COVID-19 diagnosis, vaccination during pregnancy, and adverse pregnancy outcomes of 865,654 women in England and Wales: a population-based cohort study

Summary Background The extent to which COVID-19 diagnosis and vaccination during pregnancy are associated with risks of common and rare adverse pregnancy outcomes remains uncertain. We compared the incidence of adverse pregnancy outcomes in women with and without COVID-19 diagnosis and vaccination during pregnancy. Methods We studied population-scale linked electronic health records for women with singleton pregnancies in England and Wales from 1 August 2019 to 31 December 2021. This time period was divided at 8th December 2020 into pre-vaccination and vaccination roll-out eras. We calculated adjusted hazard ratios (HRs) for common and rare pregnancy outcomes according to the time since COVID-19 diagnosis and vaccination and by pregnancy trimester and COVID-19 variant. Findings Amongst 865,654 pregnant women, we recorded 60,134 (7%) COVID-19 diagnoses and 182,120 (21%) adverse pregnancy outcomes. COVID-19 diagnosis was associated with a higher risk of gestational diabetes (adjusted HR 1.22, 95% CI 1.18–1.26), gestational hypertension (1.16, 1.10–1.22), pre-eclampsia (1.20, 1.12–1.28), preterm birth (1.63, 1.57–1.69, and 1.68, 1.61–1.75 for spontaneous preterm), very preterm birth (2.04, 1.86–2.23), small for gestational age (1.12, 1.07–1.18), thrombotic venous events (1.85, 1.56–2.20) and stillbirth (only within 14-days since COVID-19 diagnosis, 3.39, 2.23–5.15). HRs were more pronounced in the pre-vaccination era, within 14-days since COVID-19 diagnosis, when COVID-19 diagnosis occurred in the 3rd trimester, and in the original variant era. There was no evidence to suggest COVID-19 vaccination during pregnancy was associated with a higher risk of adverse pregnancy outcomes. Instead, dose 1 of COVID-19 vaccine was associated with lower risks of preterm birth (0.90, 0.86–0.95), very preterm birth (0.84, 0.76–0.94), small for gestational age (0.93, 0.88–0.99), and stillbirth (0.67, 0.49–0.92). Interpretation Pregnant women with a COVID-19 diagnosis have higher risks of adverse pregnancy outcomes. These findings support recommendations towards high-priority vaccination against COVID-19 in pregnant women. Funding BHF, ESRC, Forte, HDR-UK, MRC, NIHR and VR.


Sample size justification
The sample size in this study included all pregnant women who gave birth in hospitals in England and Wales during the study period and were registered in a primary care practice.

Time varying exposures
Based on previous studies (e.g.Knight et al. 2022), we hypothesised that associations between COVID-19 diagnosis and adverse pregnancy outcomes would be strongest immediately after a COVID-19 diagnosis, and then decline with time since diagnosis (i.e., non-proportional hazards).We therefore calculated hazard ratios in separate time periods (in days: [0,14), [14+,]) after diagnosis of COVID-19, selected to ensure adequate numbers of adverse outcomes each time periods.

Adjustment for propensity score
To enhance efficiency in the estimation of magnitude of the associations, we incorporated a propensity score allowing for more flexible control of confounding; Specifically, we calculated the propensity score as the conditional probability of exposure, given the observed covariates, by using logistic regression models.To ensure model fitting and capture any nonlinear relationships between the propensity score and the outcome, we applied a spline transformation to the propensity score (3 knots) a necessary adjustment due to the high number of covariates (ethnic group (considering three categories for adjustment: White, Other ethnic groups, Unknown ethnic group), previous pregnancy (yes/no), smoking status (smoker, former smoker, nonsmoker), history of hypertension (yes/no), history of diabetes (yes/no), history of haematological and cardiovascular diseases (yes/no), overweight/obesity (yes/no), history of depression (yes/no) and other conditions (yes/no, including at least one of chronic obstructive pulmonary disease, liver disease, chronic kidney disease, cancer and surgical intervention).The models were further adjusted for age (continuous + quadratic term), deprivation index (quintile), calendar week (as a spline term with 3 knots), region in England.
Distribution of the probability of having COVID-19 diagnosis in COVID-19 cases and non-cases in England.
For the overall period and common outcomes (i.e., gestational diabetes and preterm birth) in England, we compared estimates adjusted with the propensity score approach to those from a traditional approach, which incorporated all potential confounders into the model.There was robust consistency between the two analytical methods: HR= 1.22 (95% CI: 1.18-1.26)with both method for gestational diabetes and HR=1.63 (1.57-1.69)using propensity score adjustment vs 1.64 (1.57-1.70)using covariate adjustment for preterm birth.

Gestational diabetes
Hyperglycemia (high blood glucose levels) that is first recognized during pregnancy.
Secondary and primary care records (ICD-10 and SNOMED-CT, Read Codes -validated by a clinician)

Gestational hypertension
The onset of hypertension (blood pressure ≥ 140/90 mmHg) after 20 weeks of gestation in women who were normotensive before pregnancy.
Secondary and primary care records (ICD-10 and SNOMED-CT, Read Codes -validated by a clinician)

Preeclampsia
A multisystem disorder of pregnancy marked by the onset of hypertension and proteinuria or hypertension and significant end-organ dysfunction with or without proteinuria, after the 20th week of gestation.
Secondary and primary care records (ICD-10 and SNOMED-CT, Read Codes -validated by a clinician)

Stillbirth
Fetal death that occurs at or beyond 24 weeks of gestation.This includes both deaths occurring before (antepartum death) and during (intrapartum death) delivery.
Secondary and primary care records (ICD-10 and SNOMED-CT, Read Codes -validated by clinician), information on stillbirth at the delivery, for pregnancy with estimated gestational age > 24 weeks

Preterm birth
The delivery of an infant before 37 completed weeks of gestation.
< 37 estimated gestational weeks (reported in electronic health records) at delivery

Very preterm birth
The delivery of an infant before 32 completed weeks of gestation.
< 32 estimated gestational weeks (reported in electronic health records) at delivery

Small for gestational age
A neonate whose birth weight lies below the 5th percentile for their gestational age.
Lower than the 5th percentile of the weight stratified for the gestational week at birth and sex based on a national birth cohort in England of singleton births (1998-2015)

Venous thrombotic events
The occurrence of thromboembolic events (formation of a blood clot in a vein) during pregnancy.
Secondary and primary care records (ICD-10 and SNOMED-CT, Read Codes -validated by a clinician) Table S1.Definition of adverse pregnancy outcomes.

Sensitivity analysis Rationale Association between COVID-19 diagnosis and adverse pregnancy outcomes
Women without recorded estimated gestational age at the delivery to examine the impact of the magnitude of the association of unrecorded estimated gestational age at delivery (n=200,020, n=199,145 in England and 875 in Wales) Women with estimated pregnancy start date < 11 th March 2020 to account for unobserved confounding with regards to awareness of the pandemic at conception Women with estimated pregnancy start date ≥ 11 th March 2020 to account for unobserved confounding with regards to awareness of the pandemic at conception Women who had received at least one dose of COVID-19 vaccine up to 31 st December 2021 to account for unobserved confounding with regards to vaccination status Women who had not received any doses of COVID-19 vaccine up to 31 st December 2021 to account for unobserved confounding with regards to vaccination status Women who received at least one dose of COVID-19 vaccine from 18 th June 2021 to 31 st December 2021 to account for unobserved confounding with regards to vaccination status Excluding outcomes that occurred starting from day 1 following the exposure to check whether the cases that occurred on the same day as the exposure have an impact on the effect estimates Non-hospitalised COVID-19 diagnosis as the exposure to check whether effect estimates were driven mainly by severe COVID-19 cases Hospitalised COVID-19 diagnosis as the exposure to estimate the effect estimates in severe COVID-19 cases Follow-up censored at the maximal outcome week for preterm and very preterm to account for bias related to including a COVID-19 diagnosis as an exposure when it occurs after the timeframe in which the outcome could potentially occur (e.g., a COVID-19 diagnosis at 38 weeks of gestation when examining preterm birth as the outcome of interest) Women with spontaneous preterm and very preterm to account for bias, we have excluded non-spontaneous preterm births, as induced labor may be requested due to other conditions that could be influenced by a COVID-19 infection (e.g.preeclampsia)

Association between COVID-19 vaccination and adverse pregnancy outcomes
Censoring at COVID-19 diagnosis to account for confounding with regards to COVID-19 diagnosis Women with estimated gestational age at the delivery to examine the impact of unrecorded estimated gestational age at delivery Women vaccinated after 18 th June 2021 to account for unobserved confounding with regards to comorbidities and risk of infection, since women vaccinated prior to that date belonged to special high vulnerability groups and/or were frontline healthcare workers      The RECORD statement -checklist of items, extended from the STROBE statement, that should be reported in observational studies using routinely collected health data.

Figure S1 .
Figure S1.Flow chart -derivation of the pregnancy cohort

Table S7 . Characteristic of the pregnancy cohort: dose 2 analysis (only England). a before pregnancy
If linkage between databases was conducted for the study, this should be clearly stated in the title or abstract.Authors should provide information on how to access any supplemental information such as the study protocol, raw data, or programming code.