In many developing countries, pneumonia and sepsis are the major causes of neonatal mortality.1 The organisms causing neonatal infection in developing countries are generally gram-negative (Klebsiella pneumoniae and Escherichia coli), whereas group B streptococci is often the organism causing neonatal sepsis in developed countries.1 These organisms likely originate in the maternal genital tract and are transmitted to the fetus during labor. Decreasing mother-to-newborn transmission of these organisms has been advocated as a potential approach to reducing perinatal infections.
It has been proposed that peripartal infection of the fetus and mother may be prevented by intrapartum vaginal and neonatal wipes with an antiseptic solution.2,3 One such antiseptic solution is chlorhexidine, which has a wide range of effectiveness against aerobic and anaerobic gram-negative and gram-positive organisms. In an attempt to reduce neonatal infections, vaginal chlorhexidine in varying strengths (0.25–0.5%) has been tested as a means to reduce perinatal infections. At these concentrations, chlorhexidine is essentially free of side effects. Concentrations of 1% or greater are associated with increased vaginal burning and itching.4
Scandinavian studies showed reduced neonatal bacterial colonization after chlorhexidine vaginal wiping. However, the developed countries studies, with neonatal sepsis end points, were inconsistent.5–9 Developing-country quasirandomized studies showed more promise. Taha et al10 in Malawi and Bakr et al11 in Egypt both appeared to show that 0.25% chlorhexidine vaginal and neonatal wiping reduced neonatal deaths as a result of sepsis and improved other neonatal and maternal outcomes. In a subgroup analysis of a community study in Nepal, neonatal wiping with 0.25% chlorhexidine was associated with a reduction in mortality only among low birth weight neonates.12 A randomized trial showed no maternal or neonatal benefit.13
Because of the potential for chlorhexidine to reduce perinatal infections, because a pilot study suggested possible reductions in newborn infections,14 and because of methodological shortcomings in some prior studies, we conducted a randomized trial of chlorhexidine vaginal and baby wipes to evaluate their effect on fetal and neonatal mortality and infection-related morbidity. For the wipes, we used a strongly bacteriocidal (0.6%) chlorhexidine solution and compared the outcomes with those seen with wipes in a control group that received normal saline.
MATERIALS AND METHODS
Between June 13, 2005, and May 27, 2008, three hospitals that provide services to the poor, urban population of Karachi, Pakistan, participated in this study. Civil Hospital Karachi is affiliated with Dow University, and the remaining two hospitals are affiliated with the provincial and city governments. We had two types of teams for the study. The hospital study teams of doctors and nurses were trained on study screening, randomization and enrollment procedures, performing the vaginal and baby wipes, and data collection. The field teams conducted follow-up home visits at days 7 and 28 in which they examined mothers and neonates and collected data. The teams worked independently and were masked to treatment allocation.
Women aged 18–45 years who were admitted to the study hospitals were screened for eligibility. Women were excluded if they had a contraindication to digital cervical examination (eg, placenta previa), active genital herpes, vulvovaginal ulceration, face presentation, known fetal death, allergy to chlorhexidine, a planned cesarean delivery, arrived in labor at full dilatation, were from areas of Karachi where follow-up was risky, or were unwilling or unable to give consent. If an enrolled woman had twins, only the first born was counted as a study participant.
Commercially available chlorhexidine diacetate powder was purchased from Degussa (Frankfurt, Germany). The research pharmacist at the Aga Khan University prepared the 0.6% solution using sterile water (7.14 g/L). The solution was buffered to pH 8 with the addition of hydrochloric acid and, in four sterile 100-mL amber-colored bottles, packed in a cardboard box. Sterile saline solution in identical bottles and boxes served as the placebo.
The data center prepared a randomization list for each of the hospitals with treatment allocation generated in blocks of four and eight. Because both solutions were clear and odorless, neither the study team members nor the participants could differentiate them. Two separate sterilized packs of gloves, plastic trays, and cotton balls for application of study solution were provided. The study solution was warmed before use. The vaginal wipe was performed by the study doctor or nurse once the woman was deemed in labor and then every 4 hours (up to three times) until delivery. Using sterile gloves, a study solution-soaked cotton ball was rotated by the fingers circumferentially to cover the entire cervix, the vagina, and external genitalia. Enrolled women received a minimum of one wash at least 1 hour before delivery. The study staff wiped the neonate (but not the head and face) with 100 mL of study solution, usually immediately after delivery, within 6 hours of birth. To preserve the vernix, the team gently patted the newborn and then, to prevent hypothermia, immediately dried and wrapped him or her.
To ensure that the women received the assigned treatment and that the chlorhexidine concentration was appropriate and bacteriocidal, an independent laboratory performed tests on randomly chosen bottles. These tests confirmed that bottles presumed to contain chlorhexidine did so and in a concentration of 0.6% and were bacteriocidal, whereas the bottles presumed to contain saline had no detectable chlorhexidine and were not bacteriocidal.
The primary study outcome was the composite of fetal and neonatal mortality and neonatal sepsis at 7 days or less of life. Fetal deaths occurring during delivery were included in the final analysis because the fetuses were alive at enrollment and this was an intent-to-treat trial. Sample size calculations indicated that a trial of 5,000 women, with 2,500 in each treatment group, would have 80% power (two-tailed test at α=0.05) to detect a difference between a control group primary outcome rate of 52 per 1,000 births and the corresponding primary outcome rate of 35 per 1,000 in the chlorhexidine group. Secondary outcomes included neonatal sepsis or mortality 28 days or less, maternal mortality at 28 days or less, and measures of infection-related morbidity.
Women were asked to identify any symptoms of burning, itching, rash, or redness and to quantify the severity of these symptoms using a visual analog scale, which had been pretested in a Pakistani population.14 All neonates were examined by the physicians within 30 minutes of the infant wipe to assess any symptoms of rash, redness, tenderness, or swelling.
Newborns with Apgar scores less than 7 at 5 minutes were referred to the neonatal intensive care units and study wipes were given only after attending physician approval. The admitted newborns were followed daily until the primary outcome (ie, discharged alive, remained admitted until days 7 and 28, or death occurred). The relevant information regarding diagnosis by the attending physician, laboratory tests, and X-rays was retrieved from the hospital records.
At the day 7 and day 28 visits, study physicians assessed both the mother and newborn for illness. For newborns, the Integrated Management of Childhood Illness screening tool15 for identification of severe illness in young neonates was used. The Integrated Management of Childhood Illness screening tool includes a checklist of clinical signs of severe illness.16 At least three clinical signs and one severe symptom must be present to be classified as positive for sepsis. In addition, skin and umbilical infections were identified by visual examination of the neonate by showing pictures of neonatal skin infections to mothers and graded using standardized criteria. (All neonates were examined for skin infections on days 7 and 28, and a subgroup of 1,412 consecutively born neonates was examined on day 14.) Mothers were questioned about fever, lower abdominal pain, and vaginal discharge. Referrals to the hospital were made as appropriate according to Integrated Management of Childhood Illness criteria and the maternal condition. Whenever possible, neonates or mothers who entered a hospital were followed-up at the hospital to retrieve appropriate information.
A panel of three neonatologists and one obstetrician–gynecologist who were not members of the study team reviewed records of the neonates who were 1) admitted to the neonatal intensive care units after delivery, 2) readmitted until day 28 after discharge from the hospital, or 3) referred to the hospital by the study team because of potential severe illness. Records of neonates receiving antibiotics were also reviewed. Along with the neonatal information, maternal events at delivery were also reviewed, including highest temperature, meconium-stained amniotic fluid, foul-smelling or purulent vaginal discharge, tender uterus or abdomen, clinical diagnosis of chorioamnionitis or urinary tract infection, and duration of membrane rupture. Based on their overall judgment of the clinical and laboratory data, and maternal events at delivery, the panel classified each case into the following categories: 1) sepsis: neonates were severely ill according to Integrated Management of Childhood Illness or clinical presentation, maternal history, and involvement of at least one organ system and laboratory findings; 2) probable sepsis: neonates were severely ill according to the Integrated Management of Childhood Illness screen, clinical presentation, and had a maternal history supporting infection; 3) possible sepsis: neonates were severely ill according to Integrated Management of Childhood Illness or clinical screen and had no evidence of a nonseptic condition to account for their condition; or 4) not sepsis: these neonates were ill according to Integrated Management of Childhood Illness, had no clinical evidence of infection, and often had evidence of another condition. Neonates who were classified as having sepsis, probable sepsis, or possible sepsis were analyzed as “sepsis” for the primary outcome.
Data were analyzed using the intent-to-treat principle with SAS 9.1.3 (SAS Institute Inc, Cary, NC). Descriptive statistics (frequencies, percentages, means, standard deviations, and relative risks and 95% confidence intervals) were calculated excluding missing data from the analyses. Comparisons between the chlorhexidine and placebo groups were performed using χ2 tests for binary outcomes and t tests for continuous outcomes. In the case of sparse binary outcomes, Fisher exact tests were computed. P values were not adjusted for multiple comparisons. Relative risks and their 95% confidence intervals were used to estimate the association between chlorhexidine and placebo wash with maternal and neonatal outcomes.
Ethics approvals from the all participating institutions were obtained. The Data Monitoring Committee met every 6 months and monitored the trial throughout.
Figure 1 shows study enrollment, randomization, and follow-up. Of 20,800 women screened, 63% were not eligible for the study and were excluded. Approximately 37% of these women were excluded because their admission was for elective cesarean delivery, and 27% were excluded because they were fully dilated on arrival. Another 26% of the women were excluded because they did not live in the study catchment area. In addition, some women (4.4%) arrived at the hospital with a known fetal death. Finally, women who had conditions that contraindicated a cervical examination (2.2%) were excluded. Thus, 7,700 of the women screened were eligible. Of those, approximately 35% refused consent, often because accompanying family members did not give permission in the husbands' absence. A total of 5,008 women were randomly assigned.
The treatment and control groups were similar regarding maternal age, gravidity, number of living children, abortions, gestational age at delivery, labor duration, percent with a twin pregnancy or cesarean delivery, and type of membrane rupture (Table 1). There were also no statistically significant differences in gestational age at delivery and neonate mean birth weight or birth weight distribution, Apgar scores, and sex distribution. Of the mothers randomly assigned to chlorhexidine, 99.8% received at least one wipe as did 99.9% of those randomly assigned to placebo. The respective percentages for neonates were 98.3% and 98.8%. The mean number of maternal wipes was 1.2 and was similar between the treatment groups.
There were no major maternal or newborn complications, including death, anaphylactic shock, Stevens-Johnson syndrome, urticaria, and necrotizing fasciitis, related to chlorhexidine use. Women receiving chlorhexidine had more itching and burning than those who received saline, whereas those neonates who received chlorhexidine had less rash and skin redness (Table 2). None of these findings persisted for more than 1 hour, and none resulted in any obvious sequelae.
A total of 1,596 (32%) neonates were reviewed by the panel to assign the primary outcome. There was not a significant difference between the chlorhexidine and placebo groups in the primary outcome of neonatal death or neonatal sepsis within 7 days of birth nor in the separate components of this outcome (Table 3). Neonatal sepsis rates were not different between the groups nor was overall perinatal mortality, including stillbirths. Prospectively defined secondary neonatal outcome rates, including 28-day neonatal sepsis or mortality (analyzed as a composite or as separate component), were also not different. The mean number of neonatal hospital days before discharge was similar in both groups (data not shown), as were the number of neonates remaining in the hospital at day 7 or rehospitalized before day 28 of life (Table 3).
The percentage of neonates having a skin infection before or on day 7 was less in the chlorhexidine group compared with the placebo group, 3.3% compared with 8.2%, respectively (P<.001) (Table 4). Both bullous and pustular infections were decreased in the chlorhexidine compared with placebo group on day 7. However, at days 14 and 28, the rates of skin infection were not significantly different for chlorhexidine compared with the placebo group.
Maternal deaths were rare and did not differ significantly between the groups. Rates of foul-smelling vaginal discharge or a tender uterus before day 7 or within the first 28 days were also similar between groups. The only statistically significant difference was fewer mothers hospitalized at day 7 in the chlorhexidine group (0.6% compared with. 1.1%, P=.03). The decrease in this outcome in the chlorhexidine group was partially explained by fewer infected cesarean delivery wounds and episiotomies as well as lower rates of maternal fever (data not shown).
As the trial progressed, we became aware that peripartum antibiotic use in the study hospitals was common and collected antibiotic use after approximately 600 women were randomly assigned. Of the 4,378 women for whom we have data, 594 in the chlorhexidine group and 592 in the placebo group received no antibiotics, 246 in the chlorhexidine group and 236 in the placebo group received intrapartum antibiotics, and 1,352 women in the chlorhexidine group and 1,358 in the placebo group received only postpartum antibiotics. To test whether chlorhexidine was effective in reducing the rate of the primary outcome among neonates born to women who did not receive antibiotics during labor, a subanalysis was performed in the combined group of mothers who received no antibiotics or only postpartum antibiotics. There was no difference in the primary outcome between the groups (3.0% chlorhexidine compared with 3.1% placebo, P=.85).
Because a previous study suggested potential benefit of chlorhexidine for low birth weight neonates,12 the primary and secondary outcomes were stratified by birth weight: less than 2,000, 2,000–2,500, and more than 2,500 g. No statistically significant differences were found between the chlorhexidine and placebo groups in any outcome in any birth weight group (data not shown).
This trial from a developing country provides little evidence that the use of a combination of maternal chlorhexidine vaginal wipes during labor and neonatal wipes applied soon after birth reduces maternal and neonatal sepsis when compared with normal saline wipes. These findings are in contrast to two large nonrandomized developing-country studies that suggested an improvement in a number of infection-related perinatal outcomes in the chlorhexidine arms.10,11 However, a recent randomized trial conducted in South Africa did not show a beneficial effect of chlorhexidine vaginal and baby wipes in prevention of neonatal sepsis or the vertical transmission of potentially pathogenic bacteria among neonates, supporting our findings.13
The randomized trials of vaginal and neonatal chlorhexidine wipes published to date from developed countries5–9 and one from South Africa13 have not shown benefit in reducing neonatal sepsis or mortality. To maximize the potential effectiveness of the intervention, we increased the concentration of chlorhexidine to more than double that used in most previous studies. However, despite this higher concentration, chlorhexidine had no effect on neonatal sepsis or mortality in our trial. As in developed countries, there is now strong evidence that this intervention does not contribute to the solution of the important problem of neonatal sepsis.
The reduction in neonatal skin infections on day 7 was statistically significant and, by its magnitude, likely to be real. The result is also plausible in that a treatment at birth would be expected to show a reduction in early but not late skin infections. However, this difference in skin infections did not result in a reduction of mortality, sepsis, or other measures of maternal morbidity. Without clear improvement in these measures, it is difficult to recommend the use of perinatal chlorhexidine washes to reduce superficial skin infections not associated with increased morbidity. A large community-based trial in Nepal of 0.25% chlorhexidine neonatal skin cleansing did not show a reduction in mortality in all neonates treated.12 However, in a secondary analysis, a significant reduction in mortality was seen in low birth weight newborns. Our trial did not show any trend toward a reduction in either sepsis or mortality in low birth weight neonates in the chlorhexidine group. A large study of umbilical cord wiping over 10 days with 4% chlorhexidine found neonatal benefit.17 Our trial should not affect the interpretation of those results.
In carrying out this study, we used a validated set of criteria (Integrated Management of Childhood Illness) to screen neonates for severe illness.15,16 At the completion of the trial, but before breaking the randomization code, a panel of physicians reviewed all adverse outcomes possibly related to neonatal sepsis to arrive at a study end point of neonatal sepsis. This outcome together with death comprised our final study end point. Neonatal death can be caused by conditions other than sepsis such as asphyxia, but because it is difficult to distinguish between these diagnoses, even in settings with sophisticated laboratory and pathologic resources, we chose to include neonatal deaths from any cause as part of our primary end point. The combined rate of sepsis and death was lower than expected, but because there was not even a trend toward a reduction in mortality or sepsis in the treatment group, this finding suggests that even if the primary outcome rate were higher, there would have been no significant difference in outcome.
With the exception of the number of maternal admissions at day 7, there were also no apparent benefits to the mother from the vaginal chlorhexidine wipes. Although the reduction in hospital admissions in the chlorhexidine group may have been a chance occurrence, it is possible that vaginal cleansing with chlorhexidine resulted in fewer genital and wound infections resulting from vaginal organisms. Bakr et al11 also reported fewer hospital admissions in women who had 0.25% chlorhexidine vaginal wipes in labor. In our study, none of the other markers of maternal infection were significantly reduced in the chlorhexidine group, and in any case, the reduction in hospital stay was not associated with a decrease in other measurable sequelae.
Many study women received antibiotics, mostly in the postpartum period but also in the intrapartum period. It might be postulated that antibiotic use may have resulted in the absence of a difference in infection-related neonatal and maternal morbidity. However, analyses of women who did not receive intrapartum antibiotics provide no support for this supposition.
Despite the failure of chlorhexidine maternal and vaginal wipes to reduce sepsis, infection remains a major cause of neonatal morbidity and mortality in many developing-country settings.18 Strategies showing promise in reducing neonatal death from sepsis include antibiotics administered by community health workers and other antibiotic administration strategies.19 However, other than giving antibiotics to the mother to prevent group B streptococcus transfer to the fetus, we are not aware of other strategies that are effective in preventing newborn acquisition of vaginal organisms resulting in neonatal sepsis. Thus, it is important to develop strategies that reduce or prevent newborn bacterial colonization and strategies that reduce death once infection occurs.
1. Thaver D, Zaidi AK. Burden of neonatal infections in developing countries: a review of evidence from community-based studies. Pediatr Infect Dis J 2009;28(suppl):S3–9.
2. Goldenberg RL, McClure EM, Saleem S, Rouse D, Vermund S. Use of vaginally administered chlorhexidine during labor to improve pregnancy outcomes. Obstet Gynecol 2006;107:1139–46.
3. McClure EM, Goldenberg RL, Brandes N, Darmstadt GL, Wright LL; CHX Working Group. The use of chlorhexidine to reduce maternal and neonatal mortality and morbidity in low-resource settings. Int J Gynaecol Obstet 2007;97:89–94.
4. Wilson CM, Gray G, Read JS, Mwatha A, Lala S, Johnson S, et al. Tolerance and safety of different concentrations of chlorhexidine for peripartum vaginal and infant washes: HIVNET 025. J Acquir Immune Defic Syndr 2004;35:138–43.
5. Stade B, Shah V, Ohlsson A. Vaginal chlorhexidine during labour to prevent early-onset neonatal group B streptococcal infection. The Cochrane Database of Systematic Reviews 2001, Issue 4. Art. No.: CD003520.
6. Rouse DJ, Hauth JC, Andrews WW, Mills BB, Maher JE. Chlorhexidine vaginal irrigation for the prevention of peripartal infection: a placebo-controlled randomized clinical trial. Am J Obstet Gynecol 1997;176:617–22.
7. Rouse DJ, Cliver S, Lincoln T, Andrews W, Hauth J. Clinical trial of chlorhexidine vaginal irrigation to prevent peripartal infection in nulliparous women. Am J Obstet Gynecol 2003;189:166–70.
8. Sweeten KM, Eriksen NL, Blanco JD. Chlorhexidine versus sterile water vaginal wash during labor to prevent peripartum infection. Am J Obstet Gynecol 1997;176:426–30.
9. Facchinetti F, Piccinini F, Mordini B, Volpe A. Chlorhexidine vaginal flushings versus systemic ampicillin in the prevention of vertical transmission of neonatal group B streptococcus, at term. J Matern Fetal Neonatal Med 2002;11:84–8.
10. Taha TE, Bigger RJ, Broadhead RL, Mtimavalye LA, Justesen AB, Liomba GN, et al. Effect of cleansing the birth canal with antiseptic solution on maternal and newborn morbidity and mortality in Malawi. BMJ 1997;315:216–9; discussion 220.
11. Bakr AF, Karkour T. Effect of predelivery vaginal antisepsis on maternal and neonatal morbidity and mortality in Egypt. J Womens Health (Larchmt) 2005;14:496–501.
12. Tielsch JM, Darmstadt GL, Mullany LC, Khatry SK, Katz J, LeClerq SC, et al. Impact of newborn skin-cleansing with chlorhexidine on neonatal mortality in southern Nepal: a community-based, cluster-randomized trial. Pediatrics 2007;119:e330–40.
13. Cutland CL, Madhi SA, Zell ER, Kuwanda L, Laque M, Groome M, et al; PoPS Trial Team. Chlorhexidine maternal-vaginal and neonate body wipes in sepsis and vertical transmission of pathogenic bacteria in South Africa: a randomised, controlled trial. Lancet 2009;374:1909–16.
14. Saleem S, Reza T, McClure EM, Pasha O, Moss N, Rouse DJ, et al. Chlorhexidine vaginal and neonatal wipes in home births in Pakistan: a randomized controlled trial. Obstet Gynecol 2007;110:977–85.
15. World Health Organization. Integrated management of childhood illness. Geneva (Switzerland): WHO; 1999.
16. Young Infants Clinical Signs Study Group. Clinical signs that predict severe illness in children under age 2 months: a multicentre study. Lancet 2008;371:135–42.
17. Mullany LC, Darmstadt GL, Katz J, Khatry SK, Leclerq SC, Adhikari RK, et al. Risk of mortality subsequent to umbilical cord infection among newborns of southern Nepal: cord infection and mortality. Pediatr Infect Dis J 2009;28:17–20.
18. Bhutta ZA, Darmstadt GL, Hasan B, Haws R. Community-based interventions for improving perinatal and neonatal health outcomes in developing countries: a review of the evidence. Pediatrics 2005;115(suppl):519–617.
19. Bang AT, Bang RA, Stoll BJ, Baitule SB, Reddy HM, Deshmukh MD. Is home-based diagnosis and treatment of neonatal sepsis feasible and effective? Seven years of intervention in the Gadchiroli field trial (1996 to 2003). J Perinatol 2005;25(suppl 1):S62–71.