatic subject. Cases were defined as all symptomatic women in the study population with an RT-PCR test result from a respiratory sample collected within 10 days after the onset of symptoms and who did not have a positive RT-PCR test result in the preceding 90 days. We also calculated VE against severe COVID-19, identified through notification to SIVEP-Gripe or with a register of hospitalisation or death in the e-SUS record. Controls were defined as all women in the study population with a negative RT-PCR test result and no positive RT-PCR test in the previous 90 days or in the subsequent 14 days. The test date was defined as either the date of collecting a respiratory specimen or the date of the case registration (when the test date was missing).
The exposure studied was vaccination with CoronaVac. This was classified into partially vaccinated (≥ 14 days after the first dose and before receipt of the second dose at the time of RT-PCR testing) and fully vaccinated (≥ 14 days after the second dose at the time of RT-PCR testing). We also calculated effectiveness in the period < 14 days since vaccination as the vaccine is expected to have no or limited effectiveness in the first 13 days since vaccination. This was used as a test as high effectiveness or increased risk during this period might indicate unmeasured bias or confounding. The reference group for vaccination status was the women who did not receive a first vaccine dose before the date of sample collection.
Several risk factors may be associated with both the likelihood of the exposure (i.e. receiving a vaccine) and the likelihood of receiving an RT-PCR SARS-CoV-2 test. These include age, ethnicity, co-morbidities status, geography location, index of deprivation , time (reflecting changes in vaccination policy and disease circulation) and presence of a previous COVID-19 notification as this may be related to vaccination the risk of a second COVID-19 infection. We extracted information on these potential confounders from the e-SUS Notifica.
The test-negative design is a type of case-control study in which the study population consist of the population tested, and controls are selected from those who have a negative test . Accordingly, it was analysed using the standard methods for case-control studies [26, 27]. Logistic regression was used to estimate the odds of vaccination with CoronaVac in RT-PCR test confirmed cases compared with those who tested negative and the odds of vaccination in the severe cases compared to those who tested negative. Individuals only contributed their first positive test result from March 15, 2021 (when the vaccination programme was recommended for pregnant women nationally). Week of RT-PCR test was included in the regression models because of the variations over time in both COVID-19 incidence and vaccine delivery in Brazil. We also adjusted for age (< 20, 20–34, ≥ 35), ethnicity (white, mixed brown, black and others), presence of registered co-morbidities, geography (region) and index of deprivation (quintile). Only those with complete information were included in the adjusted analyses. We estimated the VE as one minus the corresponding odds ratio (OR), obtained from a model including the described covariates, expressed as a percentage. Numbers were not sufficient so far to estimate interactions with time since vaccination or variants. We calculated the number of expected deaths in vaccinated women by applying the case fatality rate among the unvaccinated cases to the vaccinated cases, assuming age and presence of co-morbidities were similar.
Data analyses were performed in Stata version 17.0.
This study analysed de-identified data and was approved by the National Ethics committee (CONEP) (CAAE registration no. 50199321.9.0000.0040).