|Home | About | Journals | Submit | Contact Us | Français|
Hospital-based data from Africa suggest that newborn skin-cleansing with chlorhexidine may reduce neonatal mortality. Evaluation of this intervention in the communities where most births occur in the home has not been done. Our objective was to assess the efficacy of a 1-time skin-cleansing of newborn infants with 0.25% chlorhexidine on neonatal mortality.
The design was a community-based, placebo-controlled, cluster-randomized trial in Sarlahi District in southern Nepal. Newborn infants were cleansed with infant wipes that contained 0.25% chlorhexidine or placebo solution as soon as possible after delivery in the home (median: 5.8 hours). The primary outcome was all-cause mortality by 28 days. After the completion of the randomized phase, all newborns in study clusters were converted to chlorhexidine treatment for the subsequent 9 months.
A total of 17 530 live births occurred in the enrolled sectors, 8650 and 8880 in the chlorhexidine and placebo groups, respectively. Baseline characteristics were similar in the treatment groups. Intention-to-treat analysis among all live births showed no impact of the intervention on neonatal mortality. Among live-born infants who actually received their assigned treatment (98.7%), there was a nonsignificant 11% lower neonatal mortality rate among those who were treated with chlorhexidine compared with placebo. Low birth weight infants had a statistically significant 28% reduction in neonatal mortality; there was no significant difference among infants who were born weighing ≥2500 g. After conversion to active treatment in the placebo clusters, there was a 37% reduction in mortality among low birth weight infants in the placebo clusters versus no change in the chlorhexidine clusters.
Newborn skin-wiping with chlorhexidine solution once, soon after birth, reduced neonatal mortality only among low birth weight infants. Evidence from additional trials is needed to determine whether this inexpensive and simple intervention could improve survival significantly among low birth weight infants in settings where home delivery is common and hygiene practices are poor.
DEVELOPING COUNTRIES HAVE made significant progress in reducing mortality for children who are younger than 5 years through the implementation of a variety of maternal and child health interventions in the past 3 decades. Little progress has been made, however, in reducing perinatal and neonatal mortality,1,2 and neonatal deaths now comprise nearly 40% of all mortality in children who are younger than 5 years.3 Ninety-nine percent of neonatal deaths occur in developing countries, mostly at home, and are attributable primarily to infections, birth asphyxia, and complications of prematurity.3,4 Low birth weight (LBW) is an important underlying factor for neonatal mortality and contributes to an estimated 60% to 80% of neonatal deaths in low-resource settings.1 LBW is particularly important in south Asia, where the prevalence is ~30%.2 Recent research suggests that improved care of LBW infants has the potential to improve survival significantly.5,6
Early neonatal sepsis is associated with colonization of infant skin by organisms that are present in the vaginal canal. This is supported by studies from developed countries that demonstrated reduced neonatal colonization when the vaginal canal is cleansed with antiseptic solution.7–13 Two hospital-based studies from developing countries also have addressed this issue. In nonrandomized, unmasked trials that were conducted in Malawi14 and Egypt,15 pregnant women received vaginal cleansing with 0.25% chlorhexidine during labor and their newborns were given full-body skin-cleansing with the same solution. Compared with nonintervention time periods, neonatal mortality was reduced 22% in Malawi and 33% in Egypt. However, the combination of treatment to both mother and child prevents differentiating the contribution of each intervention to these results. Generalizing these findings to other populations with high neonatal mortality also is problematic because other factors may modify the effect of such an intervention. For example, in rural Nepal, the rates of sexually transmitted infections and bacterial vaginosis are low compared with many places in Africa,16,17 and the vast majority of deliveries occur at home, a highly contaminated environment. Therefore, colonization of the infant with potential pathogens from the mother during delivery may play a smaller role in South Asia than in Africa. The causative organisms for neonatal infections also suggest that environmental sources are common.18,19 Furthermore, the treatment effect could be modified by the birth weight distribution of the population, because skin barrier function is compromised in preterm, LBW infants, leading to increased risk for infection.20–22 Given the marked differences in LBW rates in Asia and Africa, it is important to understand the impact of the intervention in both contexts. On the basis of the potential for skin-cleansing of newborns with chlorhexidine to be a simple, inexpensive, and safe intervention for improving neonatal survival, we conducted a community-based trial of this intervention on neonatal mortality.
The Nepal Newborn Washing Study was a cluster-randomized, placebo-controlled, community-based trial that began in September 2002 among newborn infants who were delivered to women who lived in Sarlahi District in south-central Nepal. This area is typical of much of the Indian subcontinent. The population is considered poor, even in Nepal; almost three fourths of families live below the poverty line.23 The area was divided into sectors (N = 413) on the basis of the population that 1 local female worker (ward distributor [WD]) could service, ~40 to 50 households.
Pregnancies in the study area were identified by the WDs, who went door to door on a monthly basis. At ~6 months' gestation, women were recruited for participation. At the time of recruitment, all women received weekly vitamin A supplementation, iron–folic acid supplements, and albendazole. These benefits were provided because previous research in this area had demonstrated their efficacy. Tetanus immunization history was assessed and, when deficient, provided by study staff, and all women were provided with a clean birthing kit. Workers provided education regarding proper nutrition during pregnancy and hygienic delivery and neonatal care, including clean cord care and prevention of hypothermia. Although the prevalence of HIV infection and group B streptococcal colonization in pregnant women is not known in this area, they are likely to be very low.
Infants in this population typically are delivered by family members or untrained, traditional birth attendants, and >95% of births occur in the home. The WD was notified by families as soon as possible after the birth of a child. The WD then walked to the home to enroll the newborn infant and conducted the skin-cleansing.
Randomization was conducted at the sector level, stratified by geographic area and tertiles of infant mortality risk as measured during an antenatal micronutrient supplementation trial that was completed in the same area 1 year before the start of this study.24 Concurrent with this trial, a subset of infants were included in a nested trial of 3 approaches to care of the umbilical cord using omphalitis and mortality as end points (Fig 1). Details of this trial were presented in a separate article.25
Informed consent was obtained at the community, household, and individual levels. Community consent was obtained during meetings with community leaders. In addition, verbal consent was obtained at the household level from the parents of enrolled infants.
Newborn infants were sector-randomized to receive 1 of 2 skin-cleansing regimens when the WD arrived at the house after delivery: (1) wiping of the total body excluding the eyes and ears with Pampers infant wipes (Procter and Gamble Co, Cincinnati, OH) that released a solution that contained 0.25% free chlorhexidine (equivalent to 0.44% chlorhexidine digluconate) or (2) wiping with the same infant wipes that lacked chlorhexidine (placebo). All wipes were alcohol-free, produced by Procter and Gamble Co, and packaged in sterile plastic sachets that contained 6 wipes. The allocation codes were kept at Proctor and Gamble, and investigators and all study staff were masked to the treatment assignment. Demonstrations with life-sized dolls and an instructional video were used to train WDs to deliver the intervention using a standard protocol.26
The primary outcome was all-cause mortality within the first 28 days of life. Cause of death was assessed by verbal autopsy and classified by an algorithmic approach defined previously.27 Sepsis deaths were defined by the presence of 2 or more of the following signs or symptoms: (1) caregiver's report of fever; (2) vomiting more than half of feeds; (3) unconsciousness; (4) bulging fontanelle; (5) feeding difficulty (not able to suck before death or feeding less than normal); (6) skin or umbilical cord infection (pus discharge from the cord stump); (7) jaundice; and (8) difficulty breathing and either rapid breathing or chest indrawing. A hierarchy of diagnoses was applied during cause-of-death analysis as follows: congenital abnormality, tetanus, prematurity, birth asphyxia, sepsis, acute lower respiratory infection, and diarrhea.27 For deaths that were attributed to congenital abnormality or tetanus, a confirmation by physician review of the verbal autopsy record was required.
Community-level data on the presence of economic, educational, and health facilities were collected during interviews with community leaders. Household-level data on socioeconomic status, education, maternal health indicators including reproductive history, and household structure were collected during a household interview. After the intervention visit to the home, a staff member (not the WD) visited to collect data on the delivery process and the condition and immediate care of the newborn infant and to weigh the infant using a digital infant scale. Visits were conducted on days 2, 3, 4, 6, 8, 10, 12, 14, 21, and 28 to assess infant vital status and morbidity. At each visit, the staff member examined the infant for signs of umbilical cord and skin infection, measured the axillary temperature and respiratory rate, and recorded other morbidity symptoms and signs. On day 14, an interview was conducted with the mother on newborn care practices. Infants who had specific sets of signs and symptoms at the time of household visits were referred to the local health system for care. All infants who were alive at 28 days were discharged from the study.
The study originally was designed to detect a minimum reduction in all-cause neonatal mortality of 20% (relative risk [RR]: 0.80) given 80% power, a 2-sided type I error of 5%, 5% loss to follow-up, and a neonatal mortality rate of 50 per 1000 live births in the placebo group. Because neonatal mortality did not cluster at the sector level during a previous study,24 no adjustment was made for intracluster correlation. This resulted in a sample size of 13 500 live births. A lower-than-expected neonatal mortality rate by the time of the first Data and Safety Monitoring Board (DSMB) meeting prompted the recommendation that the sample size be expanded to ~17 000 live births.
Statistical analysis was performed using Stata 8.0 (Stata Corp, College Station, TX). Treatment groups were compared on baseline household, maternal, and delivery characteristics. Treatment effect on neonatal mortality was done on 2 populations. The first used all live births that occurred in the cluster irrespective of whether they received their assigned intervention. The second used only newborns who were alive at the time of the WD visit to the home and who received their assigned intervention. Mortality was compared using live births as the denominator and deaths within the first 28 days as the numerator. Survival analysis, including Kaplan-Meier survival curves, also was conducted. Multivariable binomial regression with a log link function was used to model the risk for mortality adjusted for potentially confounding factors imbalanced across treatment groups and to model effect modification. Estimates of the RRs and their SEs were adjusted to account for the clustered randomization using generalized estimating equations with an exchangeable correlation structure.28
Children who migrated out of the study area or whose parents refused additional participation before day 28 were censored at the time when they left the study. Stratified analyses were planned for selected variables that are known to be important risk factors for neonatal mortality, such as LBW, previous child death in the family, gender, and ethnic group.
This study received approval from the Nepal Health Research Council and the Committee on Human Research of the Johns Hopkins Bloomberg School of Public Health. It is registered at www.clinicaltrials.gov (NCT00109616). An independent DSMB met 3 times to review the protocol and the data for safety and efficacy. At the meeting in January 2005, the DSMB recommended that the study be stopped because of adequate evidence for a beneficial effect of chlorhexidine newborn skin-cleansing among LBW infants. The ethical committees approved this decision, and from March 8, 2005, through January 2006 (end of enrollment), all newborn infants in the study area received skin-cleansing with the chlorhexidine solution.
Financial supporters and the commodity supplier played no role in the design, conduct, management, analysis, or interpretation of the results or in the preparation, review, or approval of this article.
Between September 1, 2002, and March 8, 2005, there were 17 530 live-born infants from women who agreed to participate during their pregnancy (Fig 2). Parents of 18 (0.2%) infants in each group refused to participate when asked again at the time of infant enrollment. More infants did not receive their assigned intervention in the chlorhexidine group, 27 for other reasons and 86 because the infant died before the time the WD arrived to conduct the intervention (total = 113 [1.3%]) vs 20 for other reasons and 55 pretreatment deaths in the placebo group (total = 75 [0.8%]; Fig 2). The reason for this imbalance is unclear because the distribution of the time between birth and intervention was the same in the 2 treatment groups.
Baseline characteristics of the families, deliveries, and newborn infants who were enrolled in the trial were well balanced across treatment groups (Table 1). More than 90% of deliveries occurred at home. Approximately 60% of birth attendants washed hands with soap and water before the delivery; 80% of infants were fed colostrum, but only approximately half began breast-feeding within 12 hours after delivery. Birth weight distributions in the treatment groups were similar as were a number of other characteristics related to delivery hygiene and status of the infant at birth (Table 1). The LBW rate was 30%.
Because study staff typically were not in attendance at the delivery, WDs traveled to the house where the delivery took place. Despite this inherent delay, newborn skin-cleansing occurred soon after delivery at a median time of 5.8 hours after birth (interquartile range: 2.1− 11.8 hours); 91.4% of infants were cleansed within the first 24 hours. No adverse reactions to the intervention were reported.
A concern regarding this intervention was the potential for hypothermia as a result of the wetting action of the wipes, especially because the World Health Organization (WHO) recommends that bathing after birth be delayed for at least 6 hours.29 Previous research in our study area showed that 95% of families conduct a wet wash of the newborn within the first 12 hours after delivery.30 This suggested that an additional skin-wiping followed promptly by wrapping of the infant and accompanied by behavioral messages to limit exposure and manage hypothermia would be unlikely to produce additional risk as a result of hypothermia. A pilot study of the wiping procedure confirmed that little moisture was left on the skin, little to no vernix was removed, and a small drop in body temperature (mean: 0.4°C) resulted.26 A previous trial of antenatal micronutrient supplementation on birth weight and early infant health that was conducted in the same study area used an identical method for measurement of infant temperature (P. Christian, DrPH, written communication, 2005). This allowed a comparison of rates of hypothermia between these studies. Adjusted for month of birth, there was a 12% (95% confidence interval [CI]: 5%−19%) increased odds of mild hypothermia (36.5°C–35.8°C) and a 16% (95% CI: 22%−10%) reduced odds of moderate to severe hypothermia (<35.9°C) in the skin-cleansing trial compared with the previous study.
On the basis of the intention-to-treat analysis of all live births, there was no difference in neonatal mortality (RR: 1.04; 95% CI: 0.87−1.24; Table 2). Among those who actually received their assigned intervention, there was a nonsignificant 11% lower neonatal mortality rate in the chlorhexidine group compared with placebo (Table 2). However, among LBW infants, the chlorhexidine group had a significant, 28% lower mortality than the placebo group (RR: 0.72; 95% CI: 0.55− 0.95). Infants who weighed ≥2500 g showed no significant difference in mortality (Table 2; Fig 3). Among LBW infants, baseline comparisons between the treatment groups also were well balanced (data not shown). Alternative definitions of LBW, such as <2000 g, showed similar treatment effects as for those <2500 g (data not shown). There was no evidence of interaction between the skin-cleansing intervention and the treatments in the umbilical cord care trial; both chlorhexidine treatments acted independently.25 There was no evidence that other factors modified the effect of skin-cleansing with chlorhexi-dine on mortality risk (Table 2). Among LBW infants, chlorhexidine skin-cleansing was equally effective when stratified by whether the birth attendant washed her hands before delivery (washed hands RR: 0.77 [95% CI: 0.51−1.18]; did not wash hands RR: 0.78 [95% CI: 0.53− 1.16]). Among LBW infants, there was a trend toward reduction in sepsis-specific deaths in the chlorhexidine group (RR: 0.83; 95% CI: 0.53−1.29).
Once the DSMB recommended that the randomized-trial portion of the study be stopped, all subsequent live births (n = 6055) were eligible to receive the active treatment. After conversion of the placebo clusters to chlorhexidine, there was a statistically nonsignificant 6% reduction in overall mortality as compared with no change in the chlorhexidine clusters (Table 3). Among LBW infants, there was a 37% reduction in mortality among infants in the placebo clusters, whereas no change was found from before conversion to after conversion in the chlorhexidine clusters.
Overall, there was no evidence that chlorhexidine cleansing of the skin was associated with reduced neonatal mortality in this population. However, among LBW infants who actually received their assigned intervention, there was a 28% reduction in neonatal mortality among those whose skin was wiped with chlorhexi-dine as soon as possible after birth. The causal evidence for the LBW results were strengthened by the observation of a significant decline in mortality among LBW infants after the randomized portion of the study among sectors that originally were assigned to the placebo group, whereas mortality in the clusters that originally were assigned to chlorhexidine did not change. These data suggest that, in a high-risk subgroup of infants, early postdelivery skin-cleansing with chlorhexidine may reduce the risk for neonatal mortality. This finding needs confirmation, however, from other studies of LBW infants in settings where home deliveries are common. These data lend partial support to 2 previous, nonrandomized, hospital-based studies in Malawi and Egypt, where significant reductions in early neonatal mortality were found among infants who were washed with 0.25% chlorhexidine and whose mothers received vaginal cleansing with the same solution.14,15 Whether the combined use of vaginal and newborn skin-cleansing would have produced greater benefit or spread the benefit across all infants in our study population is unknown. More than 90% of births in Nepal occurred at home, where there is a much higher potential for environmental contamination of the newborn infant than in hospital settings.
As expected for an intervention that focuses on reducing the risk for infection, fewer deaths among LBW infants as a result of sepsis were observed in cause-specific analysis. However, results of verbal autopsies for very young infants must be viewed with caution because many of the signs and symptoms that were used in the algorithms for cause of death can apply to a variety of causes. A reduction in sepsis deaths also was observed in the Malawi study, in which there was a two-thirds reduction in clinically diagnosed early neonatal sepsis deaths.14
The results from these studies suggest that the importance of a well-developed and healthy skin barrier to infection may have been underestimated to date, especially in settings where environmental contamination and the prevalence of LBW are high. This idea is supported by recent results from trials of therapy with topically applied emollients (eg, sunflower oil), which showed 54% and 41% reductions in nosocomial sepsis among preterm infants in Egypt21 and Bangladesh,22 respectively. Moreover, invasion through the skin of pathogens in neonatal sepsis among very LBW infants in developed countries has been documented.31–33 Evidence suggests that chlorhexidine-mediated reductions in numbers of colonizing pathogens, alternations in the local balance of microbes, and/or in skin barrier immune function may act to reduce risk for development of systemic infections via cutaneous and/or umbilical portals of entry.34–37
Chlorhexidine was chosen for this study because of its established safety profile at concentrations well above those used in this trial and because it leaves a residual bactericidal effect on the skin.38 It is included on the WHO essential drug list, and the WHO recommends it when antiseptic applications to the umbilical cord are indicated.39 We found no adverse effects as a result of its use as a skin-cleansing agent.
This study does have limitations. Specifically, there was imbalance in the number of deaths that occurred between birth and the time when our local worker arrived at the house to enroll the infant officially and provide the assigned skin-cleansing. We have no explanation for this difference in preenrollment mortality because the distribution of time between birth and the application of the assigned treatment was identical in the 2 treatment groups. Treatment groups also were well balanced on a variety of other factors that are associated with early infant mortality.
The results from this trial in rural Nepal and the hospital-based studies in Africa suggest that a simple, safe, and inexpensive newborn chlorhexidine skin-cleansing may reduce significantly neonatal mortality among infants in these low-resource settings. However, the limitation of a positive treatment effect to a subgroup of this study population requires that additional data that directly address the role of chlorhexidine skin-cleansing among LBW infants be collected in populations with high risk for neonatal mortality before this intervention can move from research to programs and policy.
This study was conducted by the Department of International Health, Bloomberg School of Public Health, Johns Hopkins University (Baltimore, MD), under grants HD 44004 and HD 38753 from the National Institutes of Health (Bethesda, MD); grant 810 −2054 from the Bill and Melinda Gates Foundation (Seattle, WA); and Cooperative Agreements HRN-A-00−97−00015−00 and GHS-A-00−03−000019−00 between Johns Hopkins University and the Office of Health and Nutrition, US Agency for International Development (Washington, DC). Commodity support was provided by Procter and Gamble Co (Cincinnati, OH).
Special appreciation goes to Data and Safety Monitoring Board members Drs P. S. S. Sundar Rao, Pushpa Sharma, Dharma Manandhar, and Martin Bloem.
Financial Disclosure: Dr Darmstadt has collaborated with Proctor and Gamble on testing of skin emollient products for premature newborns. The other authors have indicated they have no financial relationships relevant to this article to disclose.
A preliminary set of these data was presented at the annual meeting of the Pediatric Academic Societies; May 16, 2005; Washington DC.