Evaluation of Clinical Compendia Used for Medicare Part D Coverage Determinations for Off-label Prescribing in Dermatology

Key Points

Question

Are compendia used for Medicare Part D to make coverage determinations adequate for ensuring that patients have access to necessary off-label dermatology treatments?

Findings

In this cross-sectional study, 73 of 238 (31%) evidence-based treatments were included in either compendium evaluated. Of the 22 diseases evaluated, 10 (45%) had 1 or fewer treatments included in the DRUGDEX Information System and 15 (68%) had 1 or fewer treatments included in the American Hospital Formulary Service Drug Information compendium.

Meaning

The findings suggest that these compendia are inadequate for determining coverage for off-label dermatology treatments and that policies to reduce the reliance on the compendia to ensure that patients can access treatments for their disease should be developed.

Abstract

Importance

When making coverage determinations for off-label prescribing, Medicare Part D recognizes 2 compendia: the American Hospital Formulary Service (AHFS) Drug Information and the DRUGDEX Information System. Deficiencies in the accuracy and completeness of these compendia could result in coverage denials for necessary, effective, evidence-based treatments.

Objective

To evaluate these compendia for dermatologic conditions, with a focus on less common conditions that often require systemic treatment.

Design, Setting, and Participants

This cross-sectional study was conducted from July 1, 2018, through September 30, 2018. To identify diseases for which dermatologists may often need access to off-label systemic treatments, a list of 22 chronic, noninfectious, nonneoplastic diseases with at least 4 systemic therapies (including 1 in the first-line therapies and less than 25% approved by the US Food and Drug Administration) were selected for evaluation. With use of Treatment of Skin Disease, 5th Edition, a list of first-, second-, and third-line medications was created, including the level of evidence for each disease. A search of AHFS and DRUGDEX compendia was performed to evaluate for inclusion of the evidence-based therapies. In addition, the references cited in the compendia to justify inclusion of the therapy were examined qualitatively.

Main Outcomes and Measures

Percentage of treatment options included in each compendium, stratified by level of evidence and position on the therapeutic ladder. Concordance between the 2 compendia was assessed using Cohen κ.

Results

Overall, 73 of 238 treatments (30.7%) evaluated were included in either compendium. Among individual diseases, 10 of 22 (45%) had 1 or fewer treatments included in the DRUGDEX compendium and 15 of 22 (68%) had 1 or fewer treatments included in the AHFS compendium. Discrepancies in which a medication was included in one compendium but not in the other compendium occurred for 53 of the 238 medications (22.3%) evaluated. Literature use did not follow a discernible pattern and was often based on decades-old sources.

Conclusions and Relevance

The findings suggest that treatment options listed in these compendia are incomplete, outdated, idiosyncratic, and unpredictable. To ensure that patients can access treatments for their disease, it appears that policies to reduce the reliance on these compendia for coverage determinations should be developed.

This cross-sectional study examines the 2 clinical compendia used for Medicare part D to evaluate the inclusion of evidence-based off-label uses of treatments for dermatologic diseases.

Introduction

Medicare Part D uses drug formularies as an administrative tool to give patients access to medications and constrain costs at the same time. Although individual plans may have different formularies and prior authorization requirements, the Centers for Medicare & Medicaid Services limits the type of medications that can be considered for patients not undergoing oncology treatment strictly to those that have Food and Drug Administration (FDA) approval or are otherwise listed by indication in 1 of 2 compendia: the American Hospital Formulary Service (AHFS) Drug Information or the DRUGDEX Information System.¹ Medicare Part D sponsors are required to adhere to these medically accepted indications and use prior authorization for drugs with the highest likelihood of uses not covered by Medicare Part D. Non–FDA-approved drugs that are prescribed outside of the compendia’s guidelines must be denied without exception.

Although these formularies may be designed with the best intentions, incomplete or inadequate formulary construction could prevent patients from having access to necessary, evidence-based treatments, leading to worse outcomes for patients.^2,3 This issue is particularly relevant in dermatology, where off-label use is common because dermatologists manage more than 3000 unique conditions, some of which are devastating but rare.⁴ Although off-label use of prescriptions is seen for common disease, evidence and FDA approvals for rare, refractory, or severe diseases are even more limited.⁵ Lack of access to medications owing to deficiencies in these compendia could harm patients who are the most ill and at risk for adverse outcomes. The purpose of this study was to examine the adequacy of commonly used compendia for dermatologic conditions, with a focus on less common conditions with few or no FDA-approved medications.

Methods

This study was conducted from July 1, 2018, through September 30, 2018. To compile a list of evidence-based treatments, a search of Treatment of Skin Disease, 5th Edition, was performed.⁶ This text, last updated in 2017, lists treatments for common dermatologic conditions that are categorized into first-, second-, and third-line agents. The evidence for each treatment is categorized as follows: A, double-blind study; B, clinical trial with 20 or more participants; C, clinical trial with fewer than 20 participants; D, series with 5 or more participants; and E, individual case reports with published experiences of fewer than 5 cases. This study was deemed to be exempt from review by the Institutional Review Board of the John H. Stroger, Jr, Hospital of Cook County under category 4 because it involved collection and study of existing publicly available data.

One of the authors (J.S.B.) manually evaluated disease entries in Treatment of Skin Disease, 5th Edition⁶ to select disease categories. This text includes 259 chapters, each on a separate disease process. To identify dermatologic conditions for which there is consensus around off-label prescribing, the following inclusion criteria were specified. Diseases had to be defined entities that are chronic, noninfectious, nonneoplastic, not a wound (eg, stasis ulcers), and not a recognized genetic disease. The disease should have at least 4 systemic therapies (including 1 in the first-line therapies and <25% approved by the FDA) to ensure that non–FDA-approved medications are critical for real-world management. Dermatologic diseases that are a manifestation of another systemic disease process (eg, pretibial myxedema) were excluded. The chapter for each of these diseases was examined in Treatment of Skin Disease, 5th Edition,⁶ and a list of first-, second-, and third-line medications was extracted including the level of evidence for each disease.

Subsequently, a search of the 2 compendia recognized by Medicare, the AHFS and the DRUGDEX Information System, was performed by 2 of the authors (J.S.B. and K.S.) to evaluate for inclusion of identified therapies. In addition, data on the references cited in these compendia to justify inclusion of the therapy were abstracted. These references were qualitatively examined to the point of saturation with a focus on the type of evidence cited (eg, case report, retrospective case series, or randomized clinical trial). For cases in which the evidence cited did not include the treatments being evaluated, a search of MEDLINE was performed to compare the citations that were abstracted from the compendia with the relevant literature published on the disease.

Statistical Analysis

Data are presented using descriptive statistics as appropriate. Concordance between the 2 compendia was assessed using Cohen κ. Statistical analyses were performed using Stata, version 15 (StataCorp).

Results

Inclusion of Evidence-Based Therapies in Compendia

We identified 22 of 259 diseases (8.5%) listed in Treatment of Skin Disease, 5th Edition⁶ for inclusion, encompassing connective tissue and autoimmune disease (n = 8), blistering disorders (n = 5), papulosquamous disorders (n = 2), granulomatous diseases (n = 2), inflammatory dermatoses (n = 3), and neutrophilic dermatoses (n = 2) (eTable in the Supplement). Six of the 22 dermatologic diseases (27%) evaluated were not included as a disease entity in the compendium.

Overall, 73 of 238 evidence-based treatments (30.7%) were included in either compendium. Twenty of 55 first-line therapies (31%), 27 of 87 second-line therapies (36%), and 26 of 96 third-line therapies (27%) were included in either compendium. Eighteen of 32 therapies (56%) with evidence graded as level A, 3 of 30 therapies (10%) with evidence graded as level B, 5 of 43 therapies (12%) with evidence graded as level C, 8 of 61 of therapies (13%) with evidence graded as level D, and 0 of 72 therapies with evidence graded as level E were included in either compendium. In general, fewer medications were included in the AHFS compendium than in DRUGDEX (Table 1).

Table 1. Inclusion of Dermatologic Treatments in Compendia Accepted by Medicare by Therapeutic Ladder Position and Evidence Grade.

Category	Disease Included as Indication, No./Total No. (%)
	DRUGDEX	AHFS	Eithera	Discrepancyb
Therapeutic ladder
First line	16/55 (29)	8/55 (15)	20/55 (36)	16/55 (29)
Second line	25/87 (29)	8/87 (9)	27/87 (31)	21/87 (24)
Third line	23/96 (24)	13/96 (14)	26/96 (27)	16/96 (17)
Evidence gradec
A	15/32 (47)	10/32 (31)	18/32 (56)	11/32 (34)
B	2/30 (7)	2/30 (7)	3/30 (10)	2/30 (7)
C	5/43 (12)	1/43 (2)	5/43 (12)	4/43 (9)
D	6/61 (10)	8/61 (13)	8/61 (13)	2/61 (3)
E	0/72 (0)	0/72 (0)	0/72 (0)	0/72 (0)
Overall	64/238 (26.9)	29/238 (12.2)	73/238 (30.7)	53/238 (22.3)

Abbreviation: AHFS, American Hospital Formulary Service Drug Information.

^aMedication was included in at least 1 of the compendium evaluated.

^bMedication was included in only 1 of the 2 compendia evaluated.

^cEvidence grades are categorized as A, double-blind study; B, clinical trial with 20 or more participants; C, clinical trial with fewer than 20 participants; D, series with 5 or more participants; and E, individual case reports amounting to published experience of fewer than 5 cases.

Among individual diseases, 10 of 22 diseases (45%) evaluated had 1 or fewer treatments included in the DRUGDEX compendium, and 15 of 22 (68%) had 1 or fewer treatments included in the AHFS compendium. Neither compendium had any treatments listed for autoimmune progesterone dermatitis, eosinophilic fasciitis, lichen myxedematosus, linear IgA bullous dermatosis, or folliculitis decalvans. No treatments were included for pityriasis rubra pilaris and palmoplantar pustulosis in DRUGDEX, and no treatments were included for relapsing polychondritis, epidermolysis bullosa, mucous membrane pemphigoid, prurigo pigmentosa, granuloma annulare, and hidradenitis suppurativa in AHFS (Table 2).

Table 2. Inclusion of Dermatologic Treatments in Compendia Accepted by Medicare, by Diagnosis.

Diagnosis	Disease Included as Indication, No./No. (%)
	DRUGDEX	AHFS	Eithera
Connective tissue and autoimmune diseases
Autoimmune progesterone dermatitis	0/4 (0)	0/4 (0)	0/4 (0)
Cutaneous lupus or systemic lupus erythematosus	9/14 (64)	8/14 (57)	12/14 (86)
Dermatomyositis	4/16 (25)	2/16 (13)	4/16 (25)
Eosinophilic fasciitis	0/11 (0)	0/11 (0)	0/11 (0)
Raynaud disease	6/12 (50)	2/12 (17)	6/12 (50)
Relapsing polychondritis	1/10 (10)	0/10 (0)	1/10 (10)
Lichen myxedematosus	0/7 (0)	0/7 (0)	0/7 (0)
Scleroderma or systemic sclerosis	5/13 (38)	1/13 (8)	6/16 (46)
Blistering disorders
Bullous pemphigoid	5/14 (36)	1/14 (7)	6/14 (43)
Epidermolysis bullosa acquisita	2/7 (29)	0/7 (0)	2/7 (29)
Linear IgA bullous dermatosis	0/11 (0)	0/11 (0)	0/11 (0)
Mucous membrane pemphigoid	6/12 (50)	0/12 (0)	6/12 (50)
Pemphigus vulgaris	4/6 (67)	3/6 (50)	4/6 (67)
Papulosquamous disorders
Pityriasis rubra pilaris	0/10 (0)	1/10 (10)	1/10 (10)
Prurigo pigmentosa	1/6 (17)	0/6 (0)	1/6 (17)
Granulomatous diseases
Granuloma annulare	2/15 (13)	0/15 (0)	2/15 (13)
Sarcoidosis	5/16 (31)	4/16 (25)	6/16 (38)
Inflammatory dermatoses
Folliculitis decalvans	0/7 (0)	0/7 (0)	0/7 (0)
Hidradenitis suppurativa	1/8 (13)	0/8 (0)	1/8 (13)
Palmoplantar pustulosis	0/5 (0)	1/5 (20)	1/5 (20)
Neutrophilic dermatoses
Pyoderma gangrenosum	6/16 (38)	3/16 (19)	7/16 (44)
Behçet syndrome	7/18 (39)	3/18 (17)	7/18 (39)

Abbreviation: AHFS, American Hospital Formulary Service Drug Information.

^aMedication was included in at least 1 of the 2 compendia evaluated.

Comparison of Medications Included in the Compendia

Discrepancies in which a medication was included in one compendium but not in the other compendium occurred for 53 of 238 medications (22.3%) evaluated. Discrepancies were found for 16 of 55 first-line therapies (29%), 21 of 87 second-line therapies (24%), and 16 of 96 third-line therapies (17%). Discrepancies were found for 11 of 32 therapies (34%) with evidence graded as level A, 2 of 30 therapies (7%) with evidence graded as level B, 4 of 43 therapies (9%) with evidence graded as level C, 2 of 61 therapies (3%) with evidence graded as level D, and 0 of 72 therapies with evidence graded as level E included in either compendium (Table 1). Cohen κ was 0.31 (95% CI, 0.20-0.43) for concordance of medications included between compendia.

Qualitative Assessment of Evidence Cited in Compendia

In the qualitative assessment of the literature cited by these compendia as justification for inclusion of medications, there was not a clear pattern with respect to the level of evidence that was required or how often references were updated. Evidence ranging from single-patient case reports to personal communication from pharmaceutical companies (eg, a discussion with a Celgene representative was cited as evidence that thalidomide is effective for pyoderma gangrenosum) and randomized clinical trials were cited to support these indications. Some indications were supported by single references, although others had multiple studies cited to justify their inclusion. Frequently, the literature cited was decades old, with some citations to literature published before the 1980s. Other times, more recent references were included, such as those for rituximab for pemphigus vulgaris. In addition, many of these references could not be evaluated because for some indications, AHFS does not list references before 1984 in its online version.

For many of the treatments evaluated, the cited references in the compendia were incomplete or outdated. For instance, although both DRUGDEX and AHFS include cyclophosphamide as a treatment for systemic lupus erythematosus, only DRUGDEX includes mycophenolate mofetil despite the publication of a large randomized clinical trial that showed that mycophenolate mofetil is more effective and safer than cyclophosphamide for the treatment of lupus nephritis.⁷ Similarly, neither DRUGDEX nor AHFS includes tetracycline-class antibiotics (eg, doxycycline) for the treatment of bullous pemphigoid despite the publication of a large randomized clinical trial showing that doxycycline was noninferior to standard treatment with oral prednisolone.⁸ In addition, although the inclusion of isotretinoin for pityriasis rubra pilaris in AHFS is supported by references published between 1976 and 1984, literature from the 1960s supporting the use of methotrexate for pityriasis rubra pilaris is not included.^{9,10,11,12,13,14,15} Newer literature, such as evidence supporting the use of acitretin since 2000 and tumor necrosis factor inhibitors since 2003, has also not been added for this disease.^{16,17,18,19,20} Similar inconsistencies and deficiencies were noted for other evaluated diseases.

Discussion

Although compendia are an administratively convenient approach to determine medical indications for coverage decisions, fewer than one-third of dermatologic treatments evaluated in this study were included in either compendium. For several of the diseases examined, none of the treatment options evaluated were included in either compendium, and for many others a single medication was included. These shortcomings mean that patients with rare but treatable diseases may not be able to access necessary, evidence-based therapies when these compendia are used to make coverage determinations.

The inconsistent and incomplete nature of these compendia is highlighted further by discrepancies between the DRUGDEX and AHFS in drugs listed. For therapies with a level A category of evidence, which was defined as a double-blind clinical trial, there were discrepancies found for 34% of medications evaluated. Correspondingly, the evidence cited by these compendia in support of their decisions was often inconsistent and poorly justified. Results from randomized clinical trials were ignored, although citations from single-patient case reports and even personal communication with pharmaceutical companies were included. In addition, it was not uncommon for first-line therapies with a high evidence grade (level A or B) to be missing from the compendia and for second- or third-line therapies with a lower evidence grade to be included. In an era of increasing efforts to promote evidence-based medicine, the seemingly random inclusion and exclusion of medications in these compendia is not aligned with this goal.²¹

The net result of this system is the abdication of treatment choice from physicians and insurers to unregulated and unaccountable third parties. Improved transparency around compendia development and update cycles is necessary to clarify decision making. The AHFS reports that it is “prepared by a professional editorial and analytical staff, who critically evaluate published evidence on the drug,” and DRUGDEX states that it is reviewed by “international experts”; however, the specific composition of these reviewers is unclear.^22,23 Although AHFS has defined criteria for appraising the strength and quality of evidence, multiple therapies evaluated with supporting literature that would be defined as high strength and quality by their guidelines were not included in this compendium.²⁴ Neither compendium clearly outlines the frequency with which it is updated or the search process to identify the evidence base to be appraised. This approach leads to significant opacity around decision making for specific therapies and limits the ability of clinicians to understand the rationale for why a therapy may be excluded and of the medical community to address discrepancies or omissions in these compendia.

Of note, many cost-effective therapies were excluded from these compendia, thus limiting access to their use. Antimalarial agents such as hydroxychloroquine are broadly effective, inexpensive, and safe medications for the treatment of connective tissue disease such as cutaneous lupus,²⁵ but both hydroxychloroquine and chloroquine were absent from DRUGDEX. The lack of inclusion of these cost-effective and established therapies is likely to inadvertently shift prescribing practices toward costlier FDA-approved on-label treatments because they are easier to access. For instance, none of the 7 off-label treatments evaluated for hidradenitis were included in either compendium, but adalimumab, which is FDA approved for this indication, was included in DRUGDEX. If clinicians are unable to access cost-effective off-label treatments because they are not included in the compendia, they may be forced to prescribe costlier FDA-approved treatments that may not be preferred by either the patient or clinician, ultimately resulting in worse outcomes and increased costs to the patient and health system.

There are several potential approaches to address the shortcomings of these compendia. One option would be to develop compendia specific to the unique challenges of managing dermatologic illnesses, as has been done in oncology.³ However, given that dermatologists manage over 3000 diagnoses, it may be impossible or impractical to develop an accurate and up-to-date compendium specific to dermatology.⁴ Another potential option to address the inadequacies of these compendia would be to require Medicare Part D to consider evidence from the literature presented by clinicians during the prior authorization process to support appeals of coverage determinations. Alternatively, an expert panel could be developed to review appeals for therapies not included in the compendia. Given the inherent complexity and diversity of dermatologic disease, it is likely that there must always be an option to use supplementary evidence to support necessary treatments for patients with rare diseases and special conditions. Any reasonable solution should be built on the core principles of transparency, physician and patient input, and a process for proposing updates and challenging current decisions.

Limitations

The results of this study must be interpreted in the context of the study design. Although the definition of evidence-based treatments can be debated, all of the medications that we evaluated are used in daily practice and accepted by consensus to be reasonable treatments for dermatologic disease. By using a systematic approach to categorize therapies according to level of evidence and location on the therapeutic ladder, we attempted to define the quality of support for the treatments evaluated. Owing to a lack of evidence-based literature available, this study could not examine extremely rare diseases, for which there may be even fewer options included in these compendia.

Conclusions

The findings suggest that compendia recognized by Medicare for making coverage determinations have significant inadequacies with respect to inclusion of common, first-line, evidence-based treatments for dermatologic illnesses. We found significant variation between compendia, suggesting a lack of quality and accuracy. To ensure that patients can access treatments for their disease, it appears that policies to reduce the reliance on these compendia or supplemental compendia to capture the unique needs of patients with dermatologic illness should be developed.

Supplement.

eTable. Inclusion of Dermatologic Treatments in Compendia Accepted by Medicare

eReferences