To conclude the workshop, Catherine Spong and Zsakeba Henderson were invited to reflect on the evidence presented and highlight topics or methodology issues to consider for future research. This chapter summarizes their reflections. As a reminder, the intention of the workshop was not to reach consensus or make recommendations. The suggestions summarized here reflect only the personal observations and thoughts of two individual participants. Box 9-1 summarizes key points made by the two speakers. Also included in this chapter is a summary of closing remarks by Maxine Hayes and Ruth Lubic.
CONSIDERATIONS WHEN EVALUATING STUDIES ON BIRTH SETTINGS1
Catherine Spong offered some suggestions to help interpret the literature that was presented over the course of the 1.5-day workshop, particularly with respect to information presented on health outcomes (as summarized in Chapter 4), and to help plan a future birth settings research agenda. She noted that some of the issues she identified were issues in 1982 as well (IOM and NRC, 1982).
1This section summarizes information presented by Catherine Spong, M.D., National Institute of Child Health and Human Development, Rockville, Maryland.
- Catherine Spong identified several factors to keep in mind when evaluating the evidence presented at this workshop, especially the evidence on outcomes among different birth settings. These factors include: how outcomes are driven by the institutional norms and policies of a birth setting, regardless of type of birth setting; caregiver staffing and roles; types of patients studied; selection bias; and outcomes measured (e.g., is a 5-minute Apgar score enough?).
- Spong emphasized the need for more research on long-term outcomes, on women at increased risk, and on neural-immune connections in obstetric populations.
- Spong also stressed the importance of recognizing the limitations of available data and the limitations of generalizing research findings.
- Zsabeka Henderson noted there are many knowledge gaps still remaining, even though several additional birth setting data sources have become available since the 1982 Institute of Medicine (IOM) and National Research Council (NRC) report. In particular, birth certificate data do not capture planned home births transferred to hospitals or intended place of birth for either hospital or birth center births; very large datasets are needed to detect differences in perinatal mortality; and there is no uniform data platform to adequately compare birth settings.
- Henderson identified several key research needs discussed by workshop participants that could serve as a starting point for a future research agenda. These research needs range from evaluation of outcomes across birth settings to research on access to care in various birth settings.
- So that research can inform policy and practice, Henderson also identified several nonresearch gaps that need to be addressed. Most importantly, in Henderson’s opinion, and the most important take-home message of the workshop for her, is the need to improve interprofessional education, communication, and interaction.
Alternative Versus Conventional Settings
An important conclusion and major message for Spong from Ellen Hodnett’s presentation on alternative hospital settings versus conventional hospital settings was that alternative settings impact, and in many cases reduce, interventions.
Spong identified several factors to keep in mind when evaluating the evidence that Hodnett presented and other similar evidence: how outcomes are driven by the institutional norms and policies of the birth setting, regardless of type of birth setting; caregiver staffing and roles, including what types of caregivers are present and the timing of their care (e.g., Are caregivers present for only 8 hours at a time, or did they stay for 24-36
hours?); and the types of patients studied (e.g., the type of patient interested in one setting might be different than the type of patient interested in another setting).
UK Collaborative Group Studies
A key point for Spong from Jane Sandalls’s presentation on the UK Collaborative Group Studies was that low-risk women have very rare adverse perinatal outcomes. Other important messages were that women in their first pregnancy have higher risks than women who have had a prior successful pregnancy and that interventions occur more often in hospital settings.
In Spong’s opinion, factors to consider when evaluating the results from these and other similar studies include choice of birth setting, patient bias, and caregiver staffing and roles.
Process of Care
Spong said that a major component from Carol Sakala’s presentation on process of care was that the overall goal is fewer interventions.
Factors to consider when evaluating the evidence presented by Sakala and other similar evidence include institutional polices (e.g., the goal may be fewer interventions, but if there is a policy in place that all patients must receive a particular intervention, then all patients will receive it regardless of whether they need it), caregiver staffing and roles, selection bias, and outcomes measured (e.g., is a 5-minute Apgar score enough?).
Effect of the Built Environment
A significant point for Spong from Esther Sternberg’s presentation was that environment affects health, with both neuroendocrine and physiologic responses to what is happening in the environment impacting both mental and physical health.
An important factor to consider when evaluating the research presented by Sternberg and other similar research is that much of the data are from nonobstetrical populations. It is unclear how applicable the results are to obstetrical populations.
Spong identified several overarching factors to keep in mind when evaluating the evidence from these studies on birth settings and health outcomes, indeed when evaluating the evidence from any of the research described during the workshop or when planning future research:
- When evaluating any study, it is important to keep selection bias in mind. For example, women who plan home births typically have lower obstetrical risk and a later gestational age, desire fewer interventions, and prefer a specific care model.
- When interpreting cohort and historical studies, it is important to consider not just selection bias but also variation in outcomes measured and institutional policies.
- The availability of some obstetrical services is limited in some settings for certain types of conditions (e.g., vaginal birth after Cesarean deliveries), which can drive a woman’s decision to have her birth in one setting versus another.
- It is difficult to conduct randomized controlled trials in birth setting research; thus, few such studies have been done.
- Different studies define “low risk” differently. Even slight differences in how “low risk” is defined can impact how results are interpreted. Likewise with “normal.”
- Timing of delivery decision can impact how results are interpreted (e.g., whether it is decided in the first trimester versus the third trimester).
- Most studies report on short-term outcomes, for example Apgar scores, mortality, and hospital discharge. The field needs long-term outcomes. For example, how well does the child do in kindergarten? How well adjusted is the adolescent? How does birth setting impact the mother’s reproductive and gynecologic health?
- Most studies involve women at low obstetric risk. The implications of those study results for women at increased risk are unknown. Yet, some women at increased risk might want to have a home birth. Because they have not been included in most studies, those decisions need to be made very carefully.
- Understanding neural-immune connections in obstetric populations will be critical to moving the field forward.
Overriding all of these factors to consider when evaluating studies on birth settings is the importance of taking into account the limitations of available data. It will also be important to recognize the limitations of generalizing research findings. For example, again, it is unclear whether and how findings from studies conducted with low-risk women can be generalized to higher-risk women. Spong concluded, “There are many unanswered questions … there is much work to be done.”
REFLECTIONS ON WORKSHOP AND FUTURE RESEARCH NEEDS2
Learning without reflection is waste. Reflection without learning is dangerous.
Zsakeba Henderson remarked that while many of the research plans made in 1982 remain incomplete (IOM and NRC, 1982), some progress has been made. The present time represents a unique opportunity to build on this progress and chart out a new research agenda. Her remarks were based on the presentations and discussions throughout the workshop.
Childbirth Trends and Statistics: What Has Been Learned?
The landscape of births in the United States has changed significantly over the past 30 years, with the risk profile of women giving birth being very different now than it was then. Specifically, there have been substantial increases in births to women who are older (30 years and older); there are more births to women of Hispanic ethnicity; and more women are gaining more weight (greater than 40 pounds) during pregnancy.
Data also show that Cesarean deliveries in the United States, although they have risen significantly, nearly 60 percent from 1996 to 2009, have steadied in the past few years. There has even been a small decline, from 32.9 to 32.8 percent. Other successes include substantial decreases in low-birth-weight rate and the number and rate of triplet and higher-order multiple births.
Since 1900, the birthplace for most children born in the United States has shifted from the home to the hospital, with the rate of out-of-hospital births remaining fairly steady for decades until recently. The percentage of births outside the hospital increased rapidly from 2004 to 2010, mostly among non-Hispanic white women. Despite this recent rapid increase, the actual number of births outside the hospital is still very small, with only 1.2 percent of births in 2010. More out-of-hospital births occur among older, multiparous women with lower risk profiles. In 2010, 67 percent of out-of-hospital births were home births, and 28 percent were birthing center births. The majority of out-of-hospital births are planned (88 percent in 2010).
2This section summarizes information presented by Zsakeba Henderson, M.D., Centers for Disease Control and Prevention, Atlanta, Georgia.
Childbirth Trends and Statistics: Knowledge Gaps
Although more is known about where women are giving birth, Henderson observed that we still do not really have a good sense of national trends for intended place of delivery. Many states do not report the planning status of home birth, with only 31 states and the District of Columbia doing so, amounting to about 60 percent of U.S. births; and different states are at varying stages of implementing the revised birth certificate (although, by 2014, all states should be implementing it).
Nor do we have a good sense of trends in transfers from alternative settings to hospitals. It is not always possible to determine transfers from the birth certificate, as such reporting is not required in all states. One state, Oregon, was mentioned during the workshop as having added transfers from home to a hospital setting on their birth certificate. Henderson expressed hope that more states would do the same, given the likelihood that the U.S. Standard Certificate of Live Birth will not be revised again in the near future.
Yet another gap in our knowledge about childbirth trends and statistics stems from limiting reporting of birth attendants. Part of the problem is that those data are not being captured; another part of the problem is that states vary in terms of licensure and in terms of who is able to attend a birth.
Assessment of Risk in Pregnancy: What Has Been Learned?
Risk assessment in pregnancy is a very complex task. There is no clear definition of “low risk,” with different studies defining low risk in different ways. Moreover, risk is dynamic and subject to change. Additionally, risk to the mother must be balanced with risk to the fetus. More important than the actual risks, as presented during the workshop, is that risk perception varies between providers and patients, with providers and patients placing different values on different risks. Cultural views, women’s views, and structural conditions can also affect risk perception.
These challenges aside, Henderson remarked that overall absolute risk of adverse events in all birth settings is low.
Assessment of Risk in Pregnancy: Knowledge Gaps
Improved risk assessment in pregnancy will require uniform definitions of outcomes. Henderson noted there have been several efforts to develop these uniform definitions. Also needed are risk-assessment tools for maternal morbidity and mortality. Considerable attention has been focused on neonatal risk and levels of care, but not on maternal risk and levels of care.
Specifically, consistent “low-risk” criteria for maternal risk are needed, as are descriptors for maternal resources and levels of maternal care. Other gaps include incomplete knowledge about predictors of neonatal and maternal complications and incomplete knowledge about predictive triggers for elevation of care or transport.
More information is needed on the role of providers and the care system in contributing to risk. Henderson described her impression of the workshop participants when she first walked into the room: she knew who was a midwife and who was a physician, not based on appearance, but based on conversations. The dialogues were separate. She said, “There is a definite need for improvements in the interprofessional working relationships.”
Finally, Henderson observed that workshop participants had repeatedly mentioned that one of the reasons women are choosing home births is their perception of the risks associated with interventions. More work needs to be done on patient perception of care and thresholds for intervention in high-level care facilities.
Birth Settings and Health Outcomes: What Has Been Learned?
Alternative birth settings have been associated with less intervention, fewer complications, high transfer rates in some instances, and no differences in perinatal death rates. However, in one study, home, freestanding, and “alongside” midwifery units were associated with decreased obstetrical interventions, transfer rates of more than 20 percent, and increased neonatal risks for first pregnancies with home births (Brocklehurst et al., 2011). Another study, albeit controversial, also associated home births with increased neonatal mortality (Wax et al., 2010).
Researchers have also learned more about the process of care and that it does impact health outcomes, and that the built environment impacts neural-immune connections and health. Henderson noted that the last finding is true of any setting and encouraged more research on how environment affects patients and patient outcomes.
Birth Settings and Health Outcomes: Knowledge Gaps
Although much has been learned over the past 30 years about birth settings and health outcomes, there is still a great deal left to learn. In Henderson’s opinion, the field needs an evaluation of all birth settings, comparing women of equal risk across all settings. Henderson noted how workshop participants had highlighted the fact that there have been no randomized controlled trials of freestanding birth centers. And the one meta-analysis of home births relied on only one randomized controlled trial (Wax et al., 2010), with the remainder being observational studies. Henderson
remarked that although randomized controlled trials are considered the gold standard, it is difficult to randomize patients across all birth settings. Moreover, not all birth settings within the same category are the same. That is, not everyone’s home is the same; nor are all birthing centers the same. A comparison of health outcomes across birth settings is “definitely a gap … that needs more attention,” Henderson said.
Additionally, the field needs studies with consistent process and outcome measures. It is very difficult to compare studies when the methodologies and outcome measures are so different. Examples of research areas needing this kind of work include assessment of pain relief, effects of pain management on neonates, effects of successful breastfeeding, and physiologic and biochemical measures.
The field needs more studies on longer-term outcomes, that is, outcomes beyond the immediate neonatal period. Because of its implications for future health, also needed is more research on the developmental origins of health and disease. Lastly, the field needs more research on the optimal process of care.
Workforce Issues: What Has Been Learned?
Recognizing that discussion of workforce research needs was limited by who was present at the workshop (i.e., midwives, nurses, physicians) and that there are other members of the team not represented, Henderson identified some of what has been learned over the past 30 years about the workforce. Researchers have learned that supply trends are variable by profession; that the number of births to attendants is shifting, with increasing numbers of midwife-attended births both in and out of hospital settings; that there is state variability in who is licensed to do what; and that certain staffing models, including competent nursing staff and collaborative teams of care, contribute to improved patient outcomes.
Workforce Issues: Knowledge Gaps
Gaps in knowledge include the role of education and certification in quality of care; ideal staffing models to optimize care quality (i.e., composition of collaborative teams, provider ratios); impact of “missed nursing care” in out-of-hospital settings; how nurse staffing affects quality, safety, and cost of hospital-based care; and the impact of technology on workforce training needs and demand. With respect to ideal staffing models, Henderson noted that things are done very differently in the United Kingdom, where midwife-attended and doctor-attended births do not necessarily have labor and delivery nurses on their care teams.
Data Systems and Measurement: What Has Been Learned?
Since the 1982 IOM and NRC report, several additional data sources have become available to inform outcomes of birth settings: the 2003 U.S. Standard Certificate of Live Birth, linked birth certificate datasets (i.e., with discharge and Medicaid data), registries (e.g., Midwives Alliance of North America Stats, American Association of Birth Centers Uniform Data Set), data from payers, data from state and regional perinatal quality collaboratives, and data from professional organizations.
Additionally, the Center for Medicare and Medicaid Innovation (CMMI) Strong Start Initiative represents another opportunity to gain more information on outcomes, in particular outcomes related to preterm birth and the cost of care.
Data Systems and Measurement: Knowledge Gaps
While more types of data are available now than in 1982 (IOM and NRC, 1982), workshop participants identified several limitations to the data being collected: birth certificate data do not capture planned home births transferred to hospitals, intended place of birth is not captured on the birth certificate for either hospital or birth center births, very large numbers are needed to detect differences in perinatal mortality (i.e., there have been no randomized controlled trials of sufficient size), and there is no uniform data platform to adequately compare birth settings.
Cost, Value, and Reimbursement Issues: What Has Been Learned?
Considerable emphasis was placed during the workshop discussion on Medicaid, and “rightly so,” according to Henderson. Medicaid is the payer for 40 percent of U.S. births. CMMI is realigning incentives to reward providers for lower-cost, high-quality care. However, those incentives do not really help in situations where care is not covered by Medicaid—some states do not cover home births. State-to-state variability in coverage limits the ability to create a national agenda around this issue.
Some data were presented on the cost of care, with Washington State Medicaid expenditures for hospital-based Cesarean and vaginal deliveries being higher than for birth center and home births. More data are needed from other states and on a national level.
Cost, Value, and Reimbursement Issues: Knowledge Gaps
Gaps in knowledge include the lack of data from Medicaid managed care organizations, incomplete cost-comparison data (i.e., such data
may not include all costs associated with each birth setting), and lack of national-level cost data (i.e., because of state variability in reimbursement and state variability in linkage of Medicaid claims to vital records data).
Future Research Needs
Henderson identified several key research needs discussed by workshop participants that serve as a starting point for a future research agenda:
- Randomized controlled trials to evaluate outcomes in freestanding birth centers, outcomes in other birth settings (e.g., Snoezelen rooms, ambient rooms), and the impact of interventions in the hospital setting in terms of poor outcomes for patients. Henderson highlighted the need to consider all settings in these randomized controlled trials and not just focus on any one setting.
- An evaluation of organizational models of care across all settings.
- An examination of effective methods to transition care from out-of-hospital settings to the hospital.
- An examination of the impact of transfer on women and on care providers.
- A determination of predictors of neonatal and maternal complications.
- An evaluation of the potential unintended impact of intrapartum care processes.
- A cost assessment of birth settings.
- A cost-effective analysis of birth settings.
- Research on access to care in various birth settings (i.e., what is available to women in various communities throughout the United States?).
- An evaluation of continuity of caregiver.
- An evaluation of the experience of maternity care in various settings.
- Research on the effect of the environment on neural-endocrine-immune interactions and physiologic responses.
Henderson stated the purpose of research is to inform policy and practice, and in order to best inform policy and practice, there are other nonresearch gaps that need to be addressed as well. Workshop participants touched on several of these. Henderson emphasized the importance of maintaining and supporting the National Vital Statistics System. She reiterated what others had said about many providers not really understanding the importance of the data being collected on birth and death certificates (e.g., how those data are used). The quality of those data needs to be
improved, likewise with the quality of data on transfer to hospital care. Although transfer data are not something that can be immediately added to the National Standard Certificate of Live Birth, it is something that can be discussed and lobbied in preparation for future revisions.
Another important nonresearch, but research-related, area is the need for measurement and reporting of perinatal morbidity and mortality for all settings. It could take the form of either passive or active surveillance, or state-based review committees. Such a system is needed for all settings, not just the hospital setting. Additionally, there are needs for development of clear protocols for consultation and transfer of care; development of risk-assessment tools for maternal morbidity and mortality; and development of consistent policies for education, certification, and licensing of care providers. Related to the need for consistent policies around licensing is the need to address cost and reimbursement issues for care provided in out-of-hospital settings.
Most importantly, in Henderson’s opinion, and the most important take-home message of the workshop for her, is the need to improve interprofessional education, communication, and interaction.
There were several comments over the course of the 1.5-day workshop related to the reality that substantial gaps in knowledge remain even after more than three decades of research following the IOM and NRC (1982) research recommendations. In her closing remarks, Maxine Hayes, M.D., M.P.H., Washington State Department of Health, Olympia, Washington, reflected that there is more incentive and motivation today than there was following publication of the 1982 IOM and NRC report to move the research agenda forward. Moreover, more information about birth settings is available now than was available 30 years ago.
Hayes invited Dr. Ruth Lubic, a certified nurse midwife with a storied career, to offer some additional closing remarks. Lubic encouraged redefining “perinatal” to include conception, or even preconception, through the third year of life. “We are talking about families,” she said. “We are not talking about silos.” She remarked that birth centers have in fact been trying to conduct randomized controlled trials, but it is difficult to randomize women among birth settings. Finally, she called for more funding to do these studies, noting that most out-of-hospital care providers do not have access to the same funds that support hospital care providers (e.g., university funds).
In Hayes’ opinion, there are many ways that the information discussed at the workshop and summarized in this report can be used to move the research agenda, noting that the workshop does not provide recommenda-
tions but a statement of facts. It may be that an IOM committee is formed to examine in more detail a certain element, or that the Kellogg Foundation, the sponsor of this meeting, may want to fund a certain activity related to the information communicated here. Hayes urged states, in collaboration with federal agencies, to review their state-level policies, given the many components of obstetric care, even in hospitals, that could be improved. She also called for greater collaboration among the professions. More broadly, she stated that the unnecessary medicalization of birth has created fears about a natural process and that the whole culture of birth needs to be changed. Finally, she emphasized the importance of considering long-term outcomes and how what happens during the perinatal period impacts child health and learning.