Pregnancy Outcomes According to Gestational Timing of Maternal SARS-CoV-2 Infection: A Narrative Review

Abstract

Maternal severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection is associated with heterogeneous pregnancy outcomes. This narrative review examines associations between the gestational timing of maternal SARS-CoV-2 infection and maternal, obstetric, fetal, and neonatal outcomes, including maternal well-being. The evidence base comprised English-language publications identified through PubMed/MEDLINE and relevant reference lists between January 2020 and September 2026. Timing-specific observational studies were prioritised, and findings were synthesised qualitatively because of methodological and clinical heterogeneity. Mild first-trimester infection is not consistently associated with pregnancy loss or impaired fetal growth. Large registry data do not show increased major congenital anomalies among live births, although a smaller prospective study reports a conflicting signal. Selected cohorts associate broader early- or midpregnancy exposure windows with stillbirth, but these windows do not isolate the first trimester. Evidence does not establish a trimester-specific preeclampsia gradient. Third-trimester and peripartum infection show the most consistent associations with preterm birth and acute maternal and neonatal morbidity, partly reflecting provider-initiated delivery and prematurity. Placental injury can contribute to fetal compromise without confirmed congenital infection; selected pathology series cannot estimate population risk. Congenital infection remains uncommon, and neonatal positivity alone does not establish intrauterine transmission. A small early-pregnancy cohort documents anxiety symptoms but cannot establish infection-attributable psychological risk. Different gestational exposure periods are associated with distinct outcome profiles, without demonstrating an independent timing effect or universal trimester-based gradient. Interpretation requires attention to severity, vaccination, variant period, healthcare access, and the interval between infection and delivery.

1. Introduction

Gestational timing is clinically relevant when interpreting pregnancy outcomes after maternal severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection, which causes coronavirus disease 2019 (COVID-19). The consequences of infection may differ according to the developmental stage of the fetus, placental maturation, the physiological respiratory and cardiovascular adaptations of pregnancy, and the interval between infection and delivery. Early pregnancy raises questions about pregnancy loss, embryonic development, and placentation; later infection is more directly connected with maternal respiratory compromise, medically indicated delivery, preterm birth, and neonatal admission. The published evidence remains heterogeneous because many early studies enrolled hospitalised women near delivery, whereas later population studies included milder and asymptomatic infections.

During the first trimester, organogenesis and placental establishment represent biologically plausible periods of vulnerability to maternal systemic inflammation, fever, vascular dysfunction, and placental injury. However, the available evidence regarding pregnancy loss and congenital anomalies remains heterogeneous, with differences in study design, sample size, exposure ascertainment, and outcome definitions contributing to inconsistent findings [1,2,3,4,5].

During the second trimester, continued placental development and rapid fetal growth may shape fetal and placental vulnerability to the consequences of maternal SARS-CoV-2 infection. Associations with adverse pregnancy and neonatal outcomes vary across studies, particularly because gestational exposure windows, comparator groups, and outcome definitions differ [6]. Evidence concerning hypertensive disorders and preeclampsia also remains insufficient to establish a specific trimester of greatest risk [7,8].

During the third trimester, increased maternal cardiorespiratory demands, physiological adaptations of haemostasis, and proximity to delivery may influence the clinical consequences of SARS-CoV-2 infection. Associations with adverse obstetric and perinatal outcomes, including preterm birth and emergency caesarean delivery, require careful interpretation because maternal disease severity, obstetric management, and the interval between infection and delivery may influence the observed patterns [9]. Vaccination status and circulating viral variants represent additional factors that should be considered when evaluating these associations.

This review examines how pregnancy outcomes vary according to the trimester or gestational period of maternal SARS-CoV-2 infection. It addresses five questions: whether early infection is associated with pregnancy loss, fetal growth abnormalities, stillbirth, or congenital anomalies; whether second-trimester infection is linked to later placental or hypertensive complications; whether third-trimester and peripartum infection increase preterm birth and caesarean delivery; whether maternal severe morbidity varies by timing; and whether neonatal infection or other morbidity reflects gestational timing rather than prematurity alone. The review also evaluates how variant period, vaccination, disease severity, and study design explain apparently conflicting findings, and considers maternal mental health as an additional patient-centred outcome. These questions require separation of biological timing from ascertainment and management effects, which guides the methods and evidence tables below. The contribution of this review is to distinguish direct timing-specific comparisons from broader pregnancy-wide evidence and to identify which apparent trimester effects remain vulnerable to ascertainment and management bias.

2. Materials and Methods

A focused narrative literature review was conducted to identify studies examining maternal, obstetric, fetal, and neonatal outcomes according to the gestational timing of maternal SARS-CoV-2 infection. The literature search was performed using PubMed, including its MEDLINE component, and covered English-language publications from 1 January 2020 to 16 September 2026, the original search cutoff. Reference lists of relevant primary studies and reviews were also examined to identify additional publications. Selected publications were subsequently checked against the original sources to verify bibliographic details, study characteristics, and reported numerical findings. This targeted verification did not constitute an additional literature search or alter the original search cutoff of 16 September 2026.

The PubMed search strategy was organised into five components, covering SARS-CoV-2 infection, pregnancy, gestational timing, clinical outcomes, and publication restrictions.

The first component (#1), SARS-CoV-2 infection, included the terms “SARS-CoV-2”[tiab] OR “COVID-19”[tiab].

The second component (#2), pregnancy, included pregnan*[tiab] OR gestation*[tiab].

The third component (#3), gestational timing, included trimester*[tiab] OR “gestational age”[tiab] OR “gestational timing”[tiab] OR “early pregnancy”[tiab] OR “late pregnancy”[tiab] OR peripartum[tiab].

The fourth component (#4), pregnancy outcomes, included outcome*[tiab] OR “preterm birth”[tiab] OR stillbirth[tiab] OR preeclampsia[tiab] OR “pre-eclampsia”[tiab] OR miscarriage[tiab] OR “fetal growth”[tiab] OR “congenital anomalies”[tiab] OR “neonatal outcomes”[tiab] OR “vertical transmission”[tiab].

The fifth component (#5), publication restrictions, limited the search to 1 January 2020:16 September 2026[dp] AND english[la].

Search terms within components #1–#4 were combined using OR, while the five components were combined using AND (#1 AND #2 AND #3 AND #4 AND #5). The field tags [tiab], [dp], and [la] denote Title/Abstract, date of publication, and language, respectively.

Publications were considered for the timing-specific synthesis when they evaluated maternal SARS-CoV-2 infection during pregnancy, reported clinically relevant maternal, obstetric, fetal, or neonatal outcomes, and provided usable information on gestational timing. Eligible exposure definitions included infection according to trimester, gestational week, early versus late pregnancy, or the interval between infection and delivery. Titles and abstracts were assessed for relevance to these criteria, and potentially informative publications were examined in full text where accessible. Case reports, editorials, opinion articles, and publications without relevant pregnancy outcomes or interpretable gestational timing were excluded from the timing-specific clinical synthesis.

Primary observational studies, particularly population-based cohorts, prospective and retrospective comparative cohorts, and national surveillance studies, were prioritised. Greater interpretative weight was assigned to studies providing direct gestational-timing comparisons, clearly defined exposure and comparator groups, adequate sample sizes, adjusted effect estimates, and sufficiently documented outcomes. Systematic reviews and meta-analyses were used primarily to contextualise pregnancy-wide associations and identify relevant primary evidence, rather than to infer trimester-specific effects from pooled estimates. Directly relevant placental pathology studies and diagnostic consensus publications were considered for mechanistic interpretation, while evidence concerning maternal psychological well-being and pandemic-related prenatal healthcare access was used to address secondary clinical and methodological considerations. Abstract-only conference reports were not used to support trimester-specific risk conclusions when sufficiently detailed methodological and statistical information was unavailable.

The initial PubMed search identified 245 records. Following title and abstract screening to exclude publications not relevant to the review objectives, 127 records were retained for further assessment. After evaluation of the available full texts and application of the eligibility and relevance criteria described above, 53 publications were retained in the final narrative review. Of these, 13 primary studies were selected for individual presentation in Table 1 because they provided clinically relevant information on gestational exposure timing and pregnancy outcomes. The remaining publications contributed to the broader evidence synthesis, methodological interpretation, and supporting clinical and mechanistic context.

Relevant findings were synthesised qualitatively because the included studies differed in design, population characteristics, gestational exposure definitions, disease severity, comparator groups, vaccination status, circulating viral variants, and outcome ascertainment. Particular attention was given to distinguishing spontaneous from medically indicated preterm birth, gestational timing from the interval between infection and delivery, and direct trimester comparisons from broader early-versus-late pregnancy analyses. Adjusted estimates, confidence intervals, and absolute event frequencies were reported as presented in the original publications where available.

3. Evidence Synthesis

3.1. Characteristics of the Evidence Base

The available studies associate gestational timing with different outcome profiles, but no single trimester uniformly carries the greatest risk for every endpoint. Selected datasets associate early exposure with stillbirth or subsequent preterm birth, whereas evidence for excess early pregnancy loss remains inconclusive and several prospective cohorts report reassuring common outcomes after mild disease. Late pregnancy and peripartum infection show the most consistent relationship with preterm delivery and acute maternal and neonatal morbidity. These patterns may reflect biological pathways as well as testing, surveillance, and obstetric intervention; they do not by themselves identify an independent effect of timing.

The evidence base varies substantially in sample size and clinical spectrum. Large studies include the Israeli cohort of 2753 infected women, Canadian surveillance, the French national cohort of 244,645 births, and the United States (US) electronic-health-record cohort drawn from 73,666 deliveries [10,11,12,13]. Multinational surveillance and meta-analytic studies increase statistical power but often combine pandemic phases and heterogeneous infection and outcome definitions [14,15].

Early studies were enriched for women admitted with symptomatic disease. In a review of 62 studies involving 31,016 pregnant women, 77.7% were in the third trimester and 16.4% developed severe disease [16]. The United Kingdom (UK) national cohort likewise consisted of 427 women admitted to hospital, most of whom were in the late second or third trimester [17]. Such studies describe the clinical burden of diagnosed disease but cannot estimate the risk associated with mild infection in the general pregnant population.

Comparative studies generally support increased maternal morbidity and preterm birth after infection, although estimates vary. An individual-participant meta-analysis found increased maternal and neonatal morbidity but no association with stillbirth or intrauterine growth restriction [14]. Other reviews reported varying associations with preeclampsia, preterm birth, stillbirth, caesarean delivery, and perinatal death [18,19,20]. The early review combining severe acute respiratory syndrome (SARS), Middle East respiratory syndrome (MERS), and COVID-19 provides historical coronavirus context, not a SARS-CoV-2-specific timing estimate [19].

Table 1 compares selected timing-specific studies individually. Exposure windows and comparators are retained as published; however, conventional trimester definitions and broader gestational intervals, such as 0–20 weeks or <28 weeks, do not represent equivalent exposure periods and therefore cannot support direct comparisons of trimester-specific effects across studies. Early serologically detected infection, infection diagnosed during acute illness, and infection at delivery are also not interchangeable exposures. Smaller uncontrolled cohorts receive less inferential weight than comparative studies.

Early cohorts from Wuhan, Sweden, Massachusetts, and New York documented the clinical spectrum and perinatal care practices but were limited by small samples, hospital selection, or incomplete timing information [21,22,23,24]. They provide contextual evidence and do not establish trimester-specific causal effects.

3.2. First- and Second-Trimester Infection

Evidence concerning early infection is mixed. The early-pregnancy systematic review used exposure before 20 weeks, not an isolated first trimester, and included 18 studies and 10,147 infected women; pooled miscarriage odds were 1.44 (95% CI 0.96–2.18) compared with controls [25]. The interval includes both no association and a clinically relevant increase. A longitudinal Chinese cohort reported pregnancy loss before 28 weeks in 30/498 first-trimester-exposed pregnancies (6.02%) versus 41/654 controls (6.27%), without differences in several birth outcomes [26]. That broader loss endpoint included terminations for fetal indications and should not be equated with spontaneous miscarriage alone.

Stillbirth findings are less consistent. In the US electronic-health-record study by Lyu et al., the early category was 0–20 weeks in the methods and tables, followed by 21–27 and 28–43 weeks [27]. Adjusted stillbirth odds were higher for the first two categories versus pregnancies without recorded infection; these are broader gestational windows, not proof of a first-trimester effect (Table 1). In CRONOS, infection < 28 weeks versus ≥28 weeks was associated with stillbirth (aRR 5.76, 95% CI 3.07–10.83) and birth < 32 weeks (aRR 6.07, 95% CI 3.65–10.09) [28]. This comparison is confined to infected pregnancies and is sensitive to unequal time at risk. By contrast, pregnancy-wide individual-participant meta-analysis did not identify a stillbirth association [14]. Ghosh et al. found a first-trimester association with a composite coded fetal-death endpoint (adjusted RR 2.36, 95% CI 1.04–5.36), but this endpoint included stillbirth, fetal death, and abortion codes and is not equivalent to stillbirth alone [29].

Fetal growth findings do not support a universal adverse gradient. Prospective assessment of fetal biometry and Doppler parameters after infection in the second half of pregnancy yielded reassuring findings [30], and the Israeli population cohort found comparable SGA rates across infection trimesters [10]. Danish register data identified increased SGA risk, particularly among hospital-admitted women [31]. Garcia-Aguilar et al. reported SGA in 46/404 infected pregnancies (11.4%) versus 31/404 reference pregnancies (7.7%; aRR 1.60, 95% CI 1.03–2.48), with an association in the first-trimester subgroup (aRR 2.13, 95% CI 1.08–4.20) [8]. Referral-centre recruitment, severity enrichment, and incomplete exclusion of asymptomatic infection in controls limit generalisability; significance within one subgroup does not establish a between-trimester difference.

Table 1. Selected primary studies of pregnancy outcomes according to gestational timing of maternal SARS-CoV-2 infection.

Study	Design and Population	Exposure and Comparator	Pandemic Context	Main Findings	Principal Limitation
Magnus et al., 2024 [5]	Nordic registry; 343,066 liveborn singletons; 10,229 exposed.	PCR-positive test through 13 + 6 weeks; no T1 infection as reference.	Pregnancies starting March 2020–February 2022; index, Alpha and Delta analyses; vaccination-adjusted.	Major anomalies: 494/10,229 exposed versus 17,210/332,837 reference; aOR 0.96 (95% CI 0.87–1.05).	Live births only; infections and anomalies ending in loss/termination incompletely represented.
Luo et al., 2026 [4]	Prospective, three Chinese hospitals; 535 infected and 592 controls analysed.	RT-PCR-confirmed infection through 13 + 6 weeks; uninfected comparator; singleton defect proportions reported separately.	Recruitment December 2022–February 2023; all vaccinated before pregnancy; individual variant NR.	Total defects among singletons: 9.2% versus 4.7%; adjusted reported RR 2.25 (95% CI 1.38–3.56).	Smaller tertiary-hospital cohort; broader defect ascertainment than Nordic registry; limited follow-up.
Badr et al., 2021 [6]	Retrospective, four European hospitals; 393 RT-PCR-positive among 10,925 pregnancies.	Gestational-week analysis; uninfected/untested cohort; live fetus at 10 weeks required.	February–November 2020; before vaccination.	After covariate matching, adverse obstetric composite increased after >20 weeks; neonatal composite after >26 weeks. Published results do not supply a single adjusted trimester RR with CI.	No systematic serial testing; early losses and terminations excluded; composites contain management-dependent outcomes.
Fallach et al., 2022 [10]	Israeli population cohort; 2753 RT-PCR-positive women matched 1:1.	T1 0–12, T2 13–26, T3 ≥ 27 weeks; matched uninfected reference.	February 2020–July 2021; no variant-specific effect or vaccination-stratified estimate used here.	T3 PTB: 51/886 (5.8%) versus 20 reference events (2.3%); aOR 2.76 (95% CI 1.63–4.67). T1/T2 PTB and SGA showed no association.	PTB analysis excludes index dates ≥ 37 weeks; diagnostic timing and clinical intervention affect interpretation.
Piekos et al., 2022 [13]	US retrospective multicentre EHR cohort; 882 PCR-positive and 889 matched negative controls.	T1/T2/T3: 85/226/571; study-defined timing groups; PCR-negative matched reference.	Deliveries March 2020–July 2021; none vaccinated at infection.	Stillbirth: 7/882 (0.8%) versus 1/889 (0.1%); earlier exposure accounted for much of the PTB/stillbirth signal. Outcome comparisons were matched/descriptive, not adjusted trimester RRs.	Very few stillbirths; tested delivery cohort; prediction models do not establish causality.
Lyu et al., 2023 [27]	US retrospective EHR study; 191,403 pregnancies; 2342 early, 2075 mid, 12,697 late infections.	Diagnosis or positive PCR/antibody evidence; 0–20, 21–27, 28–43 weeks; no recorded infection reference.	March 2020–May 2021; vaccination/variant-specific estimates NR.	Stillbirth: early 35/2342 (1.5%), aOR 1.75 (95% CI 1.25–2.46); mid 36/2075 (1.7%), aOR 2.09 (1.49–2.93); respective references 0.8%.	EHR exposure/outcome classification; varying risk sets. Early category includes T1 and part of T2.
Iannaccone et al., 2024 [28]	Prospective CRONOS register; 8032 infected pregnancies in Germany/Austria.	Confirmed infection < 28 weeks (3119) versus ≥ 28 (4913); no uninfected reference.	April 2020–December 2022; wild-type through Omicron; 2156 vaccinated, 702 status missing.	Stillbirth: 46/3119 (1.47%) versus 24/4913 (0.49%); aRR 5.76 (95% CI 3.07–10.83). PTB < 32 weeks: 3.62% versus 1.05%; aRR 6.07 (3.65–10.09).	Hospital/tertiary-centre selection; unequal opportunity for early delivery; spontaneous/indicated PTB not separated.
Hughes et al., 2023 [7]	Retrospective, 17 US hospitals; 402 exposed pregnancies (414 fetuses/neonates) and 11,705 controls.	Positive viral test < 28 weeks before delivery admission; no recorded positive test reference.	March–December 2020; before vaccination.	Fetal/neonatal death: 2.9% versus 1.5%, aRR 1.97 (95% CI 1.01–3.85). PTB: 19.6% versus 13.8%, aRR 1.29 (1.02–1.63). HDP with preterm delivery: 7.2% versus 4.1%, aRR 1.74 (1.19–2.55).	Deliveries < 20 weeks excluded; death endpoint is composite; no overall HDP or severe-preeclampsia association.
Hu et al., 2025 [26]	Single-centre longitudinal Chinese cohort; 498 exposed and 654 controls.	Positive RT-PCR through 13 + 6 weeks; comparator without T1 infection.	Last menstrual period October 2022–April 2023; 429/468 infected women who delivered vaccinated.	Loss before 28 weeks: 30/498 (6.02%) versus 41/654 (6.27%); no difference in several birth outcomes. Gestational hypertension increased, distinct from preeclampsia.	Loss includes fetal-indication terminations; single centre; predominantly mild infection.
Khalil et al., 2024 [32]	Single-centre Egyptian cohort; 105 RT-PCR-positive women; T1/T2/T3 14/25/66.	Trimester of infection; infected groups only; numerical boundaries NR.	December 2021–May 2022; vaccination and individual variants NR.	Hospitalisation: 7/14 (50%), 13/25 (52%), 55/66 (83.3%). Severity, rather than an adjusted trimester contrast, predicted admission: aOR 3.9 (95% CI 1.6–9.2).	No uninfected control; severity imbalance; internal reporting inconsistencies limit precision and generalisability.
Ghosh et al., 2024 [29]	Mexican retrospective IMSS cohort; 17,340 tested singleton pregnancies; 5122 positive, 12,218 negative.	RT-PCR/antigen-positive versus test-negative; author-defined T1/T2/T3.	January 2020–November 2021; evolving variants/vaccination; no stratified effects used.	T1 coded fetal-death composite: adjusted RR 2.36 (95% CI 1.04–5.36). Overall fetal deaths 34/5122 (0.7%) versus 71/12,218 (0.6%); overall PTB RR 0.96 (0.88–1.06).	Composite includes fetal death, stillbirth and abortion codes; tested-only and complete-case selection; overall frequencies are not T1 frequencies.
Seif et al., 2023 [33]	Retrospective single-health-system, recovered-only cohort; 298 women; T1/T2/T3 48/123/127.	Positive test > 10 days before delivery; infected timing groups only.	March 2020–June 2022; vaccination status similar between groups; variant detail NR.	Unadjusted PTB < 37 weeks: T1 16/48 (33.3%), T2 23/123 (18.7%), T3 15/127 (11.8%); overall p = 0.006. Acute admission/oxygen use was greater in T2/T3. No adjusted trimester effects reported.	Small recovered-only cohort; no uninfected comparator; time-at-risk and clinical/management confounding; PTB pathways not separated.
Garcia-Aguilar et al., 2026 [8]	Prospective Barcelona cohort; 404 PCR-positive and 404 reference pregnancies; T1/T2/T3 57/171/176.	<14, 14–27, >27 weeks; reference pregnancies without clinical suspicion of infection.	March 2020–June 2023; mixed pandemic periods; vaccination/variant-stratified estimates NR.	SGA: 46/404 (11.4%) versus 31/404 (7.7%); aRR 1.60 (95% CI 1.03–2.48). Iatrogenic PTB: 11.4% versus 3.7%, aRR 2.44 (1.38–4.33); spontaneous PTB not increased.	Severe/referral cases enriched; controls not systematically tested; timing/severity subgroup precision limited.

Notes: Estimates are reproduced as reported; no new adjusted estimates were calculated. Event frequencies are unadjusted and use the published denominator, which may differ by outcome. T1, T2, and T3 denote author-defined trimesters; numerical boundaries are provided when verified. Lyu et al. use 0–20 weeks in the methods and tables, although the abstract labels early infection < 20 weeks. Neither that category nor CRONOS < 28 weeks isolates T1. Calendar dates do not establish individual viral variants. Rows are not independent pooled replications, and study overlap is possible. Abbreviations: aOR, adjusted odds ratio; aRR, adjusted risk ratio; CI, confidence interval; CRONOS, COVID-19 Related Obstetrics and Neonatal Outcome Study; EHR, electronic health record; HDP, hypertensive disorders of pregnancy; NR, not reported or not verified in accessible source detail; PTB, preterm birth; PCR, polymerase chain reaction; RR, risk ratio; RT-PCR, reverse transcription polymerase chain reaction; SGA, small for gestational age.

Table 1. Selected primary studies of pregnancy outcomes according to gestational timing of maternal SARS-CoV-2 infection.

Study	Design and Population	Exposure and Comparator	Pandemic Context	Main Findings	Principal Limitation
Magnus et al., 2024 [5]	Nordic registry; 343,066 liveborn singletons; 10,229 exposed.	PCR-positive test through 13 + 6 weeks; no T1 infection as reference.	Pregnancies starting March 2020–February 2022; index, Alpha and Delta analyses; vaccination-adjusted.	Major anomalies: 494/10,229 exposed versus 17,210/332,837 reference; aOR 0.96 (95% CI 0.87–1.05).	Live births only; infections and anomalies ending in loss/termination incompletely represented.
Luo et al., 2026 [4]	Prospective, three Chinese hospitals; 535 infected and 592 controls analysed.	RT-PCR-confirmed infection through 13 + 6 weeks; uninfected comparator; singleton defect proportions reported separately.	Recruitment December 2022–February 2023; all vaccinated before pregnancy; individual variant NR.	Total defects among singletons: 9.2% versus 4.7%; adjusted reported RR 2.25 (95% CI 1.38–3.56).	Smaller tertiary-hospital cohort; broader defect ascertainment than Nordic registry; limited follow-up.
Badr et al., 2021 [6]	Retrospective, four European hospitals; 393 RT-PCR-positive among 10,925 pregnancies.	Gestational-week analysis; uninfected/untested cohort; live fetus at 10 weeks required.	February–November 2020; before vaccination.	After covariate matching, adverse obstetric composite increased after >20 weeks; neonatal composite after >26 weeks. Published results do not supply a single adjusted trimester RR with CI.	No systematic serial testing; early losses and terminations excluded; composites contain management-dependent outcomes.
Fallach et al., 2022 [10]	Israeli population cohort; 2753 RT-PCR-positive women matched 1:1.	T1 0–12, T2 13–26, T3 ≥ 27 weeks; matched uninfected reference.	February 2020–July 2021; no variant-specific effect or vaccination-stratified estimate used here.	T3 PTB: 51/886 (5.8%) versus 20 reference events (2.3%); aOR 2.76 (95% CI 1.63–4.67). T1/T2 PTB and SGA showed no association.	PTB analysis excludes index dates ≥ 37 weeks; diagnostic timing and clinical intervention affect interpretation.
Piekos et al., 2022 [13]	US retrospective multicentre EHR cohort; 882 PCR-positive and 889 matched negative controls.	T1/T2/T3: 85/226/571; study-defined timing groups; PCR-negative matched reference.	Deliveries March 2020–July 2021; none vaccinated at infection.	Stillbirth: 7/882 (0.8%) versus 1/889 (0.1%); earlier exposure accounted for much of the PTB/stillbirth signal. Outcome comparisons were matched/descriptive, not adjusted trimester RRs.	Very few stillbirths; tested delivery cohort; prediction models do not establish causality.
Lyu et al., 2023 [27]	US retrospective EHR study; 191,403 pregnancies; 2342 early, 2075 mid, 12,697 late infections.	Diagnosis or positive PCR/antibody evidence; 0–20, 21–27, 28–43 weeks; no recorded infection reference.	March 2020–May 2021; vaccination/variant-specific estimates NR.	Stillbirth: early 35/2342 (1.5%), aOR 1.75 (95% CI 1.25–2.46); mid 36/2075 (1.7%), aOR 2.09 (1.49–2.93); respective references 0.8%.	EHR exposure/outcome classification; varying risk sets. Early category includes T1 and part of T2.
Iannaccone et al., 2024 [28]	Prospective CRONOS register; 8032 infected pregnancies in Germany/Austria.	Confirmed infection < 28 weeks (3119) versus ≥ 28 (4913); no uninfected reference.	April 2020–December 2022; wild-type through Omicron; 2156 vaccinated, 702 status missing.	Stillbirth: 46/3119 (1.47%) versus 24/4913 (0.49%); aRR 5.76 (95% CI 3.07–10.83). PTB < 32 weeks: 3.62% versus 1.05%; aRR 6.07 (3.65–10.09).	Hospital/tertiary-centre selection; unequal opportunity for early delivery; spontaneous/indicated PTB not separated.
Hughes et al., 2023 [7]	Retrospective, 17 US hospitals; 402 exposed pregnancies (414 fetuses/neonates) and 11,705 controls.	Positive viral test < 28 weeks before delivery admission; no recorded positive test reference.	March–December 2020; before vaccination.	Fetal/neonatal death: 2.9% versus 1.5%, aRR 1.97 (95% CI 1.01–3.85). PTB: 19.6% versus 13.8%, aRR 1.29 (1.02–1.63). HDP with preterm delivery: 7.2% versus 4.1%, aRR 1.74 (1.19–2.55).	Deliveries < 20 weeks excluded; death endpoint is composite; no overall HDP or severe-preeclampsia association.
Hu et al., 2025 [26]	Single-centre longitudinal Chinese cohort; 498 exposed and 654 controls.	Positive RT-PCR through 13 + 6 weeks; comparator without T1 infection.	Last menstrual period October 2022–April 2023; 429/468 infected women who delivered vaccinated.	Loss before 28 weeks: 30/498 (6.02%) versus 41/654 (6.27%); no difference in several birth outcomes. Gestational hypertension increased, distinct from preeclampsia.	Loss includes fetal-indication terminations; single centre; predominantly mild infection.
Khalil et al., 2024 [32]	Single-centre Egyptian cohort; 105 RT-PCR-positive women; T1/T2/T3 14/25/66.	Trimester of infection; infected groups only; numerical boundaries NR.	December 2021–May 2022; vaccination and individual variants NR.	Hospitalisation: 7/14 (50%), 13/25 (52%), 55/66 (83.3%). Severity, rather than an adjusted trimester contrast, predicted admission: aOR 3.9 (95% CI 1.6–9.2).	No uninfected control; severity imbalance; internal reporting inconsistencies limit precision and generalisability.
Ghosh et al., 2024 [29]	Mexican retrospective IMSS cohort; 17,340 tested singleton pregnancies; 5122 positive, 12,218 negative.	RT-PCR/antigen-positive versus test-negative; author-defined T1/T2/T3.	January 2020–November 2021; evolving variants/vaccination; no stratified effects used.	T1 coded fetal-death composite: adjusted RR 2.36 (95% CI 1.04–5.36). Overall fetal deaths 34/5122 (0.7%) versus 71/12,218 (0.6%); overall PTB RR 0.96 (0.88–1.06).	Composite includes fetal death, stillbirth and abortion codes; tested-only and complete-case selection; overall frequencies are not T1 frequencies.
Seif et al., 2023 [33]	Retrospective single-health-system, recovered-only cohort; 298 women; T1/T2/T3 48/123/127.	Positive test > 10 days before delivery; infected timing groups only.	March 2020–June 2022; vaccination status similar between groups; variant detail NR.	Unadjusted PTB < 37 weeks: T1 16/48 (33.3%), T2 23/123 (18.7%), T3 15/127 (11.8%); overall p = 0.006. Acute admission/oxygen use was greater in T2/T3. No adjusted trimester effects reported.	Small recovered-only cohort; no uninfected comparator; time-at-risk and clinical/management confounding; PTB pathways not separated.
Garcia-Aguilar et al., 2026 [8]	Prospective Barcelona cohort; 404 PCR-positive and 404 reference pregnancies; T1/T2/T3 57/171/176.	<14, 14–27, >27 weeks; reference pregnancies without clinical suspicion of infection.	March 2020–June 2023; mixed pandemic periods; vaccination/variant-stratified estimates NR.	SGA: 46/404 (11.4%) versus 31/404 (7.7%); aRR 1.60 (95% CI 1.03–2.48). Iatrogenic PTB: 11.4% versus 3.7%, aRR 2.44 (1.38–4.33); spontaneous PTB not increased.	Severe/referral cases enriched; controls not systematically tested; timing/severity subgroup precision limited.

Notes: Estimates are reproduced as reported; no new adjusted estimates were calculated. Event frequencies are unadjusted and use the published denominator, which may differ by outcome. T1, T2, and T3 denote author-defined trimesters; numerical boundaries are provided when verified. Lyu et al. use 0–20 weeks in the methods and tables, although the abstract labels early infection < 20 weeks. Neither that category nor CRONOS < 28 weeks isolates T1. Calendar dates do not establish individual viral variants. Rows are not independent pooled replications, and study overlap is possible. Abbreviations: aOR, adjusted odds ratio; aRR, adjusted risk ratio; CI, confidence interval; CRONOS, COVID-19 Related Obstetrics and Neonatal Outcome Study; EHR, electronic health record; HDP, hypertensive disorders of pregnancy; NR, not reported or not verified in accessible source detail; PTB, preterm birth; PCR, polymerase chain reaction; RR, risk ratio; RT-PCR, reverse transcription polymerase chain reaction; SGA, small for gestational age.

Congenital-anomaly evidence is conflicting. Small early cohorts did not identify a clear excess, but their ability to detect rare defects was limited [1,2]. The Chinese prospective study reported total birth defects in 9.2% of exposed singleton pregnancies versus 4.7% of controls and an adjusted relative risk of 2.25 (95% confidence interval 1.38–3.56) [4]. Its tertiary-hospital recruitment and limited follow-up constrain generalisability. In contrast, the much larger Nordic registry study did not identify an increase in major anomalies after first-trimester infection [5]. The studies differ in outcome definitions and ascertainment, and live-birth-based analyses may miss anomalies associated with pregnancy loss or termination. Together, they support continued research rather than a conclusion that SARS-CoV-2 is an established human teratogen.

Published evidence on hypertension warrants an outcome-specific interpretation. Hughes et al. compared infection < 28 weeks with no recorded positive test: HDP accompanied by delivery < 37 weeks occurred in 7.2% versus 4.1% (aRR 1.74, 95% CI 1.19–2.55), while overall HDP and preeclampsia with severe features were not significantly increased [7]. Hu et al. reported more gestational hypertension after first-trimester infection, a finding distinct from preeclampsia [26]. Garcia-Aguilar et al. found no clear preeclampsia association overall (aRR 1.17, 95% CI 0.60–2.29) or by trimester; angiogenic assessment was used to distinguish a preeclampsia-like syndrome [8]. These cohorts do not establish a trimester-specific preeclampsia gradient.

3.3. Third-Trimester and Peripartum Infection

Late-pregnancy infection has a reproducible association with preterm birth, but spontaneous and medically indicated delivery must be separated. In the Israeli cohort, third-trimester infection was associated with PTB (aOR 2.76, 95% CI 1.63–4.67), particularly after 34 weeks (aOR 7.10, 95% CI 2.44–20.61) [10]. The Spanish universal-screening cohort reported PTB in 13.8% of infected versus 6.7% of uninfected pregnancies (aOR 2.12, 95% CI 1.32–3.36); iatrogenic delivery unrelated to preterm prelabour rupture of membranes was more frequent (4.9% versus 1.3%), whereas spontaneous PTB proportions were similar (6.1% versus 4.7%) [34]. Garcia-Aguilar et al. likewise found an association with iatrogenic PTB (aRR 2.44, 95% CI 1.38–4.33), but not spontaneous PTB (aRR 0.88, 95% CI 0.48–1.62); third-trimester iatrogenic PTB had an aRR of 4.38 (95% CI 2.38–8.06) [8].

This distinction is not uniform across all datasets: the French national cohort reported increased spontaneous as well as induced PTB [12]. Pregnancy-wide meta-analysis estimated a PTB RR of 1.71 (95% CI 1.28–2.29), and predominantly hospital-based reports described high absolute PTB proportions [14,16,35]. These findings do not identify a trimester-specific spontaneous-labour mechanism. Respiratory deterioration and maternal hypoxaemia may prompt delivery, while fetal compromise, preeclampsia, or other obstetric indications represent separate pathways. Infection diagnosed during a delivery admission can also be detected because an obstetric event brought the patient into care. These indications and the infection-to-delivery interval should be reported separately.

Preterm birth is a multifactorial outcome encompassing spontaneous preterm labour with intact membranes, preterm prelabour rupture of membranes (pPROM), and medically indicated preterm delivery, which may involve distinct underlying pathways. Risk factors unrelated to SARS-CoV-2 infection, including previous preterm birth, cervical or uterine factors, multiple gestation, and maternal or placental complications, may also contribute. As contextual evidence, Swiercz et al. investigated 1164 pregnancies and reported associations between lower first-trimester pregnancy-associated plasma protein A (PAPP-A) and free β-human chorionic gonadotropin concentrations and subsequent preterm birth or pPROM, although their predictive performance was limited [36]. These findings illustrate the complexity of preterm birth risk and should not be interpreted as evidence of an infection-specific causal mechanism.

The small uncontrolled Iranian cohort reported PTB in 3/20 (15.0%), 10/87 (11.5%), and 29/110 (26.4%) women infected in T1, T2, and T3, respectively [37]. These descriptive proportions cannot establish infection-attributable or adjusted trimester risks. In Seif et al.’s retrospective cohort of 298 recovered women (positive test > 10 days before delivery), PTB < 37 weeks occurred in 16/48 (33.3%), 23/123 (18.7%), and 15/127 (11.8%) pregnancies following T1, T2, and T3 infection, respectively (overall p = 0.006) [33]. These unadjusted comparisons do not separate PTB pathways; recovered-only selection, small sample size, the absence of an uninfected comparator, and potential time-at-risk and clinical/management confounding preclude inference of an independent timing effect.

Preeclampsia and caesarean delivery are more frequent in several pregnancy-wide or at-birth comparisons, but these are not direct early-versus-late contrasts. Meta-analysis found increased preeclampsia odds after infection [18,38], and the published INTERCOVID cohort reported an adjusted RR of 1.76 (95% CI 1.27–2.43) for preeclampsia/eclampsia [15]. English at-birth surveillance also identified increased preeclampsia and emergency caesarean delivery [9], while French data found no significant difference in stillbirth or placental abruption [12]. These observations should not be used to assign a highest-risk trimester for preeclampsia.

3.4. Maternal Outcomes Across Gestational Timing

Severe maternal morbidity is associated with infection throughout pregnancy, but the absolute clinical burden often increases when infection occurs in late gestation or close to delivery. The individual-participant meta-analysis reported relative risks of 7.68 for maternal mortality (95% confidence interval 1.70–34.61), 3.81 for intensive-care admission (2.03–7.17), 15.23 for mechanical ventilation (4.32–53.71), and 5.50 for thromboembolic disease (1.12–27.12), compared with uninfected pregnancies [14]. The wide intervals for rare outcomes indicate substantial uncertainty in effect magnitude. These estimates refer to infection at any pregnancy stage and should not be interpreted as trimester-specific effects. A US hospital cohort similarly found increased maternal mortality or serious morbidity from obstetric complications among infected pregnant and postpartum individuals [39].

Population data clarify timing and severity. Quebec infection-associated severe maternal morbidity had an RR of 2.02 (95% CI 1.76–2.31); severe antepartum and peripartum infection had RRs of 13.60 (10.72–17.26) and 20.93 (17.11–25.60), respectively [40]. Scottish data found adjusted odds of 1.72 (95% CI 1.39–2.12) for critical care or death and 2.53 (1.47–4.35) for venous thromboembolism (VTE) [41]. In Khalil et al.’s 105-woman infected cohort, hospitalisation occurred in 55/66 women with T3 infection versus 7/14 with T1 and 13/25 with T2 infection; marked severity differences and reporting inconsistencies limit inference about an independent timing effect [32]. Table 2 separates pregnancy-wide estimates from timing interpretation.

Table 2. Maternal outcomes and timing-related interpretation.

Maternal Endpoint	Evidence	Timing-Related Interpretation
Maternal mortality	RR 7.68 (95% CI 1.70–34.61), infected versus uninfected pregnancies [14]	Increased across pregnancy, but not reported as a uniform trimester effect
Intensive care	RR 3.81 (95% CI 2.03–7.17), same meta-analysis [14]	Strongly influenced by acute disease severity
Mechanical ventilation	RR 15.23 (95% CI 4.32–53.71), same meta-analysis [14]	Reflects severe respiratory disease rather than gestational timing alone
Thromboembolic disease	RR 5.50 (95% CI 1.12–27.12); Scottish VTE aOR 2.53 (95% CI 1.47–4.35) [14,41]	Risk is compatible with combined infection-related and pregnancy-related thrombosis
Severe maternal morbidity	Overall RR 2.02 (95% CI 1.76–2.31); severe peripartum RR 20.93 (95% CI 17.11–25.60) [40]	Very high estimate pertains to severe infection near delivery, not timing alone
Preeclampsia	Pregnancy-wide pooled aOR 1.58 (95% CI 1.39–1.80) [38]; published comparative cohorts give differing findings [7,8,15]	Pregnancy-wide association does not establish a highest-risk trimester; HDP with preterm delivery is a distinct endpoint.

Note: Estimates use different populations, outcomes, and comparators and are not interchangeable. Severe peripartum infection is distinct from all peripartum infection. Abbreviations: RR, risk ratio; aOR, adjusted odds ratio; CI, confidence interval; HDP, hypertensive disorders of pregnancy; VTE, venous thromboembolism.

Large peripartum estimates are strongly influenced by acute disease severity and hospital selection; they should not be presented as effects of late gestational timing alone.

3.5. Neonatal Outcomes and Vertical Transmission

Neonatal morbidity is associated with maternal infection in broad comparative analyses, but prematurity, maternal illness, and observation practices contribute. Individual-participant meta-analysis found increased neonatal admission and low birth weight [14]. Spanish screening data reported neonatal intensive-care admission in 9.3% versus 2.4% (aOR 4.62, 95% CI 2.43–8.94) [34]. The New York dyad cohort found greater prematurity and intensive-care use with symptomatic maternal infection [22]. Swedish national data similarly found a small increase in neonatal admission (11.7% versus 8.4%; matched OR 1.47, 95% CI 1.26–1.70), interpreted as exploratory [42].

These neonatal outcomes should be interpreted according to their underlying clinical pathways rather than attributed directly to SARS-CoV-2 infection. Neonatal intensive-care unit (NICU) admission may reflect prematurity, maternal disease severity, institutional admission policies, or precautionary monitoring. Similarly, low birth weight may result from shortened gestation rather than fetal growth restriction. Neither outcome independently demonstrates direct fetal or neonatal infection. Evidence of congenital infection requires specific virological assessment, interpreted in relation to specimen type, sampling time, and the possibility of intrapartum or postnatal acquisition.

Direct neonatal infection remains uncommon in the cited cohorts. UK surveillance detected SARS-CoV-2 ribonucleic acid (RNA) in 12 of 265 liveborn infants, six within 12 h of birth [17]. WAPM identified one positive infant among 251 liveborn infants [35]. A French prospective study enrolled 165 dyads and confirmed one congenital infection, with a reported rate of 0.63% [43]. Early prospective cohorts found no detected intrauterine transmission [44,45]. The Wuhan report of early neonatal positivity and Massachusetts perinatal data describe neonatal test results, not automatic confirmation of congenital infection [21,46].

A positive neonatal nasopharyngeal test shortly after birth does not by itself establish intrauterine transmission. Classification requires the timing and type of samples, evidence of fetal or placental exposure, and confirmation of neonatal infection; contamination, intrapartum exposure, and postnatal acquisition must be considered [43]. Consequently, positivity rates and confirmed congenital-transmission rates should be reported separately.

The French antibody-transfer ratio increased with the interval between maternal infection and delivery, providing a more direct explanation than trimester alone [43]. Infection close to birth leaves less time for transfer and may increase neonatal testing or observation. A late-pregnancy cohort reported slightly more neonatal positivity and longer hospital stay, but no significant difference in selected neonatal complications or height and weight measures at six months [47]. These anthropometric findings do not constitute a comprehensive neurodevelopmental assessment.

Table 3 separates neonatal admission, low birth weight, positive neonatal swabs, confirmed congenital infection, and follow-up endpoints. None should be used as a surrogate for the others.

Table 3. Fetal and neonatal outcomes according to timing of maternal infection.

Neonatal or Fetal Endpoint	Early or Midpregnancy Evidence	Late or Peripartum Evidence
Neonatal intensive-care admission	Pregnancy-wide association; no separate early/midpregnancy estimate [14]	Increased in symptomatic maternal disease and late infection cohorts [22,34]
Low birth weight	Modest pregnancy-wide increase; no separate early/midpregnancy estimate [14]	Also reported in reviews of predominantly hospitalised women [16]
Vertical transmission	No viral RNA in 22 tested amniotic-fluid samples from a 55-woman cohort, or 140 samples from 134 women in another study [44,45]	Neonatal positivity is distinct from confirmed congenital infection [35,43]
Neonatal SARS-CoV-2 positivity	Early studies mainly assessed amniotic-fluid transmission; no comparable neonatal-swab positivity rate was reported [44,45]	12 positive infants among 265 live births in UK surveillance; one confirmed congenital infection in the French study [17,43]
Antibody transfer	More efficient with earlier maternal infection	Transfer depends on time from infection to delivery, not simply trimester label [43]
Six-month infant anthropometry	Limited evidence for early/midpregnancy exposure	No difference in reported height and weight measures at six months in one late-infection cohort [47]

Note: Neonatal SARS-CoV-2 positivity and confirmed congenital infection are different endpoints. The pregnancy-wide neonatal admission and low-birth-weight estimates are not specific to early or midpregnancy exposure. Absence of detected transmission in a small cohort does not establish zero risk. NICU admission and low birth weight are non-specific outcomes that may be influenced by prematurity, maternal illness, and clinical management practices; neither should be interpreted as direct evidence of neonatal viral injury. Abbreviations: SARS-CoV-2, severe acute respiratory syndrome coronavirus 2; UK, United Kingdom.

Earlier infection may allow a longer interval for antibody transfer, whereas late or peripartum illness more often coincides with prematurity and neonatal observation. The cited cohorts support uncommon congenital infection, with substantial uncertainty for rare events [17,35,43,44,45].

3.6. Integrated Summary by Gestational Exposure Window

Table 4 summarises outcome-specific findings using study-defined exposure windows. Broader early categories are not reassigned to a single trimester. The strongest late-pregnancy associations concern PTB and acute morbidity; early loss, stillbirth, growth, and hypertensive findings remain dependent on their specific endpoint and comparator.

Table 4. Outcome-specific synthesis retaining the original gestational exposure definitions.

Outcome	Gestational Exposure Definition	Evidence Synthesis	Interpretation
Pregnancy loss	T1 cohorts; broader early exposure < 20 weeks in meta-analysis.	No consistent excess; pooled miscarriage OR 1.44 (95% CI 0.96–2.18) [25]. Hu loss endpoint was before 28 weeks and included terminations [26].	Imprecision remains; do not treat all pregnancy losses as spontaneous miscarriage.
Stillbirth and fetal death	Lyu: 0–20 and 21–27 weeks; CRONOS: <28 versus ≥28 weeks.	Stillbirth associations in these broader windows [27,28]. Ghosh T1 result concerns a coded fetal-death composite [29].	No isolated T1 stillbirth inference from Lyu/CRONOS; different comparators and time at risk.
Preeclampsia and HDP	Pregnancy-wide/at-birth comparisons; Hughes exposure < 28 weeks.	Preeclampsia association in meta-analysis and INTERCOVID [15,38]; HDP with preterm delivery in Hughes [7]; no clear preeclampsia association in Garcia-Aguilar [8].	No established highest-risk trimester; HDP with preterm delivery is not equivalent to all preeclampsia.
Preterm birth	Author-defined trimesters, recovered infection, or diagnosis near birth.	Late infection most consistently associated with PTB [10,34]; earlier subsequent PTB in some cohorts [7,13,33].	Spontaneous and indicated PTB differ; severity, delivery indications, recovery interval and risk sets matter.
Fetal growth and SGA	T1 or broader early infection; infection in second half of pregnancy.	Many cohorts reassuring [10,30,44]; SGA associations in Danish and Barcelona data [8,31].	No stable trimester gradient; SGA at birth differs from antenatal growth restriction.
Major congenital anomalies	T1 infection, with live-birth or broader prospective ascertainment.	Large Nordic registry reassuring; smaller Chinese prospective study conflicts [4,5].	Outcome definitions, losses, terminations and diagnostic follow-up affect comparability; no established teratogenic mechanism.
Neonatal outcomes and transmission	Early/midpregnancy versus late/peripartum; infection-to-delivery interval.	Admission often reflects prematurity and acute maternal illness [14,22,34,42]. Congenital infection uncommon [43]; earlier exposure allows more time for antibody transfer.	Neonatal swab positivity, placental infection and confirmed congenital infection are distinct endpoints.

Note: This is a qualitative synthesis, not pooled trimester-specific risk estimation. Early exposure categories of 0–20 weeks or <28 weeks span more than the first trimester. A significant result in one subgroup and a nonsignificant result in another does not establish a between-group difference. Abbreviations: CI, confidence interval; CRONOS, COVID-19 Related Obstetrics and Neonatal Outcome Study; HDP, hypertensive disorders of pregnancy; PTB, preterm birth; SGA, small for gestational age.

4. Biological Mechanisms and Modifiers of Gestational Risk

4.1. Interpreting Outcome-Specific Timing Patterns

The literature supports an outcome-specific pattern rather than a uniform increase in risk from one trimester to the next. Associations with later fetal death after broader early exposure differ from associations with respiratory deterioration and delivery near birth [7,9,14,27,28]. Timing may identify a developmental window, a period of acute maternal illness, or simply the interval available for an outcome to occur. These explanations should be assessed before interpreting a gestational category as a biological risk period.

4.2. Early and Midpregnancy Vulnerability

Early and midpregnancy may represent periods of placental or fetal vulnerability, but clinical categories must remain intact. Lyu et al. analysed 0–20 and 21–27 weeks, and CRONOS compared <28 with ≥28 weeks; neither provides an isolated first-trimester stillbirth estimate [27,28]. A delayed placental pathway is plausible, but these associations cannot distinguish it fully from different time at risk, selection of ongoing pregnancies, or infection ascertainment. The coded composite outcome in the Mexican cohort also differs from stillbirth alone [29].

The hypertension evidence supports similar caution. Pregnancy-wide preeclampsia associations [15,38] coexist with no clear preeclampsia association in the prospective Garcia-Aguilar cohort [8]. The Hughes early-exposure result concerns HDP with preterm delivery, not an established first- or second-trimester preeclampsia effect [7]. Thus, a placental vascular mechanism remains a hypothesis rather than a demonstrated timing-specific causal pathway.

The congenital-anomaly question should be separated from miscarriage. The Nordic registry findings provide stronger statistical reassurance for major anomalies among live births than the small early cohorts, whereas the Chinese prospective signal requires independent replication [4,5]. Neither reassuring live-birth findings nor an isolated positive study resolves risks among pregnancies ending before birth or establishes a causal teratogenic mechanism. The gestational age at which a malformation is diagnosed must also be distinguished from the gestational age at which it developed.

4.3. Placental and Inflammatory Pathways

The biological interpretation must remain proportional to the clinical evidence. Systemic inflammation, endothelial dysfunction, respiratory compromise, and thromboembolic disease provide plausible pathways connecting SARS-CoV-2 infection with adverse outcomes, all of which are relevant to pregnancy physiology. The association between infection and preeclampsia, severe maternal morbidity, and thrombosis is compatible with vascular and inflammatory pathways [14,38]. It is less certain whether these pathways differ intrinsically by trimester or whether later pregnancy amplifies their clinical expression.

Placental dysfunction is a plausible explanation for stillbirth, fetal growth abnormalities, and preeclampsia, but fetal growth studies do not show a consistent pattern. Prospective fetal biometry and Doppler assessments found no effect after infection in the second half of pregnancy [30], while Danish register data found increased small-for-gestational-age risk, particularly in women admitted to hospital [31]. This contrast suggests that growth-related associations may depend on severity and population characteristics; it does not exclude acute placental injury in women with mild maternal symptoms.

SARS-CoV-2 placentitis typically combines chronic histiocytic intervillositis, trophoblast necrosis, and extensive perivillous fibrin deposition; severe cases may meet criteria for massive perivillous fibrin deposition [48,49]. In the selected 68-case fatal series, all placentas showed trophoblast necrosis and increased fibrin, 66 showed chronic histiocytic intervillositis, and 63 massive perivillous fibrin deposition [48]. Virus was localised predominantly to syncytiotrophoblast using immunohistochemistry, with RNA in situ hybridisation in some cases; some placentas were positive only by PCR, so diagnostic certainty was not uniform. Severe injury occurred without severe maternal respiratory symptoms. The study supports placental insufficiency and hypoxic injury but, being selected for placental positivity and perinatal death, cannot estimate incidence or trimester-specific population risk.

The National Institutes of Health/Eunice Kennedy Shriver National Institute of Child Health and Human Development (NIH/NICHD) consensus grades placental infection by the strength of virological evidence: definite requires localised evidence of viral replication; probable requires localised viral RNA or protein without demonstrated replication; possible includes PCR-positive homogenates or viral-like particles without localisation; placentas with negative test results are classified as unlikely cases of infection, whereas untested placentas are classified separately [49]. Validated immunohistochemistry for nucleocapsid or spike protein and RNA in situ hybridisation permit cellular localisation, but routine positive staining does not itself prove active replication. Detection of negative-strand or double-stranded replication intermediates provides stronger evidence. Appropriate positive and negative controls, sampling from multiple sites, preservation, and assay validation matter: infection can be focal, and tissue PCR alone may reflect blood contamination. These categories describe placental evidence, not confirmed fetal transmission.

Placental viral infection, secondary placental injury, and fetal infection are separate findings. Nonspecific inflammatory or maternal/fetal vascular malperfusion lesions, and even the placentitis triad in isolation, do not demonstrate SARS-CoV-2 tissue infection [49]. Placental damage can compromise oxygen exchange without fetal infection. Conversely, detecting virus in a fetal sample does not establish that it caused death: the international series included occasional fetal tissue positivity but found no evidence that direct fetal infection explained the deaths [48]. Congenital transmission requires appropriately timed fetal or neonatal confirmation, as discussed in Section 3.5.

4.4. Maternal Severity and Baseline Vulnerability

Severity can have more than one role in timing analyses. Preexisting vulnerability can affect both the likelihood of severe infection and pregnancy outcome, while respiratory deterioration after infection can mediate the decision to deliver. Severe rather than mild COVID-19 is associated with higher odds of several adverse outcomes in meta-analysis [18]. Adjusting for severity or hospitalisation may therefore answer a different question from estimating the total association with infection. Analyses should distinguish baseline risk factors, clinical severity after infection, and consequences of management rather than treating all three as interchangeable confounders.

For pregnancy outcomes, maternal respiratory severity and placental injury should therefore be considered related but distinct domains. Severe COVID-19 strongly influences maternal critical-care and delivery outcomes [14,39,40], whereas the placentitis series identifies a route to fetal compromise in the absence of severe maternal respiratory illness [48]. A clinically mild maternal course does not prove that every fetal endpoint has zero risk; neither does a fatal pathology series imply that most mild infections damage the placenta.

4.5. Vaccination, Viral Variants, and Pandemic Period

Variant and vaccination periods further limit direct comparison. UK surveillance found more severe maternal disease during Alpha and Delta dominance than during wild-type circulation [50]. Quebec data found substantially greater severe maternal morbidity during earlier variant periods than during Omicron [40]. Vaccination was associated with a lower stillbirth risk in CRONOS, and the Scottish matched cohort found no evidence of increased adverse outcomes after vaccination [28,41]. A study that combines unvaccinated wild-type infections with vaccinated Omicron infections may therefore produce an attenuated association unrelated to gestational biology.

National surveillance also identifies social and demographic differences in hospitalisation or infection burden [15,17]. Access to antenatal assessment, recognition of deterioration, and baseline medical vulnerability should consequently be considered when applying results across settings. Observational vaccination findings inform prevention and counselling, while specific vaccination schedules should follow current applicable recommendations [28,41].

Variant period is often assigned from calendar dates rather than sequencing of every infection. Vaccination, prior infection, testing access, clinical treatments, and admission policies can change during the same period. Consequently, a lower adverse-outcome rate in an Omicron-era cohort should not be attributed to viral biology alone. Comparisons within similar calendar periods and with explicit vaccination histories are more informative than unstratified comparisons across the entire pandemic [28,40,41,50].

5. Clinical Assessment and Follow-Up

5.1. Maternal Assessment and Competing Diagnoses

Maternal assessment should consider the respiratory course, baseline medical vulnerability, hypertensive or thrombotic findings, and gestational age together [14,39,40]. The comparative studies associate acute severity with critical-care needs and serious maternal morbidity, making current clinical status more informative for immediate assessment than the infection date alone. They do not, however, establish a trimester-specific laboratory threshold or treatment algorithm.

Follow-up should reflect the maternal clinical course and the obstetric findings rather than a trimester label alone. Reassuring common outcomes after mild first-trimester infection and the Nordic registry findings for major congenital anomalies support proportionate counselling, while rare outcomes and conflicting prospective findings remain uncertain [4,5,25]. These observational studies do not establish a universal enhanced-surveillance schedule, a specific ultrasound frequency, or a threshold for delivery after recovery. Individual assessment should incorporate symptoms, comorbidity, existing obstetric risk, and applicable local protocols.

When maternal deterioration is atypical, reassessment should distinguish respiratory illness, hypertensive or thrombotic complications, and documented secondary infection. Recording the reason for admission and the need for respiratory support separately helps distinguish infection severity from obstetric indications and coincident disease [14,39,40].

5.2. Fetal Assessment and Timing of Anomaly Detection

Antenatal follow-up must distinguish fetal growth, structural anatomy, placental function, and acute fetal well-being. Reassuring fetal biometry after infection does not exclude every placental or congenital endpoint, and a small-for-gestational-age newborn is not equivalent to antenatally confirmed fetal growth restriction [10,30,31]. Similarly, a structural anomaly first recognised after maternal infection may have developed earlier; chronological proximity alone does not establish a causal association.

Comparative studies should record ultrasound access and timing, fetal losses, terminations for anomaly, and postnatal diagnostic follow-up consistently. Differences in detection can otherwise be mistaken for differences in biological risk [4,5]. Pandemic-related changes in prenatal-service use are considered in Section 7.2.

5.3. Delivery Decisions and Neonatal Follow-Up

Patients infected later in pregnancy require assessment for respiratory deterioration, hypertensive disease, thrombosis, fetal status, and preterm delivery risk. The evidence does not support caesarean delivery solely because of infection. A systematic review concluded that high caesarean rates did not establish a clinical indication for elective caesarean birth [51]. Delivery decisions should therefore be based on maternal condition, fetal status, obstetric indications, and gestational age.

The distinction between spontaneous and indicated delivery in Section 3.3 should inform interpretation of individual cases; the reason for delivery and the infection-to-birth interval are essential [9,34,40].

For newborns, reporting should distinguish admission for prematurity or respiratory illness from admission for observation or infection-control reasons. Where testing is positive, sampling time and confirmatory evidence are needed before classifying congenital transmission [17,22,34,43]. The available six-month follow-up data are useful for short-term growth and development, but do not resolve later neurodevelopmental outcomes or outcomes after early maternal exposure [47].

6. Maternal Mental Health and Risk Communication

In a single-centre Romanian cohort of 45 women with mild or moderate SARS-CoV-2 infection before 24 weeks, four had Generalized Anxiety Disorder seven-item scale (GAD-7) scores of 10–14 and three had a score of 15 [52]. These descriptive findings should be interpreted cautiously, as the absence of an uninfected comparator, preinfection assessment, and longitudinal follow-up precludes attributing anxiety symptoms to SARS-CoV-2 infection or establishing a gestational timing effect. Moreover, screening scores do not constitute psychiatric diagnoses. Clinical communication should acknowledge these limitations, address maternal concerns, and facilitate appropriate perinatal mental health assessment when clinically indicated.

7. Evidence Limitations and Research Priorities

7.1. Exposure Measurement and Causal Interpretation

Testing strategy also changes the apparent effect of timing. Universal screening detects asymptomatic infection, while hospital-based surveillance preferentially detects symptomatic women. The Generation C cohort found no difference in selected outcomes between seropositive and seronegative women among live births, but seropositivity represented earlier infection and excluded women positive by reverse transcription polymerase chain reaction (RT-PCR) at delivery from the principal comparison [53]. In contrast, UK and WAPM studies recruited many women through hospital care and reported higher rates of intensive care, caesarean delivery, prematurity, and neonatal admission [17,35].

Gestational-age-dependent time at risk is especially important for PTB and stillbirth comparisons. To enter a late-infection group, a pregnancy must remain ongoing until that exposure; earlier deliveries or losses cannot enter that group. Treating infection as fixed from conception may introduce immortal-time or time-dependent exposure bias. Infection after an outcome threshold cannot cause an earlier event, and testing during an outcome-related admission can create ascertainment bias. Gestational-week risk sets, time-varying exposure models, and explicit intervals from infection to outcome are preferable to crude trimester contrasts. These considerations complement the clinical distinction between spontaneous and indicated PTB in Section 3.3; they do not imply that every published timing association is biased in the same direction.

Retrospective designs, incomplete testing of asymptomatic women, heterogeneous controls, and changing vaccination and variant contexts limit inference. Broad categories may conceal differences between infection at 21 and 27 weeks or at 32 and 39 weeks. PCR, serology, and hospital diagnosis also identify different exposure windows. Register studies may misclassify timing or severity, while small prospective cohorts lack power for uncommon outcomes. The principal modifiers are discussed in Section 4.4 and Section 4.5 rather than treated as interchangeable explanations for every result.

Comparator choice determines the question being answered. Comparing infected with uninfected pregnancies addresses a different association from comparing early with late infection among infected women. An association detected in one gestational subgroup and not another is not, by itself, evidence that the subgroup effects differ; that claim requires a direct interaction or between-group comparison. Estimates from cohorts with different baseline risks should also be accompanied by their absolute outcome frequencies [10,14,27,28].

7.2. Prenatal Healthcare Access and Outcome Ascertainment

Healthcare changes during the pandemic may alter recorded outcomes independently of maternal infection. A nationwide Polish cross-sectional study found 126,552 participants in the publicly funded prenatal screening programme in 2019 versus 110,844 in 2020, a reported 12.41% reduction [54]. The mix of screening and diagnostic procedures also changed, including fewer amniocenteses. This concerns one national programme, whose eligibility, coverage, and annual number of births affect interpretation; it cannot quantify an infection-attributable fetal risk or represent all healthcare systems. Changes in attendance, ultrasound timing, and diagnostic access may delay or reduce detection of fetal anomalies and high-risk pregnancies. Apparent outcome differences may therefore reflect ascertainment and care pathways as well as biological consequences of infection.

7.3. Limitations of This Narrative Review

This narrative review was based on a focused PubMed/MEDLINE search, English-language publications, and purposive study selection. Although direct timing-specific comparative evidence was prioritised, the use of a single bibliographic database and the narrative approach may have limited the completeness of the available evidence. No formal risk-of-bias assessment or standardised certainty grading was performed. Publication, selection, and language biases cannot be excluded.

The tables and text draw on potentially overlapping cohorts and reviews, not independent sets of replications. Broad meta-analytic estimates cannot establish a trimester-specific effect. The maternal anxiety cohort, prenatal-service study, and selected placental series address different questions and are not pooled with comparative timing cohorts. Diagnostic certainty also varies: histological injury, placental viral detection, neonatal positivity, and congenital infection require distinct interpretation. A mechanism demonstrated in selected perinatal deaths does not establish its population frequency.

7.4. Priorities for Future Research

Early gestation is underrepresented. The early-pregnancy meta-analysis found substantial heterogeneity and could pool only miscarriage [25]. Congenital anomalies, fetal neurodevelopment, placental pathology, and childhood health require prospective cohorts with standardised ultrasound, ascertainment of losses and terminations, and longer child follow-up. Six-month anthropometric findings after late infection cannot resolve later neurodevelopmental outcomes [47].

A useful minimum reporting set would connect the gestational age at infection with the interval to each outcome; distinguish diagnosis date from symptom onset; specify previous infection and vaccination; describe how severity and variant period were determined; and record the indication for delivery. Pregnancy losses and terminations should be included where ascertainment permits. Prespecified analyses of time-dependent exposure, missing data, and between-group differences would reduce the risk that testing patterns or unequal observation periods are interpreted as biological trimester effects.

Linked placental examination and maternal–infant follow-up would help distinguish inflammation, placental infection, congenital transmission, prematurity, and longer-term child outcomes. Psychological follow-up should use comparable uninfected groups and repeated measures beginning before or early in infection. These approaches would test mechanisms suggested by current studies while retaining outcomes that matter to both mother and child [5,25,43,47,48,52].

8. Conclusions

Our synthesis indicates that gestational timing is associated with distinct patterns of pregnancy outcomes after maternal SARS-CoV-2 infection, without establishing an independent timing effect or a consistent risk gradient across all endpoints. Mild early infection has not consistently been associated with excess pregnancy loss or impaired fetal growth. Large registry studies provide reassurance regarding major congenital anomalies among live births, although conflicting findings and incomplete ascertainment of pregnancies ending before birth limit certainty. Stillbirth associations in broader early-exposure groups must retain the original 0–20-week or <28-week definitions; they do not establish a first-trimester-specific effect. Published hypertension studies do not establish a highest-risk trimester for preeclampsia.

Late pregnancy and peripartum infection have the most consistent association with preterm birth and acute maternal and neonatal morbidity. Maternal respiratory severity, provider-initiated delivery, and prematurity account for part of this pattern. Placental injury represents an additional pathway to fetal compromise and may occur without severe maternal respiratory illness or confirmed fetal infection. Congenital transmission appears uncommon, and neonatal test positivity alone is insufficient to establish intrauterine infection.

Gestational age at infection should therefore be interpreted together with the maternal clinical course, baseline obstetric risk, vaccination status, circulating variants, and the interval between infection and delivery. Proportionate follow-up and clear counselling should address both physical outcomes and maternal anxiety, while delivery decisions should remain based on maternal condition, fetal status, and obstetric indications. Heterogeneous timing definitions, incomplete detection of mild infections, and limited data on rare early-pregnancy and long-term child outcomes constrain the conclusions of this narrative review. Prospective studies with standardised exposure definitions and complete maternal–infant follow-up are needed to clarify these remaining uncertainties.