Intrapartum fever is associated with neonatal morbidity and infection-related mortality.1 Obstetricians may perceive competing risks when contemplating additional cervical examinations, weighing the need for examinations to optimize labor management against the theoretical risk of infection incurred with increased number of examinations. This dilemma can be magnified in the tertiary care teaching setting, where the potential for additional examinations is enhanced.
Publications often suggest that multiple cervical examinations in labor is a risk factor for fever and intraamniotic infection.2–5 In fact, very little scientific evidence is published to guide clinicians, and what exists must often be extrapolated from populations of patients with preterm or premature rupture of membranes. Furthermore, existing publications do not use time-dependent analyses to estimate the risk between number of cervical examinations and fever, leaving the question of whether the risk factor for a particular febrile patient was the length of labor, the number of examinations, both, or neither.
We undertook this study to estimate the risk of number of cervical examinations with maternal fever in labor using time-dependent analysis to account for both length of labor and length of membrane rupture.
MATERIALS AND METHODS
This was a retrospective study of the first 2,400 women of a 4-year cohort of all consecutive term births at a large, tertiary care teaching medical center in the Midwest. The parent study aimed to estimate the association between electronic fetal heart rate patterns in the second stage of labor and acidemia at birth. Women were included in the parent cohort if they were at or beyond 37 weeks of gestation and reached the second stage of labor. For this study, the charts of women who labored at term but underwent cesarean delivery in the first stage were also extracted to assure that women with fever were not delivered before reaching complete dilation, creating selection bias. Eligibility criteria were singleton gestation, cephalic presentation, and no known fetal anomalies. Women were excluded if they were febrile on admission. Gestational age was determined by the first day of the woman's last menstrual period. If the estimated due date from the last menstrual period was consistent (±7 days in the first trimester, ±14 days in the second trimester, and ±21 days in the third trimester) with that obtained from growth measurements at the first ultrasonogram, the due date was not changed. If the last menstrual period and first ultrasonographic due dates were not consistent, the ultrasonogram-obtained due date was used to define gestational age. This study was conducted with the approval of the Washington University School of Medicine Human Research Protection Office.
The primary outcome, maternal fever, was defined as at least one recorded temperature greater than or equal to 38.0°C during labor, delivery, or the first 6 hours postpartum. The primary exposure was the number of digital cervical examinations as documented in the medical record. The digital examination and the time it was performed were extracted from both the nursing and physician progress notes. The institutional culture at our teaching center dictates that women are primarily examined by residents and nurses and all examinations are documented; medical students are generally not permitted to perform examinations on laboring patients.
Extensive labor and delivery data were collected from the electronic medical record, including obstetric and medical history, antepartum complications, medication exposures, and admitting diagnoses. Details regarding labor and labor progress were collected in addition to the cervical examinations and timing to allow reconstruction of labor curves, including the time of membrane rupture, protraction disorders, and mode of delivery. Specific to this analysis, maternal vital signs, use of internal heart rate or contraction monitors, and regional anesthesia use were recorded. We also collected neonatal and maternal birth outcomes.
The rate of fever within the cohort was estimated. To assure that there was no selection bias introduced by those who underwent cesarean delivery in the first stage, the rate of intrapartum fever was also estimated among those excluded for a first-stage cesarean delivery. Women who developed an intrapartum fever were compared with those who did not with respect to baseline and labor characteristics. The number of cervical examinations and number of hours from first cervical examination until delivery per patient within the cohort, and by presence or absence of fever, were described. [Chi] square test or Fisher's exact test was used for categorical variables. Continuous variables were examined for normality using the Shapiro-Wilk test; normally distributed continuous variables were compared using the Student's t test and those nonnormally distributed were compared using the Mann-Whitney U test. Relative risks and 95% confidence intervals were generated. Stratified analyses were used to identify potentially confounding effects. Imputed values were not used because data were nearly complete; 0–1.8% of fields were missing for any given variable.
Because women who are in labor for a longer period often have more examinations and thus time rather than the examinations themselves may be the risk factor for peripartum fever, time-to-event analyses were used, which allowed us to estimate the effect of number of cervical examinations on the risk of fever while accounting for the length of labor. Time zero was defined as the time of first examination. The number of cervical examinations was considered categorically, assigning one to three examinations as the reference with incremental increases by two examinations. Cox proportional hazard regression was used to model the effect of number of cervical examinations on the risk of fever while adjusting for potentially confounding factors identified in the stratified analyses and those that have been historically described to be associated with the risk of maternal intrapartum fever. Nonsignificant potentially confounding factors, including parity, use of internal monitors, and regional anesthesia, were removed in a backward, stepwise fashion and were not included in the final models. The proportional hazards assumption was tested using the global test of Grambsch.6 Analyses were repeated stratifying by type of labor (induced compared with spontaneous) as well as including only cases of clinically diagnosed chorioamnionitis treated with antibiotics. Lastly, to specifically examine the effect of number of cervical examinations after membrane rupture on fever risk, we redefined time zero as the time of membrane rupture and repeated the analyses with examinations performed after the time of rupture. Because length of time of rupture violated the proportional hazards assumption, it was dichotomized at 12 hours for this subanalysis. Because our sample size was fixed by the ongoing parent cohort study, we performed a post hoc calculation to determine our detectable difference. With an incidence of fever of 7.2%, assuming an α error of 0.05, we had 90% power to detect a 1.8-fold increased risk of fever with each additional examination over our reference (one to three examinations). All statistical analyses were completed using STATA 10 software package.
Of 2,400 women with singleton pregnancies reaching the second stage of labor and delivering at term, five women who were febrile on admission were excluded. Of the remaining 2,395 women, 174 (7.2%) developed an intrapartum fever. Women undergoing cesarean delivery before complete dilation (848) were excluded from this analysis. There was no significant difference in the incidence of fever between those women included and those excluded as a result of cesarean delivery in the first stage: of the 332 excluded women who labored before their first-stage cesarean delivery, 22 (6.6%) developed an intrapartum fever (P=.40).
Women who developed a fever, compared with those who did not, were statistically similar with respect to mean maternal age and gestational age at the time of delivery. The incidence of hypertensive disorders, diabetes mellitus, gestational diabetes, and rupture of membranes on admission were also statistically similar between the two groups. However, there was a number of differences. Patients who developed a fever were of lower gravidity, more likely to be nulliparous, and more likely to require induction of labor for any cause and specifically induction for premature rupture of membranes. In addition, women with fever had less cervical dilation and were more likely to have a Bishop score of 5 or less on admission and receive prostaglandins and regional anesthesia during labor. They were also less likely to have a spontaneous vaginal delivery (Table 1).
Women were examined between one and 14 times. Those who developed a fever underwent two more digital cervical examinations on average than those who remained afebrile (7.6±2.8 compared with 5.6±2.5, P<.001) and spent a significantly longer time in labor (16.3±11.1 compared with 10.3±9.6 hours, P<.001). Table 2 presents the results of the final Cox proportional hazard model accounting for spontaneous labor, rupture of membranes on admission, and Bishop score less than or equal to five at admission. Compared with women who had one to three cervical examinations, women who had a greater number of examinations were not at increased risk of fever. Even for the 505 women who had greater than seven examinations during labor, there was no statistically significant increased risk of fever (hazard ratio [HR] 0.9, 95% confidence interval [CI] 0.4–2.0). Spontaneous labor was associated with a reduced risk of intrapartum fever (HR 0.5, 95% CI 0.3–0.8), whereas rupture of membranes at admission and unfavorable Bishop score were associated with an increased risk of fever (HR 2.4, 95% CI 1.3–4.2; HR 2.2, 95% CI 1.4–3.3, respectively). Although epidural anesthesia and internal monitor use were more common in women with a fever, they did not remain significant factors in the final models.
To further delineate the relationship between number of cervical examinations and intrapartum fever, the analysis was stratified by labor type. After adjusting for rupture of membranes on admission and Bishop score, there was no significant association between number of cervical examinations in induction of labor and maternal fever (Table 3). Similar to the findings in the analysis of the entire sample, even induced women (n=320) who had more than seven cervical examinations during induction were not at increased risk of fever (HR 0.8, 95% CI 0.1–6.1). The lack of a significant association between number of examinations and fever remained when this analysis was performed for women in spontaneous labor (Table 4). Increasing number of cervical examinations during spontaneous labor did not result in increased risk of fever, even for those women (n=185) with greater than seven examinations (HR 0.7, 95% CI 0.3–1.7). Additionally, in the repeated analysis in women diagnosed with clinical chorioamnionitis and treated with antibiotics, there was no statistically significant association between number of examinations and diagnosis of chorioamnionitis.
Finally, to examine the risk of fever associated with cervical examinations after membrane rupture, the risk associated with increasing number of examinations was estimated in women ruptured less than 12 hours and also in those ruptured for 12 or more hours (Table 5). For membrane rupture less than 12 hours, there was no significant association between increasing number of examinations after membrane rupture and fever. Similarly, for membrane rupture of 12 hours or more, the risk association between increasing number of examinations after membrane rupture and fever was also not significant. In this group, women (n=101) who had greater than seven cervical examinations after membrane rupture did not have a statistically significant increased risk of fever (HR 2.4, 95% CI 0.6–10.6).
Our study shows that the number of cervical examinations in term labor is not an independent risk factor for intrapartum fever. The lack of association between examinations and fever persisted in both spontaneous and induced labor. We also found that the number of cervical examinations performed after rupture of membranes is not associated with an increased risk of fever. Our findings differ from previous reports of independent risk factors for maternal intrapartum infection.7–10
Iatrogenic labor interventions, including cervical examinations, are believed to increase the risk of intra-amniotic infection through the introduction of vaginal flora into the upper genital tract.4 Several prospective studies of both preterm and term deliveries have used multivariate analysis to identify independent risk factors for maternal infection.7–10 In a large cohort of 5,409 consecutive patients, Soper et al8 found that four or more vaginal examinations was independently associated with a threefold increased risk of intra-amniotic infection in term deliveries (odds ratio [OR] 3.07, 95% CI 2.53–3.73). Similarly, Seaward et al,9 using data from a large randomized controlled trial of 5,028 women with premature rupture of membranes at term, found that greater than eight vaginal examinations was associated with a fivefold increased risk of clinical chorioamnionitis (OR 5.07, 95% CI 2.51–10.25). These studies and others have included the length of active labor7,9 and the duration of membrane rupture7–9 as independent variables in the same logistic regression model as the number of cervical examinations. Because the number of examinations in labor is inherently related to both the time spent in labor and thus the duration of membrane rupture as well, this type of analysis may not give the most accurate estimate of the risk association between increasing number of cervical examinations and maternal fever. The strong associations between cervical examinations and maternal infection found in these studies may in fact reflect the increased risk of infection over time.
In the second phase of the cohort study by Soper et al,8 infection control measures targeting vaginal examinations and the insertion of internal monitors failed to reduce the incidence of intrapartum fever in 2,549 consecutive deliveries at a large teaching hospital. The authors suggested that lack of compliance or the limited effectiveness of the infection control measures in preventing contamination may have accounted for these results and emphasized the need to limit multiple examinations in the teaching setting. However, we offer the possibility that the lack of effect seen in this study may be explained by the true effect of length of labor on the increased risk of infection in women with multiple cervical examinations and not the number of examinations per se.
Publications examining the effect of digital cervical examinations on maternal and neonatal outcomes have relied primarily on high-risk preterm populations. In retrospective studies of preterm premature rupture of membranes, digital cervical examination, compared with no digital examination or sterile speculum examination, was associated with a significantly decreased latency period but not with increased risk of infection.11–13 However, Lewis et al13 reported a nonsignificant trend toward increased maternal infection in those women who had one or more digital cervical examinations compared with sterile speculum examination (44.1% compared with 33.1%, P=.09). These results did not adjust for potential confounders and, moreover, apply only to the risk of infection in the setting of preterm premature rupture of membranes.
Our study offers several strengths over previous publications. Our large sample size and robust data set allowed us to adjust for potential clinical confounders and to stratify our analysis by relevant subgroups. We were also able to estimate the risk association between number of cervical examinations and intrapartum fever both with spontaneous and induced labor, and we examined the association between examinations performed after membrane rupture and risk of fever. Most importantly, we used time-dependent analyses to account for the length of labor to more accurately estimate the independent effect of cervical examinations on risk of maternal fever. Lastly, we were able to test our hypothesis in a group of consecutive term labor patients who completed the first stage and thus had a similar time at risk for cervical examinations. However, if informative censoring resulting from fever affected the physician decision for first-stage cesarean delivery, a significant selection bias could have been created. For this reason, we re-extracted the charts of all women who labored and underwent first-stage cesarean delivery to demonstrate that they did not have a significantly different, and specifically higher, rate of fever.
Potential limitations of our study are important to consider when interpreting our results. The retrospective design dictated that only documented cervical examinations could be analyzed. Although efforts are made to accurately record all examinations in the medical record, it is possible particularly in the first few months of the academic year that for a single documented examination, a second examination was performed to check the quality of the first. This practice is standard at teaching centers where new residents have limited experience in performing cervical examinations and their clinical assessment needs validation. We would offer that this misclassification would most likely lead to a nondifferential bias, but this remains a potential source of bias. Our large cohort and post hoc power calculation revealed power to detect clinically significant risk of maternal fever conferred by increasing cervical examinations in our cohort. However, estimates below that threshold, which some may still consider clinically significant, and those within subgroup analyses were limited in their precision and should be interpreted with care. Our cohort was recruited from a single tertiary care medical center with urban, mostly African American patients. However, the incidence of fever in our population was 7.4%, which is similar to the incidence of maternal fever found in large cohorts of women at other teaching hospitals8 and to the incidence of clinical chorioamnionitis found in a multicenter trial of premature rupture of membranes at term.9 Thus, although our results may not be generalizable to all populations of obstetric patients, our findings are relevant to large teaching medical centers where the possibility of multiple cervical examinations is greatest. Lastly, despite an increase rate of epidural use in febrile women, this was not a statistically significant factor in the final models for risk of fever despite the fact that prior work has demonstrated an association between regional anesthesia and maternal fever. If an association exists, we believe that our data did not support that finding because so few women had no regional anesthesia, limiting our power to detect this association. This remains an important consideration for the generalizability of our results, because our study was primarily one of women with epidural anesthesia.
In conclusion, our large retrospective cohort of consecutive term second-stage deliveries found no significant association between number of cervical examinations and risk of maternal fever, even when multiple cervical examinations were performed after rupture of membranes. Clinicians may be reassured that increasing the number of cervical examinations in term labor does not place their patients at increased risk of peripartum fever.
1. Petrova A, Demissie K, Rhoads GG, Smulian JC, Marcella S, Ananth CV. Association of maternal fever during labor with neonatal and infant morbidity and mortality. Obstet Gynecol 2001;98:20–7.
2. Duff P. Maternal and perinatal infections—bacterial. In: Gabbe SG, Niebyl JR, Simpson JL, editors. Obstetrics: normal and problem pregnancies. 5th ed. Philadelphia (PA): Churchstone Livingston Elsevier; 2007. p. 1237–9.
3. Duff P, Sweet RL, Edwards RK. Maternal and fetal infections. In: Creasy RK, Resnik R, Iams JD, editors. Creasy and Resnik's maternal fetal medicine. 6th ed. Philadelphia (PA): Saunders Elsevier; 2012. p. 757–9.
4. Gibbs RS, Duff P. Progress in pathogenesis and management of clinical intraamniotic infection. Am J Obstet Gynecol 1991;164:1317–26.
5. Newton ER. Chorioamnionitis and intraamniotic infection. Clin Obstet Gynecol 1993;36:795–808.
6. Grambsch PM. Goodness-of-fit and diagnostics for proportional hazards regression models. Cancer Treat Res 1995;75:95–112.
7. Soper DE, Mayhall CG, Dalton HP. Risk factors for intraamniotic infection: a prospective epidemiologic study. Am J Obstet Gynecol 1989;161:562–6.
8. Soper DE, Mayhall CG, Froggatt JW. Characterization and control of intraamniotic infection in an urban teaching hospital. Am J Obstet Gynecol 1996;175:304–9.
9. Seaward PG, Hannah ME, Myhr TL, Farine D, Ohlsson A, Wang EE, et al.. International Multicentre Term Prelabor Rupture of Membranes Study: evaluation of predictors of clinical chorioamnionitis and postpartum fever in patients with prelabor rupture of membranes at term. Am J Obstet Gynecol 1997;177:1024–9.
10. Newton ER, Prihoda TJ, Gibbs RS. Logistic regression analysis of risk factors for intra-amniotic infection. Obstet Gynecol 1989;73:571–5.
11. Adoni A, Chetrit AB, Zacut D, Palti Z, Hurwitz A. Prolongation of the latent period in patients with premature rupture of the membranes by avoiding digital examination. Int J Gynaecol Obstet 1990;32:19–21.
12. Alexander JM, Mercer BM, Miodovnik M, Thurnau GR, Goldenberg RL, Das AF, et al.. The impact of digital cervical examination on expectantly managed preterm rupture of membranes. Am J Obstet Gynecol 2000;183:1003–7.
© 2012 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.
13. Lewis DF, Major CA, Towers CV, Asrat T, Harding JA, Garite TJ. Effects of digital vaginal examinations on latency period in preterm premature rupture of membranes. Obstet Gynecol 1992;80:630–4.