Objective : To compare midwifery and medical care practices and measure optimal perinatal out-comes using a new clinimetric instrument.
Design : Prospective descriptive cohort design. Setting : A large, inner city obstetric service with medical and midwifery services.
Participants : Three hundred seventy-ﬁ ve of 400 consecutively enrolled patients were participated (25 excluded due to extreme risk status or missing data); 92% were of minority race/ethnicity and 54% had less than a high school education. Of the 375 pa-tients, 179 received physician care and 196 received nurse-midwife care.
Main Outcome Measures : The Optimality Index-US was measured. Health record data were extracted and scored using the Optimality Index-US to summa-rize the optimality of processes and outcomes of care as well as the woman ’ s preexisting health status.
Results : Midwifery patients had more optimal care processes (less use of technology and interven-tion) with no difference in neonatal outcomes, even when preexisting risk was taken into account.
Conclusion : Even among moderate-risk patients, the midwifery model of care with its limited use of inter-ventions can produce outcomes equivalent to or better than those of the biomedical model. JOGNN,35, 779-785; 2006. DOI: 10.1111/J.1552-6909.2006.00106.x
Keywords : Midwifery — Obstetric outcomes — Optimality
Accepted: March 2006
When childbearing is viewed from the biomedical perspective, reproductive danger is placed in the fore-ground, and the normal processes of birth and related contextual issues are seen as less important to the
struc-ture or processes of clinical care. Support of those nor-mal processes is easily sacriﬁ ced when technology is believed to be superior in achieving a “ better ” birth out-come, especially when medical risk factors are present. How can perinatal care be most optimally deliv-ered? Optimality in perinatal health is deﬁ ned as “the maximal perinatal outcome with minimal interven-tion placed against the context of the woman ’ s social, medical, and obstetrical history ” ( Kennedy, 2006 , p. 763). The purpose of this study was to pilot test a new measurement of optima lity (the Optimality Index-US [ OI-US ]) by comparing nurse-midwife and physician care among women at moderate risk for poor pregnancy outcome.
Nurse-Midwifery Outcomes Research
Risk can be understood as the chance that loss or harm will occur, implying a higher than normal pos-sibility for a negative outcome. Medicine is appropri-ately focused on the reduction of biomedical risk. The perspective of nurse-midwives regarding risk is different. Even when risk factors are present, “ the vast majority of these births have good outcomes ” ( Rooks, 1997 ). Risk for harm is simply that the pos-sibility that injury or loss might occur. Within nurs-ing or midwifery practice, risk is not the sole basis on which to base clinical decisions.
Many researchers have assumed that women receiving midwifery care were at low risk for poor obstetric outcomes, and that midwives have excellent outcomes only because they care for low-risk women. This view is difﬁ cult to challenge because research has been conducted primarily with this population
Linking Obstetric and Midwifery Practice
With Optimal Outcomes
( Declercq, 1995; Rooks, 1997 ). The practice of using sam-ples of women at low risk in midwifery research has been driven by scientiﬁ c and contextual issues. First, by using “ low-risk ” samples, researchers can more easily account for confounding variables. Second, the samples reﬂ ected how midwifery practice is most often characterized in medical and popular literature.
any midwives care for women who are at
moderate risk for poor pregnancy outcomes
due to medical and obstetric problems or
because of sociodemographic variables
However, these samples have not accurately represented the characteristics of women cared for by midwives in the United States. Many midwives care for women who are at moderate risk for poor pregnancy outcomes ( Clarke, Martin, & Taffel, 1997; Declercq, 1995; Visintainer et al., 2000 ), either due to medical and obstetric problems or because of sociodemographic variables, such as age, race/ethnicity, ﬁ nancial status, geographic location, and migrant/immigrant status ( Mac Dorman & Singh, 1998; Scupholme, De Joseph, Strobino, & Paine, 1992 ). Further research about processes and outcomes of nurse-midwifery care is important to families, clinicians, policymakers, and health insurers. Comparisons Between Physician and
Nurse-Midwife Processes and Outcomes
In addition to women ’ s preexisting health status, pro-cesses of care are likely to contribute to health outcomes. Rosenblatt et al. (1997) examined the differences in processes of care used by nurse-midwives, family practice physicians, and obstetricians using a stratiﬁ ed random sample of providers of obstetric care in the state of Washington. Their purpose was to test whether there were systematic differences in the style and resource in-tensity of care provided to similar groups of women by the various provider groups. Similar groups were deﬁ ned as low-risk women having vaginal births. The researchers assumed that by excluding women with medical and ob-stetric risks, they had created equal groups of women whose outcomes could be fairly compared. The specialty of the provider explained the greatest portion of variance in outcomes after all other factors were held constant. Nurse-midwives were less likely to use technologically based processes of care than either obstetricians or family practice physicians.
By contrast, in a large prospective study ( n = 1,181), Oakley et al. (1995) examined the connections among
type of provider, processes of care, and outcomes. In addition to demographic and risk status variables, they measured other characteristics of the women, including income risk factors, locus of control, and anxiety, and women ’ s prefer-ences for methods of pain relief and prenatal opinion about the type of care processes preferred. In regression analysis, type of provider explained only a small amount of the variance in many of the outcomes. For example, provider group explained 7.6% of the variance in amount of technology-based care used. The characteristics of the women had a signiﬁ cant effect on all nine of the outcomes ( p < .01).
These studies suggest a relationship between how care is provided and actual perinatal outcomes. However, until recently, there has been little attempt to describe, measure, or predict the variables in this relationship.
Murphy and Fullerton (2001) sought to develop a mea-sure to investigate whether style of care affected the end result of that care. They updated and adapted an instru-ment developed for use in Europe by Wiegers et al. (1996) to create the OI-US . The OI-US was designed to capture a picture of processes and outcomes of care against the con-text of the woman ’ s preexisting health status. Although comparison of physician and midwifery care was the focus of the present research, any advanced practice obstetric nursing model of care can be evaluated using the OI-US .
Items in the OI-US were selected based on the most complete evidence available about characteristics and care processes inﬂ uencing optimal outcomes. The context in which a woman enters the index pregnancy is measured by the Perinatal Background Index ( PBI ), consisting of 14 standard sociodemographic and health status factors ex-isting before the current pregnancy. It is scored separately and reported as a percentage score. This score depicts pre-existing medical and social conditions that may inﬂ uence processes of care or outcomes of that care. A higher score reﬂ ects better health status.
The Optimality Index ( OI ) itself is a list of 40 care pro-cesses and outcomes across four perinatal domains (preg-nancy, parturition, neonatal condition, maternal postpartum condition). Examples include the woman ’ s current health status (anemia), processes of care (presence of a support person in labor, no episiotomy), and out-comes of care (antibiotics in the ﬁ rst 3 days postpartum, transfer of the infant to the neonatal intensive care unit). Each item meeting the criteria for optimality (e.g., method of delivery = vaginal) is scored “ 1. ” Those considered non-optimal are scored “ 0 ” (e.g., method of delivery = cesarean section). The total scores on the OI are reported as a pro-portion or percentile: the sum of all items scored, divided by total number of factors. In essence, when the PBI and OI are viewed together, a 100% score reﬂ ects the best possible perinatal health outcome within the context of
the woman ’ s health status and including appropriate uses of technology or interventions, or both in the pregnancy and childbearing process.
In comparisons between groups, higher average OI scores reﬂ ect a more optimal balance of interventions and outcomes, given the women ’ s health status. For example, two women with similar PBI scores (preexisting health history) and no history of current medical problems will have signiﬁ cantly different OI scores if one goes into spontaneous labor at term and has an unmedicated, unas-sisted vaginal delivery of a healthy infant and the other has an elective induction of labor, an epidural for pain relief, and a vacuum assisted delivery. It must be under-scored, however, that the OI-US scores are calculated for groups of women and are not to be used for risk assess-ments of individuals.
The OI-US has been used in research with samples of essentially healthy women. Murphy and Fullerton (2001) applied the instrument to a large dataset ( n = 1,286) of women who intended to give birth at home to assess its util-ity. As might be expected, the results reﬂ ected a healthy co-hort, with a PBI score of 94.8% ( SD = .065). The mean OI score was 89.2% ( SD = .059). However, the OI-US has not previously been tested with women above minimal risk.
The purpose of this prospective descriptive cohort study was to compare midwifery and medical care practices and measure optimal perinatal outcomes in a convenience sample of women at moderate risk for poor pregnancy outcome at a large, inner city obstetric service with medi-cal and midwifery services. Moderate risk was deﬁ ned as having three or more medical or psychosocial risk factors for poor pregnancy outcomes. High-risk women were de-ﬁ ned as having any one of 52 conditions (see Table 1 ) and were excluded from the study.
Data were collected on a consecutive sample of 400 women giving birth at an urban hospital, attended by either the physician or the midwifery faculty of an afﬁ liated uni-versity ’ s department of Ob/Gyn. The women giving birth in this setting historically have life situations that place them at greater risk because of poverty, immigrant status, lack of social support, among other factors. Unless women have a preexisting condition that places them at high risk, they are given their choice of midwifery or medical care.
Excluded were women ( n = 15) who had very high-risk conditions, including, but not limited to, abdominal preg-nancy, conditions requiring surgery during the pregpreg-nancy, cancer, coagulation disorders, positive HIV status, severe preeclampsia, and no prenatal care. An additional 10 women were excluded due to signiﬁ cant missing data in the patient record or data collection tool. The ﬁ nal sample
for analysis ( n = 375) consisted of 196 women cared for by a certiﬁ ed nurse-midwife (CNM) and 179 women cared for by a physician. The sample was primarily comprised women from minority populations (92%); 54% had less than a high school education and 24% were not married or partnered at the time of birth. Table 2 contains a more detailed demographic description of the sample.
Permission to conduct the study was obtained from the University of California, San Francisco Committee on
Abdominal mass Abdominal pregnancy Abscess, kidney
Anomaly, fetal incompatible with life Aortic aneurysm Aplastic anemia Appendicitis Arteriovenous malformation Bowel obstruction Brain tumor Brain death Cancer CVA
Scheduled cesarean section Cardiac arrhythmias/disease/ surgery Chest mass Coagulation defects Colostomy Dandy-Walker cyst Diabetes B-R Encephalitis Gastrostomy Guillian-Barre syndrome Hirshprungs disease HIV positive Hodgkin ’ s disease
Hypertension (on medication) Ileostomy
Incompetent cervix with cerclage Kidney/organ transplant Leukemia Listeriosis Lupus, systemic Meningitis No prenatal care Paraplegia
Pelvic mass/hx injury Renal failure Rh isoimmunization Rheumatoid arthritis
Schizophrenia/major psychosis Severe preeclampsia <36 weeks Shock/trauma (acute)
Sickle cell disease TB (active)
Thyroiditis/hyperthyroid Ulcerative colitis
Note. CVA = cardio vascular accident; B-R = class B through R; hx = history; TB = tuberculosis.
Human Research and the Investigational Review Board of San Francisco General Hospital. Waiver of informed consent was granted because the data were routinely col-lected for quality improvement and risk management activities and no additional identifying information was collected.
The information required for the OI-US was incorpo-rated into ﬁ elds on a data collection instrument already in place for the midwifery service and was pilot tested for ease of use and accuracy on 10 patient records. A study number was generated for each woman who was admitted into the intrapartum unit. Research staff, who were either experienced obstetric researchers or medical and mid-wifery students, reviewed patient records, entered the data on the forms, and maintained a diary of any questions. The forms were checked for completeness, and missing in-formation was obtained by retrieving the medical record using the linked study number. Interrater reliability was checked at the start of the project and found to be 95%. Data were entered by the principal investigator and two research assistants.
The data for women who transferred to the other group for delivery (e.g., a CNM patient who was subsequently transferred to obstetric care, most often for a cesarean deliv-ery) were analyzed as belonging to the group in which they were originally included. Style of care was operationally de-ﬁ ned as the provider group, assuming that the style used by each group was consistent with their beliefs and training.
Analyses included Students t test for between-group (CNM to physician) comparisons of OI-US scores. Linear regression was used to test for difference between the
groups on the total OI score. Logistic and linear regres-sion tests were used to examine effects of style of care on perinatal outcomes while controlling for the two preexist-ing conditions that contributed to provider group differ-ences on the PBI : use of nontherapeutic drugs or alcohol before/during pregnancy and chronic medical problems.
The mean PBI score, an indication of the medical and psychosocial factors that have potential inﬂ uences on ob-stetric outcomes, was 73% ( SD = 0.10) for women in the midwifery group and 67% ( SD = 0.14) ( p < .001) for the physician group. A higher proportion of women in the physician practice had used nontherapeutic drugs or alco-hol before or during pregnancy and a higher proportion had chronic medical problems. These two factors were primarily responsible for the between-group differences in the PBI scores for women in this study.
The mean OI-US score for the women cared for by mid-wives was 79% ( SD = 0.10) compared to 70% ( SD = 0.12) for women in the physician group ( t = -7.62, p < .001). The standard deviation of the OI scores demonstrated that the within-group variance in the midwives ’ care practices was minimal, while the physician group reﬂ ected far more variance.
In logistic and linear regression, the type of provider explained 13% of the variance in OI and the PBI score explained an additional 7% ( p < .001). Although the PBI scores indicated signiﬁ cant differences in demographic and medical risk between the CNM and the MD groups, the type of provider explained twice as much variance in OI-US scores as did the woman ’ s background health risk.
The lower frequency of practices associated with “ tech-nologically appropriate ” care in the physician patient group was not explained by higher rates of preexisting conditions associated with poor pregnancy outcomes. For example, although physicians cared for a signiﬁ cantly higher proportion of women with history of chronic ill-ness and drug abuse during pregnancy, these conditions did not explain the increase in rates of cesarean section in the physician group. The cesarean delivery rate for women cared for by midwives was 13% compared to 34% among physicians ’ patients. Women cared for by physicians were 1.7 [Conﬁ dence Interval (CI) 1.3-2.03, p < .001] times more likely to have a cesarean birth than those cared for by midwives.
In order to clarify the differences between the provider groups, the risk levels of the patient groups were made more similar by excluding women with chronic conditions as well as those with no risks or complications. Cases with preexisting chronic conditions were excluded, and only those with one or more prenatal complications were included ( n = 297). The groups had the same rates of
Characteristic % Race/ethnicity Non-Hispanic White 8 Hispanic 62 Black 16 Asian/Paciﬁ c Islander 14 Marital status Married 48
Living with partner 28
Age (years) M (years)
Range = 16-46 27
Range = 0-19 10
antepartum complications such as preeclampsia, anemia, use of antenatal testing for well-being, and use of prescrip-tion drugs, but physician patients were more likely to have had an episode of vaginal bleeding during the ﬁ rst and second trimester ( p < .02), amniocentesis ( p < .003), and inadequate prenatal care ( p < .002). As seen in Table 3 , there were signiﬁ cant differences between the midwifery and the physician groups in the use of speciﬁ c interven-tions in these moderate-risk patients. No differences were found for admissions to the neonatal intensive care unit.
This is the ﬁ rst reported study using the OI-US to ex-amine optimality in women at moderate risk. The PBI was able to discriminate between lower and higher levels of preexisting health conditions. Women in Murphy and Fullerton ’ s original analysis ( 2001 ) on a very low-risk group of women had a mean PBI score of 94.8%, com-pared to these women, whose mean PBI score was 67%-73% depending on provider group. The lower PBI scores in our sample supported our assumptions that these women had factors, particularly socioeconomic, that placed them at great health risk during pregnancy. Despite their preexisting risks, when like groups of moderate-risk women were compared using the OI-US , those cared for by midwives achieved a higher optimality score (less use of technology and equal or better health outcomes) than those cared for by physicians, with equally positive neona-tal outcomes.
scores indicated differences
in risk between the CNM and the MD groups,
provider type explained twice as much
variance in optimality scores.
This study sheds further light on the clinical practice manifestations of two concepts: optimality and risk. Risk, both its assessment and its management, is the driving per-spective in contemporary perinatal obstetric and midwifery practice. In Kennedy ’ s (2006) concept analysis of opti-mality, she discussed how optimality moves beyond risk because it measures outcomes from a positive perspective. Optimality is measurement of what should happen (achiev-ing potential) for each woman, rather than measurement of what should not happen (counting negative events).
The evidence to date on the midwifery model of care indi-cates that limited use of interventions in low-risk popula-tions results in outcomes equivalent to or better than those of the biomedical model ( Mac Dorman & Singh, 1998; Visin-tainer et al., 2000 ). However, prior to this study, it was un-clear whether the current predominantly biomedical approach to care is more optimal for mothers and neonates with some risk of poor perinatal outcomes. The current study extended the evidence on the effectiveness of midwifery care to indicate that less use of technology can result in equal or better health outcomes for moderate-risk women.
Practice MD ( n = 135), n (%) Certiﬁ ed Nurse-Midwife
( n = 162), n (%) p Value
NPO in labor 26 (20) 8 (5.1) .001
Mobility in labor: ambulatory or frequent positional changes 31 (28.4) 104 (68.4) .001 Pain relief
Nonpharmacologic method of pain relief 48 (51.1) 134 (88.1) .001
Any pharmacologic pain relief in labor 103 (82.4) 101 (63.5) .001
Epidural 63 (51.2) 49 (30.8) .001
Type of delivery .001
NSVD 85 (63.0) 129 (79.6) .004
Primary cesarean 21 (15.6) 9 (5.6) .004
Note. NPO = nothing by mouth; NSVD = normal spontaneous vaginal delivery.
a Excluding women with preexisting chronic conditions and ﬁ ltering to include only women with one or more prenatal complications.
The ﬁ ndings are limited by the use of a relatively small convenience sample. Further research using the OI-US is being conducted with similar populations of women; the additional data will assist with the assessment of the clini-cal signiﬁ cance of the OI-US scores.
Although conducted on a small sample in a single set-ting, the study ﬁ ndings are strengthened by the use of the statistical approach that retained women in their original groups for analysis. This approach insures that OI-US scores of midwifery patients transferred to physician care (most commonly for operative delivery) include all processes and outcomes of care, including those delivered after transfer. Previous midwifery outcomes research fre-quently did not track women transferred from care or was restricted only to those with vaginal deliveries. Addition-ally, the regression analysis allowed risk status to be con-sidered in the comparison of the OI-US scores between the two provider groups.
he ﬁ ndings demonstrate that
use of technology is
associated with more optimal outcomes
such as increased rates of spontaneous
This study has important implications for all practitio-ners involved in obstetric care. The ﬁ ndings challenge the pervasive view that technologic interventions protect the infant and mother from poor outcomes. For example, scientiﬁ c evidence does not demonstrate better neonatal outcomes after continuous electronic fetal monitoring ( American College of Obstetricians and Gynecologists, 2005 ), yet many perinatal clinical practitioners utilize this technology, either by choice or as mandated by institu-tional policy. The ﬁ ndings presented here demonstrate that appropriate rather than routine use of technology is associated with more optimal outcomes, such as increased rates of normal spontaneous vaginal delivery for mothers. This should be reassuring to nurses who, like other health care practitioners, ﬁ nd their management altered in the current litigious climate.
The clinical differences in patient groups and care prac-tices are demonstrated in the speciﬁ c processes of care ex-perienced by women in the current study. For example, while the rates of induction and augmentation of labor
(not optimal) were similar in each provider group, signiﬁ -cantly more women in the CNM group than women in the physician group had a support person in labor. Support in labor has been shown to decrease both length of labor and cesarean delivery rates ( Hodnett, Gates, Hofmeyr, Sakala, 2003 ). While the women cared for by physicians could be labeled as signiﬁ cantly higher risk by virtue of both the PBI scores and the speciﬁ c conditions contributing to those scores, only type of provider predicted higher cesar-ean delivery rates. The use of the OI-US permits assess-ment of the type of care that is linked to positive outcomes while taking preexisting conditions into consideration.
The OI-US holds the potential to inform the practice, research, and education paradigms for childbearing women. A model of care based on the concept of optimality in peri-natal health permits conservation of critical health care re-sources to achieve best possible outcomes with minimum intervention, within the context of the woman ’ s health sta-tus. Additionally, repositioning women ’ s health models to-ward an assumption of health rather than risk potential may begin to shift attitudes of women and clinicians away from a fear-based perception of birth. Given the current malpractice insurance crisis, a shift in this view could ben-eﬁ t all obstetric care providers: nurse clinicians, obstetric nurse practitioners as well as physicians and midwives.
As previously noted, the type of care provided may be guided by different professional philosophies and assump-tions. This study lends support to this perspective. Acknowledging the connections between the beliefs under-lying different models of maternity care and the outcomes of that care is also an important potential of research using the OI-US . In order for health care systems themselves to change, the fundamental beliefs of nursing, midwifery, and medicine must be carefully examined in relation to the outcomes achieved by each model. From this knowledge, the groundwork can be laid for a health system that pro-motes health across all dimensions, focuses on appropriate use of technology, and celebrates the strength of women ’ s bodies to bear children with little or no intervention.
Supported by a grant from the University of California, San Francisco Academic Senate and adapted from a sym-posium paper presented at the National Congress on the State of the Science in Nursing Research, October 7, 2004, Washington, DC. The authors thank the students of the 2004 nurse-midwifery class at the University of Califor-nia, San Francisco who collected data.
American College of Obstetricians and Gynecologists ( 2005 ). In-trapartum fetal heart rate monitoring . ACOG Practice Bul-letin No. 62 . Obstetrics and Gynecology , 105 , 1161 - 1169 .
Clarke , S . , Martin , J . , & Taffel , S . ( 1997 ). Trends and characteris-tics of births attended by midwives . Statistical Bulletin , 9 - 18 . Declercq , E . ( 1995 ). Midwifery care and medical complications .
Birth , 22 , 68 - 73 .
Hodnett , E. D. , Gates , S . , Hofmeyr , G. J. , & Sakala , C . ( 2003 ). Continuous support for women during childbirth (Co-chrane Review) . In The Cochrane Library (Version 3) . Kennedy , H. P. ( 2006 ). A concept analysis of ‘ optimality ’ in
peri-natal health . Journal of Obstetric, Gynecologic, & Neona-tal Nursing, 35, 763-769.
Mac Dorman , M . , & Singh , G . ( 1998 ). Midwifery care, social and medical risk factors, and birth outcomes in the USA . Jour-nal of Epidemiology and Community Health , 52 , 310 - 317 . Murphy , P. A. , & Fullerton , J. T. ( 2001 ). Measuring outcomes of midwifery care: Development of an instrument to assess optimality . Journal of Midwifery and Women ’ s Health ,
46 , 274 - 284 .
Oakley , D . , Murtland , T . , Mayes , F . , Hayashi , R . , Petersen , B. A. , Rorie , C ., et al . ( 1995 ). Processes of care. Comparisons of certiﬁ ed nurse-midwives and obstetricians . Journal of Nurse-Midwifery , 40 , 399 - 409 .
Rooks , J . ( 1997 ). Midwifery and childbirth in America . Philadel-phia : Temple University Press .
Rosenblatt , R . , Dobie , S . , Hart , L. G. , Schneeweiss , R . , Gould , D . , Raine , T ., et al . ( 1997 ). Interspecialty differences in the obstetric care of low-risk women . American Journal of Public Health , 87 , 344 - 351 ,
Scupholme , A . , De Joseph , J . , Strobino , D . , & Paine , L . ( 1992 ). Nurse midwifery care to vulnerable populations phase I: Demographic characteristics of the national CNM sample .
Journal of Nurse-Midwifery , 37 , 341 - 347 .
Visintainer , P. F. , Uman , J . , Horgan , K . , Ibald , A . , Verma , U . , & Tejani , N . ( 2000 ). Reduced risk of low weight births among indigent women receiving care from nurse-mid-wives . Journal of Epidemiology and Community Health ,
54 , 233 - 238 .
Wiegers , T. A. , Keirse , M. J. , Berghs , G. A. , & van der Zee , J . ( 1996 ). An approach to measuring quality of midwifery care . J Clinical Epidemiology , 49 , 319 - 325 .
Leslie Cragin, PHD, CNM, FACNM is an associate clinical professor in the Department of Obstetrics, Gynecology & Reproductive Sciences at the University of California, San Francisco.
Holly Powell Kennedy, PHD, CNM, FACNM is an associate professor in the Department of Family Health Care Nursing at the University of California, San Francisco.
Address for correspondence: Leslie Cragin, PHD, CNM, FACNM, Department of Obstetrics, Gynecology & Reproduc-tive Sciences, University of California, San Francisco, 1001 Potrero Avenue, Room 6D-29, San Francisco, CA 94110. E-mail: email@example.com