Early Epidural Analgesia Tied to Higher Maternal Fever Risk in Retrospective Cohort

Women who received epidural pain relief early in labor—when the cervix was still barely dilated—developed fever during delivery more often than women who waited until labor had progressed further, according to a retrospective cohort study published in Anesthesiology and Perioperative Science. Among 1,744 propensity-matched nulliparous women with term singleton pregnancies and spontaneous labor, maternal fever of at least 38 degrees Celsius occurred in 12.20 percent of those whose epidural catheters were placed at 0 to 2 centimeters of cervical dilation, compared with 7.30 percent of those whose catheters were placed at 3 centimeters or more, a difference the authors report as statistically significant (P = 0.001).

The study, led by researchers at Peking University Shenzhen Hospital, drew on 2,991 women who received epidural analgesia at the single academic center between March 2019 and December 2020. After exclusions for multiparity, preterm delivery, induced labor, missing records, and pre-epidural temperatures of 37.5 degrees Celsius or higher, 967 women formed the early-initiation group and 872 formed the late-initiation group. Propensity score matching retained all late-group patients and selected an equal number from the early group, yielding two cohorts of 872 women each whose measured baseline characteristics—including maternal age, height, weight, body mass index, hypertensive disorders of pregnancy, diabetes, group B streptococcus colonization, and rupture of membranes before analgesia—were balanced with absolute standardized mean differences below 0.1.

Because the study is observational, the authors are careful to frame the result as an association rather than proof of cause. In their conclusion, they state that causal inference cannot be drawn due to the study design and that prospective studies are warranted to confirm the associations and clarify underlying mechanisms. Still, the pattern held up across multiple statistical approaches. Multivariable logistic regression, adjusted for the covariates above, produced an adjusted odds ratio of 0.57 for maternal fever with late initiation (95 percent confidence interval, 0.41 to 0.80; P < 0.001), meaning late initiation was associated with a 43 percent lower odds of fever relative to early initiation in this cohort.

The fever findings extend to sub-threshold elevations as well. In the matched cohort, an axillary temperature of 37.5 degrees Celsius but below 38.0 degrees Celsius occurred in 27.40 percent of the early group versus 17.80 percent of the late group (P < 0.001). Temperatures were recorded with axillary thermometry at two-hour intervals from the onset of labor until delivery, with additional measurements taken when clinically indicated. The authors acknowledge in their limitations that axillary measurement may not optimally represent core body temperature, noting that maternal oral temperature correlates better with direct intrauterine monitoring, and that some fevers could have stemmed from subclinical infections such as chorioamnionitis rather than the epidural itself, since the retrospective design precluded systematic infection screening.

Epidural-related maternal fever is a recognized phenomenon in obstetric anesthesia. The paper’s introduction notes that epidural analgesia has been associated with greater temperature elevation and clinical fever in roughly 15 to 25 percent of recipients compared with other analgesic modalities, with study-to-study variation from 1.6 to 46 percent when fever is defined as a temperature of at least 38 degrees Celsius. The prevailing hypothesis holds that the fever is non-infectious, arising from thermoregulatory dysfunction and an altered inflammatory response rather than bacterial infection, and that the duration of epidural exposure is a key determinant of risk. Early initiation, by this logic, extends the window during which a progressive inflammatory process can manifest clinically as fever.

That exposure-duration logic is central to how the authors interpret their own results. In their cohort, the average duration of labor analgesia was about eight hours in the early group and six hours in the late group, because the epidural was started on maternal request whenever a woman asked for it. By contrast, the authors point to earlier randomized trials—including studies from 2006, 2009, and 2011 that found no significant fever difference between early and late initiation—in which the median duration of analgesia was around or less than six hours. Intrapartum fever typically develops five to six hours after epidural initiation, they note, suggesting those shorter exposures may have limited the observable febrile response. Supporting this view, they cite reports that fever incidence among epidural recipients rose from 5 percent when the block lasted under six hours to 28 percent beyond six hours, and that higher analgesic dosage, prolonged epidural duration, and longer total labor time are risk factors for epidural-related fever.

Mediation analysis in the study lends quantitative weight to the exposure-time explanation. Using the unmatched cohort, the researchers modeled epidural-to-delivery time as a potential mediator between analgesia timing and fever. The total effect of timing on fever was significant (coefficient −0.524, P = 0.01), but the direct effect lost significance once the mediator entered the model (coefficient −0.275, P = 0.11), while the indirect effect was significant (bootstrap 95 percent confidence interval, −0.36 to −0.15). In plain terms, longer epidural-to-delivery time partially explains the increased fever risk observed with early initiation, though the authors stress that the observational design precludes definitive confirmation of the causal assumptions and that the mediation analysis should be read as exploratory.

Fever was not the only outcome that differed. Late initiation was also associated, in adjusted analyses, with lower odds of uterine inertia (adjusted odds ratio 0.68, 95 percent confidence interval 0.48 to 0.96), oxytocin use (0.73, 0.60 to 0.88), cesarean delivery (0.51, 0.38 to 0.69), and episiotomy (0.41, 0.32 to 0.51), along with lower log-transformed blood loss in the first 24 hours after delivery (adjusted coefficient −0.12, 95 percent confidence interval −0.17 to −0.07). In the matched cohort, conversion to cesarean delivery occurred in 16.6 percent of the early group versus 9.4 percent of the late group, uterine inertia in 9.9 percent versus 6.9 percent, oxytocin administration in 55.7 percent versus 47.7 percent, and episiotomy in 33.7 percent versus 17.1 percent. Mean blood loss within 24 hours of delivery was 282.43 milliliters in the early group versus 247.93 milliliters in the late group. Neonatal outcomes told a different story: median 1- and 5-minute Apgar scores and the rates of neonatal asphyxia and hyperbilirubinemia were comparable between groups.

The cesarean finding sits uneasily with prior evidence, and the authors confront that tension directly. Current guidelines hold that maternal request is a sufficient indication for labor pain relief, a position grounded in randomized trials showing that early-labor neuraxial analgesia does not increase cesarean delivery rates compared with later initiation—though fever was not a primary outcome in those trials. The authors attribute their divergent cesarean result to several factors: observational designs are susceptible to residual confounding by indication, since women requesting very early analgesia may differ in pain perception, fetal position, or underlying labor dystocia in ways that predispose to both early request and cesarean delivery; the substantially longer analgesia duration in their early group may mediate time-dependent complications; and their institution’s low overall cesarean rate of 13.6 percent may reflect more active intervention patterns that lower the threshold for proceeding to cesarean when complications arise.

Methodological safeguards and their limits frame the study’s conclusions. The authors calculated an E-value of 2.92 for the primary outcome, meaning an unmeasured confounder would need to be associated with both the exposure and the outcome by a risk ratio of at least 2.92 to explain away the observed association—evidence, they argue, that residual confounding is unlikely to account fully for the findings. Sensitivity analyses using multivariable adjustment in the unmatched cohort, an alternative timing cutoff of 0 to 3 versus at least 4 centimeters, and inverse probability weighting all yielded results consistent with the primary analysis, and the authors note that cervical examinations are inherently subjective, which may introduce misclassification bias. They also acknowledge that potentially important intrapartum variables—detailed analgesic drug consumption, number of vaginal examinations, total local anesthetic dose, and intravenous fluid volume—were unavailable in the retrospective dataset. The single-center design, with its uniform anesthesia protocols and drug concentrations, strengthens internal comparability but may limit generalizability to other clinical settings. The authors’ bottom line is measured: early initiation of epidural analgesia at 0 to 2 centimeters of cervical dilation was associated with higher rates of intrapartum fever, cesarean delivery, and several maternal interventions in this population, the associations persisted after adjustment for measured confounders, and randomized controlled trials are needed to determine whether analgesia timing independently influences maternal and neonatal outcomes.

Subject of Research: The association between the timing of labor epidural analgesia initiation and maternal fever and other maternal outcomes in nulliparous women.

Article Title: Association between the timing of labor epidural analgesia and maternal fever in nulliparous women: a retrospective cohort study

Article References: Zhang, Y., Guo, Y., Shan, Y., Zhou, W., Pan, Y., Xu, Q., Ouyang, Y., & Luo, T. (2026). Association between the timing of labor epidural analgesia and maternal fever in nulliparous women: a retrospective cohort study. Anesthesiology and Perioperative Science, 4(1), Article 36. https://doi.org/10.1007/s44254-026-00182-1

Image Credits: AI Generated

DOI: 10.1007/s44254-026-00182-1

Keywords: epidural analgesia, maternal fever, labor pain relief, nulliparous women, retrospective cohort study, propensity score matching, cervical dilation, intrapartum fever, cesarean delivery, epidural-to-delivery time, obstetric anesthesia, observational study