Effect of cleansing the birth canal with antiseptic solution on maternal and newborn morbidity and mortality in malawi: clinical trialBMJ 1997; 315 doi: http://dx.doi.org/10.1136/bmj.315.7102.216 (Published 26 July 1997) Cite this as: BMJ 1997;315:216
- Taha E Taha, field directora,
- Robert J Biggar, senior investigatorb,
- Robin L Broadhead, professor of paediatricsc,
- Laban A R Mtimavalye, professor of obstetrics and gynaecologyc,
- Paolo G Miotti, associate scientista,
- Aafke B Justesen, obstetricianc,
- George N Liomba, associate professor of pathologyc,
- John D Chiphangwi, professor of obstetrics and gynaecologyc
- a Infectious Diseases Programme, Department of Epidemiology, School of Hygiene and Public Health, Johns Hopkins University, Rm E6011, 615 N Wolfe St, Baltimore, MD 21205, USA
- b Viral Epidemiology Branch, National Cancer Institute, Bethesda, MD 20852, USA
- c College of Medicine, University of Malawi, Blantyre, Malawi
- Correspondence to: Dr Taha
- Accepted 15 April 1997
Objective: To determine if cleansing the birth canal with an antiseptic at delivery reduces infections in mothers and babies postnatally.
Design: Clinical trial; two months of no intervention were followed by three months of intervention and a final month of no intervention.
Setting: Queen Elizabeth Central Hospital (tertiary care urban hospital), Blantyre, Malawi.
Subjects: A total of 6965 women giving birth in a six month period and their 7160 babies.
Intervention: Manual wipe of the maternal birth canal with a 0.25% chlorhexidine solution at every vaginal examination before delivery. Babies born during the intervention were also wiped with chlorhexidine.
Main outcome measures: Effects of the intervention on neonatal and maternal morbidity and mortality.
Results: 3635 women giving birth to 3743 babies were enrolled in the intervention phase and 3330 women giving birth to 3417 babies were enrolled in the non-intervention phase. There were no adverse reactions related to the intervention among the mothers or their children. Among infants born in the intervention phase, overall neonatal admissions were reduced (634/3743 (16.9%) v 661/3417 (19.3%), P<0.01), as were admissions for neonatal sepsis (7.8 v 17.9 per 1000 live births, P<0.0002), overall neonatal mortality (28.6 v 36.9 per 1000 live births, P<0.06), and mortality due to infectious causes (2.4 v 7.3 per 1000 live births, P<0.005). Among mothers receiving the intervention, admissions related to delivery were reduced (29.4 v 40.2 per 1000 deliveries, P<0.02), as were admissions due to postpartum infections (1.7 v 5.1 per 1000 deliveries, P=0.02) and duration of hospitalisation (Wilcoxon P=0.008).
Conclusions: Cleansing the birth canal with chlorhexidine reduced early neonatal and maternal postpartum infectious problems. The safety, simplicity, and low cost of the procedure suggest that it should be considered as standard care to lower infant and maternal morbidity and mortality.
Neonatal and maternal postpartum morbidity and mortality due to bacterial infections are substantial in developing countries. In areas with a high prevalence of HIV infection, such as sub-Saharan Africa, the HIV/AIDS epidemic has further increased paediatric and maternal morbidity and mortality.1 2 In countries poor in resources, where many women do not receive antenatal care early in pregnancy,3 simple and low cost methods to control infectious complications of the newborn babies and their mothers are greatly needed.
We conducted a large trial of cleansing the birth canal with chlorhexidine at Queen Elizabeth Central Hospital in Blantyre, Malawi. The primary objectives of this trial were to determine the effects of this cleansing on perinatal transmissions of HIV and other infections. Chlorhexidine is a well tolerated broad spectrum antiseptic which has been shown to reduce neonatal morbidity due to group B streptococcus4 5; it is capable of neutralising HIV.6 The cleansing procedure at delivery did not significantly reduce the transmission of HIV from mother to child, except when membranes had been ruptured for more than four hours.7 We report the effect of cleansing the birth canal on neonatal and maternal morbidity and mortality.
Women giving birth at Queen Elizabeth Central Hospital from June through November 1994 were studied. The protocol and consent forms were approved by institutional review boards in Malawi and the United States. Consent for washing the birth canal was sought when the woman was admitted to hospital during the intervention months. For all women, consent to be tested for HIV and enrolled in the study was obtained after delivery and before discharge from the hospital, when the woman was able to make decisions in private. The details of the study design and procedures have been described previously.7 Individual randomisation into intervention and non-intervention was not possible in a busy labour room (about 15 000 deliveries a year), so the intervention and control groups were enrolled in the trial in time periods. During the first two months of the trial (June and July) there was no intervention. In the subsequent three months, August through October, the intervention was introduced. In November, the chlorhexidine solution was removed from the labour room and another control group was enrolled.
Women delivering in the non-intervention months received conventional delivery procedures as routinely practised in this hospital. This consisted of cleaning the external genitalia when conducting vaginal examinations, using sterile gloves with a solution of Savlon (cetrimide 1% and chlorhexidine gluconate 0.1%). There was no attempt to clean the interior mucosal surfaces of the birth canal. Also, after delivery, the babies were not washed with any solution. The Savlon solution was not available in the labour room during the months of the intervention.
The intervention consisted of manual cleansing of the birth canal with a 0.25% chlorhexidine gluconate solution in sterile water. Chlorhexidine was chosen on the basis of safety and its reported effects on group B streptococcus and HIV.4 6 A nurse-midwife wrapped a cotton swab soaked in chlorhexidine solution on the examining fingers and wiped in a clockwise manner the entire birth canal and the external genitalia.7 Vaginal examinations were done at admission and every four hours thereafter. Many women had only one (29.1%) or two (38.4%) washes because deliveries often occurred soon after admission.3 Contraindications to vaginal examination included preterm labour, second stage labour, and placenta praevia. Babies born during the intervention phase were wiped with pads soaked in 0.25% chlorhexidine immediately after delivery. Midwives, nurses, and doctors who wiped the birth canal or the baby wore sterile gloves.
Babies developing neonatal problems before discharge were admitted to the special care baby unit, and readmissions (after discharge) were admitted to the paediatric nursery. Neonatal admissions included infants sent to the special care unit or paediatric nursery. The diagnosis of neonatal sepsis was made by paediatricians on the basis of clinical criteria of temperature >38.0°C, poor feeding, and apnoea or irregular respiration. The clinical records for all babies who died in the hospital were reviewed by a paediatrician to verify causes of death.
The diagnosis of maternal postpartum infection was made by an obstetrician if the woman had frank puerperal sepsis or fever (temperature >38°C) and had any of the following: offensive vaginal discharge, infected lochia, infected episiotomy or caesarean section wound, retained products of conception, or secondary postpartum haemorrhage. All women enrolled in the study were tested for HIV by examining cord blood samples for HIV antibodies (Genetic Systems HIV Enzyme Immuno-Assay, Seattle, USA).
Statistical analysis included comparison of rates and calculation of relative risks and 95% confidence intervals. An intent to treat analysis was performed, for which women and children were assigned to intervention or non-intervention phase according to time of delivery. The intervention phase therefore included 379 women (10% of women enrolled in the intervention group) who did not receive the intervention owing to unavoidable clinical contraindications such as premature labour and placenta praevia. Duration of hospitalisation was based on dates of admission to and discharge from the hospital. Analysis of hospital stay was restricted to women who gave birth vaginally to singletons with vertex presentation. As there was a preponderance of normal deliveries, the distribution of data on hospital stay was skewed; we therefore rank ordered the observations and used a non-parametric statistical test (Wilcoxon).
The distribution of study participants and the characteristics of women and children enrolled in the trial are shown in tables 1 and 2. A total of 6965 women were enrolled; they had a total of 7160 babies. There were no complaints or complications related to the intervention in either the mother or the child. The staff had no difficulty in administering the intervention, which added little time to the routine vaginal examination.
The effects of the intervention on infant morbidity and mortality are summarised in table 3). During the intervention and non-intervention phases, 1295 infants were admitted to the hospital with neonatal problems. Among 3743 infants born during the intervention period, 634 (16.9%) were admitted, while among 3417 infants born during the non-intervention periods, 661 (19.3%) were admitted (relative risk=0.88, 95% confidence interval 0.79 to 0.97).
Neonatal sepsis was diagnosed in 4.6% (29/634) of the admissions during the intervention period compared with 9.2% (61/661) in the non-intervention period (0.50, 0.32 to 0.76). Among 1134 babies admitted to the special care baby unit within 48 hours of delivery, neonatal sepsis was diagnosed in 2.8% (16/571) of the admissions in the intervention phase and 6.4% (36/563) in the non-intervention phase (0.60, 0.40 to 0.91). For babies admitted to the special care baby unit or paediatric nursery after 48 hours of delivery, the rate of sepsis was also lower, but the difference was not significant: 19% (13/70) among intervention babies and 28% (25/91) among non-intervention babies (P=0.24).
Since fewer infants were admitted to the hospital during intervention months, we compared the rates of admissions for sepsis by month, using live births as the denominator. Figure 1) shows that the rates of admissions due to sepsis dropped immediately after the intervention was introduced and stayed substantially lower during the intervention compared with the non-intervention months (7.8 v 17.9 per 1000 live births; 0.43, 0.28 to 0.67).
Early neonatal mortality was lower during the intervention than the non-intervention periods (28.6 v 36.9 per 1000 live births; 0.78, 0.60 to 1.00). Mortality was 25% lower in August, when the intervention was introduced, than in July, when no intervention was in place (31.6 v 42.1 per 1000 live births; 0.75, 0.49 to 1.13). Likewise, mortality was 40% lower in October, when the intervention was still in progress, than in November, when the intervention was not being used (22.8 v 32.0 per 1000 live births; 0.71, 0.44 to 1.16). Table 4 shows the causes of death. Mortality from infectious causes was significantly lower during the intervention than the non-intervention phase (2.4 v 7.3 per 1000 live births; 0.50, 0.29 to 0.88). There were no significant differences in the other main causes of infant death.
Of 3635 women who gave birth during the intervention period, 107 (29.1 per 1000 deliveries) were admitted to the hospital with postpartum problems, compared with 134 of 3330 women (40.2 per 1000 deliveries) who delivered during the non-intervention months (0.73, 0.57 to 0.94; table 3). Postpartum infection was diagnosed in 6 (5.6%) of the 107 women who delivered in the intervention period, compared with 17 (12.7%) of the 134 women who delivered in the non-intervention period. With the number of deliveries as the denominator, the postpartum infection rates were 1.7 and 5.1 per 1000 deliveries, respectively (0.37, 0.13 to 0.82). Postpartum infections were reduced in women infected with HIV and those not infected.
The median duration of hospitalisation was two days (range 0-64 days for women given intervention and 0-55 days for women not given intervention). Among 2947 intervention and 2686 non-intervention women who had uncomplicated vaginal deliveries of singletons, the mean duration of stay was 48.7 hours during the intervention period and 50.2 hours during the non-intervention period (Wilcoxon P=0.008). The duration of stay for children was identical to that of their mothers.
During the intervention we found significantly fewer neonatal admissions, fewer cases of neonatal and maternal postpartum sepsis, lower early neonatal mortality rates, and shorter duration of hospital stay. These results indicate that cleansing the birth canal at delivery has important public health benefits. We speculate that colonisation of the birth canal with potential pathogens is reduced, resulting in fewer infectious complications. We were not able to document infections by laboratory means because of limited facilities, but a follow up study to isolate and test the sensitivity of common organisms is being planned. The presence of pathogens, particularly group B streptococcus, in the birth canal is a known predisposing factor for neonatal sepsis.4 5 8 9 10 11 In the pilot study of this trial during pregnancy, cervical and vaginal swabs for bacterial culture taken before and after a single birth canal cleansing with chlorhexidine showed that group B streptococcus and Staphylococcus aureus occurred less frequently.12
The reduction in neonatal sepsis during the intervention occurred mainly in babies admitted to the special care baby unit during the 48 hours after delivery. The beneficial effect may have been due to prevention of acute infections originating in the birth canal, such as early group B streptococcus infections. Reductions in the number of admissions for neonatal sepsis after 48 hours could be due to pathogens leading to late manifestations of sepsis.
There was an abrupt 25% reduction in early neonatal mortality with introduction of the intervention, and mortality increased when the intervention was stopped. Furthermore, only mortality due to infectious causes was significantly reduced. The reduction in overall mortality (8.3 per 1000 live births) beyond that explained by lower rates of infectious causes (4.9 per 1000 live births) may be due to underdiagnosis of sepsis, especially among deaths attributed to respiratory causes. Hospital data from the previous year showed no seasonal trend in neonatal mortality, but these data have limitations in comparability of the definition of early neonatal death and criteria for inclusions and exclusions. Rates of sepsis did not vary among babies born outside Queen Elizabeth Central Hospital but referred because of problems, suggesting lack of seasonal trends.
Alternative explanations for our results are unlikely. The study design allowed comparison of outcomes in intervention and non-intervention periods, as well as observations about changes in outcome related to the introduction or termination of the intervention. Comparison of data on women and children enrolled in the intervention and non-intervention phases (table 2) showed similar baseline characteristics, and because of our alternating scheme of enrolment, bias due to selection of participants is unlikely. More non-intervention than intervention women refused to give consent for HIV testing after delivery (table 1), but these women represented less than 2% (130/7959) of hospital deliveries. Perhaps the non-intervention women did not consent to HIV testing because they were approached about the study only after delivery, when they were eager to leave the hospital. It was not possible for women delivering in the non-intervention phase to receive the wash since neither the 0.25% chlorhexidine solution nor the sterile swabs were available in the labour room. The diagnoses of neonatal and maternal postpartum sepsis were made by hospital clinicians independent of the study. Hospital practices of admissions, readmissions, and discharges were the same during the intervention and non-intervention phases. Bias due to lack of blindness in the assessment of outcome is possible, but clinicians evaluating sepsis were not formally part of this study and information on characteristics of individual participants (such as enrolment in the study, whether the woman or baby has been washed, the number of washes, and the HIV status of the woman) was not available to clinicians. Cause of death of the child (before discharge from hospital) was abstracted from clinical records and verified by a paediatrician on criteria not based on the phase of the study. The modest reduction in hospital stay (3%) probably reflects inadequate measurement of duration.
The cleansing procedure was easily administered and took almost no extra time, and the cost of the chlorhexidine solution and cotton was less than US$0.10 per patient. None of the patients had complaints or complications related to the intervention. In a large Swedish study, however, a hypotensive episode occurred in one person (0.02%) exposed to chlorhexidine by lavage4 and with many years of prior occupational exposure to chlorhexidine. Repeated deliveries using this intervention modality might thus result in sensitisation to chlorhexidine on rare occasions.
The beneficial effects, safety, simplicity, and low cost of the intervention encourage its adoption to reduce maternal and neonatal morbidity and mortality.
We thank the Ministry of Health of Malawi and the staff of the Johns Hopkins University-Ministry of Health Project for their active and dedicated collaboration.
Funding: National Cancer Institute, National Institutes of Health, Bethesda, MD, USA. The support of the PAVE and HIVNET programs, Division of AIDS, the National Institute of Allergy and Infectious Diseases, has been critical in providing infrastructure and supplemental funding.
Conflict of interest: None.