Do group practices have lower caesarean rates than solo practice obstetric clinics? Evidence from Taiwan
1Department of Health Services Policy and Management, Arnold School of Public Health, University of South Carolina, Columbia, South Carolina, USA 2School of Health Care Administration, Taipei Medical University, Taipei, Taiwan and 3Department of Public Finance, National Taipei University, Taipei, Taiwan
Correspondence: Herng-Ching Lin, School of Health Care Administration, Taipei Medical University, 250 Wu-Hsing St., Taipei 110, Taiwan. Tel: +886-2–2776–1661 ext. 3613; Fax: +886-2–2378–9788; E-mail: henry11111{at}tmu.edu.tw
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionBiographiesReferences |
|---|
Objective: This study examined physicians’ propensity for caesarean deliveries at solo versus group practice obstetrics/gynaecology (ob/gyn) clinics in Taiwan.
Method: We used population-based (National Health Insurance) claims data covering all 253 618 singleton deliveries conducted at ob/gyn clinics, during 2000–02. The dependent variable, delivery mode, was treated as dichotomous [caesarean section (CS) = 1, vaginal delivery (VD) = 0]. The independent variable of interest was practice size, classified into four categories: 1, 2, 3 and 4+ physicians. Multilevel logistic regression modelling, accounting for clinic-level variation in CS rates, was used to examine CS likelihood by practice size, among the total delivery sample and among the sub-samples disaggregated by obstetric complication status.
Results: Solo practices have 7% excess caesarean cases relative to large group practices. After controlling for patient's age, physician demographics, the clinic's geographic location and size of delivery service, and clinic-level random effect, solo practice physicians were 5.38 times as likely as 4+ physician practices to provide caesarean delivery (CI = 4.18 
6.93), 2-physician practices were 3.87 times (CI = 2.99 5.01) and 3-physician practices 2.72 times (CI = 2.06 3.59) as likely as 4+ physician practices to provide caesarean delivery. This effect is driven by higher CS propensity among solo and small groups among cases with obstetrically less salient complications and the ‘no complications’ subset of patients. Wide confidence intervals for odds ratios in these sub-samples also attest to wide variations in clinic-level CS rates among these patient groups.
Conclusions: Solo physicians are the most likely to provide caesarean delivery, and CS likelihood decreases with increasing number of physicians in the practice. Group practice support may reduce the CS likelihood, when it is not clinically indicated. Policy makers should consider initiatives to limit full service delivery privileges to group practice obstetric clinics, in order to reduce unnecessary CS. Solo practice clinics should, at best, be licensed as birthing centres, required to transfer patients needing CS to a larger facility.
Key Words: caesarean delivery, physician behaviour, clinic practice variation
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionBiographiesReferences |
|---|
Caesarean section (CS), particularly, clinically unnecessary CS, is a matter of serious concern. CS rates in many countries are climbing to record highs, without demonstrable health benefits for mothers and neonates. Chile has recorded a CS rate of 40% (Belizan et al. 1999
), South Korea 40% (Lee et al. 2004), Taiwan 32.3% (Lin et al. 2004a) and the US 27.5% in 2003 (Martin et al. 2005). In other Latin American countries, rates similar to Chile are observed (Belizan et al. 1999). The World Health Organization (WHO) recommends a primary CS rate of 15%, and several countries, including the US, have adopted this norm.
In many developing countries, including Taiwan, delivery service is highly decentralized, typically provided through small proprietary clinics, solo or group practices, with basic inpatient and operating room facilities. These obstetrics/gynaecology (ob/gyn) clinics fulfil an important community need for accessible delivery services, geographically and culturally close to where people live and work. Group practice is gaining ground in many countries due to the advantages of increasing returns to scale (Kimbell and Lorant 1977), sharing of infrastructure, personnel, fixed costs and technical investment (Graham 1972), higher physician income (Romano 2001) and better production efficiencies (Brown 1988). In addition to the above advantages, ob/gyn group practice may confer additional advantages that permit a higher clinician threshold for performing CS. These advantages include sharing after-hours care coverage and accessing colleagues’ clinical expertise in difficult situations, such as if a CS decision late in labour endangers the mother or foetus. There is no documentation on comparative clinical outcomes in group versus solo practice settings. In this study, we explore whether a group practice set-up at ob/gyn clinics, on average, predicts better quality of delivery care in the form of reduced rates of clinically unnecessary CS.
We hypothesized that deliveries at solo practice ob/gyn clinics are more likely to culminate in caesarean delivery than those in a group practice setting, due to a lower threshold for a CS decision at solo practices. We also hypothesized that increasing size of the group will be associated with reducing CS propensity. The study findings have policy implications regarding the appropriateness of clinical settings for operative delivery, and suggest the need to encourage group practice formation in specialties with emergent care responsibilities such as obstetrics. Major quality of care issues, maternal and foetal outcomes, and costs are at stake for many countries.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionBiographiesReferences |
|---|
Data source and study sample
We use 3-year, publicly available population-based data from Taiwan's ob/gyn clinics to test our hypothesis. Our data source are the National Health Insurance (NHI) claims data from 2000–02, covering all patient encounters of Taiwan's population of over 23 million. The majority of group practices in Taiwan are single specialty practices (Lin and Chen 2003; Lin et al. 2004b). Specifically, all multi-physician ob/gyn clinics are single specialty groups. Other favourable system characteristics for testing our study hypothesis are universal health care coverage, a single-payer payment system with the government as the sole insurer, and all citizens enjoying access to any medical institution of their choice.
The database provides one ICD-9CM principal diagnosis code and up to four secondary diagnoses for each inpatient. All ob/gyn clinic claims for singleton deliveries between 1 January 2000 and 31 December 2002 were selected that had the ICD-9CM codes of 72 or 73 (representing CS) or 740, 741, 742 or 744 (representing vaginal delivery). All hospital deliveries were excluded. In Taiwan, hospital deliveries almost solely comprise patients who have attended the hospital's outpatient clinic and been admitted by the hospital-employed physicians. Clinic physicians cannot admit their patients to a hospital and continue to attend on the patient. If they refer a patient to a hospital, she is transferred to the care of the hospital's attending physicians. Thus, clinic and hospital deliveries are mutually exclusive populations except for transfer patients, who then get transferred to the hospital's treatment policies.
Among CS cases, the Bureau of NHI recognizes two categories of CS for reimbursement purposes: those provided caesarean delivery at the physician's initiative (NHI's DRG code, 0373A) and maternal request CS (DRG code, 0373B). We considered restricting the sample to physician-directed CS, to examine the effect of practice size on physicians’ CS decisions. We note, however, that a physician may code a maternal request CS as clinically indicated CS to accommodate the mother's preference, and to help the patient avoid the 50% cost-share out-of-pocket that is required by NHI for maternal request CS. The Bureau of NHI reimburses providers for maternal request CS at the same rate as vaginal delivery, which is half the rate of clinically indicated caesarean delivery; the provider has to bill the patient for the difference. It is possible that some clinics, particularly solo clinics, may be more inclined to upcode a maternal request CS to clinically indicated CS. Moreover, ‘maternal request CS’ itself could be influenced by physician preferences. To avoid these sources of bias, we included maternal request CS cases in our sample.
We accounted for clinic-wise variations in CS propensities and diagnosis/DRG upcoding by applying multilevel logistic regression analysis, specifying clinic as a source of random effects. The details are described under ‘Statistical analysis’ below. A total of 253 618 women were included in the study, the universal sample of all clinic deliveries in Taiwan in the study period (96 435, 80 480 and 76 703 cases in 2000, 2001 and 2002, respectively).
Statistical analysis
The statistical package STATA (Version 9.0, STATA Corporation) was used. The dependent variable, delivery mode was treated as dichotomous [caesarean section (CS) = 1, vaginal delivery (VD) = 0]. The independent variable of interest was practice size, classified into four categories: 1, 2, 3 and 4+ physicians. (As noted earlier, all group practices in ob/gyn clinics were single specialty, obstetric practices.)
Multilevel modelling was used to examine the relative likelihood of caesarean delivery by practice size. It accounts for clinic-wise clustering of patients and the associated CS propensity of the practice. Since inter-clinic variations in diagnosis upcoding may confound the CS odds, we analysed seven separate models, the total sample and six obstetric complication categories.
Our data consisted of cross-sectional 3-year data. Two sources of variation in CS odds are possible: fixed effects and random effects. To determine whether the random (clinic-level) effect is a significant predictor of CS likelihood, we evaluated the statistical significance of the panel-level variance component (
). The likelihood ratio test of = 0 and 
= 5978.89 with p value <0.001 was applied (Greene 2003). The test rejected the null hypothesis that the panel estimator is not different from the pooled estimators. This indicated the presence of a significant random effect, to be accommodated in the model estimation.
We controlled for physician characteristics (age and gender), patient characteristics (age, clinical complication and maternal request for CS) and clinic characteristics (geographic location and size of delivery service). All clinical complications were classified according to a hierarchy of mutually exclusive diagnoses, devised by Lin and Xirasagar (2004) drawing from Anderson and Lomas (1984):
- Previous caesarean (ICD-9-CM 654.2);
- Indisputable indications for CS, covering breech presentation (652.2 and 669.6), dystocia (653 and 660–662, excluding 661.3) and foetal distress (656.3);
- Other ante-partum or intra-partum complications potentially justifying a caesarean;
- Complications reflecting pelvic floor/perineal/birth canal injury;
- Other co-morbidities not ordinarily indications for caesarean; and
- No complications, i.e. no secondary diagnosis.
Although the three intra-partum complications covered under ‘Indisputable indications’ could be questioned as being unequivocal indications for CS, the intent here is to control for clinical conditions that cannot be reasonably disputed as an indication for caesarean.
Geographic location was categorized as northern, central, southern and eastern Taiwan. We also controlled for the clinic's size of delivery service because logistic, infrastructure and revenue factors associated with various case loads might influence physicians’ caesarean decisions. Size of delivery service was a continuous variable: the total number of all claims for delivery, vaginal or caesarean (ICD-9CM codes of 72 or 73, 740, 741, 742, and 744), submitted by the clinic during the study period. A significance level of 0.05 for the regression coefficients was selected.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionBiographiesReferences |
|---|
Table 1 shows the distribution of study patients by delivery mode, practice size, patient's age and presence of obstetric complications, and physician's age. The overall CS rate among clinic deliveries was 33.8%. Solo clinics conducted the majority of clinic deliveries (57%). Table 1 also shows the contribution of each obstetric complication group to CS in each practice category. Previous CS and indisputable indications (breech presentation, dystocia and foetal distress) account for 88.9% of caesareans at group practices with 4+ physicians. For clinics with 3, 2 and 1 physician(s) respectively, the corresponding percentages were 79%, 81% and 80.2%.
|
Table 2 shows the adjusted relationship between size of practice setting and CS likelihood, examined for the total sample and separately for each clinical complication group, totalling seven models estimated. After controlling for patient's age, physician demographics and the clinic characteristics (geographic location and size of delivery service), solo practices had the highest caesarean likelihood relative to 4+ physician practices (OR = 5.38; range 4.18 to 6.93), and the odds decline with increasing number of physicians in the practice. Two-physician practices were 3.87 times as likely as 4+ physician practices to provide caesarean delivery (OR range 2.99 to 5.01), compared with 3-physician practices, who were 2.72 times as likely to provide caesarean delivery (range 2.06 to 3.59).
|
When disaggregated by obstetric complication status, we find that among the ‘No complication’ group (total 164 829), solo practices were 35.5 times as likely to offer caesarean delivery as the 4+ practices, and 2-physician practices 49.53 times as likely as 4+ practices to offer CS. Both solo and 2-physician practice types show very wide confidence intervals (though not spanning 1.00), indicating that these two practice types are significantly more likely to offer CS delivery than 4+ physician practices. The 3-physician practices are not statistically significantly different from 4+ physician practices.
Among ‘previous CS’ and ‘pelvic floor injury’ samples, the models did not converge, due to zero or near zero denominators. The ‘previous CS’ subset of the study sample had a CS rate of 99.8% across all practice sizes, and ‘pelvic floor injuries’ had a vaginal delivery rate of 99.9% across all practice sizes (Table 1). The sub-sample ‘Indisputable CS indications’ had a CS rate of 99.9%, and although this model converged to produce point estimates and confidence intervals for each practice type, all the confidence intervals are extremely wide, spanning 1.00, implying statistically insignificant odds ratios. Among the sample of ‘complications justifying CS’, the CS rate is 92.8%, and the model results show that practices with three or less physicians have significantly higher CS likelihood than 4+ physician practices. Wide confidence intervals suggest considerable inter-clinic variation in CS likelihood among cases with these complications. Among the sub-sample of ‘co-morbidities not ordinarily indications for CS’, the CS rate is 90.6%, but the delivery numbers in the 3-physician and 4+ physician practice groups are too small (150 and 71, respectively), causing loss of statistical power.
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionBiographiesReferences |
|---|
With increasing emphasis on market mechanisms in health care internationally, it is essential to identify the entrepreneurial arrangements that best foster high quality care. Many authors have documented associations between hospital institutional characteristics and caesarean delivery (McKenzie and Stephenson 1993
; Di Lallo et al. 1996; Clark et al. 1998). There is no documentation on the technical quality of health care at small proprietary facilities that dominate the health services landscape in many countries. This study examined the effect of group versus solo practice, and among group practices, the size of practice, on physicians’ CS propensity, using 3-year population-based data from Taiwan. We find that solo physicians are consistently the most likely to provide caesarean delivery, and among group practices, smaller groups are more likely to offer CS delivery than larger groups, after controlling for physician's gender and age, patient's age, geographical location of clinic and size of delivery service. When disaggregated by obstetric complication status, we find consistently high CS rates among all practice sizes for patients with obstetrically salient complications. This suggests that the differences in CS propensities are driven by cases with complications that test the physician's judgment on the delivery mode, and by those with no complications. Adjusted odds ratios support this expectation: among the ‘no complications’ sub-sample, the model shows significantly high CS odds among solo and 2-physician practices relative to larger practices. Wide confidence intervals reflect high inter-clinic variation in CS propensities, a random effect issue that is accounted for and yet unequivocally shows decreasing CS odds with increasing number of physicians in the practice. Collectively, our results provide empirical support for our hypothesis.
It could be argued that the high odds ratios between solo and lower order groups may be largely a function of very large sample sizes, rather than clinically significant differences in CS rates. However, the complication groups with uniform, near 100% CS rates at all clinic categories (previous CS and indisputable indications for CS) account for a significantly higher proportion of the case loads at 4+ physician group practices relative to solo and lower order practices (88.9% versus approximately 80% at lower order practices). This substantiates the case that solo and smaller practices have a significantly increased CS propensity, driven by lower threshold for CS in cases with clinically insignificant or no complications.
Overall, comparing crude rates in solo practices and large (4+) group practices (34% versus 31.7%), solo practices had an excess of 3841 caesareans relative to 4+ groups, accounting for 7% of their total caesarean deliveries (applying 31.7% to 144 362, solo clinics should have had 45 762 caesarean deliveries instead of 49 604). Considering that almost 57% of clinic deliveries in Taiwan (144 362 out of 253 618) took place in solo practices during 2000–02, a 7% excess rate translates into a substantial excess CS rate attributable to delivery services at solo practices.
We propose three possible reasons for the heightened caesarean propensity of solo physicians. Time pressure or convenience may play a significant role in obstetricians’ CS decisions, particularly in solo practice settings. Obstetricians in group practice clinics can share after-hours call coverage, which increasingly resembles a hospital's 24-hour, regular delivery service. Carpenter et al. (1987) found that CS rates could be reduced if obstetricians shared night calls with at least two other colleagues.
In countries with third-party payment systems, group practice formation is facilitated, because the elaborate documentation and claims procedures required by third- party payers prompts physicians to consider hiring skilled administrative staff to deal with the administrative burden. Affordability of administrative staff greatly improves if multiple physicians form groups and share costs. In the process, there arises an opportunity for clinical production efficiencies, sharing call coverage and sharing clinical expertise to back up one another in difficult clinical situations. Empirical studies by Brown (1988) and Burns et al. (1995) strongly suggest that under time pressure, physicians may choose caesarean delivery, the oft-cited ‘physician convenience’ factor. Past data from Taiwan have shown evidence of high time pressures among solo practice physicians, and high dissatisfaction with the working hours (Lin 1999; Lin et al. 2003). In this study, we are unable to identify the extent to which time pressure or lack of access to back-up expertise contribute to the increased CS propensity of solo practices and small groups.
Although many western countries, such as the US, do not currently permit CS in clinics, the outpatient practice setting is likely to affect caesarean propensity even if the delivery itself is conducted in a hospital setting. This is because group practice physicians typically share their off-duty patient care responsibilities through call rotations, which helps when an obstetrician is concurrently called upon to attend to delivery cases admitted to different hospitals.
Apart from clinical outcomes, group practice confers other advantages for patients. An earlier study by Lin et al. (2004b) in Taiwan found that group practice outpatients, regardless of specialty, perceived significantly better service quality than those who received service at a solo practice office. The current study, showing significantly better technical quality of obstetric care, extends the earlier study's findings on consumer-perceived service quality in group practices. Together, these studies substantiate the intuitive expectations that group practice may be more satisfying and clinically beneficial for patients, while also logistically efficient and economically beneficial for providers, as documented in other studies.
Policy implications
Our findings readily generalize to many low- and middle-income countries. Similar to Taiwan, in many countries, physicians practicing in freestanding clinics cannot admit their patients to hospitals. Many low-income countries have a substantial presence of small proprietary obstetric clinics providing delivery services. Many also have a substantial, if not, major presence of third-party payer systems. Our findings suggest that policymakers and payers should consider initiatives to encourage group practice formation in high-risk specialties such as obstetrics, as a pre-condition for providing full service delivery care or interventional ambulatory care procedures. Decentralized delivery services fulfil an important community need for accessible and convenient obstetric care. Our findings, however, make a clear case for re-designating solo practice obstetric clinics as birthing centres, required to transfer patients needing CS to a larger institution, in line with the current practice of birthing centres in many western countries.
Our study also offers a precautionary note for many western countries, where the privatization of health care and its financing is ushering in ingenious ways of cost containment. An increasingly popular option, particularly in the United States, is to establish freestanding ambulatory care clinics and short-stay surgical centres, often owned by physicians. These operative care facilities are thought to cut costs by enabling low-risk inpatient care procedures to be carried out in a facility without the expensive back-up of intensive care units. This concept appears likely to extend to obstetric care in many western countries in the foreseeable future. The findings of our study have precautionary relevance for this issue.
Study limitations and future research
There are a few limitations to this study. First, being an administrative claims dataset, it lacks information on the physicians’ ownership or employee relationships within group practices. Groups may be organized as a partnership, sole proprietorship with employment of the remaining members, or as a lease-based relationship to share facilities and expenses. Physician incentives to favour CS or otherwise, as well as the pattern of sharing patient care responsibilities, may vary by the type of association. Accounting for these factors would enable more refined conclusions and, therefore, targeted policy initiatives.
Secondly, being an administrative (claims) database, it lacks information on the stage of labour or antecedent duration of labour prior to the CS operation. This information could better substantiate the antecedent factors driving higher CS propensity in solo and small group practices, the time pressure issue or lack of professional back-up support. Our database also lacks information on maternal socioeconomic characteristics, parity and neonatal birth weight which may influence physicians’ considerations about the delivery mode.
Thirdly, the cross-sectional nature of the data limits the scope of conclusions about causality. Longitudinal studies would enable tracking of physicians through their stints in solo practice, moving into group practice, and vice versa. Such studies would facilitate a conclusive determination of whether a group practice setting reduces CS propensity as a cause-and-effect sequence.
Finally, it is essential to substantiate the professional factors underlying this phenomenon, using survey research or other qualitative research methods such as focus groups. Such studies will enable an understanding of the extent to which time pressures, convenience or lack of back-up expertise drive solo or small group practice physicians’ CS decisions. Explication of the causes will provide additional information to policymakers to devise appropriate policy initiatives, with the unequivocal approval of obstetric professional bodies. Despite the above limitations, our findings of increasing CS propensity with decreasing practice membership have important policy implications for the quality of obstetric care, internationally.
|
Biographies |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionBiographiesReferences |
|---|
Dr Sudha Xirasagar is a Research Assistant Professor of health services policy and management at the University of South Carolina. Her research interests include health policy and system development, health care financing and reform, and clinical leadership development. Dr Xirasagar received her medical degree from Bangalore University in India and her Ph.D. in health services administration from the University of South Carolina at Columbia, in the United States.
Dr Herng-Ching Lin is an Associate Professor of health care administration at Taipei Medical University, Taipei, Taiwan. His research interests include payment system, primary health care and quality of medical care. Dr Lin received a Ph.D. in health care administration from the University of South Carolina at Columbia.
Dr Tsai-Ching Liu has a Ph.D. in health economics from the University of North Carolina at Chapel Hill, USA. She has been engaged in research in the areas of health policy and health care financing. At present, she works as a Professor in the Department of Public Finance, National Taipei University, Taiwan.
|
Acknowledgements |
|---|
This study is based on data from the National Health Insurance Research Database provided by the Bureau of National Health Insurance, Department of Health, Taiwan and managed by the National Health Research Institutes. The interpretations and conclusions contained herein do not represent those of the Bureau of National Health Insurance, Department of Health or the National Health Research Institutes.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionBiographiesReferences |
|---|
Anderson GM and Lomas J. 1984. Determinants of the increasing cesarean birth rate. Ontario data 1979 to 1982. New England Journal of Medicine 311:887–92.[Abstract]
. Anonymous. 1996. Doctors flocking to group practices. Hospital and Health Networks 70:47.
Belizan JM, Althabe F, Barros FC, et al. 1999. Rates and implications of caesarean sections in Latin America: ecological study. British Medical Journal 319:1397–400.
Brown DM. 1988. Do physicians underutilize aides? Journal of Human Resources 23:342–55.[CrossRef]
Burns LR, Geller SE, Wholey DR. 1995. The effect of physician factors on the cesarean section decision. Medical Care 33:365–82.[ISI][Medline]
Carpenter MW, Soule D, Yates WT, et al. 1987. Practice environment is associated with obstetric decision making regarding abnormal labor. Obstetrics and Gynecology 70:657–62.
Clark SL, Xu W, Porter TF, et al. 1998. Institutional influences on the primary cesarean section rate in Utah, 1992 to 1995. American Journal of Obstetrics and Gynecology 179:841–845.[Medline]
Di Lallo D, Perucci CA, Bertollini R, et al. 1996. Cesarean section rates by type of maternity unit and level of obstetric care: an area-based study in central Italy. Preventive Medicine 25:178–85.[Medline]
Goer H. 2001. The case against elective cesarean section. Journal of Perinatal and Neonatal Nursing 15:23–38.[ISI][Medline]
Graham FE. 1972. Group versus solo practice: arguments and evidence. Inquiry 9:49–60.[Medline]
Greene W. 2003. Econometric analysis (5th edn) New Jersey Prentice Hall, Inc.
Kimbell LJ and Lorant JH. 1977. Physician productivity and returns to scale. Health Services Research 12:367–79.[ISI][Medline]
Lee SI, Khang YH, Lee MS. 2004. Women's attitudes toward mode of delivery in South Korea – a society with high cesarean section rates. Birth 31:108–16.[Medline]
Lin YY. 1999. The impact of National Health Insurance on physician productivity and income. Hospital 32:1–14 (in Chinese).
Lin HC and Chen CC. 2003. Self-evaluation of group practice physicians on their practice models in Taiwan. Medical Education 3:258–70 (in Chinese).
Lin HC and Xirasagar S. 2004. Institutional factors in cesarean delivery rates: policy and research implications. Obstetrics and Gynecology 103:128–36.
Lin HC, Chang WY, Tung YC. 2003. Factors related to dissatisfaction with the National Health Insurance among primary care physicians in Taiwan. Chang Gung Medical Journal 26:81–90.[Medline]
Lin HC, Sheen TC, Tang CH, et al. 2004a. Association between maternal age and the likelihood of a cesarean section: a population-based multivariate logistic regression analysis. Acta Obstetricia et Gynecologica Scandinavica 83:1178–83.[Medline]
Lin HC, Xirasagar S, Laditka JN. 2004b. Patient perceptions of service quality in group versus solo practice clinics. International Journal for Quality in Health Care 16:437–45.
Martin JA, Hamilton BE, Sutton PD, et al. 2005. Births: Final data for 2003. National Vital Statistics Reports Hyattsville, MD National Center for Health Statistics volume 54:no 2 Available at: [http://www.cdc.gov/nchs/data/nvsr/nvsr54/nvsr54_02.pdf].
McKenzie L and Stephenson PA. 1993. Variation in cesarean section rates among hospitals in Washington State. American Journal of Public Health 83:1109–12.
Morrison J and MacKenzie IZ. 2003. Cesarean section on demand. Seminars in Perinatology 27:20–33.[CrossRef][Medline]
Romano M. 2001. United we do quite well, thank you. Physicians say practicing in groups is a better way to provide care. Modern Healthcare 31:34–5.[Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||