Novel inflammatory indices compared with CRP for diagnosing pelvic inflammatory disease: a case-control study

Abstract

Background and objectives

Pelvic inflammatory disease (PID) is a common upper genital tract infection associated with serious long-term sequelae. Its clinical diagnosis is challenging due to non-specific symptoms, creating an urgent need for accurate, accessible biomarkers. While C-reactive protein (CRP) is commonly used, its specificity is limited. Novel, composite inflammatory indices derived from routine complete blood counts (CBC) have emerged as promising biomarkers in other fields. This study aims to evaluate the diagnostic performance of CBC-derived inflammatory indices against the conventional marker CRP in differentiating PID from non-inflammatory gynecological conditions and to exploratorily assess their association with descriptive ultrasound categories.

Methods

In this retrospective case-control study, 135 PID patients were compared with 135 age-matched non-PID controls. Key hematological indices—including the systemic immune-inflammation index (SII), systemic inflammation response index (SIRI), neutrophil-to-lymphocyte ratio (NLR), monocyte-to-lymphocyte ratio (MLR), and platelet-to-lymphocyte ratio (PLR)—were calculated from routine CBCs obtained at initial clinical presentation. Diagnostic accuracy was evaluated using receiver operating characteristic (ROC) curve analysis, and independent associations were assessed via multivariable logistic regression. As an exploratory analysis, the PID cohort was also subcategorized according to a novel, purely descriptive classification of ultrasound findings to investigate patterns of systemic inflammation across different anatomical presentations.

Results

PID patients exhibited significantly elevated levels of all novel inflammatory indices compared to controls (all p < 0.001). In multivariable analyses adjusted for clinical and laboratory confounders, SII, SIRI, NLR, MLR, and PLR remained strong and independent factors associated with PID (all p < 0.01). ROC analysis revealed that SII possessed the highest diagnostic accuracy (AUC = 0.870, 95% CI: 0.830–0.911), significantly outperforming CRP (AUC = 0.648, p < 0.001) and other single ratios. SIRI demonstrated comparable efficacy to SII (AUC = 0.832). The inflammatory indices showed a distinct pattern of strong intercorrelations specifically within the PID group. In the exploratory ultrasound analysis, the highest median levels of SII, SIRI, and NLR were observed in patients with isolated pelvic effusion, with lower levels seen in groups with findings such as tubo-ovarian abscess.

Conclusions

The CBC-derived indices SII and SIRI showed higher diagnostic accuracy for PID compared to CRP. As inexpensive, routine tests, they represent highly promising tools to aid timely and accurate PID diagnosis in clinical practice. The pattern of their association with different ultrasound phenotypes suggests they may reflect distinct inflammatory states, a hypothesis requiring validation in future studies.

Introduction

Pelvic inflammatory disease (PID) is a common infectious syndrome of the upper female genital tract, encompassing endometritis, salpingitis, and tubo-ovarian abscess [1]. It primarily affects sexually active young women and poses a significant public health burden [2, 3]. In the United States, national survey data indicate that a substantial proportion of reproductive-aged women are affected. For example, a study based on the National Health and Nutrition Examination Survey (NHANES) estimated that approximately 4.4% of sexually experienced women aged 18–44 reported a lifetime diagnosis of PID [4], with the prevalence rising to about 10% among those with a prior sexually transmitted infection (STI). More recent syntheses of national data estimate that roughly 2 million American women of reproductive age have experienced PID [1]. In China, a large case-control study further highlights the disease burden, confirming that a history of PID is a strong, independent risk factor for female infertility, underscoring its profound and lasting impact on reproductive health [5]. A declining trend in PID diagnoses has been noted in some high-income settings, potentially linked to widespread chlamydial screening programs [6, 7]. Nevertheless, significant disparities persist, with ethnic/racial minorities, particularly Black, non-Hispanic women, continuing to bear a disproportionate burden of the disease [2, 8].

A primary clinical challenge is the heterogeneous and often nonspecific presentation of PID. Symptoms can range from severe acute pelvic pain to being entirely asymptomatic [9]. Consequently, PID is both underdiagnosed and prone to misdiagnosis, as a purely clinical diagnosis based on pelvic tenderness lacks specificity [10, 11]. This diagnostic uncertainty is consequential, as it can lead to delayed or unnecessary antibiotic treatment [12]. More importantly, untreated or subclinical infection is a leading preventable cause of severe long-term sequelae, most notably tubal factor infertility, ectopic pregnancy, and chronic pelvic pain [11, 13, 14]. Therefore, improving the accuracy of diagnostic tools for PID remains a critical unmet need in clinical practice.

Given the diagnostic challenge, clinicians rely on adjunctive laboratory markers. In clinical practice, serum C-reactive protein (CRP) is a widely utilized acute-phase reactant that provides a non-specific measure of systemic inflammation [15]. CRP serves as a cornerstone biomarker in the clinical evaluation of suspected PID due to its rapid response to infection and inflammation [16, 17]. Elevated CRP levels can support the diagnosis and are often used to assess disease severity and monitor treatment response [18]. However, the diagnostic utility of CRP is inherently constrained by its lack of disease specificity. As a component of the innate immune response, CRP synthesis is upregulated by a broad spectrum of stimuli, including infections (bacterial, viral), trauma, surgery, and non-infectious inflammatory states [19]. This lack of discriminatory power necessitates the exploration of biomarkers that might better capture the specific inflammatory milieu of PID.

The search for improved diagnostics has extended beyond single protein biomarkers. Recent efforts include multiplex PCR assays from lower genital tract samples to quantify pathogen loads associated with endometritis [20], and sophisticated blood transcriptomic analyses to identify host-response signatures specific to PID [21]. While these approaches offer high biological specificity, their complexity, cost, and technical requirements limit their feasibility for widespread, rapid clinical adoption — particularly in primary care or emergency settings where timely PID diagnosis is most critical.

Recently, novel hematologic inflammation indices derived from routine complete blood count (CBC) parameters—such as the systemic immune-inflammation index (SII), systemic inflammation response index (SIRI), neutrophil-to-lymphocyte ratio (NLR), monocyte-to-lymphocyte ratio (MLR), and platelet-to-lymphocyte ratio (PLR)—have emerged as promising biomarkers in fields like oncology and cardiology [22, 23]. Their appeal lies in being inexpensive, routinely available, and integrating information from multiple immune and inflammatory pathways (e.g., neutrophils, lymphocytes, monocytes, and platelets), potentially offering a more comprehensive and stable reflection of the host’s inflammatory status than single-marker assays like CRP [24, 25].

Despite their promise in other diseases, a significant knowledge gap exists regarding their application in PID. First, comprehensive studies directly comparing the diagnostic performance of these composite indices against CRP are lacking. Second, their value as independent risk factors remains unclear. Most critically, the relationship between the levels of these systemic inflammatory markers and the diverse anatomical phenotypes of PID observable on imaging remains largely unexamined.

This study was therefore conducted with two primary aims. First, to determine whether novel CBC-derived inflammatory indices offer higher diagnostic accuracy for PID compared to conventional CRP, and to evaluate their independent association with the disease. Second, to exploratorily investigate whether the levels of these systemic markers show any discernible pattern across different anatomical presentations of PID seen on ultrasound. To enable this exploratory aim, we developed a novel, purely descriptive classification system for ultrasound findings, intended solely as a tool for hypothesis generation and not as a validated grading scale of disease severity. We hypothesized that composite inflammatory indices (SII and SIRI) would demonstrate higher diagnostic accuracy compared to CRP alone, and that their levels would exhibit distinct patterns across the descriptive ultrasound categories.

Materials and methods

Population and study design

This retrospective case-control study was conducted at Dongzhimen Hospital, Beijing University of Chinese Medicine. The study utilized the hospital’s standardized clinical pathway for acute PID. In this pathway, patients with suspected PID are first evaluated in the outpatient gynecology clinic. Upon clinical suspicion, routine blood tests (including a complete blood count and C-reactive protein) and a pelvic ultrasound are obtained during the outpatient visit. Patients meeting criteria for hospitalization are then admitted based on these initial findings. Notably, systemic antibiotic therapy for PID is initiated only after admission. This pathway ensures that all laboratory and imaging data used in this study were collected prior to hospitalization and before any definitive inpatient PID treatment. Data were retrieved from the electronic medical record system for patients admitted between May 2023 and December 2025.

The case group consisted of consecutive patients admitted with a primary diagnosis of acute PID, as defined by the U.S. Centers for Disease Control and Prevention (CDC) [10]minimum clinical criteria. This requires the presence of lower abdominal tenderness, if one or more of the following on pelvic examination: cervical motion tenderness, uterine tenderness, or adnexal tenderness. Although laparoscopic or microbiological confirmation was not available due to the retrospective nature of the study, all patients were hospitalized with a working diagnosis of acute PID and received inpatient intravenous antibiotic therapy. To reduce diagnostic misclassification, in cases where cross-sectional imaging (CT or MRI) was performed during hospitalization, patients found to have alternative diagnoses explaining their symptoms (e.g., acute appendicitis, ureteral calculi, ovarian torsion) were excluded from the case group.

The control group consisted of patients with non-inflammatory gynecological conditions (e.g., uterine fibroids, abnormal uterine bleeding). Controls were matched 1:1 to cases based on age (± 3 years), resulting in a total sample size of 270 (135 cases and 135 controls). This control group was selected to provide a preliminary, proof-of-concept comparison between PID and a baseline state of non-inflammatory pelvic conditions. We acknowledge that it does not represent the more clinically challenging differential diagnoses (e.g., acute appendicitis, ectopic pregnancy, ovarian torsion), and that this is a major limitation of the present study.

This control group was selected to provide a preliminary, proof-of-concept comparison between PID and a baseline state of non-inflammatory pelvic conditions. This design addresses the differential diagnostic question of PID versus other gynecological pathology in a hospital setting, rather than the detection of PID in asymptomatic community screening. We acknowledge that it does not represent the more clinically challenging differential diagnoses (e.g., acute appendicitis, ectopic pregnancy, ovarian torsion) and that this is a major limitation of the present study. Accordingly, the cut-off values identified herein (e.g., SII > 636) should not be considered generalizable to primary care or community settings without external validation.

Inclusion criteria for both groups were:

• (1) Age between 18 and 50 years;

• (2) Availability of a complete blood count (CBC), C-reactive protein (CRP), coagulation profile, and hepatic/renal function test results obtained within 24 hours of admission and prior to any surgical intervention or systemic treatment.

Exclusion criteria for both groups were:

• (1) pregnancy or within 42 days postpartum;

• (2) presence of other acute systemic infections, active autoimmune diseases, hematological disorders, or any malignancies;

• (3) Documented use of antibiotics, corticosteroids, or immunosuppressants within the 7 days prior to admission as recorded in the medical history.

• (4) History of trauma, surgery, or an invasive procedure within the past 4 weeks.

The study protocol was approved by the Institutional Ethics Committee of Dongzhimen Hospital (Approval No: [2026DZMEC-224-02]). The requirement for informed consent was waived due to the retrospective design and the use of anonymized clinical data.

Baseline and laboratory data of patients

Baseline sociodemographic and clinical data were obtained from electronic medical record reviews by trained personnel. The data included patient age, height, weight, smoking history, alcohol consumption, medical insurance type, and gynecological history (including gravidity, parity, history of abortion, and pelvic inflammatory disease). Body mass index (BMI) was calculated as weight (kg) divided by the square of height (m2). According to Chinese guidelines, BMI was categorized as underweight (< 18.5 kg/m2), normal (18.5 ≤ BMI < 24.0 kg/m2), overweight (24.0 kg/m2 ≤ BMI < 28.0 kg/m2), and obese (≥ 28.0 kg/m2).

Laboratory testing was conducted using blood samples collected within 24 h of patient admission, prior to the initiation of any intravenous antibiotic therapy. The laboratory indicators included white blood cell count, neutrophils, urea, creatinine, albumin, CRP, and hemagglutination index.

The novel inflammatory markers were calculated from CBC results as follows:

• SII = (platelet count × neutrophil count)/lymphocyte count.

• SIRI = (monocyte count × neutrophil count)/lymphocyte count.

• NLR = neutrophil count/lymphocyte count.

• MLR = monocyte count/lymphocyte count.

• PLR = platelet count/lymphocyte count.

Ultrasonographic classification and assessment

The pelvic ultrasound images archived for patients in the PID group were retrospectively retrieved and reviewed. To enable a blinded and unbiased assessment of the proposed classification system, all personally identifiable information and original radiologic reports were removed from these images prior to review. Given the absence of a universally accepted, standardized system for categorizing the sonographic manifestations of PID, we developed a descriptive categorical classification system encompassing the key anatomical presentations documented in the radiological literature [26]. This system aims to objectively categorize the predominant imaging finding at presentation without implying hierarchical severity grades or pathological progression.

The system aligns with clinical needs to distinguish presentations amenable to medical management from those requiring more aggressive intervention [27]. A key feature is the distinction between the tubo-ovarian complex and the tubo-ovarian abscess, which are recognized as distinct entities with different management implications [28].

Five descriptive categories were defined, each denoted by a letter reflecting its key anatomical feature:

• Category N (No Definitive Abnormality): No sonographic findings suggestive of acute PID. Physiologic pelvic fluid (<1 cm) may be present.

• Category N (No Definitive Abnormality): No sonographic findings suggestive of acute PID. Physiologic pelvic fluid (<1 cm) may be present.

• Category I (Tubal or Ovarian Involvement): Presence of sonographic signs of localized tube or ovary infection without complex adnexal mass infection. This includes:

• I-S: Acute Salpingitis/Pyosalpinx: Visualized as an abnormally dilated tubular structure with wall thickening (>5 mm) and/or thickened endosalpingeal folds appearing as incomplete septa or the cogwheel/beads on a string sign. The tubal content may be anechoic, echogenic , or demonstrate an air-fluid level.

• I-O: Acute Oophoritis: Manifests as ovarian enlargement with loss of corticomedullary differentiation.

• Category C (Tubo-ovarian Complex): Inflammatory adhesion between the tube and ovary, where both structures remain distinguishable (tubo-ovarian complex).

• Category A (Tubo-ovarian Abscess): Complex, poorly defined adnexal mass with merged, indistinguishable tubal and ovarian structures (tubo-ovarian abscess, TOA), characterized by thick walls, internal debris, and increased Doppler vascularity.

Assignment Rule: For the purpose of assigning a single, primary category, findings meeting the criteria for a more anatomically specific category took precedence over findings that defined a less specific category. The five categories are defined as descriptive labels and are not intended to constitute a validated severity grading scale. To ensure objectivity, classification was performed independently by two reviewers blinded to all clinical and laboratory data. Prior to assessment, both reviewers underwent a calibration session using 20 pilot images to ensure consistent application of criteria. Each reviewer assessed all images and assigned a category (N, E, I, C, A). Inter-rater agreement was assessed using Cohen’s kappa (κ) statistic. Discrepancies were resolved by consensus, with a third senior radiologist acting as an adjudicator if needed. The final agreed-upon category was used for all subsequent analyses.

Statistical analysis

Statistical analyses were performed using SPSS (version 26.0; IBM Corp., Armonk, NY) and GraphPad Prism (version 9.3.1). Continuous variables with a normal distribution were presented as the mean ± standard deviation (SD) and compared using independent sample t-tests; non-normally distributed data were expressed as the median (interquartile range, IQR) and compared using the Mann-Whitney U test for two groups, or the Kruskal-Wallis test for more than two groups with post-hoc Mann-Whitney U comparisons and adjustment for multiple comparisons. Categorical variables were described as frequencies (percentages) and compared using the Chi-square test or Fisher’s exact test, as appropriate. Baseline and clinical variables with a univariate association of P < 0.1 were entered into a multivariate binary logistic regression model to identify independent factors associated with PID. The diagnostic performance of SII, SIRI, and CRP was evaluated and compared using receiver operating characteristic (ROC) curve analysis (DeLong test). For all multiple comparisons, P < 0.05 was considered significant.

Results

Patients’ characteristics

This study included 270 patients, comprising 135 (50%) diagnosed with PID and 135 (50%) non-PID controls. Regarding diagnostic confirmation, all 135 PID patients met the CDC minimum clinical criteria (lower abdominal tenderness plus ≥ 1 of: cervical motion tenderness, uterine tenderness, or adnexal tenderness). Additional supportive findings were documented as follows: leukocytosis (WBC > 10 × 10⁹/L) in 35 patients (25.9%), and documented fever (oral temperature ≥ 38.3 °C) in 4 patients (3.0%). No patient underwent an endometrial biopsy or laparoscopic confirmation.

A documented medical history of PID was significantly more common in the case group (62.96% vs. 0%, p < 0.001), which should be interpreted in the context of potential ascertainment bias, as such history is more thoroughly sought in patients presenting with relevant symptoms. Additionally, concurrent vaginitis was more prevalent in the PID group (48.15% vs. 17.04%, p < 0.001), a finding that aligns with the shared etiological pathways and anatomical proximity between these conditions.

Importantly, the two groups were well-matched on key baseline characteristics, including age, height, weight, BMI, smoking, alcohol consumption, gravidity, parity, history of abortion, and IUD use (all p > 0.05) (Table 1).

Table 1.

Clinical characteristics of the study population

Variables	PID (n = 135)	non-PID (n = 135)	p-value
Age (years)	37.00(31.00,43.00)	38.00(32.00,44.00)	0.790a
Height (m)	1.61(1.57,1.65)	1.61(1.58,1.66)	0.271a
Weight (kg)	59.00 (54.00, 68.00)	60.00(55.00,68.00)	0.552a
BMI (kg/m2)	23.34(20.69,25.91)	22.68(20.70,25.26)	0.898a
Underweight, n (%)	16(11.85)	11(8.15)	/
Normal, n (%)	57(42.22)	73(54.07)	/
Overweight or obese, (%)	62(45.93)	51(37.78)	/
Smoking, n (%)	3(2.22)	4(2.96)	0.082b
Alcohol consumption, n (%)	0(0)	3(2.22)	0.702b
Previous pregnancies, n (%)	89(65.93)	86(63.70)	0.073b
Gravidity	1(0,2)	1(0,2)	0.559a
Parity	1(0,1)	1(0,1)	0.898a
History of abortion, n (%)	71(52.59)	56(42.48)	0.068b
Number of abortions	1(0,2)	0(0,1)	0.180a
Presence of an IUD, n (%)	14(10.37)	14(10.37)	1.000b
Current vaginitis, n (%)	65(48.15)	23(17.04)	< 0.001 b
History of PID, n (%)	85(62.96)	0(0)	< 0.001 b

All values given as numbers, mean ± standard deviation or median (IQR), BMI body mass index, IUD Intrauterine Device, PID Pelvic inflammatory disease, a Mann–Whitney U test, b Chi-square test

Patients in the PID group exhibited significantly elevated levels across all novel systemic inflammatory indices, including SII, SIRI, NLR, PLR, and MLR (p < 0.001).

Consistent with a state of systemic inflammation and potential subclinical coagulation activation, PID patients also demonstrated significantly higher levels of fibrin degradation products (FDP) and D-dimer (D-D) (both p < 0.001). In contrast, no significant differences were observed in markers of routine renal function (Cr, Urea), liver function (ALT, AST), or nutritional status (TP, ALB, ALP) between the two groups (Table 2).

Table 2.

Results of the comparisons between the case (PID) and control (non-PID) groups

Variables	PID (n = 135)	non-PID (n = 135)	p-value
CRP(mg/L)	1.00 (1.00, 6.02)	1.00 (1.00,1.00)	< 0.001
HGB (g/L)	127.00 (120.00,135.0)	127.00 (113.00,136.00)	0.366
WBC (× 109/L)	8.00 (6.40,10.10)	5.80 (4.90,6.80)	< 0.001
Neutrophils (× 109/L)	5.50 (3.90,7.10)	3.40 (2.70,4.10)	< 0.001
Lymphocytes (× 109/L)	1.60 (1.30,2.00)	1.80 (1.50,2.20)	0.007
Monocytes (× 109/L)	0.50 (0.40,0.60)	0.40 (0.30,0.50)	< 0.001
PLT(× 109/L)	272.00 (235.00,315.00)	256.00 (219.00,285.00)	0.005
SII	808.74 (600.47,1277.54)	501.67 (378.44,583.82)	< 0.001
SIRI	1.62 (0.93, 2.77)	0.72(0.50,0.98)	< 0.001
NLR	3.09 (2.16, 4.85)	1.89(1.47,2.38)	< 0.001
MLR	0.30 (0.22, 0.42)	0.21(0.17,0.26)	< 0.001
PLR	161.76 (133.68, 213.64)	142.38(109.44,168.95)	< 0.001
UREA (mmol/L)	4.15 (3.15, 5.05)	4.01(3.46,4.95)	0.879
Cr (µmol/L)	55.50(50.00,62.30)	56.70(51.10,61.10)	0.839
ALT (U/L)	14.30 (10.50, 24.40)	15.00(10.70,20.50)	0.838
AST (U/L)	17.50 (15.00, 23.60)	18.40(15.40,22.10)	0.754
TP (g/L)	71.30(66.30,76.30)	71.90(69.10,76.70)	0.153
ALB (g/L)	43.10(39.20,45.60)	43.80(41.70,45.60)	0.083
ALP (U/L)	54.30(45.60,68.10)	56.60(49.40,67.30)	0.325
PT(S)	11.20(10.70,12.20)	11.20(10.70,11.70)	0.068
APTT (S)	31.80 (29.50, 34.00)	31.60(29.80,33.70)	0.583
FIB (g/L)	2.86(2.36,3.28)	2.80(2.43.3.14)	0.340
TT(S)	14.90 (14.20, 16.00)	15.10(14.50,16.00)	0.106
FDP (mg/mL)	0.85 (0.53, 1.44)	0.54(0.33,0.84)	< 0.001
D-D (µg/mL)	97.00(59.00,144.00)	76.00(40.00,100.00)	< 0.001

All values given are numbers, mean ± standard deviation or median (IQR), CRP C-reactive protein, WBC white blood cell count, SII systemic immune-inflammation index, SIRI systemic inflammation response index, NLR neutrophil-to-lymphocyte ratio, MLR monocyte-to-lymphocyte ratio, PLR platelet-to-lymphocyte ratio, TP total protein, ALB albumin, FDP fibrin degradation products

Correlation analysis of inflammatory markers

Spearman correlation analysis was performed to examine the pairwise correlations among inflammatory markers in both groups (Table 3).

Table 3.

Spearman correlation coefficients among inflammatory markers in PID patients and non-PID controls

Indicator Pair	PID Group (n = 135)	Non-PID Group (n = 135)	Group Difference
	r (p-value)	r (p-value)	Z-value (p-value)
CRP-SII	0.405 (***<0.001)	0.068 (ns 0.433)	2.94 (0.003)
CRP-SIRI	0.512 (***<0.001)	0.029 (ns 0.738)	4.87 (***<0.001)
CRP-NLR	0.399 (***<0.001)	0.001 (ns 0.988)	3.85 (***<0.001)
CRP-MLR	0.479 (***<0.001)	-0.018(ns 0.837)	4.85 (***<0.001)
CRP-PLR	0.135 (ns 0.119)	-0.015(ns 0.861)	1.44 (0.151)
SII-SIRI	0.852 (***<0.001)	0.620 (***<0.001)	4.38 (***<0.001)
SII-NLR	0.892 (***<0.001)	0.694 (***<0.001)	4.68 (***<0.001)
SII-MLR	0.671 (***<0.001)	0.397 (***<0.001)	3.58 (***<0.001)
SII-PLR	0.596 (***<0.001)	0.518 (***<0.001)	0.86 (ns 0.390)
SIRI-NLR	0.891 (***<0.001)	0.718 (***<0.001)	4.52 (***<0.001)
SIRI-MLR	0.871 (***<0.001)	0.715 (***<0.001)	3.74 (***<0.001)
SIRI-PLR	0.333 (***<0.001)	-0.021 (ns 0.810)	2.99 (**0.003)
NLR-MLR	0.752 (***<0.001)	0.450 (***<0.001)	3.58 (***<0.001)
NLR-PLR	0.477 (***<0.001)	0.156(ns 0.071)	2.78 (**0.003)
MLR-PLR	0.493 (***<0.001)	0.383 (***<0.001)	1.22(0.222)

In the PID group, strong correlations were observed among the composite hematologic indices sharing neutrophil and lymphocyte components, with SII–NLR (r = 0.892), SIRI–NLR (r = 0.891), and SII–SIRI (r = 0.852) all showing very high correlations (all P < 0.001). CRP showed moderate correlations with SII (r = 0.405) and SIRI (r = 0.512; both p < 0.001).

In the control group, correlations among SII, SIRI, and NLR were weaker (SII–NLR: r = 0.694; SII–SIRI: r = 0.620; both p < 0.001), and CRP was not significantly correlated with SII (r = 0.068, p = 0.433) or SIRI (r = 0.029, p = 0.738).

Fisher Z‑tests indicated that the correlation coefficients for most marker pairs differed significantly between groups (Z = 2.78–4.87, p < 0.01), except for CRP–PLR (Z = 1.44, p = 0.151) and SII–PLR (Z = 0.86, p = 0.390), confirming stronger co‑variation among the cellular inflammatory indices in PID patients.

Heatmap visualization illustrated these differences (Fig. 1). In the PID group, SII, SIRI, and NLR formed a cluster of strong correlations (r > 0.85), consistent with their shared neutrophil and lymphocyte components. In the control group, correlations among these indices were weaker, and CRP showed no significant association with the composite indices.

Fig. 1.

Spearman correlation heatmaps of inflammatory markers. A Spearman correlation heatmap of the PID group (n = 135) showing strong correlations among SII, SIRI, and NLR (r > 0.85, red box). B Spearman correlation heatmap of the non-PID group (n = 135) with weaker correlations and CRP disconnection (p > 0.05, blue box). ***p < 0.001, ns: not significant

Independent associations of individual inflammatory markers with PID

To evaluate the independent association of each inflammatory marker with PID, multivariable binary logistic regression analyses were performed. For each of the six markers (CRP, SII, SIRI, NLR, MLR, and PLR), a separate series of three sequential models was constructed. This approach was used to mitigate multicollinearity, as strong intercorrelations were observed among the composite inflammatory indices in network analysis. The Crude Model was unadjusted. Model II was adjusted for sociodemographic and clinical factors. Model III was further adjusted for laboratory parameters. The full list of covariates for each model is provided in the footnote of Table 4.

Table 4.

Associations of inflammatory indices with PID

Variables	Crude model			Model II			Model III
	OR	95% CI	p-value	OR	95% CI	p-value	OR	95% CI	p-value
CRP	1.178	1.061–1.308	0.002	1.200	1.074–1.340	0.001	1.208	1.076–1.356	0.001
SII	1.008	1.006–1.011	<0.001	1.010	1.007–1.013	<0.001	1.011	1.004–1.017	< 0.001
SIRI	9.770	5.019–19.017	<0.001	9.505	4.516–20.006	<0.001	10.816	4.709–24.842	< 0.001
NLR	4.111	2.715–6.226	<0.001	4.444	2.790–7.078	<0.001	4.602	2.786–7.601	< 0.001
MLR	8.012	4.071–15.767	<0.001	9.823	4.626–20.860	<0.001	8.626	3.906–19.048	< 0.001
PLR	1.013	1.007–1.018	<0.001	1.016	1.009–1.022	<0.001	1.014	1.007–1.021	<0.001

Crude model: unadjusted

Model II: adjusting for age, BMI, alcohol consumption, smoking, history of pregnancy, gravidity, parity, history of abortion, number of abortions, and presence of vaginitis

Model III: adjusting for age, BMI, alcohol consumption, smoking, history of pregnancy, gravidity, parity, history of abortion, number of abortions, presence of vaginitis, ALT, AST, TP, ALB, ALP, PT, APTT, TT, FIB, FDP, and D-dimer

Odds ratios for SII and PLR are reported per 1-unit increase; clinical interpretation is provided in the text

After full adjustment in Model III, all six inflammatory markers remained independently and significantly associated with PID (all P < 0.001). The adjusted odds ratios (aORs) and 95% confidence intervals from all models are presented in Table 4. These findings confirm that elevated levels of these hematological indices in patients with PID are not attributable to the confounding factors adjusted for in the models.

Notably, the odds ratios for markers such as SII and PLR, which are reported per one-unit increase, should be interpreted with reference to their clinically relevant ranges. For example, an SII increase of 100 units corresponds to approximately a 3.0-fold increase in the odds of PID, given an OR of 1.011 per unit.

Diagnostic performance of inflammatory markers

The diagnostic performance of the inflammatory markers was assessed using receiver operating characteristic (ROC) curve analysis. CRP, SII, SIRI, NLR, MLR, and PLR all showed significant diagnostic value for PID (all p < 0.001). SII demonstrated the highest area under the curve (AUC = 0.870, 95% CI: 0.830–0.911), with a sensitivity of 71.9% and a specificity of 83.7% at the optimal cut-off value of 635.96 (Fig. 2). Pairwise comparison of AUCs using DeLong’s test showed that SII had a significantly higher AUC than CRP (AUC difference = 0.222, Z = 6.956, p < 0.001), as well as NLR, MLR, and PLR (all p < 0.05). The difference between SII and SIRI (AUC difference = 0.038, Z = 1.924, P = 0.054) did not reach statistical significance, suggesting comparable diagnostic performance between these two indices. The AUC, sensitivity, and specificity values for all markers are shown in Table 5.

Fig. 2.

The ROC curves of inflammatory markers for diagnosing PID

Table 5.

The AUC, sensitivity and specificity of inflammation markers for PID diagnosis

Variables	AUC (95% CI)	p-value	Cut-off value	Sensitivity	Specificity
CRP	0.648 (0.582–0.714)	<0.001	2.535	0.341	0.926
SII	0.870 (0.83–0.911)	<0.001	635.963	0.719	0.837
SIRI	0.832 (0.785–0.880)	<0.001	1.191	0.652	0.874
NLR	0.825 (0.777–0.872)	<0.001	2.513	0.659	0.844
MLR	0.741 (0.682-0.800)	<0.001	0.265	0.63	0.778
PLR	0.661 (0.597–0.725)	<0.001	177.157	0.444	0.815

Association between descriptive categories and the systemic inflammatory burden

The inter-rater reliability between the two independent gynecologists for the descriptive categories was excellent, with a Cohen‘s kappa of 0.92 (95% CI: 0.862–0.972, p < 0.001). The two assessors assigned identical categories in 127 of 135 cases (94.1%). All discrepant cases (n = 8) involved disagreements between categories with similar anatomical features and were resolved by consensus.

Among the 135 PID patients, the distribution according to the predefined descriptive categories was as follows: 49 (36.30%) with Category N (No definitive abnormality), 28 (20.74%) with Category E (Pelvic effusion), 41 (30.37%) with Category I ( Indicating tubal or ovarian involvement), 13 (9.63%) with Category C (Tubo-ovarian complex), and 4 (2.96%) with Category A (Tubo-ovarian abscess). The small sample size in Category A limits definitive conclusions for this subgroup.

For descriptive purposes, we also performed the Kruskal-Wallis test, which treats the categories as unordered groups. It indicated significant overall variation for NLR and CRP (p = 0.018 and p = 0.019, respectively), but not for the other indices (Table 6). The distribution of all six indices across categories is visualized in Fig. 3. Post-hoc pairwise comparisons were performed using the Mann-Whitney U test with Bonferroni correction. Results indicated that the significant overall differences for NLR and CRP were primarily driven by elevated levels in Categories E and I compared to Category N.

Table 6.

Systemic inflammatory markers stratified by descriptive ultrasound categories in patients with PID

Variables	Category N (n = 49)	Category E (n = 28)	Category I (n = 41)	Category C (n = 13)	Category A (n = 4)	p-value (K-W)
CRP (mg/L)	1.00(1.00,1.21)	1.00(1.00,30.24)	2.57(1.00,11.23)	1.58(1.00,11.01)	2.08(1.07,4.29)	0.019
SII	705.39 (561.84, 961.34)	1005.28(637.76,1889.09)	906.00(599.50,1662.67)	765.57(517.73,1205.66)	837.27(707.94,1505.00)	0.104
SIRI	1.27 (0.90, 2.09)	2.23(0.86,5.07)	1.88(1.02,3.48)	1.61(0.69,2.22)	1.53(1.45,2.19)	0.132
NLR	2.60(2.03,3.71)	3.94(2.60,6.95)	3.38(2.08,5.66)	2.69(1.96,4.34)	2.70(2.29,3.30)	0.018
MLR	0.28 (0.21, 0.36)	0.30(0.22,0.57)	0.35(0.24,0.43)	0.29(0.21,0.37)	0.31(0.29,0.33)	0.467
PLR	160.83 (121.39, 201.33)	170.63(144.20,247.22)	161.76(133.72,237.85)	174.74(127.14,202.67)	178.47(141.15,208.50)	0.356

Data are presented as the median (interquartile range [IQR])

K-W: Kruskal-Wallis test (treats categories as unordered groups)

p-values in bold indicate statistical significance (p < 0.05)

Fig. 3.

Distribution of systemic inflammatory indices across ultrasonographic descriptive categories. Box and whisker plots (min to max) display the levels of six indices: (A) SII, (B) SIRI, (C) NLR, (D) CRP, (E) MLR, and (F) PLR. The whiskers extend from the minimum to the maximum value, and the box itself spans the interquartile range (IQR, 25th to 75th percentile)

Descriptively, the highest median levels of SII, SIRI, and NLR were observed in Category E (pelvic effusion) rather than in categories denoting more localized or organized findings (C and A).

Discussion

Main findings and implications

This case-control study systematically evaluated novel inflammatory indices for the diagnosis of PID. We demonstrated that SII and SIRI are independently associated with PID and showed higher diagnostic accuracy compared to conventional CRP in this cohort. Notably, SII exhibited the highest AUC among the markers evaluated. Furthermore, the strong intercorrelations among these indices, as revealed by correlation analysis, suggest a coordinated cellular immune response in PID.

The baseline characteristics and laboratory profile of our cohort delineate a distinct clinical and systemic inflammatory picture of PID. Critically, the two groups were well matched on key demographic and obstetric factors, including age, BMI, and IUD use (all p > 0.05). The comparable rate of IUD use between groups (10.37% vs. 10.37%, p = 1.000) aligns with contemporary evidence suggesting that modern IUDs, when placed with appropriate screening, confer a much lower PID risk than historically perceived [29]. Notably, a study focusing on IUD-associated PID found that culture positivity did not correlate with disease severity or outcomes, further underscoring that the presence of an IUD itself may be less determinative of the inflammatory course than the host response [30]. This reinforces that the observed biomarker differences are more likely attributable to the infectious-inflammatory process of PID itself rather than confounding variables. As expected and consistent with the pathophysiology of ascending infection, concurrent vaginitis was significantly more prevalent in the PID group (48.15% vs. 17.04%, p < 0.001) [31, 32]. Contemporary insights into the cervicovaginal microbiome suggest that this dysbiotic lower tract environment acts as a critical risk factor, facilitating pathogen ascent and predisposing to PID [33]. Beyond local infection, PID patients exhibited a pronounced systemic response. They demonstrated not only elevated novel hematologic inflammation indices (SII, SIRI, NLR, MLR, PLR) but also significantly higher levels of fibrin degradation products (FDP) and D-dimer (D-D) (p < 0.001). This co-elevation suggests a state of subclinical coagulation activation intertwined with systemic inflammation. Our finding of elevated D-dimer in PID patients supports and extends prior observations linking D-dimer levels to disease severity in PID [34]. Collectively, these findings paint a picture of PID as a condition involving both a localized infectious focus and a measurable, multifaceted systemic host response. encompassing immune cell dysregulation and coagulation pathway engagement.

Our multivariate logistic regression analysis revealed that all novel inflammatory indices were independently associated with PID after rigorous adjustment for sociodemographic, clinical, and laboratory confounders. The composite indices yielded adjusted odds ratios of substantial magnitude (all aOR > 4.0 for SIRI, MLR, and NLR), whereas the association for CRP was more modest (aOR = 1.21 per 1 mg/L). The diagnostic performance assessed via ROC curve analysis further supported the potential utility of these markers. SII achieved the highest AUC (0.870), with an optimal cut-off value of 636, providing a sensitivity of 71.9% and specificity of 83.7%. SII showed a significantly higher AUC than CRP (AUC difference = 0.222, p < 0.001) and other individual ratios, while its performance was comparable to SIRI (p = 0.054). It should be noted, however, that CRP demonstrated very high specificity (92.6%) in this cohort, whereas SII primarily improved the balance between sensitivity and specificity. These markers may therefore offer complementary rather than replacement value.

While direct comparisons of SII and SIRI for PID diagnosis are lacking in the literature, the predictive value of composite inflammatory indices aligns with a broader principle observed in other chronic inflammatory and infectious conditions. For instance, SII has been established as a significant predictor of cardiovascular events and mortality, often outperforming single-parameter indices due to its integration of platelet, neutrophil, and lymphocyte pathways [35, 36]. Similarly, NLR has been consistently identified as an independent risk factor for adverse outcomes in various systemic inflammatory states [37]. Our findings extend this paradigm to PID, demonstrating that SII and SIRI may serve as useful diagnostic adjuncts and are independently associated with PID, as confirmed by multivariable logistic regression, even after adjusting for key clinical confounders, including concurrent vaginitis. This robustness supports their potential as adjunctive diagnostic tools that may reflect the integrated burden of dysregulated innate immunity, an adaptive immune response, and thrombotic activity characteristic of pelvic infection.

Beyond their diagnostic utility, the interrelationships among these markers offer insights into the inflammatory profile of PID. We observed that in PID patients, SII, SIRI, and NLR formed a strongly correlated cluster, suggesting that neutrophils, lymphocytes, and platelets do not vary independently but rather in a coordinated manner during active infection. In contrast, CRP was only moderately linked to this cellular cluster. This observed dissociation suggests that PID may involve two concurrent inflammatory axes: a prominent, coordinated cellular immune response (reflected by SII/SIRI) and a separate acute-phase protein response (reflected by CRP). The higher diagnostic accuracy of the composite indices may stem from their ability to capture the dominant cellular axis, which appears more specific to the infectious process in the pelvis than the generic acute-phase response captured by CRP.

The observed dissociation between the cellular cluster (SII/SIRI/NLR) and CRP aligns with their distinct biological origins. CRP, a prototypical acute-phase protein, is primarily synthesized by hepatocytes in response to pro-inflammatory cytokines such as IL-6,IL-1β, and TNF [38]. The regulation of these cytokines and their downstream signaling pathways (e.g., NF-κB) has been extensively reviewed in the context of anti-inflammatory mechanisms [39]. It represents a rapid but generalized systemic response to diverse inflammatory insults, including infection, trauma, and autoimmune processes [15]. In PID, as in other conditions, its elevation is a sensitive but non-specific indicator of inflammation. In contrast, indices like SII are derived from the direct enumeration of circulating immune cells—neutrophils, lymphocytes, and platelets—each representing a distinct but interacting arm of the host response. Our correlation analysis revealed that in PID, the counts of these cells change in a coordinated fashion. Therefore, the diagnostic advantage of SII and SIRI may stem from their capacity to quantitatively capture this multi-lineage immune perturbation, which appears to be a more pronounced feature of active pelvic infection than the isolated rise in CRP. This principle—that composite indices may better reflect complex, coordinated pathophysiology—is consistent with their established prognostic utility in diverse systemic conditions such as cardiovascular diseases and autoimmune disorders [36, 40, 41], and may now extend to PID.

A key strength of our study is the clear temporal sequence between biomarker measurement and treatment. Due to our hospital’s standardized clinical pathway, all inflammatory indices and ultrasound data were captured during the initial outpatient evaluation, prior to hospital admission, and the initiation of any systemic antibiotic therapy. This design feature mitigates a major confounding factor common in retrospective studies of infectious diseases and strengthens the inference that the observed biomarker elevations are directly related to the acute infectious process rather than being a consequence of treatment.

Importantly, the composition of our control group warrants specific discussion. We deliberately selected patients hospitalized for other non-inflammatory gynecological conditions (e.g., uterine fibroids) rather than healthy individuals. While this limits direct comparisons of absolute biomarker levels to a “healthy” baseline, it more accurately reflects the common clinical challenge of distinguishing PID from other pelvic pathologies in a hospital setting. This design strengthens the evaluation of the diagnostic specificity of the inflammatory indices for PID, as it tests their ability to differentiate PID from a clinically relevant differential diagnosis. Future prospective, multi-center studies with standardized diagnostic protocols are needed to validate our findings.

Clinical implications and pragmatic utility

The higher diagnostic accuracy of SII and SIRI, derived from routine complete blood counts, suggests potential pragmatic value for managing patients with suspected PID. Their utility may lie in providing an objective, supplemental measure of the systemic inflammatory burden to aid in clinical triage.

A key potential application of these indices is as a triage aid for identifying higher-risk presentations. In our cohort, an elevated SII (> 636) served as an objective marker of a pronounced systemic inflammatory response. In a symptomatic patient, such a finding could reasonably heighten clinical concern and support a more urgent management pathway, including timely initiation of empiric antibiotic therapy. The imperative for urgent imaging remains, but with adjusted expectations: prompt pelvic ultrasound serves the dual purpose of evaluating potential complications of infection and helping exclude other acute abdominal or pelvic pathologies.

Our data further introduce an important nuance: the composite hematologic indices (SII, SIRI, NLR) and CRP showed different distributions across the descriptive ultrasound categories. The highest median values for SII, SIRI, and NLR were observed in Category E (isolated pelvic effusion), with lower values in categories describing more localized or organized findings (I, C, A). In contrast, CRP levels were highest in Category I (tubal or ovarian involvement). This dissociation suggests that the CRP and the systemic cellular immune response (reflected by SII, NLR, SIRI) may reflect distinct aspects of the host inflammatory reaction. The pattern of higher cellular indices in patients with isolated pelvic effusion is an exploratory finding that requires validation and underscores that these markers likely capture a distinct immunologic state—possibly driven by acute, diffuse vascular permeability and leukocyte recruitment—which is not simply a proxy for the variety of anatomical patterns observed on ultrasound. These observations carry potential implications for clinical decision-making.

Building on these findings, in clinical assessment, these indices and ultrasound findings might be best viewed as complementary sources of information. An elevated SII in the context of less pronounced ultrasound findings could indicate a patient in an active inflammatory state where aggressive medical therapy might be most effective in this setting. Furthermore, the trade-off between sensitivity and specificity should be considered: CRP demonstrated very high specificity (92.6%) but modest sensitivity (34.1%), whereas SII achieved a more balanced profile (sensitivity 71.9%, specificity 83.7%). This suggests that SII and CRP may complement rather than replace each other—CRP could serve as a high-specificity rule-in test, while SII may help identify additional cases missed by CRP alone. Future studies should explore whether combining these markers could optimize diagnostic performance. This integrated perspective, informed by our specific findings, encourages a tailored approach that considers both the dynamic host response (via indices) and the anatomical sequelae (via imaging).

Limitations

Several limitations of this study should be acknowledged. First, this was a retrospective, single-center, case-control study, which may introduce selection bias and limit generalizability. The case-control design inherently precludes determination of the temporal relationship between biomarker elevation and PID diagnosis; therefore, our findings demonstrate associations but cannot establish causality. Additionally, the small number of patients in Category A (n = 4) limits statistical power for subgroup analyses. Multi-center prospective studies with larger sample sizes are needed to validate these findings.

Second, the diagnosis of PID relied on the CDC minimum clinical criteria without laparoscopic, endometrial biopsy, or universal microbiological confirmation. Although all patients were hospitalized with a clinical diagnosis of PID, the absence of a true gold standard may affect the estimated diagnostic performance. Furthermore, because inflammatory markers themselves may have influenced clinicians’ diagnostic decisions, there is a potential for incorporation bias. Reliance on the CDC minimum clinical criteria alone may have resulted in the inclusion of patients with pelvic pain from other etiologies, potentially inflating the reported AUC values. Although additional supportive findings were documented (leukocytosis in 25.93%, concurrent vaginitis in 48.15%, and fever in 3.00%), the absence of a true gold standard remains a major limitation that should be carefully considered when interpreting the diagnostic performance of the indices. It should also be noted that transvaginal ultrasound, performed in all cases to assess for sonographic features of PID, is limited to evaluation of the pelvic reproductive organs and cannot exclude extra-pelvic or non-gynecologic causes of acute pelvic pain. Future prospective studies incorporating standardized diagnostic confirmation (e.g.,, laparoscopic or microbiological evaluation), biomarker-independent enrollment protocols, and comprehensive imaging to exclude alternative diagnoses are needed to address these limitations.

Third, the control group consisted of patients with non-inflammatory gynecological conditions, which does not reflect the clinically challenging differential diagnosis of acute pelvic pain in emergency settings (e.g., acute appendicitis, ectopic pregnancy, ovarian torsion). The reported diagnostic performance may therefore be overestimated relative to real-world clinical scenarios. Additionally, the cut-off values identified in this study (e.g., SII > 636) require external validation before they can be considered generalizable to other populations or clinical settings.

Fourth, several methodological limitations should be acknowledged. Regarding the ultrasonographic classification, the descriptive system, although demonstrating excellent inter-rater reliability, was developed specifically for this exploratory analysis and has not been externally validated. Its potential clinical utility requires prospective evaluation. Regarding the inflammatory indices, the strong intercorrelations among SII, SIRI, and NLR are partially attributable to their shared neutrophil and lymphocyte components; therefore, their biological independence should not be overstated. Although separate regression models were used to avoid multicollinearity, these indices are mathematically derived from overlapping CBC parameters and may reflect overlapping aspects of the host inflammatory response rather than distinct biological pathways. The sensitivity of SII at the optimal cut-off was 71.9%, indicating that it is insufficient as a standalone diagnostic test and should complement, not replace, clinical judgment and current CDC diagnostic criteria. Regarding methodological considerations, internal validation was not performed for the ROC analysis, and the reported AUC values may be optimistic. Additionally, information on microbial etiology (e.g., Chlamydia trachomatis, Neisseria gonorrhoeae) was not available; yet, different pathogens may elicit distinct inflammatory responses.

Fifth, despite adjustment for multiple confounders, residual confounding from unmeasured variables (e.g., detailed sexual history, specific microbial etiology) cannot be excluded. Furthermore, antibiotic use prior to presentation was assessed retrospectively based on medical record review; patients may have received antibiotics elsewhere or self-medicated without this being documented. This potential information bias could affect the generalizability of findings to the true untreated PID population. Additionally, as a retrospective analysis, data on other specific inflammatory cytokines (e.g., IL-6, TNF-α) were not routinely available, preventing a direct comparison of these hematologic indices with established inflammatory mediators.

Despite these limitations, our findings have important clinical implications. The higher diagnostic accuracy of SII and SIRI, combined with their derivation from routine, low-cost CBC parameters, positions these indices as promising tools for improving PID diagnosis in resource-limited settings. Future research should focus on validating these indices in prospective cohorts, developing clinical decision algorithms that incorporate these markers alongside clinical symptoms and anatomical presentation, and exploring their utility in guiding treatment intensity and predicting long-term sequelae such as infertility and chronic pelvic pain.

Conclusion

In conclusion, the SII and SIRI, derived from routine blood tests, showed higher diagnostic accuracy for PID compared to conventional CRP in this cohort and were independently associated with the disease. An exploratory analysis of descriptive ultrasound categories further suggested that these cellular markers may reflect distinct inflammatory states, with the highest levels observed in patients with isolated pelvic effusion. As an adjunct to clinical evaluation, these indices may help improve diagnostic precision, particularly in ambiguous cases, but should complement, not replace, clinical judgment and current CDC diagnostic criteria. Future prospective studies are warranted to validate these indices and explore their prognostic utility.

Authors’ contributions

Designed study: JWK, LT; Performed research: JWK, JJZ, XHS, CY, LT; Analyzed data: JWK; Critical revision: JWK, JJZ, XHS, CY; Wrote the paper: JWK.

Funding

The Fourth National Medical Master Inheritance Workshop Construction Project (401091401).

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request, subject to institutional ethics.

Declarations

Ethics approval and consent to participate

The Ethics Committee of Dongzhimen Hospital, Beijing University of Chinese Medicine, approved this study (2026DZMEC-224-02). The requirement for informed consent was waived by the Institutional Ethics Committee of Dongzhimen Hospital due to the retrospective design and the use of anonymized clinical data. All procedures performed in this study involving human participants were in accordance with the ethical standards of the institutional research committee and with the Declaration of Helsinki.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.