RECALIBRATING "EXPERIMENTAL TREATMENT EXCLUSION": AN EMPIRICAL ANALYSIS Won Bok Lee* Health insurers usually exclude treatments without proven safety and effectiveness. When a claim is denied on such grounds and the denial is disputed in court, it is generally believed that the insurersface an uphill battle in front of sympathetic judges. Some ,suggest that the insurers' perceived odds in adjudications in turn make them hesitant to deny coverage of experimental treatments in thefirst place, which may not be socially desirable. Through an empirical analysis of all experimental exclusion litigationsince 2003, this article will address thatperception. In addition, through contents analysis, this article will show that the courts have become more sophisticated in judging experimental treatmentexclusion cases than in the past. The results of this analysis have policy implications. As the new federal regulations under the Affordable Care Act require internal appeal and external review process, there will be a likely increase of coverage disputes. If health insurers and policy holders become more aware of the recent trends in court, their preference for settlement may shift. More importantly, the results of this study have bearingon the evidencebased coverage policy in general. If health insurers still consider the court as the bottleneck to adopting evidence-based coverage policye.g., cost-effectiveness-it may be time for reassessment and recalibration. I.Introduction ........................................................................................ II. Experimental Exclusion and Dispute .............................................. III. Recent Trend in Experimental Exclusion Litigation: An Em pirical Analysis ..................................................................... A . Data & M ethods .................................................................... B . R esu lts ................................................................................... C. Reason for Rejecting Experimental Exclusion ..................... D . Study Lim itations .................................................................. E. Recalibrating Experimental Exclusion Litigation ................. IV. Role of Experimental Exclusion in New Technology, Insurance 172 173 178 178 183 188 195 196 * Associate Professor of Law, Ewha Womans University School of Law. I thank Steven Shavell, Louis Kaplow, Aaron Kesselheim, Glen I. Cohen, Christopher Robertson and participants at the Law and Economics Seminar at Harvard Law School for helpful comments. All errors remain mine. This work was supported by the National Research Foundation of Korea Grant funded by the korean Government (NRF-2012S 1A5B8A03034450). UNIVERSITY OF CINCINNATI LAW REVIEW [VOL. 83 Coverage & Health Care Spending Triad .................................. A. New Grievance Process Requirement under the Patient Protection and Affordable Care Act ..................................... B. Big Picture: Experimental Exclusion & Health Care Spending ............................................................................... V . C onclusion ....................................................................................... 196 196 198 20 1 I. INTRODUCTION Health insurer's denial of coverage is plentiful.1 Yet, coverage denial of treatments with unproven safety and efficacy, also known as experimental treatments, 2 has not only created headline-grabbing jury verdicts, 3 but also present important policy implications. If a treatment, whether drug, medical device, or surgical procedure, 4 does not have proven safety and efficacy as defined by the Food, Drug, and Cosmetic Act, why would the treating physician prescribe it or perform it in the first place? A treating physician prescribes experimental treatments when a patient is suffering from a terminal illness and has run out of standard treatment options; experimental treatment may be the only possibility of recovery. In other cases, there may be other treatment options but the physician may believe that emerging technology is promising and better than standard treatments.5 Do health insurers readily pay for experimental treatment? Unfortunately, health insurers usually exclude experimental treatments 1. One survey estimates that the aggregate rate of (partial or full) coverage denial ranges from OFFICE, GAO- 11-268, PRIVATE HEALTH INSURANCE: DATA ON APPLICATION AND COVERAGE DENIALS (2011). 2. Throughout this article, "experimental," "investigational," and "unproven" will be used interchangeably. 3. E.g.,Erik Eckholm, $89 Million Awarded Family Who Sued H.MO., N.Y. TIMES (Dec. 30, 1993), http://www.nytimes.com/1993/12/30/us/89-million-awarded-family-who-sued-hmo.html. 4. Under the Food, Drug and Cosmetic Act, drugs and certain medical devices cannot be marketed until they have proven safety and efficacy. That does not foreclose physicians from prescribing FDA-approved drugs and devices for indications that were not approved by the FDA. Prescription of a drug or medical device for a non-FDA approved indication is called "off-label" use. "Off-label" treatments are often treated as experimental by health insurers, as they lack safety and efficacy data. See infra text accompanying note 68. 5. Sometimes physicians based their belief on uncontrolled, observational reports. Without a control group, observational reports of only treated patients can fail to assess the true benefit of the treatment masked by 'placebo effect'. A. G. Johnson, Surgery as a Placebo, 344 LANCET 1140 (1994). The long list of treatments that were accepted based on uncontrolled, observational studies later to be found to be ineffective or even harmful include liberation therapy for chronic cerebrospinal venous insufficiency in multiple sclerosis, vertebroplasty for osteoporotic vertebral fracture, arthroscopic surgery for arthritis, high dose chemotherapy/autologous bone marrow transplant for breast cancer, radical mastectomy for breast cancer, extracranial/intracranial arterial bypass for atherosclerosis, jejunoileal (intestinal) bypass for weight reduction, gastric freezing for peptic ulcer, internal mammary artery ligation for angina, colectomy for epilepsy and nephropexy for "floating-kidney." 11% to 24% of all claims. UNITED STATES GOV'T ACCOUNTABILITY 2014] RECALIBRATING EXPERIMENTAL TREATMENT EXCLUSION 173 from their provided coverage because they do not rise to the standard of care, the cost involved, and the uncertainty of safety. By excluding experimental technology from coverage, insurers aim to strike a balance between providing adequate coverage to their policy holders and preventing actuarially uncalculated payout that can undermine the financial stability of the insurer and force increases in insurance premiums. If the policy holder believes the adverse determination was unwarranted, litigation will ensue. As will be closely examined below, scholars have believed that health insurers particularly fare badly in courts in litigating coverage denial of experimental treatment, not least due to sympathetic judges. This article addresses that belief against empirical data. The issue warrants a closer look in light of the new regulations implementing the Patient Protection and Affordable Care Act (Affordable Care Act).6 Part II of this article will map out the overall landscape of disputes between health insurers and policy holders when coverage is denied because the treatment is experimental in addition to exploring the dominant perception in literature that health insurers have difficulty justifying their coverage denial in experimental treatment litigation. That perception is tested in Part III through an empirical analysis of all experimental exclusion litigation since 2003. Part IV will examine the policy implications of the analysis in light of the new federal regulations requiring external review of experimental exclusions. It will also examine the implication on the much broader issue of evidence-based coverage criteria that health insurers may consider. II. EXPERIMENTAL EXCLUSION AND DISPUTE Experimental technology is usually carved out from health insurance coverage through an exclusionary provision that requires one or more of the following validations for coverage: safety and efficacy proven through clinical trials (as reported in peer-reviewed medical literature), general acceptance as safe and effective treatment by the medical community, and/or Food and Drug Administration (FDA) approval when applicable. 7 Below is an example definition of "experimental" 6. Pub. L. 111-148. 7. Experimental treatments were at first excluded based on the general "medical necessity" clause, but a separate "experimental exclusion" provision began to be used in the 1970s when insurers were unsuccessful in excluding experimental technology simply based on the broad "medical necessity" requirement. Mark A. Hall and Gerard F. Anderson, Health Insurers'Assessment of Medical Necessity, 140 U. PA. L. REv. 1637, 1644-47 (1992). For the difference in function between "experimental exclusion" and "medical necessity," see Mary Ader, Access to InvestigationalTreatments, 6 HEALTH MATRIX 187, 190 (1996). Medicare, on the other hand, does not have a stand-alone exclusionary provision and relies on the "reasonable necessity" standard instead to exclude experimental treatment UNIVERSITY OF CINCINNATI LAW REVIEW [VOL. 83 from an actual case8: "[E]xperimental or investigational of for research purposes" is defined as a drug, biological product, device, treatment or procedure that meets any one of the following criteria, as determined by us: - Cannot be lawfully marketed without the final approval of the United States Food and Drug Administration (FDA) and which lacks such final FDA approval for the use or proposed use, unless (a) found to be accepted for that use in the most recently published edition of the United States Pharmacopeia-Drug Information for Healthcare Professional (USP-DI) or in the most recently published edition of the American Hospital Formulary Service (AHFS) Drug Information, or (b) identified as safe, widely used and generally accepted as effective for that use as reported in nationally recognized peer reviewed medical literature published in the English language as of the date of service; or (c) is mandated by state law; * Is a device required to receive Premarket Approval (PMA) or 510K approval by the FDA but has not received a PMA or 510K approval; - Is not identified as safe, widely used and generally accepted as effective for the proposed use as reported in nationally recognized peer reviewed medical literature published in the English language as of the date of service; - Is the subject of a National Cancer Institute (NCI) Phase I, II or III trial or a treatment protocol comparable to a NCI Phase I, II or III trial, or any trial not recognized by NCI regardless of phase; or * Is identified as not covered by the Centers for Medicare and Medicaid Services (CMS) Medicare Coverage Issues Manual, a CMS Operational Policy Letter or a CMS National Coverage Decision, except as required by state or federal law. How often are coverage claims denied on account of experimental or investigational treatment exclusion? According to one report, 9.59% and 5.77% of all claims denied by two major health insurers were coded as experimental exclusion and investigational treatment exclusion, respectively. 9 from coverage. 42 U.S.C. § 1395y(a)(1)(A). 8. Boldon v. Humana Ins. Co., 466 F.Supp.2d 1199, 1204-05 (D. Ariz. 2006). 9. AM. MED. ASS'N, NATIONAL HEALTH INSURER REPORT CARD 2013 (2013), available at http://www.nhxs.com/docs/files/File/2013-nhirc-results.pdf. The exact statistics are hard to come by, as 2014] RECALIBRATING EXPERIMENTAL TREATMENT EXCLUSION 175 When the policy holder cannot agree with the adverse determination by the insurer based on experimental exclusion, he or she can pursue a grievance process with the insurer and, if unsuccessful, further challenge the denial in litigation. The grievance process that a beneficiary can pursue with the insurer after claim denial includes internal appeal and external review. In an internal appeal, the claim denial is reviewed by the health insurer's internal personnel, while external review is handled by an outside party independent from the health insurer. Once the policy holder has exhausted the internal appeal and external review, the policy holder can file a lawsuit to challenge the claim denial. When an experimental exclusion case is brought to the court, literature tends to show that health insurers face an uphill battle, '0 with sympathetic judges often siding with the patient and creating "judgemade" coverage of experimental technology." This rather pessimistic view of health insurers' fate in experimental exclusion litigation finds support in a series of litigation in the late 1980s through the early 2000s arising from insurer's denial of coverage for one particular cancer therapy: high-dose chemotherapy and autologous bone marrow transplant (HDC/ABMT) for advanced breast cancer. 12 In a HDC/ABMT procedure, bone marrow is harvested from the cancer patient, and an intensified dose of cancer-killing alkylating agents is subsequently administered to the patient.1 3 The bone marrow is then salvaged by transfusing the pre-harvested bone marrow of the patient. The dose intensification of chemotherapy is designed to kill a not all insurers will strictly and consistently adhere to the administrative codes. 10. E.g., Grace Powers Monaco & Rebecca L. Burke, Insurer as Gatekeeper - Part Two: Policy Obstacles in Unproven Methods Litigation, 20 FORUM 400 (1984); Lee N. Newcomer, Defining Experimental Therapy - A Third-PartyPayer's Dilemma, 323 NEW ENGL. J. MED. 1702 (1990); Frank P. James, The Experimental Treatment Exclusion Clause, 12 J. LEGAL MED. 359 (1991); Paul J. Molino, Reimbursement Disputes Involving Experimental Medical Treatment, 24 J. HEALTH HOsP. L. 329 (1991); Hall & Anderson, supra note 7; Richard S. Saver, Reimbursing New Technologies: Why Are the Courts Judging Experimental Medicine?, 44 STAN. L. REV. 1095 (1992); M. Dubinsky & J.H. Ferguson , Court-Ordered Reimbursement for Unproven Medical Technology: Circumventing Technology Assessment, 269 JAMA 2116 (1993); Lee N. Newcomer, Technology Assessment, Benefit Coverage, and the Courts, in ADOPTING NEW MEDICAL TECHNOLOGY (Annetine Gelijns and Holly V. Dawkins ed., 1994); Ader, supra note 7; Natalie L. Regoli, Insurance Roulette: The Experimental Treatment Exclusion and DesperatePatients, 22 QUtIPIAC L. REV. 697 (2003); Peter D. Jacobson & Shannon Brownlee, The Health Insurance Industry and the Media: Why the Insurers Aren't Always Wrong, 5 Hous. J. HEALTH L. & POL'Y 235 (2005). 11. Michelle M. Mello & Troyen A. Brennan, The Controversy Over High-Dose Chemotherapy With Autologous Bone Marrow TransplantForBreast Cancer,20 HEALTH AFF. 101, 109 (2001). 12. This ill-fated procedure provided fodder large enough for numerous articles and a 350 pagesplus tome to be written about it. RICHARD A RETTIG ET AL., FALSE HOPE: BONE MARROW TRANSPLANTATION FOR BREAST CANCER (2007). 13. Raymond B. Weiss, The Randomized Trials of Dose-Intensive Therapyfor Breast Cancer: What Do They Mean for Patient Care and Where Do We Go From Here?, 4 ONCOLOGIST 450, 451 (1999). UNIVERSITY OF CINCINNATI LAW REVIEW [VOL. 83 greater number of cancer cells than regular dosage would, but, at the same time, it eradicates stem cells in the bone marrow, an otherwise lethal outcome but for the bone marrow transplant. Based on inconclusive evidence of the superiority of this treatment, 14 HDC/ABMT gained rapid acceptance in the 1990s. HDC/ABMT was an expensive procedure. It typically costs anywhere from $80,000 to $150,000, whereas conventional therapy costs between $15,000 and $40,000.15 Many health insurers initially denied coverage of the treatment on the ground that the technology had not fully proven its effectiveness through clinical trials. Inevitably, lawsuits followed, and courts were divided and outcome unpredictable.' 6 A number of factors influenced the court in deciding whether HDC/ABMT for breast cancer should fall under "experimental" treatment. These factors included whether the experimental exclusion clause was clearly defined in the insurance policy; whether the insurer had taken reasonable steps in arriving at its coverage denial; whether the medical community generally accepted HDC/ABMT for breast cancer; and whether the insurer's denial of coverage was solely driven by the 17 insurer's pecuniary interest. Under the threat of litigation as well as the mounting public 14. E.g., W.P. Peters et al., High-Dose Chemotherapyand Autologous Bone Marrow Support as Consolidation After Standard-Dose Adjuvant Therapy for High-Risk Primary Breast Cancer, II J. CLINICAL ONCOLOGY 1132 (1993). This phase 2 clinical trial used historical results as control, rather than a prospective randomized controlled study. Id. 15. U.S. GEN. ACCOUNTING OFFICE, HEHS-96-83, COVERAGE OF AUTOLOGOUS BONE MARROW TRANSPLANTATION FOR BREAST CANCER 3 (1996). 16. Michelle M. Mello & Troyen A. Brennan, supra note 11, at 108-09 For example, the court ordered payment in Kekis v. Blue Cross & Blue Shield, 815 F. Supp. 571 (N.D.N.Y. 1993); Wilson v. Group Hospitalization & Medical Services., Inc., 791 F. Supp. 309 (D.D.C. 1992); Kulakowski v. Rochester Hosp. Serv. Corp., 779 F. Supp. 710 (W.D.N.Y. 1991); White v. Caterpillar, Inc., 765 F. Supp. 1418 (W.D. Mo. 1991); Bucci v. Blue Cross-Blue Shield of Connecticut, Inc., 764 F. Supp. 728 (D. Conn. 1991); Adams v. Blue Cross & Blue Shield of Maryland, Inc., 757 F. Supp. 661 (D.Md. 1991); Pirozzi v. Blue Cross-Blue Shield of Virginia, 741 F. Supp. 586 (E.D.Va. 1990); Bailey v. Blue Cross/Blue Shield of Virginia, 866 F. Supp. 277 (E.D.Va. 1994), aff'd 67 F.3d 53 (4th Cir. 1995), Scalamandre v. Oxford Health Plans, Inc., 834 F. Supp. 1050 (E.D.N.Y. 1993), Simkins v. Nevadacare, 229 F.3d 729 (9th Cir. 2000), Henderson v. Bodine Aluminum, Inc., 70 F.3d 958 (8th Cir. 1995). On the other hand, the court denied payment in Fuja v. Benefit Trust Life Ins. Co., 809 F. Supp. 1333 (N.D. Ill. 1992), rev'd, 18 F.3d 1405 (7th Cir. 1994); Nesseim v. Mail Handlers Benefit Plan, 792 F. Supp. 674 (D.S.D. 1992), rev'd 995 F.2d 804 (8th Cir. 1993); Caudill v. Blue Cross-Blue Shield of North Carolina, 1992 WL 486661 (E.D.N.C. 1992), affrd 999 F.2d 74 (4th Cir. 1993); Harris v. Mutual of Omaha Cos., 992 WL 421489 (S.D. Ind. Aug. 26, 1992), aff'd 992 F.2d 706 (7th Cir. 1993); Clark v. K-Mart, No. 91-3723, 1992 WL 106935 (3rd Cir. 1992); Holder v. Prudential Ins. Co., 951 F.2d 89 (5th Cir. 1992); Boland v. King County Medical Blue Shield, 798 F. Supp. 638 (W.D. Wash. 1992); Thomas v. Gulf Health Plan, Inc., 688 F. Supp. 590 (S.D. Ala. 1988); Peruzzi v. Summa Medical Plan, 137 F.3d 431 (6th Cir. 1998). 17. Denise S. Wolf, Who Should Payfor "Experimental" Treatments?Breast Cancer Patients v. Their Insurers,44 AM. U. L. REV. 2029 (1995). 2014] RECALIBRATING EXPERIMENTAL TREATMENT EXCLUSION 177 pressure, 18 many health insurers acquiesced and reluctantly started to cover HDC/ABMT. 19 An estimated 42,680 HDC/ABMT procedures were performed on breast cancer patients during the 1990's and cost probably more than $3.4 billion.20 The controversy within the medical community surrounding the true benefit of the procedure, however, never subsided, due to the lack of any definitive results from statistically robust, randomized controlled clinical trials. After much difficulty-partly because breast cancer patients preferred to receive HDC/ABMT rather than enroll in a clinical trial in which they stood a chance of being assigned to the placebo control group--large-scale randomized controlled studies with statistical robustness only became available in the late 1990s. 21 Dashing hopes of many breast cancer patients and their families, these studies concluded that patients undergoing HDC/ABMT did not survive any longer than those that received conventional chemotherapy. 22 To make matters worse, patients who underwent HDC/ABMT suffered more serious side effects, such as infection, diarrhea, and vomiting.23 Subsequently, the procedure was gradually phased out as treatment for advanced breast cancer. The dramatic rise and fall of HDC/ABMT has left behind lessons for the medical community, insurers, patient advocacy groups, policy makers, and the courts.24 Over a decade has passed since the HDC/ABMT controversy. Is 18. If litigation was an effort to force health insurers to pay for the experimental HDC/ABMT procedure on an individual basis, a campaign with a far wider reach and more lasting effects on health insurance coverage for this experimental procedure was rolled out by the breast cancer patients and advocacy groups through legislative lobbying. On this battle front, the breast cancer advocacy groups were very successful. Many states passed legislation requiring health insurers to cover HDC/ABMT for breast cancer. Id. at 2094-2102. In addition, Office of Personal Management required that all Federal Employees Health Benefits Program cover the procedure. U.S. GENERAL ACCOUNTING OFFICE, supra note 15, at 3. These legislations went into effect before any conclusive evaluation of the procedure through randomized, controlled clinical trials had been completed. 19. U.S. GENERAL ACCOUNTING OFFICE, supra note 15, at 8-9. 20. Mello & Brennan, supranote 11, at 110. 21. Sjoerd Rodenhuis et al., Randomised Trial of High-Dose Chemotherapy and Haemopoietic Progenitor-CellSupport in Operable Breast Cancer with Extensive Axillary Lymph-Node Involvement, 352 LANCET 515 (1998); Edward A. Stadtmauer et al., Conventional-Dose Chemotherapy Compared with High-Dose Chemotherapy Plus Autologous Hematopoietic Stem-Cell Transplantation for Metastatic Breast Cancer, 342 1069 (2000); Gabriel N. Hortobagyi et al., Randomized Trial of HighDose Chemotherapyand Blood Cell Autografts for High-Risk Primary Breast Carcinoma,92 J. NAT'L CANCER INST. 225 (2000). 22. One South African physician reported positive results from randomized, controlled clinical trials of HDC/ABMT on breast cancer patients. His studies, however, were later found to be based on serious scientific misconduct. Raymond B. Weiss et al., High-Dose Chemotherapy for High-Risk PrimaryBreast Cancer:An on-Site Review of the Bezwoda Study, 355 LANCET 999 (2000). 23. Edward A. Stadtmauer et al., supra note 21, at 1074-75. • 24. E.g., Mello & Brennan, supranote 11; Peter D. Jacobson and Stefanie A. Doebler, We Were All Sold a Bill of Goods: Litigating the Science of Breast Cancer Treatment, 52 WAYNE L. REV. 43 (2006). UNIVERSITY OF CINCINNATI LAW REVIEW [VOL. 83 experimental exclusion still vulnerable to reversal by judges, as literature seems to believe? 25 How often do courts disagree with an insurer's findings that a technology-whether procedure, drug, or medical device-is experimental? When the court reverses the insurer's experimental exclusion, what is the reasoning behind the court's decision? In the next part, I will try to answer these questions by looking into court opinions in experimental exclusion litigation since 2003. III. RECENT TREND IN EXPERIMENTAL EXCLUSION LITIGATION: AN EMPIRICAL ANALYSIS The purpose of this empirical analysis is to assess the frequency and reasoning in the courts' disagreement with health insurer's determination of experimental technology. A. Data & Method 26 All data in this empirical study were collected using the advanced search in WestLaw Next. First, a WestLaw Next search string was built to capture all cases since 2003 in which insurance coverage was denied by an insurer or a third-party payer on the ground that the treatment technology involved was "experimental," "investigational" or "unproven." The year 2003 was selected as the cut-off in light of the fact that HDC/ABMT litigation for breast cancer largely disappeared by 2003.27 The HDC/ABMT litigation has been chosen for comparison for a couple of reasons: (1) there have been studies that looked into the HDC/ABMT litigation data and (2) the HDC/ABMT could have been the turning point of the court's attitude towards experimental treatment. The following was the search string used in WestLaw: DA(aft 01-01-2003 & bef 08-31-2013) & OP((experimental investigational unproven) /s (procedure drug device therapy treatment transplant) and (deni! /4 (cover! benefit! reimburs! authoriz! claim claims 28))) 25. E.g., Mello & Brennan, supra note 11, at 112-14; P. D. Jacobson et al., Litigating the Science of Breast CancerTreatment, 32 J. HEALTH POL. POL'Y & L. 785, 809-10 (2007). 26. For methodologies on case selection and coding for content analysis in general, see Mark A. Hall and Ronald F. Wright, Systematic Content Analysis of Judicial Opinions, 96 CALIF. L. REV. 63 (2008). 27. As late as 2002, HDC/ABMT coverage denial was still litigated. E.g., Zervos v. Verizon New York, Inc. 277 F.3d 635 (2d Cir. 2002). 28. WestLaw has recently banned a wild card use of the word "claim" as in "claim!". 2014] RECALIBRATING EXPERIMENTAL TREATMENT EXCLUSION 179 The initial search resulted in a total of 597 opinions, 29 of which 519 were classified by WestLaw Next as civil cases and 155 as criminal. 30 The vast majority of the civil cases were disability cases, employee benefit cases, or patent cases containing all search keywords or quoting a precedent in which all of the search keywords were included. This required skimming all 519 civil cases to select only those in which experimental exclusion was actually disputed, 3' netting 135 cases of genuine experimental exclusion dispute.32 I further applied additional inclusion and exclusion criteria to these 135 experimental exclusion cases. Excluded from the analysis were cases in which the court did not review the merit of the insurer's experimental exclusion. For example, cases decided purely on a defect in the insurer's grievance proceduresuch as the insurer's failure to send proper notification-were excluded.33 By the same rationale, Medicare cases in which the decision was based on the exclusionary rule-making process of the program, rather than on the third-party payer's proper application of the exclusionary rule, were excluded.34 On the other hand, opinions that did not dispose of the case-i.e., opinions that did not result in a "win" for either the plaintiff or the defendant-were included, as long as the opinion provided the court's substantive reasoning for either accepting or rejecting the insurer's For example, a decision to deny the experimental exclusion. defendant's motion to dismiss and instead to send the case to trial would be included if the opinion passed its judgment on the insurer's application of the experimental exclusion, even though the case was not disposed of. 35 Similarly, decisions that remanded the case to the insurer for further review consistent with the court's opinion were retained. Also included were both trial and appellate court decisions of the 29. As WestLaw constantly adds cases to their database, running the same search string may turn up different results depending on when the search is run. The data in this study are based on the search results as of November 1, 2013. Upon inquiry, the author will provide the list of the 519 civil cases categorized into (1) non-experimental exclusion cases, (2) experimental exclusion cases without relevance, and (3) experimental exclusion cases with relevance for this study. 30. The sum of these two categories exceeds 597, because some of the cases were categorized as both civil and criminal by WestLaw. 31. Because this determination of relevance requires judgment of the researcher, it may affect the general reproducibility of the study. The reasonably objective case selection criteria somewhat mitigates that risk. Hall & Wright, supra note 26, at 106. 32. One case where the defendant was an external review agency but the result was imputed to the insurer was also included. Vellios v. IPRO, External Review Agent, 765 N.Y.S.2d 222 (Sup. Ct. 2003). 33. E.g., Zalduondo v. Aetna Life Ins. Co., 941 F. Supp. 2d 125 (D.D.C. 2013). 34. For example, the plaintiff may challenge an exclusionary provision in the federal regulation itself (e.g., 42 C.F.R. § 411.15(k)(1)) rather than challenge the interpretation or application of that rule. E.g., Yale-New Haven Hosp. v. Leavitt, 470 F.3d 71 (2d Cir. 2006). 35. E.g., Chapman v. Anthem Health Plans of New Hampshire, Inc., no. 03-CV-480-PB, 2005 WL 1123949 (D.N.H. May 12, 2005). UNIVERSITY OF CINCINNATI LAW REVIEW [VOL. 83 same case, if both reviewed the merits of the insurer's coverage denial.36 To avoid "double-counting" of the same case in the analyses below, however, only the highest-level opinion is tallied when the dependent 37 variable being examined is the reversal of the insurer's determination. Lastly, a few first-party insurance disputes were also included, in which the treatment's experimental nature was advanced as defense for non-coverage by the insurer, even in the absence of a specific experimental exclusionary provision in the policy. Although they are not technically health insurance litigation, they nonetheless qualify as reflection of the courts' reasoning on the appropriateness of the insurer's determination of experimental medical technology. Applying these exclusion and inclusion criteria resulted in sixty-three court opinions, representing fifty-seven unique cases when multiple opinions (e.g., district court opinion and appellate court opinion) of the same case are counted as one. The final results of the selection process are compiled in Table 1. Figure 1 Case Selection Process Initial Search I(nm697)I 36. By the same token, both preliminary injunction opinion and the subsequent opinion on the merits of the same case would have been included, but there was no such case. 37. Mark A Hall et al., JudicialProtectionof Managed Care Consumers: An EmpiricalStudy of Insurance Coverage Disputes, 26 SETON HALL L. REV. 1055 (1995) took this approach and included the highest-level decision when appeals resulted in multiple decisions from a single case. 2014] RECALIBRATING EXPERIMENTAL TREATMENT EXCLUSION 181 Table 1 Total Number of Opinions Found Published Unpublished Total Federal 14 36 50 State 7 6 13 Total 21 42 63 1. Coding All selected opinions were then coded according to a code book for content analysis. 38 The code book contained variables that identified the case (title, citation, court, date of decision, reporter), parties (plaintiff type, insurer type 3 9), technology in dispute (description of technology, target disease, seriousness of the disease, technology category, 41 validation of technology at the time of coverage denial 42), procedural attributes (granularity of experimental definition in the policy, preauthorization, initial coverage reviewer, 4 3 number of appeals, appeal reviewers, 44 state-mandated reviews), court's decision (standard of review, judgment & relief 45), and court's reasoning.46 The coding was performed solely by the author to ensure consistency. 47 It turns out that many opinions did not contain sufficient information to fill all the variables in the code book. Consequently, not all variables were utilized in the ensuing analyses. 38. Copies of the code sheet and the raw data used in this study are on file with the author and available by request. 39. The vast majority were health plans under ERISA, but some were public programs such as Medicare or Medicaid. As mentioned earlier, first-party insurance disputes are also included to the extent the court passed its judgment on the insurer's denial of benefit based on experimental exclusion. 40. This variable had a Boolean value of either true or false and would have "True" value if the disease is a life-threatening disease such as cancer. 41. Drug, device, procedure or combination. 42. Under clinical trials, safety/efficacy proven for other indication, or unproven. 43. Internal or external. 44. Internal or external. 45. The options for this variable include "judgment in favor of plaintiff/defendant," "remand to insurer for further review consistent with the court's opinion," "send to trial," "preliminary injunction granted/denied," and "plaintiff prevails on experimental exclusion but loses on other ground." 46. The options for this variable include "interpretation of the policy by insurer rejected," "other studies supporting the technology at issue," "expert opinions," "treating physician's opinion," "general acceptance in medical community," "other insurer's coverage," "irregularity in insurer's past determination." "insurer's burden to prove experimental nature not met," and "miscellaneous." 47. Reproducibility of coding is an important issue in content analysis. Mark A. Hall & Ronald F. Wright, Systematic Content Analysis of Judicial Opinions, 96 CALIF. L. REv. 63, 112-16 (2008). Since most of the coding variables used in this study are meant to capture objective characteristics of the opinions, a separate statistical testing of the reliability of subjective coding was not conducted. UNIVERSITY OF CINCINNATI LAW REVIEW [VOL. 83 2. Why No Statistical Tests? Not all opinions written by judges are published in the official court reporters, such as the Federal Reporter. When the authoring judge believes that the opinion does not have precedential value, he or she can suppress publication. 48 "Unpublished" does not mean "unavailable," however. Commercial legal databases such as WestLaw carry a large number of "unpublished" court opinions, both federal and state, in addition to all published opinions. 49 WestLaw chooses which unpublished opinions will be made available on their database based on certain proprietary metrics, including the importance of the decision. 50 Unpublished federal court opinions are said to be more widely available than unpublished state court opinions in commercial legal databases.51 Overall, both published opinions and unpublished opinions available on WestLaw are "selected" based on certain criteria by judges in the former and by WestLaw personnel in the latter. Consequently, the opinions available on WestLaw-the sum of published opinions and Westlaw-available-unpublished opinions-may not share the same traits as the rest of all written opinions. It follows that making an inference to the population of interest in this analysis-all court opinions between 2003 and 2013-from the WestLaw-available opinions using a formal statistical test is likely inappropriate and potentially misleading. That is the reason this study presents only descriptive analyses and does not attempt to make statistical inferences. 48. Andrew T. Solomon, Making Unpublished Opinions Precedential:A Recipe for Ethical Problems & Legal Malpractice,26 MIss. C. L. REV. 185, 189-191 (2006). "Unpublished federal circuit opinion" has actually become a misnomer because, since 2005, even those opinions chosen to become "unpublished" by the court are now published in the FederalAppendix. Id. at 207-08. 49. Unreported Decisions, ZIMMERMAN'S RESEARCH GUIDE, http://Iaw.lexisnexis.com/infopro/zimmermans/disp.aspx?z=208 (last visited Apr. 1, 2014). 50. E-mail from WestLaw representative to author (June, 2013) (on file with the author). 51. Solomon, supra note 48, at 210. (citing two reasons for the less availability of state unpublished opinions: (1) less interest in aggregating state unpublished opinions on the part of commercial legal database vendors, and (2) no comprehensive statutory requirement, like the federal EGovernment Act of 2002, Pub. L. No. 107-347, for the state courts to provide online access to opinions at the state level). 2014] RECALIBRATING EXPERIMENTAL TREATMENT EXCLUSION 183 B. Results 1.Overall Trend Table 2 Overall Outcomes E~rimento! Eprmental Detwniarop 00tDulnaion 2003 2 1 3 2004 1 8 9 2005 3 3 6 2006 4 4 8 2007 1 1 2 2008 1 5 6 2009 1 4 S 2010 1 3 4 2011 1 8 9 2012 1 4 5 Year Total 16 (2Z.1%) 41 71.9%) Total 57 Table 2 is a break-down of opinions in the data by the year in which each opinion was rendered. 52 It counts only the highest-level opinion if multiple opinions of the same case are available, as mentioned above.5 3 There is no noticeable increase or decrease in the number of cases during the ten-year period. No previous study gathered the results of experimental exclusion litigation in a similar manner, but Jacobson's study of reported cases of HDC/ABMT litigation between 1988 and 2002 provides meaningful comparison.54 During that period, patients won forty-four cases and insurers won forty-three out of a total of 87 HDC/ABMT cases, providing a more even split between patients and 52. In Table 2 and Table 3, caution is warranted in interpreting the "experimental determination rejected" and "experimental determination upheld" columns. As the column legends read, the categorization is based on whether the courts reversed the insurer's determination that some treatment was experimental. As such, the numbers under the "experimental determination rejected" column do not necessarily mean "wins" for the patient. For example, in one rare instance, the court rejected the insurer's determination of the technology at issue as experimental but still found for the insurer on some other ground. McHenry v. PacificSource Health Plans, 679 F. Supp. 2d 1226 (D. Or. 2010). In addition, one case was remanded back to the insurer for redetermination of coverage, which ultimately may or may not have reversed the initial exclusion, and another was sent to trial. 53. This explains why the total number of opinions below are slightly smaller than the total tallied in Table 1. 54. P. D. Jacobson et al., Litigatingthe Science of Breast Cancer Treatment, 32 J. HEALTH POL. POL'Y & L. 785, 790-92 (2007). UNIVERSITY OF CINCINNATI LAW REVIEW [VOL. 83 insurers. 55 In contrast, as can be seen above, the results from the more recent ten-year period are far more favorable to the insurers. Table 3 is a tabulation of the results segmented by various independent variables. Table 3 Analysis of Experimental Exclusion Cases Experimental Determination Rejected 16 (28.1%) Total U _ -h-ed blJ~* Published Experimental Determination Upheld 41 (71 SM) Total 67 ------- Unpublished 10 (50.0%) 6 (16.2%) 10 (50%) 31 (83.8%) 20 37 Federal 12 (26.7%) 33 (73.3%) 45 4(33.%) 8(66.7%) 12 6 (71A%) 10 (83.3%) 24(66.7%) 1 (100.0%) 1 (100l0%) 7 12 36 State Device Drug Procedure Device & Procedure Drug & Procedure 2(28.6%) 2 (16.7%) 12(33.3%) Ufe-Threaening Not Life-Threatening 6 (30.0%) 10 (27.0%) 14 (70.0%) 27 (73.0%) 20 37 ERISA Self-insured ERISA Non-Self Insured 2 (22.2%) 10 (313%) 7 (77Z%) 22[68.8%) 9 32 1 1 Other Prvate 2(100.0%) 2 First-Party Insurer Medicare Other Federa Programs I 110.0%) 3 (100A0%) 1 500%) I 3 2 (26.0%) 3 (76A%) 4 1 (33-3%) 1 (100.0%) 2 (66.7%) 3 1 Medicaid Other State Programs Not clear 1 (50.0%) 2. Published vs. Unpublished Unpublished opinions available on WestLaw, both federal and state, were included in this study for three reasons. 56 First, for content 55. Id. Jacobson et al. also included only reported cases in which the court addressed substantive legal questions and excluded those cases decided on procedural issues. 56. The citation format for the unpublished opinions in this study contains the designation "WL," 2014] RECALIBRATING EXPERIMENTAL TREATMENT EXCLUSION 185 analysis that examines the reasoning behind the ruling, the lack of precedential value should not be as important. Second, one of the objectives of this study is to test the belief found in literature that experimental exclusion determined by insurers is vulnerable to reversal by the court. 57 Since past literature on experimental exclusion cases cited both published and unpublished opinions, it is only appropriate to include unpublished opinions in this study as well. Last, inclusion of unpublished opinions would mitigate publication biases that may arise if only published opinions are used in the analysis. 58 By including as many opinions as available in the analysis, the data will gain more resemblance to the population of all written opinions. As referenced in Table 3, the publication status of the case is the only independent variable which seems to affect the overall outcome. For all other independent variables, the outcome is roughly 3:7 in favor of the insurer, unless the sample size is too small. In previous studies exploring other types of litigation, published opinions as a whole had different outcomes from the unpublished opinions. 59 The discrepancy between the published opinions and the unpublished ones is probably the result of selection bias-i.e., a case where an individual defeats a corporate insurer is more likely to interest judges and commercial database vendors for publication. This raises the possibility that unpublished opinions unavailable on WestLaw are more likely to be in favor of the insurer than the patient. If that is the case, the actual outcome of all experimental exclusion litigation would be even more favorable towards insurers than the numbers suggest in this study. 3. Federal Court Opinions vs. State Court Opinions The data in this study contain far more federal court opinions than state court opinions.60 The difference can easily be explained with the which stands for WestLaw. 57. See supra note 25. 58. W.M. Sage, Judicial Opinions Involving Health Insurance Coverage: Trompe L 'Oeil or Window on the World, 31 IND. L. REv. 49, 65-66 (1998); Michelle M. Mello & Kathryn Zeiler, Empirical Health Law Scholarship: The State of the Field, 96 GEO. L. J. 649, 658 (2007). Previous empirical analyses of health insurer's denial of coverage relied on published opinions only. See Mark A Hall et al., Judicial Protection of Managed Care Consumers: An Empirical Study of Insurance Coverage Disputes, 26 SETON HALL L. REv. 1055 (1995) (coverage disputes in general); P.D. Jacobson et al., The Role of the Courts in Shaping Health Policy: An EmpiricalAnalysis, 29 J.L. MED. & ETHICS 278 (2001) (litigation against managed care organizations in general). 59. See Peter Siegelman & John J. Donohue, Studying the Icebergfrom Its Tip: A Comparisonof Published and Unpublished Employment Discrimination Cases, 24 LAW & Soc'Y REv. 1133 (1990) (discovering difference in patterns between published and unpublished opinions in employment discrimination cases). 60. This finding is shared by two previous empirical studies on health insurance litigation. See UNIVERSITY OF CINCINNATI LAW REVIEW [VOL. 83 preemption under the federal Employee Retirement Income Security Act (ERISA) for employer-sponsored health insurance. 6 1 Adding to the difference is WestLaw's tendency to catalogue federal court opinions more extensively. In Table 3, reversal rates between federal and state courts are very close. In ERISA litigation, coverage denial is usually reviewed under an abuse of discretion standard, rather than the more strict de novo standard, if the claim administrator has discretion to interpret the plan. 62 Under the abuse of discretion standard, the court considers whether-the denial to provide a benefit was arbitrary and capricious, as opposed to whether the court would have reached the same conclusion as the claim administrator for the insurer. The abuse of discretion standard thus makes it somewhat more difficult for the plaintiff to reverse the administrator's denial of coverage.6 3 In light of the abuse of discretion standard that many of the 64 federal ERISA cases applied, the similar reversal rate is very interesting. 4. By Technology Type Categorization of the disputed technology into either drug, device or procedure can be tricky in certain situations. The boundary between one technology category and another becomes blurred when multiple technologies are involved. In such cases, the coding was based on the most salient experimental aspect of the treatment. For example, intraarterial delivery of radioactive element ittrium-90 to treat liver cancera "procedure" using a new "device"-was categorized as a "device" Hall et al., supra note 37; Peter D. Jacobson, Role of ERISA Preemption in Health Reform: Opportunitiesand Limits, 37 J. L. MED. & ETHICs 88 (2009). 61. 29 U.S.C. § 1144(a); see also Pilot Life Ins. Co. v. Dedeaux, 481 U.S. 41 (1987) (finding that ERISA preempts state law tort and contract claims for improper processing of a benefit claim). The ERISA preemption has also helped the availability of written opinions by foreclosing jury trials. Although ERISA does not expressly state whether the beneficiary is entitled to jury trial, it has been generally viewed by the courts that recovery of denied benefit under 29 U.S.C. §§ 1132(a)(l)(B) and 1132(a)(3)(B) is equitable in nature and not eligible for jury trial. Sage, supra note 58, at 54-55; but see Donald T. Bogan, ERISA: Re-Thinking Firestonein Light of Great-West - Implicationsfor Standard of Review and the Right to a Jury Trial in Welfare Benefit Claims, 37 J. MARSHALL L. REv. 629 (2003). For a primer on ERISA preemption of health insurance coverage litigation, see Jacobson, supra note 60. 62. Firestone Tire & Rubber Co. v. Bruch, 489 U.S. 101, 109 (1989). Jacobson, supranote 60, at 90-91; Patricia C Kuszler, FinancingClinical Research and ExperimentalTherapies: Payment Due, but from Whom, 3 DEPAUL J. HEALTH CARE L. 441, 470-74 (1999). 63. P. D. Jacobson et al., Litigatingthe Science of Breast Cancer Treatment, 32 J. HEALTH POL. POL'Y & L. 785, 792-795 (2007); E. Haavi Morreim, Benefits Decisions in ERISA Plans: Diminishing Deference to Fiduciariesand an Emerging Problemfor Provider-SponsoredOrganizations,65 TENN. L. REv. 511,517-23 (1998). 64. However, during the HDC/ABMT litigation in the late 1980s through the early 2000s, the plaintiffs did slightly better than the defendants even in ERISA cases - winning 31 cases and losing 25 cases between 1988 and 2002, many of which were presumably reviewed under the abuse of discretion standard also. Jacobson et al., supranote 54, at 793. 2014] RECALIBRATING EXPERIMENTAL TREATMENT EXCLUSION 187 case, because the effectiveness of the radioactive element on liver cancer had not been validated, whereas the delivery method was not experimental per se. 65 If a case equally involved multiple treatments of different categories, then multiple categories were assigned to the variable.66 While some scholars suggest that the lack of regulatory governance of clinical procedures is partly to blame for the large number of HDC/ABMT disputes, 67 off-label utilization of drugs and medical devices, which do have to receive regulatory clearance, still represent one-third of all cases as shown in Table 3. In almost all of the cases where the disputed technology was a drug or a device, they were an FDA-approved drug or device, prescribed by the treating physician for off-label use, rather than an experimental drug without any FDAapproved indication. 68 The reversal rate is very similar across all types of technologies. 5. Life-Threatening Condition Experimental exclusion cases often deal with life-threatening situations for the patient who runs out of conventional treatment options and has to resort to unproven technology as a last hope. In that situation, the court may lend a sympathetic ear to the plaintiff.69 One possible measurement of the court's sympathy towards the patients' agony is the correlation between the seriousness of the patient's condition and the outcome of the case. According to the data in this study, whether the patient's condition was life-threatening or not does not seem to have affected the outcome. An earlier empirical study 70 examining "medical necessity" disputes reported similar results. 65. See Boldon v. Humana Ins. Co., 466 F. Supp. 2d 1199 (D. Ariz., 2006). 66. In Jacobs v. Guardian Life Ins. Co. of Am., 730 F. Supp. 2d 830 (N.D. IT1.,2010), both experimental procedure (intensity modulated radiation therapy) and off-label use of drugs (Avastin and Abraxane) were at issue. 67. E.g., Jacobson et al., supra note 54, at 796. 68. There was only one case in which the drug had not received approval for any indication at all at the time of litigation. It was Dr. Burzynski's antineoplaston in Emerson v. Medical Mutual of Ohio, 2004-Ohio-3892 (Ohio App. July 23, 2004). Dr. Burzynski's confrontation with the FDA is welldocumented in U.S. v. Burzynski Cancer Research Inst., 819 F.2d 1301 (5th Cir. 1987). 69. See Mark A. Hall & Gerard F. Anderson, Health Insurers'Assessment of Medical Necessity, 140 U. PA. L. REv. 1637, 1657-1662 (1992) (arguing that the court's sympathy can lead to undesirable consequences, such as pricing policy holders out of the market by raising premiums and discouraging insurers from cost-containing efforts). 70. Mark A. Hall et al., JudicialProtection ofManaged Care Consumers: An Empirical Study of Insurance Coverage Disputes, 26 SETON HALL L. REV. 1055, 1064-65 (1995) (concluding from sample data that the likelihood of death did not have consistent effect on the outcome in litigation involving managed care's medical necessity cases). UNIVERSITY OF CINCINNATI LAW REVIEW [VOL. 83 6. By Insurer Type Since health insurance companies are sometimes vilified as profitoriented businesses, 71 one may wonder if not-for-profit insurers will be viewed more generously by the court. For example, in self-insured group health plans sponsored by employer, the financial risk is born by the plan, not by an insurer, although an insurance company may still handle claims and benefit determination as an administrator. Likewise, public programs like Medicare or Medicaid are not businesses and may escape any possible bias of the court. From Table 3, it is difficult to tell if any type of insurer fares better or worse than the other. Self-insured ERISA plans do not seem to have done better than health plans fully funded by health insurers, although statistical significance is unavailable. Likewise, public programs do not appear to have had their decisions accepted by the court more often, although Medicare prevailed in all three cases. C. Reasonfor Rejecting ExperimentalExclusion This subpart analyzes the reasoning of the court in all of the opinions in which the court disagreed with the insurer's experimental exclusion.72 In total, there were eighteen cases in which the court rejected the insurer's experimental exclusion. When more than one type of reasoning was employed in an opinion, all of them were counted. The reasons are further grouped into: (1) rejection of insurer's interpretation of the exclusionary provision in the policy; 74 (2) evidentiary grounds; and (3) other reasons. 71. Peter D. Jacobson & Shannon Brownlee, supranote 10, at 235. In one of the opinions in the data set, the court expressed distaste for the way coverage was denied based on experimental exclusion. Czarnopys v. Crystal Flash L.P., 200 F.Supp.2d 780, 781 (W.D. Mich. 2002) ("Let this case stand as notice, health care plans do deny coverage for life saving procedures with nary a care for the carnage left in their wake."). Incidentally, the defendant in this case was a self-insured employer-sponsored health plan. 72. For this analysis, multiple (i.e., both lower and appellate) opinions of the same case are counted, whereas in the preceding analyses only the highest-level opinion was counted. A larger pool of opinions, regardless of whether or not they were upheld on appeal, will facilitate the objective of this exercise, which is to analyze the reasoning of the court whenever it reverses the insurer's experimental exclusion. By similar logic, one opinion that disagreed with the insurer's determination of experimental nature of the therapy but still ruled in favor of the insurer on some other ground is also included. McHenry v. PacificSource Health Plans, 679 F. Supp. 2d 1226 (D. Or. 2010). 73. This explains why the numerical sums of the reasons exceed 20. 74. This group includes Medicare or Medicaid cases, in which the court turned on the proper interpretation of the governing statute on the benefit coverage. E.g., Edmonds v. Levine, 417 F. Supp. 2d 1323 (S.D. Fla. 2006). 75. These include coverage of the same technology by other insurer, a balancing between the likelihood of success on the merits and irreparable harm in preliminary injunction cases, lack of individualized determination, FDA approval obtained during dispute, etc. 2014] 189 RECALIBRATING EXPERIMENTAL TREATMENT EXCLUSION Table 4 is a summary chart of this break-down, and Table 5 shows what reason(s) the court provided for reversing the insurer's exclusion in each case. Table 4 Reasons for Overturning Insurer's "Experimental" Exclusion Court rejected insurer's interpretation of experimental exclusion provision Burden of proof not met by insurer Other studies supporting effectiveness of treatment 4 (22.2%) 7 (38.9%) Wide use/general acceptance 5 (27.8%) Expert opinion 3 (16.7%) Treating physician's opinion Other reasons 7 (38.9%) 11 (61.1%) Evidence on safety/efficacy of technology at issue 1 (5.6%) 2 (11.1%) During the HDC/ABMT litigation era prior to 2003, the dispute centered around the interpretation of the insurance policy. The key issue was usually whether terms such as "experimental" or "unproven" were ambiguous and should be interpreted against the drafter under the doctrine of contra preferentum or not. 76 In contrast, in the data set examined by this study, the interpretational question of the underlying insurance policy was the decisive factor in only seven cases, representing 38.9% of all cases in which the experimental exclusion was rejected by the court.77 Why is the interpretation less of an issue now than during the HDC/ABMT days? One possible answer is that the insurers may have learned from past litigation and refined the vulnerable parts of the insurance policy, such as the definition of "experimental," leaving little room for court to maneuvering around semantics. 7' But considering that the court can always find its way around even a very detailed exclusionary provision if it chooses to, which it sometimes did in HDC/ABMT litigation, 79 a more accurate assessment is that courts in general do not strain the interpretation to provide coverage to patients as 76. Michelle M. Mello & Troyen A. Brennan, The Controversy Over High-Dose Chemotherapy With Autologous Bone Marrow TransplantFor Breast Cancer, 20 HEALTH AFF. 101, 108-09 (2001). See also Mark A. Hall & Gerard F. Anderson, Health Insurers'Assessment of Medical Necessity, 140 U. PA. L. REv. 1637, 1644-57 (1992) (reviewing court opinions that the authors found to have employed strained logic in order to reverse health insurers' denial of coverage based on lack of medical necessity). 77. Interpretation was a decisive issue in 6 cases, but often in combination with other noninterpretational grounds. 78. Mary Ader, Access to InvestigationalTreatments, 6 HEALTH MATRIX 187, 192 (1996). 79. In one case, the court found the exclusionary phrase "tissue transplant coverage is limited to allogenic bone marrow only" (allogenic meaning from another person) in the policy would not be understood by an "average person" to exclude stem cell transplant because an average person would not understand the term "tissue" to encompass stem cell. Simkins v. NevadaCare, Inc., 229 F.3d 729, 735 (9th Cir. 2000). UNIVERSITY OF CINCINNATI LAW REVIEW [VOL. 83 often now. Even when the court took issue with the insurer's interpretation, the reasoning was often, though not always, more nuanced. For example, in one opinion, the insurer tried to equate "a demonstrated value based on clinical evidence reported by peer-review medical literature or by generally recognized academic experts" with "standard of care." The court rejected that construction and instead ruled that a technology doesn't need to become a standard of care to be considered nonexperimental. 80 Another opinion interpreted the drug reimbursementeligibility criteria under the Medicaid Act which permitted reimbursement of all compendia-listed drugs, 81 not as being conditional upon proof of efficacy for the prescribed indication through randomized controlled clinical trials. 82 In another case, the court rejected the insurer's argument that scientific evidence of long-term improvement, as opposed to short-term improvement, is required, if there is no such 83 specific requirement in the policy. Table 5 Reasons for Rejecting Insurer's Experimental Exclusion A B-1 B-2 B-3 B-4 B-5 C Wheeler v. Aetna Life Ins. Co., 2003 WL 21789029 Vellios v. IPRO, External Review Agent, 1 Misc.3d 468 0 Chambers v. Coventry Health Care of Louisiana, Inc.,318 F.Supp.2d 382 0 Hospital Authority of Houston County v. Bohannon, 272 Ga.App. 96 0 0 Chapman v. Anthem Health Plans of New Hampshire, Inc.,2005 WL 1123949 Tarraferro v. State ex rel. Wyoming Medical Com'n, 123 P.3d 912 Edmonds v. Levine, 417 F.Supp.2d 1323 0 0 0 Johnson v. Blue Cross & Blue Shield of Alabama, Inc., 457 F.Supp.2d 1288 Boldon v. Humana Ins. Co., 466 F.Supp.2d 1199 80. 81. 82. 83. 2009). 0 Chambers v. Coventry Health Care of La., Inc., 318 F. Supp. 2d 382, 391 (E.D. La. 2004). 42 U.S.C. § 1396r-8(g)(1)(B)(i). Edmonds v. Levine, 417 F. Supp. 2d 1323 (S.D. Fla. 2006). Welch v. HMO Louisiana/Blue Cross, no. 08-4576, 2009 WL 3401046 (E.D. La. Oct. 20, 2014] RECALIBRATING EXPERIMENTAL TREATMENT EXCLUSION 0 0 Whitley v. Carolina Care Plan, Inc., 2006 WVL 3827503 Pinckney v. Blue Cross Blue Shield of Tennessee, Inc., 2007 WL 108886 Summers v. Touchpoint Health Plan, Inc., 309 Wis.2d 78 * Klein v. Central States, Southeast & Southwest Areas Health & Welfare Plan, 2009 WL 455342 Welch v. HMO Louisiana/Blue Cross, 2009 WL 3401046 0 0 0 0 0 McHenry v. PacificSource Health Plans, 679 F.Supp.2d 1226 * Lafferty v. Providence Health Plans, 706 F.Supp.2d 1104 Prowell v. UPS Flexible Benefits Plan, 2011 WL 5110291 191 O O O 0 O O 0 * Berge v. U.S., 879 F.Supp.2d 98 O * reversed in appeal A: Court rejected insurer's interpretation of the exclusionary provision B-I Burden of proof not met by insurer B-2 Other studies supporting effectiveness of treatment B-3: Wide use/general acceptance in the medical community B-4: Expert opinion B-5: Treating physician's opinion C: Other reasons The courts reversed the insurer's experimental exclusion on evidentiary grounds rather than on interpretational grounds, when the court either: (a) did not find sufficient evidence to support the insurer's determination that the technology at issue was experimental: (b) found evidence submitted by the plaintiff that supported the non-experimental nature of the technology; (c) found wide (or general) acceptance of the technology in the medical community; (d) relied on expert testimony; or (e) relied on the treating physician's opinion. Of special interest from a policy perspective is the court's receptiveness towards expert testimony and the treating physician's opinion as evidence supporting the non-experimental nature of the excluded technology, which is widely accepted in the medical community. In many of the HDC/ABMT cases, the court equated wide acceptance of the procedure by the medical community with safety and UNIVERSITY OF CINCINNATI LAW REVIEW [VOL. 83 effectiveness of the treatment. 84 On one hand, it is difficult to find fault with the courts for believing a treatment so widely practiced by the medical community to be safe and effective. 85 Some argue that it is the medical community that should be responsible for disseminating a technology before its full validation. 86 In this study's data, wide acceptance is less often relied on by the courts. When the court ruled based on the wide acceptance of the technology in the medical community, 87 it was never the sole ground for the court's ruling; there were always other reasons, such as interpretational reasons or other studies supporting safety and effectiveness of the technology. Maybe it is the court's tacit acknowledgement that "wide acceptance" does not test the scientific validity of the technology and that it needed additional reasons to reverse the experimental exclusion in its decision. 1. Expert Witness Testimony Literature predicts that coverage denial litigation entails "war of expert witnesses" in an adversarial setting. 88 In the HDC/ABMT litigation, expert witnesses indeed played a critical role in persuading courts. 89 Interestingly, however, this study actually discovered paucity of court's reliance on expert testimony. The unforeseen paucity in the role of expert witness in experimental exclusion litigation can perhaps be explained by three factors. The first is the emergence of new doctrines on expert witness testimony under the three Supreme Court decisions known as the Daubert Trilogy, 90 which limit the admissibility 91 of expert witnesses' testimony to those that are reliable and relevant. The Daubert standard, which emerged during the height of the 84. Peter D. Jacobson & Stefanie A. Doebler, We Were All Sold a Bill of Goods: Litigating the Science of Breast Cancer Treatment, 52 WAYNE L. REv. 43, 93 (2006). The HDC/ABMT for breast cancer gained wide acceptance before any statistically robust clinical trials validated its safety and effectiveness. See supra text accompanying notes 12-23. 85. P. D. Jacobson et al., Litigating the Science of Breast Cancer Treatment, 32 J. HEALTH POL. POL'Y & L. 785, 799-801 (2007). 86. Id. at 815 (arguing that the HDC/ABMT controversy was a reflection of the lack of a good technology evaluation mechanism in the medical community). 87. From a litigation perspective, what amounts to "wide acceptance in the medical community" may be critical. The Court relied on expert opinions or literature to determine whether the technology is in wide use. 88. See e.g., id. at 105; Arnold J. Rosoff, Evidence-Based Medicine and the Law: The Courts Confront Clinical PracticeGuidelines, 26 J. HEALTH POL. POL'Y & L. 327, 331 (2001); David M. Eddy, Commentary: The Use of Evidence and Cost Effectiveness by the Courts: How Can It Help Improve Health Care?, 26 J. HEALTH POL. POL'Y & L. 387, 393 (2001). 89. Jacobson & Doebler, supra note 24, at 66-68. 90. Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993); Gen. Electric Co. v. Joiner, 522 U.S. 136 (1997); Kumho Tire Co. v. Carmichael, 526 U.S. 137 (1999). 91. Daubert, 509 U.S. at 584-87. 2014] RECALIBRATING EXPERIMENTAL TREATMENT EXCLUSION 193 HDC/ABMT litigation, may have placed greater scrutiny on the expert witnesses' qualification in recent years compared to the HDC/ABMT litigation era. 92 The second possible explanation is the procedural aspect of experimental exclusion disputes. When policy holders are entitled to appeal for denial of coverage,9 3 each side's experts may have had ample chance to submit letters and opinions, which become part of the litigation record and lessenes the need to bring the expert witness to the stand in trial. This exchange of evidence prior to trial may also explain the predominantly large proportion of cases disposed of by summary judgment, without going to trial for further fact-finding. A third explanation has to do with the nature of the central issue in experimental exclusion litigation-the objective safety and effectiveness of the technology. This type of validation requires hard, objective data from a large number of trials or observations that a personal expertise or experience cannot replace. An expert witness, even if present in the courtroom, can94 hardly take a role as more than an "interpreter" of scientific data. 2. The Treating Physician Rule The courts can also rely on the treating physician's opinion to determine the safety and effectiveness of technology. In certain contexts of health care, the treating physician's opinion is sometimes accorded extra weight, creating what is called the "treating physician rule." Originally, the Courts of Appeals developed the treating physician rule as a federal common law in determining disability claims under the Social Security Act. 95 Under the treating physician rule, the treating physician's determination of disability will be given more 92. I ran a second search within the 519 cases from the initial search of this study and found only 3 cases in which admissibility of expert witness testimony was debated. The search string used for this nested search was "(((702 703)/6 (expert evid! rule)) ((admiss! inadmiss! admit! exclud! preclud! strike stricken unqualifl disqualifl qualifl bar barred barring) /8 (expert /3 (witness testi! affidavit))))" was adopted from LLOYD DIXON AND BRIAN GILL, CHANGES INTHE STANDARDS FOR ADMITTING EXPERT EVIDENCE IN FEDERAL CIVIL CASES SINCE THE DAUBERT DECISION (2001). Since many of the decisions ruling on the admissibility of an expert witness testimony may not have been captured by WestLaw, it is difficult to tell how often the admissibility of expert witness testimony is challenged in experimental litigation setting. 93. Under ERISA, group health plans are required to provide beneficiaries "a reasonable opportunity for a full and fair review" of adverse benefit determination. 29 C.F.R. § 2560.503-1(h). In addition ERISA group health plans are prohibited from requiring more than two appeals before the beneficiary can bring a civil action. 29 C.F.R. § 2560.503-1(c)(2)). Medicare also establishes appeal process. 42 U.S.C. § 1395ff(b). Under some state laws, a health insurance beneficiary is also entitled to a state-mandated appeal procedure. 94. Eddy, supra note 89, at 393-94 ("Thus the expert is not there to introduce a different kind of evidence that is non-empirical, but to introduce the conclusions learned from the empirical evidence."). 95. Black & Decker Disability Plan v. Nord, 538 U.S. 822, 829 (2003). UNIVERSITY OF CINCINNATI LAW REVIEW [VOL. 83 weight than the opinions of others who have not treated the subject patient. The rule was later formally codified into the federal regulation. 96 For disability determination under the ERISA, however, the Supreme Court specifically denied that the physician treating rule applies, as ERISA-in contrast to the Social Security Act-has not legislatively adopted the treating physician rule.97 For medical necessity determination 98 under Medicare, court opinions are split. Some federal courts show deference to the treating physician's opinion of medical necessity, 99 while others do not. 100 Still others concede that the answer is unclear. 10 1 For medical necessity determination under the ERISA, the 02 treating physician rule has been denied by some federal district courts.' The experimental exclusion presents a slightly different flavor.' 0 3 It is debatable whether the treating physician should receive any deference at all as to the scientific validity (i.e., safety and effectiveness) of an emerging technology that the insurer has classified as experimental. Validation of technology is different from determining disability or medical necessity. The correct diagnosis of the patient is not disputed in an experimental exclusion case. The dispute is about the benefit of a new technology, given the diagnosis of the patient. Assessment of the safety, effectiveness and the overall benefit of a new technology is made in isolation from the patient, mostly based on objective clinical trials data found in medical literature. The advantage that the treating physician may have over third-party physicians (i.e., knowledge about the individual patient) should play little role in determining the benefit of the technology at issue. In the cases examined in this study, one court explicitly rejected the applicability of the treating physician rule.'0 4 Only two out of the sixtythree opinions examined here accorded substantial weight on the 96. 20 C.F.R. §§ 404.1527(c)(2), 416.927(c)(2). 97. Nord, 538 U.S. at 830-31. 98. For example, the medical necessity of certain treatment for a given disease. 99. E.g., Executive Dir. of Office of Vermont Health Access ex rel. Carey v. Sebelius, 698 F. Supp. 2d 436 (D.Vt. 2010). 100. E.g., Almyv. Sebelius, 749 F. Supp. 2d 315,332 (D. Md. 2010). 101. E.g., Precision Therapeutics v. Sebelius, no. 09-451, 2010 WL 3924323 (W.D. Pa. June 25, 2010). 102. E.g., Cummings v. Union Sec. Ins. Co., No. 1:06cv667WJG-JMR, 2008 WL 410644 (S.D. Miss. Feb. 11, 2008); Jon N. v. Blue Cross Blue Shield of Mass., 684 F. Supp. 2d 190, 20304 (D. Mass. 2010). 103. Almost without exception, the treating physician recommended the technology at issue in the cases in this study. Otherwise, the patient would not have been introduced to the technology in the first place. Emerson v. Medical Mutual of Ohio, 2004 WL 1635604, 2004-Ohio-3892 (Ohio App. July 23, 2004) is an exception in which the patient proactively sought the treatment, although the primary physician did not recommend it or make a referral. 104. Riley v. Blue Cross & Blue Shield of Miss., No. 3:09CV674HTW-LRA, 2011 WL 2976926 (S.D. Miss. July 21, 2011). 2014] RECALIBRATING EXPERIMENTAL TREATMENT EXCLUSION 195 treating physician's opinion against other evidence.' 0 5 The insignificant role of the treating physician's opinion seen in the cases here represents another change from the HDC/ABMT litigation from a decade ago, in which the treating physician was usually the plaintiff's strongest witness,' 0 6 and, in some cases, received more deference than a renowned 10 7 academic. D. Study Limitations The WestLaw-available opinions are only a fraction of all written opinions, which represent only a portion of all disputes arising out of the experimental exclusion. 108 The cases examined in this study, while carefully coded and analyzed, represent only a small proportion of the entire universe of similar litigation-or the proverbial tip of the iceberg. 109 It is an unavoidable limitation of empirical studies based on opinions available on a commercial database. 110 In its defense, this study was not meant to explore the entire iceberg to begin with. It was designed to test the prevailing perception in literature that health insurers have difficulty defending their determination to deny coverage of experimental technology." '1 There is a bigger caveat. There is no guarantee that the data obtained from litigation records at least constitute an accurate representation of the experimental technologies being used in the medical field. For example, in the current data set, there are five cases of manipulation under anesthesia (MUA) and four cases of autism therapies, out of a total of fifty-seven unique cases. They do not come across as cuttingedge or breakthrough like HDC/ABMT for breast cancer, prone to controversy and confrontation between patient and insurer. That makes one wonder if these technologies were also proportionately abundant in the real world as well, or if these technologies were the kind of 105. Dubaich v. Connecticut General Life Ins. Co., No. CV 11-10570 DMG (AJWx), 2013 WL 3946108 (C.D. Cal. July 31, 2013). 106. Peter D. Jacobson & Stefanie A. Doebler, We Were All Sold a Bill of Goods: Litigating the Science of Breast Cancer Treatment, 52 WAYNE L. REV. 43, 66 (2006). 107. Id. at 72-73. 108. One study observed a falling proportion of federal civil cases ending by trial from 11.5% in 1962 to 1.8% in 2002. The same study also noted similar decline in the state courts as well. Marc Galanter, The Vanishing Trial: An Examination of Trials and Related Matters in Federal and State Courts, I J. OF EMPIRICAL LEGAL STUD. 459 (2004). 109. Peter Siegelman & John J. Donohue, Studying the Iceberg from Its Tip: A Comparison of Published and Unpublished Employment Discrimination Cases, 24 LAW & SOc'Y REV. 1133, 1146 (1990) (explaining the stages of judicial process from dispute to published opinion). 110. W.M. Sage, Judicial Opinions Involving Health Insurance Coverage: Trompe L'Oeil or Window on the World, 31 IND. L. REV. 49, 61-68 (1998). 111. See supra note 25. UNIVERSITY OF CINCINNATI LAW REVIEW [VOL. 83 technologies that simply happen to be taken to the court more often. That question is beyond the scope of this study, as electronically available court opinions are not sufficient to answer it. E. RecalibratingExperimentalExclusion Litigation The findings from the empirical analyses show that the results are quite different from the HDC/ABMT "battle" era and portray a different litigation landscape than scholars have presumed. The analyses of the WestLaw-available opinions suggest that third-party payers do not face particularly harsh review by the courts. Instead, the reasoning provided by the courts in the small number of cases in which they rejected the insurer's experimental exclusion tend to indicate that the courts' focus has shifted towards the evidence of the technology's scientific validity rather than on interpretational weakness. Courts now place less deference on expert testimony or a treating physician's opinion, creating a different litigation environment from the HDC/ABMT era of a decade ago. Indeed, the changes from the HDC/ABMT litigation appear both quantitative and qualitative-the courts reject the insurer's experimental exclusion sparingly and for more sophisticated reasons. What could have brought about the changes? The HDC/ABMT litigation in the late 1980s through the early 2000s may have catalyzed a shift in experimental treatment exclusion litigation by heightening the awareness of the policy implications of experimental exclusion litigation in patients, insurers, courts, and society. 1 2 Courts are more focused on the validity of scientific evidence than before. Moreover, experimental exclusion, the purpose of which is to exclude treatments without proven benefit, is functioning closer to its design than in the HDC/ABMT years. IV. ROLE OF EXPERIMENTAL EXCLUSION IN NEW TECHNOLOGY, INSURANCE COVERAGE & HEALTH CARE SPENDING TRIAD A. New Grievance Process Requirement under the PatientProtection and Affordable CareAct Prior to the promulgation of the Affordable Care Act, some states did not require internal appeal or external review of health insurance coverage denial." 3 Even in states that mandated internal appeal and 112. A possible alternative explanation is that the courts have maintained a similar stance all along, but the HDC/ABMT cases were just outliers. 113. E.g., North Dakota, South Dakota, Alabama, Mississippi and Nebraska had such requirement. Phil Galewitz, New Rules GuaranteePatients' Right to Appeal Insurance Claim Denials, KAISER HEALTH NEWS (Jul 22, 2010), www.kaiserhealthnews.org/stories/2010/july/22/insurance- 2014] RECALIBRATING EXPERIMENTAL TREATMENT EXCLUSION 197 external review, certain types of insurances-individual plans and church-sponsored plans, for example-were often excluded. 114 The Affordable Care Act amended certain provisions of the Public Health Service Act (PHS Act) relating to health insurance, and § 2719 of PHS Act and the regulations implementing it will make internal appeal and external review available to all types of health insurance.," 5 First, these regulations require that health insurers implement an effective internal claims and appeals process, with slightly different requirements between group insurance and individual insurance.1" 6 In addition, unless the state in which the health insurer operates legislates its own external review requirements meeting certain minimum standards, 117 a federal requirement for independent external review will apply directly to the insurer. 118 The minimum standards that the state's external review legislation must meet are largely based on the National Association of Insurance Commissioners Uniform Health Carrier External Review Model Act (NAIC Uniform Model Act). 119 Experimental or investigational exclusion is one of the types of claim denial eligible for 120 external review. Section Ten of the NAIC Uniform Model Act,' 2 1 which forms the minimum standards for state legislation, has very detailed and elaborate processes for external review of coverage denial on account of experimental exclusion. For example, the treating physician has to certify that: (a) standard health care services or treatments have not been effective in improving the condition of the patient; (b) standard health care services or treatments are not medically appropriate-for the patient; or (c) there is no available standard health care service or treatment denials.aspx. 114. Interim Final Rules for Group Health Plans and Health Insurance Issuers Relating to Internal Claims and Appeals and External Review Processes Under the Patient Protection and Affordable Care Act, 75 Fed. Reg. 43330, 43336. 115. Id.; Group Health Plans and Health Insurance Issuers: Rules Relating to Internal Claims and Appeals and External Review Processes, 76 Fed. Reg. 37208. 116. Interim Final Rules for Group Health Plans and Health Insurance Issuers Relating to Internal Claims and Appeals and External Review Processes Under the Patient Protection and Affordable Care Act, 75 Fed. Reg. 43330, 43332-34. 117. Id. at43334. 118. Initial review by the Department of Health and Human Services concluded that not all State external review processes provide the minimum consumer protection of the NAIC Uniform Model Act, leaving the possibility for the federal version to become applicable in those states. Id. at 43336. 119. Id. at 43335. Based on the comments received after the initial regulations, the minimum requirements for consumer protection have been narrowed in 2011. Group Health Plans and Health Insurance Issuers: Rules Relating to Internal Claims and Appeals and External Review Processes, 76 Fed. Reg. 37208, 37216. 120. Id. at 37229, 37231, 37233. 121. NAIC Uniform Model Act http://www.dol.gov/ebsa/pdf/extemalreviewmodelact.pdf. § 10 (2010), available at UNIVERSITY OF CINCINNATI LAW REVIEW [VOL. 83 covered by the health carrier that is more beneficial than the recommended or requested health care service. 122 Treating physicians must also certify that scientifically valid studies using accepted protocols demonstrate that the experimental treatment is likely to be more beneficial to the 23patient than any available standard health care services or treatments. As the new federal regulations under the Affordable Care Act require internal appeal and external review processes, there likely will be an increase of coverage disputes. If the health insurers and policy holders become more aware of the recent trends in court that this study has shown, their strategy going into the grievance procedure prior to litigation may shift. If health insurers have underestimated their likelihood of prevailing in litigation, as has been the perception in scholarship, then they will be more willing to litigate going forward. B. Big Picture: ExperimentalExclusion & Health Care Spending Thus far, this article has focused on the mechanism through which health insurance excludes drugs, devices or procedures without proven safety and effectiveness. Now this article will explore whether and how health insurance impacts the development and adoption of health technology. Some believe that increased health insurance enrollment may have a positive correlation with the development and utilization of technology. 124 Consumers will find new technology more affordable because they only have to pay a portion of the cost. Because new technology will be more affordable to the consumers, suppliers of new technology will be motivated to develop new technologies. On the other hand, the correlation between health insurance enrollment and new technology may trim out to be negative. Large health insurers may dampen technological development and25utilization, using their monopolistic powers to cut reimbursement price. 1 In which direction will increased availability of health insurance impact development and diffusion of technology in the aggregate? The question requires empirical research, little of which is available. 126 Regardless of the answer to that empirical question, one thing is clear: health insurance can affect development and adoption of technology in 122. Id. § 1O.B.(c). 123. Id. § 10.B.(d). 124. Burton A. Weisbrod, The Health Care Quadrilemma:An Essay on Technological Change, Insurance, Quality of Care,and Cost Containment, 29 J. ECON. LITERATURE 523, 534-41 (1991). 125. Joan Costa-Font et al., The "Weisbrod Quadrilemma" Revisited: Insurance Incentives on New Health Technologies, 37 GENEVA PAPERS ON RISK & INS. ISSUES & PRAC. 678, 679-80 (2012). 126, Id. 2014] RECALIBRATING EXPERIMENTAL TREATMENT EXCLUSION 199 one direction or another. Anecdotal experience supports that prediction. For example, when percutaneous transluminal coronary angioplasty (PTCA) was invented in the late 1970s, it was initially assigned to the same reimbursement code as surgical coronary angioplasty by Medicare, although PTCA was less costly to perform than surgery. 127 This extra financial incentive for PTCA may have made the procedure more attractive to health providers and aided in the rapid adoption of the procedure. 128 Cochlear implant succumbed to the opposite fate. It was regarded as a promising innovation for people with hearing disability when it obtained FDA approval in 1984. However, due to its below-cost reimbursement pricing under Medicare, few patients ended up receiving the transplant, until the manufacturer stopped its production 29 altogether.1 Technology, in turn, is connected with a bigger issue of today: health care spending. The soaring health care expenditure in the U.S., which took up 17.2% of its gross domestic product (GDP) in 2012,130 will be at an unsustainable level if it continues to rise. Of the factors that have contributed to the increase in health care spending over the last few decades, technology-related changes in medical practice are considered to be very significant-believed by some to be the single biggest factor. 131 Why is the U.S. so receptive toward new technologies? Desire for better care, peer pressure, reputation, and pecuniary interest have been cited as the driving forces behind health providers' decision to adopt new technologies at an early stage.' 32 Another important factor is the third-party payers' reimbursement policy. The empirical data in this study have not shown any collective tendency of the courts to frustrate the substantive standards for experimental treatments that health insurers established, by creating "judge-made" insurance. How does this empirical analysis that can be 127. Ethan A. Halm & Annetine C. Gelijns, An Introduction to the Changing Economics of Technological Innovation in Medicine, in THE CHANGING ECONOMICS OF MEDICAL TECHNOLOGY 1, 11 (Institute of Medicine ed., 1991). 128. N. T. Kouchoukos, Percutaneous Transluminal Coronary Angioplasty: A Surgeon's View, 72 CIRCULATION 1144, 1146 (1985). 129. Nancy M. Kane and Paul D. Manoukian, The Effect of the Medicare Prospective Payment System on the Adoption of New Technology, 321 NEW ENG. J. MED. 1378 (1989). 130. Anne B. Martin et al., National Health Spending In 2012: Rate Of Health Spending Growth Remained Low For The Fourth Consecutive Year, 33 HEALTH AFF. 67 (2014). 131. CBO, through literature review, estimated that roughly half of the health care cost increase was attributable to technological change. CONG. BUDGET OFFICE, TECHNOLOGICAL CHANGE AND THE GROWTH OF HEALTH CARE SPENDING (2008). Smith et al. attributed 27-48% of health spending growth since 1960 to medical technology. Sheila Smith et al., Income, Insurance, And Technology: Why Does Health Spending Outpace Economic Growth?, 28 HEALTH AFF. 1276, 1281 (2009). 132. John B. McKinlay, From "Promising Report" to "Standard Procedure ": Seven Stages in the Career of a Medical Innovation, 59 MILBANK Q. 374, 381-382 (1981). UNIVERSITY OF CINCINNATI LAW REVIEW [VOL. 83 summarized as "reserved reaction from the courts" relate to the function of health insurance coverage as a "spigot" of health technology that will consequently impact health care spending? First, it opens up the possibility to redefine the "experimental exclusion." As the new federal regulations on external review of experimental exclusion-as well as section ten of the NAIC Uniform Model Act that the regulations refer to--concern only procedural rules, the substantive standards for "experimental technology" are up to each insurer or health plan administrator. 133 For example, "wide acceptance in the medical community' 134 as negating condition for experimental treatment may be re-considered, as it can be at odds with objective validation of safety and effectiveness. The weight to be accorded on the opinion of the treating physician can also be specified in the individual health insurance or health plan, so as to minimize dispute. Regardless of whether the experimental exclusion is heightened or relaxed, making objective data an integral part of the determination can have the salutary effect of facilitating more evidence-gathering in medical practice and technology development. Second, the empirical results of this study should remove preconception about how new efforts to make efficient use of technology will be reviewed by the courts. There is nothing in the empirical data that suggests judicial review adversely foreshadows the possible implementation 1of other evidence-based metrics, such as 35 clinical practice guideline. Cost-effectiveness may warrant re-analysis. Despite the economic appeal of tying insurance coverage to the cost-effectiveness of technology, it has not been generally introduced in the U.S. Among the cited hurdles to the implementation of cost-effectiveness in the U.S. is the fear of litigation and the courts' possible distaste for coverage limitation that involve any calculation of cost. 136 However, the outcome from the empirical analysis of experimental litigation can be interpreted such that the courts do not foreclose consideration of evidence-based metrics that can be used to limit health insurance coverage. Instead, the outcome of this analysis urges health insurers and policy makers to re133. Some state laws may require coverage of certain technologies. For example, Massachusetts requires health insurers to cover off-label use of certain H1V/AlDS drugs. MASS. GEN. LAWS ch. 175, § 47P. 134. See supra text accompanying notes 84-87. 135. Arnold J. Rosoff, Evidence-Based Medicine and the Law: The Courts Confront Clinical Practice Guidelines, 26 J. HEALTH POL. POL'Y & L. 327,353 (2001). 136. See Stirling Bryan et al., Has the Time Come for Cost-Effectiveness Analysis in US Health Care?, 4 HEALTH ECONOMICS, POL'Y & L. 425, 437-38 (2009) (finding from survey that litigation risk is one of the big reasons health insurers hesitate to introduce cost-effectiveness into their coverage decisions). 2014] RECALIBRATING EXPERIMENTAL TREATMENT EXCLUSION 201 assess their pessimism towards the judicial fate of evidence-based coverage determination and to re-engage in the discussion. V. CONCLUSION With the roll-out of the individual shared responsibility payment (so called "individual mandate") under the Affordable Care Act in March of 2014, 137 as well as the implementation of the external review requirement, it is likely that more disputes will arise out of experimental exclusion in the coming years. It makes an opportune time to reexamine the disputes arising out of experimental exclusion, and this article refutes the popular belief that the courts tend to favor policy holders in experimental exclusion litigation. Experimental exclusion functions to ensure validation of health technology with evidence, which in turn will affect the development and adoption of technology, directly impacting total health care expenditures. If the court does not act as a bottleneck to such functioning as far as experimental exclusion is concerned, then health insurers and policy makers may need to recalibrate their timidity towards other evidence-based metrics of coverage determination as well and open up the discussion more actively. 137. 26 U.S.C. § 5000A(b). i !