OBJECTIVE: To examine factors affecting participation in obstetrics among obstetrician–gynecologists and changes in participation over time.
METHODS: Using physician billings from Ontario, Canada, from 1992/1993 to 2001/2002, we examined the impact of physician age, gender, practice location, and years of practice on participation in obstetrics with multiple logistic regression and repeated measures analyses. We also examined differences in practice patterns between obstetrics providers and nonproviders using linear and log-linear regressions.
RESULTS: Obstetrics participation declined with age, from 96% among physicians under age 35, to 34% among those aged 65 and over (2001/2002 figures). Regressions demonstrated a lower likelihood of performing obstetrics in successive years (odds ratio [OR] 0.95 per year; 95% confidence interval [CI] 0.93, 0.98) and among physicians who were older (OR 0.91 per year of age; 95% CI 0.90, 0.93), female (OR 0.57; 95% CI 0.36, 0.91), and practicing in cities with medical schools (OR 0.58; 95% CI 0.44, 0.78). The crude obstetrics participation rate dropped from 82% to 77%, from 1992/1993 to 2001/2002. The physician age-sex–adjusted participation rate dropped from 80% in 1992/1993 to 77% in 2001/2002. Obstetrics providers had almost double the annual billings of nonproviders ($364,000 verus $187,000; P < .001), but more on-call days worked (105 versus 13; P < .001). Nonproviders of obstetrics derived more of their billings from outpatient visits, psychotherapy, and diagnostic tests. The likelihood of an obstetrics nonprovider resuming obstetrics was 1.1% per year.
CONCLUSION: The proportion of obstetrician–gynecologists practicing obstetrics declined modestly in the last decade, partly because of more female physicians in the workforce who were less likely to practice obstetrics. Planners should consider these trends when estimating how many obstetrician–gynecologists to train to meet future societal needs.
LEVEL OF EVIDENCE: II-2
Participation in obstetrics by obstetrician&#x2013;gynecologists declined in the 1990s, and nonparticipants were more likely to be older, female, working in cities with medical schools, and infrequently on call.
From the Institute for Clinical Evaluative Sciences, Toronto; the; Departments of Health Policy, Evaluation and Management; Family and Community Medicine; and Public Health Sciences, University of Toronto, Toronto; and;Sault Area Hospitals, Sault Ste. Marie, Ontario, Canada.
Received September 11, 2003. Received in revised form October 21, 2003. Accepted November 21, 2003.
Address reprint requests to: Dr. Ben Chan, Senior Scientist, Institute for Clinical Evaluative Sciences, 2075 Bayview Avenue, Suite G-106, Toronto, ON, M4N 3M5, Canada; e-mail: firstname.lastname@example.org.
Funding support was provided by the Institute for Clinical Evaluative Sciences.
Ensuring an adequate supply of health care personnel to meet the obstetrical needs of the population is a critical task for health system planners. In the past 2 decades, considerable attention has been focused on the declining participation of family physicians in obstetrics. This trend has been described in both the United States1 and Canada,2 and the proportion withdrawing from obstetrics in both countries is nearly identical.3 Analysts have suggested that a variety of factors may have contributed to these declining participation rates, such as fear of malpractice lawsuits or the desire for a balanced lifestyle.4 The reduced participation in obstetrics by family physicians could potentially lead to increased reliance on midwives or specialists to handle the burden of remaining cases.
There is scant information, however, on whether participation in obstetrics among obstetrician–gynecologists has also declined. Some studies have reported that obstetrician–gynecologists decrease their participation in obstetrics near the end of their career and that female physicians have a lower overall workload than their male counterparts.5 However, there is no information on whether women participate in obstetrics any differently than men and whether overall participation in obstetrics is decreasing, independent of changes to the gender structure of the workforce. One review cited personal communication referring to unpublished survey data on the declining participation in obstetrics,6 but these figures have not been validated elsewhere.
A declining participation in obstetrics among obstetrician–gynecologists would represent an even greater challenge to policy makers because of the central role that they play in the delivery of obstetrical services. In addition to handling an important share of routine deliveries, these specialists also manage high-acuity cases and provide back-up to other obstetrical providers. If this trend is indeed occurring, then policy makers would need to consider measures to ensure that this critical service is maintained.
This study uses data from the Canadian province of Ontario to determine which physician factors influence the participation rate in obstetrics and whether or not this participation rate is indeed declining. This study also describes the practice characteristics of providers and nonproviders of obstetrics in terms of types of services provided, frequency of on-call duties, and total remuneration. Lastly, this study examines the likelihood that a physician who has abandoned obstetrics will return to obstetrics.
MATERIALS AND METHODS
All legal residents of Ontario are enrolled in the Ontario Health Insurance Plan, which provides coverage for all essential medical services with no copayments or deductibles. There are no other insurance carriers for obstetrical care. The vast majority of obstetrician–gynecologists in Ontario (97%) are remunerated on a fee-for-service basis and bill Ontario Health Insurance Plan for each service rendered. A small number of physicians in Kingston, Ontario, were excluded because they were remunerated through a special payment plan for academic physicians. Each billing claim contains identifiers for the physician, patient, and hospital where the service was performed (if applicable), a fee code describing the service provided, the number of services provided, and amount paid. We had access to the full Ontario Health Insurance Plan claims database under a comprehensive research agreement with the Ontario Ministry of Health and Long-Term Care. Individual physician and patient identifiers were scrambled to protect confidentiality. Data were available from fiscal years 1992/1993 to 2001/2002. In Ontario, fiscal years begin on April 1 and end March 31. The Research Ethics Board of Sunnybrook and Women's College Health Sciences Centre approved this study.
The Ontario Physician Human Resource Data Centre database contains additional information on each physician's specialty, gender, age, year of graduation, and postal code of practice. We used this database to report baseline information about the demographic characteristics of obstetrician–gynecologists. Two-tailed t tests were used to test differences in age between male and female physicians. Next, we linked the Ontario Physician Human Resource Data Centre database to the Ontario Health Insurance Plan claims database, and then linked the physician's postal code in the database to the 1996 Census Canada Postal Code Conversion File. The latter permitted us to classify a physician's practice location into 2 categories: major cities with university teaching centers (Toronto, London, Ottawa, and Hamilton) and all other communities.
We considered an obstetrician–gynecologist to be participating in obstetrics if he or she performed at least 5 deliveries per year (either spontaneous vaginal deliveries, assisted deliveries, or cesarean deliveries). This threshold was increased to 25 and 50 per year in a sensitivity analysis. The results were consistent using each cutoff, and the results from the sensitivity analysis are available from the authors upon request.
We calculated the obstetrical participation rate by physician age and gender and conducted bivariate analyses using two-tailed t tests to examine differences in obstetrical participation by gender, stratified by age group. Next, we calculated crude and physician age-sex standardized rates of participation in obstetrics. Direct standardization techniques were used for the latter, and 2001/2002 was the standard year. Then, we tested the significance of changes in the likelihood of doing obstetrics from 1992/1993 to 2001/2002 using logistic regression models, employing generalized estimating equations and a repeated measures analysis with autoregressive error terms. These analyses employed the generalized linear model (GENMOD) procedure in SAS 8.0 (SAS Institute Inc, Cary, NC).
We also examined changes in propensity to perform obstetrics using a cohort analysis. We identified all individuals performing obstetrics in 1992/1993 and classified them into 3 age categories (less than 35, ages 35–64, and age 65 years and over). We then calculated the obstetrical participation rate in each age group in each subsequent year up to 2001/2002. The participation rate in a given year was defined as the number of obstetrician–gynecologists in the cohort performing obstetrics, divided by the number in the cohort still in practice of any sort in the given year.
We compared practice patterns between providers and nonproviders of obstetrics. These patterns included total annual billings; annual number of days required to work while on call (defined as the number of calendar days in which a physician billed for a special visit to a hospital, emergency department, or other practice setting); and proportion of total billings on obstetrical deliveries and related services, gynecological and other surgery, other hospital care, prenatal office visits, office psychotherapy, other office visits, ultrasound, and other diagnostic tests. We used regression analyses to test the significance of differences between the 2 groups. For total billings, linear regression was used; for days on call, Poisson regression; and for proportions of billings, linear and log-linear models, with the latter reserved for cases where the distribution of the proportions was heavily skewed.
The number of obstetrician–gynecologists increased by 3.2% from 1992/1993 to 2001/2002 (Table 1), and the proportion who were women rose steadily. The age structure of the workforce was similar at the start and at the end of the study period, but the proportion of the workforce age 65 and over reached a peak in 1996/1997 and has since declined to its previous level. In 2001/2002, female obstetrician–gynecologists tended to be younger than their male counterparts (mean age 42 versus 53 years; two-tailed t test, P < .001). Among obstetrician–gynecologists practicing obstetrics, women performed 11% fewer deliveries per physician in fiscal year 2001/2002 than did men (mean 211 versus 237, P < .008).
Participation in obstetrics decreased as physicians grew older. In 2001/2002, the vast majority of physicians under age 35 performed obstetrics, compared with approximately one third of those aged 65 and over (Table 2). The cohort analysis was consistent with these findings and demonstrated that physicians gradually relinquish obstetrical practice over time. Among doctors providing obstetrics in 1992/1993, the proportion that continued obstetrics in 2001/2002 was 92%, 82%, and 18% for physicians under age 35, ages 35 to 64, and age 65 and over, respectively. Female physicians appeared to have lower participation rates in obstetrics than male physicians in the 35-to-64-year-old age range (Table 2), although in this simple bivariate analysis using t tests, the difference in participation rate between genders was not statistically significant.
Table 3 describes the results of a logistic multivariable analysis of the factors that predict obstetric practice. The decline in likelihood of obstetric practice over time was significant. Increasing physician age was related to a lower likelihood of practicing obstetrics. Female gender was significantly related to a lower likelihood of obstetric practice, after adjusting for differences in age between men and women. Physicians in cities with medical schools were also less likely to practice obstetrics.
Consistent with the multivariable analysis above, the crude rate of participation in obstetrics declined from 82% in 1992/1993 to 77% in 2000/2001, with the lowest point reached in 1999/2000 (75%). Part of the decline in obstetrical participation, however, was due to the changing demographics of the physician pool, with a higher proportion of women in the workforce who were less likely to perform obstetrics. Even after adjusting for changes in physician demographics, however, the participation rate still declined modestly, from 80% to 77%.
The decline in obstetrical participation was accompanied by a 5.6% increase in the mean annual number of deliveries per physician among those performing deliveries, from 216 per year in 1992/1993 to 228 per year in 2001/2002. This increase was not statistically significant (P = .20, based on a repeated measures analysis, adjusting for aging of the same physician over time). However, the magnitude of the increase in obstetrical workload differed by physician gender. Among female physicians, mean deliveries rose 4.4%, from 202 to 211 per year, and among male physicians, 7.8%, from 219 to 236 per year. The increase was not statistically significant for women (P = .54) but was significant for men (P = .03).
Among obstetrician–gynecologists who were doing obstetrics in any given year from 1992/1993 to 2001/2002, the proportion who relinquished obstetrics in the following year was 3.3%. Among those who were in practice but not doing deliveries in any given year from 1992/1993 to 2001/2002, the proportion who resumed obstetrics in the following year was 1.1%.
There were marked differences in practice patterns between providers and nonproviders of obstetrics. Most nonproviders participated in almost no call activities that required a special visit to an institution (Table 4). These physicians devoted more of their practice to office visits, performing ultrasounds, and other diagnostic procedures. Furthermore, nonproviders of obstetrics had total billings that were 46% lower than providers.
This study documents a modest decline in the past decade in the likelihood that an obstetrician–gynecologist will perform obstetrics. Seventy-seven percent of the obstetrics and gynecology workforce practices obstetrics at present. Approximately one half of this decline can be attributed to the fact that female physicians are less likely than males of the same age to perform obstetrics and that women constitute a growing proportion of the workforce. Physicians also tend to relinquish obstetrics as they grow older. Only 1.1% of those who have given up obstetrics will return to it in a subsequent year. The decrease in obstetrical participation coincides with a 6% increase in average number of deliveries among those who continue to provide the service.
Physicians in the youngest age category (less than 35 years) had very high obstetric participation rates (96% in 2001/2002). This finding suggests that almost all obstetrician–gynecologists do perform obstetrics shortly after completing residency. Such a finding tends to refute the hypothesis that prospective physicians are “streamed” into subspecialties during their training and therefore begin their medical career with a very narrow scope of practice that excludes obstetrics. In Canada, there is a greater emphasis on producing generalist physicians than in the United States,7 which may account for the observed results. Further research should examine whether this trend also exists in the United States, where earlier enthusiasm for subspecialization8 has been mitigated by recent declines in subspecialist enrollment.9
Participation in obstetrics declines with age, and there is an accelerated decline from age 65 onwards. This finding is consistent with the literature. Other authors have suggested that either the energy and skills for treating pregnancy decline with age, or that there is a life-cycle pattern in the doctor/patient relationship, with pregnancy serving as an entry point.10 Furthermore, physicians with more years of experience may have had greater exposure to malpractice claims. Further research should examine whether those physicians previously exposed to a malpractice claim are more likely to relinquish obstetrical practice.
This study demonstrated that female physicians were more likely than male physicians to relinquish obstetrics, particularly in the prime years in the middle of their career. Furthermore, as an increasing proportion of obstetrician–gynecologists withdraw from obstetrics, the obstetrical workload increases for those obstetrician–gynecologists who continue to do obstetrics, especially if they are male. The lower participation in obstetrics by women is congruent with past research showing that female physicians work fewer hours than men.5 The reasons for this finding deserve further research. One obvious possibility is that concerns about balance between work and family life may encourage women to relinquish obstetrics more readily than men. Such a hypothesis raises the sensitive issue of how to balance the wishes of women in the workforce for quality of life against the imperative of meeting societal needs for obstetrical services. The fact that the medical workforce in both Canada and the United States is composed of a growing proportion of women,11,12 may contribute further to the decline in obstetrical participation. Planners should account for such trends when assessing future physician supply requirements.
We observed a relatively modest decline in obstetrical participation over time, even after adjusting for demographic changes in the workforce. One potential factor affecting this trend may be the increasingly litigious environment surrounding obstetrics. Over the past decade in Canada, the number of obstetrical malpractice claims per year has been constant, but the size of malpractice awards has doubled.13 In Ontario specifically, the premiums for malpractice have risen by 369%, from $14,796 in 199214 to $69,360 in 2002,15reflecting a more rapidly growing level of damages awarded in the province. These premiums, however, are heavily subsidized by the Ontario government, such that the actual out-of-pocket cost to physicians has been frozen over time. The level of premiums per se is generally not associated with lower participation in obstetrics.16 However, rising premiums could potentially reflect changes in other aspects of the malpractice experience, which may alter the emotional impact on the physician and lead to reduced obstetrical participation. Further research is needed to determine if this hypothesis is true.
The likelihood of performing obstetrics is lower in major cities with teaching hospitals. This trend is consistent with findings in other studies.17 One reason for this finding may be that, in large cities, the critical mass of patients may exist to support a core of subspecialists. Further research should examine the extent to which such subspecialization is being encouraged within academic tertiary referral centers.
Did remuneration influence the likelihood of providing obstetrics? In this study, physicians who performed deliveries had significantly higher billings than nonproviders of obstetrics. Furthermore, the decline in interest in obstetrics occurred despite the fact that, in Ontario, the fee for a basic obstetrical delivery rose by 34%, from $245 in 1992/1993 to $329 in 2001/2002, compared with only 3.5% for services such as consultations or gynecological surgery during the same time period.18,19 Although we do not know whether the increase in fees prevented the participation rate from falling even further, we suggest that remuneration increases of the scale noted during this time period have had little impact in improving participation in obstetrics. Policy makers may wish to consider other factors of potentially greater importance to the decision to practice obstetrics if they wish to encourage physicians to deliver babies.
Can obstetrician–gynecologists be enticed into returning to obstetrics once they have left? This study shows that once a physician has abandoned obstetrical practice, the likelihood of returning to obstetrics is very low, approximately 1% per year. Furthermore, this study suggests that those who leave obstetrics have a significantly easier lifestyle. Most nonproviders of obstetrics take almost no call duties that result in their having to make a special visit to a hospital or other care setting. Such individuals instead provide office consultations, psychotherapy, scheduled diagnostic or laboratory services, or elective surgery, all of which offer a predictable, scheduled work environment. One hypothesis that deserves further research is that physicians, once they have stopped obstetrics, may realize that the lifestyle is so much less stressful that they will not return to obstetrics no matter what the remuneration.
This study has a number of strengths and limitations. One strength is that it is based on a near census of all specialists practicing in a defined jurisdiction. Second, we believe the data to be accurate, as physicians have a strong financial incentive to code deliveries accurately. The study is limited by the fact that only utilization patterns can be discerned, and information is not available on the underlying motives of physicians behind their withdrawal from obstetrics. The activity of 3% of practitioners was not captured. The measures of on-call participation do not include instances in which a physician is on call but does not physically make a special visit to the hospital; hence, they may underestimate the actual disruption to the physician's lifestyle. Lastly, this study traces only services funded by Ontario's publicly funded insurance plan and excludes most assisted reproductive technologies, such as in vitro fertilization, which are not covered. This study cannot therefore track whether doctors who are relinquishing obstetrics are moving into this potentially lucrative field of practice.
The findings of this study raise concern about who will provide obstetrical services for the population in the future. Fortunately for obstetrical planners in Ontario, total births declined by 13% in the past decade.20 The alternatives to care by obstetrician–gynecologists are family physicians and midwives. However, family physicians have radically decreased their participation in obstetrics in the past decade in Ontario as well.2 Midwifery was introduced in Ontario in 1994,21 and as the number trained increases, midwives will play an important role in the delivery of routine obstetrical care in the future. However, midwives cannot replace specialists for more complex cases. The fact that workload is rising among the declining number of obstetrician–gynecologists who do obstetrics raises concern about sustainability of workload. Maintaining access to obstetrical services will remain an important challenge for policy makers in light of the trends identified in this study.
1. Nesbitt TS, Davidson RC, Paliescheskey M, Fox-Garcia J, Arevalo JA. Trends in maternity care by graduates and the effect of an intervention. Fam Med
2. Chan BT. The declining comprehensiveness of primary care. CMAJ
3. Kruse J, Phillips D, Wesley RM. Withdrawal from maternity care: a comparison of family physicians in Ontario, Canada, and the United States. J Fam Pract
4. Kruse J, Phillips D, Wesley RM. Factors influencing changes in obstetric care provided by family physicians: a national study. J Fam Pract
5. Pearse WH, Haffner WH, Primack A. Effect of gender on the obstetric–gynecologic work force. Obstet Gynecol
6. Nesbitt TS, Baldwin LM. Access to obstetric care. Prim Care
7. Petersdorf RG. An American's view of Canadian medical education. CMAJ
8. Merrill JA. (Sub)specialization in obstetrics and gynecology: results of a survey by the American Board of Obstetrics and Gynecology. Am J Obstet Gynecol
9. Pearse WH, Gant NF, Hagner AP. Workforce projections for subspecialists in obstetrics and gynecology. Obstet Gynecol
10. Baumgardner JR, Marder WD. Specialization among obstetrician/gynecologists: another dimension of physician supply. Med Care
11. Chan BTB. From perceived surplus to perceived shortage: what happened to Canada's physician workforce in the 1990s? Ottawa: Canadian Institute for Health Information; 2002.
12. Jacoby I, Meyer GS, Haffner W, Cheng EY, Potter AL, Pearse WH. Modeling the future workforce of obstetrics and gynecology. Obstet Gynecol
13. Canadian Medical Protective Association. Special Report: 1998 fees driven by court awards and settlements. Ottawa, 1998.
14. Canadian Medical Protective Association. Fee Schedule for 1992. Ottawa, 1992.
15. Canadian Medical Protective Association. Fee Schedule for 2002. Ottawa, 2002.
16. Grumbach K, Vranizan K, Rennie D, Luft HS. Charges for obstetric liability insurance and discontinuation of obstetric practice in New York. J Fam Pract
17. Rosenblatt RA, Weitkamp G, Lloyd M, Schafer B, Winterscheid LC, Hart LG. Why do physicians stop practicing obstetrics? The impact of malpractice claims. Obstet Gynecol
18. Ontario Ministry of Health. Schedule of benefits: physician services under the Health Insurance Act. Toronto: Ontario Ministry of Health; 1992.
19. Ontario Ministry of Health and Long-Term Care. Schedule of benefits: physician services under the Health Insurance Act. Toronto: Ontario Ministry of Health; 2000.
20. Statistics Canada. Report on the demographic situation in Canada 2001. 91-209-XPE. Ottawa: Federal Publications Inc; 2002.
21. MacDonald M. Postmodern negotiations with medical technology: the role of midwifery clients in the new midwifery in Canada. Med Anthropol