This prospective study enrolled 150 pregnant women diagnosed with PROM according to the 2015 Guidelines for Diagnosis and Management of PROM [18] from Chongqing University Central Hospital and the People’s Hospital of Yubei District between February 2023 and February 2024. Initial screening identified 132 eligible participants meeting all inclusion criteria: (1) age $\ge$18 years; (2) singleton pregnancy; (3) regular prenatal care without significant trauma or recent sexual intercourse; and (4) complete clinical documentation. Subsequently, we excluded 12 participants based on predefined exclusion criteria: (1) recent infections (respiratory/urinary tract) or antibiotic use within two weeks; (2) pregnancy complications (e.g., hypertension or diabetes); (3) severe systemic disorders (affecting liver, kidney, or heart function) or malignancy history; and (4) inability to comply with study protocols. The final study population comprised 120 participants. The procedure for participant recruitment is summarized in the flow chart of Fig. 1.

Fig. 1.

The study flow chart. PROM, premature rupture of membranes.

Participants were stratified based on the degree of neutrophil infiltration in placental membrane tissues under high magnification into non-infection, mild infection, moderate infection, and severe infection categories. The Ethics Committee of Chongqing University Central Hospital approved the study protocol (2023-8). All participants were fully briefed on the study’s objectives and procedures, from which informed consent was voluntarily obtained.

Participant data collected included age, gestational age, length of hospital stay, parity, delivery mode, sex of the fetus, and newborn weight. On admission, 3 mL of venous blood was drawn, stored in dry vacuum coagulation tubes, and centrifuged at 3000 rpm for 15 minutes. The serum was then transferred to Eppendorf tubes and stored at –80 °C. Placental and membrane tissues were collected within 10 minutes post-delivery from near the rupture site for pathological analysis, fixed in formaldehyde, and processed with hematoxylin and eosin staining.

Serum $\beta$-hCG levels were measured using the electrochemiluminescence method on a Roche Cobas e411 analyzer (Roche Diagnostics GmbH, Mannheim, Germany), and SF levels were assessed with the chemiluminescence method on a DiaSorin LIAISON XL analyzer 2210 (DiaSorin S.p.A., Saluggia, Italy), following all manufacturer instructions for calibration and quality control.

The placental membranes collected postpartum were fixed in formalin, processed into pathological sections, and stained with hematoxylin and eosin for microscopic examination. The severity was determined by examining placental membrane tissue slices for neutrophil infiltration per high-power field: none (5 neutrophils), mild (5–10 neutrophils), moderate (11–30 neutrophils), and severe ($>$30 neutrophils) [19].

Participants were categorized into non-infected and infected groups, and then further subdivided by infection severity. Data are presented as mean $\pm$ SD or median values (25th–75th percentile) as appropriate. Group comparisons were made using one-way ANOVA for continuous variables. Variables with skewed distributions, like serum $\beta$-hCG and ferritin levels, were logarithmically transformed before analysis.

Spearman’s rank correlation analysis was conducted to evaluate the association between risk factors and intrauterine infection. Logistic regression analyses were performed to identify the risk factors for intrauterine infection. Two logistic regression models were developed: Model 1 adjusted for age, hospital stay, white blood cells (WBCs), neutrophils, and hemoglobin; Model 2 adjusted for gestational age and log-transformed serum levels. Multinomial logistic regression was used to assess the predictive value of serum markers and gestational age.

The C-statistics were calculated to measure the concordance between model-based risk estimates and intrauterine infection in PROM. Net reclassification improvement (NRI) and integrated discrimination improvement (IDI) were determined to measure the incremental prognostic effect of adding $\beta$-hCG, serum ferritin, and gestational age to conventional risk factors. The basic prediction model included conventional risk factors for intrauterine infection in PROM, such as age, length of hospital stay, white blood cell count, neutrophil count, hemoglobin, and newborn weight.

The gold standard for diagnosis was pathological examination. The efficacy of combined serum $\beta$-hCG, SF levels, and gestational age as predictors of intrauterine infection in PROM was analyzed using receiver operating characteristic (ROC) curves, evaluating the area under the curve (AUC).

All statistical analyses were conducted by using SAS version 9.4 (SAS Institute Inc, Cary, NC, USA), or SPSS version 22.0 (IBM Corp., Armonk, NY, USA), with significance set at a two-sided p-value of 0.05.

The study included 120 patients with PROM, among whom 32 (26.67%) were diagnosed with intrauterine infection. The infected patients exhibited significantly higher median levels of serum $\beta$-hCG and SF compared to non-infected patients (serum $\beta$-hCG: 43,104.00 vs. 22,375.00 mIU/mL, p 0.0001; SF: 34.14 vs. 27.81 ng/mL, p = 0.0020). Additionally, gestational age was significantly shorter in the infected group (38.63 vs. 37.78 weeks, p = 0.0040) (Table 1). Table 2 presents the basic characteristics of patients stratified by intrauterine infection levels in cases of PROM. Significant trends were observed with shorter gestational age, increasing levels of SF and $\beta$-hCG related to the severity of infection (p values were 0.01, 0.001, and 0.0001, respectively).

In the correlation analysis (Table 3), log₁₀-$\beta$-hCG and log₁₀-SF demonstrated significant positive correlations with intrauterine infection (correlation coefficients of 0.39, p 0.0001, and 0.23, p = 0.0100, respectively). In contrast, gestational age showed a negative correlation with intrauterine infection (correlation coefficient of –0.26, p = 0.0020). Similarly, when analyzing the severity of infections, log₁₀-$\beta$-hCG (correlation coefficient of 0.41, p 0.0001), log₁₀-SF (correlation coefficient of 0.26, p = 0.0030), and gestational age (correlation coefficient of –0.29, p = 0.0010) remained significant predictors.

For the logistic regression analysis (Table 4), elevated log₁₀-$\beta$-hCG and log₁₀-SF levels, as well as lower gestational age, were significantly associated with a higher risk of intrauterine infection. In Model 1, after adjusting for age, hospital stay, WBC count, neutrophil ratio, hemoglobin levels, and newborn weight, the odds ratio (OR) for log₁₀-$\beta$-hCG was 31.42 (95% confidence interval [CI]: 5.11–193.23, p = 0.0002), for log₁₀-SF was 13.45 (95% CI: 2.79–64.72, p = 0.0010), and for gestational age was 0.54 (95% CI: 0.36–0.81, p = 0.0020). In Model 2, after further adjustment for gestational age, log₁₀-$\beta$-hCG, and log₁₀-SF, the associations remained significant, with ORs of 22.41 (95% CI: 3.39–148.19, p = 0.0010) for log₁₀-$\beta$-hCG, 6.45 (95% CI: 1.10–37.81, p = 0.0300) for log₁₀-SF, and 0.61 (95% CI: 0.39–0.97, p = 0.0300) for gestational age.

When analyzing infection severity (non-infected, mildly, moderately, severely), the risk increased significantly with higher log₁₀-$\beta$-hCG and log₁₀-SF levels, and shorter gestational age. In Model 1, log₁₀-$\beta$-hCG had an OR of 16.88 (95% CI: 4.47–63.53, p 0.0001), while log₁₀-SF had an OR of 10.74 (95% CI: 3.49–33.05, p 0.0001), and gestational age had an OR of 0.59 (95% CI: 0.43–0.81, p = 0.0010). After full adjustment in Model 2, the ORs were 13.56 (95% CI: 3.51–52.92, p = 0.0001) for log₁₀-$\beta$-hCG, 5.47 (95% CI: 1.70–17.52, p = 0.0040) for log₁₀-SF, and 0.71 (95% CI: 0.52–0.97, p = 0.0300) for gestational age (Table 4).

Fig. 2 presents the receiver operating characteristic (ROC) curves, illustrating the diagnostic performance of these biomarkers. The combined approach using log₁₀-$\beta$-hCG, log₁₀-SF, and gestational age significantly enhanced diagnostic accuracy, as evidenced by the highest area under the curve (AUC) of 0.78 (Table 5). These findings underscore the value of integrating multiple biomarkers to improve the early diagnosis and management of intrauterine infection in PROM patients.

Fig. 2.

The receiver operating characteristic analysis for the risk of intrauterine infection. This figure displays the ROC curve, plotted using sensitivity (true positive rate) against 1-specificity (false positive rate) based on the data from the predictive model. The AUC value included in the figure quantifies the model’s accuracy in distinguishing cases with and without the risk of intrauterine infection, reflecting its diagnostic ability. AUC, area under the curve; ROC, receiver operating characteristic.

Next, we compared the predictive performance of $\beta$-hCG, SF, and gestational age, alongside basic risk factors, in forecasting incident intrauterine infection in patients with PROM. Incorporating $\beta$-hCG, SF, and gestational age into the conventional risk models significantly improved C-statistics, IDI, and NRI, compared to adding each variable individually to the models (all p values 0.05, except for NRI in the models that included gestational age alone, or both gestational age and SF) (Table 6).

This study has several limitations that should be acknowledged. Firstly, although participants were enrolled from two medical centers, the sample size remains relatively modest, which may limit the statistical power and the generalizability of the findings. Expanding the sample size in future studies could enhance the robustness of the conclusions. Secondly, while serum $\beta$-hCG, SF, and gestational age demonstrated strong diagnostic value, variations in clinical protocols, diagnostic thresholds, and testing equipment across different institutions may influence the reproducibility of results. Additionally, some potential confounding variables, such as CRP, procalcitonin, maternal microbiome profiles or environmental factors, were not included in the analysis and may contribute to variability in outcomes. Finally, emerging biomarkers such as the neutrophil-to-lymphocyte ratio [27] were not included in this study but could be valuable additions to future investigations. Future research should prioritize larger, multicenter trials to validate these findings and improve diagnostic models for intrauterine infections in PROM cases. Incorporating genetic, immunological, and microbial assessments may further enhance diagnostic precision and support the development of personalized management strategies for high-risk patients.