A total of 23,662 live births occurred between September 1st, 2002 and January 31st, 2006 and were eligible for enrollment in the study. Over 90% of births occurred at home or outdoors during transport to a facility. The median time to the first study visit was 19.3 hours and 63% were first visited within 24 hours. Less than 5% of eligible subjects were not enrolled for various reasons including being unable to meet the mother and newborn within the first 28 days of life, refusal, emigration, and infant death prior to the first study visit. Maternal hand-washing status was not available for infants who died prior to the first post-delivery study visit. Characteristics of the study population are presented in . There were slightly more males than females enrolled and approximately 70% were from the Madeshi ethnic group. Only about a quarter of mothers had ever attended primary school and other socioeconomic indicators classified this population as poor, even for rural Nepal. Approximately 30% of infants were low birth weight (< 2500 g) and about 18% were preterm ().
Selected Characteristics by Birth Attendant Hand-Washing
The overall mortality rate among enrolled infants was 32.1 per 1,000 live births. Birth attendants washed their hands prior to delivery for 59.2% of live births, whereas only 14.8% of mothers reported washing their hands with soap and water or antiseptic prior to handling their infant (). Neonatal mortality was significantly lower among infants whose birth attendant and/or mother washed their hands with soap and water or antiseptic. Newborns whose birth attendant washed hands before assisting with delivery had a 25% lower risk of death compared to newborns whose birth attendant did not wash her hands [Relative Risk (RR)=0.75, 95% Confidence Interval (CI): 0.65−0.86] (). Infants whose mothers washed their hands prior to handling their infant had a 60% lower risk of neonatal death compared to those whose mothers did not wash their hands (RR=0.40, 95% CI: 0.28−0.59) (). These effects were not independent as shown by the combined effect of both the birth attendant and mother washing their hands (RR=0.44, (95% CI: 0.28−0.68) (). Excluding the 271deaths that occurred prior to the first post-delivery study visit made only modest changes to the estimates of the effects of birth attendant hand-washing behavior ().
Unadjusted Association of Birth Attendant and Maternal Hand-Washing Behavior and Neonatal Mortality
The population attributable risk percent for hand-washing by the birth attendant assisting with delivery was 12.2% (31.9 − 28.0 / 31.9 = 12.2%). Among those infants who survived the first few days of life, the population attributable risk percent associated with maternal hand-washing with soap and water or antiseptic prior to the handling of their neonate was 55.8% (19.9 − 8.8 / 19.9 = 55.8%).
Stratified analyses were conducted to evaluate the presence of confounding or effect modification. Potential confounders were identified from various maternal, infant and care-practice covariates. Some of these covariates have a recognized association with neonatal mortality or neonatal infection (e.g. birthweight, gestational age, cord cleansing with chlorhexidine (21
)). Binomial regression with a log link function was used to model the RR of these covariates with neonatal mortality. Similar models were created to estimate the association of these covariates with birth attendant and maternal hand-washing behaviors.
There was no evidence for effect modification of the association of hand-washing behavior on risk of mortality by sex or treatment group assignment. However, hand-washing behaviors tended to have larger effects on mortality among infants with indicators of higher underlying risk such as Madeshi ethnicity, low birthweight, preterm birth, low maternal education, and those without a latrine in the household, although the strength of evidence for interaction was only modest ().
Unadjusted Association of Hand-Washing Behavior and Neonatal Mortality by Selected Characteristics of the Population
Adjustment for a number of potentially confounding variables including birth weight (LBW), gestational age, mother's age, receipt of colostrum, breastfeeding initiation time and treatment groups did not materially change the association of birth attendant hand-washing and neonatal mortality (adjusted RR=0.80, 95% CI: 0.65−0.98). After adjusting for these same covariates, the magnitude of the association of maternal hand-washing and neonatal mortality was reduced from a 60% reduction to a 44% reduction (adjusted RR=0.56, 95% CI: 0.38−0.82). Similarly, the strength of the association with combined birth attendant and maternal hand-washing was reduced from a 56% to a 41% reduction in mortality (adjusted RR=0.59, 95% CI: 0.37−0.93).
In addition, RRs were calculated conditioned on survival of infants until days 2, 3 and 7 (). This was done to evaluate the impact of each hand-washing exposure after the exclusion of early deaths. It was initially hypothesized that excluding these deaths would show a greater impact of birth attendant hand-washing relative to the impact of maternal hand-washing, because the majority of very early neonatal deaths may be due to causes that might not be readily impacted by hand-washing, such as birth asphyxia, prematurity or congenital abnormalities. However, this was not supported by the data as there was little change in the relative risks after removing these early deaths (). There was also no substantive change in the relative risk of death associated with maternal hand-washing as early deaths were excluded ().
Adjusted Associations of Birth Attendant and Maternal Hand-Washing with Neonatal Mortality Beginning at Selected Times Since Delivery