Dr M. C. Klein, BC Research Institute for Children's and Women's Health, Room L-309B, 4500 Oak Street, Vancouver, BC Canada V6H 3N1.
Objective To compare family physicians', obstetricians' and midwives' self-reported practices, attitudes and beliefs about central issues in childbirth.
Design Mail-out questionnaire.
Setting/Population All registered midwives in the province, and a sample of family physicians and obstetricians in a maternity care teaching hospital. Response rates: 91% (n= 50), 69% (n= 97) and 89% (n= 34), respectively.
Methods A postal survey.
Main outcome measures Twenty-three five-point Likert scale items (strongly agree to strongly disagree) addressing attitudes toward routine electronic fetal monitoring, induction of labour, epidural analgesia, episiotomy, doulas, vaginal birth after caesarean section (VBACs), birth centres, provision educational material, birth plans and caesarean section.
Results Cluster analysis identified three distinct clusters based on similar response to the questions. The ‘MW’ cluster consisted of 100% of midwives and 26% of the family physicians. The ‘OB’ cluster was composed of 79% of the obstetricians and 16% of the family physicians. The ‘FP’ cluster was composed of 58% the family physicians and 21% the obstetricians. Members of the ‘OB’ cluster more strongly believed that women had the right to request a caesarean section without maternal/fetal indications (P < 0.001), that epidurals early in labour were not associated with development of fetal malpositions (P < 0.001) and that increasing caesarean rates were a sign of improvement in obstetrics (P < 0.001). The ‘OB’ cluster members were more likely to say they would induce women as soon as possible after 41 3/7 weeks of gestation (P < 0.001) and were least likely to encourage the use of birth plans (P < 0.001). The ‘MW’ cluster's views were the opposite of the ‘OBs’ while the ‘FP’ cluster's views fell between the ‘MW’ and ‘OB’ clusters.
Conclusions In our environment, obstetricians were the most attached to technology and interventions including caesarean section and inductions, midwives the least, while family physicians fell in the middle. While generalisations can be problematic, obstetricians and midwives generally follow a defined and different approach to maternity care. Family physicians are heterogeneous, sometimes practising more like midwives and sometimes more like obstetricians.
The practice of maternity care is characterised by close co-operation between several health care disciplines—obstetricians, midwives, family physicians and nurses in the same hospital and caring for the same patient. This can lead to levels of conflict or co-operation depending on the culture of the working environment. It is widely accepted that each discipline has a unique body of expertise and ideology. It was our view that identifying areas of agreement, disagreement or ambiguity in knowledge or attitudes could help reduce misunderstanding, improve co-operation among care providers and potentially improve the quality of care for mothers.1 In British Columbia, family physicians provide intrapartum care and attend to approximately 64% of births, obstetricians attend about 30% and midwives 6%. Outside of the major two population centres of Vancouver and Victoria, obstetricians work principally as consultants to family physicians and midwives. Midwives became regulated, autonomous and fully a part of the health care system in 1998 and attend births in both hospital and at home, both services covered by the provincial Medicare system.
We had four study objectives: (1) Design a tool to measure attitudes, beliefs and self-reported practices pertaining to key aspects of maternity care including attitudes toward interventions and the role of the mother in the birth process. (2) To determine if there were similar attitudes, beliefs and practices within a given professional group. We did this for three professional groups who provide intrapartum care: family physicians, obstetricians and midwives. (3) To compare the attitudes, beliefs and practices among the three professional groups to see if each professional group had a distinct approach to childbirth. (4) To compare the degree to which the three professional groups share a consensus around some common clinical approaches.
A preliminary set of questions covering key approaches to maternity care was drawn from several sources—informed opinion, professional association guidelines and hospital policy and procedure manuals. A draft questionnaire was developed and sent to 54 family physicians at two external hospitals. An accompanying cover letter gave some background about the questions and asked the respondents to add any questions they thought should be added to the questionnaire. Twenty-three family doctors returned questionnaires suitable for analysis. Based on comments and suggestions, the questionnaire was revised. The revised questionnaire was sent to a convenience sample of seven family practice leaders across Canada. Again, subjects were asked to complete the questionnaire and to provide comments and suggestions. The final questionnaire consisted of demographic and practice-related questions and 76 statements to be responded to on a seven-point scale (very strongly agree to very strongly disagree), all addressing key maternity care areas. Ethical approval for the project was obtained from the University of British Columbia. From these core questions two additional questionnaires with five-point Likert scales (strongly agree to strongly disagree) were developed specifically for obstetricians and midwives. Based on statistical advice, the agreement scale was changed from a seven-point scale to a five-point scale. All questionnaires were again pilot tested with an external group. All three maternity care provider questionnaires included a core set of 23 questions. The remaining questions were unique for each professional group and are not reported here. Examples of the 23 common items included
Knowledge (e.g. ‘The use of epidural analgesia in early labour [before 4 cm] increases the caesarean section rate’, ‘A women's sexual abuse has no impact on the course of labour’),
Personal preferences for practice (e.g. ‘If it were available I would consider attending births at a freestanding birth centre’),
Beliefs about childbirth (e.g. ‘Childbirth is only normal in retrospect’).
The family physician questionnaire was mailed out in October 1999 with a self-addressed envelope to all 141 family physicians who attended one or more births during the period 1 April 1997 to 20 August 1998 at the BC Women's Hospital. Included with each questionnaire was a cover letter stating the purpose of the study, guaranteeing confidentiality and the opportunity to win a gift certificate for a local restaurant. The response rate after second mailing was 68.8%. Because some key areas covered in the obstetrician and midwife questionnaires were not covered in the initial family physician questionnaire, in May 2001 a follow up questionnaire with 14 additional questions was sent to the same mailing list. The response rate was 53.9% (76/141). The obstetrician questionnaire was mailed to obstetricians with privileges at the BC Women's Hospital and/or St. Paul's Hospital in June 2001 (response rate was 88.5%). The midwives survey was sent out in June 2002 to the list of 55 practising midwives registered with the College of Midwives of British Columbia (response rate was 90.9%). BC Women's Hospital is the largest maternity hospital in Canada (7000 births per year) and considered the leader in maternity care in British Columbia.
We used cluster analysis to create categories of care providers based on the similarity or differences in the way they responded to questions in the survey.2 Cluster analysis is an exploratory data analysis tool for solving classification problems. Its object is to sort cases (people, things, events, etc.) into groups, or clusters, so that the degree of association is strong among members of the same cluster and weak among members of different clusters. Each cluster thus describes the class to which its members belong. Cluster analysis is a tool of discovery. It may reveal associations and structure in data which, although not previously evident, may be sensible and useful once found. The results of cluster analysis may contribute to the definition of a formal classification scheme, such as taxonomy for related animals, insects or plants.
Reliability was analysed by measuring Cronbach's α and validity was tested using principal component factor analyses (rotation = varimax). χ2 statistics were conducted to compare the percentages of agreement (‘somewhat agree’ and ‘very much agree’) to the particular items among the three professions. Additionally, to examine whether there were distinct subgroups of individuals with differentiating attitudes—independent of their profession, cluster analyses (k-means-cluster) were performed. χ2 statistics were conducted to test whether there were demographic differences among the three clusters.
Reliability analyses with the original 23 items revealed a satisfactory reliability (Cronbach's α= 0.76). After deletion of the items ‘interventions for legal reasons’, ‘oxytocin’, and ‘assess women in the office’, reliability was improved to Cronbach's α= 0.83.
In order to examine construct validity of our tool, we conducted principal factor analysis (rotation = varimax) with the remaining 20 five-point scale items (n= 181). We found four factors contributing >5% to the explained variance, resulting in 49% of explained variance. The factor-eigenvalues (5.3, 1.9, 1.4, 1.2) showed one strong and three minor factors. Items on the first factors referred to the item nos. 1, 2, 4, 5, 13 and 20. The second factor consisted of the items 3, 11, 12, 15, 18 and 19. Factor 3 contained the items 6, 8, 10 and 17. The fourth factor consisted of the items 7, 9, 14 and 16. The item loadings were >0.40 for every item. In summary, the data suggested good reliability and validity for our tool.
There were significant differences among the three maternity provider groups in 20 out of 23 items (Table 1): Every midwife agreed that she would consider birth in a freestanding birth centre, compared with 25.8% of the obstetricians and 44.8% of the family physicians (P < 0.001). The majority of family physicians (67.8%) and midwives (72.2%) agreed that the administration of epidural early in labour is associated with the development of fetal malpositions. On the contrary, only 24.2% of obstetricians agreed with this statement (P < 0.001). While all obstetricians agreed that they preferred to induce patients after 41 3/7 weeks of gestation, fewer family physicians (80.5%) and only 30% of midwives shared this approach (P < 0.001). More than half of the obstetricians agreed that an elective caesarean without clinical reason is a women's right, while only a small minority of midwives (14.6%) and family physicians (13.6%) agreed with this philosophy. There were no significant differences among the three provider groups regarding the question, acknowledging that interventions or tests are sometimes performed due to legal reasons rather than clinical indications. In general, obstetricians saw themselves as performing interventions more often than midwives. On the other hand, most midwives agreed that a woman's attitude and personal history plays an important role during childbirth, while many obstetricians disagreed. Family physicians' attitudes most often fell in between those of obstetricians and midwives. The attitudes of both urban and rural midwives were similar (data not shown).
Table 1. Summary of data for provider groups.
Percentage agreement (%)
1. I would consider attending births at a freestanding centre.
2. Childbirth is only normal in retrospect.
4. Few women would choose to have a VBAC if they knew of the consequences of uterine rupture.
5. A woman's history of sexual abuse has an impact on the course of labour.
13. I provide written educational material on labour and delivery to my patients.
20. I prefer to induce my post-date patients at or as soon as possible after 41 3/7 weeks.
3. Doulas improve maternal and newborn outcomes.
11. I prefer routine continuous EFM over intermittent auscultation for all births.
12. 3rd and 4th degree tears are an inevitable consequence of difficult deliveries, largely due to maternal or fetal characteristics.
15. I prefer to employ episiotomy routinely, because it is easier to repair than lacerations that result when episiotomy is not used.
18. I regularly employ episiotomy to prevent pelvic floor relaxation.
19. The increasing C/S rate in our country is a sign of improvement in obstetric care.
6. I encourage my patients to develop a birth plan.
8. Many women have painful early labour that requires admission.
10. The administration of epidurals early in labour is associated with the development of fetal malpositions (OT and OP).
17. I always offer alternative forms of pain relief before offering an epidural.
7. I believe the most important determinant of a successful birth is the woman's own confidence and determination.
9. I feel that it is a woman's right to elect to have a caesarean section even if there are no clear maternal or fetal indications.
14. We could safely lower our caesarean section rate through organized peer review of caesarean sections for dystocia.
16. We should perform routine cord gases on all births.
Eliminated from factor analysis
At times I perform tests and procedures for medico-legal reasons rather than clinical indications.
If oxytocin was used more often routinely we would do fewer caesarean sections for dystocia.
I often arrange to examine a woman in my office when I am trying to assess if she is in labour.
We identified three distinct clusters showing significant differences in every item (Fig. 1).
There was no significant difference in age among the clusters (Table 2). Cluster 1 (‘MW’ cluster) included 100% of the midwives and 26% of the family physicians. Cluster 2 (‘OB’ cluster) included 79% of the obstetricians (including all the perinatologists) and 16% of the family physicians. Cluster 3 (‘FP’ cluster) included 58% of the family physicians and 21% of the obstetricians. Women predominated in the ‘MW’ cluster and were under-represented in the ‘OB’ cluster. The number of attended births was highest in the ‘OB’ cluster. Nineteen percent of members in the MW cluster predicted that they would be leaving the profession in five years compared with 29% of members of the OB cluster and 46% of members of the FP cluster. These results were not age dependent.
Table 2. Description of the three clusters.
Cluster 1 (MW)
Cluster 2 (OB)
Cluster 3 (FP)
100% (n= 50)
0% (n= 0)
0% (n= 0)
26% (n= 25)
16% (n= 16)
58% (n= 56)
0% (n= 0)
79% (n= 27)
21% (n= 7)
Mean number of births per year
Staying in profession in five years
Leaving profession in five years
Undecided if staying or leaving profession in five years
The MW cluster (n= 75) most strongly agreed that doulas improve maternal and newborn outcomes (P < 0.001), that they encouraged their clients to make a birth plan (P < 0.001), that the administration of epidurals in early labour is associated with fetal malpositions (P < 0.001), that epidurals early in labour increased the caesarean section rate (P < .001) and they would consider attending births in a freestanding birth centre (P < 0.001). The MW cluster most strongly disagreed that the increasing caesarean section rate was an expression of improving childbirth in our country (P < 0.001), that women had a right to an elective caesarean section without obstetric indications (P < 0.001), that many women have painful early labour that requires admission to hospital (P < 0.001) and that they preferred to induce clients at or as soon as possible after 41 3/7 weeks (P < 0.001). For these items, the OB cluster (n= 43) showed the opposite pattern to the MW cluster, while the attitudes of the FP cluster (n= 63) fell in the middle between the MW cluster and the OB cluster (see Fig. 1).
The MW cluster most strongly agreed that sexual abuse has an impact on the woman's birth experience and that doulas improve maternal and newborn outcomes. The FP cluster most often disagreed with these two items while the OB cluster fell in the middle (P < 0.001). The MW and FP clusters both strongly agreed that they would offer alternative pain control methods before using epidural analgesia, that the woman's self-confidence is the most important determinant of a successful birth and that peer review would lower the caesarean section rate for dystocia. The OB cluster more often disagreed with these statements. The OB and FP clusters were in agreement with the statement that childbirth is only normal in retrospect while the MW cluster disagreed. The OB cluster most strongly agreed that cord gases should be performed routinely on all births, while the FP cluster and MW cluster strongly disagreed.Figure 1 shows the separation of the three clusters according to the questions that were in common on all three provider surveys.
We have studied differences and similarities among three maternity care provider groups: family physicians, obstetricians and midwives. We have tested a tool to measure attitudes and self-reported practices pertaining to key aspects of maternity care. The tool has good reliability and validity. We have shown that viewpoints and ideologies are not consistent within members of each of the three professional groups. However, cluster analysis revealed that each maternity care professional group had distinctive and shared beliefs/opinions and approaches. While there was some overlap among professions within each of the three clusters, this was mostly due to the variability in beliefs among the family physicians. Family physicians were found in each of the three clusters, shifting in their views and cluster membership according to various questions—although the majority of family physicians were found in the ‘FP’ cluster.
Divergent view points on maternity care have been reported in Quebec.3 Midwives in Quebec held more client-centred and less interventionist attitudes than physicians, which agrees with our findings, although differences among types of doctors was not reported. Differences in opinion about vaginal birth after caesarean section (VBAC) have been reported among different types of maternity care providers.4
We found significant differences among the three maternity care provider groups on some knowledge-based questions, for example, ‘Doulas improve maternal and newborn outcomes’. This question illustrates that midwives are either more aware of the evidence that concludes that doulas improve outcomes or are more likely to ‘believe’ the evidence.5–8 Other issues such as early epidural use and utilisation of electronic fetal monitoring showed that midwives and family physicians had attitudes/beliefs more in line with current evidence than obstetricians. Obstetricians' beliefs are more likely to be aligned with recent evidence about induction for post-term pregnancy than were the beliefs of the other clusters, while midwives and family physicians were more sceptical about the external validity of the literature on induction at 41 3/7 weeks of gestation.9–12
For some questions that were purely opinion based such as ‘I feel that it is a woman's right to elect to have a caesarean section even if there are no clear maternal or fetal indications’, there was clear separation among the professional groups. These differences in opinion may be the result of differences in professional education as well as reflecting recent editorials from obstetrician opinion leaders.13–17 It is also possible that certain professions attract people with a pre-existing world view, so that group opinions may be a result of self-selection into a professional group.
VBAC is an area under great pressure in the obstetrician community. While it was fashionable to attempt VBAC only a few years ago, today especially in the USA, but also in Canada, obstetricians are feeling more uncomfortable in encouraging VBAC.18 The evidence on the risks and benefits of VBAC is controversial.19–25 Obstetricians tend to accept the literature that points out that VBAC is dangerous while family physicians and midwives are more doubtful about this literature.
While the sample of midwives represented all midwives in BC, the family physician and obstetrician samples were based in the Vancouver teaching hospitals. This might have skewed the results toward those physicians who favour more interventions. We believe that the influence of this hospital on the opinions of other care providers in the province is strong due to considerable outreach and training provided to family physicians and obstetricians in the province over the past 19 years. Therefore, we believe the opinions reflected in this survey are probably representative of care providers in this province as a whole. However, we have no idea if the diversity of opinion, knowledge and practice preferences that we found would be found in other countries or health care systems. We suspect in that in a country with fewer types of health care professionals providing intrapartum care, these difference might be more marked. We also suspect that there might be less disagreement regarding technology among care providers in countries that use less technology in birth. Finally, in countries with a strong financial incentive toward evidence based medicine, there might be less divergent opinions. Thus, we suggest that further validation of this tool in other countries will be useful.
We have developed a unique tool that separates midwives, obstetricians and family physicians providing intrapartum care according to attitudes and beliefs regarding key issues in maternity care. Using this tool, we have shown differences of opinion and beliefs among the three different maternity care provider groups. Midwives and obstetricians often have significantly divergent views on key issues in maternity care. Family physicians, on the other hand, are more of a diverse group, having some members in each of the three clusters, but the majority have relatively neutral positions compared with midwives and obstetricians. Recognition, by both the woman and her health care provider, of the areas of agreement and disagreement among the maternity care professional groups offers an opportunity for patient/client choice and for education and dialogue to the benefit of the childbearing woman.
We will use the results of this study to determine if there is a link between these beliefs and actual behaviour by using our database of hospital utilisation data to confidentially link attitudes/knowledge with provider behaviour and maternal and newborn outcomes.