Labor epidural analgesia has been associated with increased intrapartum temperature in previous randomized controlled trials and retrospective cohort studies.1–6 However, in these analyses of the effects of labor epidural analgesia, study methods have often been hindered by patient crossover in randomized trials and self-selection bias in cohort studies. A natural experiment, examining an outcome of interest in a population before and after an intervention has been made available as an alternative scientific method to examine the effects of labor epidural analgesia without such crossover or selection bias effects. Although several previous natural experiments have noted an increased incidence of maternal intrapartum fevers associated with epidural analgesia, none have systematically examined the broader issue of maternal hyperthermia and associated adverse maternal and neonatal effects.7,8 The purpose of our investigation was to quantify the effect of introducing an on-request labor epidural analgesia on the incidence of maternal intrapartum febrile morbidity and the resultant impact on neonatal care using a natural experiment analytic method.
MATERIALS AND METHODS
Tripler Army Medical Center is a large tertiary care facility located on the island of Oahu and is the sole military facility providing in-patient obstetric care to military personnel and their family members located in the Hawaiian Islands. There is a resident training program in obstetrics and gynecology, and approximately 3000 deliveries per year are performed at this institution. The patient population is relatively homogenous in socioeconomic status, with housing provided for all families, regular income, and direct access to health care—including prenatal and obstetric care—which is provided without any direct cost to the patient. The racial composition of the population served is approximately 60% white, 17% Asian, 14% black, and 8% Hispanic, with the remainder being of other ethnicity. The majority of prenatal and intrapartum care is provided by residents in training under the supervision of faculty physicians and midwives, functioning as dedicated teams who provide coverage for 12- or 24-hour shifts. These individuals have no other scheduled care responsibilities during their shifts. All labor and delivery unit rooms have continuous temperature monitoring and conditioning equipment to maintain the ambient room temperature at 75–79F.
Before October 1993, epidural analgesia in laboring women was only available to patients deemed to have a medical indication such as severe preeclampsia or severe cardiovascular disease. A Department of Defense mandate became effective in October 1993 with subsequent funds also becoming available for the provision of round-the-clock on-demand labor epidural analgesia capability at military centers. At our facility, this service became fully functional in January 1994 with a certified registered nurse anesthetist or anesthesiologist assigned to labor and delivery 24 hours a day. All epidurals were placed and maintained by these nurse anesthetists or anesthesiologists.
All nulliparous women with singleton gestations in vertex presentation and spontaneous onset of labor at 37–41 weeks, who gave birth from October 1, 1992 to September 30, 1993, and from October 1, 1995 to September 30, 1996 were selected for study inclusion, and a stepwise exclusion process, as detailed in Table 1, was then used to select groups of uncomplicated term parturients. The selection criteria were developed to minimize differences between the patient populations and practice patterns between the two study periods. Inpatient medical records of all patients meeting the above criteria during the respective study periods were then obtained and reviewed. These two study groups will be called “Before” and “After” groups, respectively, hereafter. Information regarding the patient's sociodemographic characteristics, prenatal care, admission assessment, course of labor and delivery, analgesia use, postpartum complications, and neonatal outcome were extracted from each patient's record. The study protocol was approved by the Human Use Committee at Tripler Army Medical Center. Investigators adhered to the policies for protection of human subjects as prescribed by US Military regulation.
The specific techniques and dosing used for labor epidural analgesia were similar between the two study periods. The general policy during the study period was to initiate epidural analgesia when the patient was at least 4-cm dilated, except in women with a severe cardiovascular disorder, such as valvular heart disease or severe preeclampsia. After placement, a test dose of 3 mL 1.5% lidocaine with 1:200,000 epinephrine was injected, followed by an 8-mL bolus of 0.125% bupivacaine with 100 μg of fentanyl. The catheters were then infused with 0.125% bupivacaine with 2–4 μg/mL of fentanyl begun at 10 mL per hour. Thereafter, women were assessed at hourly intervals and the infusion rate was adjusted to maintain a T10 sensory level with minimal motor block. Infusions were generally continued through the second stage of labor.
Recording of cervical assessment was standardized with dilation measured in 0.5-cm increments, effacement in percent of cervical length, and station of the presenting part divided into thirds above and below the ischial spines (−3 to +3). Maternal temperature was generally assessed every 2–4 hours with the use of an infrared tympanic probe (Genius Infrared Tympanic Thermometer 3000A, Sherwood Medical, St. Louis, MO). Intraoperative temperature values in women undergoing cesarean delivery were not included in the analysis. Chorioamnionitis was diagnosed in the woman with an intrapartum temperature of at least 100.4F when two of the following three criteria were present, or of at least 99.5F on more than two occasions if all three of the following were present: 1) fetal tachycardia, defined as baseline heart rate exceeding 160 beats per minute; 2) maternal tachycardia, defined as maternal baseline heart rate exceeding 100 beats per minute; and 3) uterine tenderness assessed during the interval between uterine contractions. In the absence of the above, intrapartum maternal temperatures of at least 100.4F were considered to be a fever of undetermined origin. Use of broad-spectrum antimicrobial therapy was generally prescribed in the setting of clinical chorioamnionitis unless delivery appeared imminent. Use of antimicrobial therapy was generally prescribed in the setting of intrapartum fever of undetermined origin at the attending health care provider's discretion. Postpartum endometritis was diagnosed based upon persistent postpartum fever (greater than or equal to 100.4F) with associated lower abdominal or uterine tenderness.
Neonatal care was based on clinical findings and intrapartum factors. Fellowship-trained neonatologists directed all neonatal care. There were no standard protocols requiring specific laboratory or radiographic assessment in neonates delivered to women with intrapartum fevers or antibiotic therapy, and decisions regarding laboratory analysis and antibiotic therapy were decided on a case-by-case basis during both study periods. Antibiotic therapy was generally reserved for infants with symptoms suggestive of early-onset sepsis.
Comparison of categorical variables was performed with χ2 analysis with Yates' correction or Fischer exact test, as appropriate. Mantel-Haenszel χ2 was used for comparison of categorical variables with multiple factor levels. Continuous variables were compared with the unpaired t test. Multivariable analysis was performed with logistic regression control for chronological variables. A P value of less than .05 was considered statistically significant for all analyses. Relative risk (RR) and 95% confidence intervals (CI) were calculated where appropriate. Statistical analysis was performed with Epi-Info 6.0 (Centers for Disease Control and Prevention, Atlanta, GA) and SAS 6.12 (SAS Institute Inc., Cary, NC).
The methods for study population selection are described in Table 1. Data were initially analyzed for 507 women who fulfilled eligibility criteria in the study period before the availability of on-request labor epidural analgesia and 581 women delivered during the period of on-request labor epidural analgesia. Maternal temperature of at least 99.5F was detected upon admission in nine women in the Before group and nine women in the After group. These women were excluded from further analysis (Table 1). Demographic and labor admission characteristics are listed in Table 2. Labor epidural analgesia was used by five of 498 (1.0%) of nulliparas delivered before the availability of on-request epidural analgesia and 475 of 572 (83.0%) of nulliparous parturients after on-request epidural analgesia was available (P < .001). Labor characteristics and outcome, including the mean maximal intrapartum temperatures are listed in Table 3. There were no statistically significant differences in the number of vaginal examinations or the duration of membrane rupture between the two groups. Figure 1 is a graphic portrayal of the proportion of undelivered women with a maximal intrapartum temperature of at least 100.4F relative to the time since admission. This temporal reference point was used given the inherent vagaries of establishing the onset of labor for all patients. The percentage of each population remaining undelivered over time since admission was statistically similar between the two periods (data not shown). The relative incidence of maternal febrile morbidity is presented in Table 4. The mean body temperature was examined among women who used epidural analgesia. The mean body temperature rose progressively right after epidural placement and reached a plateau after about 10 hours after placement (Figure 2).
Further analysis was performed solely on the After group by substratifying patients by labor epidural analgesia use. The mean maximal intrapartum temperature was higher in women who had labor epidural analgesia compared with those who did not have epidural analgesia (99.1 ± 1.1F versus 98.4 ± 0.8F, P < .001). A maximal temperature of at least 99.5F was detected in 142 of 480 (30%) women with epidural analgesia compared with eight of 92 (8.7%) parturients who did not use epidural analgesia (RR = 3.7, 95% CI 1.9, 7.5, P < .001). A maximal temperature of at least 100.4F was detected in 62 of 480 (12.9%) women with epidural analgesia and in one of 92 women (1.1%) women without epidural analgesia (RR = 11.3, 95% CI 1.6, 79.4, P < .001). Thus, when labor epidural analgesia was available, maternal febrile morbidity predominantly occurred in women who used epidural analgesia.
Multivariable analysis was performed using logistic regression to control for duration of labor, duration of rupture of membranes, number of vaginal exams, body temperature on admission, and maternal prepregnancy weight. Using this model, on-request labor epidural analgesia was associated with an intrapartum temperature of at least 99.5F (RR = 3.0, 95% CI 2.3, 3.6, P < .001) and intrapartum temperature of at least 100.4F (RR = 20.2, 95% CI 7.0, 86.0, P < .001).
Comparison of the frequency of neonatal laboratory analysis and neonatal length-of-stay is reported in Table 5. There was a statistically significant increase in the frequency of screening complete blood count and blood cultures in the After period, as well as an increase in the median length of stay. There were no cases of culture-proven neonatal sepsis in either group. The median length of stay remained the same between the two study periods; however, there was a statistically significant increase in the length of stay in the After period. This was largely due to a combination of a slight increase in postnatal antibiotic therapy for presumed sepsis and surveillance of neonates delivered to women colonized with group B streptococcus during the later study period.
Larue et al initially proposed a link between labor epidural analgesia and maternal hyperthermia in 1987.9 This association was strengthened by Fusi et al in 1989 who found women receiving labor epidural analgesia to have an increased mean intrapartum temperature when compared to a control group of women receiving intravenous narcotic analgesia.10 Subsequently, numerous randomized, controlled trials and retrospective cohort studies have demonstrated an increase in intrapartum fevers in women with intrapartum epidural analgesia compared with women who did not receive epidural analgesia; however, such findings are somewhat limited by the inability to control for other important risk factors for intrapartum fever, to include duration of labor, duration of rupture of membranes, and the number of vaginal examinations.1–6 However, some investigators have controlled for such factors through multivariate modeling, with results that continue to suggest a link between epidural analgesia and intrapartum hyperthermia.2,11 The estimated incidence of maternal temperature of at least 100.4F associated with epidural analgesia is 0–30%.12
There has been substantial controversy in the scientific literature regarding the potential impact of epidural analgesia on the course and outcome of labor. Although the randomized, controlled trial is typically considered the most scientifically rigorous means of study, interpretation of the results of randomized, controlled trials of labor epidural analgesia have been problematic; most investigations have reported a substantial number of patients crossing over between study groups.3,4 Retrospective cohort analysis comparing women who receive epidural analgesia with a concurrent cohort of women who elect not to receive epidural analgesia are confounded by a predilection of women destined to have relatively longer labors and higher operative delivery rates selecting epidural analgesia for intrapartum pain management.1,5,10,13,14 Thus, both approaches are hampered by potential bias resulting from patient self-selection. Natural experiments, such as this one, examine the potential impact of an intervention by examining the multiple variables of interest in populations before and after the introduction of the intervention. This method has been used, in the case of labor epidural analgesia, as a means of attempting to overcome some of the limitations of randomized trials and cohort analysis, by eliminating the potential for study group crossover and patient self-selection. The limitation of a natural experiment is the potential for the introduction of other factors between the two outcome periods that could alter the primary outcome variable.
In this analysis of febrile morbidity associated with labor epidural analgesia, we have attempted to examine our two study groups as carefully as possible for the presence of potential alterations in practice patterns or interventions that could have impacted the incidence of intrapartum maternal hyperthermia and have been unable to discern any substantive change between the two study groups, aside from the dramatic increase in the use of intrapartum epidural analgesia. Further attempts to control for potential confounders through the use of multivariate modeling demonstrate that the liberal use of on-request labor epidural analgesia in a population of nulliparas in spontaneous labor results in a 20-fold increase in the incidence of intrapartum fever of at least 100.4F.
The precise etiology of the maternal hyperthermia related to epidural analgesia remains unknown. Given the dramatic rise of intrapartum fever in our population without any evidence of altered intrapartum management patterns, or associated surrogate indicators of ascending genital tract infection, such as neonatal sepsis or postpartum maternal infections, we think that it is highly unlikely that an infectious origin can be implicated in the majority of cases of intrapartum fever in women with epidural analgesia. It is therefore likely that the rise in maternal temperature associated with epidural analgesia is due to an alteration in maternal thermoregulatory physiology. Proposed theories include an alteration in the hypothalamic thermoregulatory set point, similar to what occurs with centrally mediated fever, impairment of peripheral thermoreceptor input to the central nervous system with selective blockage of warm stimuli, or an imbalance between heat production and heat loss.12,13 The latter theory is attractive, as effective labor pain relief with lumbar epidural analgesia is known to reduce maternal hyperventilation while impairing sweat production in the lower half of the body, factors that impair heat loss.15 This occurs simultaneously with an increased potential for heat production through maternal shivering, occurring in up to 66% of laboring women with epidural analgesia.12 Furthermore, intrapartum fevers have been noted more commonly in women who shiver after epidural dosing.16 However, an infectious etiology can never be completely discounted. Although it appears that epidural analgesia does not cause protracted labor, it is clear that the woman who requests labor epidural analgesia is more likely to have a protracted labor relative to the parturient who does not use epidural analgesia. It is therefore likely that the etiology of intrapartum fevers associated with epidural analgesia is multifactorial in origin, creating a clinically challenging situation.
The clinical significance of the association between epidural analgesia and intrapartum fever likely has much more to do with the response of the obstetric and neonatal providers to the fever than the maternal and uteroplacental physiologic alterations associated with the increase in maternal core temperature. Previous investigations have suggested that the woman with intrapartum fever is more likely to undergo operative abdominal or vaginal delivery, and her neonate is more likely to undergo evaluation for neonatal sepsis and have antibiotic therapy initiated awaiting culture results.1,17 In our population, neonates delivered during the period of on-request epidural use were more likely to have a screening blood count performed, but there was no statistically significant increase in antibiotic use and only a modest increase in the length of stay. This is likely due to neonatal practice patterns with sepsis suspected on clinical grounds rather than solely on maternal intrapartum fever and intrapartum antibiotic therapy. As expected, the incidence of proven neonatal sepsis in our population and other series of epidural-associated maternal fevers is extremely low, much lower than would be expected had the origin of the maternal fever been infectious in origin. An enhanced awareness of the association of epidural analgesia with intrapartum maternal hyperthermia may help health care providers avoid unnecessary intervention; however, there is currently no reliable means of accurately differentiating the woman with an infectious-mediated fever from the parturient with hyperthermia resulting from physiologic alterations induced by epidural analgesia. Hopefully, future research will either provide an effective prevention for the epidural-mediated hyperthermia or a reliable means of identifying the parturient with an underlying infection whose neonate is at risk for serious morbidity and mortality.
1. Lieberman E, Lang JM, Frigoletto F Jr, Richardson DK, Ringer SA, Cohen A. Epidural analgesia, intrapartum fever, and neonatal sepsis evaluation. Pediatrics 1997;99:415–9.
2. Philip J, Alexander JM, Sharma SK, Leveno KJ, McIntire DD, Wiley J. Epidural analgesia during labor and maternal fever. Anesthesiology 1999;90:1271–5.
3. Sharma SK, Sidawi JE, Ramin SM, Lucas MJ, Leveno KJ, Cunningham FG. Cesarean delivery: A randomized trial of epidural versus patient-controlled meperidine analgesia during labor. Anesthesiology 1997;87:487–94.
4. Ramin SM, Gambling DR, Lucas MJ, Sharma SK, Sidawi JE, Leveno KJ. Randomized trial of epidural versus intravenous analgesia during labor. Obstet Gynecol 1995;86:783–9.
5. Dashe JS, Rogers BB, McIntire DD, Leveno KJ. Epidural analgesia and intrapartum fever: Placental findings. Obstet Gynecol 1999;93:341–4.
6. Herbst A, Wolner-Hanssen P, Ingemarsson I. Risk factors for fever in labor. Obstet Gynecol 1995;86:790–4.
7. Fogel ST, Shyken JM, Leighton BL Mormal JS, Smeltzer JS. Epidural labor analgesia and the incidence of cesarean delivery for dystocia. Anesth Analg 1998;87:119–23.
8. Lyon DS, Knuckles G, Whiteaker E, Salgado S. The effect of instituting an elective labor epidural program on the operative delivery rate. Obstet Gynecol 1997;90:135–41.
9. Larue F, Labaille Th, Mazoit X, Mezzaroba PH, Benlabed M, Benhamou D. Anesthésie péridurlae et surveillance de la temperature au cours du travail. Ann Fr Anesth Réanim 1987;6S:R163.
10. Fusi L, Steer PJ, Maresh MJA, Beard RW. Maternal pyrexia associated with the use of epidural analgesia in labor. Lancet 1989;1250–2.
11. Lieberman E, Cohen A, Lang J, Frigoletto F, Goetzl L. Maternal temperature elevation as a risk factor for cesarean delivery and assisted vaginal delivery. Am J Public Health 1999;89:506–10.
12. Mercier FJ, Benhamou D. Hyperthermia related to epidural analgesia during labor. Int J Obstet Anesth 1997;6:19–24.
13. Camann WR, Hortvet LA, Hughes N, Bader AM, Datta S. Maternal temperature regulation during extradural analgesia for labour. Br J Anaesth 1991;67:565–8.
14. Vinson DC, Thomas R, Kiser T. Association between epidural analgesia during labor and fever. J Fam Pract 1993;36:617–22.
15. Goodlin RC, Chapin JW. Determinants of maternal temperature during labor. Am J Obstet Gynecol 1982;143:97–101.
16. Gleeson NC, Nolan KM, Ford MRW. Temperature, labour, and epidural analgesia. Lancet 1989;11:861–2.
© 2001 by The American College of Obstetricians and Gynecologists.
17. Traynor JD, Dooley SL, Seyb S, Wong CA, Shadron A. Is the management of epidural analgesia associated with an increased risk of cesarean delivery? Am J Obstet Gynecol 2000;182:1058–62.