Long COVID after COVID-19 infection during pregnancy in a large Italian hospital

SARS‐CoV‐2 infection has been associated with some unique risks during pregnancy. Pregnant women are more likely to experience severe illness compared to the nonpregnant population and may require more intensive life‐supporting measures, admission to the intensive care unit (ICU), mechanical ventilation, or ECMO (extracorporeal membrane oxygenation).1 Moreover, SARS‐CoV‐2 infection during pregnancy has also been linked to several pregnancy‐specific complications, like preterm birth and stillbirth.2, 3 This increased vulnerability has been hypothesized to be related, among other changes, to the pregnancy‐specific immune modifications, such as a shift from Th1 to Th2 immune responses and reduced Natural Killer (NK) cell activity, that may alter viral clearance mechanisms.4, 5 The postacute sequelae of SARS‐CoV‐2 (PASC), or Long COVID, are now a recognized consequence of the pandemic, affecting potentially 10–30% of infected individuals. This syndrome involves symptoms that might not be present before acute illness and that linger or appear a few months after what should have been a resolved case of COVID‐19.6 With its status as a newly recognized condition, the World Health Organization has defined Long COVID as symptoms occurring within 3 months of SARS‐CoV‐2 infection, persisting for at least 2 months, and not explained by alternative diagnoses.7 Common manifestations of Long COVID include fatigue, shortness of breath, and cognitive dysfunction.8 These symptoms appear to affect women approximately 13% more often than men.6 In the pregnant population, prevalence rates range widely from as low as 9.3% up to 74.7%. This high variability is probably partly dependent on the definition of Long COVID the authors employed as well as different inclusion criteria and information‐gathering protocols3, 9, 10

COVID‐19 vaccination during pregnancy has demonstrated protective effects against adverse outcomes.11, 12 Recent meta‐analyses show that vaccination reduces the risk of SARS‐CoV‐2 infection (by ~60%), hospitalization (~53%), and ICU admission (~82%) in pregnant women.13 Additionally, vaccination has been linked to a reduced risk of preterm birth and stillbirth1

Nonetheless, while systematic reviews suggest vaccination decreases Long COVID risk by about 24% in general populations,14 data specific to pregnant women remain scarce. Our study aims to evaluate the development of Long COVID in pregnant women following SARS‐CoV‐2 infection, characterizing the symptom profile and identifying potential risk or protective factors within this specific population.

This observational study combined retrospective data collection from medical records with prospective symptom assessment via questionnaires. It was conducted at the Obstetrics and Gynecology Unit of ASST‐Spedali Civili of Brescia (Italy), a tertiary care center that served as a regional reference hub for COVID‐19 pregnancies from March 2020 to May 2022. Informed consent was obtained from all participants prior to their inclusion in the study.

The study included pregnant women aged ≥18 years with confirmed SARS‐CoV‐2 infection via reverse transcription polymerase chain reaction (RT‐PCR) or antigen testing, either during pregnancy at any gestational age or upon admission for labor. Long COVID was defined as persistence of symptoms or emergence of new ones for more than 12 weeks following acute infection. Data were primarily extracted from electronic medical records using institutional hospital digital systems. Information collected included demographic data (age, body mass index (BMI), country of origin), maternal comorbidities (obesity, diabetes, hypertension, cardiovascular problems, history of respiratory disease, autoimmune conditions), COVID‐19 infection details (date of admission, severity, hospitalization duration, ICU admission, pneumonia diagnosis), pregnancy‐related data (gestational age at infection, obstetric complications, mode of delivery and pregnancy outcome).

A standardized questionnaire detailed in Tablewas specifically developed for this study and administered to eligible patients through telephone interviews. We asked about symptoms characteristics, duration, severity, and effect on quality of life, other than vaccination status at the moment of admission. The list of symptoms was developed based on the most frequently described manifestations in the literature on post‐acute COVID‐19 sequelae, as reported in major clinical studies, while at the same time allowing participants to specify which symptoms they experienced. This approach allowed participants to report additional manifestations beyond the predefined list. Based on questionnaire responses, participants were divided into two groups: those who experienced only acute symptoms and those who developed long‐term sequelae persisting beyond 12 weeks. S1

A total of 510 women were identified. Of these, 133 could not be reached and 29 declined to participate. In the end, 348 pregnant women, hospitalized with COVID‐19 across all levels of severity, were included in the study (68.2%): 103 (30%) were vaccinated and 245 (70%) were unvaccinated. Among them, 94 (27.0%) developed Long COVID: the most frequently reported persistent symptoms were chronic fatigue (46.8%), memory impairment (38.3%), anxiety disorders (20.4%), attention deficit, dyspnea, and myalgia (each 13.8%) (Figure 1). The mean number of symptoms reported per patient was 1.9 (range 1–7), with 63.8% of patients reporting multiple persistent symptoms. Symptom patterns differed markedly by vaccination status (Figure 2). Among vaccinated patients with Long COVID (n = 16), chronic fatigue was most common (75.0%), followed by dyspnea (31.2%) and arthralgia (12.5%). In contrast, unvaccinated patients (n = 78) exhibited a broader symptom profile, with memory impairment (44.9%), chronic fatigue (41.0%), and anxiety disorders (23.4%) being most common.

Baseline characteristics of the study population stratified by Long COVID status are reported in Table 1 as well as univariate logistic regression analysis. Compared to those without Long COVID (n = 254), affected women had significantly longer hospital stays (mean 7.6 ± 7.5 vs. 4.6 ± 3.7 days, p = 0.002), higher rates of ICU admission (6.4% vs. 1.2%, p = 0.025), were more likely to have radiologically confirmed pneumonia (35.1% vs. 8.3%, p < 0.001) and a history of pre‐existing respiratory disease (9.6% vs. 2.4%, p = 0.013). Notably, patients who developed Long COVID were infected at an earlier gestational age compared to those who did not (mean 33.3 ± 8.4 vs. 36.2 ± 6.4 weeks, p = 0.001). Vaccination rates were also markedly lower in the Long COVID group (17.0% vs. 34.3%, p = 0.002). No significant differences were observed between groups regarding other comorbidities, including diabetes, chronic hypertension, cardiac disease, and autoimmune or endocrine disorders. Women with respiratory disease had significantly higher rates of Long COVID compared to those without (60.0% vs. 25.5%, p = 0.013).

In the multivariable logistic regression model, after adjustment for potential confounders, including age, BMI, pneumonia at admission, ICU admission, cardiovascular risk factors, gestational age at admission, diabetes, and vaccination status, preexisting respiratory disease remained independently associated with Long COVID development (adjusted OR = 3.17, 95% CI 1.21–12.39, p = 0.045; E‐value = 5.8).

Pneumonia at admission was the strongest predictor of Long COVID in our cohort. Women diagnosed with pneumonia at admission had significantly higher rates of Long COVID compared to those without pneumonia (61.9% vs. 21.2%, p < 0.001). After multivariable analysis, pneumonia at admission remained strongly associated with Long COVID development (adjusted OR = 4.481, 95% CI: 2.16–9.28, p < 0.001; E‐value 9.51).

Gestational age at SARS‐CoV‐2 infection emerged as a significant independent predictor of Long COVID development: each additional week of pregnancy at the time of infection was associated with a 3.9% reduction in the odds of developing Long COVID (adjusted OR = 0.961 per week, 95% CI: 0.928–0.995, p = 0.027). This inverse relationship remained significant after adjusting for maternal age, BMI, preexisting respiratory disease, pneumonia at admission, vaccination status, diabetes, and cardiovascular risk factors.

Vaccination status significantly influenced Long COVID risk. In unadjusted analysis, 15.5% of vaccinated patients developed Long COVID versus 31.8% of unvaccinated patients (p = 0.002; risk ratio = 0.394). After adjusting for relevant covariates, including disease severity indicators, the association was attenuated and no longer statistically significant (adjusted OR = 0.60, 95% CI: 0.31–1.16, p = 0.13), although the direction of effect remained consistent.

To better estimate the overall impact of vaccination, including indirect effects via reduced disease severity, an inverse probability of treatment weighting (IPTW) analysis was conducted on a sub‐cohort of 326 women (89 vaccinated, 237 unvaccinated). This sample excluded women for whom the infection was diagnosed at the time of admission for delivery, for whom gestational age at infection was not applicable for propensity score modeling. The estimated propensity scores in this sample ranged from 0.101 to 0.508, and the trimmed stabilized weights ranged from 0.694 to 1.703. Baseline characteristics of this analytical sample (N = 326) before and after IPTW weighting are presented in Table 2. After applying the trimmed stabilized IPTW, we achieved a balance across all measured covariates included in the propensity score model, with postweighting SMDs below 0.1 (e.g., for age, postweighting SMD = 0.011; for BMI, postweighting SMD = 0.041).

The results from the IPTW‐weighted logistic regression analyses for Long COVID, performed on the N = 326 cohort, are shown in Table 3. This approach yielded a weighted odds ratio of 0.38 (95% CI: 0.20–0.71, p = 0.003), indicating a statistically significant 62% reduction in the odds of Long COVID among vaccinated women. The corresponding E‐value of 4.70 (2.17 for the lower CI bound) indicated that a relatively strong unmeasured confounder would be needed to nullify this effect, supporting a potentially causal relationship between vaccination and reduced Long COVID risk.

Regarding specific symptom differences by vaccination status in the overall cohort (N = 348): memory impairment was significantly less prevalent in vaccinated compared to unvaccinated patients (6.2% vs. 44.9%, p = 0.004). Similarly, attention deficit (0% vs. 16.7%), infectious diathesis (0% vs. 12.8%), anosmia (0% vs. 5.1%), and ageusia (0% vs. 9.0%) were entirely absent in vaccinated patients but present in varying proportions of unvaccinated patients.

Our study in women hospitalized with COVID‐19 at the regional hub of one of the most affected Italian cities by COVID‐19 found that approximately one‐quarter (27.0%) experienced Long COVID. Our analysis points to several key factors increasing this risk: underlying respiratory illness before pregnancy, developing pneumonia during the acute infection, and getting infected earlier in gestation all independently predicted Long COVID. Conversely, COVID‐19 vaccination appeared to be a protective factor, associated with both a lower overall chance of developing Long COVID and a different pattern of symptoms.

Comparing our findings directly to other recent studies can be difficult. The landscape of Long COVID research in pregnancy is characterized by different patient groups, definitions, and ways of collecting data. Our 27.0% rate falls within the broad range (9.3% to over 90%) found across different studies.19 It is substantially higher than the 9.3% seen in the large US RECOVER‐Pregnancy Cohort,3 a group that was not exclusively hospitalized and was mostly infected later in the pandemic (Omicron wave). The incidence of Long COVID in our cohort seems to be higher also than rates derived from massive US electronic health record databases using computational phenotypes, ranging from 4.4% to 16.5%.20 This difference likely stems from our focus on hospitalized patients—a group inherently likely to have had more severe acute illness, which is itself a major driver of Long COVID risk.3, 9, 10 Our rate does align more closely with the 34.2% reported by a Spanish study that included outpatients but used a similar ≥3‐month definition.9 It remains much lower, though, than the 74.7% reported in a single‐center Turkish study which employed a broader >4‐week persistence definition.10 This sheer variability probably reflects the different study design choices: inclusion criteria (hospitalized vs. all pregnant individuals), Long COVID definition (different duration), and how information was gathered (patient interviews vs. record mining). The period of investigation also seems to be relevant, with Omicron having reported substantially lower absolute incidences.21

A consistent finding across our study and others is the strong association between acute COVID‐19 severity and subsequent Long COVID risk. Our identification of pneumonia at admission (aOR 4.48) and ICU admission (significant in univariate analysis) aligns well with findings associating oxygen requirement,3 hospitalization,9, 10 or moderate/severe acute disease with increased Long COVID risk.9 Our finding that prior respiratory disease might also independently raise Long COVID risk (aOR 3.17) further highlights underlying vulnerability. Other studies based on the general population pointed out that preexisting asthma and COPD were associated with increased odds of developing Long COVID by 41% and 32%, respectively.22 Perhaps altered local immune function, reduced viral clearance capacity, or simply less physiological buffer make persistent symptoms more likely or noticeable. This is consistent with observations in non‐pregnant groups and aligns with the broader comorbidity risks observed in pregnancy cohorts.3, 20

We found that increasing gestational age has a 4% reduced odds per week of developing Long COVID. Zang et al. (2025) found higher risks with first/second‐trimester infection, while Muñoz‐Chápuli Gutiérrez et al. (2024) with first and third trimester infection.9, 20 Our result, coming from hospitalized patients, could imply that needing hospitalization due to COVID‐19 early in pregnancy might signify greater maternal susceptibility or a different interaction with the developing immune system at that stage, potentially hindering full recovery. Improving maternal antibody function may also have a role later in gestation.23 However, some symptoms could be confounded by overlapping pregnancy physiology (e.g., fatigue/dyspnea from metabolic demands). Mitigating this, assessments occurred postdelivery (mean 12–18 months post‐infection). More work is clearly needed to dissect how gestational timing, immune changes, viral factors, and Long COVID truly intersect.

Our data show that COVID‐19 vaccination offers protection against Long COVID in our cohort. While the effect was attenuated after adjusting for acute severity markers—suggesting preventing severe illness is a key benefit—the strong protective signal seen in unadjusted and IPTW analyses (weighted aOR 0.38) supports a meaningful overall effect, with an approximately 62% lower odds of Long COVID in vaccinated individuals. This finding aligns with other research9 and systematic reviews, identifying nonvaccination as a risk factor and suggesting vaccination helps reduce Long COVID.9, 13, 14, 19 The absence of a significant adjusted association in the RECOVER cohort could arise from differences in study populations (predominantly Omicron era), statistical power, or adjustment strategies.24

Intriguingly, our study also hints that vaccination might change how Long COVID presents during pregnancy. We found a marked reduction in neuropsychiatric symptoms, especially memory impairment, among vaccinated women compared to unvaccinated women (6.2% vs. 44.9%). Although fatigue remained frequent in both groups, this apparent shift away from cognitive issues in vaccinated individuals mirrors some observations in general populations.25

Our results identify specific high‐risk groups—particularly pregnant women with respiratory comorbidities or severe acute disease—who are at higher risk of Long COVID development.

These risk factors—respiratory comorbidities (aOR = 3.17), pneumonia (aOR = 4.48), and nonvaccination (weighted OR = 0.38)—mirror those in nonpregnant populations, where meta‐analyses show that severe acute disease and preexisting conditions (like asthma, OR = 2.15) are significant risk factors.6 This implies shared mechanisms, such as endothelial inflammation and viral persistence.26 However, pregnancy introduces unique nuances not captured in general population studies. Most notably, gestational timing emerged as an independent predictor, which may stem from trimester‐specific immune adaptations amplifying vulnerability early in gestation. Vaccination also modified symptoms (e.g., reduced neuropsychiatric issues), possibly via pregnancy‐specific immune interactions.

However, the landscape of SARS‐CoV‐2 continues to evolve. Recent surveillance data from CDC shows that the LP.8.1 variant has been increasing rapidly in 2025.17 Our study did not assess how the follow‐up care might play into the next stage of the pandemic nor does it give advice about when and how pregnant women should get vaccinated. Thus, the implications of variant shift for vaccine effectiveness against Long COVID remain unclear, underscoring the need for continued monitoring and updated research.

There are some key limitations to our study that the reader should consider when interpreting our findings. First, the study execution within a single tertiary care center potentially limits the direct generalizability of our results to broader community settings or healthcare systems with different patient demographics or resources. Furthermore, our study lacked a concurrently recruited nonpregnant control group. This restricts our ability to isolate aspects of Long COVID solely attributable to the pregnant state versus postviral sequelae in general. We acknowledge that the overlap between some Long COVID symptoms and common pregnancy‐related complaints represents a limitation of our study. Although most women were interviewed several months after delivery and clearly described these manifestations as new or persistent sequelae following acute infection—rather than typical pregnancy‐related conditions—we recognize that a certain degree of symptom overlap cannot be completely ruled out, as also noted in previous research on Long COVID during pregnancy.

There are also several potential limitations related to our methods. Our reliance on retrospective chart review combined with prospective telephone interviews carries an inherent risk of recall bias concerning the precise timing, nature, and duration of symptoms, as patient recollection can change over time. We also acknowledge a form of selection bias as our cohort focused exclusively on hospitalized individuals—likely experiencing more severe acute COVID‐19 than the general pregnant population. Consequently, the observed 27.0% prevalence of Long COVID might overestimate the burden across all pregnant individuals infected with SARS‐CoV‐2, including those with milder outpatient cases.

Finally, information on potentially significant confounding variables, such as detailed socioeconomic indicators, specific patterns of healthcare utilization, or occupational exposures, was not uniformly available. Similarly, while data on country of origin were collected, we lacked consistent details on length of residence or citizenship status. This heterogeneity prevented a reliable categorization of “immigrant status,” a variable known to correlate with vaccination uptake.

While statistical approaches like IPTW and E‐value calculations were employed to mitigate confounding, the possibility of residual bias from unmeasured factors persists. We also faced limitations in consistently capturing precise details about vaccination, including exact timing relative to infection or gestation, or specific vaccine types received, hindering analyses of optimal vaccine strategies. Similarly, the lack of individual‐level viral variant data restricts the applicability of our findings, collected across multiple pandemic waves, to the current or future variant landscape.

Additionally, calendar time represents a significant potential confounder unaccounted for in our models. Vaccination was unavailable pre‐2021, and the study spans multiple pandemic waves with variant shifts (e.g., Wild‐type to Omicron), which may differ in Long COVID propensity.27 We also lacked data on prior infections, limiting adjustment for hybrid immunity. Restricting analyses to vaccine‐available eras would underpower the study; thus, the vaccination effect (IPTW OR = 0.38) should be interpreted cautiously, as residual time‐dependent bias may persist despite IPTW and E‐value assessments.

Our study found that approximately one in four pregnant women with COVID‐19 developed Long COVID syndrome. Preexisting respiratory disease and severe illness during acute infection emerged as strong risk factors for Long COVID development. Moreover, we identified a significant temporal relationship between infection timing, early in pregnancy, and Long COVID risk. Vaccination was associated with a significantly lower risk of Long COVID and a modified symptom profile.

These findings, supported by emerging evidence from recent systematic reviews and meta‐analyses, highlight the importance of preventive strategies and careful monitoring of high‐risk pregnant women following COVID‐19 infection.

Future research should address several key questions: the interaction between vaccination, disease severity, and Long COVID development; variant‐specific effectiveness; value of preexisting comorbidities that can alter immune response. Prospective studies with larger cohorts and longer follow‐up are needed to further understand the natural history, pathophysiology, and optimal management of Long COVID in the obstetric population. Given the evolving nature of SARS‐CoV‐2, continued surveillance and updated analyses will be essential to inform clinical practice and public health policies.

The clinical study was designed, conceptualized, and supervised by Anna Fichera. Eleonora Biancareddu collected data and wrote the original manuscript draft. Marco Bozzo conducted data analysis and created visualizations. Mirabella Ezenwa and Emma Paola Ongarini collected and analyzed data. Anna Fichera and Federico Prefumo were responsible for the interpretation of the data. All authors reviewed, edited, and approved the final version.

The authors declare no conflicts of interest.

This study has obtained ethical approval from the Regional Ethical Review Authority (Comitato Etico Lombardia 6; NP 6458) on January 14, 2025. The study was conducted according to principles of the declaration of Helsinki. Informed consent was obtained from all participants prior to their inclusion in the study.

We gratefully acknowledge the contribution of Roberta Castellani, Dafne Corbelli, Irene Di Stasio, Angelica Fiorini, Marina Momi, Rossella Monaci, Marta Papaccio, and Chiara Ventriglia who performed the initial data collection on patient characteristics.

The data that support the findings of this study are available from the corresponding author upon reasonable request.