Abstract
Background
SARS‐CoV‐2 infection during pregnancy has been linked to preterm birth, but this association is not well understood.
Objectives
To examine the association between SARS‐CoV‐2 infection and spontaneous and provider‐initiated preterm birth (PTB), and how timing of infection, and race/ethnicity as a marker of structural inequality, may modify this association.
Methods
We conducted a retrospective cohort study among pregnant people who delivered singleton, liveborn infants (22–44 weeks gestation) from 1 March 2020 to 31 March 2021 (n = 68,288). We used Cox proportional hazards models to compare the hazard of PTB between pregnant people with and without laboratory‐confirmed SARS‐CoV‐2 infection during pregnancy. We evaluated this association according to the trimester of infection, timing from infection to birth, and timing of PTB. We also examined the joint associations of SARS‐CoV‐2 infection and race/ethnicity with PTB using the relative excess risk due to interaction (RERI).
Results
Positive SARS‐CoV‐2 tests were identified for 2195 pregnant people (3.2%). The prevalence of PTB was 7.2% (3.8% spontaneous, 3.6% provider‐initiated). SARS‐CoV‐2 infection during pregnancy was associated with an increased risk of PTB overall (adjusted hazard ratio [HR] 1.53, 95% confidence interval [CI] 1.34, 1.74), and provider‐initiated PTB (HR 1.79, 95% CI 1.50, 2.12) but not spontaneous PTB (HR 1.09, 95% CI 0.89, 1.36). Second trimester infections were associated with an increased risk of provider‐initiated PTB, and third trimester infections were associated with an increased risk of both PTB subtypes. A joint inverse association between White non‐Hispanic race/ethnicity and SARS‐CoV‐2 infection and spontaneous PTB (HR 0.56, 95% CI 0.34, 0.94; RERI −0.6, 95% CI −1.0, −0.2) was also observed.
Conclusions
SARS‐CoV‐2 infections were primarily associated with an increased risk for provider‐initiated PTB in this study. These findings highlight the importance of promoting infection‐prevention strategies among pregnant people.
Keywords: infection, pregnancy, preterm birth, SARS‐CoV‐2
Synopsis.
Study question
What is the impact of SARS‐CoV‐2 infection during pregnancy on the risk of preterm birth?
What's already known
Some studies from mostly clinical settings have suggested that SARS‐CoV‐2 infection during pregnancy may increase the risk of preterm birth.
What this study adds
Our study demonstrates the robustness of associations between SARS‐CoV‐2 infection and spontaneous and provider‐initiated preterm birth after taking into account several possible sources of bias that have not previously been considered, and by illustrating how SARS‐CoV‐2 infection may interact jointly with race/ethnicity.
1. BACKGROUND
Novel coronavirus SARS‐CoV‐2 infection has posed a threat to human health since its emergence in 2019 and will likely do so until vaccines and therapeutics are equitably accessible and widely utilised. At the end of August 2022, close to 94 million cases of SARS‐CoV‐2 infection had been detected in the United States. 1 Like other newly emergent viruses such as H1N1 2 and Zika, 3 accumulating evidence suggests that infection during pregnancy may lead to adverse pregnancy consequences. 4
Multiple studies have described an association between SARS‐CoV‐2 infection during pregnancy and PTB (before 37 weeks of gestation). 4 , 5 , 6 , 7 , 8 , 9 , 10 , 11 , 12 , 13 , 14 Approximately 1 in 10 births in the United States occurred before 37 weeks of gestation in 2019, before the onset of the SARS‐CoV‐2 pandemic. 15 Further elucidation of the impact of SARS‐CoV‐2 infection on PTB is important as infants born preterm face increased risks of short‐ and long‐term neurodevelopmental, gastrointestinal, and respiratory complications. 16 Moreover, an association between SARS‐CoV‐2 infection during pregnancy and PTB has the potential to exacerbate existing racial and ethnic inequities given that pregnant people of racially and ethnically minoritised groups are disproportionately affected by both SARS‐CoV‐2 infection and PTB. 9
Although an association between SARS‐CoV‐2 infection during pregnancy and PTB has been observed with some consistency, the association between infection and type of PTB is more uncertain. Authors of the INTERCOVID study, which involved 2130 pregnant people in 18 countries, found an elevated risk for PTB among those diagnosed with COVID‐19 during pregnancy, but this association was only present among those who experienced a provider‐initiated PTB involving induction of labour or a caesarean section without labour. 4 Two large population‐based cohort studies, however, have reported that the SARS‐CoV‐2 infection is associated with an increased risk of both spontaneous and provider‐initiated preterm birth. 9 , 11
Disentangling the pathways through which SARS‐CoV‐2 infection may lead to premature onset of labour versus provider‐initiated preterm delivery is necessary to successfully prevent these outcomes among those infected. In addition, more research is needed regarding how timing of infection during pregnancy (i.e., trimester of infection), timing from infection to birth, and racial and ethnic inequities may modify this association.
2. METHODS
2.1. Cohort selection
We performed a retrospective cohort study of all singleton live births that occurred in Massachusetts (MA) between March 2020 and March 2021 that were issued a birth certificate by the MA Registry of Vital Records and Statistics and contained plausible gestational ages for live birth, which we defined as greater than 22 or less than 44 weeks of gestation. 17 We linked birth certificate data to SARS‐CoV‐2 infections confirmed by positive reverse transcription polymerase chain reaction test results (RT‐PCR) reported to the MA Bureau of Infectious Diseases and Laboratory Sciences in accordance with state policy.
2.2. Exposure
We considered pregnant persons to be exposed to SARS‐CoV‐2 infection if we identified a positive RT‐PCR test result during the first 36 weeks of gestation, since any infection occurring at 37 weeks or later would be outside the risk period for PTB. We considered pregnant persons without positive SARS‐CoV‐2 RT‐PCR test to be unexposed. We did not consider positive antigen test results as an inclusion criterion for several reasons: (1) in MA, only antigen tests performed in health facilities are reported to the state laboratory, and the majority of these are expected to have RT‐PCR confirmation; (2) at‐home antigen testing could not be captured, indicating incomplete ascertainment of antigen testing; and (3) antigen testing is less sensitive compared to RT‐PCR. 18 To evaluate the impact of infection timing, we further categorised infections as occurring during the first (1–13 weeks), second (14–26 weeks), or third (27–36 weeks) trimesters.
2.3. Outcomes
We defined PTBs as those occurring before 37 weeks of gestation based on the best clinical estimate of gestational age as provided on the birth certificate. To further classify PTBs into spontaneous and provider‐initiated, we applied the algorithm developed by Klebanoff and colleagues. 18 This algorithm uses delivery information from the birth certificate to determine whether PTB occurred spontaneously due to premature rupture of membranes or contractions, or by provider initiation through labour induction or pre‐labour caesarean. Validation data suggest that this algorithm correctly classifies 86% of PTBs. 19
2.4. Statistical analysis
We estimated unadjusted and adjusted hazard ratios (HR) and 95% confidence intervals (CI) for the association between SARS‐CoV‐2 infection during pregnancy and PTB in stratified Cox proportional hazards regression models 20 with gestational week as the time scale. This approach enabled us to model infection status as a time‐varying variable such that pregnant people would contribute unexposed person time until their positive test. It also allowed the baseline hazard of delivery to vary according to week of gestation.
Using available data from the birth certificate, we adjusted for: maternal age in 5‐year intervals, parity, pre‐pregnancy body mass index (BMI), nativity status, education, insurance type, marital status, Kotelchuck Index of prenatal care utilisation, 21 race/ethnicity, pre‐existing diabetes, gestational diabetes, and chronic hypertension, as potential confounders. These confounders were chosen a‐priori based on a directed acyclic graph (Figure S1). We note that race/ethnicity in this study is considered a marker for socially‐mediated exposures such as racism, discrimination, and access to resources, as well as occupational and environmental factors that have placed communities at disproportionate risk of SARS‐CoV‐2 infection, not as a biologic difference. 22 Collinearity of model covariates was assessed using variance inflation factors (VIF) from a series of linear regression models in which each covariate was modelled against the other covariates. We then compared VIF values to a published threshold (<10). 23
We repeated the above analysis for each type of PTB separately (spontaneous or provider‐initiated) by trimester of first positive SARS‐CoV‐2 test result (first, second, or third trimester), and by timing of PTB (<28 , 28–32 , or 33–36 weeks), and time from test to birth (0–3 , 4–30 , or 31 days or more). These cut‐offs were selected to differentiate testing that was likely performed during the delivery hospitalisation as part of universal screening for SARS‐CoV‐2 from earlier ad hoc testing. We also examined the possibility of an additive interaction between SARS‐CoV‐2 infection and race/ethnicity by calculating the relative excess risk due to interaction (RERI), 24 which indicates the amount by which a combination of exposures produces a greater or lesser than expected risk for an outcome than is observed when adding the individual exposures. A RERI greater than 0 denotes a synergistic interaction, whereas a RERI less than 0 denotes antagonistic interaction. Additionally, we repeated these analyses within strata of educational level to determine whether the associations are consistent across a maternal characteristic that proxies for advantage.
2.5. Missing data
The proportion of participants with missing covariate values ranged from <1% to 9%. To address possible bias from missing data, we used chained equation multiple imputations to generate 50 imputed datasets.
2.6. Sensitivity analyses
More infections may have been detected among PTBs than among term births because asymptomatic infections detected through universal SARS‐CoV‐2 screening at delivery were only included in this study if that delivery occurred before 37 weeks. We therefore conducted a probabilistic bias analysis 25 in which we assumed an exposure classification sensitivity ranging from 0.90 to 1.00 among the PTBs, consistent with published estimates for RT‐PCR testing, 26 and a lower sensitivity ranging from 0.60 to –0.80 among the term births to account for undetected infections. We assumed that differential detection of infections through universal screening would not impact specificity, which we set to 0.97–1.00 26 for both groups. We conducted 1000 simulations to calculate a bias‐adjusted risk ratio and 95% simulation interval. We also applied the lower sensitivity to both groups to explore the potential impact of non‐differential misclassification.
We further considered the possibility of selection bias if some participants who had experienced a SARS‐CoV‐2 infection during the study period delivered term births after the cut‐off date of 31 March 2021, whereas PTBs occurring before this date were captured. We addressed this in a sensitivity analysis limited to pregnancies with dates of conception between 1 March 2020 and 14 July 2020 (n = 17,018). These individuals could have had full‐term pregnancies during the study period.
Lastly, we examined the robustness of our results to the exclusion of antigen tests by randomly assigning 1% of participants to have an “antigen positive” test at a random week of gestation in each of the 50 separate datasets that were pooled for multiple imputation. Previously unexposed participants in this group with a simulated “antigen positive” test result prior to 37 weeks were recoded as exposed. We chose 1% based on the approximate prevalence of SARS‐CoV‐2 infection in our data at the time of delivery where universal screening was conducted, as it is our best estimate of the population prevalence of SARS‐CoV‐2 among pregnant people in MA during the study period.
2.7. Ethics approval
The Institutional Review Board at the Massachusetts Department of Public Health approved all study activities.
3. RESULTS
From 1 March 2020 to 31 March 2021, 69,960 births were recorded in MA. Of these, 536 birth records had a missing or implausible gestational age for live birth, and 1136 births were multiple gestations. The final analytic cohort included 68,288 births. Positive SARS‐CoV‐2 tests were identified for 2195 pregnant people in this cohort (3.2%). Compared to uninfected individuals, infected individuals were more frequently younger, multiparous, classified as overweight or obese, of Hispanic or non‐Hispanic Black race/ethnicity, or born outside the United States; and less frequently college educated, privately insured, or married (Table 1).
TABLE 1.
Baseline characteristics of people with singleton liveborn deliveries by SARS‐CoV‐2 infection status during pregnancy status, Massachusetts, 1 March 2020 to 31 March 2021 (n = 68,288)
| SARS‐CoV‐2 Infection during Pregnancy | Full cohort | |||||
|---|---|---|---|---|---|---|
| No | Yes | |||||
| N | % | N | % | N | % | |
| Maternal age (years) | ||||||
| <25 | 8086 | 12.2 | 395 | 17.9 | 8481 | 12.4 |
| 25–29 | 14,218 | 21.5 | 599 | 27.1 | 14,817 | 21.7 |
| 30–34 | 25,336 | 38.3 | 726 | 32.8 | 26,062 | 38.2 |
| 35–39 | 15,158 | 22.9 | 387 | 17.5 | 15,545 | 22.8 |
| 40+ | 3278 | 5.0 | 105 | 4.8 | 3383 | 5.0 |
| Parity (births) | ||||||
| 0 | 29,287 | 44.3 | 787 | 357 | 30.074 | 44.0 |
| 1 | 22,260 | 33.7 | 656 | 29.7 | 22,196 | 33.6 |
| 2 | 8783 | 13.3 | 353 | 16.0 | 9136 | 13.4 |
| 3+ | 4457 | 6.9 | 203 | 9.2 | 4660 | 6.8 |
| Pre‐pregnancy BMI (kg/m2) | ||||||
| <18.5 | 2096 | 3.2 | 44 | 2.0 | 2140 | 3.1 |
| 18.5 to <25 | 30,946 | 46.8 | 754 | 34.1 | 31,700 | 46.4 |
| 25 to <30 | 17,510 | 26.5 | 697 | 31.5 | 18,207 | 26.7 |
| 30+ | 14,495 | 21. | 654 | 29.6 | 15,149 | 22.2 |
| Race/ethnicity | ||||||
| American Indian or other non‐Hispanic persons | 647 | 1.0 | 22 | 1.0 | 669 | 1.0 |
| Asian American/Pacific Islander, non‐Hispanic persons | 6075 | 9.5 | 126 | 5.7 | 6201 | 9.1 |
| Black, non‐Hispanic persons | 6606 | 10.3 | 340 | 15.4 | 6946 | 10.2 |
| Hispanic persons | 13,321 | 20.8 | 939 | 42.5 | 14,260 | 20.2 |
| White, non‐Hispanic persons | 37,126 | 58.3 | 705 | 31.9 | 37,831 | 55.4 |
| Nativity status | ||||||
| U.S. born | 44,259 | 67.0 | 1043 | 47.2 | 45,302 | 66.3 |
| Foreign born | 21,543 | 32.5 | 1162 | 52.5 | 22,705 | 33.3 |
| Education | ||||||
| Less than high school | 4935 | 7.5 | 345 | 15.6 | 5280 | 7.7 |
| High school | 10,817 | 16.4 | 495 | 22.4 | 11,312 | 16.6 |
| Some college | 10,890 | 16.5 | 502 | 22.7 | 11,392 | 16.7 |
| Associate's degree | 4142 | 6.3 | 155 | 7.0 | 4297 | 6.3 |
| Bachelor's degree or higher | 33,851 | 51.23 | 627 | 28.4 | 34,478 | 50.5 |
| Unknown | 1441 | 2.2 | 88 | 4.0 | 1529 | 2.2 |
| Insurance | ||||||
| Private insurance | 39,743 | 60.2 | 875 | 39.6 | 40,618 | 59.5 |
| Medicaid | 24,244 | 36.7 | 1292 | 58.4 | 25,536 | 37.4 |
| None/self‐pay/other/unknown | 2088 | 3.2. | 45 | 2.0 | 2133 | 3.1 |
| Kotelchuck index a | ||||||
| Inadequate | 4357 | 6.6 | 178 | 8.1 | 4535 | 6.6 |
| Intermediate | 28,468 | 43.1 | 931 | 42.1 | 29,399 | 43.1 |
| Adequate | 30,914 | 46.8 | 942 | 42.6 | 31,856 | 46.7 |
| Adequate Plus | 1155 | 1.8 | 43 | 1.9 | 1198 | 1.8 |
| Unknown | 1182 | 1.8 | 118 | 5.3 | 1300 | 1.9 |
| Married (yes/no) | ||||||
| No | 22.039 | 33.4 | 1014 | 45.8 | 23,053 | 33.8 |
| Yes | 44,037 | 66.7 | 1198 | 5.2 | 45,235 | 66.2 |
| Chronic hypertension | ||||||
| No | 64,537 | 97.7 | 2.155 | 97.4 | 66,692 | 97.7 |
| Yes | 1524 | 2.3 | 57 | 2.6 | 1581 | 2.3 |
| Pre‐existing diabetes | ||||||
| No | 65,434 | 99.0 | 2176 | 98.4 | 67,610 | 99.0 |
| Yes | 627 | 1.0 | 36 | 1.6 | 663 | 1.0 |
| Preterm birth | ||||||
| No | 61,421 | 93.0 | 1958 | 88.5.5 | 63,379 | 92.8 |
| Yes | 4655 | 7.0 | 254 | 11.5 | 4909 | 7.2 |
Abbreviation: BMI, body mass index.
See Kotelchuck [21].
The overall prevalence of PTB in the cohort was 7.2% (n = 4909/68,288). Spontaneous PTBs accounted for 51% (n = 2513/4909) of PTBs, and 3.8% (2513/65,892) of total births (excluding the unknown and provider‐initiated PTB subtypes), while provider‐initiated PTBs accounted for 49% (n = 2372/4909) of PTBs and 3.6% (n = 2372/65,751) of total births (excluding the unknown and spontaneous PTB subtypes). Twenty‐four PTBs (0.4%) could not be classified according to the algorithm and were classified as unknown. The prevalence of PTB was higher among infected individuals compared to those uninfected (11.5% vs. 7.0%) (Table 1).
Kaplan–Meier plots showing the cumulative incidence of PTB overall and each subtype by SARS‐CoV‐2 infection status are shown in Figure S2. These plots show an earlier time to birth overall and to provider‐initiated birth for those with SARS‐CoV‐2 infection, but an earlier time to spontaneous birth among those uninfected. Adjustment for maternal characteristics attenuated the HR (HR 1.53, 95% CI 1.34, 1.74) (Table 2). The adjusted HR for spontaneous PTB was lower than the overall HR (HR 1.09, 95% CI 0.89, 1.36), while the adjusted HR for provider‐initiated PTB was higher (HR 1.79, 95% CI 1.50, 2.12). VIF values did not suggest strong collinearity between covariates (Table S1).
TABLE 2.
Unadjusted and adjusted a associations between SARS‐CoV‐2 infection during pregnancy and preterm birth, by type, among singleton live birth deliveries — Massachusetts, 1 March 2020 to 31 March 2021
| Gestational age | n (%) | Unadjusted HR (95% CI) | Adjusted HR a (95% CI) |
|---|---|---|---|
| Term births (37–44 weeks) | 63,379 (92.8) | ||
| Preterm births (22–36 weeks) | 4909 (7.2) | 1.80 (1.59, 2.05) | 1.53 (1.34, 1.74) |
| Unknown b | 24 (0) | – | – |
| Spontaneous | 2513 (3.8) c | 1.27 (1.04, 1.57) | 1.09 (0.89, 1.36) |
| Provider‐initiated | 2372 (3.6) d | 2.13 (1.80, 2.52) | 1.79 (1.50, 2.12) |
Abbreviation: HR, hazard ratio.
Adjusted for age category, parity, pre‐pregnancy body mass index (BMI) category, nativity status, education, insurance type, marital status, Kotelchuck index, race/ethnicity, pre‐existing diabetes, and chronic hypertension.
24 preterm births had insufficient information for classification into a subtype. These cases have been included in the numerator and denominator for all preterm births but not for either subtype.
Calculated as 2513/2513 + 63,379.
Calculated as 2372/2372 + 63,379.
Among pregnant people with second trimester infections, an association was only observed for provider‐initiated PTB (HR 2.15, 95% CI 1.63, 2.83) (Table 3). Among those with third trimester infections, elevated hazards of both spontaneous (HR 1.63, 95% CI 1.26, 2.10) and provider‐initiated (HR 2.14, 95% CI 1.71, 2.69) PTB were observed, though the magnitude of the association was greater for provider‐initiated PTBs. Stronger associations were observed for PTBs that occurred before 28 weeks and between 28 and 32 weeks among provider‐initiated PTBs. While substantially elevated hazard ratios for both types of PTB were observed for infections that were detected within 3 days of delivery and between 4 and 30 days of delivery, those detected more than 30 days before delivery were associated with a reduced hazard of spontaneous PTB and were not associated with provider‐initiated PTB. SARS‐Cov2‐infection was not associated with spontaneous PTB among pregnant people with more than a high school education (Table S2).
TABLE 3.
Adjusted a associations between SARS‐CoV‐2 infection during pregnancy and preterm birth, by trimester of infection, time between infection and delivery, and timing of preterm birth among singleton live birth deliveries — Massachusetts, 1 March 2020 to 31 March 2021
| All preterm birth | Spontaneous preterm birth | Provider‐initiated preterm birth | |
|---|---|---|---|
| aHR (95% CI) | aHR (95% CI) | aHR (95% CI) | |
| Trimester of infection | |||
| First (1–13 weeks) | 0.99 (0.66, 1.48) | 0.83(0.45, 1.55) | 1.15 (0.68, 1.94) |
| Second (14–26 weeks) | 1.48 (1.17, 1.86) | 0.82 (0.53, 1.26) | 2.15 (1.63, 2.83) |
| Third (≥27 weeks) | 1.86 (1.57, 2.21) | 1.63 (1.26, 2.10) | 2.14 (1.71, 2.69) |
| Time between infection and delivery, days | |||
| 0–3 | 14.30 (10.99, 18.60) | 13.71 (9.38, 20.05) | 14.09 (9.63, 20.62) |
| 4–30 | 4.29 (3.30, 5.56) | 3.31 (2.18, 5.06) | 5.19 (3.71, 7.26) |
| ≥30 | 0.86 (0.73, 1.03) | 0.54 (0.40, 0.73) | 1.04 (0.83, 1.31) |
| Timing of preterm birth, weeks | |||
| <28 | 2.19 (1.33, 3.63) | 1.31 (0.57, 3.05) | 2.50 (1.22, 5.14) |
| 28–32 | 1.87 (1.17, 2.97) | 0.62 (0.20, 1.94) | 2.63 (1.52, 4.57) |
| 33–36 | 1.48 (1.29, 1.71) | 1.09 (0.87, 1.37) | 1.73 (1.43, 2.10) |
Abbreviation: HR, hazard ratio.
Adjusted for age category, parity, pre‐pregnancy body mass index (BMI) category, nativity status, education, insurance type, marital status, race/ethnicity, Kotelchuck index, pre‐existing diabetes, and chronic hypertension.
Compared to pregnant people who were not of White, non‐Hispanic race/ethnicity and were uninfected during pregnancy, those of White, non‐Hispanic race/ethnicity who experienced SARS CoV‐2 infections during pregnancy had a reduced risk of spontaneous PTB (HR 0.56, 95% CI 0.34, 0.94) (Table 4). The RERI of −0.6 (95% CI −1.0, −0.2) was indicative of an antagonistic interaction.
TABLE 4.
Adjusted a joint associations for SARS‐CoV‐2 infection during pregnancy and race/ethnicity and preterm birth among singleton live birth deliveries — Massachusetts, March 2020 to March 2021
| Race/ethnicity and SARS‐COV‐2 infection status | Spontaneous preterm birth | Provider‐initiated preterm birth | ||
|---|---|---|---|---|
| aHR (95% CI) | RERI | aHR (95% CI) | RERI | |
| American Indian or other non‐Hispanic persons | ||||
| No, SARS‐CoV‐2 uninfected | 1.00 (Reference) | 0.1 (−2.0, 2.0) | 1.00 (Reference) | 1.0 (−2.5, 4.5) |
| No, SARS‐CoV‐2 infected | 1.11 (0.90, 1.37) | 1.79 (1.51, 2.13) | ||
| Yes, SARS‐CoV‐2 uninfected | 0.97 (0.64, 1.46) | 0.80 (0.49, 1.28) | ||
| Yes, SARS‐CoV‐2 infected | 1.14 (0.16, 8.14) | 2.67 (0.67, 10.57) | ||
| Asian American/Pacific Islander persons | ||||
| No, SARS‐CoV‐2 uninfected | 1.00 (Reference) | 0.8 (−0.4, 2.0) | 1.00 (Reference) | 0.3 (−1.1, 1.7) |
| No, SARS‐CoV‐2 infected | 1.06 (0.85, 1.33) | 1.79 (1.50, 2.14) | ||
| Yes, SARS‐CoV‐2 uninfected | 1.10 (0.94, 1.28) | 0.99 (0.84, 1.14) | ||
| Yes, SARS‐CoV‐2 infected | 2.00 (1.04, 3.85) | 2.06 (1.04, 4.08) | ||
| Black, non‐Hispanic persons | ||||
| No, SARS‐CoV‐2 uninfected | 1.00 (Reference) | 0.3 (−0.4, 1.0) | 1.00 (Reference) | 0.2 (−0.7, 1.1) |
| No, SARS‐CoV‐2 infected | 1.06 (0.84, 1.34) | 1.80 (1.49, 2.18) | ||
| Yes, SARS‐CoV‐2 uninfected | 1.02 (0.89, 1.16) | 1.22 (1.08, 1.39) | ||
| Yes, SARS‐CoV‐2 infected | 1.38 (0.86, 2.19) | 2.23 (1.52, 3.26) | ||
| Hispanic persons | ||||
| No, SARS‐CoV‐2 uninfected | 1.00 (Reference) | 0.1 (−0.4, 0.6) | 1.00 (Reference) | 0.3 (−0.4, 1.0) |
| No, SARS‐CoV‐2 infected | 1.04 (0.77, 1.40) | 1.77 (1.40, 2.23) | ||
| Yes, SARS‐CoV‐2 uninfected | 1.14 (1.02, 1.27) | 1.07 (0.95, 1.20) | ||
| Yes, SARS‐CoV‐2 infected | 1.31 (0.97, 1.77) | 1.95 (0.51, 2.54) | ||
| White, non‐Hispanic persons | ||||
| No, SARS‐CoV‐2 uninfected | 1.00 (Reference) | ‐0.6 (−1.0, −0.2) | 1.00 (Reference) | −0.4 (−1.0, 0.1) |
| No, SARS‐CoV‐2 infected | 1.29 (1.02, 1.63) | 1.92 (1.56, 2.35) | ||
| Yes, SARS‐CoV‐2 uninfected | 0.87 (0.78, 0.96) | 0.86 (0.77, 0.96) | ||
| Yes, SARS‐CoV‐2 infected | 0.56 (0.34, 0.94) | 1.32 (0.95, 1.84) | ||
Abbreviations: HR, hazard ratio; RERI, relative excess risk due to interaction.
Adjusted for age category, parity, pre‐pregnancy body mass index (BMI) category, nativity status, education, insurance type, marital status, Kotelchuck index, race/ethnicity, pre‐existing diabetes, and chronic hypertension.
3.1. Sensitivity analyses
In a probabilistic bias analysis to assess the possibility of exposure misclassification due to undetected SARS‐CoV‐2 infections among those without a positive SARS‐CoV‐2 test on record, the corrected risk ratio for the association between SARS‐CoV‐2 infection and PTB was 1.58 (95% simulation interval 1.16, 2.98) assuming differential misclassification. Assuming non‐differential misclassification, the corrected risk ratio for this association was 2.22 (95% simulation interval 1.70, 2.97).
When the cohort was restricted to pregnancies with dates of conception between 1 March 2020 and 14 July 2020 who were likely to have had a birth outcome by the cut‐off date for our cohort formation (n = 17,018), elevated associations were observed for PTB overall and provider‐initiated PTB (Table 5), but these associations were attenuated compared to those from the primary analysis. SARS‐CoV‐2 infection was not associated with spontaneous PTB in this group, but the association between SARS‐CoV‐2 infection and provider‐initiated PTB remained apparent (Table 5). Simulating “antigen‐positive” test results among 1% of the cohort indicated a small amount of bias in our uncorrected results, which was away from the null (Table S3).
TABLE 5.
Adjusted a associations between SARS‐CoV‐2 infection during pregnancy and preterm birth, by trimester of infection, among singleton live birth deliveries with dates of conception between 1 March 2020 and 14 July 2020 (n = 17,018)
| All infections | First trimester (1–13 weeks) | Second trimester (14–26 weeks) | Third trimester (27–36 weeks) | |
|---|---|---|---|---|
| aHR (95% CI) | aHR (95% CI) | aHR (95% CI) | aHR (95% CI) | |
| Preterm type | ||||
| All | 1.30 (1.07, 1.59) | 0.88 (0.49, 1.60) | 1.62 (1.14, 2.30) | 1.38 (1.06, 1.79) |
| Spontaneous | 0.96 (0.69, 1.34) | 0.68 (0.26, 1.83) | 1.25 (0.70, 2.21) | 1.10 (0.73, 1.68) |
| Provider‐initiated | 1.34 (1.15, 1.94) | 1.07 (0.50, 2.25) | 1.99 (1.29, 3.09) | 1.66 (1.19, 2.32) |
Abbreviation: HR, hazard ratio.
Adjusted for age category, parity, pre‐pregnancy body mass index (BMI) category, nativity status, education, insurance type, marital status, Kotelchuck index, race/ethnicity, pre‐existing diabetes, and chronic hypertension.
3.2. Comment
3.2.1. Principal findings
In this large, population‐based study, we found that SARS‐CoV‐2 infection during pregnancy was associated with both spontaneous and provider‐initiated PTB, but particularly provider‐initiated PTB. We also observed that the timing of SARS‐CoV‐2 infection may influence the risk of PTB, and infection was most strongly associated with early PTBs. Our results also suggest that the combination of SARS‐CoV‐2 infection and White, non‐Hispanic race/ethnicity was associated with a larger‐than‐expected reduction in the risk of spontaneous PTB.
3.2.2. Strengths of the study
Strengths of this study include its large sample size, population‐based design, use of laboratory‐confirmed SARS‐CoV‐2 test results, and rigorous analytic approach, including bias analyses for exposure misclassification and selection bias.
3.2.3. Limitations of the data
Some limitations should be noted. First, RT‐PCR‐based test positivity rates underestimate the population prevalence of SARS‐CoV‐2 infection. 27 , 28 , 29 Testing was widely available in MA throughout most of the study period, though social vulnerability was associated with testing gaps. 30 At‐home testing was limited during this time. Exposure misclassification may therefore have occurred among those who had asymptomatic or otherwise undetected infections. Although most MA hospitals implemented universal screening of pregnant people at delivery by May 2020, 31 our results may consequently be subject to differential misclassification because asymptomatic infections detected through universal SARS‐CoV‐2 screening at delivery were included only for pregnant people who delivered preterm. Infections detected at delivery among term pregnant people were excluded because those individuals were no longer at risk for the outcome. Our probabilistic bias analysis quantifying the impact of possible differential exposure misclassification found that the results did not appreciably change. Given the substantially elevated associations between SARS‐CoV‐2 infections detected around the time of delivery and PTB, however, we cannot rule out incidental detection during hospitalisations for PTB.
We may also have introduced some exposure misclassification by not including positive antigen test results, which are included in the Council of State and Territorial Epidemiologists definitions as “probable” COVID‐19 cases, 32 though some of these cases may have become confirmed cases if individuals subsequently underwent RT‐PCR testing. This misclassification likely would have led us to slightly overestimate the hazards of PTB as shown in the results of our sensitivity analysis.
In addition, some selection bias may have occurred given that pregnant people who experienced full‐term deliveries after the participation cut‐off date of 31 March 2021 were excluded from the study, whereas those with similar conception dates who experienced preterm deliveries were included. When we restricted the analysis to pregnant people who had the opportunity to have a full‐term birth during the study period, we observed an attenuation of the association between third trimester SARS‐CoV‐2 infection and spontaneous PTB. While the sample size reduction may have limited the precision of the estimate, this sensitivity analysis suggests some caution may be warranted in interpreting the association between third trimester SARS‐CoV‐2 infection and spontaneous PTB. Other limitations include potential confounder misclassification and residual confounding due to inaccuracies in the birth certificate data. Lastly, we lacked medical record data necessary for examining infection severity.
3.2.4. Interpretation
Our results are consistent with several other studies associating increased risk of PTB with SARS‐CoV‐2 infection. 4 , 5 , 6 , 7 , 8 , 9 More disagreement exists, however, regarding the association between SARS‐CoV‐2 infection and PTB type. The INTERCOVID study found no association between COVID‐19 diagnosis and spontaneous PTB, but an elevated association with provider‐initiated PTB similar in magnitude to that observed in our study (RR 1.97, 95% CI 1.56, 2.51). 4 While we similarly found stronger evidence for an association between SARS‐CoV‐2 and provider‐initiated PTB, the existence of an association between SARS‐CoV‐2 and spontaneous PTB is supported by findings from two large population‐based cohort studies, one which used data from California birth certificates and the other of which used national insurance claims data. These studies showed that prenatal SARS CoV‐2 infection was associated with 1.5 (95% CI 1.2, 1.7) and 1.79 (95% CI 1.37, 2.34) times the risk of spontaneous PTB, respectively.
Different mechanisms likely underlie the association between SARS‐CoV‐2 infection and type of PTB. The established role of inflammation in preterm labour 33 lends biologic plausibility to our observation that SARS‐CoV‐2 infection was associated with an increased risk of spontaneous PTB in the third trimester. An estimated 20%–40% of preterm deliveries occur due to maternal infection. 34 Authors of a chart review study recently reported an increased prevalence of intrauterine infection and inflammation among pregnant people with SARS‐CoV‐2 infections at delivery compared to those uninfected (8.8% vs. 1.4%), though they did not observe an increased risk for PTB associated with SARS‐CoV‐2 infection. 35 Alternatively, the lack of an association among those with more than a high school education suggests that residual confounding by structural inequities may partly explain the association between SARS‐CoV‐2 infection and spontaneous preterm birth. Further research is needed to elucidate potential pathways through which SARS‐CoV‐2 may trigger spontaneous preterm labour.
Mechanisms through which SARS‐CoV‐2 infection may lead to provider‐initiated PTB are similarly unclear, but biologically plausible. First, preterm delivery may have been initiated as a direct result of worsening maternal illness. Secondly, evidence suggests that PTB overall and provider‐initiated PTB declined during the early COVID‐19 pandemic. 36 This decline has been attributed to obstetric service disruptions. 37 It is possible that obstetric interventions were preferentially allocated to patients with acute SARS‐CoV‐2 during this time. Thirdly, SARS‐CoV‐2 infection may increase the risk of pregnancy complications that are themselves indications for preterm delivery, such as preeclampsia. 38 , 39 Our observation that the increased risk for provider‐initiated PTB began in the second trimester aligns with the hypothesis that pregnancy complications may mediate the association between SARS‐CoV‐2 infection and this PTB subtype. Nevertheless, infections detected 30 days or more before delivery were not associated with provider‐initiated PTB which suggests that the biologic underpinnings of an association between SARS‐CoV‐2 and provider‐initiated PTB operate during a relatively short timeframe.
Previous studies of the impact of time between infection detection and delivery on the association between SARS‐CoV‐2 and PTB are lacking. As noted above, we observed considerably high hazard ratios for both preterm subtypes when restricting to infections detected within 30 days of delivery that were not seen for older infections. It is possible that those with full‐term gestations had more opportunity to have had an infection that occurred more than 1 month before their delivery. This positive correlation between time since infection and gestational duration may have resulted in some downward bias.
In a population‐based study from California, Karasek et al. 9 noted the absence of any interactions between SARS‐CoV‐2 infection and race/ethnicity, though they did observe disproportionate burdens of both infection and of PTB among Black, Indigenous, and other People of Colour. We examined potential interaction on the additive scale through the calculation of the RERI. Although, in the past, the RERI has been interpreted to signify “biologic interaction,” 40 an interaction between SARS‐CoV‐2 infection and race/ethnicity might be related to systemic racial/ethnic inequities. Here, we observed that White, non‐Hispanic individuals were not only both less likely to experience SARS‐CoV‐2 infection and PTB compared to individuals of other race/ethnicity groups, but those who became infected with SARS‐CoV‐2 had an even greater reduction in the risk of spontaneous PTB than would be expected from the combination of these exposures. We interpret this finding to suggest that factors related to structural racism may underlie the association between SARS‐CoV‐2 infection and spontaneous PTB, since the systemic advantage of “whiteness” seems to confer some protection from it. Reasons for this protection are not clear but could include greater structural access to and utilisation of healthcare. Future research that addresses such systemic barriers to care access and utilisation should be prioritised.
4. CONCLUSION
In conclusion, our results support an association between SARS‐CoV‐2 infection and both spontaneous and provider‐initiated PTB for third trimester infections. The association was stronger, however, for provider‐initiated PTB. Our findings add to the existing literature by demonstrating robustness of these to several possible sources of bias that have not previously been considered and by illustrating how SARS‐CoV‐2 infection may interact jointly with race/ethnicity. They also underscore the importance of evaluating the impacts of asymptomatic versus symptomatic infection in PTB as well as the impacts of different variants of SARS‐Cov2‐on PTB in future prospective cohorts. Lastly, they highlight the importance of promoting infection‐prevention strategies among pregnant people, and particularly for pregnant persons of racially and ethnically minoritised groups.
AUTHOR CONTRIBUTIONS
Drs Darling and Yazdy had full access to the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Concept and design: Darling, Manning, Nestoridi, Shephard, Yazdy. Acquisition, analysis, or interpretation of data: Darling, Manning, Nestoridi, Shephard, Yazdy. Statistical analysis: Darling. Drafting of the manuscript: Darling. Critical revision of the manuscript for important intellectual content: Darling, Manning, Nestoridi, Shephard, Yazdy.
FUNDING INFORMATION
This work was supported in part by the Centers for Disease Control and Prevention cooperative agreement “Building and Enhancing Epidemiology, Laboratory and Health Information Systems Capacity in Massachusetts” grant (NU50CK000518), and by an appointment to the Applied Epidemiology Fellowship Program administered by the Council of State and Territorial Epidemiologists (CSTE) and funded by the Centers for Disease Control and Prevention (CDC) Cooperative Agreement Number 1NU38OT000297–03‐00.
CONFLICT OF INTEREST
None.
ROLE OF THE FUNDER/SPONSOR
The funding source had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Supporting information
Appendix S1
ACKNOWLEDGEMENT
The authors wish to acknowledge the expertise contributed by Dominique Heinke, PhD in conducting a full review of the code used for statistical analysis.
Darling AM, Shephard H, Nestoridi E, Manning SE, Yazdy MM. SARS‐CoV‐2 infection during pregnancy and preterm birth in Massachusetts from March 2020 through March 2021. Paediatr Perinat Epidemiol. 2022;00:1‐11. doi: 10.1111/ppe.12944
The findings and conclusions in this presentation are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention/Agency for Toxic Substances and Disease Registry.
DATA AVAILABILITY STATEMENT
Research data are not available.
REFERENCES
- 1. Center for Disease Control and Prevention . CDC Data Tracker. Accessed December 31, 2021. https://covid.cdc.gov/covid‐data‐tracker/#datatracker‐home
- 2. Siston AM, Rasmussen SA, Honein MA, et al. Pandemic 2009 influenza a (H1N1) virus illness among pregnant women in the United States. Jama. 2010;303(15):1517‐1525. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Moore CA, Staples JE, Dobyns WB, et al. Characterizing the pattern of anomalies in congenital zika syndrome for pediatric providers. JAMA Pediatr. 2017;171(3):288‐295. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Villar J, Ariff S, Gunier RB, et al. Maternal and neonatal morbidity and mortality among pregnant women with and without COVID‐19 infection: the INTERCOVID multinational cohort study. JAMA Pediatr. 2021;175:817‐826. doi: 10.1001/jamapediatrics.2021.1050 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Woodworth KR, Olsen EO, Neelam V, et al. Birth and infant outcomes following laboratory‐confirmed SARS‐CoV‐2 infection in pregnancy—SET‐NET, 16 jurisdictions, march 29–October 14, 2020. Morb Mortal Wkly Rep. 2020;69(44):1635‐1640. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Allotey J, Stallings E, Bonet M, et al. Pregnancy: living systematic review and meta‐analysis. BMJ. 2019;2020:370. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Gurol‐Urganci I, Jardine JE, Carroll F, et al. Maternal and perinatal outcomes of pregnant women with SARS‐CoV‐2 infection at the time of birth in England: national cohort study. Am J Obstet Gynecol. 2021;225(5):522.e1‐522.e11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Crovetto F, Crispi F, Llurba E, et al. Impact of severe acute respiratory syndrome coronavirus 2 infection on pregnancy outcomes: a population‐based study. Clin Infect Dis. 2021;73(10):1768‐1775. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Karasek D, Baer RJ, McLemore MR, et al. The association of COVID‐19 infection in pregnancy with preterm birth: a retrospective cohort study in California. Lancet reg Health Am. 2021;2:100027. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Blitz MJ, Gerber RP, Gulersen M, et al. Preterm birth among women with and without severe acute respiratory syndrome coronavirus 2 infection. Acta Obstet Gynecol Scand. 2021;100(12):2253‐2259. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Regan AK, Arah OA, Fell DB, Sullivan SG. SARS‐CoV‐2 infection during pregnancy and associated perinatal health outcomes: a national US cohort study. J Infect Dis. 2022;225(5):759‐767. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. McClymont E, Albert AY, Alton GD, et al. Association of SARS‐CoV‐2 infection during pregnancy with maternal and perinatal outcomes. Jama. 2022;327(20):1983‐1991. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Ferrara A, Hedderson MM, Zhu Y, et al. Perinatal complications in individuals in California with or without SARS‐CoV‐2 infection during pregnancy. JAMA Intern Med. 2022;182(5):503‐512. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Sentilhes L, De Marcillac F, Jouffrieau C, et al. Coronavirus disease 2019 in pregnancy was associated with maternal morbidity and preterm birth. Am J Obstet Gynecol. 2020;223(6):914.e1‐914.e15. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Martin J, Hamilton B, Osterman M. Births in the United States, 2019. US Department of Health and Human Services, CDC, National Center for Vital Statistics; 2020. Accessed December 1, 2021. https://www.cdc.gov/nchs/data/databriefs/db387‐H.pdf [Google Scholar]
- 16. Saigal S, Doyle LW. An overview of mortality and sequelae of preterm birth from infancy to adulthood. Lancet. 2008;371(9608):261‐269. [DOI] [PubMed] [Google Scholar]
- 17. Talge NM, Mudd LM, Sikorskii A, Basso O. United States birth weight reference corrected for implausible gestational age estimates. Pediatrics. 2014;133(5):844‐853. [DOI] [PubMed] [Google Scholar]
- 18. Matsuda EM, de Campos IB, de Oliveira IP, Colpas DR, Carmo AM, Brigido LF. Field evaluation of COVID‐19 antigen tests versus RNA based detection: potential lower sensitivity compensated by immediate results, technical simplicity, and low cost. J Med Virol. 2021;93(7):4405‐4410. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Klebanoff MA, Yossef‐Salameh L, Latimer C, et al. Development and validation of an algorithm to determine spontaneous versus provider‐initiated preterm birth in US vital records. Paediatr Perinat Epidemiol. 2016;30(2):134‐140. [DOI] [PubMed] [Google Scholar]
- 20. Kleinbaum DG. The stratified Cox procedure. Survival Analysis. Springer; 1996:171‐210. [Google Scholar]
- 21. Kotelchuck M. An evaluation of the Kessner adequacy of prenatal care index and a proposed adequacy of prenatal care utilization index. Am J Public Health. 1994;84(9):1414‐1420. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Adkins‐Jackson PB, Chantarat T, Bailey ZD, Ponce NA. Measuring structural racism: A guide for epidemiologists and other health researchers. Am J Epidemiol. 2022;191(4):539‐547. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Dormann CF, Elith J, Bacher S, et al. Collinearity: a review of methods to deal with it and a simulation study evaluating their performance. Ecography. 2013;36(1):27‐46. [Google Scholar]
- 24. Rothman KJ. Interactions between causes. Mod Epidemiol. 1986;3:11‐326. [Google Scholar]
- 25. Lash TL, Fox MP, Fink AK, eds. Applying Quantitative Bias Analysis to Epidemiological Data. Springer; 2009. [Google Scholar]
- 26. Tsang NN, So HC, Ng KY, Cowling BJ, Leung GM, Ip DK. Diagnostic performance of different sampling approaches for SARS‐CoV‐2 RT‐PCR testing: a systematic review and meta‐analysis. Lancet Infect Dis. 2021;21(9):1233‐1245. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Bruckner TA, Parker DM, Bartell SM, et al. Estimated seroprevalence of SARS‐CoV‐2 antibodies among adults in Orange County, California. Sci Rep. 2021;11(1):1‐9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Lamba K, Bradley H, Shioda K, et al. SARS‐CoV‐2 cumulative incidence and period seroprevalence: results from a statewide population‐based serosurvey in California. Open Forum Infect Dis. 2021;8(8):ofab379. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Giuliano AR, Pilon‐Thomas S, Schell MJ, et al. SARS‐CoV‐2 period seroprevalence and related factors, Hillsborough County, Florida, USA, October 2020–march 2021. Emerg Infect Dis. 2022;28(3):556‐563. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Dryden‐Peterson S, Velásquez GE, Stopka TJ, Davey S, Lockman S, Ojikutu BO. Disparities in SARS‐CoV‐2 testing in Massachusetts during the COVID‐19 pandemic. JAMA Netw Open. 2021;4(2):e2037067. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Angelidou A, Sullivan K, Melvin PR, et al. Association of maternal perinatal SARS‐CoV‐2 infection with neonatal outcomes during the COVID‐19 pandemic in Massachusetts. JAMA Netw Open. 2021;4(4):e217523. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Council of State and Territorial Epidemiologists . Standardized surveillance case definition and notification for 2019 novel coronavirus disease (COVID‐19). Accessed December 1, 2021. https://cdn.ymaws.com/www.cste.org/resource/resmgr/2020ps/Interim‐20‐ID‐01_COVID‐19.pdf
- 33. Romero R, Espinoza J, Gonçalves LF, Kusanovic JP, Friel L, Hassan S. The role of inflammation and infection in preterm birth. Semin Reprod Med. 2007;25(1):21‐39. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Gibbs RS, Romero R, Hillier SL, Eschenbach DA, Sweet RL. A review of premature birth and subclinical infection. Am J Obstet Gynecol. 1992;166(5):1515‐1528. [DOI] [PubMed] [Google Scholar]
- 35. Nguyen P, El‐Kady D, Rosner JY, Rahimi S, Dinglas C. COVID‐19 and pregnancy outcomes: an increased risk of intrauterine inflammation/infection. Am J Perinatol. 2021;38(S2):A019. [Google Scholar]
- 36. Margerison C, Bruckner TA, MacCallum‐Bridges C, Catalano R, Casey JA, Gemmill A. Exposure to the early COVID‐19 pandemic and early, moderate and overall preterm births in the United States: a conception cohort approach. Paediatr Perinat Epidemiol. 2022. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Lisonkova S, Joseph KS. Why did preterm birth rates fall during the COVID‐19 pandemic? Paediatr Perinat Epidemiol. 2022. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Papageorghiou AT, Deruelle P, Gunier RB, et al. Preeclampsia and COVID‐19: results from the INTERCOVID prospective longitudinal study. Am J Obstet Gynecol. 2021;225(3):289.e1‐289.e17. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Conde‐Agudelo A, Romero R. SARS‐COV‐2 infection during pregnancy and risk of preeclampsia: a systematic review and meta‐analysis. Am J Obstet Gynecol. 2022;226(1):68‐89.e3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Andersson T, Alfredsson L, Källberg H, Zdravkovic S, Ahlbom A. Calculating measures of biological interaction. Eur J Epidemiol. 2005;20(7):575‐579. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Appendix S1
Data Availability Statement
Research data are not available.