Long COVID before or during pregnancy was not linked to preterm birth in a US cohort

Whether Long COVID complicates a subsequent pregnancy is a question a lot of people have been asking without much to go on. An analysis published in Obstetrics & Gynecology on August 27 supplies the first substantial answer, and it is mostly a null [s1].

What was analysed

The data come from RECOVER-Adult, a multicentre prospective longitudinal cohort of adults with and without prior SARS-CoV-2 infection, enrolled from October 2021 to January 2024 [s1]. This analysis included participants with a SARS-CoV-2 infection and at least one symptom survey recorded before or during pregnancy [s1].

Participants were classified using the Long COVID Research Index: a score of 11 or higher meant likely Long COVID, and 0 to 10 meant Long COVID-indeterminate [s1]. The primary outcome was preterm birth before 37 weeks; secondary outcomes were hypertensive disorders of pregnancy, caesarean delivery, neonatal intensive care admission, and small-for-gestational-age birthweight below the 10th percentile [s1].

To handle the fact that people with Long COVID differ systematically from people without it, the authors used propensity score methods with full matching to estimate an average treatment effect, with a sensitivity analysis estimating the average treatment effect among the treated [s1].

Among 603 participants, 118 (19.6%) were classified as likely having Long COVID before or during pregnancy [s1].

The groups were not alike

Before any outcome is discussed, the baseline differences are worth stating, because they are the finding a reader is most likely to be able to act on.

Participants classified as likely having Long COVID were more likely to report specific adverse social determinants of health — difficulty covering expenses and paying bills, food insecurity, missed care because of cost, and medical discrimination — and had higher prepregnancy body mass index than those classified as indeterminate [s1].

Whatever else Long COVID classification tracks in this cohort, it tracks economic precarity and experience of discrimination in care.

The results

In the primary average treatment effect analysis, there was no association between likely Long COVID and preterm delivery: adjusted outcome rates of 8.1% in the exposed group versus 9.6% in the unexposed, with an absolute risk reduction of 0.82 (95% CI 0.36–1.90) [s1]. None of the secondary outcomes were associated either [s1].

In the sensitivity analysis restricted to the treated, likely Long COVID was again not associated with preterm delivery (ARR 1.11, 95% CI 0.60–2.06) [s1]. It was, however, associated with an increased risk of hypertensive disorders of pregnancy — adjusted outcome rates of 39.0% versus 25.9%, ARR 1.51 (95% CI 1.12–2.03) — and with no other secondary outcome [s1].

Reading the one positive result

The hypertensive-disorders signal is the part most likely to be reported on its own, so it is worth being precise about what it is: a secondary outcome, significant in a sensitivity analysis and not in the primary one, among five secondary outcomes tested.

That combination is exactly the shape a chance finding takes. It is also the shape a real, modestly sized effect takes in an underpowered study. The paper does not resolve which, and the authors' conclusion says only that the association highlights the need for further research [s1].

The wide confidence intervals around the preterm result are the other thing to notice. An interval running from 0.36 to 1.90 does not establish that Long COVID is harmless for preterm birth; it establishes that 603 pregnancies cannot detect an effect of the size most people would care about. A null result and an absence of evidence look identical from the outside, and here it is closer to the latter.

What "likely Long COVID" means

Everything above depends on a score threshold, and how well that threshold identifies the condition is an open question in its own right.

A validation study published on July 24 within the same RECOVER programme trained a computable phenotype for Long COVID against patient-reported symptoms in 1,501 participants with linked electronic health record data, of whom 376 (25%) met criteria for highly symptomatic Long COVID on the same research index [s2]. Features associated with the condition included clinician diagnosis of shortness of breath, malaise and fatigue, and cardiac dysrhythmias; treatment with albuterol, gabapentin or duloxetine; and elevated heart rate [s2].

The resulting algorithm achieved an AUROC of 0.80 (95% CI 0.74–0.85) and an area under the precision-recall curve of 0.58 (95% CI 0.47–0.69) [s2]. That is useful discrimination and it is not a diagnostic test. Any exposure defined this way carries misclassification, and misclassification of exposure biases estimates toward the null — which is one more reason to read the preterm result as uninformative rather than reassuring.

What to watch

A larger pregnancy cohort with the same index applied, and whether the hypertensive-disorders association appears in a study designed to test it as a primary outcome.

This article is informational and is not medical advice.

Sources

• Pregnancy Outcomes in Individuals With Long COVID — Obstetrics & Gynecology, 2026-08-27
• Characterization and Validation of EHR Computable Phenotypes for Long COVID Using Patient-Reported Symptoms: Insights from the Nationwide RECOVER Program — Journal of the American Medical Informatics Association, 2026-07-24