- Email: [email protected]
Physician gender and cesarean sections
Journal of Clinical Epidemiology 53 (2000) 1030–1035
Physician gender and cesarean sections Lloyd K. Mitlera, John A. Rizzoa,b, Sarah M. Horwitza,b,c* a
Department of Epidemiology & Public Health, Yale University School of Medicine, 60 College St., Box 208034 New Haven, CT 06520, USA b Institution for Social and Policy Studies, Yale University, New Haven, CT 06520, USA c Child Study Center, Yale University School of Medicine, New Haven, CT 06520, USA Received 1 January 1998; received in revised form 9 January 2000; accepted 26 January 2000
Abstract Background: Among consumers insurers, and providers there is pervasive concern regarding the high incidence of cesarean section delivery. To date, attempts to reduce these rates have focused on the clinical behavior of providers resulting in only minimal changes. Therefore, non-medical variables must be investigated as potential explanatory factors for the decision to perform cesarean delivery. Methods: Data were collected on clinical and non-clinical factors for obstetrician-gynecologists delivering at Yale-New Haven Medical Center to measure the impact of these factors on the performance of cesarean sections. Specifically, variation in patient demographic, ante- and intra-partum risk variables, practice setting, and doctor-specific characteristics were examined. Using contingency table and logistic regression analyses the contribution of selected factors was evaluated. Results: Multivariate modeling revealed that male physicians were significantly more likely than their female colleagues to perform cesarean section. This relationship was particularly strong in the university practice setting. Conclusions: Efforts to reduce the incidence of cesarean section need to focus on the continuing education of health care providers and the delineation of non-clinical factors as essential elements in the election of specific clinical therapies. © 2000 Elsevier Science Inc. All rights reserved. Keywords: Cesarean section; Physicians’ practice patterns; Clinical uncertainty; Physician gender
1. Introduction In 1995 cesarean sections accounted for 21.0% of live births in the United States [1], a rate that has nearly quadrupled over the prior two decades. While 14.5% of all deliveries are now accomplished by a primary cesarean operation, repeat cesarean surgery accounts for 8.3% of total births [2]. Cesarean section comprises 4% of all surgical procedures in short-term, non-federal hospitals, at a cost twice that of routine vaginal birth [3]. The greater expense of cesarean section reflects both higher hospital and physician charges, as well as the substantial expenditures resulting from increased maternal and neonatal morbidity [4]. Among consumers, insurers, and providers there is pervasive concern regarding the high incidence of cesarean section delivery. To date, attempts to reduce these rates have focused on the appropriateness of clinical indicators, resulting in only minimal changes. Moreover, cesarean section rates have been found to vary substantially by patient and other nonclinical factors as well [5]. Consequently, it is important to understand any non-medical factors driving the decision to employ the cesarean operation. * Corresponding author. Tel.: 203-786-2854; fax: 203-785-6287. E-mail address: [email protected] (S.M. Horwitz)
Patient, medical practice, and physician-specific factors may each play a role in shaping the cesarean section decision. Patient demographic characteristics associated with the incidence of cesarean surgery include advancing maternal age, white race, first pregnancy, and higher socioeconomic status [6–8]. Practice setting may influence the decision to perform cesarean delivery as well. For example, the probability of cesarean section is lower for pregnant women receiving pre-natal care in a staff model HMO compared to women in a fee-for-service plan [9]. Additionally, provider concerns about the malpractice environment, preferences regarding convenience and time allocation, the quality of physician–patient interactions, events during medical training, and the perceived safety of cesarean section may also contribute to the final treatment decision. Because there are few adequate measures for such factors, the impact of provider attitudes and beliefs has remained largely unstudied. Physician characteristics that have been shown to affect the incidence of cesarean delivery include graduation from a foreign medical school and board certification, possibly reflecting factors relating to provider level of training or to patient case-mix [10]. Interestingly, previous research examining physician gender and the cesarean decision has found little association [5,10,11]. This is puzzling because in the context of the
0895-4356/00/$ – see front matter © 2000 Elsevier Science Inc. All rights reserved. PII: S0895-4356(00)00 2 2 1 - 3
L.K. Mitler et al. / Journal of Clinical Epidemiology 53 (2000) 1030–1035
birthing process, female obstetricians may on average have better information about the costs and benefits of the alternative modes of delivery and may be more likely to communicate this knowledge to the patient [12]. Another important, though largely ignored, factor is managed care because it may also affect the cesarean decision by limiting physician discretion. Non-clinical factors such as the opportunity cost of time and provider preferences may have less impact on the cesarean section decision in a managed care setting where individual decision making is more limited [13]. The thesis of the present investigation is that non-clinical factors such as physician gender are important in situations involving clinical uncertainty and provider discretion. Consequently, this study evaluates the impact of such factors in three practice settings. Specifically, the study tests for gender differences in the cesarean section decision in university practice, private practice, and health maintenance organization (HMO) practice. The university practice in our sample treats substantially more complicated maternal or intrauterine fetal illnesses; hence there is potentially more uncertainty as to the suitable course of medical treatment. Further, clinical uncertainty may be exacerbated by the youth and inexperience of the house staff. At the opposite extreme, within HMOs, physicians have less latitude and fewer options in their treatment decisions, and patients tend to be younger, at lower risk, and thus less likely to be candidates for cesarean birth. Relative to the university setting, clinical ambiguity should be less
1031
evident in private pay patients receiving care in the community, due to referral of the most medically ill and obstetrically complex mothers to the high-risk practices at medical centers. We anticipate that gender-specific differences in the cesarean section decision will be most pronounced in the university practice setting, both because clinical uncertainty in deciding appropriate treatment course is likely to be greatest, and because physicians have greater discretion in choosing the appropriate treatment course than in the HMO setting.
2. Methods 2.1. Research design This study examines the choice of cesarean delivery or vaginal birth at Yale-New Haven Hospital. The obstetrical service includes a clinic linked to a faculty-supervised highrisk service, two staff model health maintenance organizations, and a large private practice service that provides a majority of the total pregnancy-related care in the community. This data set is based on a sample frame selected from among 4460 total deliveries including 960 cesarean sections for the year 1990 and 3875 total deliveries including 837 cesarean sections for the year 1995. An unmatched case–control study was performed by examining the first 65 available records by date of discharges from the obstetrical service for vaginal and cesarean deliveries from among the selected months, yielding a total sam-
Table 1 Sociodemographic characteristics of mothers, neonates, and delivering physicians at Yale-New Haven Hospital in 1990 and 1995—overall and by practice setting
Characteristic MD characteristics Male (%) Mean physician age (years) Teaching appointment (%) Board certified (%) Fetal characteristics Abnormal fetal test (%) Mean number other conditions Birth weight (1000s grams) Apgar score Oxytocin infusion (%) Evening delivery (%) Delivery year 1995 (%) Maternal characteristics Mean maternal age (years) Black race (%) Nulliparity (%) Smoker (%) Substance abuser (%) Toxemia (%) Comorbid illness (%) Infection (%)
University practice (n ⫽ 308)
Private practice (n ⫽ 546)
HMO practice (n ⫽ 146)
C-sec (n ⫽ 140)
No C-sec (n ⫽ 168)
C-sec (n ⫽ 293)
No C-sec (n ⫽ 253)
C-sec (n ⫽ 66)
No C-sec (n ⫽ 80)
C-sec (n ⫽ 499)
No C-sec (n ⫽ 501)
68.6 36.3 90.1 72.9
41.1 33.9 29.8 22.6
75.1 45.3 95.2 94.5
54.5 44.1 73.1 72.7
19.7 36.2 78.8 74.2
10.0 35.8 27.5 20.0
65.9 41.6 91.8 85.8
42.9 39.4 51.3 47.5
40.1 0.2 2.7 8.2 38.6 42.9 55.6
20.8 0.05 3.0 8.5 34.5 48.2 57.1
15.4 0.1 3.4 8.8 39.6 37.9 48.1
9.1 0.04 3.4 8.9 39.5 45.8 49.4
15.2 0.05 3.6 8.9 53.0 45.5 47.0
8.8 0.04 3.3 8.6 33.8 48.8 36.3
22.4 0.1 3.3 8.7 41.1 40.3 50.0
13.0 0.04 3.3 8.7 36.9 47.1 50.0
26.4 42.9 43.6 49.3 25.0 10.7 26.4 16.4
24.2 50.6 33.3 38.1 25.0 4.2 16.1 10.1
31.9 6.8 46.4 13.0 8.2 6.5 13.0 3.1
29.9 5.1 43.1 12.6 5.9 2.4 6.7 0.004
31.4 16.7 54.5 18.2 12.1 9.1 15.2 10.6
29.7 22.5 41.3 20.0 8.8 0.05 0.05 0.04
30.3 18.2 46.7 23.8 13.4 8.2 17.0 7.8
28.0 23.2 39.5 22.4 12.8 3.4 9.6 4.2
C-sec ⫽ patient has had prior cesarean section; No C-sec ⫽ patient has not had prior cesarean section.
All practices (n ⫽ 1000)
1032
L.K. Mitler et al. / Journal of Clinical Epidemiology 53 (2000) 1030–1035
ple of 1000 births divided evenly for the years 1990 and 1995. Charts were chosen equally from the months of February, May, August, and November to balance the effect of house staff training, which begins each July, on the clinical performance of the physicians who staff the university service as resident physicians or fellows. Patient sociodemographic characteristics, prominent ante-partum exposures and intra-partum events, practice and insurance setting, and type of delivery, were abstracted from the hospital records by one of he authors (L.K.M.). Physician demographic data were supplied by the Office of the Chief of the Medical Staff. To estimate appropriate sample size, a ratio of control to cases of 1:1, 80% power, and an alpha (Type I) error of P ⫽ 0.05 were assumed. Clinical characteristics of pregnancies included comorbid illness, testing indicating fetal jeopardy, toxemia, multiple gestation, presence of bleeding or infection including sexually transmitted disease, congenital anomalies of the newborn, and the use of oxytocin infusion. Patient-specific variables included maternal age, race, parity, marital status, geographic residence, smoking behavior and substance abuse profile, type of insurance, and year of birth. Physician and practice identifiers included provider age, gender, university teaching appointment, board certification, graduation from foreign medical school, and practice setting (uni-
versity, community, or HMO). Because previous studies have found a relationship between the time of delivery (6:00 p.m.–6:00 a.m.) and the admission to delivery interval to the performance of cesarean section, this information was collected as well [14–16]. 2.2. Analyses Contingency table analysis was used to determine the association of potential explanatory factors to the outcome of interest. Logistic regression analysis was performed to isolate the individual effects of the various explanatory variables. Separate models were estimated for the entire data set and for each of the three practice settings. Due to the strong correlation between past and current cesarean sections, we elected not to include previous cesarean section as an explanatory variable in these models. Instead, we estimated additional models on a subset of mothers without a previous cesarean delivery, to see if important differences in model estimation results emerged. 3. Results Table 1 provides summary characteristics for the overall sample and for each of the three practice settings. Separate
Table 2 Contingency table analysis of cesarian section controlling for practice type Percent cesarean sectiona Characteristics MD characteristics Male Physician age ⬎42 Teaching appointment Board certified Fetal characteristics Abnormal fetal test Number other conditions Birth weight ⬍4000 gm Low apgar scorec Oxytocin infusion Evening delivery Delivery year 1995 Maternal characteristics Maternal age ⬎35 Black race Nulliparity Smoker Substance abuser Toxemia Comorbid illness Infection
University practice
Private practice
HMO practice
All practices
Odds ratiob
58.2 (96/165) 45.7 (16/35) 71.8 (127/177) 72.9 (102/140)
61.5 (220/358) 53.6 (143/267) 60.1 (279/464) 60.1 (277/461)
61.9 (13/21) 58.3 (7/12) 70.3 (52/74) 75.4 (49/65)
60.5 (329/544) 52.9 (166/314) 64.1 (458/715) 64.3 (428/666)
2.57* 1.19 10.61* 6.66*
62.0 (57/92) 79.5 (31/39) 43.7 (125/286) 57.1 (12/21) 48.2 (54/112) 42.6 (60/141) 44.8 (78/174)
66.2 (45/68) 67.7 (21/31) 52.0 (232/446) 77.8 (7/9) 53.7 (116/216) 48.9 (111/227) 53.0 (141/266)
58.8 (10/17) 50.0 (3/6) 40.8 (49/120) 0.0 (0/3) 56.5 (35/62) 43.5 (30/69) 51.7 (31/60)
63.3 (112/177) 72.4 (55/76) 47.7 (406/852) 57.6 (19/33) 52.6 (205/390) 46.0 (201/437) 50.0 (250/500)
1.94* 2.83* 0.54* 1.38 1.19 0.76* 1.00
75.0 (18/24) 41.4 (60/145) 52.1 (61/117) 51.9 (69/133) 45.5 (35/77) 68.2 (15/22) 57.8 (37/64) 57.5 (23/40)
66.0 (64/97) 60.6 (20/33) 55.5 (136/245) 54.3 (38/70) 61.5 (24/39) 76.0 (19/25) 69.1 (38/55) 90.0 (9/10)
60.0 (15/25) 37.9 (11/29) 52.2 (36/69) 42.9 (12/28) 53.3 (8/15) 60.0 (6/10) 71.4 (10/14) 70.0 (7/10)
66.4 (97/146) 44.0 (91/207) 54.1 (233/431) 51.5 (119/231) 51.2 (67/131) 70.2 (40/57) 63.9 (85/133) 65.0 (39/60)
2.37* 0.68* 1.34* 1.09 1.06 2.48* 1.94* 1.94*
a For each practice setting, these percentages represent the fraction of all deliveries with a given characteristic that wree delivered by cesarean section. Thus, for the characteristic “delivered by a male physician,” 96/165, or 58.2% of such deliveries were by cesarean section in the university practice setting, 61.5% in the private practice setting, and so on. Figures in parentheses thus represent the numbers of deliveries with that characteristic that were delivered via cesarean section divided by the total number of deliveries with that characteristic. b For each percent cesarean section characteristic, this is the odds ratio across all practices. Thus, the odds ratio that a delivery by a male physician across all practices is via cesarean section is 2.57. c A low Apgar score is defined as a score below 7. *Statistically significant at the 1% level.
L.K. Mitler et al. / Journal of Clinical Epidemiology 53 (2000) 1030–1035
summary statistics are provided according to whether delivery was via cesarean section or vaginal routes. For mothers who received cesarean sections, mean age was 30.3 years and 18.2% of these mothers were black. Nearly a quarter (23.8%) of these mothers were smokers. Comorbid illness was present in 17.0% and toxemia in 8.2% of all mothers who had cesarean section. Abnormal fetal tests occurred in 22.4% of these neonates. The mean physician age for cesarean delivery was 41.6 years, and 65.9% of these attending physicians were male. Male physicians performed 543 deliveries and female doctors 457. For the cohort delivered vaginally, mean maternal age is slightly lower. These mothers are slightly more likely to be black and to have other children. The incidence of abnormal fetal tests, toxemia, comorbid illness, and infection are all lower among the vaginal delivery cohort. Physician characteristics differ as well. Mothers delivered vaginally are much less likely to have had a male physician, a board certified physician, or a physician with a teaching appointment. Table 1 also reveals dramatic differences in the characteristics of mothers and neonates across practice settings. In the university practice, 49.3% of mothers receiving cesarean sections are smokers—more than three times the incidence of smokers in private settings, and twice the incidence in the HMO. Fetal risk is substantially higher in the university practice setting, as indicated by the markedly higher rates of comorbid illness and abnormal test for fetal well-being. Relative to private practice physicians, mean physician age is substantially lower in the university practice and HMO practice settings. Table 2 shows the results of the contingency table analysis. A characteristic associated with a cesarean section rate substantially above or below 50% suggests that cesarean surgery rates differ along this feature. The cesarean section rate is markedly higher among male doctors: regardless of practice setting, the cesarean section rate was approximately 60% among male physicians. University teaching appointment and board certification are other physician characteristics associated with a higher incidence of cesarean section. Fetal characteristics associated with a higher incidence of cesarean section include: abnormal test for fetal wellbeing and other fetal comorbidities (bleeding, congenital defect, birth injury, and multiple pregnancy). Evening deliveries were less likely to have been cesarean sections. Older age of the mother and nulliparity are associated with a higher incidence of cesarean section. Mothers who are black are less likely to receive cesarean sections. Comorbid medical illness in the mother, clinically apparent maternal infection (e.g., sexually transmitted disease), and toxemia were all associated with a significantly greater probability of cesarean section. Contingency table analysis thus provides a provocative result: physician gender appears to affect the choice of cesarean section quite strongly. Several factors, including provider age as a surrogate for practice experience, teaching appointment, and board-certification status, are potentially
1033
Table 3 Logistic regression for the outcome of cesarian section for the entire data set Characteristics Physician Male Physician age Teaching appointment Board certified Fetal Abnormal fetal test Number other conditions Birth weight Apgar score Oxytocin infusion Evening delivery Delivery year 1995 Maternal Maternal age Black race Nulliparity Smoker Substance abuser Toxemia Comorbid illness Infection
Odds ratio
95% CI for odds ratio
1.38** 0.99 6.73* 1.52
1.00–1.88 0.97–1.00 3.29–13.76 0.78–2.96
1.83* 2.27* 1.39* 1.03 0.94 0.75 0.78
1.16–2.88 1.22–4.22 1.09–1.79 0.89–1.19 0.68–1.31 0.56–1.02 0.58–1.06
1.07* 1.45 1.66* 1.57** 1.24 2.09** 1.61** 2.70*
1.04–1.10 0.95–2.20 1.21–2.29 1.06–2.33 0.76–2.03 1.02–4.28 1.03–2.54 1.39–5.26
*Statistically significant at the 1% level. **Statistically significant at the 5% level.
correlated with gender and might confound any association. Therefore, using multivariate logistic regression, the outcomes for the entire data set are provided in Table 3. Table 4 reports gender-specific effects estimated from separate multivariate models for each of the three practice settings. As the results in Table 3 indicate, the odds ratio for male physician gender is 1.38 (95% CI: 1.00–1.88) in the full sample. Table 4 points to substantial differences in this relationship across practice settings. In particular, the odds ratios for male physician gender is 2.82 (95% CI: 1.43–5.55) for the university practice setting, and 1.65 (95% CI: 1.05– 2.60) for private practice. Physician gender has no discernible effect on the performance of cesarean surgery in the HMO setting.
Table 4 Odds ratios for the outcome of cesarian section for a male MD Practice setting University practice All patients No prior c-section Private practice All patients No prior c-section HMO practice All patients No prior c-section All practice settings All patients No prior c-section
Odds ratio
95% CI for odds ratio
2.82* 2.75*
1.43–5.55 1.35–5.60
1.65** 1.75**
1.05–2.60 1.06–2.88
0.46 0.43
0.11–1.86 0.10–1.90
1.38** 1.31
1.00–1.88 0.93–1.83
*Statistically significant at the 1% level. **Statistically significant at the 5% level.
1034
L.K. Mitler et al. / Journal of Clinical Epidemiology 53 (2000) 1030–1035
Other findings from Table 3 indicate that abnormal fetal tests, toxemia, the presence of comorbid illness, higher birth weight, nulliparity, increasing maternal age, and smoking are independently associated with a greater likelihood of cesarean delivery. Physicians who hold teaching appointments are also significantly more likely to perform cesarean sections. 4. Discussion This study has investigated the determinants of the cesarean section decision. While clinical differences do play an important role in explaining outcomes, the results of this investigation suggest that non-clinical variables are important predictors of cesarean delivery as well. It is reasonable to assume that non-clinical factors are likely to influence the decision-making process when there is uncertainty regarding the appropriateness of a medical judgment [17]. In such cases, uncertainty affords the physician latitude to be more sensitive to non-medical influences such as potential malpractice repercussions, patient satisfaction, time constraints including coverage arrangements or office scheduling, and physician characteristics [18–21]. Other authors that have addressed the issue of gender effect and cesarean surgery [5,11,12]; ours finds an important relationship between physician gender and the probability of cesarean section. The precise reasons for the differences between our findings and those of previous researchers are unclear. One possibility is that different types of data sets were used. Previous studies employed data that were statewide in nature but were not based on chart review; our sample was gleaned from patient medical charts at a major referral center. Each approach has advantages and limitations. Statewide data are more generalizable, yet they often lack the clinical information and possibly the accuracy from a detailed chart review by an experienced clinician. Perhaps the most important reason for the difference between our results and those of earlier research is that we tested for differences in physician gender across practice settings. Gender differences are greatest in the university practice setting. As noted above, neonates in the university practice setting are substantially sicker than in the other two settings, and their mothers are much more likely to be smokers. With such a high-risk patient mix, there is likely to be more uncertainty as to the appropriate delivery mode; vaginal delivery is not a clear choice in many of these patients. When it is less clear as to which treatment option is preferable, physicians may rely more on their own beliefs and preferences in reaching a decision. It is precisely under these circumstances that we would expect gender-specific differences in treatment philosophies to be most pronounced. The gender effect is highly significant in the private practice setting, but smaller in magnitude, a pattern which may reflect the reduced clinical uncertainty associated with a substantially healthier patient population. We find no significant relationship between gender and the cesarean section decision in the HMO setting, where physician choice has significant boundaries. While this
finding suggests that managed care limits the importance of non-clinical factors such as physician gender on the cesarean section decision, the available sample of HMO patients was relatively small so that this result should be interpreted with caution. These gender differences are similar to those observed in the use of preventive services between male and female general practitioners. Specifically, female GPs offer their women patients more gender-specific preventive services, such as mammograms and PAP smears, than do their male counterparts [22,23]. Moreover, differences persist even after patient mix and practice characteristics have been controlled for [23]. Provider gender has also been shown to be an important predictor of surgery, with female physicians reported to perform numerically one-half the number of hysterectomies as compared to male doctors [24,25]. Although the reasons for the gender differences need to be explained, in the final analysis, the reduction in the number of cesarean deliveries among women physicians may occur because they have a better innate understanding of the female body and its physical processes such as child birth, and they also communicate and educate their patients before and during labor more effectively than their male counterparts. 5. Conclusion This examination of cesarean section has revealed the important effect of physician gender on the cesarean section decision in non-managed care settings, an effect which was particularly large in the university practice. These findings may reflect gender-specific differences in the perception and limits of the doctor–patient relationship, or gender differences in treatment philosophy. Further investigation is required to explore gender differences in terms of the structure, process, and outcome of cesarean delivery and other medical treatments, and to better understand how and why treatment patterns may differ among male and female providers.
Acknowledgments The authors gratefully acknowledge the cooperation of Dr. Edwin Cadman, Chief of Medical Staff at Yale-New Haven Hospital for his assistance in providing physician demographic data.
References [1] Sachs BP, Kobelin C, Castro MA, Frigoletto F. The risks of lowering the cesarean-delivery rate. N Engl J Med 1999;340:54–7. [2] Morbidity and Mortality Weekly Report. Rates of cesarean delivery: United States, 1993. Morbidity and Mortality Weekly Report 1995; 44:303–7. [3] Mushinski MA. Average charges for uncomplicated cesarean and vaginal deliveries in the United States, 1993. Stat Bull Metro Ins 1994;75:27–36. [4] Gardner LP. Economic considerations in cesarean section use, In:
L.K. Mitler et al. / Journal of Clinical Epidemiology 53 (2000) 1030–1035
[5] [6] [7] [8] [9]
[10]
[11]
[12] [13] [14] [15]
Flamm, BL, Quilligan, EJ, editors. Cesarean Section. New York: Springer Verlag, 1995. pp. 173–90. Burns LR, Geller SE, Wholey DR. The effect of physician factors on the cesarean section decision. Med Care 1993;33:365–82. Peipert JF, Bracken MB. Maternal age: an independent risk factor for cesarean delivery. Obstet Gynecol 1993;81:200–5. Placek PJ, Taffel SM. Recent patterns in cesarean delivery in the United States. Obstet Gynecol Clinics N Am 1988;15:607–27. Shearer EL. Cesarean section: medical benefits and costs. Soc Sci Med 1993;37:1223–331. Stafford RS. Cesarian section use and source of payment: an analysis of California hospital discharge abstracts. Am J Pub Health 1990;80: 313–5. Tussing AD, Wojtowycz MA. The effect of physician characteristics on clinical behavior: cesarian section in New York state. Soc Sci Med 1993;37:1251–60. Berkowitz G, Fiarman G, Mojica M, Bauman J, Haynes de Regt R. Effect of physician characteristics on the cesarean birth rate. Am J Obstet Gynecol 1989;360:146–9. Sakala C. Midwifery care and out-of-hospital birth settings: how do they reduce unnecessary cesarean births? Soc Sci Med 1993;3:1233–50. Luft HS. Assessing the evidence on HMO performance. Milbank Quart 1980;58:501–36. Evans MI, Richardson DA, Sholl JS. Cesarean section: assessment of the convenience factor. J Reprod Med 1984;29:670–6. Seitchik J, Holden AE, Castillo M. Amniotomy and oxytocin treat-
[16] [17] [18] [19]
[20] [21] [22]
[23]
[24] [25]
1035
ment of functional dystocia and route of delivery. Am J Obstet Gynecol 1986;155:585–92. Goyert G, Bottoms SF. The physician factor in cesarean birth rates. N Engl J Med 1989;320:706–9. Wennberg JE, Barnes B, Zubkoff M. Professional uncertainty and the problem of supplier-induced demand. Soc Sci Med 1978;6:811–24. Paul RH. Reducing the cesarean delivery rate, In: Phelan JP, Clark SL, editors. Cesarean Section. New York: Elsevier, 1988. pp. 462–6. Localio AR, Lawthers AG, Bengsten JM, Herbert LE, Weaver SL, Brennan TA, Landis JR. Relationship between malpractice claims and cesarean delivery. JAMA 1993;296:366–73. Gruber J, Owings M. Physician financial incentives and cesarean section delivery. Rand J Econ 1996;27:99–123. Keeler EB, Brodie M. Economic incentives in the choice between vaginal delivery and cesarean section. Milbank Quart 1993;71:365–404. Lurie N, Slater J, McGovern P, Ekstrom J, Quam L, Margolis K. Preventive care for women: does the sex of the physician matter? N Engl J Med 1993;329:478–82. Brit H, Bhasale A, Miles DA, Meza A, Sayer G, Angelis M. The sex of the general practitioner: a comparison of characteristics, patients, and medical conditions managed. Medical Care 1996;34:403–15. Domenighetti G, Luraschi P, Marazzi A. Hysterectomy and the sex of the gynecologist. (letter) N Engl J Med 1985;313:1482. Bickell NA, Earp JA, Garrett JM, Evans AJ. Gynecologist’s sex, clinical beliefs, and hysterectomy rates. Am J Pub Health 1994;84: 1649–52.
Physician gender and cesarean sections Lloyd K. Mitlera, John A. Rizzoa,b, Sarah M. Horwitza,b,c* a
Department of Epidemiology & Public Health, Yale University School of Medicine, 60 College St., Box 208034 New Haven, CT 06520, USA b Institution for Social and Policy Studies, Yale University, New Haven, CT 06520, USA c Child Study Center, Yale University School of Medicine, New Haven, CT 06520, USA Received 1 January 1998; received in revised form 9 January 2000; accepted 26 January 2000
Abstract Background: Among consumers insurers, and providers there is pervasive concern regarding the high incidence of cesarean section delivery. To date, attempts to reduce these rates have focused on the clinical behavior of providers resulting in only minimal changes. Therefore, non-medical variables must be investigated as potential explanatory factors for the decision to perform cesarean delivery. Methods: Data were collected on clinical and non-clinical factors for obstetrician-gynecologists delivering at Yale-New Haven Medical Center to measure the impact of these factors on the performance of cesarean sections. Specifically, variation in patient demographic, ante- and intra-partum risk variables, practice setting, and doctor-specific characteristics were examined. Using contingency table and logistic regression analyses the contribution of selected factors was evaluated. Results: Multivariate modeling revealed that male physicians were significantly more likely than their female colleagues to perform cesarean section. This relationship was particularly strong in the university practice setting. Conclusions: Efforts to reduce the incidence of cesarean section need to focus on the continuing education of health care providers and the delineation of non-clinical factors as essential elements in the election of specific clinical therapies. © 2000 Elsevier Science Inc. All rights reserved. Keywords: Cesarean section; Physicians’ practice patterns; Clinical uncertainty; Physician gender
1. Introduction In 1995 cesarean sections accounted for 21.0% of live births in the United States [1], a rate that has nearly quadrupled over the prior two decades. While 14.5% of all deliveries are now accomplished by a primary cesarean operation, repeat cesarean surgery accounts for 8.3% of total births [2]. Cesarean section comprises 4% of all surgical procedures in short-term, non-federal hospitals, at a cost twice that of routine vaginal birth [3]. The greater expense of cesarean section reflects both higher hospital and physician charges, as well as the substantial expenditures resulting from increased maternal and neonatal morbidity [4]. Among consumers, insurers, and providers there is pervasive concern regarding the high incidence of cesarean section delivery. To date, attempts to reduce these rates have focused on the appropriateness of clinical indicators, resulting in only minimal changes. Moreover, cesarean section rates have been found to vary substantially by patient and other nonclinical factors as well [5]. Consequently, it is important to understand any non-medical factors driving the decision to employ the cesarean operation. * Corresponding author. Tel.: 203-786-2854; fax: 203-785-6287. E-mail address: [email protected] (S.M. Horwitz)
Patient, medical practice, and physician-specific factors may each play a role in shaping the cesarean section decision. Patient demographic characteristics associated with the incidence of cesarean surgery include advancing maternal age, white race, first pregnancy, and higher socioeconomic status [6–8]. Practice setting may influence the decision to perform cesarean delivery as well. For example, the probability of cesarean section is lower for pregnant women receiving pre-natal care in a staff model HMO compared to women in a fee-for-service plan [9]. Additionally, provider concerns about the malpractice environment, preferences regarding convenience and time allocation, the quality of physician–patient interactions, events during medical training, and the perceived safety of cesarean section may also contribute to the final treatment decision. Because there are few adequate measures for such factors, the impact of provider attitudes and beliefs has remained largely unstudied. Physician characteristics that have been shown to affect the incidence of cesarean delivery include graduation from a foreign medical school and board certification, possibly reflecting factors relating to provider level of training or to patient case-mix [10]. Interestingly, previous research examining physician gender and the cesarean decision has found little association [5,10,11]. This is puzzling because in the context of the
0895-4356/00/$ – see front matter © 2000 Elsevier Science Inc. All rights reserved. PII: S0895-4356(00)00 2 2 1 - 3
L.K. Mitler et al. / Journal of Clinical Epidemiology 53 (2000) 1030–1035
birthing process, female obstetricians may on average have better information about the costs and benefits of the alternative modes of delivery and may be more likely to communicate this knowledge to the patient [12]. Another important, though largely ignored, factor is managed care because it may also affect the cesarean decision by limiting physician discretion. Non-clinical factors such as the opportunity cost of time and provider preferences may have less impact on the cesarean section decision in a managed care setting where individual decision making is more limited [13]. The thesis of the present investigation is that non-clinical factors such as physician gender are important in situations involving clinical uncertainty and provider discretion. Consequently, this study evaluates the impact of such factors in three practice settings. Specifically, the study tests for gender differences in the cesarean section decision in university practice, private practice, and health maintenance organization (HMO) practice. The university practice in our sample treats substantially more complicated maternal or intrauterine fetal illnesses; hence there is potentially more uncertainty as to the suitable course of medical treatment. Further, clinical uncertainty may be exacerbated by the youth and inexperience of the house staff. At the opposite extreme, within HMOs, physicians have less latitude and fewer options in their treatment decisions, and patients tend to be younger, at lower risk, and thus less likely to be candidates for cesarean birth. Relative to the university setting, clinical ambiguity should be less
1031
evident in private pay patients receiving care in the community, due to referral of the most medically ill and obstetrically complex mothers to the high-risk practices at medical centers. We anticipate that gender-specific differences in the cesarean section decision will be most pronounced in the university practice setting, both because clinical uncertainty in deciding appropriate treatment course is likely to be greatest, and because physicians have greater discretion in choosing the appropriate treatment course than in the HMO setting.
2. Methods 2.1. Research design This study examines the choice of cesarean delivery or vaginal birth at Yale-New Haven Hospital. The obstetrical service includes a clinic linked to a faculty-supervised highrisk service, two staff model health maintenance organizations, and a large private practice service that provides a majority of the total pregnancy-related care in the community. This data set is based on a sample frame selected from among 4460 total deliveries including 960 cesarean sections for the year 1990 and 3875 total deliveries including 837 cesarean sections for the year 1995. An unmatched case–control study was performed by examining the first 65 available records by date of discharges from the obstetrical service for vaginal and cesarean deliveries from among the selected months, yielding a total sam-
Table 1 Sociodemographic characteristics of mothers, neonates, and delivering physicians at Yale-New Haven Hospital in 1990 and 1995—overall and by practice setting
Characteristic MD characteristics Male (%) Mean physician age (years) Teaching appointment (%) Board certified (%) Fetal characteristics Abnormal fetal test (%) Mean number other conditions Birth weight (1000s grams) Apgar score Oxytocin infusion (%) Evening delivery (%) Delivery year 1995 (%) Maternal characteristics Mean maternal age (years) Black race (%) Nulliparity (%) Smoker (%) Substance abuser (%) Toxemia (%) Comorbid illness (%) Infection (%)
University practice (n ⫽ 308)
Private practice (n ⫽ 546)
HMO practice (n ⫽ 146)
C-sec (n ⫽ 140)
No C-sec (n ⫽ 168)
C-sec (n ⫽ 293)
No C-sec (n ⫽ 253)
C-sec (n ⫽ 66)
No C-sec (n ⫽ 80)
C-sec (n ⫽ 499)
No C-sec (n ⫽ 501)
68.6 36.3 90.1 72.9
41.1 33.9 29.8 22.6
75.1 45.3 95.2 94.5
54.5 44.1 73.1 72.7
19.7 36.2 78.8 74.2
10.0 35.8 27.5 20.0
65.9 41.6 91.8 85.8
42.9 39.4 51.3 47.5
40.1 0.2 2.7 8.2 38.6 42.9 55.6
20.8 0.05 3.0 8.5 34.5 48.2 57.1
15.4 0.1 3.4 8.8 39.6 37.9 48.1
9.1 0.04 3.4 8.9 39.5 45.8 49.4
15.2 0.05 3.6 8.9 53.0 45.5 47.0
8.8 0.04 3.3 8.6 33.8 48.8 36.3
22.4 0.1 3.3 8.7 41.1 40.3 50.0
13.0 0.04 3.3 8.7 36.9 47.1 50.0
26.4 42.9 43.6 49.3 25.0 10.7 26.4 16.4
24.2 50.6 33.3 38.1 25.0 4.2 16.1 10.1
31.9 6.8 46.4 13.0 8.2 6.5 13.0 3.1
29.9 5.1 43.1 12.6 5.9 2.4 6.7 0.004
31.4 16.7 54.5 18.2 12.1 9.1 15.2 10.6
29.7 22.5 41.3 20.0 8.8 0.05 0.05 0.04
30.3 18.2 46.7 23.8 13.4 8.2 17.0 7.8
28.0 23.2 39.5 22.4 12.8 3.4 9.6 4.2
C-sec ⫽ patient has had prior cesarean section; No C-sec ⫽ patient has not had prior cesarean section.
All practices (n ⫽ 1000)
1032
L.K. Mitler et al. / Journal of Clinical Epidemiology 53 (2000) 1030–1035
ple of 1000 births divided evenly for the years 1990 and 1995. Charts were chosen equally from the months of February, May, August, and November to balance the effect of house staff training, which begins each July, on the clinical performance of the physicians who staff the university service as resident physicians or fellows. Patient sociodemographic characteristics, prominent ante-partum exposures and intra-partum events, practice and insurance setting, and type of delivery, were abstracted from the hospital records by one of he authors (L.K.M.). Physician demographic data were supplied by the Office of the Chief of the Medical Staff. To estimate appropriate sample size, a ratio of control to cases of 1:1, 80% power, and an alpha (Type I) error of P ⫽ 0.05 were assumed. Clinical characteristics of pregnancies included comorbid illness, testing indicating fetal jeopardy, toxemia, multiple gestation, presence of bleeding or infection including sexually transmitted disease, congenital anomalies of the newborn, and the use of oxytocin infusion. Patient-specific variables included maternal age, race, parity, marital status, geographic residence, smoking behavior and substance abuse profile, type of insurance, and year of birth. Physician and practice identifiers included provider age, gender, university teaching appointment, board certification, graduation from foreign medical school, and practice setting (uni-
versity, community, or HMO). Because previous studies have found a relationship between the time of delivery (6:00 p.m.–6:00 a.m.) and the admission to delivery interval to the performance of cesarean section, this information was collected as well [14–16]. 2.2. Analyses Contingency table analysis was used to determine the association of potential explanatory factors to the outcome of interest. Logistic regression analysis was performed to isolate the individual effects of the various explanatory variables. Separate models were estimated for the entire data set and for each of the three practice settings. Due to the strong correlation between past and current cesarean sections, we elected not to include previous cesarean section as an explanatory variable in these models. Instead, we estimated additional models on a subset of mothers without a previous cesarean delivery, to see if important differences in model estimation results emerged. 3. Results Table 1 provides summary characteristics for the overall sample and for each of the three practice settings. Separate
Table 2 Contingency table analysis of cesarian section controlling for practice type Percent cesarean sectiona Characteristics MD characteristics Male Physician age ⬎42 Teaching appointment Board certified Fetal characteristics Abnormal fetal test Number other conditions Birth weight ⬍4000 gm Low apgar scorec Oxytocin infusion Evening delivery Delivery year 1995 Maternal characteristics Maternal age ⬎35 Black race Nulliparity Smoker Substance abuser Toxemia Comorbid illness Infection
University practice
Private practice
HMO practice
All practices
Odds ratiob
58.2 (96/165) 45.7 (16/35) 71.8 (127/177) 72.9 (102/140)
61.5 (220/358) 53.6 (143/267) 60.1 (279/464) 60.1 (277/461)
61.9 (13/21) 58.3 (7/12) 70.3 (52/74) 75.4 (49/65)
60.5 (329/544) 52.9 (166/314) 64.1 (458/715) 64.3 (428/666)
2.57* 1.19 10.61* 6.66*
62.0 (57/92) 79.5 (31/39) 43.7 (125/286) 57.1 (12/21) 48.2 (54/112) 42.6 (60/141) 44.8 (78/174)
66.2 (45/68) 67.7 (21/31) 52.0 (232/446) 77.8 (7/9) 53.7 (116/216) 48.9 (111/227) 53.0 (141/266)
58.8 (10/17) 50.0 (3/6) 40.8 (49/120) 0.0 (0/3) 56.5 (35/62) 43.5 (30/69) 51.7 (31/60)
63.3 (112/177) 72.4 (55/76) 47.7 (406/852) 57.6 (19/33) 52.6 (205/390) 46.0 (201/437) 50.0 (250/500)
1.94* 2.83* 0.54* 1.38 1.19 0.76* 1.00
75.0 (18/24) 41.4 (60/145) 52.1 (61/117) 51.9 (69/133) 45.5 (35/77) 68.2 (15/22) 57.8 (37/64) 57.5 (23/40)
66.0 (64/97) 60.6 (20/33) 55.5 (136/245) 54.3 (38/70) 61.5 (24/39) 76.0 (19/25) 69.1 (38/55) 90.0 (9/10)
60.0 (15/25) 37.9 (11/29) 52.2 (36/69) 42.9 (12/28) 53.3 (8/15) 60.0 (6/10) 71.4 (10/14) 70.0 (7/10)
66.4 (97/146) 44.0 (91/207) 54.1 (233/431) 51.5 (119/231) 51.2 (67/131) 70.2 (40/57) 63.9 (85/133) 65.0 (39/60)
2.37* 0.68* 1.34* 1.09 1.06 2.48* 1.94* 1.94*
a For each practice setting, these percentages represent the fraction of all deliveries with a given characteristic that wree delivered by cesarean section. Thus, for the characteristic “delivered by a male physician,” 96/165, or 58.2% of such deliveries were by cesarean section in the university practice setting, 61.5% in the private practice setting, and so on. Figures in parentheses thus represent the numbers of deliveries with that characteristic that were delivered via cesarean section divided by the total number of deliveries with that characteristic. b For each percent cesarean section characteristic, this is the odds ratio across all practices. Thus, the odds ratio that a delivery by a male physician across all practices is via cesarean section is 2.57. c A low Apgar score is defined as a score below 7. *Statistically significant at the 1% level.
L.K. Mitler et al. / Journal of Clinical Epidemiology 53 (2000) 1030–1035
summary statistics are provided according to whether delivery was via cesarean section or vaginal routes. For mothers who received cesarean sections, mean age was 30.3 years and 18.2% of these mothers were black. Nearly a quarter (23.8%) of these mothers were smokers. Comorbid illness was present in 17.0% and toxemia in 8.2% of all mothers who had cesarean section. Abnormal fetal tests occurred in 22.4% of these neonates. The mean physician age for cesarean delivery was 41.6 years, and 65.9% of these attending physicians were male. Male physicians performed 543 deliveries and female doctors 457. For the cohort delivered vaginally, mean maternal age is slightly lower. These mothers are slightly more likely to be black and to have other children. The incidence of abnormal fetal tests, toxemia, comorbid illness, and infection are all lower among the vaginal delivery cohort. Physician characteristics differ as well. Mothers delivered vaginally are much less likely to have had a male physician, a board certified physician, or a physician with a teaching appointment. Table 1 also reveals dramatic differences in the characteristics of mothers and neonates across practice settings. In the university practice, 49.3% of mothers receiving cesarean sections are smokers—more than three times the incidence of smokers in private settings, and twice the incidence in the HMO. Fetal risk is substantially higher in the university practice setting, as indicated by the markedly higher rates of comorbid illness and abnormal test for fetal well-being. Relative to private practice physicians, mean physician age is substantially lower in the university practice and HMO practice settings. Table 2 shows the results of the contingency table analysis. A characteristic associated with a cesarean section rate substantially above or below 50% suggests that cesarean surgery rates differ along this feature. The cesarean section rate is markedly higher among male doctors: regardless of practice setting, the cesarean section rate was approximately 60% among male physicians. University teaching appointment and board certification are other physician characteristics associated with a higher incidence of cesarean section. Fetal characteristics associated with a higher incidence of cesarean section include: abnormal test for fetal wellbeing and other fetal comorbidities (bleeding, congenital defect, birth injury, and multiple pregnancy). Evening deliveries were less likely to have been cesarean sections. Older age of the mother and nulliparity are associated with a higher incidence of cesarean section. Mothers who are black are less likely to receive cesarean sections. Comorbid medical illness in the mother, clinically apparent maternal infection (e.g., sexually transmitted disease), and toxemia were all associated with a significantly greater probability of cesarean section. Contingency table analysis thus provides a provocative result: physician gender appears to affect the choice of cesarean section quite strongly. Several factors, including provider age as a surrogate for practice experience, teaching appointment, and board-certification status, are potentially
1033
Table 3 Logistic regression for the outcome of cesarian section for the entire data set Characteristics Physician Male Physician age Teaching appointment Board certified Fetal Abnormal fetal test Number other conditions Birth weight Apgar score Oxytocin infusion Evening delivery Delivery year 1995 Maternal Maternal age Black race Nulliparity Smoker Substance abuser Toxemia Comorbid illness Infection
Odds ratio
95% CI for odds ratio
1.38** 0.99 6.73* 1.52
1.00–1.88 0.97–1.00 3.29–13.76 0.78–2.96
1.83* 2.27* 1.39* 1.03 0.94 0.75 0.78
1.16–2.88 1.22–4.22 1.09–1.79 0.89–1.19 0.68–1.31 0.56–1.02 0.58–1.06
1.07* 1.45 1.66* 1.57** 1.24 2.09** 1.61** 2.70*
1.04–1.10 0.95–2.20 1.21–2.29 1.06–2.33 0.76–2.03 1.02–4.28 1.03–2.54 1.39–5.26
*Statistically significant at the 1% level. **Statistically significant at the 5% level.
correlated with gender and might confound any association. Therefore, using multivariate logistic regression, the outcomes for the entire data set are provided in Table 3. Table 4 reports gender-specific effects estimated from separate multivariate models for each of the three practice settings. As the results in Table 3 indicate, the odds ratio for male physician gender is 1.38 (95% CI: 1.00–1.88) in the full sample. Table 4 points to substantial differences in this relationship across practice settings. In particular, the odds ratios for male physician gender is 2.82 (95% CI: 1.43–5.55) for the university practice setting, and 1.65 (95% CI: 1.05– 2.60) for private practice. Physician gender has no discernible effect on the performance of cesarean surgery in the HMO setting.
Table 4 Odds ratios for the outcome of cesarian section for a male MD Practice setting University practice All patients No prior c-section Private practice All patients No prior c-section HMO practice All patients No prior c-section All practice settings All patients No prior c-section
Odds ratio
95% CI for odds ratio
2.82* 2.75*
1.43–5.55 1.35–5.60
1.65** 1.75**
1.05–2.60 1.06–2.88
0.46 0.43
0.11–1.86 0.10–1.90
1.38** 1.31
1.00–1.88 0.93–1.83
*Statistically significant at the 1% level. **Statistically significant at the 5% level.
1034
L.K. Mitler et al. / Journal of Clinical Epidemiology 53 (2000) 1030–1035
Other findings from Table 3 indicate that abnormal fetal tests, toxemia, the presence of comorbid illness, higher birth weight, nulliparity, increasing maternal age, and smoking are independently associated with a greater likelihood of cesarean delivery. Physicians who hold teaching appointments are also significantly more likely to perform cesarean sections. 4. Discussion This study has investigated the determinants of the cesarean section decision. While clinical differences do play an important role in explaining outcomes, the results of this investigation suggest that non-clinical variables are important predictors of cesarean delivery as well. It is reasonable to assume that non-clinical factors are likely to influence the decision-making process when there is uncertainty regarding the appropriateness of a medical judgment [17]. In such cases, uncertainty affords the physician latitude to be more sensitive to non-medical influences such as potential malpractice repercussions, patient satisfaction, time constraints including coverage arrangements or office scheduling, and physician characteristics [18–21]. Other authors that have addressed the issue of gender effect and cesarean surgery [5,11,12]; ours finds an important relationship between physician gender and the probability of cesarean section. The precise reasons for the differences between our findings and those of previous researchers are unclear. One possibility is that different types of data sets were used. Previous studies employed data that were statewide in nature but were not based on chart review; our sample was gleaned from patient medical charts at a major referral center. Each approach has advantages and limitations. Statewide data are more generalizable, yet they often lack the clinical information and possibly the accuracy from a detailed chart review by an experienced clinician. Perhaps the most important reason for the difference between our results and those of earlier research is that we tested for differences in physician gender across practice settings. Gender differences are greatest in the university practice setting. As noted above, neonates in the university practice setting are substantially sicker than in the other two settings, and their mothers are much more likely to be smokers. With such a high-risk patient mix, there is likely to be more uncertainty as to the appropriate delivery mode; vaginal delivery is not a clear choice in many of these patients. When it is less clear as to which treatment option is preferable, physicians may rely more on their own beliefs and preferences in reaching a decision. It is precisely under these circumstances that we would expect gender-specific differences in treatment philosophies to be most pronounced. The gender effect is highly significant in the private practice setting, but smaller in magnitude, a pattern which may reflect the reduced clinical uncertainty associated with a substantially healthier patient population. We find no significant relationship between gender and the cesarean section decision in the HMO setting, where physician choice has significant boundaries. While this
finding suggests that managed care limits the importance of non-clinical factors such as physician gender on the cesarean section decision, the available sample of HMO patients was relatively small so that this result should be interpreted with caution. These gender differences are similar to those observed in the use of preventive services between male and female general practitioners. Specifically, female GPs offer their women patients more gender-specific preventive services, such as mammograms and PAP smears, than do their male counterparts [22,23]. Moreover, differences persist even after patient mix and practice characteristics have been controlled for [23]. Provider gender has also been shown to be an important predictor of surgery, with female physicians reported to perform numerically one-half the number of hysterectomies as compared to male doctors [24,25]. Although the reasons for the gender differences need to be explained, in the final analysis, the reduction in the number of cesarean deliveries among women physicians may occur because they have a better innate understanding of the female body and its physical processes such as child birth, and they also communicate and educate their patients before and during labor more effectively than their male counterparts. 5. Conclusion This examination of cesarean section has revealed the important effect of physician gender on the cesarean section decision in non-managed care settings, an effect which was particularly large in the university practice. These findings may reflect gender-specific differences in the perception and limits of the doctor–patient relationship, or gender differences in treatment philosophy. Further investigation is required to explore gender differences in terms of the structure, process, and outcome of cesarean delivery and other medical treatments, and to better understand how and why treatment patterns may differ among male and female providers.
Acknowledgments The authors gratefully acknowledge the cooperation of Dr. Edwin Cadman, Chief of Medical Staff at Yale-New Haven Hospital for his assistance in providing physician demographic data.
References [1] Sachs BP, Kobelin C, Castro MA, Frigoletto F. The risks of lowering the cesarean-delivery rate. N Engl J Med 1999;340:54–7. [2] Morbidity and Mortality Weekly Report. Rates of cesarean delivery: United States, 1993. Morbidity and Mortality Weekly Report 1995; 44:303–7. [3] Mushinski MA. Average charges for uncomplicated cesarean and vaginal deliveries in the United States, 1993. Stat Bull Metro Ins 1994;75:27–36. [4] Gardner LP. Economic considerations in cesarean section use, In:
L.K. Mitler et al. / Journal of Clinical Epidemiology 53 (2000) 1030–1035
[5] [6] [7] [8] [9]
[10]
[11]
[12] [13] [14] [15]
Flamm, BL, Quilligan, EJ, editors. Cesarean Section. New York: Springer Verlag, 1995. pp. 173–90. Burns LR, Geller SE, Wholey DR. The effect of physician factors on the cesarean section decision. Med Care 1993;33:365–82. Peipert JF, Bracken MB. Maternal age: an independent risk factor for cesarean delivery. Obstet Gynecol 1993;81:200–5. Placek PJ, Taffel SM. Recent patterns in cesarean delivery in the United States. Obstet Gynecol Clinics N Am 1988;15:607–27. Shearer EL. Cesarean section: medical benefits and costs. Soc Sci Med 1993;37:1223–331. Stafford RS. Cesarian section use and source of payment: an analysis of California hospital discharge abstracts. Am J Pub Health 1990;80: 313–5. Tussing AD, Wojtowycz MA. The effect of physician characteristics on clinical behavior: cesarian section in New York state. Soc Sci Med 1993;37:1251–60. Berkowitz G, Fiarman G, Mojica M, Bauman J, Haynes de Regt R. Effect of physician characteristics on the cesarean birth rate. Am J Obstet Gynecol 1989;360:146–9. Sakala C. Midwifery care and out-of-hospital birth settings: how do they reduce unnecessary cesarean births? Soc Sci Med 1993;3:1233–50. Luft HS. Assessing the evidence on HMO performance. Milbank Quart 1980;58:501–36. Evans MI, Richardson DA, Sholl JS. Cesarean section: assessment of the convenience factor. J Reprod Med 1984;29:670–6. Seitchik J, Holden AE, Castillo M. Amniotomy and oxytocin treat-
[16] [17] [18] [19]
[20] [21] [22]
[23]
[24] [25]
1035
ment of functional dystocia and route of delivery. Am J Obstet Gynecol 1986;155:585–92. Goyert G, Bottoms SF. The physician factor in cesarean birth rates. N Engl J Med 1989;320:706–9. Wennberg JE, Barnes B, Zubkoff M. Professional uncertainty and the problem of supplier-induced demand. Soc Sci Med 1978;6:811–24. Paul RH. Reducing the cesarean delivery rate, In: Phelan JP, Clark SL, editors. Cesarean Section. New York: Elsevier, 1988. pp. 462–6. Localio AR, Lawthers AG, Bengsten JM, Herbert LE, Weaver SL, Brennan TA, Landis JR. Relationship between malpractice claims and cesarean delivery. JAMA 1993;296:366–73. Gruber J, Owings M. Physician financial incentives and cesarean section delivery. Rand J Econ 1996;27:99–123. Keeler EB, Brodie M. Economic incentives in the choice between vaginal delivery and cesarean section. Milbank Quart 1993;71:365–404. Lurie N, Slater J, McGovern P, Ekstrom J, Quam L, Margolis K. Preventive care for women: does the sex of the physician matter? N Engl J Med 1993;329:478–82. Brit H, Bhasale A, Miles DA, Meza A, Sayer G, Angelis M. The sex of the general practitioner: a comparison of characteristics, patients, and medical conditions managed. Medical Care 1996;34:403–15. Domenighetti G, Luraschi P, Marazzi A. Hysterectomy and the sex of the gynecologist. (letter) N Engl J Med 1985;313:1482. Bickell NA, Earp JA, Garrett JM, Evans AJ. Gynecologist’s sex, clinical beliefs, and hysterectomy rates. Am J Pub Health 1994;84: 1649–52.