Use of clinical outcome assessments in specialty drug coverage policies

Abstract

BACKGROUND:

Clinical outcome assessments (COAs) are tools widely used in clinical trials to evaluate treatment efficacy by capturing patient experience. However, little is known about how COAs are used in specialty drug coverage decisions.

OBJECTIVE:

To describe the frequency, type, and role of COAs in US commercial health plans’ specialty drug coverage policies.

METHODS:

We analyzed coverage data from the Tufts Medical Center Specialty Drug Evidence and Coverage (SPEC) Database, which tracks specialty drug coverage policies from 18 large US commercial health plans. Researchers reviewed each policy to identify disease-specific COAs, excluding tools assessing objective physiological data or clinical algorithms (eg, biomarkers, risk prediction scores). We categorized COAs by (1) type (ie, patient-reported outcomes [PROs], clinician-reported outcomes [ClinROs], etc), (2) time point (initial approval vs reauthorization), and (3) application (ie, supporting diagnosis, documenting baseline, limiting access for initial criteria; or defining meaningful treatment response for reauthorization criteria).

RESULTS:

As of April 2024, SPEC included 13,933 coverage policies for 440 specialty drugs for 386 indications. These policies incorporated COAs for 169 (43.8%) indications. In total, 2,188 (15.7%) policies referenced at least 1 COA, and of these 763 (34.9%) referenced multiple COAs. Because policies can incorporate COAs into both initial and reauthorization criteria, we identified 4,305 total COA references. Inclusion of COAs varied across plans, ranging from 5% to 27.2% of coverage decisions. COAs appeared 2,077 times in initial criteria, most commonly as ClinROs (69.0%) and infrequently as PROs (5.7%). In initial criteria, plans primarily used COAs to limit access (50.5%), whereas in reauthorization criteria, plans used COAs to determine treatment response.

CONCLUSIONS:

COA inclusion varied across plans, with most using them to limit access. ClinROs were the most frequently used COA type, underscoring the pivotal role of providers on behalf of patients and health plans in shaping treatment access decisions. Plans used PROs relatively infrequently, with some including PROs in their policies more than others, suggesting a missed opportunity to consistently incorporate patients’ voices to complement the assessment of treatment outcomes when determining coverage. Future research should explore the rationale behind COA use and its implications for treatment access.

Plain language summary

Clinical outcome assessments (COAs) help show how well a treatment works from the patient’s point of view. Health plans use COAs in different ways when deciding which specialty drugs to cover. Some plans rely more on clinician-reported outcomes and use patient-reported outcomes less often. This means they might miss important details about how patients feel or what matters most to them. Using COAs more often could lead to better decisions and improve patient care overall.

Implication for managed care pharmacy

Commercial health plans vary in their use of COAs in coverage decisions, most often relying on clinician-reported outcomes while rarely including patient-reported measures. This inconsistency may increase provider burden and limit patient access. Missed opportunities to integrate the patient perspective and reliance on impractical trial-based tools highlight the need for standardization. Emphasizing feasible, patient-centered measures could streamline processes, reduce inequities, and improve outcomes.

To manage budgets while ensuring appropriate access to specialty drugs (typically high-cost or other complex therapies), health plans frequently apply prior authorization criteria through utilization management tools.¹ These include step therapy protocols, clinical eligibility thresholds, and other restrictions that can create administrative burden for providers and patients.^2,3 One understudied aspect of utilization management is the use of clinical outcome assessments (COAs), instruments often designed for use in clinical studies or practice.

COAs are instruments that quantify or qualify patient experience.⁴ Typically administered in clinical trials, COAs evaluate whether a given treatment benefits patients by monitoring its impact on quality of life, symptoms, and daily function.⁵ In November 2021, the US Food and Drug Administration (FDA) published the Center for Drug Evaluation and Research (CDER) COA Compendium, which summarizes how clinical trials have used COAs to assess patient experience and support labeling claims.⁶ For example, the compendium lists the Psoriasis Area and Severity Index and the Physician’s Global Assessment as measures that pivotal clinical studies use to evaluate ustekinumab’s impact on body surface area involvement of a patient’s plaque psoriasis from a clinician’s perspective.⁶

Additionally, the CDER Patient-Focused Drug Development initiative provides guidance for systematically integrating patient perspectives into drug development and evaluation.⁷ As the FDA notes, any activities related to patients’ health and well-being should be patient-centered, meaningfully incorporating their experiences, needs, and priorities.

Although COAs were developed primarily for use in clinical studies, some health plans incorporate them into specialty drug coverage policies, in which they can directly influence patient eligibility.³ For example, a plan’s coverage of biologics for atopic dermatitis may require documentation that a specific percentage of body surface area is affected. Incorporating COAs into coverage criteria can enhance decision-making by providing insights into the patient experience beyond traditional symptom reporting.⁸ Still, COAs represent just one element of the broader evidence base that plans consider when formulating coverage criteria, which also draws on clinical trial data, real-world evidence, economic analyses, and clinical guidelines.^9–11 However, the extent to which COAs affect patients’ ability to receive treatment through health plan coverage remains largely unexplored.

Because health plans use COAs in a variety of ways, it is important to document and quantify the real-world application of these measures.

This study aims to address this gap by examining how US commercial health plans use COAs in specialty drug coverage determinations, specifically, by assessing their frequency, type, and application across coverage policies and across health plans.

Methods

DATA SOURCE

This study uses data from the Tufts Medical Center Specialty Drug Evidence and Coverage (SPEC) Database, which tracks publicly available specialty drug coverage decisions from 18 large US commercial health plans (Supplementary Table 1 ^{(521KB, pdf)} , available in online article). Details of the SPEC database have been described in previous research.^12,13 As of April 2024, the database tracked coverage decisions for 440 drugs and 386 indications, corresponding to 975 drug-indication pairs. Of the 975 drug-indication pairs, 442 (45.3%) are approved through at least 1 expedited review pathway (ie, fast-track approval, priority review, accelerated approval, or breakthrough therapy), 407 (41.7%) are orphan-designated drugs, 197 (20.2%) are approved in pediatric populations, 190 (19.5%) are biosimilars, and 349 (35.8%) are oncology treatments. Since the FDA may approve a drug for multiple indications, eg, FDA approved natalizumab for Crohn disease and multiple sclerosis (MS), the database includes a separate coverage decision for each drug-indication pair. After excluding decisions labeled as “no policy” (no publicly available coverage information was found) or “not covered” (the plan deemed the use of the drug to be experimental/investigational), 13,933 active coverage decisions remained for analysis.

DATA COLLECTION

To assess the frequency and types of COAs in specialty drug coverage decisions, 3 researchers manually reviewed the 13,933 active coverage decisions, divided equally by indication. Each researcher identified disease-specific COAs and categorized them by type using the CDER COA Compendium⁶ and the Mapi Research Trust ePROVIDE COA database.¹⁴ A second researcher then reviewed each file to validate findings and flag any missed COAs. The team resolved discrepancies through consensus discussions with senior staff. The final results were compiled into a master file listing each indication, the associated COAs, and their types. Supplementary Table 2 ^{(521KB, pdf)} lists the indications with the most individual COAs documented in our sample.

Using this master file, each researcher searched 2 SPEC Database text fields within their assigned coverage decisions: (1) initial coverage criteria, which outlines the requirements to begin therapy, and (2) reauthorization criteria, also known as continuation criteria, which outlines plans’ criteria for patients to be eligible to continue therapy. Researchers recorded the number of decisions that included at least 1 identified COA. For each referenced COA, they documented (1) type, (2) time point, and (3) application, as described below.

COA Type

We categorized COAs into 6 groups based on the National Institutes of Health Biomarkers, EndpointS, and other Tools resource.⁵ Patient-reported outcomes (PROs) reflect the patient’s direct report of their health status, symptoms, or treatment effects. Observer-reported outcomes (ObsROs) describe observable signs or behaviors related to a health condition, reported by someone other than the patient or a health professional, such as a caregiver. Clinician-reported outcomes (ClinROs) are based on a health care professional’s clinical assessment. Performance outcomes (PerfOs) involve standardized tasks that the patient actively performs, such as the 6-min walk test.⁵ We also included a composite category for assessments that combine multiple COA types. Finally, we created a “COA unspecified” category for outcomes referenced in coverage policies that do not cite a specific measurement tool or instrument, eg, a requirement stating “improvement in bone and joint pain” for reauthorization of velaglucerase alfa for Gaucher disease.

Because clinical tools assessing objective physiological data or clinical algorithms, such as biomarkers (eg, blood, urine, or genetic tests; physiologic measures; imaging tests) and risk prediction scores based on patient characteristics (eg, 10-year probability of hip fracture), do not meet common definitions of COAs,^4,14 we excluded these instruments from our analysis.

COA Time Point

We categorized COAs based on when plans applied them – either during the initial approval process or at the time of reauthorization.

COA Application

We further categorized COAs included in the initial coverage text into 3 primary applications. First, supporting diagnosis, in which COAs helped confirm a disease diagnosis, eg, requiring a Mayo score of 6 or greater typically indicates moderate to severe ulcerative colitis. Second, documenting baseline, in which COAs established a patient’s initial disease status as a reference for evaluating future changes, eg, such as using the Children’s Hospital of Philadelphia Infant Test of Neuromuscular Disorders to assess baseline motor function before nusinersen treatment. Third, limiting access, in which COAs set disease severity thresholds for eligibility, eg, requiring a Myasthenia Gravis–specific Activities of Daily Living (MG-ADL) score of 6 or greater for access to eculizumab (Supplementary Table 3 ^{(521KB, pdf)} ).

We did not further classify COAs found in reauthorization criteria, as plans uniformly used them to assess treatment response, ie, to determine whether ongoing therapy was justified based on observed outcomes.

A second researcher and a senior staff member reviewed each coverage decision file to confirm that all COAs were correctly identified and documented.

ANALYSES

The unit of analysis for this study was each unique application of a COA within a health plan’s coverage policy. For example, if a policy for efgartigimod for myasthenia gravis referenced the MG-ADL both at initial authorization (to limit access) and at reauthorization (to define treatment response), we counted each reference as 2 separate entries in our dataset.

We conducted 4 descriptive analyses:

Analysis 1: COA Use Across Coverage Decisions and Health Plans

We assessed COA use across all coverage decisions, including (1) the proportion of decisions referencing at least 1 COA; (2) the proportion referencing multiple COAs; and (3) the total number of COAs referenced in initial and reauthorization text. We also examined variation in COA use by health plan.

Analysis 2: COA Use by Drug Characteristics

We described the sample of included drugs and characteristics of drugs associated with COA-referencing coverage decisions, focusing on the following attributes tracked in the SPEC database: orphan drug designation (ie, designated for rare diseases affecting <200,000 people in the United States), oncology indication, age indication, and inclusion in FDA expedited review pathways (ie, drugs approved through ≥1 of the 4 FDA expedited review programs: priority review, fast-track designation, accelerated approval, or breakthrough therapy).³

Analysis 3: How COA Types May Differ When Used for Initial and Reauthorization Coverage Criteria

We analyzed the distribution of COA types (ie, PROs, ClinROs, PerfOs, etc) referenced in initial and reauthorization criteria, then evaluated variation in COA type across health plans.

Analysis 4: How Plans Use COAs in Initial Coverage Determinations

We examined how COA applications (ie, documenting baseline, limiting access, etc) varied in initial coverage texts and summarized findings by health plan.

Case Study

We conducted a case study of the indication with the highest number of individual COAs in our sample, describing variation in COA use, type, and application.

All analyses were conducted using Stata Statistical Software (Release 18) and Microsoft Excel.

Results

As of April 2024, the SPEC database included 13,933 coverage decisions with initial coverage criteria, of which 9,053 (65.0%) included accompanying reauthorization criteria.

COA USE ACROSS COVERAGE DECISIONS AND HEALTH PLANS

Of the included 13,933 coverage decisions, 2,188 (15.7%) included at least 1 COA, whereas the remaining 11,745 (84.3%) did not include any COAs (Figure 1). COA inclusion varied substantially by health plan, with the proportion of decisions within each plan referencing at least 1 COA, ranging from 5.0% to 27.2% (Supplementary Figure 1 ^{(521KB, pdf)} ).

FIGURE 1.

COA Availability in Specialty Drug Coverage Decisions

Excluded “not covered” and “no policy” coverage decisions.

COA = clinical outcome assessment; SPEC = Tufts Medical Center Specialty Drug Evidence and Coverage Database.

Among the 2,188 decisions that included at least 1 COA, 1,425 (65.1%) referenced 1 COA and 763 (34.9%) referenced multiple COAs. Counting each instance of a COA within both initial and reauthorization criteria, we identified a total of 4,305 COA references across all coverage decisions (Figure 1).

COA USE BY DRUG CHARACTERISTICS

Certain drug characteristics were associated with a higher likelihood of COA inclusion in coverage decisions. Among the 2,188 decisions that included at least 1 COA, drugs approved through at least 1 expedited review pathway had the highest rate of COA use (45.2%), followed by orphan-designated drugs (40.4%), drugs approved for pediatric indications (24.7%), and oncology therapies (17.4%) (Table 1).

TABLE 1.

COA Availability by Drug Characteristics

Characteristic	Coverage decisions with COAs (n = 2,188)
≥1 FDA expedited review pathway, n (%)
Yes	989 (45.2)
No	1,199 (54.8)
Orphan designation, n (%)
Yes	885 (40.4)
No	1,303 (59.6)
Age indication, n (%)
Pediatric	540 (24.7)
Adult	1,648 (75.3)
Oncology indication, n (%)
Yes	381 (17.4)
No	1,807 (82.6)
Top 10 indications with decisions that included ≥1 COA, n (%)
Myasthenia gravis (n = 78)	70 (89.7)
Pulmonary arterial hypertension (n = 178)	155 (87.1)
Psoriasis (n = 367)	250 (68.1)
Duchenne muscular dystrophy (n = 82)	52 (63.4)
Neutropenia-myelosuppressive chemotherapy (n = 211)	75 (35.5)
Polyarticular juvenile idiopathic arthritis (n = 206)	53 (25.7)
Multiple sclerosis (n = 283)	64 (22.6)
Rheumatoid arthritis (n = 436)	92 (21.1)
Ulcerative colitis (n = 298)	60 (20.1)
Crohn disease (n = 298)	53 (17.8)

COA = clinical outcome assessment; FDA = US Food and Drug Administration.

COA inclusion also varied by indication. Among the indications with the highest number of COA-referencing coverage decisions, myasthenia gravis (89.7%), pulmonary arterial hypertension (87.1%), psoriasis (68.1%), and duchenne muscular dystrophy (63.4%) showed the highest rates of COA use (Table 1).

HOW COA TYPES MAY DIFFER WHEN USED FOR INITIAL AND REAUTHORIZATION COVERAGE CRITERIA

Across all 4,305 referenced COAs, plans referenced COAs in reauthorization criteria more frequently (2,228, 51.8%) than in initial coverage criteria (2,077, 48.2%) (Figure 2).

FIGURE 2.

COA Types in Initial vs Reauthorization Coverage Criteria

ClinRO = clinician-reported outcome; COA = clinical outcome assessment; ObsRO = observer-reported outcome; PerfO = performance outcome; PRO = patient-reported outcome.

Initial Coverage Criteria

Among the 2,077 COAs referenced in initial coverage criteria, most (69.0%) were ClinROs. Other COA types included composites (10.3%), PerfOs (9.6%), PROs (5.6%), COA unspecified (4.6%), and ObsROs (1.0%) (Figure 2). All 18 health plans most frequently used ClinROs, though the proportion varied substantially, from 42.9% to 90.2% of referenced COAs (Figure 3).

FIGURE 3.

COA Types in Health Plan Initial Coverage Criteria (n = 2,077)

Plans are listed according to magnitude of ClinROs in initial coverage criteria; no specific order is assigned to the included health plans, ie, plan 1 in Supplementary Figure 1 ^{(521KB, pdf)} does not reflect plan 1 in Figure 3.

ClinRO = clinician-reported outcome; COA = clinical outcome assessment; ObsRO = observer-reported outcome; PerfO = performance outcome; PRO = patient-reported outcome.

Reauthorization Coverage Criteria

Among the 2,228 referenced COAs in reauthorization criteria, ClinROs remained the most commonly used type (27.7%), though their share was lower than initial coverage.

Other COA types included COA unspecified (27.0%), composites (22.2%), PROs (12.7%), PerfOs (10.2%), and ObsROs (0.2%) (Figure 2).

The use of ClinROs in reauthorization criteria ranged from 15.4% to 64.8% across plans, whereas the use of COA unspecified ranged from 2.7% to 66.7% (Supplementary Figure 2 ^{(521KB, pdf)} ).

HOW PLANS USE COAs IN INITIAL COVERAGE DETERMINATIONS

Initial Coverage Criteria

Among the 2,077 COAs referenced in initial coverage criteria, plans most often applied COAs to limit access to treatment (50.5%), followed by to support diagnosis of disease (29.9%) and to document baseline disease status (19.6%) (Supplementary Figure 3 ^{(521KB, pdf)} ).

Application patterns varied widely across plans. Use of COAs to limit access ranged from 26.4% to 66.7%, to support diagnosis from 17.7% to 46.3%, and to document baseline from 0.0% to 47.0% (Supplementary Figure 4 ^{(521KB, pdf)} ).

CASE STUDY

To explore COA variation in greater depth, we conducted a case study of MS, the indication with the highest number of distinct COAs in our dataset (n = 13). Focusing on a subset of 20 therapies for relapsing/remitting MS, we identified 55 coverage decisions referencing 1 or more of 12 MS COAs: 2 (16.7%) ClinRO, 2 (16.7%) PROs, 3 (25.0%) COA unspecified, and 5 (41.7%) PerfOs (Table 2). The number of coverage decisions referencing each COA ranged from 2 to 49.

TABLE 2.

Relapsing/Remitting MS Case Study: COA Type, Time Point, and Application

COA	COA type	Number of decisions featuring COAs	COA in initial coverage text, n (%)	COA in reauthorization coverage text, n (%)	COA application in initial coverage
					Limiting access, n (%)	Documenting baseline, n (%)	Supporting diagnosis, n (%)
Expanded Disability Status Score	ClinRO	49	22 (44.9)	27 (55.1)	2 (9.1)	20 (90.9)	0 (0)
Some functional arm/hand use consistent with performing activities of daily living	COA, unspecified	20	20 (100)	0 (0)	20 (100)	0 (0)	0 (0)
Objective measure of walking ability/walking speed	COA, unspecified	18	18 (100)	0 (0)	18 (100)	0 (0)	0 (0)
Multiple Sclerosis Walking Scale	PRO	5	0 (0)	5 (100)	0 (0)	0 (0)	0 (0)
Multiple Sclerosis Functional Composite Score	PerfO	5	0 (0)	5 (100)	0 (0)	0 (0)	0 (0)
6-Minute Walk Test	PerfO	5	0 (0)	5 (100)	0 (0)	0 (0)	0 (0)
Fatigue Symptom and Impact Questionnaire Relapsing Multiple Sclerosis Scale	PRO	5	0 (0)	5 (100)	0 (0)	0 (0)	0 (0)
Stabilization, slowed progression, or improvement in ≥1 symptom, such as motor function, fatigue, vision, bowel/bladder function, spasticity, walking/gait, or pain/numbness/tingling sensation	COA, unspecified	4	0 (0)	4 (100)	0 (0)	0 (0)	0 (0)
New York Heart Association Functional Class	ClinRO	4	4 (100)	0 (0)	4 (100)	0 (0)	0 (0)
Ambulate without aid or rest for ≥5 meters/ambulate without aid or rest for ≥100 meters	PerfO	3	3 (100)	0 (0)	3 (100)	0 (0)	0 (0)
9-Hole Peg Test	PerfO	2	1 (50)	1 (50)	0 (0)	1 (100)	0 (0)
Timed 25-foot Walk Test	PerfO	2	1 (50)	1 (50)	0 (0)	1 (100)	0 (0)
Overall	NA	122	69 (56.6)	53 (43.4)	47 (68.1)	22 (31.9)	0 (0)

We collected 13 individual COAs for MS, but excluded 1 COA from our case study to focus exclusively on drugs indicated for relapsing/remitting MS.

ClinRO = clinician-reported outcome; COA = clinical outcome assessment; MS = multiple sclerosis; NA = not available; ObsRO = observer-reported outcome; PerfO = performance outcome; PRO = patient-reported outcome.

Initial Coverage Criteria

Plans included at least 1 COA in 56.6% of initial coverage criteria for treatments for relapsing/remitting MS. Among these references, 68.1% were used to limit access and 31.9% to document baseline disease status (Table 2). No plans used COAs to support diagnosis. The most commonly referenced COA was the Expanded Disability Status Score (EDSS), a ClinRO used to measure disability. Plans predominantly used the EDSS to document baseline.

Reauthorization Coverage Criteria

Plans referenced at least 1 MS COA in 43.4% of reauthorization criteria, and all were used to assess treatment response (Table 2).

As with initial coverage, the EDSS was the most frequently referenced COA (55.1%) in reauthorization decisions, but here it was used exclusively to evaluate treatment response.

Discussion

US commercial health plans incorporate COAs into a subset of specialty drug coverage decisions (in 16% of coverage decisions), shaping both initial authorization and reauthorization criteria. To our knowledge, this is the first study to systematically examine how commercial health plans use COAs to guide their enrollees’ access to specialty therapies. Our findings have important implications.

First, the use of COAs varied substantially: although all plans in our sample referenced COAs at least once, the frequency of use varied widely, ranging from 5% to 27% of coverage decisions. Our MS case study illustrates the wide variation in COA use. This variation introduces administrative complexity for providers navigating prior authorization processes and trying to align clinical assessments with plan-specific requirements.¹⁵ Health plan requirements may also affect how clinicians practice medicine, influencing how, why, when, and how often they perform clinical assessments of their patients.

Second, our study shows that COA use is often vague. Although fit-for-purpose, disease-specific COAs can help ensure appropriate use of therapies, inconsistent or vague requirements, such as those referencing unspecified COAs, can engender uncertainty and impose barriers to access.¹⁶ Plans also relied more heavily on unspecified COAs and composite COAs in reauthorization criteria, in which COAs were uniformly used to assess treatment response. Additionally, although we identified COAs described in plan policies, providers may use additional COAs in clinical practice that are not reflected in the documentation. Although such flexibility may facilitate individualized decision-making, it also introduces ambiguity in clinical expectations and documentation standards.

Third, ClinROs were the most frequently incorporated COA type in initial criteria (69%), typically used to limit access. In reauthorization criteria, COA types were more evenly distributed, though ClinROs still appeared most often. Compared with ClinROs, health plans rarely included PROs in coverage decisions, a finding that is consistent with prior research highlighting limited integration of patient perspectives in payer decisions.¹⁷ PRO inclusion varied considerably across plans. Some never used PROs, whereas for others, PROs represented 15% of the COAs included in initial coverage criteria. For conditions like major depressive disorder, coverage policies often reference PROs such as the Patient Health Questionnaire-9 or Beck Depression Inventory to assess patient experience. Among the 11 major depressive disorder coverage decisions in our sample that included a COA, 73% included at least 1 PRO. However, in some cases, plans relied solely on ClinROs, potentially overlooking valuable patient input. In contrast, for Crohn disease, an indication in which more reliance on PROs was expected given the subjective nature of its symptoms, 94% of coverage decisions including a COA relied on composite measures, and the remaining 6% used PROs alone. These patterns suggest that health plans recognize the value of integrating patient perspectives, but their integration is inconsistent and indication-specific.

Fourth, plans often used COAs in initial authorization criteria to limit access. Prior research has shown that plans frequently impose restrictions in their coverage policies, often aligning them with the eligibility criteria used in a drug’s FDA pivotal clinical trials.³ Plans may use COAs to reinforce that alignment. For instance, coverage decisions for myasthenia gravis therapies often required the same primary endpoints used in trials: the Myasthenia Gravis Foundation of America Clinical Classification (MGFA) and MG-ADL scores.¹⁸ Among the 70 myasthenia gravis coverage decisions referencing COAs, 53% required both MGFA and MG-ADL, whereas 4% required only 1 of the measures. An additional 43% required both MGFA and MG-ADL along with the Quantitative Myasthenia Gravis Score, a secondary endpoint used in the drugs’ clinical trials.

Finally, these findings raise important questions about the suitability of COAs for use in routine clinical care and prior authorization. Although many COAs are developed for use in clinical trials, they may not always reflect outcomes meaningful to patients or be practical in real-world settings. Some instruments, such as Hamilton Depression Rating Scale (a ClinRO that plans in our sample used for major depressive disorder, treatment-resistant depression, and postpartum depression), require structured interviews lasting up to 30 minutes.¹⁹ Others, such as the Multiple Sclerosis Functional Composite, involve multiple physical and cognitive performance tests and can take up to 1 h depending on the patient’s disability level.²⁰ These time and resource demands may pose significant burdens for patients and clinicians.

Future research should explore why plans incorporate COAs and how these decisions affect care and approval/denial rates for coverage. Studies could examine variation in COA scoring thresholds. It would also be valuable to highlight clinicians’ perspectives on plans’ use of COAs, including their inconsistent use, the administrative effort required for documentation, and how these factors influence access and clinical workflow. Understanding the downstream effects of COA use is essential to ensure that they support, rather than hinder, patient care.

LIMITATIONS

This study has several limitations. First, our analysis was limited to 18 large US commercial health plans and may not reflect practices among smaller health plans or public payers such as Medicaid or Medicare. Second, the findings may not generalize to specialty drugs not included in our sample. Third, some COAs were categorized broadly (eg, COA unspecified), which may obscure nuanced distinctions. Fourth, we limited our sample by excluding risk prediction scores and biomarkers from the included COA types. Finally, the manual review of coverage decisions introduces the potential for misclassification.

Conclusions

We found that commercial health plans used COAs, tools designed for use in clinical trials, for a range of purposes. COAs were most often used to limit access, with ClinROs being the most commonly included COA type. Some plans incorporated PROs more often than others, suggesting a missed opportunity to consistently incorporate patient perspective to complement the clinical data when making coverage decisions. The wide variation in COA use across plans underscores the need for further research to assess their impact on patient access, clinical practice, and patient outcomes.

Disclosures

This research study was supported by Eli Lilly and Company.

Dr Poon is an employee of Eli Lilly and Company and owns stock in the company. Ms Klopchin was an employee of Eli Lilly and Company at the time this work was initiated and still owns stock in the company. Ms Rucker, Ms LaMountain, Ms Lin, Ms Beinfeld, and Dr Chambers have no conflicts of interest to disclose.