Many reports have chronicled the large variation in US obstetric care with total cesarean rates ranging from 6.1% to 69.9% ; hospital rates of early elective delivery rates varying from 0% to 83% ; and third- and fourth-degree laceration rates swinging from 0.5% to 9.5% among delivery services.

In California, we also see significant variation within all levels of care, among university hospitals, among large urban facilities, and even among small rural facilities, for morbidity indicators as well as for process indicators such as failed labor induction. If we could only understand what drives such variation, we would be well on our way toward improving care for our entire birthing population and not just facility by facility, or so quality improvement science would suggest. There must be some objective criteria that establish a recipe for success.

Would that it be so easy. In this issue, Korst et al present a series of well-designed and carefully executed studies to search for this elusive goal. They first describe a comprehensive survey with 185 questions that covered hospital staffing (eg, all types of direct and support providers), clinical resources (eg, pharmacy and blood bank 24 hours a day 7 days a week, resources for severe obese patients), and patient care activities (eg, protocols for emergencies, drills, and educational programs).

The surveyors were persistent and they achieved a remarkable response rate of > 96%. Not surprisingly, they found that the 26 staff-model health maintenance organization facilities had 100% in-house obstetricians and 24 hours a day 7 days a week availability of specialists. The 26 teaching facilities were more varied, but the 187 community facilities with private practice providers showed great variation for most of the services. Clearly there is more opportunity for investigation among subdivisions of this population of hospitals that accounts for the large majority of maternity care in the United States.

But the overarching question remains: does the variation in maternity services and capabilities help explain the variation in performance and outcomes? The authors address this in a second study in which they examined the role of a key variable: will the presence of a laborist program reduce cesarean delivery rates and improve maternal morbidity? Surprisingly, for the 43 hospitals with laborists the answer was no on all counts. Here again, the authors did a careful job of adjusting for the variety of other factors that varied among the facilities and examined a wide range of outcomes.

Previous studies were much more encouraging on this count. Iriye and colleagues found that full-time laborists were associated with a 27% reduction in the cesarean delivery rate, whereas a community laborist model was not associated with any benefit at all. Most recently a single institution study of a switch from a private practice model to a staff-model with combined laborist and midwife care showed a nearly 50% reduction in both total and nulliparous term single vertex cesarean delivery rates.

How can we reconcile the current study with these very positive findings? I believe the answer lies in the limitations of the survey approach. It appears that all laborist programs are not the same; job descriptions vary: some may be there for emergencies, others to supervise women’s labor after cesarean deliveries, still others to oversee labor until the private attending physician appears, and still others are there to completely manage labor and delivery.

The two case studies showing great success were on the active end of this spectrum. In the survey study presented here, we do not know the full job description of the laborist: providing an opportunity for follow-up analysis. Indeed, there are a number of survey questions that retrospectively may have been more precisely asked. For example, almost all facilities answered that they had 24 hours a day 7 days a week blood bank coverage, but the real test may be how many units are available locally before needing to send out for additional supplies?

On the other hand, too precise a survey question about services may lead you down a false path. In the third article of the series, the authors seek to use their service availability questions to categorize hospitals into a locally developed system of maternal levels of care. A major criterion for their basic level of care was whether the hospital had the ability to perform a cesarean delivery birth within 30 minutes 100% of the time. Thirty-six percent of hospitals could not meet that standard.

Other criteria including 24 hours a day 7 days a week availability of pediatricians and radiologists excluded other hospitals so that a total of 66% of hospitals were not even considered basic level. This was certainly eye opening, but none of these criteria are in the newly released American College of Obstetricians and Gynecologists/Society for Maternal-Fetal Medicine maternal levels of care criteria set, so the applicability of this study going forward appears limited for levels of care. However, the study again illustrates the wide variation in maternity services.

Where do we go next in our quest to understand variation in maternal care? Broad categories like laborists: yes/no should yield to more nuanced questions about their exact roles in the labor and delivery department. Whereas structural factors are undoubtedly important, the roles of unit culture and staff attitudes toward childbirth appear to be very important for driving safety and reducing cesarean delivery rates. For example, attitudes toward trial of labor after cesarean delivery appear to influence primary cesarean delivery rates. These areas are in need of their own survey instruments and large-scale studies. But unfortunately, a key lesson that we have learned from these carefully crafted investigations is just how difficult it is to squeeze out answers to the question of variation in maternity care.