Incidence of postoperative fever and bloodstream infections in gynecological surgical patients: a three-year retrospective surveillance

Abstract

Background

Postoperative bloodstream infection (PBSI) is a severe complication in gynecological surgery, associated with higher morbidity than surgical site infections. However, large-scale data on PBSI epidemiology and risk factors, particularly from the Chinese population and within modern Enhanced Recovery After Surgery (ERAS) pathways, are scarce. This study aimed to define the incidence, etiology, and modifiable risk factors for PBSI in a major Chinese tertiary center.

Methods

We conducted a three-year (Jan 2022–Dec 2024) retrospective cohort study of 19,906 patients undergoing gynecological surgery. Data were extracted from electronic medical records to determine the incidence of postoperative fever (PF) and PBSI. Pathogen distribution, antimicrobial susceptibility, and risk factors were analyzed using univariable and multivariable logistic regression models.

Results

The overall incidence of PF was 2.67% and PBSI was 1.30%. The institutional rate of preoperative vaginal preparation was strongly and inversely correlated with PBSI incidence (ρ = -0.50, P = 0.002). Multivariable analysis identified several independent risk factors, including combined high-risk surgery (aOR 3.72; 95% CI, 2.84–4.89), and seasonality, with peak risk in Q3 (aOR 2.03; 95% CI, 1.41–2.96). Escherichia coli was the predominant pathogen in culture-positive PBSI (70.37%), exhibiting high resistance to the first- and second-generation cephalosporins commonly used for prophylaxis, but remained susceptible to carbapenems and β-lactam/β-lactamase inhibitors.

Conclusions

The implementation of preoperative vaginal preparation as a core component of a quality improvement initiative was associated with a lower institutional PBSI rate. The high prevalence of resistant E. coli likely neutralizes the effectiveness of standard antibiotic prophylaxis, mandating a shift towards antibiotic stewardship guided by local antibiograms. Risk stratification should also account for surgical complexity and seasonal peaks. These findings provide a data-driven framework for optimizing infection prevention strategies in modern gynecological surgery.

Clinical trial registration number

Not applicable.

Background

Gynecological surgery is a cornerstone of treatment for a wide range of women’s health conditions. The transition from traditional open procedures to minimally invasive surgery has expanded its applications, yet serious postoperative complications, particularly infections, remain a significant clinical challenge [1, 2]. Among these, postoperative bloodstream infection (PBSI) is especially dangerous, often leading to prolonged hospitalization, increased costs, and substantially higher risks of sepsis and mortality compared to more localized surgical site infections (SSIs) [3–5].

Despite its severity, PBSI remains a comparatively understudied area, with most research focusing on SSIs. Crucially, there is a scarcity of large-scale, contemporary data on PBSI risk factors specifically within the Chinese population. Furthermore, the widespread adoption of Enhanced Recovery After Surgery (ERAS) protocols, which aim to minimize surgical stress and reduce complications, necessitates an updated understanding of infection risks within these modern care pathways. Establishing a clear epidemiological baseline and identifying modifiable risk factors in this context is essential for optimizing patient safety and guiding evidence-based prevention.

To address these gaps, we conducted a three-year retrospective surveillance at a major tertiary hospital in China. This study aims to: (1) determine the incidence and trends of postoperative fever (PF) and PBSI in a large cohort of patients undergoing gynecological surgery; (2) characterize the pathogens and their antimicrobial resistance patterns; and (3) identify key clinical and procedural risk factors to provide a robust rationale for targeted infection prevention and control strategies.

Methods

Study design and setting

This retrospective cohort study was conducted at the Sichuan Academy of Medical Sciences & Sichuan Provincial People’s Hospital, a major tertiary care center in Western China with over 4,300 beds and an annual gynecological surgical volume of approximately 6,000 procedures. We examined the electronic medical records (EMRs) of patients who underwent gynecological surgery from January 2022 to December 2024. The study was approved by the hospital’s Institutional Review Board, and the requirement for individual patient consent was waived given the retrospective and de-identified nature of the data.

Participant selection

The inclusion criteria for participants were: (1) undergoing gynecological surgery at our hospital during the study period; (2) having a postoperative hospitalization of at least 48 h for observation; and (3) possessing a complete EMR. Exclusion criteria were: (1) presence of a preexisting infectious disease or receiving ongoing antimicrobial therapy prior to surgery; (2) procedures not performed in the main operating room; and (3) EMRs with critical data missing.

Data collection and definitions

Data were extracted from the EMRs, including demographic data, laboratory results, vital signs, clinical symptoms, and details of each surgical event and related medical advice.

Primary outcomes

Postoperative Fever (PF): Defined as an axillary temperature > 38.0 °C for more than two consecutive days during postoperative hospitalization [6].

Postoperative Bloodstream Infection (PBSI): Defined, with reference to previous studies [7, 8], as the simultaneous fulfillment of all four criteria: (1) an axillary temperature > 38.0 °C persisting for ≥ 2 consecutive days after surgery; (2) no other identifiable focus of infection; (3) initiation of systemic antimicrobial therapy by a clinician; and (4) at least one blood culture obtained, irrespective of the result.

Key procedures and definitions

Surgical Categories: Surgical procedures were first categorized based on the primary gynecological surgery and any concurrent operations. Informed by the initial analysis, a composite binary variable, termed ‘Combined high-risk surgery’, was then constructed for inclusion in the multivariable regression model. This variable was designed to group the surgical categories that demonstrated a statistically significant and clinically meaningful increase in PBSI incidence in the univariate comparison. Specifically, a procedure was classified as ‘high-risk’ and coded affirmatively if it involved either (1) Gynecology + Gastrointestinal Surgery or (2) Gynecology + Hepatobiliary Surgery. All other surgical categories served as the reference group in the model.

Preoperative Vaginal Preparation: The standard institutional procedure, when performed, involved mechanical cleansing of the vaginal vault and cervix in the operating room after anesthesia induction. A sterile sponge stick soaked in 10% povidone-iodine solution was used for systematic scrubbing for a duration of at least 30 s.

Prophylactic Antimicrobial Administration: Prophylaxis primarily consisted of first- or second-generation cephalosporins (most commonly Cefazolin or Cefuroxime), administered as a single intravenous dose within 30–60 min prior to surgical incision.

Statistical analysis

Data were analyzed using R statistical software (version 4.4.1). Continuous variables were reported as mean ± standard deviation (SD) or median (interquartile range, IQR), and categorical variables were reported as frequencies (%). The chi-squared test was used to compare rates, and Spearman’s rank correlation test was used to assess associations between monthly incidence and process measure rates.

To identify risk factors for PBSI, potential variables were first screened using univariable logistic regression. Variables with a P-value < 0.20 in the univariable analysis or those of known clinical importance were included in a final multivariable logistic regression model to calculate adjusted Odds Ratios (aOR) and 95% Confidence Intervals (CI). Given the significant quarterly variations observed in the initial analysis and supporting evidence from existing literature, the season (categorized as Q1, Q2, Q3, Q4) was included in the final model a priori to ensure adequate adjustment for this potential confounder. A two-sided P-value < 0.05 was considered statistically significant.

Results

Participant baseline characteristics

A total of 22,522 patient records were extracted from the gynecological surgery database. After applying the exclusion criteria, 19,906 participants were included in the study. The detailed participant selection process is shown in Fig. 1. The baseline demographic and clinical characteristics of the study participants are presented in Table 1.

Fig. 1.

Participants selection process. Flowchart detailing the inclusion and exclusion of patients from the initial database query to the final study cohort

Table 1.

Baseline demographic and clinical characteristics of participants

Characteristic	Value
Demographics
Age, years	44.0 (34.0, 52.0)
BMI (kg/m²)	22.6 (20.7, 24.8)
Ethnicity, n (%)
Han	17,897 (89.9%)
Zang	1,446 (7.3%)
Other minorities	563 (2.8%)
Comorbidities & lifestyle, n (%)
Diabetes	832 (4.2%)
Hypertension	2,080 (10.4%)
Smoking	14 (< 0.1%)
Preoperative laboratory values
WBC, ×10⁹/L	9.3 (7.5, 11.2)
HGB, g/L	117.0 (105.0, 127.0)
PLT, ×10⁹/L	208.0 (176.0, 244.0)
ALB, g/L	43.8 (41.7, 45.5)
CREA, µmol/L	80.0 (70.4, 92.7)
eGFR, mL/min/1.73 m²	108.8 (103.1, 114.9)
ALT, U/L	15.0 (12.0, 20.0)
Surgical characteristics
ASA score, n (%)
I, II	19,008 (95.5%)
III, IV, V	898 (4.5%)
Surgical approaches, n (%)
Laparoscopy	17,316 (87.0%)
Laparotomy	2,590 (13.0%)
Incision type, n (%)
I, II	4,335 (21.8%)
III, IV	15,571 (78.2%)
Surgical duration, min	65.0 (25.0, 110.0)
Intraoperative blood loss, mL	50.0 (50.0, 50.0)
Perioperative interventions, n (%)
Prophylactic antimicrobials	11,964 (60.1%)
Vaginal preparation	3,065 (15.4%)

The incidence rate of PF and PBSIs

A total of 19,906 patients who underwent gynecological surgery were included in this study. Among them, 532 developed PF, corresponding to an overall incidence of 2.67%. PBSIs occurred in 259 patients, for an incidence of 1.30%. During the surveillance period, the monthly incidence of PF ranged from 0% to 6.81%, while that of PBSIs ranged from 0% to 3.81%. The monthly trends of these incidences are shown in Fig. 2.

Fig. 2.

Incidence of PF and PBSI by month. The graph displays the temporal trends in the incidence rates (%) of PF and PBSI for each month from January 2022 to December 2024

Quarterly incidence of PF and PBSI

A significant seasonal variation was observed in the incidence of both PF and PBSI (overall P < 0.001 for both; Table 2). The rates for both conditions peaked in the third and fourth quarters. Specifically, the incidence of PF was highest in Q4 at 3.25% (95% CI, 2.74–3.83), compared to 2.12% in Q2. Similarly, the incidence of PBSI peaked in Q3 at 1.71% (95% CI, 1.38–2.11), a substantial increase from the 0.91% (95% CI, 0.66–1.21) observed in Q1.

Table 2.

Quarterly incidence and comparison of PF and PBSI

Quarter	No. of Surgeries (N)	PF Cases (n)	PF Incidence (%, 95% CI)¹	PBSI Cases (n)	PBSI Incidence (%, 95% CI)¹
Q1	4,971	110	2.21 (1.82–2.66)	45	0.91 (0.66–1.21)
Q2	5,381	114	2.12 (1.75–2.54)	61	1.13 (0.86–1.47)
Q3	5,248	168	3.20 (2.74–3.72)	90	1.71 (1.38–2.11)
Q4	4,306	140	3.25 (2.74–3.83)	63	1.46 (1.12–1.88)
Total	19,906	532	2.67 (2.45–2.91)	259	1.30 (1.15–1.47)
Overall P-value²			< 0.001		< 0.001

¹ 95% Confidence Interval calculated using the Clopper-Pearson method

² Overall P-value from Chi-square test for comparison across all four quarters

Bacteriological profile of PBSIs

Of the 259 patients who met the criteria for PBSI, 27 (10.42%) had a laboratory-confirmed bloodstream infection with a positive blood culture. The remaining 232 patients (89.58%) were classified as having clinical PBSI based on the presence of postoperative fever and the initiation of antibiotic therapy, despite having negative blood cultures. Among the 27 culture-positive cases, Escherichia coli (E. coli) was the most frequently isolated pathogenic bacteria, representing 70.37% (19/27) of all isolates. The full bacteriological profile is detailed in Table 3.

Table 3.

Pathogen distribution in Culture-positive cases

Pathogen	No. of Isolates (n)	Percentage (%)
Gram-negative bacteria	22	81.48
Escherichia coli	19	70.37
Klebsiella pneumoniae	1	3.70
Enterobacter cloacae	1	3.70
Bacteroides fragilis	1	3.70
Gram-positive bacteria	5	18.52
Staphylococcus epidermidis	3	11.11
Staphylococcus hominis	2	7.41
Total	27	100.0

Antimicrobial susceptibility profile of E. coli in PBSIs

E. coli was the most common pathogenic bacteria identified in PBSIs, with 19 strains isolated. The antimicrobial susceptibility of these isolates was evaluated. High rates of resistance (> 80%) were observed for piperacillin, ticarcillin, ampicillin, and tetracycline. Resistance was also considerable (50–80%) against several broad-spectrum antimicrobials, including levofloxacin, moxifloxacin, and third-generation cephalosporins (e.g., cefotaxime, ceftriaxone). The isolates remained highly susceptible to several antimicrobials, with resistance rates ≤ 16%. These included carbapenems (imipenem, meropenem), beta-lactam/beta-lactamase inhibitor combinations (e.g., piperacillin/tazobactam), and amikacin. The complete antimicrobial resistance profile is presented in Fig. 3.

Fig. 3.

Antimicrobial resistance profile of E. coli isolates from PBSI cases. The bar chart shows the percentage of resistance for 19 E. coli strains isolated from culture-positive PBSI cases against a panel of common antimicrobials

The incidence of PF and PBSIs by type of surgery

Similar to trends seen in PF, type of concomitant surgery was a key contributor to risk of PBSIs. Specifically, concomitant gastrointestinal surgery (4.07%, χ²=17.47, P < 0.001) and combined hepatobiliary surgery (10.00%, χ²=5.94, P = 0.015) both contributed to a significantly higher rate of PBSIs. In contrast, obstetric surgery was associated with a much lower rate of PBSIs (0%, χ²=12.33, P < 0.001). The PBSI rates for other surgical types were not significantly different from the cohort average. A comparison of PF and PBSIs incidence among different surgical types is detailed in Table 4.

Table 4.

Comparison of the incidence rates of PF and PBSI by type of surgery

Type of surgery	Incidence of PF (%, n/N)	χ² *	P value *	Incidence of PBSI (%, n/N)	χ² *	P value *
Gynecological surgery	2.71 (498/18368)	0.04	0.841	1.32 (242/18368)	0.01	0.920
Gynecology + Obstetrics Surgery	0.49 (5/1015)	17.49	< 0.001	0 (0/1015)	12.33	< 0.001
Gynecology + Gastrointestinal Surgery	6.98 (24/344)	21.88	< 0.001	4.07 (14/344)	17.47	< 0.001
Gynecology + Urological Tract Surgery	0 (0/96)	1.70	0.192	0 (0/96)	0.45	0.502
Gynecology + Hepatobiliary Surgery	10.00 (2/20)	1.78	0.182	10 (2/20)	5.94	0.015
Gynecology + Other surgeries	4.76 (3/63)	0.40	0.527	1.59 (1/63)	0.00	1.000
Total	2.67 (532/19906)	Ref.	Ref.	1.3 (259/19906)	Ref.	Ref.

* The ‘Total’ row was used as the reference group for all Chi-squared (χ²) tests and P value calculations

Correlation of preoperative prophylactic antimicrobial administration, preoperative vaginal preparation, and incidence of PF and PBSIs

The rate of preoperative vaginal preparation showed no correlation with the incidence of PF (ρ = -0.07, P > 0.05). In contrast, it was significantly and negatively correlated with the incidence of PBSIs (ρ = -0.50, P = 0.002). Furthermore, there was no significant correlation between the rate of preoperative prophylactic antimicrobial administration and the incidence of either PF (ρ = -0.04, P > 0.05) or PBSIs (ρ = -0.21, P > 0.05). This inverse relationship between the rate of preoperative vaginal preparation and the incidence of PBSIs is comprehensively illustrated in Fig. 4, which depicts the opposing temporal trends (Fig. 4A and B) and the statistically significant negative correlation (Fig. 4C).

Fig. 4.

Preoperative vaginal preparation is associated with a reduced incidence of PBSI. (A) The monthly incidence rate of PBSI from January 2022 to December 2024. (B) The monthly rate of preoperative vaginal preparation during the same period. (C) Scatter plot showing a significant negative correlation between the vaginal preparation rate and PBSI incidence (Spearman’s ρ = -0.50, P = 0.0021)

Univariable and multivariable analyses of risk factors

A total of 19,906 patients were included in the final analysis, of whom 259 (1.30%) developed a PBSI. The results of the univariable and multivariable logistic regression analyses are detailed in Table 5.

Table 5.

Univariate and multivariable analysis of risk factors for PBSI

Characteristic	n/N (%)	Univariate Analysis Crude OR (95% CI)	P-value	Multivariable Analysis Adjusted OR (95% CI)	P-value
Age (per 10 years increase)	259 / 19,906 (1.29)	1.43 (1.31–1.56)	< 0.001	1.20 (1.07–1.34)	0.001
Surgical Duration (per 30 min increase)	259 / 19,906 (1.29)	1.02 (1.01–1.04)	0.012	1.01 (1.00-1.02)	0.024
Intraoperative Blood Loss (per 100mL increase)	259 / 19,906 (1.29)	1.25 (1.18–1.31)	< 0.001	1.18 (1.12–1.24)	< 0.001
ASA Score
I/II	205 / 17,435 (1.17)	Ref	-	Ref	-
III/IV/V	54 / 2,471 (2.18)	1.89 (1.38–2.53)	< 0.001	1.20 (0.85–1.66)	0.294
Diabetes
No	240 / 19,074 (1.25)	Ref	-	Ref	-
Yes	19 / 832 (2.28)	1.84 (1.11–2.87)	0.011	1.05 (0.62–1.70)	0.836
Hypertension
No	223 / 17,826 (1.25)	Ref	-	Ref	-
Yes	36 / 2,080 (1.73)	1.40 (0.96–1.96)	0.065	0.72 (0.48–1.06)	0.105
Vaginal Preparation
No	214 / 16,843 (1.26)	Ref	-	Ref	-
Yes	45 / 3,063 (1.47)	1.16 (0.83–1.59)	0.358	0.86 (0.61–1.19)	0.379
Antibiotic Prophylaxis
No	55 / 7,939 (0.69)	Ref	-	Ref	-
Yes	204 / 11,967 (1.70)	2.47 (1.85–3.37)	< 0.001	1.68 (1.24–2.32)	0.001
Combined high risk Surgery
No	242 / 19,542 (1.24)	Ref	-	Ref	-
Yes	17 /364 (3.47)	5.12 (3.99–6.58)	< 0.001	3.72 (2.84–4.89)	< 0.001
Season of Surgery
Q1	44 / 4,970 (0.89)	Ref	-	Ref	-
Q2	61 / 5,382 (1.13)	1.28 (0.87–1.91)	0.209	1.29 (0.87–1.93)	0.205
Q3	91 / 5,249 (1.71)	1.95 (1.37–2.83)	< 0.001	2.03 (1.41–2.96)	< 0.001
Q4	63 / 4,305 (1.46)	1.66 (1.13–2.46)	0.010	1.67 (1.12–2.51)	0.012

Abbreviations: OR, Odds Ratio; CI, Confidence Interval; ASA, American Society of Anesthesiologists

In the multivariable analysis, after adjusting for all covariates, several factors remained as significant independent risk factors for PBSI. Combined high risk surgery was identified as the strongest risk factor (aOR 3.72; 95% CI, 2.84–4.89; P < 0.001). A significant seasonal pattern was observed, with the highest risk in the third quarter (Q3: aOR 2.03; 95% CI, 1.41–2.96; P < 0.001) and fourth quarter (Q4: aOR 1.67; 95% CI, 1.12–2.51; P = 0.012) compared to the first quarter.

Other independent risk factors included increasing age (per 10 years: aOR 1.20; 95% CI, 1.07–1.34; P = 0.001), greater intraoperative blood loss (per 100mL: aOR 1.18; 95% CI, 1.12–1.24; P < 0.001), and longer surgical duration (per 30 min: aOR 1.01; 95% CI, 1.00–1.02; P = 0.024). Paradoxically, the use of antibiotic prophylaxis was also associated with a higher risk (aOR 1.68; 95% CI, 1.24–2.32; P = 0.001), likely reflecting confounding by indication.

Notably, several factors that were significant in the univariable analysis, such as a higher ASA score (P = 0.294) and the presence of diabetes (P = 0.836), lost their statistical significance after adjustment in the multivariable model.

Discussion

A principal finding of our surveillance was a significant inverse association between the institutional rate of preoperative vaginal preparation and the incidence of PBSI (ρ = -0.50, P = 0.002). This association is biologically plausible, as the theoretical aim of vaginal cleansing with antiseptics is to mitigate the intra-vaginal microbial load, thereby potentially limiting the risk of ascending infection [9–12]. Our findings are consistent with major clinical stakeholder recommendations. For example, both the American College of Obstetricians and Gynecologists (ACOG) and Chinese Medical Association’s Gynecological Oncology Group (GOG-CMA) recommend preoperative vaginal preparation in all procedures that are vaginal in nature or involve a vaginal incision [13, 14]. Our results also align with existing literature. For instance, a study by Skeith et al. [15] found that patients who underwent vaginal preparation had a 38% lower rate of infection following hysterectomy (P < 0.05). Similarly, a systematic review by Liu et al. [16] concluded that preoperative vaginal preparation was associated with a significantly lower risk of PF and SSIs after cesarean delivery. However, this relationship becomes more nuanced in our multivariable analysis. While the direction of the effect remained protective (Adjusted OR = 0.86), it did not reach statistical significance (95% CI: 0.61–1.19, P = 0.379) after adjusting for a range of patient- and procedure-related confounders. This discrepancy suggests that while vaginal preparation is a critical component of a successful infection control bundle, its individual effect may be statistically confounded by more powerful predictors in the model. Therefore, the strong negative correlation observed in Fig. 4 should be interpreted as powerful real-world evidence for the effectiveness of a quality improvement program in which vaginal preparation was a cornerstone, even if its independent contribution is difficult to isolate from other co-occurring factors in a complex patient population.

According to the 2022 China Antimicrobial Resistance Surveillance Report, E. coli accounted for 49.9% of pathogens in bloodstream infections [17]. Our surveillance, however, found a much greater proportion for PBSIs, where E. coli comprised 70.37%. This may be due to the unique bacterial colonization of surgical sites in gynecological surgeries compared to elsewhere and other regional differences. The susceptibility testing for these E. coli isolates indicated an alarming resistance pattern, with high rates of resistance for piperacillin, ticarcillin, ampicillin, and tetracycline. In contrast, they demonstrated high susceptibility to β-lactam/β-lactamase inhibitor combination and carbapenems. These data should be used to assist with guiding empirical therapy when PBSI is suspected but culture data are not yet available. In accordance with the expert opinion on this issue [18, 19], and supported by our susceptibility data, we recommend a tiered strategy. For mild-to-moderate infections, a β-lactam/β-lactamase inhibitor combination (e.g., cefoperazone/sulbactam, piperacillin/tazobactam) is a reasonable first-line medication. For severe infections, and especially in the setting of suspected ESBL-producing organisms, carbapenems (e.g., imipenem, meropenem) should be considered the primary medication.

A notable finding in our study was the low blood culture positivity rate (10.42%) among patients classified with PBSI. This phenomenon is a well-documented challenge in the diagnosis of postoperative infections and can be attributed to several factors. A primary contributor is likely the widespread administration of preoperative antibiotic prophylaxis, which can suppress bacterial growth and lead to false-negative culture results even when a true infection is present. Other potential factors include the suboptimal timing or volume of blood culture collection relative to fever spikes or episodes of bacteremia, the transient nature of bacteremia in some surgical site infections that may not be captured by a single blood draw, and the possibility of infection by fastidious organisms not easily grown in standard culture media. Therefore, the low positivity rate does not necessarily negate the presence of a clinically significant infection but rather highlights the limitations of conventional culture methods in this specific patient population.

A central and complex finding of our study relates to the role of preoperative prophylactic antimicrobials. The results help to explain the paradoxical finding that antibiotic prophylaxis was associated with a higher risk of PBSI in our multivariable model (aOR 1.68), while simultaneously showing no significant correlation with the aggregate monthly incidence rates. The paradoxical association of prophylaxis with increased risk is best explained by confounding by indication. This phenomenon occurs when clinicians preferentially administer a prophylaxis to patients whom they preoperatively assess as having a higher intrinsic risk of complications (e.g., longer anticipated surgery, greater complexity). In this scenario, the prophylaxis variable becomes a surrogate marker for this underlying, unmeasured risk, rather than being the cause of the infection itself. This lack of a protective effect is then further clarified by our bacteriological findings. Recent guidelines [20] advocate for first- or second-generation cephalosporins for routine prophylaxis. However, our surveillance revealed a high prevalence of E. coli in PBSI cases, and our antibiogram demonstrates that these isolates are highly resistant to precisely those classes of cephalosporins. This microbial resistance pattern effectively neutralizes the intended prophylactic mechanism of action. Therefore, the combined interpretation is that prophylaxis was disproportionately given to high-risk patients (explaining the paradoxical high risk in the model), but it ultimately failed to mitigate that risk due to a mismatch between the chosen antibiotics and the predominant resistant pathogen (explaining the overall lack of a protective association).

A key finding from our subgroup analysis is the markedly elevated infection risk associated with surgeries involving the gastrointestinal and hepatobiliary systems. This increased risk is likely multifactorial, stemming from the convergence of a high bacterial load and increased surgical complexity. The gut’s intrinsic microbial environment serves as a potent source for severe polymicrobial infections [21, 22]. This inherent biological challenge is amplified by the procedural factors of the surgery itself. As demonstrated by Li et al. [23], the prolonged operative times typical of these complex cases extend tissue exposure, thereby increasing the window of opportunity for contamination. Therefore, the higher infection rate likely reflects a synergy between a greater bacterial challenge and a prolonged period of host vulnerability.

Our findings align with previous research indicating that the most extensive gynecologic procedures, such as pelvic exenteration, are associated with the highest rates of infectious complications [24, 25]. However, when possible, adopting minimally invasive techniques—including laparoscopic and robotic approaches—has been shown to reduce postoperative infection risk and shorten recovery time significantly [26, 27]. These findings emphasize the importance of balancing radicality with surgical invasiveness, especially in groups prone to infection.

Our surveillance demonstrated that the occurrence of postoperative infections displayed considerable seasonal variation, with the majority of PF and PBSIs occurring in the third and fourth quarters. This is congruent with prior research that reported increases in the risk of SSIs and Gram-negative bacteremia during warmer and more humid months [28, 29]. The primary hypothesis is likely due to environmental factors. High ambient temperature and humidity can provide an optimal environment for microbes to proliferate on human skin surfaces and the surrounding environment thus leading to an increased bacterial bioburden and contamination risk [29]. Even with air conditioning in hospitals, patients and physicians may act as vectors and load a greater bacterial burden from outdoors. A plausible second factor relates to the potential impact of the seasonal change on host immunity. The physiological stress of adaptation to significant temperature changes may affect a patient’s immune system capacity to resist infection. For these reasons, clinician’s should be aware of the increased seasonal risk and upsurge infection control, especially with respect to skin pre-operative preparation and the maintenance of a sterile clinical environment during this time of the year.

Finally, it is worth noting that many of the risk factors identified in our study can be addressed through comprehensive, multimodal strategies such as Enhanced Recovery After Surgery (ERAS) pathways. ERAS protocols, which optimize nutritional status, encourage early mobilization, and standardize analgesic regimens, are designed to reduce the overall physiological stress of surgery. By accelerating recovery and shortening hospital stays, ERAS programs have the potential to mitigate the risk of various postoperative complications, including infections [30]. Integrating evidence-based infection prevention bundles into a broader ERAS framework may therefore be a powerful approach to improving patient outcomes in gynecological surgery.

Limitations

This study has several limitations. First, our definition of PBSI included a substantial proportion (89.6%) of culture-negative cases, meaning the reported incidence of 1.30% may overestimate true, laboratory-confirmed bacteremia. This inclusive definition was deliberately chosen to reflect clinical reality, where treatment for postoperative fever is often initiated empirically before culture results are available. Our approach thus functions as a form of syndromic surveillance, capturing the full burden of clinically significant infections requiring intervention, not just the laboratory-confirmed subset. We acknowledge, however, that this cohort may include patients with non-infectious systemic inflammatory responses. Future studies using advanced molecular diagnostics, such as PCR or metagenomic sequencing, could better differentiate true infection from sterile inflammation in this challenging population.

Secondly, the study’s retrospective, observational design means that while it can robustly identify associations, it cannot establish causality. Therefore, although we observed a strong inverse correlation between vaginal preparation and PBSI rates, this finding should be considered hypothesis-generating. Prospective, randomized controlled trials (RCTs) are required to definitively establish a causal protective effect and provide the high-level evidence needed to change clinical practice.

Third, while our analysis includes a multivariable logistic regression model to adjust for key confounders, the potential for residual or unmeasured confounding factors inherent in any retrospective design remains. Factors such as detailed nutritional status, intraoperative fluid management, or subtle variations in surgical technique were not available for analysis and could influence the outcomes.

Conclusions

This study identifies critical, modifiable factors associated with a lower risk of postoperative infections in gynecological surgery. The findings suggest that a multi-faceted risk mitigation strategy is beneficial. Specifically, the implementation of preoperative vaginal preparation as part of a quality improvement initiative was associated with a significant reduction in PBSI at the institutional level, and our results underscore the importance of targeted risk-mitigation strategies, such as meticulous bowel preparation, in patients undergoing higher-risk operations involving the gastrointestinal tract. This should be complemented by evidence-based approaches to antibiotic stewardship; our data suggest that empirical therapy for suspected PBSI should be based on local resistance patterns, favoring a tiered approach to antibiotic selection. Finally, while this investigation highlights the importance of vaginal preparation, it also exposes a critical knowledge gap regarding the optimal approach. Therefore, a crucial future research direction is to conduct prospective, comparative-effectiveness trials to establish a definitive, evidence-based standard of care for patients undergoing vaginal preparation for gynecological surgery.

Abbreviations

ACOG

American College of Obstetricians and Gynecologists

aOR

Adjusted Odds Ratio

CI

Confidence Interval

EMR

Electronic Medical Record

ERAS

Enhanced Recovery After Surgery

E. coli

Escherichia coli

ESBL

Extended-Spectrum Beta-Lactamase

GOG-CMA

Gynecological Oncology Group of the Chinese Medical Association

IQR

Interquartile Range

PBSI

Postoperative Bloodstream Infection

PF

Postoperative Fever

RCT

Randomized Controlled Trial

SD

Standard Deviation

SSI

Surgical Site Infection

Author contributions

Yunxin Gong and Ping Luo contributed to the conception and design of the study, data acquisition, statistical analysis, interpretation of results, and drafting of the manuscript. Xuesong Wang was involved in initial data analysis, and critically reviewing the manuscript. Xuerong Zeng and Qiu Lu Xiang participated in data validation and literature review. Youya Liu and FuHua Zhou assisted with data organization and figure preparation. Qian Xiang and Mao Ran contributed to the study design, supervised the entire study, provided critical intellectual guidance, secured funding and was responsible for the final approval of the manuscript. All authors read and approved the final manuscript.

Funding

This study was supported by the National Natural Science Foundation of China (No. 22476032), 2023 Natural Science Foundation of Sichuan Province (No. 2023NSFSC0534) and the Multi-Campus Hospital Systems project (NIHA24YYDQ23) of the National Institute of Hospital Administration of China.

Data availability

The datasets used and analysed during this study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The research strictly adhered to international ethical guidelines for scientific research, including the Declaration of Helsinki. The study protocol was reviewed and approved by the Conjoint Health Research Ethics Board of Sichuan Academy of Medical Sciences & Sichuan Provincial People’s Hospital (Approval Number: 2025 − 144; Date: 28 February 2025). The requirement for informed consent from participants was waived by the same ethics board due to the retrospective nature of this study and in accordance with relevant regulations/guidelines.

Competing interests

The authors declare no competing interests.