Abstract
Objectives:
To identify the prevalence of comorbidities in pregnant women and examine the incremental costs of these conditions on the care for mothers and their newborns.
Methods:
This was a retrospective comparative cohort study of women ages 15–49 years with a documented live-birth delivery using de-identified claims from the MarketScan Research Commercial Claims and Encounters database incurred between January 1, 2007, and December 31, 2011. Total health care costs from date of first pregnancy-related claim through 3 months postdelivery were reported; pregnancy-related comorbidities prior to the pregnancy diagnosis were identified and categorized in the 12 months prior to the pregnancy diagnosis, and costs associated with each condition were compiled. A subset of newborns was matched to their mothers using a unique family identifier and their costs were captured for the three months following birth. Comparisons of costs for both mothers and newborns were made using both unadjusted and multivariate analyses between mothers with and without each condition.
Results:
A total of 322,141 women with live births were identified; 135,572 of these mothers were linked to their newborn(s). Prevalent conditions included back disorders (8.9%), mental disorders (6.5%), headache (5.5%), allergic rhinitis (5.5%), and osteoarthritis (4.8%). Diabetes (0.97%) and hypertension (1.9%) were associated with the highest adjusted incremental costs of care in both mothers ($6,211 [95% confidence interval 5,720–6,702] and $3,367 [95% CI 2,935–3,799] respectively) and newborns ($2,067 [95% CI 1,515–2618]; and $1,210 [95% CI 725–1,695] respectively). The two most common conditions, back disorders and mental disorders, were associated with unadjusted costs of $1,895/$978 (mothers/infants) and $2,097/$1,902 (mothers/infants) respectively.
Conclusion:
Preexisting conditions common in pregnant women may result in additional resource utilization and costs for both mothers and newborns.
Introduction
C
Current trends suggest that women are giving birth for the first time or having subsequent children at an older age, increasing the odds that particular chronic conditions may be present prior to childbirth. 5 A number of preexisting maternal medical conditions may add to the high cost of prenatal care, delivery, and postdischarge care for both the mother and newborn. Comorbidities, such as diabetes and hypertension, are not uncommon in pregnant women, and the prevalence of many comorbid conditions is on the rise. 6 According to hospital discharge data, between 1992 and 2006 the average number of comorbidities increased from 0.43 to 0.58 for vaginal delivery patients, 1.34 to 1.59 for Caesarean delivery patients, and 0.65 to 0.93 for pregnant women overall. 6 The presence of comorbid conditions during pregnancy often requires additional health care visits, and certain comorbidities, including obesity, type 2 diabetes, hypertension, osteoarthritis, and malignancies, increase the risk of neonatal death. 7 –11 These conditions can also increase the average cost of prenatal care and lead to significantly more Caesarean deliveries. 12,13
The cost burden of comorbidities during pregnancy and childbirth has not been well documented. In this study, we report the prevalence of comorbidities in pregnant women and compare the health care utilization and costs of pregnancy and newborn care based on the presence or absence of particular maternal comorbidities. Such an analysis will determine the incremental costs of pregnancy and newborn care complicated by comorbid conditions and provides payers and policymakers insight into the current total cost of childbirth. Results will also assist health care providers as they seek to efficiently manage certain medical conditions in both pregnant women as well as during and following delivery.
Methods
Study design and population
The study incorporated a retrospective comparative cohort study design using de-identified medical and pharmacy claims from the Truven Health MarketScan Commercial Claims and Encounters database incurred between January 1, 2007, and December 31, 2011. The database contains the inpatient, outpatient, and outpatient prescription drug experience of approximately 46 million employees and their dependents, covered under a variety of fee-for-service and managed care health plans, including exclusive provider organizations, preferred provider organizations, point of service plans, indemnity plans, and health maintenance organizations per year. The database provides detailed cost, use, and outcomes data for health care services performed in both inpatient and outpatient settings. The medical claims are linked to outpatient prescription drug claims and person-level enrollment data through the use of unique enrollee identifiers. As the MarketScan database is de-identified in compliance with the Health Insurance Portability and Accountability Act regulations, Institutional Review Board approval was not required.
The main study cohort consisted of women aged 15–49 years who had at least two outpatient or one inpatient pregnancy-related claims between January 1, 2008, and September 30, 2011, and had at least one claim related to live-birth delivery within 42 weeks following the initial pregnancy claim. The date of the first pregnancy-related claim was defined as the first index date; the date of delivery defined the second index date. Cohort inclusion required continuous medical and drug benefit coverage for at least 12 months prior to the first index date and 3 months following the second index date and with the absence of a claim indicating termination of pregnancy between the first and second index dates (Appendix 1). Only the first pregnancy in the study period was included for each woman to avoid bias in the comparative statistical tests (between pregnant women with and without comorbidities) because of within-person correlation.
A subgroup of women with live-birth deliveries that could be linked to newborns in the MarketScan database was extracted for analysis of the costs of newborn care. Newborns linked to mothers using a unique family identifier in the database were included if age was less than 1 year in the reported year of delivery and the newborn had continuous medical and drug benefit coverage for at least 90 days following birth. Cost and utilization outcomes were compared for newborns of mothers with and without evidence of a comorbid condition. For women with multiple gestations (pregnancy with two or more fetuses), each child was included as a single observation in the analyses of newborn costs. Demographic characteristics of the linked sample were examined and compared with the maternal sample to explore the possibility of selection bias.
Outcomes
Total health care costs of mothers in the main study cohort and health care costs of newborns in the subgroup of linked mothers and newborns were evaluated. Mothers' health care costs were defined as total expenditures on the mothers' health care insurance claims (including primary and secondary insurers and patient out-of-pocket payments) from the first index date through the 3 months following the second index date. Health care costs of newborns were defined as total expenditures on the health care claims of the newborns in the 3-month period following birth. For each time period average costs were reported overall and for individual components of care including inpatient care, emergency department visits, outpatient office and clinic visits, other outpatient services (such as lab and imaging), and prescription drugs. All costs were adjusted to 2011 dollars based on the medical care component of the Consumer Price Index.
The frequency of comorbid conditions among pregnant women was also assessed. Within the cohort of pregnant women, 25 medical conditions observed in the 12 months prior to the first index date were examined (i.e., prior to first pregnancy related claim). Twenty of these conditions were classified based on medical eligibility criteria for contraceptive use by the Centers for Disease Control and Prevention, 14 which identifies conditions that are associated with increased risk for adverse health events as a result of an unintended pregnancy; other conditions were selected based on frequency in the period prior to the first pregnancy claim using the Agency for Healthcare Research and Quality (AHRQ) Clinical Classification Software. 15 Additional conditions observed prior to the first index date were also considered for inclusion based on frequency among the cohorts of interest. One claim with a diagnosis in the primary position was required in order to classify a patient as having a given comorbidity category (codes are listed in Appendix A).
Statistical analysis
Descriptive analyses were used to compare baseline characteristics (as of the first index date) of pregnant women with comorbidities relative to pregnant women without comorbidities. Unadjusted differences in average costs of care were reported between these cohorts of pregnant women and their newborn(s) based on the presence or absence of each maternal comorbidity.
Two multivariable generalized linear models using log transformations of the expenditure data were estimated to further examine the difference in health care costs between pregnant women with and without specific comorbidities and between newborns of mothers with and without specific comorbidities. One model estimated the total health care costs of the mothers and the other model estimated the total newborn costs. The covariates in each model included the eight most prevalent comorbidities, (occurring in at least 0.5% of the pregnant population) plus a combined variable indicating the presence of any of the other comorbidities. Additional covariates were maternal age, year of pregnancy, insurance plan type, and geographic region. After the initial model was fit, a residual analysis was performed to assess the model's appropriateness. If a cost value produced a corresponding residual that fell outside of 6.5 sigma limits (i.e., outside of the 3.25 standard deviations below or above the mean on the residual scale), the observation was removed. The model was then re-fit with values whose residuals satisfied the 6.5 sigma limits. All analyses were conducted using SAS Version 9.2.
Results
Subject characteristics
A total of 322,141 women with live births and meeting the inclusion criteria were identified. Of these, 135,572 mothers could be linked to their newborn(s) using a unique family identifier; a total of 137,040 newborns (due to multiple gestations per mother) were analyzed. Study subjects tended to reside in the southern region of the United States and in mostly urban settings. Women with live births (total) were generally similar to the mother–newborn linked subset for most baseline characteristics (Table 1).
Because the samples are not independent (i.e., the number of women with linked births is a subset of the group of women with live births) evaluating the two groups formally using statistical tests is not feasible.
SD, standard deviation.
Comorbidities in pregnant women
Overall, the presence of selected comorbidities was rare with most occurring in less than 1% of mothers. Only five conditions were observed in approximately 5% or greater of the population: back disorders (8.9%), mental disorders (6.5%), headache (5.5%), allergic rhinitis (5.5%), and osteoarthritis (4.8%). These conditions were not among those defined using Centers for Disease Control and Prevention and AHRQ definitions but were selected for inclusion in this analysis based on their frequency within the population. Conditions prevalent in the general population, such as hypertension and diabetes, were also present in the study population: 1.9% and 1.0%, respectively. Breast and ovarian cancers were also relatively rare (<0.1%).
Unadjusted cost comparisons
Maternal care
Regardless of the condition, the presence of a comorbidity in pregnant women was associated with a significant increase (ranging from $827 for women with allergic rhinitis to $42,266 for women with history of solid organ transplant) in the total costs of care as compared with women without the comorbidity of interest (Table 2). For nearly all conditions, costs were higher for the pregnancy and delivery/postpartum periods with the exception of allergic rhinitis and tuberculosis where the absence of these conditions was observed to be more costly following live birth. With few exceptions, differences in costs were the result of higher expenditures in the period prior to delivery. Conditions that were associated with the highest difference in total costs were history of organ transplant ($42,266), HIV/AIDS ($19,028), sickle cell disease ($18,423), and lupus ($15,144), mostly due to differences in costs related to inpatient care.
All costs in U.S. dollars.
No subjects had evidence of peripartum cardiomyopathy or schistosomiasis; therefore, these conditions are not represented in the cost analyses.
Frequency of comorbidity in all women with live births.
Includes costs prior to live birth.
Includes delivery costs and those accrued in the three months following live birth.
Average cost with the condition minus the average cost without the condition.
HIV/AIDS, human immunodeficiency virus/acquired immunodeficiency syndrome.
Newborns
The costs of care for newborns in the first 3 months of life were also impacted by the presence of certain maternal comorbidities. Women with lupus ($12,296), history of bariatric surgery ($10,917), and hypertension ($8,262) had the highest costs of newborn care, again mostly inpatient care. Two maternal conditions, epilepsy and ischemic heart disease, both occurring relatively infrequently, were associated with lower newborn costs (Table 3). Cost drivers were related to inpatient treatment of newborns during the first 3 months of life, with differences in average inpatient costs between infants of mothers with and without comorbidity ranging from $607 (epilepsy) to $48,011 (sickle cell disease) (not shown).
N=137,040 (accounts for multiple gestations per mother); all costs in U.S. dollars.
Frequency of comorbidity in mother linked to the newborn.
Indicates average cost with the condition minus the average cost without the condition.
Multivariable cost comparisons
The estimated average costs of care for mothers and newborns without a maternal comorbidity were $10,221 (SE 219) and $1,653 (SE 171), respectively. Compared with unadjusted analyses, nearly all cost differences decreased following multivariate adjustment; the estimate of the incremental cost of care for mothers with allergic rhinitis was the only one that increased following adjustment.
The conditions that were associated with the highest incremental costs of care for both mothers and newborns were diabetes ($6,211 [95% confidence interval 5,720–6,702] and $2,067 [95% CI 1,515–2,618], respectively) and hypertension ($3,367 [95% CI 2,935–3,799] and $1,210 [95% CI 725–1,695], respectively) (Fig. 1).
Incremental costs of pregnancy due to maternal comorbidities (data source: 2007–2011 MarketScan Commercial Claims and Encounters Database).
About 7% of the cases in the study had maternal cost estimates that were extremely small or large relative to the main cohort and were excluded from the multivariate analysis for estimating the total maternal costs. Outliers were trimmed from this model because an adequate fit was not obtained using the entire set of data, the result of which was the lowering of cost estimates from unadjusted values for some conditions. Sensitivity analyses were conducted to determine the effect of trimming on the results. Diabetes, the highest average cost comorbidity investigated in the multivariate analysis, was affected the most by the trimming (raising the limit to $50,000 increased the incremental cost estimate by 61%) and allergic rhinitis, the lowest cost comorbidity, was affected the least (raising the limit to $50,000 increased the incremental cost estimate by 20%). The effect of raising the limit to $50,000 on the other comorbidities ranged from 37% to 45%.
This phenomenon of large residual values was not present in the modeling of the newborn data costs. The newborn data, post modeling, adhered to the requisite underlying residual distribution theory, and did not require an intervention to ensure an adequate fit.
Discussion
Pregnancies associated with preexisting conditions may lead to increased health care resource use and costs for both the mother and their newborns when compared with pregnant women without maternal comorbidities. Through this analysis, we identified the comorbid conditions that are associated with increased resource utilization and costs of care for mothers and their newborns and the magnitude of that increase. In this study, nearly 9% of all pregnant women reported back problems and 5% or more had allergic rhinitis, headache, or some level of mental disorder prior to becoming pregnant. These conditions led to a minimum incremental cost of care of $1,076 (adjusted result) when compared with mothers without these conditions prepregnancy. Higher costs of care were also observed for newborns of mothers with a comorbid condition; however, the incremental costs were significantly lower for these newborns. Such differences suggest that while comorbidities may increase the costs of care for newborns, we can expect the majority of incremental costs during and following pregnancy to be attributed to maternal treatment, particularly in the period prior to delivery.
Hypertension and diabetes—chronic conditions prevalent in the general population—were also observed in our sample of pregnant women. Previously, it was suggested that due to trends in the increasing prevalence of this condition, 3%–5% of pregnant women would also have chronic hypertension. 16,17 Our observed rate of 1.9% is in line with this suggestion for the referenced time period, although our analysis was not structured to provide a population-level estimate. Similarly, the observed diabetes prevalence in this study (1.0%) was comparable to what was seen in a previous database analysis (1.3%) specific to pregnant women diagnosed with diabetes prior to becoming pregnant. 18 Hypertension and diabetes were also associated with the highest incremental costs for mothers and newborns in both groups. While the primary driver of cost differences in women with comorbidities was resource utilization prior to delivery, higher costs for newborn care were related to inpatient treatment in the first 3 months of life. These findings highlight the need to identify and treat these conditions in mothers prior to and during pregnancy to diminish the need for costly resource utilization and to decrease morbidity and mortality in their newborns. 16,19,20 Current and expected trends in prevalence for diabetes and hypertension combined with the associated high costs of treatment have the potential to introduce significant costs to the health care system as a whole.
Importantly, this investigation also highlighted both the prevalence and associated costs of pregnancies for rarer conditions that were present prior to conception. While present in 0.12% or less of this study population, conditions such as sickle cell disease, lupus or cirrhosis and histories of bariatric surgery and solid organ transplant were associated with an increase in average costs of over $10,000 for either the mother or her newborn. Considering the high rate of resource utilization and related morbidity, health care providers should carefully monitor women of childbearing age with these conditions or procedures and offer effective family planning counseling. Future research should also address the specific cost drivers for pregnant women with rare conditions to potentially uncover ways to minimize the financial impact while maintaining, or even improving, maternal and newborn outcomes.
For two maternal conditions, absence of the condition resulted in higher maternal costs following live birth. While we did not test this hypothesis in the current analysis, these results may be because mothers in the “without” cohort had unaccounted comorbid conditions that might have led to increased costs. Two maternal conditions, primarily those with very low frequency (e.g., less than 1%), were associated with lower newborn costs. As with the maternal costs, it is possible that in the “without” cohort the mothers had other unaccounted comorbid conditions which might have led to higher newborn care costs. Additionally, these are newborn care costs and not the costs of the mothers.
This investigation has several limitations. First, as is common in cost analyses, the data are skewed and as such, inferences cannot be based solely on the summary statistics (e.g., mean/standard deviation). To address this, outliers were trimmed in order to produce reasonable estimates of the incremental costs of comorbidities on pregnancy and delivery costs for mothers. Sensitivity analyses showed that the likely effect of the trimming was to underestimate the incremental cost of the comorbidities, but because of the effect of high-cost outliers on the models and on average costs, the magnitude of the underestimation is greater for comorbidities for which the patients have higher average costs. Additionally, the costs provided for several rare conditions may not be reliable estimates due to the rarity of these conditions in the study population. Comparisons were made between cohorts of women with and without select comorbidities, relying on the presence of a claim for these conditions during a defined period. Conditions not specifically targeted for analysis may contribute to the costs of care or resource utilization of these subjects and their newborns.
Second, this analysis relies on medically attended conditions and/or coding of these conditions. While it is possible that women with a comorbidity could be misclassified based on absence of a medical claim for the condition, we believe that this misclassification would be rare, as pregnancy management for high-risk patients requires a higher level of oversight, and reimbursement arrangements are often different for these high-risk patients. The use of administrative claims limits the ability to control for variation in prenatal care between subjects, which may impact the extent to which comorbidities were addressed and treated. While the results indicate that substantial resources were consumed by pregnant women prior to giving live birth, we are unable to account for how closely guideline-recommended care was sought and received.
Third, the analysis was limited to a commercially insured population; therefore, the results are not generalizable to those covered under other types of insurance or who lack coverage. The data used in this study are generally representative of persons with employer-sponsored health insurance nationally. Although there is slight overrepresentation in the South, for example, this difference is not substantive (e.g., the percentage of overall MarketScan Commercial population in the South is 37.1%; U.S. Census in 2010 was 35.6%). Finally, multiple pregnancies were not accounted for in the analysis, however, multiple births were (i.e., each child served as the unit of analysis in the newborn portion of the study). While multiple births may be correlated events, the number of such events relative to single births was small.
Conclusions
These findings support the need for practitioners to identify and properly address chronic conditions that have the potential to impact the management and costs of care for mothers and their newborns. Counseling women of childbearing age with conditions known to complicate pregnancies allows for setting expectations and agreed-upon management between patients and providers should these women seek to become pregnant. Current trends suggest that women are giving birth for the first time or having subsequent children at an older age, increasing the odds that particular chronic conditions may be present prior to childbirth. Considering the impact that preexisting conditions may have on the costs and outcomes of care, regardless of their prevalence, it will become increasingly important to properly address and manage these disorders prior to conception, throughout pregnancy, and continuing through delivery.
Footnotes
Author Disclosure Statement
This research was supported by Bayer Healthcare Pharmaceuticals (“Bayer”), Whippany, New Jersey. Bayer participated in the study design, results interpretation, and manuscript review as reflected in the authorship by Bayer employees, AL, MM, and RL. At the time the work was completed, SMC, JG, PLJ, and PLB were employees of Truven Health Analytics, Inc., which has received research funds from Bayer.
Appendix A. Diagnosis Codes
| Comorbidity | ICD-9 CM code |
|---|---|
| Breast cancer | 174.0x–174.6x, 174.8x, 174.9x, 175.0x, 175.9x, 233.0x, V10.3x |
| Valvular heart disease | 394.0x–394.2x, 394.9x, 395.0x, 395.1x, 395.2x, 395.9x, 396.0x–396.3x, 396.8x, 396.9x, 397.0x, 397.1x, 397.9x, 424.0x–424.3x, 424.9x, 785.2x,785.3x, V42.2x, V43.3x |
| Diabetes | 249.0x–249.9x, 250.0x–250.9x, 790.2x, 791.5x, 791.6x, V45.8x,V53.9x, V65.4x 24900, 25000, 25001, 7902, 79021, 79022, 79029, 7915, 7916, V4585, V5391, V6546 |
| Endometrial/ovarian cancer | 183.0x, V10.4x |
| Epilepsy | 345.0x–345.9x, 780.3x |
| Hypertension | 401.0x, 401.1x, 401.9x, 402.0x, 402.1x, 402.9x, 403.0x, 403.1x, 403.9x, 404.0x, 404.1x, 404.9x, 405.0x, 405.1x, 405.9x, 437.2x |
| Bariatric surgery history * | V45.8X, 43644, 43645, 43770–43775, 43846, 43847, 43848, 43886, 43887, 43888 |
| HIV/AIDS | 042.xx, 079.5x, 279.1x, 795.7x, 795.8x, V08.xx |
| Ischemic heart disease | 410.0x–410.9x, 411.0x, 411.1x, 411.8x, 412.xx, 413.0x, 413.1x, 413.9x, 414.0x, 414.2x, 414.3x, 414.4x, 414.8x, 414.9x, V45.8x |
| Malignant neoplasm of placenta | 181.xx, 183.2x–183.5x, 183.8x, 183.9x, 184.0x–184.4x, 184.8x, 184.9x, 233.3x, 795.1x, V10.4x |
| Malignant liver tumors | 155.0x, 155.1x, 155.2x, 230.8x, V10.07 |
| Peripartum cardiomyopathy † | 674.5x |
| Schistosomiasis † | 120.9x |
| Severe cirrhosis | 571.2x, 571.5x, 571.6x, 571.8x, 571.9x |
| Sickle cell disease | 282.4x, 282.5x, 282.6x |
| Solid organ transplantation history | V42.0x, V42.1x, V42.7x, V43.2x, V43.8x |
| Stroke | 434.0x, 434.1x, 434.9x, 435.0x, 435.1x, 435.2x, 435.x3, 435.8x, 435.9x, V12.54 |
| Systemic lupus erythematosus | 695.4x, 710.0x |
| Thrombogenic mutations | 270.4x, 286.4x, 286.9x, 289.81 |
| Tuberculosis | 010.0x, 010.1x, 010.8x, 010.9x, 011.0x–011.9x, 012.0x–012.3x, 012.8x, 013.0x–013.6x, 013.8x, 013.9x, 014.0x, 014.8x, 015.0x, 015.1x, 015.2x, 015.5x–015.9x, 016.0x–016.7x, 016.9x, 017.0x–017.9x, 018.0x, 018.8x, 018.9x, 137.0x–137.4x, V12.01 |
| Back problems ‡ | 720.1x, 720.2x, 720.8x, 720.9x, 721.0x–721.9x, 722.0x–722.9x, 723.0x–723.9x, 724.0x–724.9x |
| Allergic rhinitis ‡ | 477.0x, 477.2x, 477.8x, 477.9x |
| Mental disorders, except psychoses ‡ | 290.0x–290.4x, 290.8x, 290.9x, 291.0x–291.5x, 291.8x, 291.9x, 292.0x, 292.1x, 292.2x, 292.8x, 292.9x, 293.0x, 293.1x, 293.89, 293.9x, 294.0x, 294.1x, 294.2x, 294.8x, 294.9x, 299.0x, 299.1x, 299.8x, 299.9x, 300.0x–300.3x, 300.5x–300.9x, 301.0x–301.9x, 302.1x–302.9x, 303.0x, 303.9x, 304.0x–304.9x, 305.0x–305.9x, 306.0x–306.9x, 307.1x, 307.4x, 307.5x, 307.7x, 307.8x, 308.0x–308.4x, 308.9x, 309.0x–309.4x, 309.8x, 309.9x, 310.0x, 310.1x, 310.2x, 310.8x, 310.9x, 312.0x–312.4x, 312.8x, 312.9x, 313.0x–313.3x, 313.8x, 313.9x, 314.0x, 314.1x, 314.2x, 314.8x, 314.9x, 315.0x–315.5x, 315.8x, 315.9x, 316.xx, 317.xx, 318.0x, 318.1x, 318.2x, 319.xx, 331.0x, 331.1x, 331.2x, 331.8x, 333.92, 357.5x, 425.5x, 535.3x, 571.0x–571.3x, 648.3x, 648.4x, 655.5x, 760.7x, 779.5x, 790.3x, 797.xx, 965.0x, 980.0x, E95.0x–E95.9x, V11.0x–V11.4x, V11.8x, V11.9x, V15.4x, V15.82, V40.0x–V40.3x, V40.9x, V62.84, V62.85, V65.42, V66.3x, V67.3x, V70.1x, V70.2x, V71.0x, V79.0x–V79.3x, V79.8x, V79.9x |
| Headache, including migraine ‡ | 339.0x–339.4x, 339.8x, 346.0x–346.5x, 346.7x, 346.8x, 346.9x, 784.0x |
| Osteoarthritis ‡ | 713.0x–713.8x, 715.0x–715.3x, 715.8x, 715.9x, 716.0x, 716.2x–716.6x, 716.8x, 716.9x, 718.1x, 718.2x, 718.5x–718.9x, 719.0x–719.9x, V13.4x |
Based on CPT-4 procedure codes except for V45.8x (ICD-9 CM).
No observations of these comorbidities were seen in the study population.
Indicate customized comorbidities used for this investigation.