Assessing the Actuarial Basis for Health-Related Underwriting in Medical and Disability Insurance
By Larry Kirsch 1
Larry Kirsch is a health economist and managing partner of IMR Health Economics, LLC, a national consulting firm focusing on problems pertaining to the health insurance sector and consumer protection.
Introduction
No two tasks are more fundamental to the health and disability insurance businesses than the evaluation of risk and the pricing of risk protection. The principal problems of risk from the insurer’s standpoint are competitive and financial: on the competition side, the insurer faces the possibility of enrolling a disproportionate share of people who are or will become severely disabled and therefore costly to insure.2 On the financial front, the insurer runs the risk of underestimating the cost of claims filed by its policyholders, thereby jeopardizing its margins and courting the possibility of insolvency. To mitigate these potential threats, insurers work to identify risk-related characteristics of insurance applicants and policyholders and to manage enrollment in a disciplined fashion.3
Health-related risk characteristics such as current and past medical conditions and utilization experience are thought to be well correlated with expected claims’ costs and are thus commonly used by insurance companies in the medical underwriting process to define eligibility criteria, separate policyholders into risk-related categories and set premiums corresponding to risk levels. 4 5
This Issue Brief will be of particular interest to attorneys and other legal advocates working in the fields of insurance and disability law.6 It is intended as a primer on the complex subject of medical underwriting and risk classification practices in health and disability insurance. Because the application of risk-related underwriting and classification guidelines almost always results in some policyholders enjoying greater access to insurance coverage, more comprehensive benefits and/or lower premiums than others, they engender concerns about basic fairness and discrimination.
Prohibitions against unfair discrimination in underwriting and classification are written into various federal and state statutes and regulations pertaining to health and disability insurance including the Americans with Disabilities Act (“ADA”) and state unfair insurance practices acts. In this Issue Brief, special attention will be paid to arguments that an insurer’s medical underwriting and risk classification practices trigger the Safe Harbor provisions of the Americans With Disabilities Act, 42 U.S.C. § 12201(c) and provide a complete defense against claims of unfair discrimination. The Safe Harbor provisions of the ADA require, among other things, evidence of consistency with state laws and ultimately some type of actuarial justification. 7,8 So far, in ADA Safe Harbor litigation, policyholders have generally tried to address the relatively easy question: has the insurance company provided any actuarial justification at all or have its underwriting and risk-classification decisions been implemented without any objective and consistently defined actuarial standards? See, e.g. Doe v. Mutual of Omaha, 179 F.3d 557 (7th Cir. 1999). In Chabner v. United of Omaha, 225 F.3d. 1042, (9th Cir. 2000), the Ninth Circuit affirmed (on state law grounds only) a district court’s ruling that an insurance company had assigned the plaintiff a mortality rating that was neither actuarially sound nor based on actual experience. The exaggerated mortality rating resulted in a premium assessment that was substantially more disadvantageous to the plaintiff, who suffered from a rare form of muscular dystrophy, than it would have been to a non-disabled person.9
In the future, one can reasonably expect insurers to begin introducing actuarial evidence at trial as a predicate for claiming Safe Harbor protection. At that point the courts will be called upon to weigh the sufficiency of the insurer’s claims and the decisive question will inevitably turn on whether the hurdle for demonstrating actuarial justification is set too high or too low. This Issue Brief will provide attorneys engaged in litigating insurance discrimination claims information relevant to defining the issues and critically evaluating technical actuarial justifications. 10
Functions and Impact of Underwriting and Risk Classification
Disability-based underwriting and risk classifications are employed by insurance companies issuing coverage in both the individual and group segments of the market. Although a carrier may and generally will apply its underwriting guidelines and risk classification plans differently in the several business segments it serves, the consistent purpose is to help insurers make eligibility, coverage and rating decisions and ipso facto to safeguard their competitiveness and solvency. 11,12
Carriers use underwriting and risk classification to help guide their enrollment, coverage and pricing decisions. In general, carriers apply underwriting and risk classification algorithms as part of their decisions whether to (a) reject (or accept) applicants; (b) write comprehensive vs. limited coverage for applicants; (c) exclude some or all covered benefits for varying durations of time; (d) manage benefits liberally or more restrictively via utilization rules and conditions or financial terms, and/or (e) apply a range of risk-related classifications and rate schedules.13 Because each carrier is free to establish its own underwriting guidelines in keeping with its particular philosophy and business objectives, underwriting determinations tend to be subjective and inconsistent on an industry-wide basis. 14
From the policyholder’s perspective, the impact of an insurer’s underwriting and classification determinations can be profound. Medical underwriting can render an individual uninsurable in the private insurance market or it can drastically limit the scope and depth of benefits an insurer is willing to offer. At a somewhat more subtle level, the underwriting process assigns people to progressively higher risk-graded premium categories, thereby potentially contributing to problems of affordability.
Problems of Unfair Discrimination in Underwriting and Classification
While policyholder challenges to discriminatory practices in private health insurance have more often than not been unavailing under interpretations of the ADA, various courts have ruled that federal as well as state laws prohibit insurers from discriminating unfairly between policyholders-- either explicitly on the basis of disability or in more general terms.15 Certain federal circuit courts have recently construed Title III of the ADA to extend to insurance policies that discriminate on the basis of disability. See, e.g., Carparts Distribution Center, Inc. v. Automotive Wholesaler’s Ass’n of New England, 37 F. 3d 12, 19-20 (1st Cir. 1994) (holding that Title III provides a cause of action against insurance companies that restrict access to their policies on the basis of disability, and suggesting that Title III may also extend to the substantive terms of the policies); and Palozzi v. Allstate Life Ins. Co., 198 F.3d 28 (2nd Cir. 1999) (holding that Title III regulates the sale of insurance policies by insurance companies, subject to the limitations of the ADA safe harbor provision). And, the National Association of Insurance Commissioners (NAIC) Model Unfair Insurance Trade Practice Act (adopted widely by the states) provides, in part, that “making or permitting any unfair discrimination between individuals of the same class and of essentially the same hazard in the amount of premium, policy fees, or rates...or contract...or in the benefits payable thereunder, or in any of the terms and conditions...or in any other manner, whatever” is impermissible. (See, for instance, Mass. Gen’l Laws Ann. c. 176D, §3(7)).
In this context, an unfair discrimination will arise if individuals or groups who are at equal risk of loss (i.e. have equivalent expected claims’ costs over the life of the policy) are charged different premiums or are permitted access to more limited coverages, etc. Federal and state laws require that policyholders who are at equal risk of loss (based on actual or reasonably expected claims experience or “sound” actuarial and underwriting principles) be treated consistently. 16
In contrast, persons whose underwriting factors place them at some elevated risk of loss may be treated differently in terms of premiums and coverages so long as these disparities are cost justified, non- invidious and do not operate as a subterfuge. This has been termed “actuarial fairness”. Those in the insurance industry have long argued that actuarially fair pricing and coverage are imperative in order to prevent adverse risk selection leading, ultimately, to the collapse of insurance markets. And courts have upheld challenges to state regulations that would prohibit insurers from denying coverage or charging differential premiums based on the results of HIV testing, finding that screening for HIV status is consistent with “sound underwriting practices.” Health Ins. Ass’n v. Corcoran, 551 NYS 615, 618 (AD 3 Dept. 1990) (denying coverage or charging a differential premium based on “sound underwriting practices” is not “by nature . . . prejudicial to policyholders”); Life Ins. Assoc. of Mass. v. Comm’r of Ins., 403 Mass. 410, 416 (1988) (“It is not seriously denied that persons who have HIV antibodies, as a group, are at greater risk of illness and have shorter life expectancies than those who do not have HIV antibodies”). 17
Litigating Claims of Unfair Discrimination in Underwriting and Risk Classification
At least one recent federal court ruling, Fletcher v. Tufts University and Metropolitan Life Ins. Co., 367 F.Supp.2d 99, 113-14 (2005) has found that under the ADA, the insurance company making a disability-based distinction assumes the burden of proving its actuarial justification, either through a demonstration of actual or anticipated cost experience or sound actuarial principles. See also the EEOC’s “Interim Enforcement Guidance on the application of the Americans with Disabilities Act of 1990 to disability-based distinctions in employer provided health insurance” (EEOC Notice No. 915.002, June 8, 1993).18 Further, if the insurer wishes to avail itself of the ADA’s Safe Harbor protection, it must prove that its classifications are not inconsistent with state law. 42 U.S.C.S. 12201(c). Compare Palozzi, 198 F.3d at 36 (holding that the plaintiff has the burden of proving that an insurance practice is inconsistent with state law or is being used as a subterfuge to evade Titles I and III, but that she is not required to plead absence of actuarial justification unless state law so requires).
How will insurance carriers most likely try to demonstrate the fairness and reasonableness of their risk classifications and underwriting guidelines pursuant to the Safe Harbor provisions of the ADA? How much and what kind of actuarial evidence will they need?
The actuarial profession’s most authoritative statement on the subject of underwriting and risk classification in health insurance is set forth in Actuarial Standards of Practice (ASOP) 12 (“Concerning Risk Classification”) first issued by the Actuarial Standards Board (ASB) in 1989 with a revision exposed in 2004.19 Further statements focusing specifically on the individual health market (“Risk Classification in Individually Purchased Voluntary Medical Expense Insurance”) and the disability income and long term care sectors (“Risk Classification in Voluntary Individual Disability Income and Long-Term Care Insurance”) were published in 1999 and 2001, respectively, as issue papers by the American Academy of Actuaries (Academy).
ASOP 12 is a highly condensed synopsis of actuarial principles and practice guidance bearing on the subjects of risk classification and underwriting. It is limited in scope and detail and remains silent on issues which counsel will no doubt wish to explore in greater depth in interpreting anti-discrimination statutes, including the Safe Harbor provisions of the ADA.
In relevant part, ASOP 12 provides actuarial guidance on:
(a) the selection of risk characteristics to be used in classifications: (1) risk characteristics should be objective and related to expected outcomes (e.g. differences in cost), (2) the relationship may be established by a correlation between risk characteristics and outcomes, (3) a causal relationship need not be demonstrated, and (4) state law may constrain the selection of risk characteristics.
(b) sources of information and acceptable methods of demonstrating a relationship between risk characteristics and outcomes include: (1) relevant information from any reliable source including analysis of any available data; the reliability of externally-supplied data falls upon the supplier, (2) expert opinion and clinical experience, (3) inferences without specific empirical demonstration, (4) statistical methods subject to the individual judgment of the actuary.
(c) constraints and limitations: the level of analytic detail provided should be weighed against practical considerations of time, cost and effort. The analysis should be consistent with industry and business practices.
In terms of supporting empirical evidence, industry practice varies. Some insurers mine their own databases for information about risk characteristics and claims’ costs. Depending on their size and sophistication, these insurers may be able to capture relationships between risk and cost which are most relevant to the population(s) subject to classification. Other insurers probably rely more extensively on proprietary underwriting guidelines generated by actuarial consulting firms. One actuarial consulting firm, The Milliman Company, offers both Small Group and Individual Medical Underwriting Guidelines covering 700 medical conditions based on longitudinal cost data derived from a “stable” 400,000- member population. Milliman assigns “debit points” to reflect their estimate of the relative costliness of different sets of risk factors. For instance, males, age 45 with medical condition X and no prior hospitalization would probably be assigned more debit points if they were smokers than if they were not. 20 Another actuarial firm, Reden and Anders, offers an alternative approach, methodology and database for predicting the relative costliness of diagnostic-related risk groups. Its model is used by insurers to calibrate renewal rates in the small group health insurance market; it can make comparisons between small groups and against a benchmark standard. 21
In applying ASOP 12 and pertinent actuarial principles to an insurer’s underwriting and risk classification guidelines, both industry and policyholders’ counsel would do well to carefully consider the following kinds of evaluative questions: 22
(a) Are the elements of the classification (e.g. medical condition, current treatments, history of prior treatment) sufficiently correlated with the relevant outcome measure (e.g. an elevated level of cost) for all members of the defined policyholder class? And, to what should costs be compared? What is the appropriate benchmark comparison group or class?
(b) Is any element of the classification superfluous? Does each element make a statistically significant contribution to the outcome?
(c) Is the correlation between classification and the outcome (cost) bona fide? Is it clear that the correlation is not an artifact of one or more unmeasured variables (i.e. statistically spurious)? 23
(d) Is it clear that an otherwise lawful classification is not operating as a surrogate for one which is impermissible, e.g. medical condition as a surrogate for race?
(e) Are the data used to establish the correlation between classification and outcome drawn from actuarial cost experience of people who are substantially similar to those subject to the classification?
(f) Are the data sufficient to ensure a high degree of accuracy (actuarial credibility)?
(g) Are members of the class similar and consistent with one another in terms of risk profiles (“homogeneity”)?. Do members of one class have risk/cost profiles which are distinct from members of other classes (“separation”)? 24, 25
(h) Where the insurer’s underwriting action is to decline a category of risks outright, can the insurer demonstrate that excess costs are not transitory and cannot be mitigated with less severe underwriting actions, e.g. a time-limited exclusionary rider?
Discussed in more depth, below, are some of the more important technical and policy issues embedded in the evaluation criteria outlined in ¶ 15. For purposes of exposition, a hypothetical medical underwriting guideline for a fictitious health insurance company will be woven into this narrative as a foil for discussing the evaluation criteria.
A Medical Underwriting Guideline
Consider the Individual Plain Vanilla Medical Indemnity Policy underwritten and sold to individuals (not employer groups) by the Stellar Insurance Company of New Mexico (“Stellar”) through its agent sales force licensed in three southwestern states. Last year, Stellar’s book of Plain Vanilla medical indemnity business generated $180 million in premium income for 20 thousand individual policyholders and their dependents. Plain Vanilla policies are medically-underwritten and require a written health statement with or without additional documentation. Stellar has developed (confidential) medical underwriting guidelines to screen applicant eligibility in the field and the home office; to determine any limitation in the standard coverages offered to the applicant by restricting the scope of benefits, the duration of waiting periods, dollar caps on benefits, exclusion riders and certain other measures, and to slot each policyholder into the appropriate risk-related premium tier. Stellar relies principally on its own claims database to establish risk profiles and underwriting guidelines.
Stellar’s underwriting guideline for depression (Dysthymic Disorder or “DD”) would permit the company:
(a) to enroll an applicant with a diagnosis of depression (DSM 4-300.4) for standard coverage at standard premiums who has never received medication or treatment (outpatient psychotherapy or hospitalization)for DD;
(b) to enroll an applicant with a diagnosis of depression at above-standard premium levels who is not currently in treatment or taking medication and has never been hospitalized for the illness;
(c) to decline an applicant with a history of major depression (single or recurrent episodes) diagnosed within the last ten years.
The Application of Evaluative Criteria to the Stellar Insurance Company’s Depression Underwriting Guideline
- Are Stellar’s Depression Underwriting Guideline and Risk Classification Plans Valid?
The first six evaluative questions (¶ 15 (a)-(f)) really address different aspects of the fundamental problem of validity. The underpinning of a permissible (“actuarially fair”) discrimination is that a sufficient and demonstrable link be established between one or more socially acceptable risk factors and some outcome, in this case, the probability of higher costs for members of the class. 26
If Stellar were to offer statistical proof of a significant association between its depression risk factors and elevated costs (the risk factors being various treatments and the duration since initial diagnosis), it would probably be subject to the same evidentiary standards which have evolved in other types of anti-discrimination litigation.27 The technical issues raised elsewhere-- and which would apply to Stellar-- pertain to the specification of the risk factors included in the statistical test(s) (e.g. multiple regression) to ascertain that they meet all conventional technical requirements, are inclusive, 28 not spurious and not redundant, and that the association between risk factors and outcomes is sufficiently strong. 29
The more industry-specific questions inherent in the insurance example, however, bring into focus a blend of technical and normative concerns.
An important and potentially contentious threshold issue deals with the choice of an appropriate basis for making cost comparisons. If the expected actuarial cost experience of the depression class is deemed to be “elevated”, then a comparative benchmark must be stipulated. One intuitive option would be to compare costs for the depression class against Stellar’s Individual members considered as a whole. By definition this would include individuals rated as “standard risks”, “sub-standard risks” (above- average risk) and “preferred risks” (below-average risk). Plainly this approach has the virtue of administrative simplicity. But attorneys representing policyholders may infer various reasons for opposing a “group average” comparison approach. In the alternative, one could argue that a better comparison might be the depression class vs. “standard risks” on the theory that “elevated cost” should not be judged against a group which included better than average (“preferred”) risks. Yet a third option would be the depression class benchmarked against Stellar’s other “sub-standard” risk categories. The argument might be that it would be unfair to rate the depression group any more stringently than any other high risk class that Stellar had underwritten. Other possibilities abound, and while it is well beyond the scope of this Issue Brief to explore each of them, counsel will want to analyze the implications of various plausible scenarios. 30
A second issue has to do with the specificity of the risk factors defined by Stellar. Stellar’s underwriting guidelines for depression could have been more complex and refined or less so. There is no particular business standard, only allusions to the cost and value of developing and applying incrementally fine risk classes. 31 In the direction of increasing complexity, Stellar’s underwriters might have tried to find an additional statistical association with age, or explored other demographic characteristics, or looked for interactions between the various risk factors. If Stellar had moved in this direction, it is likely that more applicants with depression would have been disqualified or subject to some other form of underwriting action. On the other hand, proponents of refined underwriting would argue that it produces a fairer distribution of premium burdens because costs can be more tightly aligned with expected risks.
A third question pertains to the data used by Stellar to document a relationship between risk and cost. The prevalence of depression in the adult American population is roughly 5%, meaning that one would expect there to be, at most, 1000 Stellar policyholders with the condition. The limited number of cases may give rise to some concern about the actuarial credibility and accuracy of any estimate based exclusively on Stellar’s depression experience. Alternatively, if Stellar chose to supplement its own database with external data (e.g. from one or another actuarial consulting firm) it is likely that the consultants would have created a database derived from employer groups rather than individual accounts. This may create questions of relevance since, in general, the actuarial experience in the employer group sector is demonstrably quite different from the individual sector–- that is, the group subject to Stellar’s underwriting and classification.32 Lastly, if Stellar does rely on external claims’ studies, ASOP 12 places the responsibility for data reliability on the data provider, not on the insurer. As a practical matter, the verification of data accuracy promises to be extremely challenging even without the added complications brought about by a third-party data provider claiming a proprietary interest in its work product.
A closely related industry-specific issue concerns the nature of the proof to be adduced in insurance discrimination cases. In International Brotherhood of Teamsters v. United States, 431 U.S. 324, 339 (1977), an employment discrimination case, the Court stated that “our cases make it unmistakably clear that statistical analyses have served and will continue to serve an important role in cases in which the existence of discrimination is a disputed issue...” To the extent that significant reliance is placed on ASOP 12 in establishing the ground rules for insurance discrimination litigation, the role of statistical analysis is highly uncertain. While ASOP 12 recognizes statistical analysis as one approach for establishing a link between risk factors and costs, it also offers a menu of alternative methodologies. These include inferences without empirical demonstration, expert opinion and clinical experience. And the Standard of Practice leaves it up to the individual actuary to make a choice consistent with business and industry practices and the practical constraints of cost, time and level of effort.
- Does Stellar’s Risk Classification Plan Misclassify Policyholders?
The seventh evaluative question encompasses two important actuarial principles referred to as “homogeneity” and “separation”. 33
Since any risk classification applies with equal force to all members of a class or sub-class, it is important that the membership be as similar and consistent as possible with respect to risk profiles and expected medical costs. If the risk classification is defined very broadly, it is likely that some members of the class will have risk profiles which are quite dissimilar to others. In the example of Stellar’s depression classification, if all persons with depression had been lumped into one broad category, then persons with less severe conditions would be treated the same as those with more serious chronic disease. Whether Stellar’s strategy of creating at least two sub-classes within the overall depression category achieves the objective of homogeneous risk groupings is an empirical question. Obviously creating additional sub-classes might improve homogeneity and could be desirable under certain circumstances. However, it could also be impractical since the data necessary to develop actuarial cost estimates for narrowly defined sub-classes might be difficult to procure.
The second prong relates to the principle of “separation”. Plainly stated, the intent here is to minimize any overlap between the risk and cost profiles of people in two or more risk classes or sub-classes. If a significant fraction of individuals in the never treated DD group had claims’ costs equal to or higher than persons in the not currently treated sub-class, there would be questions about whether the underwriting discrimination was fair.
- Is Stellar’s Underwriting and Risk Classification Minimally Restrictive?
The final evaluative question falls into a zone which some attorneys might view as being discretionary and/or oblique to specific ADA requirements. In effect, it recognizes that medically-impaired populations often incur initial costs in excess of “standard” populations but that this cost differential narrows over time. The term of art for this phenomenon is “underwriting wear-off”. In lieu of classifying an individual as medically uninsurable and denying coverage outright, an insurer has the option of offering coverages having time-limited restrictions. These could be in the nature of waiting periods or exclusionary riders which would allow the policyholder to attain standard coverage at the conclusion of the “wear-off” period.34 This raises a possible question whether the classification and underwriting plan fairly recognizes the actuarial probability that individuals with elevated costs in some initial time period will subsequently have costs which fall into the standard cost level.
Notes
1. The author can be reached via Larry@IMRHealth.com; 503 223 9200). He wishes to acknowledge the generous contributions of Gordon Trapnell, FSA, MAAA, Jennifer Mathis, Esq., and Mark Scherzer, Esq. Most of all, he extends his sincere appreciation to Malick Ghachem, Esq. of the firm of Zalkind Rodriguez Lunt and Duncan, LLP (Boston, MA)for insightful commentary, collegial debate and unstinting assistance with the preparation of this manuscript.
2. This is referred to as “adverse risk selection”. Presumably applicants and policyholders who are at lower risk and therefore able to secure less costly coverage will enroll with or switch to competitors.
3. Insurers also attempt to price coverage adequately, competitively and equitably-- generally defining equity in terms of risk-related premium structures
4. Underwriting is a term of art which comprehends an insurance company’s decision whether or not to issue coverage to a given applicant and which benefits to offer so that premiums reflect the actuary’s estimate of future claims’ liability. Zamora-Quezada v. HealthTexas Medical Group, 34 F.Supp.2d 433, 443 (W.D.Tx 1998).
Risk classification is a process employed by insurers for grouping policyholders having comparable risk characteristics. It is believed that persons with like risk characteristics are apt to incur similar expected levels of medical costs. Individuals in any given risk category are grouped together and are charged a consistent, class-wide premium. In essence, risk classification is used to divide a large pool of diverse policyholders (such as all individual members residing in a given state or region) into smaller, more homogeneous risk categories (e.g. males, ages 45-65, with a primary diagnosis of X and no previous history of Y and no history of tobacco use) in order to define medical eligibility standards and establish consistent, risk-related premium structures.
5. In Iwata v. Intel, 349 F. Supp.2d 135, 150 (D. Mass. 2004) Chief Judge William Young drew a sharp distinction between health insurance and long-term disability coverage. In the LTD insurance case, the cost of income replacement does not vary by disability type and thus actuarial considerations play a far less important role than they do in health insurance.
6. It may also be relevant to attorneys involved in life, long-term care and other branches of insurance.
7. The requirement that an insurance company’s underwriting and classification policies be consistent with state law is stated in Section 501(c) of the ADA. The requirement that underwriting and classification policies be grounded in actuarially sound justifications or in actual claims experience is derived from the legislative history of the ADA, specifically H.R. Rep. No. 485, 101st Cong., 2d Sess. 136-37 (1990); S. Rep. No. 116, 101st Cong., 1st Sess. 84-86 (1989). See Lewis v. Aetna Life Ins. Co., 982 F.Supp. 1158 (ED Va. 1997) revs’d, sub nom. Lewis v. K-Mart Corp., 180 F.3d 166 (4th Cir. 1999); World Ins. Co. v. Branch, 966 F. Supp. 1203 (ND Ga. 1997) vacated 13 NDLR 250 (11th Cir. 1998); Fletcher v. Tufts University and Metropolitan Life Ins. Co. 367 F. Supp. 2d 99 (D. Mass. 2005).
8. Arguments for the application of actuarial justification have also been made in state courts. Typically, however, these cases have been decided on other grounds before reaching the factual question of actuarial justification. In the Matter of Polan v. State of NY Ins. Dep’t, 2004 NY LEXIS 1608 (July 1, 2004) (holding that state insurance law does not prohibit insurers from providing shorter long-term disability benefits to mentally disabled persons relative to physically disabled persons). Most of the federal courts of appeals that have addressed the legality of providing mentally disabled persons shorter long-term disability benefits than are provided to physically disabled persons have found no violation of the ADA. See, e.g., Ford v. Schering-Plough Corp., 145 F.3d 601, 608 (3d Cir. 1998) ("So long as every employee is offered the same plan regardless of that employee's contemporary or future disability status, then no discrimination has occurred even if the plan offers different coverage for various disabilities").
9. In its brief discussion of the lack of an actuarially sound justification for the discriminatory premium, the Ninth Circuit pointed to the contours of the coming debate. “The parties argue at length about the proper definitions of "sound actuarial principles" or "actual and reasonably anticipated experience. For example, they dispute whether an insurance company must base its rating decisions on ‘hard data’ that is specific to each person, or whether it may take into account more generalized estimates of mortality when it lacks specific data. We need not resolve the debate about exactly what can justify a mortality decision as actuarially sound or related to actual and reasonably anticipated experience, for in this case there is no question that United's mortality rating was arbitrarily high.” Chabner, 225 F. 3d at 1052. Though Chabner was decided solely on state law grounds at the appellate level (the Court having found that the ADA applies only to access to an insurance policy, and not to its terms), the evidentiary struggles over actuarial justification that may ensue in other circuits will likely involve some of the same issues as those raised in Chabner.
10. Note: this Issue Brief will not cover the diverse legal issues that arise in the course of ADA insurance discrimination litigation. Instead, it will focus on the analysis of actuarial justifications.
11. State laws vary with respect to the permissibility, scope and content of underwriting and classification. As a general principle, however, in the voluntary group insurance market, underwriting determinations are applied at the group (employer) level whereas in the individual market, applicants are screened individually.
12. Insurers are also responsive to the preferences of employer group clients. One reflection can be found in the recognition of disability characteristics in benefit design. “To accommodate employers’ and employees’ needs and budget constraints, insurers typically offer a shorter term of coverage for mental disabilities than for physical disabilities...Eliminating this practice would cause a large-scale disruption among employees and the insurance industry and would frustrate...the purposes of the ADA”. Brief Amici Curiae of the Health Insurance Association of America, the Equal Employment Advisory Council, the Chamber of Commerce of the United States, and the American Council on Life Insurance in support of Defendants-Appellees. Fennell v. Aetna Life Ins. Co. , posted on www.eeac.org/briefs/fennell.pdf. (Page 43).
13. In general, see, Merlis, Mark, “Fundamentals of Underwriting in the Nongroup Health Insurance Market: Access to Coverage and Options for Reform” National Health Policy Forum Background Paper, April 13, 2005. Posted on http://www.nhpf.org/pdfs_bp/BP_Underwriting_04-13-05.pdf
14. Karen Pollitz and her colleagues at Georgetown University compared the way a number of health insurers serving different locales would have underwritten seven hypothetical applicants presenting a range of risk profiles. Not only were there discrepancies between carriers in the decision of whether or not to accept the applicant and how much coverage to offer; there were commonly 9-11 fold differences in the premiums that would have been charged by the respective insurers. Pollitz, K., R. Sorian and K.Thomas, “How Accessible is Individual Health Insurance for Consumers in Less-Than-Perfect Health?”, Henry J. Kaiser Family Foundation, 2001.
For additional detail and documentation, interested readers may compare the medical underwriting guidelines issued by Unicare Life and Health Insurance Company (now a subsidiary of Wellpoint Health Networks)and Aetna. http://www.gatewaydbu.com/forms/UNICARE/UNICARE%20Underwriting%20Guide.pdf
http://www.aetna.com/producer/data/sbc/field_uw_ind.pdf
15. Mathis, Jennifer, “The ADA’s Application to Insurance Coverage”, (June 2004). Posted on http://www.bazelon.org/issues/disabilityrights/resources/insurance.htm
16. For a discussion, see, Abraham, K. “Efficiency and Fairness in Insurance Risk Classification”, 71 Virginia L. Rev. (April 1985), Jacobi, J. “The Ends of Health Insurance”, 30 U.C. Davis L. Rev. (Winter 1997).
17. However, for principled critiques of the concept of actuarial fairness, see, Austin, R. “The Insurance Classification Controversy”, 131 U. Pa. L. Rev. (January 1983), Worthington, L. “Insurance Classification: Too Important to be Left to Actuaries”, 19 U. Mich. J. L. Reform ( 1986), Light, D. “The Practice and Ethics of Risk-Rated Health Insurance” 267 JAMA 2503 (1992); and Tom Baker, “On the Genealogy of Moral Hazard,” 75 Tex. L. Rev. 237 (1996).
18. Available at http://www.eeoc.gov/policy/docs/health.html. It is worth nothing that although the EEOC’s 1993 Interim Guidance takes the position that disparities between long-term disability benefits for the mentally and physically disabled do not violate the ADA, the EEOC has since taken the opposite position in litigation. See Equal Employment Opportunity Commission vs. Aramark Corporation, Inc., 208 F.3d 266 (D.C. Cir. 2000).
19. Actuarial Standards Board, “Proposed Revision of Actuarial Standard of Practice Number 12: Risk Classification for All Practice Areas”. (September 2004). Posted on www.actuarialstandardsboard.org/pdf/exposure/asop12_exposure.pdf
20. Shreve, Jonathan and van der Heide, M. “Medical Underwriting for Individual Medical Coverages” (2005) (http://www.milliman.com/pubs/health_newsletter_winter_05_final.pdf)
21. Anderson, L. And Minnich, J. “Small Group Renewal Rating Using the Ingenix Predictive Model” (2001) http://www.reden-anders.com/pdf/PredictiveModelingWhitePaper.pdf
22. This list is meant to be suggestive. In most instances, there will be no correct “yes/no” answer; instead there will be a range of possible answers.
23. In Goldman v. Standard Ins. Co., 341 F.3d 1023, 1034-36 (9th Cir. 2003), the Ninth Circuit found that actuarial and medical evidence offered by the plaintiffs was sufficient to create an issue of material fact whether various unmeasured variables provided a better explanation of outcomes than those measured by the insurance company.
24. Shayer, N. “Driver Classification in Automobile Insurance” in Massachusetts Division of Insurance, Automobile Insurance Risk Classification: Equity and Accuracy, 1978.
25. Michigan Insurance Bureau, Final Decision in the Matter of Golden Rule Ins. Co., Petitioner, No. 90-10830-R (1993).
26. “A relationship between a risk characteristic and an expected outcome, such as cost, is demonstrated if it can be shown that the variation in actual or reasonably anticipated experience correlates to the risk characteristic.” Exposure Draft, Proposed Revision of ASOP 12 “Risk Classification for All Practice Areas”, March 2005, § 3.2.1.
27. See, e.g., Hazelwood School District v. United States, 433 U.S. 299 (1977); Sobel v. Yeshiva University, 839 F.2d 18 (2nd Cir. 1988).
28. See note 25 above.
29. For a more detailed discussion of statistical discrimination testing in the courtroom, see, Feinberg, S., “The Increasing Sophistication of Statistical Assessments as Evidence in Discrimination Litigation”, 77 Am. Stat. A. J. 784 (1982), Rubinfeld, D. “Econometrics in the Courtroom”, 85 Colum. L. Rev.1048 (1985).
30. See, Informal Legal Opinion of the NY State Insurance Department re NY Insurance Law Section 4224 (a)(1) Unfair Discrimination Among Members of the Same Class (December 13, 2000), available at http://www.ins.state.ny.us/rg012132.htm]. In relevant part the Department defined the term “class” extremely broadly to encompass “all applicants for and insureds covered under all individual form life insurance policies”. The Department also stated that while the term “class” may be broadly construed “in appropriate circumstances”, the “plain meaning of ‘same class’ as used by Insurance Law is actuarially similar in terms of morbidity and mortality”. Citing, Health Ins. Ass’n v. Corcoran, 140 Misc.2d 255, modified and aff’d, 154 A.D.2d 61 (1990).
31. See, Chander, S. “Insurance Regulation,” available at http://encyclo.findlaw.com/5700book.pdf.
32. Hadley, J. and Reschovsky, J. “Health and the Cost of Nongroup Insurance”, 40 Inquiry 235 (Fall 2003).
33. Shayer, N. “Driver Classification in Automobile Insurance” in Massachusetts Division of Insurance, Automobile Insurance Risk Classification: Equity and Accuracy, 1978.
34. Shreve, Jonathan and van der Heide, M. “Medical Underwriting for Individual Medical Coverages”, available at http://www.milliman.com/pubs/health_newsletter_winter_05_final.pdf. |
