Home births attended by certified nurse midwives (CNMs) make up an extremely small proportion of births in the United States (<1.0%) and are not supported by the American College of Obstetrics and Gynecology (ACOG). The primary objective of this analysis was to examine the safety of certified nurse midwife attended home deliveries compared with certified nurse midwife in-hospital deliveries in the United States as measured by the risk of adverse infant outcomes among women with term, singleton, vaginal deliveries.
United States linked birth and infant death files for the years 2000 to 2004 were used for the analysis. Adverse neonatal outcomes including death were determined by place of birth and attendant type for in-hospital certified nurse midwife, in-hospital ‘other’ midwife, home certified nurse midwife, home ‘other’ midwife, and free-standing birth center certified nurse midwife deliveries.
For the 5-year period there were 1 237 129 in-hospital certified nurse midwife attended births; 17 389 in-hospital ‘other’ midwife attended births; 13 529 home certified nurse midwife attended births; 42 375 home ‘other’ midwife attended births; and 25 319 birthing center certified nurse midwife attended births. The neonatal mortality rate per 1000 live births for each of these categories was, respectively, 0.5 (deaths=614), 0.4 (deaths=7), 1.0 (deaths=14), 1.8 (deaths=75), and 0.6 (deaths=16). The adjusted odds ratio (95% confidence interval) for neonatal mortality for home certified nurse midwife attended deliveries vs in-hospital certified nurse midwife attended deliveries was 2.02 (1.18, 3.45).
Deliveries at home attended by CNMs and ‘other midwives’ were associated with higher risks for mortality than deliveries in-hospital by CNMs.
A return to more natural surroundings and to less medicalization of the birthing process have been advocated by some as a means of providing a more satisfying experience to women as well as providing substantial cost savings.1, 2 Currently, the only Western nation doing a substantial number of home births (∼30% of all births) is the Netherlands.3 The Royal College of Obstetricians and Gynecologists in conjunction with the Royal College of Midwives in the United Kingdom, ‘support the provision of home birth services for women at low risk of complications.’4 The Society of Obstetricians and Gynaecologists of Canada, ‘recognizes and stresses the importance of choice for women and their families in the birthing process.’5 The American College of Obstetrics and Gynecology, however, suggests that, ‘the safest setting for labor, delivery, and the immediate postpartum period is in the hospital, or a birthing center within a hospital complex’ and ‘does not support programs that advocate for, or individuals who provide, home births.’6
As the percentage of in-hospital births in the United States moved upward from 36.9% in 1935 to over 99% of births in 1973 the number of deliveries attended by certified nurse midwives (CNMs) has increased to 306 377 (7.4% of all births) in 2005.7, 8 The majority of the births attended by CNMs, however, are within a hospital setting (96.6%) whereas only 1.3% occur in a home setting.8 Although the data on the cost-savings of in-home births are extremely limited, one study suggests the cost of home births to be approximately one-third of that of in-hospital costs.1
In the debate of the virtues of home birthing the most important factor to consider is the risk for adverse outcome in both the infant and mother. Comparisons of outcomes for infants by site of birth vary by study design. Most studies examining neonatal mortality among pregnancies with planned home deliveries by certified midwives have found either no difference or lower neonatal mortality rates than among in-hospital births.9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19 Several studies using less strict criteria for screening pregnancies for planned home birth or with less strict criteria for the certification of the attendant at delivery, however, have reported an increased risk of neonatal mortality for home births.20, 21 The primary objective of this analysis that was limited to term (37 to 42 weeks gestation), singleton, vaginal deliveries was to compare various adverse outcomes among infants delivered in-hospital by CNMs to the outcomes of infants delivered at home by CNMs in the United States for the years 2000 to 2004. The rationale for limiting the analysis to term, singleton, vaginal deliveries were to examine the outcome of the birthing process in a population that would provide the best candidates for home delivery.
United States linked birth and infant death certificate files from the National Center for Health Statistics for the years 2000 to 2004 were used.22 These files contained maternal demographic characteristics, gravidity, medical complications of pregnancies, labor and delivery complications, labor induction, site of delivery, attendant type at delivery, method of delivery, and infant characteristics. Demographic variables in the analysis included maternal race (Hispanic, non-Hispanic Black, non-Hispanic White, American Indian/Alaskan Native, and Asian/Pacific Islander), maternal age, and maternal education. Files were created using SAS that extracted all survivors and deaths with a ‘best estimated gestational age’ between 37 and 42 weeks, and limited the births used in this analysis to vaginal deliveries of singleton pregnancies.
There are 18 medical diagnoses and 16 complications of labor and delivery listed on the standard birth certificate (see Appendix of Ref.23 for specific diagnoses and complications). The presence of one or more of the diagnoses or complications was recorded in an indicator variable for complications. There are 22 congenital anomalies listed on the standard birth certificate. The presence of one or more of these diagnoses was recorded in an indicator variable for the presence of a congenital anomaly. The adverse neonatal outcomes selected for analysis and reported on the standard birth certificate included anemia at birth, injury at birth, hyaline membrane disease, need for mechanical ventilation at 30 min or for >30 min, meconium aspiration, and a 5 min Apgar score <4. neonatal death (0 to 27 postnatal days) was obtained from the linked death certificate file.
Chi-square tests for general association were used for comparing the distributions of nominal variables. Logistic regression models were used to determine odds ratios and 95% confidence intervals for the adverse neonatal outcomes comparing site of delivery and attendant type to hospital births attended by CNMs as the reference group. Maternal age, education, parity, race, the presence of one or more medical or labor risk factors, the presence of a congenital anomaly, gestational age, and sex of the infant were included in the full models. All analyses were carried out using SAS.24 An arbitrary P-value of 0.05 was designated to indicate statistical significance.
There were 1 335 471 term, singleton, vaginal births available from the 5-year period that were used in the analysis. Ninety-three percent of the births were in-hospital births attended by CNMs, 1% were in-hospital births attended by ‘other’ midwives, 2% of births occurred in a free-standing birthing center attended by CNMs, 1% were in-home deliveries attended by CNMs, and 3% were in-home deliveries attended by ‘other’ midwives (Table 1). Characteristics of in-home deliveries attended by CNMs compared with in-hospital certified nurse midwife attended deliveries included a higher percentage of women that were >35-years old (19.6% vs 7.5%); a higher percentage of women with 16 years of education (35.8% vs 20.0%); more than one previous pregnancy (83.6% vs 67.1%); a higher percentage of non-Hispanic White births (89.4% vs 53.1%); a lower percentage of medical or labor risk factors (26.3% vs 37.9 %); and a higher percentage of pregnancies with gestations >40 weeks (25.1% vs 19.1%).
Using in-hospital certified nurse midwife attended deliveries as the reference group, in-home certified nurse midwife deliveries had higher risk of mortality (ORadj=2.02). In-home ‘other’ midwife deliveries also had a higher risk of mortality (ORadj=3.63), Table 2. Among other adverse neonatal outcomes, when compared with in-hospital certified nurse midwife deliveries, only the birthing center certified nurse midwife deliveries were at higher risk of anemia, ORadj=2.19. The risk for hyaline membrane disease was lower at birthing centers and for home deliveries by ‘other midwives.’ The risk for birth injury was lower in all situations except for in-home certified nurse midwife deliveries where there was no difference from the reference group. The risk for short-term mechanical ventilation was higher for hospital ‘other’ midwife attended deliveries, whereas the risk for mechanical ventilation for >30 min was increased for home ‘other’ midwife deliveries. The risk for seizures was increased for home deliveries attended by ‘other’ midwives (ORadj=3.15). The risk for a 5 min Apgar score <4 was greatest among home deliveries by CNMs and ‘other’ midwives (ORadj=7.83 and 3.39, respectively).
Congenital anomalies were the most common cause of neonatal death for in-hospital and home certified nurse midwife deliveries (41.7% and 42.9%) followed in prevalence by neonatal deaths attributed to conditions arising in the perinatal period (24.2% and 42.8%, respectively).
In examining the medical risk factors and labor or delivery complications with prevalences of >1 per 1000 live births, it was apparent that the pregnancies with the more significant risk factors, for example, diabetes, pregnancy-induced hypertension, abruptio placenta, fetal distress, and ‘other’ complications, were delivered in the hospital (Table 3). Of concern was the higher prevalence of excessive bleeding, precipitous deliveries, and prolonged rupture of membranes in the home deliveries.
Implementing and expanding the role of home deliveries as a potential cost-saving system in the United States has been a challenge because of concern about safety issues associated with the outcomes of home births as well as the tension between the medical establishment and home birth providers.6, 25 Among the studies that focus on low-risk deliveries that fall within certain criteria to be carried out in the home environment, equivalent or lower perinatal and neonatal mortality rates are reported.15, 16, 17 This is in contrast to the less restrictive studies that report a higher mortality rate.20, 21 Given that the provider of care in a home delivery should be a certified nurse midwife or equivalent, that the woman should not have medical risk factors that complicate the pregnancy, and that the development of the pregnancy has been such that no complications of the pregnancy have evolved, the Netherlands and the Royal College of Obstetricians and Gynecologists/Royal College of Midwives support the concept of offering women the option of home births.4 In contrast, the American College of Obstetricians and Gynecologists (ACOG), ‘does not support programs that advocate for, or individuals who provide, home births.’6
As other reports have suggested, this analysis shows that the population choosing to deliver at home in the United States for the period from 2000 to 2004 is more likely to be older, better educated, non-Hispanic White, multiparous women.7, 26, 27 This analysis selected a relatively low-risk population to examine for neonatal mortality. Women delivering single fetus term pregnancies by the vaginal route have low neonatal mortality risk.28 The fact that the deliveries in this analysis were completed at the site from which they are reported on the birth certificate biases the analysis toward the lowest risk population. One is left to assume that either the birth was intended to occur at the site reported, or that the opportunity to transfer was missed and the birth occurred emergently. Assuming that births occurring at home were intended and that any laboring women who had intended on a home birth but had encountered problems had been transferred to a hospital setting suggests that the home deliveries reported should be at very low risk. Despite the low absolute risk of neonatal mortality for this select group of home deliveries (1.0 to 1.8 per 1000 live births), the adjusted odds ratio for neonatal mortality for women delivering at home by CNMs compared with deliveries in-hospital by CNMs was significantly increased, ORadj=2.02 (1.18, 3.45). As pointed out by other reports, the attendant type at the home birth does appear to be of great importance in the determination of risk.9, 29 Those women delivered at home by ‘other’ midwives had a higher point estimate for an adjusted odds ratio for neonatal mortality (ORadj=3.63) than women delivered at home by CNMs (ORadj=2.02). In a separate logistic model with home deliveries by CNMs as the reference group, the adjusted odds ratio for ‘other’ midwife attended home deliveries, however, did not attain statistical significance (ORadj=1.71, 95% CI=0.96, 3.04).
Other neonatal outcomes of interest in this analysis that showed significant increases in risk for home births included the need for mechanical ventilation for >30 min among women delivered by ‘other’ midwives, ORadj=1.63 (1.38, 1.94); and the risk for neonatal seizures, ORadj=3.15 (2.25, 4.41). For home deliveries attended by both CNMs and ‘other’ midwives the risk for 5 min Apgar scores <4 was significantly increased compared with in-hospital deliveries by CNMs, ORadj=7.83 and 3.39, respectively. This constellation of adverse outcomes, that is, the need for mechanical ventilation, neonatal seizures, and low Apgar scores could reflect the risk for emergent asphyxiating conditions immediately after birth and the difficulty in dealing with this issue in the home environment. Among the home deliveries attended by CNMs, 3 of the 14 deaths (21%) were attributed to birth asphyxia, whereas 10 of the 75 deaths (13.3%) associated with home deliveries by ‘other’ midwives were so attributed. In contrast, 35 deaths of the total of 614 deaths (5.7%) were attributed to birth asphyxia among the in-hospital deliveries attended by CNMs and only 1 death of 16 (6.3%) in a free-standing birth center.
A large proportion of the deaths arising among the home births were attributable to congenital anomalies (42.9% for certified nurse midwife deliveries and 45.0% for ‘other’ midwife deliveries). In-hospital deliveries attended by CNMs also had a large proportion of neonatal deaths attributable to congenital anomalies as the underlying cause of death (42.9%). Despite eliminating these deaths/births from an analysis the risk of neonatal mortality remained significantly elevated for home deliveries by both attendant types compared with in-hospital certified nurse midwife deliveries.
There are a number of limitations to this study. A major shortcoming is the lack of information on maternal outcome associated with site of delivery and attendant type. Such information will be vital to overall decisions made about the safety of home births and those attended by CNMs. As with all studies using vital statistics data there is concern for the validity of some of the information recorded. Demographic characteristics, methods of delivery, and infant characteristics appear to be reliably reported on birth certificates.30, 31 Reporting of pre-existing maternal medical conditions and complications of pregnancy, however, appears to be significantly under-reported.32 This study is also unable to provide information on the intention to deliver at home and the transfer rate incurred for those pregnancies that develop complications during the course of labor. Transfer rates from heterogeneous populations of women range from 11 to 43%.16, 17, 18 Urgent intrapartum or postpartum referrals in a well screened population appear to be less frequent (3.6%).17 Neonatal mortality among Dutch women cared for by nurse midwives, which were urgently referred, however, were reported to be as high as 3 per 1000 live births.17 As stated earlier, this analysis of births in the United States for the period 2000 to 2004 only examines those home births that remained at home to deliver by intent or through the missed opportunity to transfer. Deliveries that might have contributed to a higher mortality or morbidity rate for home deliveries may have been transferred, thus the mortality and morbidity estimates presented in this analysis may underestimate the true risk.
In summary, deliveries at home by CNMs or ‘other’ midwives have a higher risk of adverse outcomes than in-hospital deliveries attended by CNMs. Women electing to deliver in the home environment should be aware of the higher risk.
I gratefully acknowledge the review of this manuscript and suggestions made by Dr Martin G Myers and Dr George Saade. Part of this analysis was accepted for presentation in poster form at the Annual Meeting of the Pediatric Academic Societies in Baltimore, MD, in May 2009.