The influence of medical insurance on obstetrical care

Abstract

Maternal and obstetrical outcomes vary widely within the United States. The impact of insurance type on health care disparities and its influence on obstetrical care and maternal outcome is not clear. We report the impact of health care insurance on obstetrical and maternal outcomes in a tertiary care health care system. Our maternal quality care database (n = 4199) was queried comparing commercial insurance to government sponsored insurance from July 1, 2015 through June 30, 2018. Parturients with commercial insurance were older, weighed more, presented with less gravidity and parity, had more advanced gestation, and had a higher neonatal 5-minute Apgar score than government insured parturients. Additionally, government insured parturients were less likely to be admitted for induction with oxytocin, receive labor epidural analgesia, and have a primary caesarean delivery. Similarly, government insured parturients were more likely to be of African American descent, be a current known smoker, have a positive urine drug screen, and receive a general anesthetic. We conclude obstetrical and maternal health care disparities exist based on medical insurance type.

INTRODUCTION

Childbirth is a leading cause for hospitalization in the United States (US) and is the largest category for hospital admissions with an estimated annual hospital cost of over $100 billion for insurance payers.¹ We report the impact of health care insurance on obstetrical and maternal outcomes in a tertiary care health care system.

BACKGROUND

Among childbirth deliveries, 49.4% were paid via private insurance, 42.6% via Medicaid, and 4.1% “self-pay,” or deliveries by uninsured mothers.² It has been reported that obstetrical and neonatal outcomes vary widely across the United States,^1,3,4 and there is growing evidence in the literature of racial and ethnic disparities in maternal and infant outcomes.^1,5

LITERATURE REVIEW

In black women compared to white women, infant mortality rates are two times higher and maternal mortality rates are three to four times higher.^1,6,7 Infant and maternal mortality rates are also higher in some Hispanic and other minority groups as compared to white women. Lack of education, poor nutritional status, smoking, and rural residence have further been associated with adverse maternal and perinatal outcomes.^1,8,9 However, the impact of medical insurance as a factor in obstetrical and maternal outcomes has not been reported with uniformity. Given the variation in health insurance type among parturient women, it is important to consider insurance coverage type as another means of health disparity.²

MATERIALS AND METHODS

After local institutional review board (IRB) approval, the West Virginia University maternal quality care assurance population database was queried comparing commercial insurance to Medicaid/Medicare insurance from July 1, 2015 through June 30, 2018 (n = 4 199). Queried database variables included the deidentified patient database number, age, gestation, admit source, admission time, discharge time, days since last urine result, live births, gravida, parity, indications for induction, height, weight, body mass index (BMI), race, ethnicity, insurance payer, smoking status, any positive urine drug screen, augmentation, induction, rupture to delivery time, antibiotics given during labor, anesthesia type (epidural, general, spinal, local, none), labor and delivery complications, delivery time, delivery method, indication for caesarean delivery, baby status (living or demise), Apgar scores at 1 minutes, 5 minutes, and 10 minutes, cesarean delivery category (primary or repeat), Cesarean priority (scheduled, nonscheduled, emergency/stat), delivery procedures, forceps attempted, vacuum attempted, estimated blood loss, blood transfusion, postdural puncture headache, and length of stay.

STATISTICAL ANALYSIS

Exploratory analysis included examining the associations between the two-group coded insurance variables against delivery outcomes. Bivariate chi-square tests were run for categorical outcome data, Fisher’s exact tests for small cell sizes, and odds ratios are reported. Independent sample t-tests or nonparametric equivalents were conducted on continuous outcome data, with means and standard deviations or medians interquartile range (IQR) reported. Multiple logistic regression, controlling for significant demographic characteristics, was conducted for significant outcomes. SAS 9.4 (Cary, NC.) was used for all analysis with an α set to .05. This study was powered on preliminary electronic medical record (EMR) data (N = 2 817), based on the general anesthesia (any type) outcome. For that sample, general anesthesia was conducted in 3.37% across the sample, with a probable increase to 5% for public insurance, requiring a sample of N = 4 171 with 0.95 power, α = .05.

RESULTS

Parturients with commercial insurance (n = 3 352) were older (mean years: 29.1 vs 26.1, P < .0001), weighed more (median BMI: 31.7 vs 30.8, P = .004), presented with less gravidity (median 2 vs 3, P < .0001) and parity (median 2 vs 2, P < .0001), had more advanced gestation (median 38.9 weeks vs 38.0 weeks, P < .0001), and had a neonate with a higher 5-minute Apgar score (median 9 vs 9, P = .009) than government insured parturients (n = 847). Additionally, government insured parturients were less likely to be admitted for induction with oxytocin (odd ratio [OR]: 0.84, P = .047), receive labor epidural analgesia (OR: 0.79, P = .003), have a delivery supplemented with local anesthesia (OR: 0.53, P = .04), and have a primary cesarean delivery (OR: 0.66, P = .002). Government insured parturients were more likely to be of African American descent (OR: 1.75, P = .008), be a current known smoker (OR: 5.37, P < .0001), have a positive urine drug screen (OR: 3.39, P < .0001), receive a general anesthetic (OR: 1.97, P < .0001), and receive a sterilization procedure (OR: 1.67, P < .0001) (Table 1, Figure 1). After adjusting for gestational age, maternal age, smoking status, and positive urine drug screen (N = 4 199), multivariate logistic regression analysis with 95% Wald confidence intervals revealed epidural anesthesia (OR: 0.77, 95% CI: 0.62, 0.96), sterilization procedure (OR: 1.79, 95% CI: 1.26, 2.55), and primary cesarean delivery (OR: 1.55, 95% CI: 1.05, 2.18) remained significantly related to government insurance factors.

Table 1:

Comparisons between and private and government payor insurance parturients (N = 4 199)

		Private (n = 3 352)	Government (n = 847)	P value
Demographic dataa
Age (yr)		29.1 ± 5.5	26.1 ± 5.9	<.0001
BMI (kg/m2)		31.7 ± 8.3	30.8 ± 8.3	.004
Gravida		2 [1–3]	3 [2–4]	<.0001
Parity		1 [0–1]	1 [0–2]	<.0001
Gestation (wk)		38.9 ± 3.5	38 ± 4.1	<.001
Caucasian/White (%)		91.1%	91.3%	.91
African American (%)		2.3%	3.9%	.008
Current smoker (%)		12.2%	42.9%	<.0001
Urine drug screen + (%)		8.8%	27.4%	<.0001
Labor induction
Oxytocin (%)		27.5%	24.1%	.047
Anesthesia
Epidural (%)		56.0%	50.3%	.003
General (%)		3.0%	5.7%	.0001
Local (%)		2.6%	1.4%	.04
Maternal outcome
Vaginal (%)		62.7%	64.8%	.25
Primary C/S (%)		61.3%	51.1%	.002
Tubal (%)		7.0%	11.1%	< .0001
Neonatal outcome	Apgar 1 < 7 (%)	19.0%	20.7%	.26
	Apgar 5 < 9 (%)	26.5%	31.1%	.009

^aMedian and IQR reported for Gravida and Parity, wk = week, yr = year, + = positive.

Figure 1:

Odds of component for having government Insurance relative to those who have private payor

DISCUSSION

The Institute of Medicine (IOM) in 2003 defined health care disparities as racial or ethnic differences in the quality of health care that are due more from bias and prejudice than to access, disease processes, and treatment intervention.^1,10 It was reported that patients who belong to racial and ethnic populations in the United States are less likely to receive needed procedures, more likely to receive less useful procedures, and overall experience a lower quality of health care.^10,11 Since the 2003 IOM report, the medical literature and medically related websites have become replete with information describing social determinants of health.¹² The social determinants of health refer to the economic, social, and cultural factors that influence individual population health both directly and indirectly, through their impact on psychosocial factors and biophysiological responses.¹³ Physicians’ negative explicit and implicit biases are known to adversely affect doctor–patient relationships and are associated with disparate outcomes.¹⁴ Systemic barriers also contribute to health inequity. More recent publications emphasize the role that structural racism plays in the root cause of health care disparities in the United States.¹⁵ Our results illustrate these disparities exist not only along race and ethnicity but extend to health insurance coverage. Whether patients have private or public health insurance coverage is closely tied to their level of education and employment, factors which may be strongly affected by structural racism. Within our study, government insured parturients are less likely to undergo labor induction, less likely to receive labor epidurals for analgesia, and more likely to have a primary cesarean delivery with general anesthesia.

Other studies have demonstrated marked racial and ethnic disparities in maternal care.^1,9,11,16,17 Both Tucker et al.¹⁸ and Rosenberg et al.¹⁹ have shown that black women are more likely to have pregnancy-associated mortality even after accounting for severity of the complication.¹¹ Black and Hispanic women have higher rates of severe maternal morbidity as well diabetes, and obesity.¹¹ Asian and Hispanic women are at greater risk of developing gestational diabetes.¹¹ Furthermore, white women are less likely to experience postpartum hemorrhage, infection, and severe perineal laceration than other racial and ethnic groups.^11,20–23 To our knowledge, this is the first study to demonstrate that maternal and neonatal care differ as a result of medical insurance.

West Virginia (WV) is a mid-Atlantic state and one of the 13 states comprising the Appalachian region. WV is the only state in the United States entirely located in the Appalachia region. WV is considered a rural state, with most of the population of roughly 1.8 million people living in rural areas (fewer than 2 500 people). The population has declined 3.3% over the past decade. Median household income is just under $45 000 per year, with nearly 18% of the population considered to be living in poverty. WV is an elderly state with over 20% of the population Medicare eligible, and 8% of those under 65 years without health insurance. The state is not very diverse based on race and ethnicity with 93.5% identifying as white non-Hispanic (Caucasian). Blacks are the largest non-white group at 3.6% followed by Latino at 1.7%. Education levels for persons age greater than 25 years show 86.5% with at least a high school diploma, but only 20% holding a four year college degree or higher.^24,25

The WV University Health System (ie WVU Medicine) is the largest nonprofit health care system in the state of WV and is the state’s largest private employer. It is affiliated with the only comprehensive academic medical center in the state under WV University encompassing five schools in dentistry, medicine, nursing, pharmacy and public health. Ruby Memorial Hospital is the largest single facility in the system housing nearly 700 of the system’s 1500 bed capacity. It also contains a level 1 trauma center. There are nine additional community hospital members and three critical access hospitals in the system. Clinics are operated across the region in four states (WV, PA, MD, and OH). Management services are provided to an additional six hospitals.

It is unknown how physician attitudes and beliefs about government insured patients in our health care system may have impacted the results that were observed in our study. Limited information exists about physician perception regarding government insured patients. Niess et al.²⁶ did a cross-sectional survey assessing physician attitudes and beliefs regarding government insured patients. In this study, a vast majority of participants perceived government insured patients to be complex, disreputable, disinhibited, and unappreciative.²⁶ These patients were also significantly less likely to have health care access due to less reimbursement from the federal government as compared to private insurance (P = .01).²⁶ Perhaps health risk managers should encourage physicians, and specifically obstetricians/gynecologists to take implicit association tests and self-reflect on their attitudes concerning health insurance status of their patients, as our data indicates that there is an association of disparate obstetric care along insurance status lines.

We found that government insured parturients were more likely to be of African American descent, be a current known smoker, have a positive urine drug screen, are less likely to undergo labor induction, less likely to receive labor epidurals for analgesia, and more likely to have a primary Cesarean delivery with general anesthesia. Obstetrical risk management includes identifying these parturients early in pregnancy and ensuring appropriate care before, during, and after delivery. Before delivery, education on smoking cessation and illicit drug use/abuse needs to be provided with testing (if indicated) and reinforcement at each antenatal appointment. During admission, ensuring appropriate labor induction with oxytocin augmentation, adequate pain control with labor epidural analgesia which can also be converted to provide anesthesia for Cesarean delivery eliminating the increased morbidity and mortality associated with general anesthesia, and immediate neonatal care after delivery including appropriate resuscitation. After delivery, ensure adequate follow-up care with the necessary resources for continuing breast feeding, childcare, and maternal well-being.

STUDY LIMITATIONS AND STRENGTHS

One of our study limitations is the quality assurance database may not have included all observable patient characteristics that could confound the association between race, ethnicity, and maternal outcome. Also, it could be argued that insurance selection bias exists on account of disproportionate numbers of racial and ethnic minorities receiving government insurance. Although this might account for some of the results seen within our study, the patient demographics within the state of WV do not demonstrate substantial diversity with regards to race and ethnicity.²⁴ Furthermore, it is possible that nonmodifiable patient factors could explain our observed results.¹¹ Many obstetric interventions such as episiotomy, vaginal exams, delayed pushing, and operative delivery do not have well-established guidelines with regard to the appropriateness of the procedure unlike other procedures in other specialties such as for cardiac catherization or stroke intervention.^10,11,27,28

CONCLUSION

Maternal health care insurance is another disparity that can factor into the quality of obstetrical care and outcome. The health care team must be acutely aware of the potential for negative bias associated with government insurance and be willing to work toward its elimination. Ideally, insurance coverage should provide similar access to care without differences between patient preferences, need for treatment, and provision. Insurance carriers and clinicians need to commit towards eliminating these disparities by enacting policies to ensure all women have access to and receive comprehensive high quality, high-value maternity care.

ABOUT THE AUTHORS

Manuel C. Vallejo, MD, DMD is an associate dean and designated institutional official for graduate medical education at West Virginia University, and vice chair for faculty development and institutional affairs. He is a professor of medical education, anesthesiology, obstetrics & gynecology. His area of specialty is in Anesthesiology and Obstetric Anesthesiology. Contact Information: PO Box 9001A, Health Science Center North, West Virginia University, Morgantown, WV 26506. Phone: 304-293-0672 Fax: 304-293-5160. vallejom@hsc.wvu.edu Robert E. Shapiro, MD is an associate professor in the West Virginia School of Medicine, department of obstetrics & gynecology. His area of specialty is in obstetrics and gynecology, and urogynecology. Contact Information: PO Box 9186, 1 Medical Center Drive, Morgantown, WV 26506. Phone: 304-293-5631 Fax: 304-293-2131 rshapiro@hsc.wvu.edu Christa L. Lilly, PhD is an associate professor in West Virginia University School of Public Health, department of biostatistics. She enjoys teaching introductory biostatistics and multivariate analysis. Contact Information: PO Box 9190, 64 Medical Center Drive, Morgantown, WV 26506. Phone: 304-293-6515 Fax: 304-293-2700 clilly@hsc.wvu.edu Linda S. Nield, MD is the associate dean of admission for the West Virginia University school of medicine and professor of pediatrics. Her area of specialty is in pediatrics. Contact Information: PO Box 9111, Department of Medical Education, School of Medicine, Morgantown, WV 26506. Phone: 304-598-4835 Fax: 304-293-1216 lnield@hsc.wvu.edu Norman D. Ferrari, MD is the vice dean for medical education, and chair of the department of medical education in the West Virginia University school of medicine. He is a professor of medicine and pediatrics. His area of specialty is in medicine and pediatrics. Contact Information: PO Box 9111, Department of Medical Education, Morgantown, WV 26506. Phone: 304-293-2408 Fax: 304-293-7814 nferrari@hsc.wvu.edu