Introduction US data reveal a Caesarean rate discrepancy between insured and uninsured patients, with the C-section rate highest among the privately insured. The data have prompted concern that financial incentives associated with insurance status might influence American physicians' decisions to perform Caesarean deliveries.
Objective To determine whether differences in medical risk factors account for the apparent Caesarean rate discrepancy between Medicaid and privately insured patients in Michigan, USA.
Method A retrospective review was performed of 617 269 live birth deliveries in Michigan hospitals during 2004–8. All live birth records that were able to be linked to their mothers' hospital discharge records were utilised. Diagnosis-related group codes from the hospitalisation records were used to identify Caesarean deliveries. Regression models determined Caesarean probability for the time period under study, adjusted for insurance type, maternal age, race, maternal medical conditions, multiple births, prematurity and birth weight.
Results From 2004 to 2008, Caesarean rates were 33% for privately insured patients and 29% for Medicaid patients. The probability of Caesarean delivery was significantly greater for privately insured than Medicaid patients on univariate analysis (OR 1.2, 95% CI 1.19 to 1.22) but not on multivariate analysis (adjusted OR 1.01, 95% CI 0.99 to 1.02).
Conclusion No significant disparity was found in the odds of Caesarean delivery between privately insured and Medicaid patients in Michigan after adjusting for other Caesarean risk factors. A positive disparity would have provided de facto evidence that financial incentives play a role in physician decision-making regarding Caesarean delivery.
- clinical ethics
- empirical bioethics
- insurance reimbursement
- interests of woman/fetus/father
- obstetrics and gynaecology
Statistics from Altmetric.com
- clinical ethics
- empirical bioethics
- insurance reimbursement
- interests of woman/fetus/father
- obstetrics and gynaecology
Caesareans are now the most commonly performed operating room procedures in the USA.1 ,2 From 1996 to 2007, the Caesarean rate increased in all 50 US states for mothers of all ages and races, regardless of pregnancy gestational age.2 In 2007, 32% of all US infants were delivered by Caesarean, representing more than a 50% increase from 1997.2 ,3 The rate is rising despite the WHO recommendation, in 1985, that Caesarean rates should not exceed 15% because rates above this level have been shown to result in increased maternal and/or neonatal morbidity.4
This increasing trend in Caesarean delivery may be causing a widespread public health problem in the USA. For example, between 1996 and 2004, there was an increase in the US preterm birth rate,5 which translates into an increase in infant mortality, even in babies who are only mildly to moderately preterm.6 This increase in the preterm birth rate occurred mainly in infants who had been delivered by Caesarean.5 A critically important unanswered question is how many of these preterm births, especially the mildly preterm babies, were iatrogenically induced secondary to elective (ie, non-medically indicated) Caesareans.
The causes of the steadily increasing Caesarean rate are believed to be multifactorial. Obstetric professionals have cited three main non-medical reasons for the increasing Caesarean rates: mounting medical-legal pressures, increasing maternal requests and decreased obstetric skill in vaginal delivery.7 ,8 Critics outside the profession have alleged that financial incentives and/or institutional convenience pressure choices for Caesarean delivery.7
Given the threat to maternal and child health posed by Caesarean procedures, it is ethically imperative to move beyond speculation and empirically investigate causal factors and their relative weight. Indeed, policy challenges posed by medically unnecessary Caesarean rates may be addressable only through empirically based approaches to bioethics. Such approaches allow social science methodologies for assessing causality and outcome to challenge philosophical appeals that fail to explain current trends—such as pervasive calls to respect principles of autonomy and beneficence. Influenced both by dissatisfaction with overly theoretical approaches to bioethics and by the rise of evidence-based medicine, ‘empirical bioethics’ posits that understanding complex causal pathways through diverse social science methods is a necessary first step towards institutionally embodying prevalent moral values.8
Among potential non-medical risk factors for C-section, the relatively greater financial reward to healthcare providers for surgical birth has attracted particular ethical concern. While it is difficult to explore the overall magnitude of potential effects from financial incentives, several studies have assessed differences in C-section rates across patient populations with different relative payment rates. A decade ago, a study by the National Bureau of Economic Research concluded that lower fee differentials between Caesarean and vaginal childbirths within the Medicaid programme versus within private insurance plans could explain more than half of the discrepancy between Medicaid and private insurance Caesarean rates.9 A study in Greece found parallel differentials in C-section rates between public and private payers (personal communication, 2008 Michigan Blue Cross/Blue Shield Reimbursement Rates, Michigan Blue Cross/Blue Shield Office, Detroit, MI, USA). Utilising the 2006 US HealthCare Cost and Utilisation Database, a more recent American study found Caesarean deliveries were performed for 34% of all privately insured births but only for 25% of uninsured births (personal communication, 2008 Medicaid Reimbursement Rates, Medicaid Division of Michigan Dept of Community Health, Lansing, MI, USA).
Like previous studies of financial influence, we do not plan to analyse the extent to which financial factors influence C-section in general. Rather, we will assess the more delineated question of whether there is a differential financial influence for private-pay versus publicly funded (Medicaid) patients. We will perform this study in the state of Michigan, USA. Answering this question may identify which, if any, Michigan subpopulations may be at greater risk of medically unnecessary C-section.
Materials and methods
We examined trends in the Caesarean delivery rate in the state of Michigan from 2004 to 2008 by tracking the overall rate of increase in Caesarean births and then by examining changes in the rate of medical risk factors for Caesarean section, such as multiple births. Caesarean rate discrepancies between privately and publicly insured (Medicaid) patients were evaluated to determine the potential influence of insurance reimbursement differentials on the Caesarean delivery rate.
Two data sources were utilised; one was the Michigan Department of Community Health's Division of Vital Records, containing live birth data, and the second source was the Michigan Inpatient Hospital Database (MIDB), which contains payer and inpatient-level data for inpatient hospitalisations at all acute care hospitals in Michigan. Linkage between these two databases was performed at a 98% matching rate, thereby effectively linking an infant birth record with their mother's hospitalisation for delivery.
Maternal demographics and insurance information were derived from the MIDB dataset and infant information was derived from the live birth records. Diagnosis-related group codes were used to identify Caesarean deliveries. Maternal medical conditions were identified by the International Statistical Classification of Diseases and Related Health Problems version 9 codes. Approval for this study was granted by both the Michigan Department of Community Health Institutional Review Board and Michigan Health and Hospital Association (MHA), legal owner of the MIDB. SAS V.8.2 software was utilised to carry out all statistical analyses.
Spearman's correlation coefficient was utilised to measure agreement between insurance type and maternal age. Logistic regression models were constructed to determine the association between Caesarean deliveries and the year of birth while adjusting for the following covariates: maternal age, insurance type (private insurance vs Medicaid), race, maternal medical conditions (diabetes, hypertension and/or cardiovascular disease), multiple births, prematurity (<37 weeks gestation), high birth weight (>4500 g) and low birth weight (<2500 g).
Within Michigan hospitals, 617 269 births occurred during 2004–8. The percentage of Caesarean deliveries increased by 12% during those years. Caesarean deliveries peaked at 33% for all babies, with no difference by race. The mean age of women delivering by Caesarean was 28.6 years (SD 6.1 years) and the mean age of those with vaginal births was 26.9 years (SD 5.9 years).
Figure 1 demonstrates the Caesarean delivery trends for Medicaid and privately insured patients. There was, on average, a 4% annual difference between the groups; the Caesarean rate was 33% for privately insured patients and 29% for Medicaid patients. The probability of Caesarean delivery was significantly greater for privately insured than Medicaid patients on univariate analysis (OR 1.2, 95% CI 1.19 to 1.22) but not on multivariate analysis (adjusted OR 1.01, 95% CI 0.99 to 1.02) as seen in table 1. There was a significant correlation demonstrated (Spearman correlation coefficient 0.36, p<0.001) between privately insured patients and increasing maternal age.
Both the unadjusted and adjusted point estimates for the logistic regression odds of Caesarean delivery are shown in table 1 for year of birth (per year increase), maternal age (per year increase), insurance type, multiple births, birth weight, race, prematurity, maternal hypertension, maternal cardiovascular disease and maternal diabetes mellitus. Year of birth had an adjusted OR of 1.05 (95% CI 1.04 to 1.05).
The primary goal of this study was to determine the impact of insurance type (private versus public insurance) on Michigan's Caesarean delivery rate, given the discrepancy observed in reimbursement fees between insurance types and birthing procedures. If the discrepancy between C-section rates is linked to insurance type, several important ethical questions are raised. Does the degree of financial reward for performing a surgical procedure influence obstetricians' decisions to perform Caesarean delivery? Are better-insured women at greater risk of having a C-section unnecessarily? (S. Selders, personal communication). We consider our study to be one step in the right direction towards the development of wider-ranging empirical analyses of C-section rates designed to inform bioethical health policy.
In general, physician payments are higher for Caesarean section than for vaginal delivery within both Medicaid and private plans, while overall payment rates for either C-section or vaginal delivery are generally higher in private plans. However, the proportional public/private payment differential is greater for C-section than for vaginal delivery. For example, in 2008, there was 46% greater reimbursement for vaginal delivery from Michigan Blue Cross/Blue Shield plans than Medicaid compared with 50% greater for C-section between insurance types (personal communication, 2008 Michigan Blue Cross/Blue Shield Reimbursement Rates, Michigan Blue Cross/Blue Shield Office, Detroit, MI, USA; personal communication, 2008 Medicaid Reimbursement Rates, Medicaid Division of Michigan Dept of Community Health, Lansing, MI, USA).
If we examine reimbursement differences within a single insurance plan, we find that in 2008, there was14.8% more reimbursement from Medicaid for Caesarean than vaginal delivery (personal communication, 2008 Medicaid Reimbursement Rates, Medicaid Division of Michigan Dept of Community Health, Lansing, MI, USA), but as high as 18% more between delivery types from Michigan Blue Cross/Blue Shield (personal communication, 2008 Michigan Blue Cross/Blue Shield Reimbursement Rates, Michigan Blue Cross/Blue Shield Office, Detroit, MI, USA). To eliminate the financial advantage of performing Caesarean delivery, the northern Michigan-based insurer, Priority Health, currently has a pilot programme underway in which physicians are offered flat fees for delivery, regardless of delivery type (S. Selders, personal communication). Future studies will be needed to examine the impact of flat-fee delivery reimbursement on Caesarean rates.
Consistent with results from national studies,10 we confirmed that changes in maternal demographics and medical conditions do not fully account for the increasing Caesarean rate for the population of Michigan. The rate of Caesareans in Michigan between the years 2004 and 2008 rose by 11.8%, with no difference by race. A baby born in any given year had a 1.05 times odds of being delivered by Caesarean compared with the previous year, even after correcting for insurance type, maternal age, maternal medical conditions, multiple births, prematurity, infant birth weight and race.
We used univariate and multivariate analysis to assess the hypothesis that insurance type was a significant predictor of C-section risk. By univariate analysis, insurance type was a significant predictor of Caesarean delivery. However, efforts to construct proxies for clearly medically indicated Caesareans in the Michigan dataset were stymied by lacunae in available population-level data, thus providing some limitation to our analysis. For example, health records do not indicate that a given Caesarean delivery had been electively chosen.
To complicate matters further, there may be subjectivity on the part of the physician as to the medical need for a Caesarean delivery; the presence of Caesarean risk factors alone may not automatically necessitate a Caesarean delivery. Penna and Arulkumaran1 noted in 2003 that non-medical indications, such as maternally perceived social convenience and planning and peer group pressures may influence Caesarean delivery rates, as well as physicians' considerations of maternal choice, medicolegal issues, avoidance of criticism, time constraints and convenience and economic gain. In their 2007 study, Savage and Francome7 noted physicians' subjective concerns regarding liability, maternal requests and/or lack of skill were some reasons noted by British medical directors. On the other hand, in regard to maternal delivery preference, Gamble and Creedy11 and Gamble et al 12 have reported that only a small number of women request a C-section in the absence of current or previous obstetric complications. Therefore, focusing on maternal C-section preferences as a reason for the increase in Caesarean rates may divert attention away from physician-led influences.11 ,12 In light of this, perhaps, we can say that when there is a borderline level of medical risk, then the presence of non-medical issues may compel a physician to document the medical need for Caesarean delivery.
By using the available population data, with acknowledgement of its limitations, we found that in a multivariate model adjusted for known Caesarean risk factors, insurance type lost its significance. Furthermore, an earlier study that utilised linked maternal–child records,13 found that after adjusting for maternal comorbidities, the total cost of healthcare related to Caesarean deliveries may be lower than that of vaginal deliveries. The authors found that the total average cost of Caesarean deliveries combined with subsequent neonatal intensive care unit utilisation by the newborn was less than the combined cost of vaginal deliveries and the subsequent need for neonatal intensive care unit utilisation.13 In particular, we identified the risk factor of maternal age as one of the main confounders in our model. The significant positive correlation found between maternal age and insurance type indicated that older mothers may be more likely to have private insurance. We found that the women in our study undergoing Caesarean delivery were on average 18 months older than those giving birth vaginally.
The results from our study were conducted on a large Michigan hospital discharge dataset over a period of 4 years but are likely to be generalisable to other US states. It would be interesting to assess whether the national private–public health disparity would persist after adjustments are made for underlying medical risk indicators for Caesarean delivery, or whether insurance type significance would disappear as it did in our study.
In a 2009 report of health indicators,14 the Organisation of Economic Cooperation and Development noted a continued rise in Caesarean deliveries across all Organisation of Economic Cooperation and Development member countries and that a number of studies from different member countries also identified higher Caesarean delivery rates in private hospitals compared with public facilities.11 However, because the organisation, structure and cost of healthcare in the USA is so different from that of other countries, our findings may not be applicable to other healthcare systems. For example, most economic healthcare analyses in the USA use charges as proxies for costs, but this proxy would not be appropriate for use in the UK.15 Also, in most industrialised countries, Caesarean deliveries generally cost more than other modes of delivery.16 However, the aggregate costs of delivering a baby in the USA is approximately five times higher than a comparable delivery performed outside the USA.16 Therefore, we again reiterate that our results may not be generalisable to other industrialised nations.
We found no significant disparity in the odds of Caesarean delivery between privately insured and Medicaid patients in Michigan after adjusting for other Caesarean risk factors. Positive disparity would have provided de facto evidence that financial incentives play a broader systemic role in physician decision-making regarding Caesarean delivery. Further studies are needed to support our findings and assess whether this null effect regarding insurance type holds true nationally.
Future studies should devise new methodologies to assess potential financial incentives within specific insurance plans or across the healthcare system generally. Customary data collection surrounding deliveries should be interrogated in terms of its adequacy to evaluate both the role of physician financial incentives and the rising C-section rates. For example, current data limitations on physician subjectivity regarding the medical need for C-section and the lack of documented scheduled C-section indications may have possibly influenced our analysis to such an extent that the role of financial incentives may not have been possible to detect. New kinds of data may need to be collected in order to assess rising Caesarean rates critically.
The authors would like to thank Glenn Copeland, State Registrar and director of the Division of Vital Records and Health Statistics and Thu Le, statistician specialist in the same division for their support and assistance. Without Thu Le's expert effort to link the live birth records successfully with hospital discharge data the authors would not have been able to conduct this study.
Competing interests None.
Ethics approval Ethics approval was received from the Institutional Review Board of Michigan Department of Community Health.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.