In recent observational studies, epidural analgesia during labor at patient request has been associated with maternal fever. The authors report a secondary analysis of fever in women who were randomized to receive either epidural or patient-controlled intravenous analgesia during labor.
Maternal tympanic temperature was measured during spontaneous labor in 715 women at term who were randomized to either epidural analgesia with bupivacaine and fentanyl or to patient-controlled intravenous analgesia with meperidine. Intent-to-treat analysis of women with fever (temperature > or = 38.0 degrees C) versus those without was performed using Student t test and Fisher exact test to determine statistical significance (P < 0.05).
Epidural analgesia was associated with maternal fever (odds ratio = 4.0; 95% confidence interval = 2.0-7.7), as was nulliparity (odds ratio = 4.1; 95% confidence interval = 1.8-9.1) and labor longer than 12 h (odds ratio = 5.4; 95% confidence interval = 2.9-9.9). These factors were all independent variables for maternal fever when analyzed using logistic regression.
Epidural analgesia is associated with maternal fever. However, nulliparity and dysfunctional labor are also significant cofactors in the fever attributed to epidural analgesia.
This article is accompanied by an Editorial View. Please see: Camann W: Intrapartum epidural analgesia and neonatal sepsis evaluations: A casual or causal association? Anesthesiology 1999; 90:1250–2.
SOME investigators have reported that epidural analgesia during labor is an independent risk factor for the development of maternal fever. [1,2] Others have indicated that epidural analgesia may induce subclinical levels of maternal hyperthermia. [3–6] However, all of these reports have been based on observational studies in which women received epidural analgesia during labor at their request. Therefore, there is the possibility that the association between maternal fever and epidural analgesia occurred as a result of selection bias for more difficult, dysfunctional labor, which in itself is an independent risk factor for the development of maternal fever.
We recently reported a randomized trial of the effects of epidural analgesia on the outcomes of labor in a carefully defined and homogenous group of parturients.  Specifically, the women studied all had low-risk term pregnancies and were admitted in the active phase of spontaneous labor. Women were randomized to receive either epidural analgesia or patient-controlled intravenous meperidine analgesia (PCIA). There were minimal cross-overs from one study arm to the other. We now report a secondary analysis of fever in these women.
Methods and Materials
Our sample for analysis included 715 women enrolled in the randomized trial of epidural versus intravenous analgesia during labor, conducted at Parkland Memorial Hospital from June 1995 through February 1996. After approval by the Institutional Review Board of the University of Texas Southwestern Medical Center at Dallas and informed consent, healthy women with singleton cephalic gestations at term and presenting in spontaneous active labor were randomly assigned to receive either epidural analgesia or PCIA during labor from a randomization sequence that was computer-derived in blocks of 20 subjects and then placed in sealed opaque envelopes.
All staff followed a written procedural manual that prescribed the intrapartum management of women admitted to the low-risk labor unit. Our labor management approach encourages amniotomy in active labor when the fetal head is applied to the cervix. Internal electronic fetal monitoring was used in those women with meconium-stained amniotic fluid, known fetal heart rate decelerations, or inadequate progress of labor. Pelvic examinations were performed approximately every 2 h to evaluate the progress of labor. Oxytocin augmentation of labor was undertaken when the rate of cervical dilation was less than 1 cm/h, and a hypotonic contraction pattern was measured using intrauterine pressure transducers.
After an intravenous bolus of 500 ml of Ringer's lactate solution, epidural analgesia was initiated at the first request for pain relief with increments of 0.25% bupivacaine and maintained with a continuous infusion of 0.125% bupivacaine with fentanyl, 2 [micro sign]g/ml. Women randomized to PCIA were given 50 mg of meperidine with 25 mg of promethazine hydrochloride intravenously at their first request for pain relief, followed by 10 mg of meperidine every 10 min for the first hour using a patient-controlled pump. Thereafter, 15 mg of meperidine was given every 10 min as needed. A more detailed description of the study methodology has been published elsewhere. 
Maternal tympanic temperature was measured during the course of labor using a Genius thermometer (Sherwood Medical, St. Louis, MO). Intrapartum fever was defined as a maternal temperature of 38.0 [degrees sign]C or greater during the course of labor. Neonates born to women with fever during labor were commonly evaluated for possible sepsis, which included a complete blood count and blood cultures, after which they were given antibiotics for 48 h. Indications for a sepsis evaluation in neonates born to women without fever in labor included temperature instability (hyperthermia or hypothermia), tachypnea, dusky spells, lethargy, or hypoglycemia. Neonates with suspected sepsis were empirically treated with antibiotics for 48 h. This was the standard of practice during the interval of the study and was not part of the trial protocol
Analysis of data was performed using an intent-to-treat approach, and tests of significance were two-tailed. The data were analyzed using SAS statistical software (SAS Institute Inc., Cary, NC). Statistical significance (P < 0.05) was determined using unpaired Student t test for the comparison of continuous data and Fisher exact test for the comparison of categorical data. Logistic regression analysis was performed to adjust for potentially confounding variables.
A total of 715 women were randomized in this investigation. Of 358 women who were allocated to receive epidural analgesia, 243 women completed the study as allocated, whereas 115 (32%) women did not comply with the epidural analgesia protocol. Of these 115 women, 78 progressed rapidly to delivery before epidural analgesia could be initiated, and 37 women refused epidural analgesia. Of 357 women who were allocated to receive PCIA, 259 completed the study as allocated, and 98 (28%) women did not comply with the protocol. Of these 98 women, 73 progressed rapidly to delivery and did not receive any analgesia, 20 refused any analgesia, and only 5 women who received meperidine as randomized later crossed-over and received epidural analgesia because of inadequate pain relief.
Sixty-eight of the 715 women (10%) developed fever during labor (Figure 1). As shown in Figure 1, 54 (15%) women randomized to epidural analgesia developed fever during labor compared with 14 (4%) who were given PCIA (P < 0.001). When stratified according to parity however, fever occurred more often with epidural analgesia in nulliparous women (24% vs. 5%; P < 0.001) but not in parous women (4% vs. 3%; P = not significant; see Figure 1).
There were no significant differences in maternal demographic characteristics, including age, weight, and height, in women who developed fever compared with those who did not. Black women had fever less often (12% vs. 24%; P = 0.03). As shown in Table 1, intrapartum factors significantly associated with maternal fever included epidural analgesia, nulliparity, prolonged labor > 12 h, internal fetal monitoring, and oxytocin augmentation of labor.
To adjust for potentially confounding variables we performed a stepwise logistic regression analysis of all factors associated with maternal fever. Epidural analgesia, nulliparity, and prolonged labor were independently associated with maternal fever. The odds ratio (95% CI) was 4.0 (2.0–7.7) for epidural analgesia, 4.1 (1.8–9.1) for nulliparity, and 5.4 (2.9–9.9) for prolonged labor.
Only one infant was born with a 5-min Apgar score of 3 or less, and the mother did not have fever during labor. Two infants, one born to a woman with fever and one without fever, had a umbilical artery blood pH < 7.0. As shown in Figure 2, more neonates born to women with fever were evaluated and treated for possible sepsis. In women without fever, there was no association between the type of analgesia and neonatal sepsis evaluations and antibiotic treatment (see Figure 2). Sepsis, defined as a positive blood culture, was not diagnosed in any of the infants, and there were no neonatal deaths in the study cohort.
In this secondary analysis, epidural analgesia was associated with maternal fever. However, nulliparity and prolonged labor were also independent risk factors for maternal fever. Epidural analgesia was associated with fever only in nulliparous women, thereby implicating nulliparity as a factor in the fever related to epidural analgesia. Labor was significantly longer in women with fever regardless of parity or the type of analgesia, thereby implicating prolongation of labor as a factor in the development of fever.
Several investigators have reported an association between epidural analgesia and either subclinical maternal hyperthermia [3–6] or overt maternal fever. [1,2] However, all of these reports have been based on uncontrolled observational studies. We analyzed a carefully defined and homogenous group of parturients who were randomized to receive either epidural analgesia or PCIA during labor. Such a study design minimizes selection bias and, in our opinion, is most appropriate to investigating the relationship, if any, of epidural analgesia to maternal fever.
The metabolic expenditure of labor increases maternal temperature, nulliparous women experiencing a greater rise in temperature than multiparous women.  Thermoregulatory changes with epidural analgesia may also exert a hyperthermic effect. Evaporative heat loss through hyperventilation is decreased as a result of pain relief.  Sweat gland paralysis resulting from neural blockade decreases sweating and evaporative heat loss. [10,11] Reactive vasoconstriction in the upper body decreases convective heat loss.  Thermal cues are modified such that cold sensation persists, whereas warm sensation is blocked, [11,12] leading to inappropriate shivering [11,12] with associated hyperthermia.  Therefore, it is possible that nulliparous women who experience prolonged labor and who receive epidural analgesia generate heat that they are unable to dissipate, thereby rendering them hyperthermic.
In our study more neonates born to women with fever were evaluated and treated for possible sepsis. Fever during labor is a well-accepted marker for maternal infection. In women with chorioamnionitis fever is considered to be the most significant clinical finding. It is difficult however to clearly determine whether fever during labor is of infectious origin. Given the risk of neonatal infection with chorioamnionitis, infants born to women with fever are frequently evaluated and treated for possible sepsis in the event that fever is due to maternal infection. Although sepsis was suspected in neonates born to women with fever, sepsis was not diagnosed in any of the infants in our study.
Evaluation of neonates for possible sepsis has been associated with epidural analgesia in the absence of maternal fever. Lieberman et al. found a threefold increase in the rate of sepsis evaluations in infants born to afebrile women who had received epidural analgesia during labor.  They concluded that the use of epidural analgesia during labor is strongly associated with the occurrence of neonatal sepsis evaluations.  However, many of these women who received epidural analgesia at their request had prolonged labor and prolonged rupture of membranes, risk factors for maternal and subsequent neonatal infection that may have prompted infant evaluations in the absence of maternal fever. In contrast, our results indicate that in the absence of maternal fever, epidural analgesia during labor has no bearing on the need for such neonatal management and therefore should not be considered a predictor per se for neonatal sepsis evaluations. We attribute this finding to the minimization of ascertainment bias as a result of randomization of analgesia.
In conclusion, epidural analgesia is associated with maternal fever. However, this association becomes apparent in a small subset of women who are usually nulliparous and who experience prolonged labor. In other words, epidural analgesia is not solely responsible for maternal fever. Nulliparity and dysfunctional labor, both circumstances wherein epidural analgesia is often used, also account for maternal fever. Similarly in the absence of maternal fever, epidural analgesia is not associated with neonatal sepsis evaluations. Therefore, we believe that epidural analgesia should not be avoided based on the fear that such analgesia induces fever or because of concerns about unnecessary neonatal evaluations for possible sepsis.