We undertook a primary data analysis using data from DONA International. The population was composed of women who received doula care provided by members of DONA International. Data on DONA doula-supported births came from DONA’s Birth Doula Data Collection Form. Data was submitted by DONA International doulas, including certified doulas, certification candidates, and non-certified doulas. Doulas used one form per birth and either mailed or faxed their completed form to DONA International, Chicago, Illinois. The data collection form has 15 categories, including birthing demographics, place of birth and type of healthcare provider, pregnancy and labor and birth information, and infant outcomes. It includes the name and zip code of the doula, but there is no identifiable information about birthing women. Doulas are not required to complete Birth Doula Data Collection Forms.

DONA volunteers entered the data from the Birth Doula Data Collection Forms into SPSS statistical software to create a dataset (N=33,247). The authors of this article verified the accuracy of the data entry and analyzed a portion of the DONA dataset about doula-supported US births.

The following exclusion criteria were applied: (1) international zip codes (n=1831); (2) missing doula last names and/or zip codes (n=232); and (3) duplicate entries (n=1686). Further exclusion criteria were applied to obtain a sample of birthing individuals: multiparous birth (n=12,160), complicated pregnancy (defined on the collection form as the presence of gestational diabetes, pregnancy-induced hypertension, gestation <38 weeks or >42 weeks, high-risk, and “other” reasons) (n=4361), and method of birth (unknown, vaginal birth after cesarean, and planned cesarean) (n=302). After exclusion and inclusion criteria were applied, 20,572 observations were removed from the dataset (Figure 1).

Delimitation began with the number of DONA International Birth Doula Data Collection Forms. Final analyses were limited to women who gave birth with a DONA International doula in the United States.

We derived variables from the DONA Birth Doula Data Collection Form. Primary outcomes of interest were characteristics related to doulas (source of doula client referrals, payment source for doula services), doula-supported birthing women (mother’s age, childbirth education, breastfeeding initiation), childbirth settings and providers, and obstetric outcomes (induction, Pitocin or other augmentation, artificial rupture of membranes (AROM), intravenous (IV) fluids, epidural administration, and method of birth). To examine the prevalence of the characteristics mentioned above, we applied an urban-rural classification to illuminate the subtle underlying geographic or demographic patterns.

Descriptive statistics were performed to examine the number of doula-assisted births per 10,000 births by state. We calculated mean maternal age and standard deviation (SD), percentage of all births for characteristics related to doulas, doula-supported birthing women, childbirth settings and providers, and obstetric outcomes. We used t tests or Pearson’s chi-squared test (X²) to compare urban to rural doula-supported births. We used rural-urban commuting area codes to derive urban versus rural residence based on zip code.¹⁹ Analyses were carried out using Stata version 13.0 (StataCorp). Missing data occurred in less than 3% of outcome variables. Therefore, we used complete case analysis to deal with missing data.

In this analysis of a national sample of doula-supported births, women experienced lower rates of certain interventions compared to national statistics for deliveries in the United States during the same time period. Compared to the average overall epidural rate (65.5%) in the general population for all births (2004 through 2012, years of available data), the rate in the sample (52.3%) was lower.²⁰ While there is controversy over whether epidurals increase the risk of cesarean births, there is evidence that they can lead to an increased likelihood of other medical interventions.^21–23 Additionally, the rates of AROM have been increasing over time (34.0% in 2010), while the analysis of the sample suggests that only about 1 in 4 doula-supported birthing women had this intervention.²⁴ The reduced rates of these interventions could be due to a variety of factors, including the continuous support provided by the doulas as well as education on a variety of coping methods for pain.

The United States has a high cesarean birth rate of about 1 in 3 births.²⁵ Compared to the low-risk cesarean birth rate in the general population, the cesarean rate was lower in the sample (26.0% vs 18.2%, respectively) during the same period.²⁶ The lower rate is consistent with evidence from previous research that continuous labor support (ie, a doula) is associated with a decrease in the likelihood of cesarean births.^7,10 This could be secondary to the increased advocacy that doulas provide to their patients when communicating with the medical team or that those with doula-supported births had higher rates of child education classes prior to birth.

The failure of intravenous synthetic oxytocin infusion for the treatment of labor arrest is the leading indication for primary cesarean births, making up one-third of all cesarean births in the United States.²⁷ Before 2005, data regarding labor augmentation was not reported in the National Vital Statistics; therefore, a comparison to the sample rate (34%) in this study was not possible. However, it can be inferred that this rate may be higher in the sample given the lower cesarean rates, which supports the American College of Obstetricians and Gynecologists’ (ACOG) recommendation to attempt synthetic oxytocin infusion prior to cesarean section for arrest of labor.²⁸ The rate of forceps/vacuum birth in the sample (8.6%) was twice that as for the general population of all births (4.74%) for the same period.²⁹ The percentage of instrumental births was 9.0% or more before 1998, but as the cesarean rate increased in the general population, the percentage of instrumental births decreased.²⁹

Compared to the average induction rate in the general US population for all births during the same period as the sample, the sample’s induction rate did not differ (23.3% vs 23.2%).³⁰ Underreporting of widespread hospital induction practices in the United States might explain why the induction rates in both the general US population and in the doula-supported population are similar.

While breastfeeding rates in the United States have risen in recent years and are now stabilized around 80%, there is an indication that doula care supports near-universal breastfeeding initiation, especially for underserved women.³¹ Given this additional encouragement and likely additional education, the sample shows that 8 out of 10 doula-supported mothers did initiate breastfeeding after giving birth, which is similar to the national rate.³¹

Despite the demonstrated benefits of doula care, the majority (52.9%) of doulas in the sample reported that the women they supported during birth were referred through means other than a hospital or DONA International. Known referral sources are likely associated with friends, family members, childbirth educators, prenatal practices, organizations, and/or forms of media. These data suggest a heightened need for hospitals and obstetricians to provide more information about doulas to their patients.

The majority (49.7%) of doulas reported compensation through self-pay clients, and a small minority (1.3%) reported third-party reimbursement. Similarly, a 2002 study found that the majority (80%) of doulas reported self-pay clients, and a minority (10%) reported third-party reimbursement.⁹ The high rate of doulas reporting compensation through self-pay, rather than by third-party reimbursement, suggests there may be a lack of insurance coverage for doula services. Insurance coverage plans vary by state. Twenty-one states had ≥21 doula-supported births per 10,000 births, while the remaining states had ≤20 doula-supported births per 10,000 births. Overall, this suggests that lack of state insurance coverage of doula care may be a barrier to use of a doula during birth.

We observed that certain regions of the country might have higher rates of doula usage. Furthermore, we noted that births by rural women were less likely to be attended by an obstetrician/gynecologist, likely reflective of the shortage of obstetricians/gynecologists in rural areas and lack of hospitals in those regions that take obstetric patients.^32,33 There are significantly higher rates of doula-supported births in urban areas, suggesting that there is potentially a heightened need for doula services in rural areas. Given that insurance coverage is a likely barrier to seeking doula services, and because Medicaid pays for nearly half of all rural births, insurance coverage of doula-supported births would likely be able to reduce this limitation in the rural population.³⁴

This study had several methodological limitations. Because there was no control group, meaningful conclusions about the impact of doula-supported care during labor and childbirth were not possible. Information about doula-supported births came from only one doula certifying organization; therefore, trends over time given the voluntary sample are reflective only of what DONA data were available. Although doulas across the United States received the same type of comprehensive training, we could not account for hospital practice variations that might have impacted obstetric care interventions. Additionally, doula training courses are cost prohibitive for many, especially certification and additional expenses associated with breastfeeding classes, reading materials, and organization membership, such that many of the doulas contributing to these data were likely from socioeconomically stable backgrounds.

As noted earlier, reporting births using the Birth Doula Data Collection Form is voluntary. We did not know how many doula-supported births went unreported and cannot know whether there were differences in outcomes between doulas who reported and those who did not report births. Information on the data collection form relied on the self-report of doulas.

Although we had information about maternal race, we did not have information about maternal ethnicity, educational attainment, insurance type, or prenatal care, which hampered our ability to assess disparities in subgroups of the sample population.

We were unable to assess how racially and ethnically diverse women have access to doula services. Extant literature suggests that increased access to doula care, especially in underserved communities, can improve a range of health outcomes for mothers and infants. For example, continuous labor support is especially vital for women of color. Studies have shown that Black women in the general population experience higher rates of pre-term birth, cesarean birth, low birth weight infants, and infant mortality.¹⁸ These outcomes have been ameliorated with doula-supported care.^18,35 Evidence also suggests that doula support may lower maternal stress and enhance women’s self-efficacy regarding their pregnancy and the ability to manage labor.^4,7,13 Given that racial concordance is known to impact health outcomes with Black patients citing poorer communication, information-giving, patient participation, and participatory decision-making than White patients, this study does not adequately cover different races and socioeconomic backgrounds.^36,37