Limited patient access to care and financial penalties to high-risk providers may be unintended consequences of this payment system.
Dr. DeRoche is Medical Director at Greater Nashville Perinatology, Nashville, Tennessee.
Dr. Guerette is a Research Psychologist at Greater Nashville Perinatology, Nashville, Tennessee.
Many healthcare delivery and payment reform options have been proposed to control rising and unsustainable costs, and are being implemented across the country. One such strategy is the state of Tennessee’s episode-based payment model, which took effect on January 1, 2015.1 It comprises 3 initial episodes of care (EOC)-asthma exacerbation, total joint replacement, and pregnancy-and a plan is in place for development and implementation of more than 70 additional episodes during the next 5 years. Several other states are currently considering a perinatal EOC model similar to the Tennessee approach.2,3
The perinatal EOC is a payment system intended to control costs while focusing on patient-centered, high-value healthcare for pregnant women. The state identified high healthcare costs for infants in the neonatal intensive care unit (NICU) as a significant health and financial issue and determined that the best way to reduce NICU costs was to improve maternal care. The state intends to accomplish this through the perinatal EOC model. The perinatal model will initially affect the deliveries of approximately 80,000 pregnant women covered by TennCare (Tennessee's Medicaid program) and certain Blue Cross/Blue Shield commercial plans. It is designed to control maternal costs by rewarding lower-cost providers and penalizing higher-cost providers. The perinatal model does not consider NICU admission rates or neonatal costs associated with an obstetrician’s deliveries, and it is not known whether controlling maternal costs will result in improved maternal care or lower NICU utilization.
Under Tennessee's perinatal model, an episode of care begins 40 weeks prior to admission for delivery and ends 60 days after discharge. The physician or midwife who delivers an infant is designated the principal accountable provider (PAP), and is responsible for all costs incurred during the episode, regardless of whether he or she provided some or all of the antenatal care, or was only present to deliver an infant in an obstetric emergency. Costs include fees paid to the PAP, medication costs, laboratory fees (including prenatal testing), emergency department visits, hospital inpatient and delivery fees, other professional fees, and anesthesia costs. The patient has no responsibility for any costs incurred. At the end of an episode, the costs are totaled and adjusted using a risk weight based on the woman’s age, health conditions, and complications during pregnancy.
Legally Speaking: Ob/gyn neglects to read test results; infant is exposed to HSV
At the end of the year, a PAP's average adjusted cost for all pregnancies delivered is determined and compared to “commendable” and “acceptable” levels established by each payer. If a PAP’s average adjusted cost is less than the 25th percentile and he or she has met state-designated quality metrics, the PAP will receive a financial rebate from the payer/state (gain sharing). If the provider’s average adjusted cost is between the 25th and 75th percentile, there is no rebate or penalty. If a PAP's average adjusted cost is between the 75th and 99.73rd percentile, that provider will be required to reimburse the payer/state one half of the difference between the 75th percentile cutoff and their average cost, multiplied by the number of pregnancies delivered (risk sharing). Pregnancies with a cost greater than the 99.73rd percentile after adjustments (ie, greater than 3 standard deviations [SDs] above the mean) will be excluded from a PAP’s annual average adjusted cost of pregnancy. Pregnancies complicated by HIV/AIDS, actively treated cancer, multiple sclerosis, and some blood clotting disorders such as hemophilia are excluded from the perinatal EOC. There are no provisions for tort reform included in the Tennessee model.
The perinatal episode model was developed for women with low-to-medium risk pregnancies.4 In an effort to equitably compare average costs for low-risk and high-risk providers, the model uses risk weights to determine an “adjusted” cost for each pregnancy. Under the Tennessee EOC model, each pregnancy has a baseline risk weight derived by payers based on the woman's age. For a woman younger than age 35, this initial risk weight is approximately 0.633 (average of weights from the 3 payers).4-6 Pre-existing medical conditions such as diabetes or hypertension incrementally increase the baseline risk weight, and additional complications during pregnancy or within 60 days after discharge (eg, preeclampsia, postpartum hemorrhage) are added to the risk weight. At the end of an episode, all costs are summed to determine the actual cost of pregnancy, which is then modified using the patient’s risk weight to determine the adjusted cost of pregnancy (Actual Cost/Risk Weight = Adjusted Cost). Because the actual cost of pregnancy is divided by the risk weight, only risk weights greater than 1 reduce the adjusted cost of pregnancy.
For example, the average baseline risk weight for a 26-year-old woman is 0.633. If this patient has an uncomplicated pregnancy and a term vaginal delivery, her risk weight will remain at 0.633. If the actual cost of her pregnancy is $7000, her adjusted cost of pregnancy will be $7000/0.633 = $11,058, which will be applied to her PAP’s year-end average. Each payer has developed its own risk weight tables and its own “acceptable” and “commendable” levels of expected cost of pregnancy.4-6
This EOC model may be suitable for the general obstetrician whose patient population consists of a high volume of lower-risk pregnancies, but applying it to maternal-fetal medicine (MFM) specialists who manage a smaller volume of predominantly or exclusively high-risk pregnancies is problematic.
First, the perinatal episode is fundamentally different from the other episodes being developed in Tennessee, and these differences will likely lead to unintended consequences and abuses in application of the model.
In other EOCs, either the PAP is the first provider to care for the patient or the episode begins once a patient is referred to a specific physician for a problem. In the perinatal EOC, the PAP is the final provider who cares for a pregnant woman (ie, the delivering provider). The PAP may not have provided care for a woman until delivery, yet is responsible for all costs accrued during her pregnancy and for 60 days after discharge. Costs begin accruing 40 weeks prior to admission for delivery, but the PAP is not able to affect costs until the woman presents to him or her for care, yet is responsible for all pregnancy-related expenses prior to that initial presentation, including follow-up infertility costs after in vitro fertilization procedures, emergency department visits, and laboratory fees. There are many ways in which abuse of the model allows one provider to shift costs of an expensive pregnancy to another, or could result in limited access to care for the most at-risk, costly pregnant women.7
A second concern relates to the basic assumptions of the perinatal EOC. The model assumes that a provider will deliver a majority of low-risk, lower-cost pregnancies and will be able to average down a few, more costly, high-risk pregnancies. This economy of scale should allow a cost-conscious, high-volume general obstetrician or midwife to maintain an adjusted yearly cost under the 75th percentile. However, the model systematically biases against MFM specialists who do not typically provide care to less-costly, lower-risk pregnant women and may not be able to average down a majority of more expensive, high-risk pregnancies, despite the risk adjustments.8 For transport or delivery services at tertiary hospitals where patients are often transported immediately prior to delivery, the PAP may receive only a delivery fee, yet will be responsible for high antenatal costs accrued prior to transport. In delivery services where this is common, this could result in a PAP having to reimburse more than he or she received in delivery fees to care for and deliver late-referral, high-risk pregnancies. This is not a financially sustainable model and may ultimately limit access to MFM specialists for management and delivery.
Third, although the use of risk weight adjustments is intended to equitably adjust costs to compensate a PAP for managing and delivering a high-risk pregnancy, the risk weights provided by payers may not appropriately equalize costs.8 The weights assigned to medical conditions commonly seen in high-risk patients do not adequately adjust for the increased time/effort/costs commonly associated with proper management of high-risk pregnant women. For example, 2 of the payers do not provide a risk weight adjustment for any multifetal pregnancy despite the increased risk of complications; the other payer has a risk weight adjustment for multifetal pregnancies but does not distinguish between types of twinning (dichorionic/diamniotic, monochorionic/diamniotic), triplets or quadruplets.9-12 In a second example, one payer has a significant difference in risk weight adjustments between Type I and Type II diabetes (0.653 vs 0.042), with the risk adjustment for Type II being the fourth-lowest among all complications listed (only gastritis, migraines, and anxiety disorder/phobia are lower).5 While it is acknowledged that pregnancy risks to a woman with Type I diabetes are increased, management of Type II diabetic patients can be just as complex and time-consuming, yet the additional costs are not adjusted by the risk weight. The other 2 payers use the same risk weights for both types of diabetes but their weights differ considerably from the first payer and from each other (0.30 vs 0.736).4-6
Fourth, there are no risk weight adjustments for complications of a prior pregnancy. When the perinatal EOC was developed, data regarding prior pregnancy complications were not considered, therefore, no risk weight adjustments were developed for prior preeclampsia, gestational diabetes, intrauterine growth restriction (IUGR), shortened cervix, preterm labor and delivery, or intrauterine fetal demise (IUFD). For example, a 25-year-old with a previous pregnancy complicated by a shortened cervix, preterm labor, and IUGR, with a cesarean delivery at 34 weeks secondary to preeclampsia, will have the same risk weight in a future pregnancy as a 25-year-old woman with a prior uncomplicated term vaginal delivery. Because the baseline risk weight for a 25-year-old woman is less than 1, the cost of additional monitoring necessitated by prior pregnancy complications will be amplified further by the lack of risk weight adjustment. This will significantly affect MFM specialists, who typically manage women with prior pregnancy complications.
A fifth important concern with the risk weight adjustments is that because a very costly (>3 SD) pregnancy is excluded from a provider's annual average only after adjustment, in some cases, the risk weights will actually work contrary to the way in which they are intended by adjusting extremely costly pregnancies down into the penalty zone (75th to 99.73rd percentile). To illustrate, an insurance payer presents an example of a complicated patient with a risk weight greater than 1.5 The payer demonstrates how a costly pregnancy is adjusted down in fairness to the PAP for caring for this complicated patient. The patient is a 41-year-old woman with Type I diabetes, class III obesity, and a conduction disorder (arrhythmia) who had vaginal bleeding during the pregnancy and a fetus in a breech presentation. Her risk weight calculated to 1.742. In this example, the payer assumed an actual total cost of pregnancy of $7000, resulting in an adjusted cost of $7000/1.742 or $4018. This risk weight adjustment appears favorable to the PAP. However, from a clinical perspective, it is likely that even with appropriate, evidence-based care, this women would have had a much more expensive pregnancy. She likely might have had one or more hospital evaluations, depending on the severity of her bleeding, as well as regular evidence-based monitoring for her diabetes and cardiac arrhythmia (eg, regular blood sugar log reviews, fetal growth ultrasounds, fetal echocardiogram, nutritional assessments, cardiology consultation, maternal echocardiogram, etc.). It is also overwhelmingly likely that she would have had a cesarean delivery secondary to the breech presentation, diabetes, and class III obesity. This care would add significantly to her actual cost of pregnancy, perhaps driving the costs to greater than 3 SDs above the mean, which should result in exclusion of this pregnancy from the PAP's annual average. However, because exclusions occur only after risk adjustment, the adjustment has the potential to move this costly pregnancy from possible exclusion down into the 75th to 99.73rd percentile, penalizing rather than rewarding the PAP for managing a complicated pregnancy.
A sixth concern with the model is that many costs are outside the PAP's control, occur before the PAP first sees the patient for antenatal care, or result from duplication of costs to ensure that quality metrics are met. The majority of pregnancy costs derive from hospital expenses at the time of delivery and from separate anesthesia costs (which can add 17%).8 The PAP cannot affect these costs, yet assumes responsibility for them. There is no incentive or penalty for a hospital or anesthesiologist to decrease costs under this model.
One payer suggested that obstetricians could control delivery costs by delivering at a different hospital with lower costs.13 This is not possible in many towns with a single hospital or when insurance restrictions, physician privileges, physician coverage of multiple hospitals, or level of neonatal care limit location of delivery. In addition, because the patient has no responsibility for any cost incurred during pregnancy on state-sponsored insurance plans, the PAP will have difficulty controlling patients’ independent access to additional care (eg, emergency department visits). Also, for many MFM specialists who receive patients in transport, there is often no ability to confirm that quality metrics mandated by the state have been met. Making the delivering physician responsible for these quality measures may result in unnecessary duplication of laboratory procedures and costs in order to meet the quality metrics mandated by the reform initiative.
A final concern is that the EOC model financially discourages practitioners from providing medically and ethically appropriate but costly, evidence-based care to high-risk pregnant women. Obstetrics is unique in that it is the only field of medicine in which there are at least 2 patients to manage simultaneously (mother and fetus). However, in Tennessee's perinatal EOC, only maternal costs are considered in determining whether cost-effective care was provided; neonatal costs and outcomes are not currently considered and a neonatal EOC will not be implemented for 5 years. Medical management decisions that lower maternal pregnancy costs have a real potential to increase neonatal costs and lifetime medical costs of the child. Management decisions that directly affect a provider’s personal financial outcome will often be at odds with the ethically appropriate choice of care for a woman and her child. For example, high-risk practices frequently manage patients with 28-week preterm premature rupture of membranes (PPROM). Appropriate management options include attempting to prolong the pregnancy to 34 weeks or delivering after 2 days of antenatal steroids, magnesium sulfate, and antibiotics. The maternal costs of managing PPROM in a hospital from 28 weeks until delivery at 34 weeks are clearly greater than the maternal costs of delivery at 28 weeks. These increased pregnancy costs will be assessed to the PAP, presenting a financial disincentive for the PAP to prolong the pregnancy. Delivery at 28 weeks may result in a significant increase in neonatal costs or lifetime costs of a severely premature infant, but these will not be the responsibility of the PAP.14
Similar arguments can be made for early preterm labor, twin-to-twin transfusion syndrome, and other serious conditions often seen by MFM specialists. There is no reward for an obstetric provider who has consistently low neonatal costs. Likewise, there is no penalty for a maternal provider who may receive gain-sharing for low-cost maternal care, even if the infants of this provider's patients have higher NICU admission rates and neonatal costs. Because maternal and neonatal costs are not linked, it will be very difficult to determine whether lower-cost maternal management techniques will result in improved quality and decreased NICU costs, a goal of the perinatal EOC. Other examples of financial disincentives exist. Although there are small risk weight adjustments for other services such as lactation consultation and family planning services, these adjustments may not offset the increased costs. Under the current EOC model, providing family planning services, placing an intrauterine device, or performing a tubal ligation within 60 days following discharge results in additional costs to the PAP. However, delaying these services until after the 60-day window could result in an unintended pregnancy, which would impose additional costs to the state.
As the EOC model drives down the average cost of pregnancy, the level set as “commendable” will progressively decrease and the money available for gain sharing will be reduced. As obstetric providers realize the personal financial costs of managing high-risk patients, practice patterns will likely change, resulting in decreased access to appropriate care for high-risk pregnant women. Table 1 lists anticipated changes in management and referral patterns as physicians attempt to avoid delivering high-risk pregnancies and assuming the high healthcare costs associated with these.
There are solutions that could improve the perinatal EOC model. The most important is to change the fundamental difference between the perinatal EOC and all other EOCs. A more sustainable model that promotes a greater continuity of care is one in which the PAP is the physician or nurse midwife who provides the majority of antenatal care; transport for delivery at a tertiary center does not change who was responsible for the majority of a woman’s care and costs. The Geisinger perinatal model, implemented in Pennsylvania, excludes all deliveries in which there was not at least 12 weeks of continuous antenatal care; late referrals of high-risk patients were also excluded from the model.15 Developers of perinatal EOC models should specify a minimum timeframe for continuous antenatal care that must be provided in order to be designated as a PAP. Like other EOCs, the perinatal EOC should begin with the initial visit to the provider, rather than 40 weeks prior to admission for delivery, thus allowing the PAP more control over total pregnancy costs.
Also deserving reconsideration is the current inclusion of the cost of elective epidurals, which makes up a disproportionate amount of the total cost of pregnancy. Exclusion of or adjustment for an elective epidural charge readily reimbursed by the payer should be considered; delivery-indicated anesthesia charges are valid inclusions in the EOC. In addition, a pregnancy with an actual cost greater than 3 SDs above the mean should be excluded before risk adjustment, rather than penalizing a PAP when a risk weight adjusts the costs down into the “penalty zone.” Further, risk weight adjustments should be developed for prior pregnancy complications such as preeclampsia, preterm labor, gestational diabetes, IUGR, and IUFD, which have an increased likelihood of recurrence in a subsequent pregnancy and necessitate closer surveillance and greater intervention. Another modification deserving consideration is establishment of a minimum number of annual deliveries performed by a provider in order to be included in the model. This would ensure that delivery of a small number of complicated, costly pregnancies would not significantly penalize a provider. Finally, differences in obstetric management options should be analyzed to assess maternal and neonatal costs and outcomes to determine which obstetric management options optimize outcomes and decrease NICU costs. This is particularly relevant for determining the most appropriate management options for high-risk pregnancy complications.
The perinatal EOC model, as currently designed, leads to issues that may limit access to care. The assignment of a PAP at the end of an episode creates the potential for financial abuse,7 while the model itself creates a financial incentive for providers to “cherry pick” low-risk pregnancies and avoid or shift the cost of high-risk pregnancies.8 With modification, this model has the potential to work for providers who deliver a high volume of low-to-medium-risk pregnancies with an occasional complicated pregnancy. However, for MFM specialists who see a lower volume of high-risk patients, this model may be untenable. This payment reform model is now in place in varying stages in Arkansas, Ohio, and Tennessee. Before it is widely accepted by other states in its current form, physicians must be actively involved in seeking modifications to ensure that women with high-risk pregnancies have access to care from those trained to manage their complicated issues.
1. Tennessee Health Care Innovation Initiative. http://www.tn.gov/HCFA/strategic.shtml Accessed November 1, 2014.
2. Arkansas Center for Health Improvement. http://www.achi.net/Pages/OurWork/Project.aspx?ID=47 and Health care payment improvement initiative: Building a healthier future for all Arkansans. http://www.paymentinitiative.org/episodesOfCare/Pages/default.aspx. Accessed March 11, 2015.
3. Governor’s Office of Health Transformation. Ohio episode-based payment overview: Summary document. http://www.healthtransformation.ohio.gov/CurrentInitiatives/ImplementEpisodeBasedPayments.aspx . Accessed January 20, 2015.
4. Blue Cross/Blue Shield of Tennessee. https://www.bcbst.com Accessed January 6, 2015.
5. Amerigroup. TN bundled payment initiative: overview of episode risk adjustment; summary of episode risk methodology and results. https://providers.amerigroup.com/ProviderDocuments/TNTN_BundledPaymentInitiative.pdf. Accessed December 15, 2014.
6. United Healthcare Community Plan (formerly Americhoice). Unpublished risk weight tables. Provided upon request. 2014.
7. DeRoche M. Tennessee payment reform initiative: 27 Ways to circumvent the obstetric episode of care. Document prepared for Blue Cross/Blue Shield of TN and TennCare (August 2014; revised to “30 Ways” March 2015).
8. Sutherland S, Farrell RM, Lockwood, C. Genetic screening and testing in an episode-based payment model. Obstet Gynecol. 2014;124:987–991.
9. Society for Maternal-Fetal Medicine (SMFM); Simpson LL. Twin-twin transfusion syndrome. Am J Obstet Gynecol. 2013;208(1):3–18.
10. Bahtiyar MO, Emery SP, Dashe JS, et al. The North American Fetal Therapy Network consensus statement: Prenatal surveillance of uncomplicated monochorionic gestations. Obstet Gynecol. 125,118–123.
11. ACOG (2004, reaffirmed 2009). Multiple gestation: Complicated twin, triplet, and higher-order multifetal pregnancy. (Practice Bulletin No. 56). Obstet Gynecol. 106,869–883.
12. Sullivan AE, Hopkins PN, Weng HY, et al. Delivery of monochorionic twins in the absence of complications: analysis of neonatal outcomes and costs. Am J Obstet Gynecol. 2012;206,257.e1-7.
13. Open Provider Advisory Meeting, sponsored by United Healthcare Community Plan (presided by Chief Medical Officers from United Healthcare/Americhoice, Amerigroup, and Blue Cross/Blue Shield of Tennessee). Belmont University, Nashville, TN, June 10, 2014.
14. Gilbert WM, Nesbitt TS, Danielsen B. The cost of prematurity: quantification by gestational age and birth weight. Obstet Gynecol. 2003;102:488-492.
15. Berry SA, Laam LA, Wary AA, et al. ProvenCare perinatal: A model for delivering evidence/guideline-based care for perinatal populations. Jt Comm J Qual Patient Saf. 2011;37(5):229–239.