Enrollee Appeals of Preservice Coverage Denials at 2 Health Maintenance Organizations FREE

Procedural protections have emerged as the keystone of reforms aimed at safeguarding the interests of patients in managed care.^{1 - 7} The perennial worry, poignantly dramatized in recent films such as John Q and The Rainmaker, is that financial incentives lead managed care organizations to refuse coverage for needed care.⁸ One policy response has been to mandate coverage of specific services, such as obstetrical care,^{2 - 3} infertility treatments,⁹ and autologous bone marrow transplantation for breast cancer.¹⁰ But most commentators regard wider prescriptions of "medical necessity" as neither possible nor desirable.¹¹ Consequently, recent regulatory interventions designed to protect consumers are primarily concerned with ensuring that the processes used to decide questions of coverage in managed care are fair, prompt, careful, and, most importantly, subject to review.

External review mechanisms now exist in more than 40 states,¹² and the US Supreme Court recently endorsed their legality in a widely publicized decision.¹³ However, the first recourse for most enrollees who are denied coverage is an appeal to the health plan itself. Previous research suggests that health maintenance organizations (HMOs) across the country adjudicate more than 250 000 such appeals annually for their privately insured enrollees alone.¹⁴ Recognizing the pivotal importance of plan-based appeals systems in protecting patients, policy makers nationwide have moved to bolster them.^{2 - 3 ,15 - 18} But despite the tremendous reliance being placed on appeals, virtually nothing is publicly known about their characteristics or outcomes.

We reviewed the administrative files associated with 3519 appeals lodged by privately insured enrollees at 2 large HMOs. Our study goals were to improve understanding of the sources of conflict between patients and managed care organizations, and to provide data to help shape effective policies for protecting patients.

Two of the largest HMOs in the country shared their data on appeals of coverage denials. Both participating plans (hereafter "plan 1" and "plan 2") are based in California, insure several million enrollees, and operate well-established appeals systems. They also deliver care through a "delegated" model in which the plans transfer authority and substantial risk for utilization review to several hundred medical groups with which they contract to provide services to their enrollees.^{19 - 20}

Approximately half of managed care enrollees nationwide receive care through this type of delivery model.^{21 - 23} However, responsibility for appeals functions generally resides with the plan²⁴ ; in California, state law mandates this. Hence, the appeals consist of plans' reconsiderations of benefits denials made by medical groups, with the medical groups customarily bearing responsibility for the costs of any services approved in the appeal.

The structures of the appeals systems operated by each plan are similar to those previously reported in surveys of HMOs.^{24 - 25} We have described them in detail elsewhere.¹⁴ In summary, enrollees who were formally denied coverage by their medical group received a letter notifying them of the denial and of their opportunity to appeal it to their plan. Plan 1 offered 3 levels of appeal and plan 2 offered 2 levels. At plan 1, enrollees who lost their appeal at level 1 could request another reconsideration (level 2) and, if the denial stood, one final reconsideration (level 3). At plan 2, enrollees who lost their appeal at level 1 had it automatically reconsidered at level 2 whenever "medically reviewable" issues were involved; for all other types of appeal there was only 1 level of review.

Nonphysician case managers handled tasks associated with the first level of appeal at both plans, including obtaining medical records and other information relevant to the denial from the treating clinician or medical group, although they were instructed to collaborate with a medical director when dealing with appeals that presented clinical issues. Level 2 at both plans triggered committee review, which included 1 or more medical directors. Level 3 reconsiderations at plan 1 consisted of review by 1 or 2 medical directors who had not previously been involved in adjudication of the appeal.

At plan 1, we randomly sampled 2491 (19%) of all 13 033 appeals lodged by privately insured enrollees between January 1, 1998, and December 31, 1999; at plan 2, we randomly sampled 1500 (67%) of all 2223 appeals lodged between January 1 and June 30, 2000. These time periods were chosen because they were recent intervals during which the structure and operation of the appeals systems at both plans were stable and comparable. The division of the sample sizes across plans balanced resource constraints, an interest in securing reasonable representation from both plans, and the larger number of appeals available at plan 1.

We used a stratified random sampling approach. One set of strata corresponded to the level at which the appeal terminated. At plan 1, we selected 15% (n = 1813) of the appeals that terminated at level 1, 95% (n = 543) of those that terminated at level 2, and all appeals that went to level 3 (n = 135). At plan 2, we selected 65% (n = 1366) of the level 1 appeals and all level 2 appeals (n = 134). In addition, to permit closer investigation of appeals involving particular services, and because existing electronic data at plan 1 provided a reasonable guide to the treatments in dispute, for plan 1 we oversampled appeals involving durable medical equipment (DME), laboratory/diagnostic testing, and surgery.

We derived weights to adjust for the sampling design, and all estimates in this study are weighted to represent the general population of appeals at each plan.

We gathered information from the administrative files associated with each sampled appeal through an explicit review methodology. The "appeals file" is the hard copy repository for all information accumulated by the plan through the life of an appeal that is deemed relevant to judging its outcome. This includes the enrollee's original request for reconsideration (a letter or written summary of a telephone call), copies of plan-medical group and plan-enrollee correspondence, and, where relevant and applicable, medical records, minutes of committee review meetings, and the evaluations of case managers, medical directors, and medical experts.

To extract detailed information about each appeal from the file, we designed an appeal file abstraction form (AFAF). The AFAF was used to collect data on the appeal, including type (preservice/postservice), reason for the medical group denial, the services or equipment in dispute (including Current Procedural Terminology code), diagnosis information (including International Classification of Diseases, Ninth Revision code), sociodemographic characteristics of the enrollee (eg, age, sex), and details of the plan's decision (eg, level, outcome).

We recruited 5 abstractors at plan 1 and 4 abstractors at plan 2 to review the files associated with sampled appeals. The abstractors, who were college undergraduate and graduate students, were trained in 1-day workshops at each site and supplied with a manual on use of the AFAF. Abstraction proceeded June through November 2000 at plan 1 and September through January 2001 at plan 2. Study personnel were on site or available to abstractors by telephone throughout the review periods. To test reliability of the abstraction process, approximately 10% of the sampled files at each plan were independently reviewed by 2 different abstractors.

The study and our confidentiality assurance plans were approved by the institutional review boards at RAND and the Harvard School of Public Health.

We focused exclusively on preservice (or prospective) appeals in this analysis. These are disputes over coverage for services or equipment that enrollees seek, as opposed to disputes over financial responsibility for services already obtained (postservice/retrospective).¹⁴

Prior to analysis we excluded appeals against denials of pharmaceuticals (n = 224) because information on these appeals was incomplete at both plans due to the involvement of external agencies in the adjudication process. We also eliminated appeals filed mistakenly (because no initial denial had occurred) (n = 70), duplicate cases (n = 36), provider-plan disputes that had been misclassified as appeals (n = 17), appeals withdrawn before adjudication at the first level (n = 5), and other disputes that were not actual appeals for miscellaneous reasons (n = 35).

Our analyses are descriptive. We first classified preservice appeals into 4 basic types: (1) out of network, (2) contractual coverage, (3) medical necessity, and (4) administrative issue.

Out-of-network appeals differ conceptually from contractual coverage and medical necessity appeals in having the identity of the provider at issue, rather than access to the service itself; in other words, enrollees in these appeals had rejected the proposed in-network provider and sought care from one unaffiliated with the medical group and/or plan. We classified appeals as contractual if the service at issue required the adjudicator to refer primarily to terms of coverage in the enrollee's insurance policy. Medical necessity appeals were those that involved services for which adjudicators' primary reference point was the prevailing clinical standard of care. We standardized classifications within specific service types based on the dominant tenor of the disputes therein. However, these demarcations are sometimes challenging. The distinction between contractual coverage and medical necessity can be particularly vexing for services such as physical therapy and certain types of DME when coverage is specified in the insurance policy but the circumstances call for a blend of contractual and clinical considerations.

We analyzed out-of-network appeals by the enrollee's stated reason for seeking the care, and contractual coverage and medical necessity appeals by the services in dispute. Within appeal types, we also tested for differences in the proportion of disputes resolved in favor of the enrollee using Pearson χ² tests corrected for the weighted design.²⁶ We performed all analyses using STATA v7.0 (STATA Corp, College Station, Tex).

We completed abstraction of 3519 appeals files, 88% of the sample (83.3% at plan 1 and 96.3% at plan 2). The rest were not locatable. Sixty-one percent (n = 2161) of the files abstracted involved preservice appeals. The exclusions left 1774 preservice appeals, of which 951 (53.6%) came from plan 1 and 823 (46.4%) came from plan 2.

Enrollees or their relatives brought 95% of these appeals. The rest were initiated on behalf of enrollees by treating physicians (4.2%), attorneys (0.2%), and other third-party representatives (1.1%). Most preservice appeals (86.3%) were resolved at the first level; 4% proceeded to the highest level at plan 1 and 1% did so at plan 2.

Approximately 93% of the appeals fell into 1 of 3 major types (Table 1). Approximately one third of appeals (36.9%) were medical necessity disputes, another third (36.6%) were over contractual limits of coverage, and 1 in 5 (19.7%) contested access to out-of-network services. Overall, the plans decided 41.9% of appeals in favor of the enrollee, overturning the medical group's denial. Enrollee wins were significantly more likely among medical necessity appeals (52.2%) than among out-of-network (35.4%) or contractual coverage appeals (33.2%) (P<.001).

There were several notable interplan differences. Out-of-network appeals accounted for a higher proportion of appeals at plan 2 than at plan 1 (26.0% vs 17.9%, P<.001). But when this type of appeal is excluded from comparisons, the proportions of contractual coverage and medical necessity appeals across plans were nearly identical. Outcomes differed markedly by plan, with enrollees winning approximately twice as frequently at plan 2 as at plan 1 in all 3 leading types of appeal. Outcome differences were particularly sharp for appeals over ancillary services and DME, which may stem partly from interplan differences in the scope of covered benefits. However, the contrasting win rates between the plans did not disrupt the overall difference in outcomes by type: medical necessity appeals were significantly more likely than other types of appeal to be decided in favor of the enrollee at both plans.

In the reliability testing, 3 of the 161 files independently reviewed by 2 abstractors differed on outcome and none conflicted directly on service type, although 4 services were not strictly consistent (3 recorded a more general description than the other and 1 recorded a Current Procedural Terminology code where the other did not). We resolved the outcome differences by reference to administrative data at the relevant plan and used the more specific of the service descriptions.

Enrollees' stated reason for seeking care outside their network in 60% of out-of-network appeals was that they believed a specific provider was of superior quality to the one available to them within their network (Table 2). For example, one appeal involved an enrollee's desire to obtain care from an out-of-network sleep specialist instead of the plan physician she was seeing in-plan because she believed the former had more experience with her disorder. Another enrollee sought access to a pediatric ophthalmologist, rather than the general ophthalmologist who was currently treating her son's nasolacrimal duct obstruction. Yet another enrollee was aware of more technologically advanced imaging equipment at an out-of-network facility and sought access to it for follow-up diagnostic work on his epileptic condition.

The other leading motivations for out-of-network appeals were geographic distance to the assigned network provider (10.8%) and enrollees' desire to consult clinicians with whom they were familiar because the preferred clinician had previously treated them or someone they knew (17.2%). There were no significant interplan differences in the frequency of any of these leading reasons.

Four service types—DME, ancillary health services, dental care, and alternative medicine—accounted for 84.4% of contractual coverage appeals (Table 3). Disputes over DME were the most common (47.5%), with foot orthotics (22.2%) being the dominant item. The next most common service types were ancillary services (18.2%), dental care (11.0%), alternative medicine (7.7%), and investigational or experimental treatments (4.7%).

Overall, several types of services showed enrollee win rates that differed significantly from the mean win rate (33.2%) for contractual coverage appeals. Relatively few DME cases were decided in the enrollee's favor (27.1%, P = .02). This result was driven largely by the exceptionally low overturn rate among appeals involving orthotics (14.9%, P<.001). By contrast, enrollees were significantly more likely to win appeals over chiropractic care (63.6%, P = .002) and dental care (54.7%, P = .004).

Three service categories—surgical procedures (29.2%), office consultations with specialists (24.0%), and diagnostic tests (19.7%)—accounted for nearly three quarters of medical necessity appeals (Table 3). Among appeals over surgery, the 3 most common types of operations were gastric bypass (7.3%), breast alteration (5.7%), and procedures to remove varicose veins (3.3%); 57% of appeals over surgery disputed coverage of 1 of these 3 procedures. One in 10 medical necessity appeals involved surgical treatments related to obesity—namely, gastric bypass, gastric stapling, liposuction, and dermolipectomy. Rulings in favor of the enrollee were significantly less likely in disputes over gastric bypass surgery (26.1%, P = .004) than the average medical necessity appeal.

Enrollees disputed coverage for access to a diverse range of specialists, with visits to dermatologists (3.4%), orthopedic surgeons (3.3%), and psychiatrists (2.8%) most frequently at issue. Together these 3 specialties accounted for nearly 40% of all appeals over the medical necessity of specialist consultations. One half of the disputed diagnostic tests were magnetic resonance imaging (4.4%) or dual-energy x-ray absorptiometry/bone density studies (4.4%). Finally, disputes over denied treatment for scars or benign skin lesions were common (12.3%); these were medical or minor surgical procedures sought from dermatologists or plastic surgeons.

Nearly 60% of the preservice appeals we studied contested the contractual scope of coverage or access to out-of-network care rather than issues of medical necessity. Among those appeals that did center on medical necessity, enrollees won approximately half. Inventory of the services involved in preservice appeals showed that both medical necessity and contractual coverage appeals were remarkably concentrated within a fairly small group of services. Cosmetic and nonessential therapies were important epicenters of dispute. The policy questions raised by these types of coverage disputes have tended to escape notice, receding into the background amid preoccupation with the insurance status of treatments with much more dramatic health consequences.^{27 - 29}

Most previous studies of coverage denials in the managed care setting have involved normative analyses of how such determinations should be made.^{11 ,30 - 32} There have been several attempts to use empirical methods to investigate actual decision making,^{33 - 36} but few have ventured into the territory of appeals^{37 - 38} and none, besides Richardson and colleagues' 1993 summary of external review decisions in the Medicare program,³⁹ have undertaken large-scale analysis of plan data. Thus, it should not be surprising that the recent wave of regulation directed at internal review procedures has approached them in a fairly monolithic fashion.

Our findings suggest that a more nuanced understanding of appeals could aid ongoing public and private efforts to protect consumers and inform the design of more effective approaches to dispute resolution. Each of the 3 major types of appeals raises distinct issues.

Restrictions on access to out-of-network care have long been a core component of strategies used by managed care organizations to control costs.^{40 - 41} Our findings highlight the friction that such rules create. They also suggest that a majority of disputes in this area spring from enrollee beliefs about the superior quality of particular outside providers. Leading patients' rights proposals^{2 - 3} aim to lower plan-imposed barriers to specialist care, but the deference to "covered" benefits in such mandates means that they are unlikely to disrupt existing limits on access to out-of-network providers.

With respect to contractual coverage appeals, foot orthotics, speech therapy, physical therapy, dental care, alternative medicine treatments, investigational therapies, and infertility treatments together accounted for 61%. Some of these disputes involve significant health issues and demand a blend of contractual and clinical considerations. But many of them challenge the denial of elective treatments by directly contesting stated exclusions in the insurance product. Appeals systems, whether internal or external, are a peculiar mechanism for resolving such disputes. More careful and concise prespecification of covered benefits would seem the obvious "ounce of cure" here.

However, purchasers' distaste for explicit rationing, together with apprehension about "incomprehensible road maps" of covered benefits, have traditionally chilled interest in a more prescriptive approach to coverage.^{42 - 44} Our findings should reinvigorate this debate. The conglomeration of disputes around relatively few service types that we observed suggests that greater contractual specification of benefits may be both feasible and valuable. Contractual specification is no panacea. External oversight is still critical to ensure that contractual clarity is not used to promote substandard care.⁴⁵ Moreover, clearer statements among plans, purchasers, and enrollees about the coverage limits that apply to commonly disputed services will not eliminate disputes over contractual coverage issues, nor will it end consumer dissatisfaction over noncovered benefits, but it would promote transparency and consumer awareness.

Three features of medical necessity appeals were particularly striking. First, they accounted for less than 40% of appeals at both plans but had quite high rates of overturn. Although these prevalence and merit statistics are clear enough within the appeals population, they do not reveal the extent to which enrollees who experience inappropriate medical necessity denials actually appeal. Similarly, within the appeals system itself, it is unclear whether the paucity of upper-level appeals reflects satisfaction with first-level decisions or enrollees simply being worn down and therefore reluctant to press on with their appeals. Research in other domains shows that significant disconnects may exist between actionable experiences and claiming behavior.^{46 - 48} However, compared with tort lawsuits such as medical malpractice, the relative ease of lodging coverage appeals and reappeals—including no fees, options for telephone initiation, mandatory notification rules, and the absence of a need for representation—suggests that the disconnect may not be as significant.

Second, we again observed a marked concentration of appeals among a small band of services: 5 service types—surgery for obesity or obesity-related conditions, breast alterations, varicose vein removal, bone density and sleep studies, and treatments of scars or benign lesions—accounted for nearly 40% of medical necessity appeals. The relatively homogeneous nature of many of the diagnoses that accompany these treatments suggests some further possibilities for prespecification in coverage contracts.

Third, the medical necessity disputes frequently converged not around life-sustaining therapies, but in areas of ongoing uncertainty about the proper limits of insurance coverage. Sabin and Daniels⁴⁹ have conceptualized "medical necessity" using a treatment/enhancement paradigm; the distinction here is between interventions meant to prevent, cure, or ameliorate impairments, and those that merely improve conditions that are part of normal human functioning. A number of the services that provoke medical necessity disputes—for example, varicose vein removal, liposuction, dermolipectomy, scar treatments, and a variety of cosmetic therapies—appear either to fall into the enhancement category, or else to make particularly "hard" cases for the treatment/enhancement distinction (the same is true in the contractual coverage area, as evidenced by the contentiousness of services like alternative medicine, infertility treatments, and certain types of DME). Thus, to a significant extent, flash points for managed care disputes appear to track areas of societal uncertainty about the legitimate boundaries of insurance coverage.

Debate, education, and better community consensus about the coverage status of services in this "gray zone" should help to avoid both inappropriate denials and conflict. Demystifying medical necessity—whether by greater attention in coverage contracts to the most contentious services or by some other means—may also permit stakeholders to incorporate considerations of clinical efficacy and cost-effectiveness into decisions about covered benefits.^{42 ,50 - 52} Most importantly, it would allow oversight authorities to hone in on the disputes with the greatest potential for harm to enrollees. For example, specific types of medical necessity appeals may require faster review and higher standards than current laws mandate; some may also demand assurances of or special access to external review; others, especially those in the enhancement category, may call for less scrutiny.

Our study is limited by the fact that we did not undertake independent clinical review of the merits of the appeals; hence, we cannot determine the extent to which denials were upheld or overturned inappropriately. Another limitation is that the appeals data come from 2 health plans in California that operate delegated models of care. A substantial proportion of HMO enrollees nationwide receive their care through this type of delivery system, but characteristics of appeals from other types of plans may differ. There was variation in overturn rates even among the 2 plans we studied, and overturn rates generally may be lower in nondelegated models when the plans themselves are at direct financial risk for any coverage approved. The prominence of certain service types, such as cosmetic therapies, may also differ regionally.

Finally, we reiterate the difficulty we faced in some cases in delineating different types of appeals, particularly in making the contractual coverage/medical necessity divide. The distinction sometimes blurs. Recent legal cases contesting the coverage of physical therapy for incurable but treatable illnesses illustrate the potential for decisions that ostensibly fall within contractual coverage and out-of-network categories to be infused with medical necessity considerations.^{53 - 54}

This analysis provides new insights into the plan-based review processes that have come to figure so prominently in patients' rights legislation. Improved knowledge of the content of appeals, particularly the services involved, should help refine public and private efforts to protect managed care enrollees. However, it will also sharpen the focus on difficult questions for society as a whole about where and how the boundaries of insurance coverage should be drawn.