Examining US commercial health plans’ use of The Institute for Clinical and Economic Review’s reports in specialty drug coverage decisions

Abstract

BACKGROUND:

The Institute for Clinical and Economic Review (ICER) has emerged in a visible role in US health care. However, it is unclear to what extent US commercial health plans use ICER value assessments in their specialty drug coverage decisions.

OBJECTIVE:

To evaluate the relationship between ICER’s reported cost-effectiveness ratios (CERs) and coverage restrictiveness. Also, to examine the frequency with which plans have cited ICER in their coverage policies and to investigate how frequently health plans adjusted their drug coverage criteria in the 12 months after ICER’s assessments.

METHODS:

We analyzed the Tufts Medical Center Specialty Drug Evidence and Coverage Database, which includes specialty drug coverage decisions issued by 17 large US commercial health plans. For ICER-assessed drugs, we recorded ICER’s estimated CERs in the form of cost per quality-adjusted life-year (QALY) gained. First, we used multivariate logistic regression to examine the association between ICER’s reported CERs and plan coverage restrictiveness, when controlling for other factors that were likely to affect decision-making. Next, we examined how often plans cited ICER’s assessments in coverage decisions issued in years 2017-2020. Lastly, we examined whether plans added or removed coverage restrictions (eg, patient subgroup restrictions or step therapy protocols) in the 12 months following ICER’s assessment.

RESULTS:

Plans tended to cover drugs with higher (less favorable) CERs more restrictively than drugs with CERs less than $100,000 per QALY: odds ratio (OR) = 4.48 if $100,000-$175,000 per QALY; OR = 2.00 if $175,000-$500,000 per QALY; and OR = 2.10 if $500,000 or more per QALY (all P < 0.01). Plans cited ICER in 0.8% (5/622) of coverage policies in 2017, 0.6% (5/833) in 2018, 1.7% (19/1,139) in 2019, and 2.4% (33/1,406) in 2020. For drugs with CERs less than $175,000 per QALY, plans adjusted coverage in 37% of cases: added restrictions in 20%, removed restrictions in 15%, and added one restriction but removed another in 2%. For drugs with CERs of $175,000 or more, plans changed coverage criteria in 29% of cases: added restrictions in 21%, removed restrictions in 5%, and added one restriction but removed another in 4%.

CONCLUSIONS:

We found that when controlling for other factors, health plans’ specialty drug coverage decisions were associated with ICER’s estimated CERs. Plans infrequently cited ICER value assessments. We did not observe a trend for plans more often narrowing coverage criteria for drugs with CERs $175,000 or more compared with drugs with CERs less than $175,000.

Plain language summary

The Institute for Clinical and Economic Review (ICER) has a prominent role in US health care. We found an association between ICER’s value assessments and how US commercial health plans cover specialty drugs. We also found that although health plans rarely cite ICER’s assessments, they are doing so more frequently.

Implications for managed care pharmacy

We found that, when controlling for other decision-making factors (eg, orphan drug status, availability of alternatives), US commercial health plans’ specialty drug coverage decisions were associated with ICER’s value assessments. In other words, plans tended to impose fewer coverage restrictions on drugs that ICER estimated to have more favorable cost-effectiveness ratios. Future research should examine plans’ responsiveness to ICER’s value assessments and how plans use this information in drug coverage policy formulation.

The Institute for Clinical and Economic Review (ICER) has emerged as a prominent presence in US health care. Surveys of health plan officials have reported that half or more of respondents say they use ICER’s value assessments (ie, ICER’s cost-effectiveness analyses) in contracting discussions or formulary decision-making.^1,2 However, whether US commercial health plans use ICER’s reports when formulating their specialty drug coverage policies is unclear.

We investigated health plans’ use of ICER’s reports in 3 ways. First, we evaluated the association between ICER’s reported cost-effectiveness ratios (CERs) and specialty drug coverage restrictiveness when controlling for other potential decision-making factors. Second, we examined the frequency with which health plans cite ICER assessments in their drug coverage policies and whether this frequency has changed over time. Third, we determined how frequently health plans adjust their coverage decisions for specialty drugs in the 1 year following an ICER assessment.

Methods

We used the Tufts Medical Center Specialty Drug Evidence and Coverage (SPEC) Database, which includes publicly available specialty therapy coverage decisions issued by 17 of the 20 largest US commercial health plans, as well as the evidence that plans cite in support of these decisions (see a list of included plans in Supplementary Exhibit 1 ^{(200.7KB, pdf)} , available in online article). Specialty therapies are typically costly treatments that require special handling, administration, and monitoring. As of August 2020, SPEC included 325 specialty therapies (including cell and gene therapies). Because the US Food and Drug Administration (FDA) may approve a therapy for multiple indications, the database lists each drug indication separately. For example, because the FDA approved bortezomib for multiple myeloma and mantle cell lymphoma, bortezomib is featured twice in the database. For this analysis, SPEC included 654 drug-indication pairs corresponding to 8,586 coverage decisions. Additional details on SPEC can be found elsewhere.^3,4

HEALTH PLAN DRUG COVERAGE

SPEC defines drug coverage restrictiveness by comparing a plan’s policy to the drug’s FDA-labeled indication. Coverage restrictions in SPEC are categorized as (1) patient subgroup restrictions (based on clinical criteria, such as disease severity), (2) step therapy protocols (requirements for patients to try and fail ≥1 alternative treatment before gaining access to a drug), and (3) other (eg, requirement for a drug to be used concomitantly with another medication).

INCLUDED ICER VALUE ASSESSMENTS

We included 37 ICER assessments performed between May 2012 and March 2020. Because ICER typically assesses multiple treatments in an assessment, we included 101 ICER-assessed therapies, corresponding to 111 drug-indication pairs. In 5 instances, ICER assessed the same indication multiple times (eg, ICER-assessed asthma treatments in February 2016 and again in November 2018). For these cases, we included ICER’s most recent assessment.

ANALYSES

Association Between ICER’s Value Assessment and Health Plan Coverage Restrictiveness. We examined the association between ICER’s value assessments and health plan coverage restrictiveness after controlling for other factors potentially relevant to decision-making. We extracted reported base-case incremental CERs from each assessment and grouped them as follows: (1) Dominant (more effective, less expensive than comparator) to less than $100,000 per quality-adjusted life-year (QALY), (2) $100,000 or more to less than $175,000 per QALY, (3) $175,000 or more to less than $500,000 per QALY, (4) $500,000 or more per QALY or dominated (more expensive, less effective than comparator), and (5) no estimate. We grouped interventions that ICER estimated to be dominant with interventions that ICER estimated to have an incremental CER of less than $100,000 per QALY because of the small number of interventions that ICER estimated to be dominant.

If ICER reported separate base-case analyses using a health care sector perspective and a modified societal perspective, we included the ratio from the health care sector perspective because this perspective may align better with health plan decision-making. If ICER assessed the same drug multiple times because of its availability in multiple formulations (eg, subcutaneous and intravenous) or assessed it for different forms of the same disease (eg, episodic and chronic migraine), we calculated an average CER and included that estimate in our analysis.

VARIABLES

The dependent variable in our analysis characterizes whether health plans imposed coverage restrictions beyond the drug’s FDA label indication. We categorized decisions as coverage with no restrictions (ie, the plan covered the drug in a manner consistent with the FDA-approved label) and coverage with restrictions or not covered (ie, the plan imposed restrictions on coverage beyond the FDA label or did not cover the drug). We grouped “not covered” decisions along with “more restrictive” because of the few not covered decisions in our sample.

There were 11 independent variables (Supplementary Table 2 ^{(200.7KB, pdf)} ). Of these, we included 9 because previous investigations have shown them to be associated with restricted drug coverage or because we considered them likely to be associated with restricted drug coverage.^3,5,6 Cancer treatment is a dichotomous variable that accounts for whether the drug is indicated to treat a cancer. Available alternative is a dichotomous variable that accounts for whether the FDA has approved another drug for the same condition. Pediatric population is a dichotomous variable that accounts for whether the drug is indicated for a pediatric patient population (eg, adalimumab for juvenile idiopathic arthritis). Safety is a dichotomous variable that accounts for whether the drug is included in a Risk Evaluation and Mitigation Strategy program or has a black box warning. FDA-expedited approval is a dichotomous variable that accounts for whether the FDA included the drug in 1 of its 4 expedited approval pathways⁷: Priority Review, Accelerated Approval, Fast Track, and Breakthrough Therapy. Self-administered is a dichotomous variable that accounts for whether the drug is self-administered or physician-administered. Orphan status is a categorical variable that accounts for whether the drug is indicated for an orphan disease and whether the orphan disease meets ICER’s ultrarare disease threshold. Annual cost is a categorical variable that accounts for the yearly per patient drug cost, which we estimated using the average wholesale price as reported in the IBM Micromedex Red Book⁸ (a proprietary source of drug pricing) and dosing information from the drug’s label.

We also included Manufacturer ICER membership (ie, whether the manufacturer was a member of ICER—members pay a fee and in return attend regular meetings with ICER officials) and Health plan ICER membership to consider the possibility that ICER membership may influence plan behavior.

ANALYSIS

We developed a multilevel, random-effect logistic regression model to estimate the association of each independent variable on coverage across all health plans while accounting for differences across plans. We report associations between independent variables and coverage restrictiveness using odds ratios (ORs). For each analysis, we considered P values less than 0.05 to be statistically significant. We performed analyses using Stata/SE, release 15.

We conducted sensitivity analyses by running the same regression as outlined above while accounting for other aspects of ICER’s value assessment, including ICER’s clinical evidence rating and discount from a drug’s wholesale acquisition cost (WAC) necessary for a drug’s cost-effectiveness to meet a $100,000/QALY threshold. Supplementary Table 1 ^{(200.7KB, pdf)} describes these variables in further detail.

In certain circumstances, ICER’s public advisory council members vote on whether an intervention represents low, intermediate, or high long-term “value for money.” According to ICER’s value framework, which was in operation from 2017 to 2019, the voting committee automatically deemed treatments with a CER greater than $175,000 per QALY to be “low value.”⁹ Because public advisory council members did not vote on every intervention, we did not include a variable that addresses the public advisory council members vote in a regression analysis. Instead, as a form of sensitivity analysis, we report the frequency with which plans imposed restrictions on drugs that ICER’s public advisory council members voted low, intermediate, or high value.

Health Plan Citation of ICER Value Assessments. We examined the frequency with which health plans cited ICER value assessments in coverage decisions in August of each of the following 4 years: 2017, 2018, 2019, and 2020. We accounted for plan citations of value assessments that ICER had posted on their website and ICER’s value assessments that were published in the peer-reviewed literature. We used a chi-square test to assess whether citation frequency differed for drugs that ICER determined to have CERs below $175,000 per QALY or CERs above $175,000 per QALY. We use the $175,000 per QALY benchmark value to be consistent with the ICER’s 2017-2019 Value Assessment Framework in which the voting committee automatically consider drugs with CERs above this benchmark to be low value.⁹

Pre/Post ICER Assessment. We examined whether plans added or removed coverage restrictions (eg, patient subgroup restrictions or step therapy protocols) following ICER’s assessments by identifying drug-indication pairs evaluated by ICER for which SPEC includes plan coverage both before and after ICER reported its results. We compared coverage decisions issued prior to the date of ICER’s value assessment, with coverage decisions issued exactly 1 year after ICER’s value assessment (eg, if ICER issued an assessment on June 1, 2018, we considered plans’ coverage policies in effect on June 1, 2019). We categorized drug-indication pairs by dichotomized ICER cost-effectiveness results (ie, below or above ICER’s benchmark of $175,000 per QALY used to classify a drug as low value without requiring a vote).⁹

We identified plan adjustments to their coverage criteria (ie, whether they changed coverage policy). A sensitivity analysis uses an alternative cutoff of $100,000 per QALY.

Results

ASSOCIATION BETWEEN ICER’s VALUE ASSESSMENT AND HEALTH PLAN COVERAGE RESTRICTIVENESS

Overall, we categorized 50% (4,258/8,586) of coverage decisions as coverage with no restrictions, 48% (4,129/8,586) as coverage with restrictions, and 2% (199/8,586) as not covered.

Health plans imposed restrictions in 61% (863/1,406) of coverage decisions for drugs assessed by ICER compared with 38% (2,697/7,180) of decisions for drugs not assessed by ICER. When considering only interventions with CERs less than $175,000/QALY, which is the benchmark ICER used to classify a drug as low value without requiring a vote, plans tended to cover interventions with more favorable CERs more generously (Figure 1).⁹ In some cases, plans restricted drugs despite favorable ICER assessments. For example, ICER found emicizumab for hemophilia A to be dominant (cost saving and improves health), compared with a bypassing agent or prophylaxis, but health plans imposed restrictions in 67% of coverage decisions (10 of 15 plans issuing decisions).

FIGURE 1.

The Frequency That Plans Imposed Coverage Restrictions on Interventions With CERs Below $175,000/QALY

Multivariate analysis showed ICER’s assessment of a drug was associated with drug-coverage restrictiveness. That is, the higher (less favorable) the CER reported by ICER, the more likely that plans imposed coverage restrictions. Compared with drug-indication pairs with CERs less than $100,000 per QALY, drug-indication pairs with higher (less favorable) CERs were associated with more restrictive coverage, although the magnitude of association varied by cost-effectiveness range: OR of 4.48 (P < 0.01) if at least $100,000 to less than $175,000 per QALY; OR of 2.00 (P < 0.01) if $175,000 to less than $500,000 per QALY; and OR of 2.10 (P < 0.01) if $500,000 or more per QALY or dominated.

Several independent variables were associated with more generous coverage: cancer treatment, the drug’s inclusion in an FDA-expedited program, orphan (and ultraorphan) status, self-administration, number of years since FDA approval, and whether the drug was indicated for a pediatric patient population. Higher annual drug costs were associated with more restrictive coverage (Table 1).

TABLE 1.

Effect of Independent Variables on the Likelihood of Restricted Drug Coverage, Accounting for ICER’s Cost-Effectiveness Estimate

Independent variable	Multivariate analysis
	Odds ratio	95% CI
Cancer treatment	0.181a	0.158	0.206
Available alternative	0.957	0.830	1.104
Pediatric population	0.772a	0.669	0.892
Years since approval	0.966a	0.959	0.972
Safety	1.047	0.942	1.164
FDA-expedited approval	0.798a	0.709	0.899
Self-administered	0.863a	0.775	0.962
Orphan status
Nonorphan	Reference
Orphan	0.682a	0.604	0.771
Ultra-orphan	0.639a	0.541	0.755
	Joint significance P = 0.000
Health plan is ICER member	0.760	0.501	1.153
Manufacturer is ICER member	0.939	0.842	1.047
Annual cost
First quartile	Reference
Second quartile	1.154b	1.009	1.320
Third quartile	1.248a	1.063	1.465
Fourth quartile	1.300a	1.145	1.557
	Joint significance P = 0.002
ICER cost-effectiveness
No ICER assessment	1.100	0.686	1.460
Dominant/<$100,000 per QALY	Reference
$100,000-$175,000 per QALY	4.479a	2.671	7.509
$175,000-$500,000 per QALY	1.998a	1.325	3.014
>$500,000 per QALY/dominated	2.104a	1.308	3.386
	Joint significance P = 0.000

This analysis included 8,586 coverage decisions. Dominated refers to more expensive, less effective than comparator treatment; dominant refers to less expensive, more effective than comparator treatment.

^aP < 0.01.

^bP < 0.05.

FDA = US Food and Drug Administration; ICER = The Institute for Clinical and Economic Review; QALY = quality-adjusted life-year.

In the first sensitivity analysis, we did not find an association between ICER’s clinical evidence rating and health plan drug coverage restrictiveness (Supplementary Table 3 ^{(200.7KB, pdf)} ). In the second sensitivity analysis, we did not find an association between ICER’s proposed price reduction to achieve a “value-based price” (ie, the reduction needed in an intervention’s WAC price to cause its CER to fall below a $100,000/QALY benchmark [Supplementary Table 4 ^{(200.7KB, pdf)} ]) and health plan drug coverage restrictiveness. There was no observable trend in ICER’s public advisory council member voting and coverage restrictiveness (Supplementary Figure 1 ^{(200.7KB, pdf)} ).

HEALTH PLAN CITATION OF ICER VALUE ASSESSMENTS

Health plans cited ICER reports in 0.8% (5/622) of eligible coverage decisions in 2017, 0.6% (5/833) in 2018, 1.7% (19/1,139) in 2019, and 2.4% (33/1,406) in 2020. Of the 17 included health plans, the number citing ICER in at least 1 coverage decision increased from 3 in 2017 to 11 in 2020.

By 2020, ICER had issued 37 assessments. Ten assessments were cited by multiple plans and 4 by a single plan. No plans cited 23 assessments (Figure 2). ICER’s value assessment for Duchenne muscular dystrophy treatments was cited most frequently (by 5 plans). We found a weak trend for plans more often citing ICER value assessments that estimated the treatment to have a CER of more than $175,000/QALY (P = 0.102).

FIGURE 2.

Frequency Health Plans Cited ICER Value Assessments

PRE/POST ICER ASSESSMENT

Our analysis included 26 drug-indication pairs (which ICER assessed in 8 different reports) corresponding to 307 cases for which SPEC included coverage information that was current before and after an ICER assessment.

For drugs that ICER did not automatically classify as low value based on a threshold of $175,000 per QALY, plans changed coverage criteria in 37% of cases: plans narrowed criteria in 20%, expanded criteria in 15%, and issued a mixed change (more restrictive in one way but more generous in another) in 2%. For low-value drugs, plans changed coverage criteria 29% of the time: plans narrowed criteria in 21%, expanded criteria in 5%, and issued a mixed change in 4%. Our findings were broadly consistent when we used a threshold value of $100,000 per QALY (Table 2).

TABLE 2.

Change in Coverage Restrictiveness in the 12 Months Following ICER Value Assessment Issuance

ICER’s estimated cost-effectiveness	No change in coverage criteria	Change in coverage criteria
		More generous	Mixed	More restrictive
<$175,000 per QALY (n = 159)	62.6	15.0	2.0	20.4
≥$175,000 per QALY (n = 116)	71.2	4.5	3.6	20.7
Not reported ICER (n = 32)	77.4	6.5	3.2	12.9
Sensitivity analysis
<$100,000 per QALY (n = 44)	63.4	12.2	0.0	24.4
≥$100,000 per QALY (n = 231)	66.8	10.1	3.2	19.8
Not reported ICER (n = 32)	77.4	6.5	3.2	12.9

All data are presented as percentage. A total of 26 drug-indication pairs (which ICER assessed in 8 different reports) corresponding to 307 coverage decisions. The following ICER reports were included in this analysis: prostate cancer, Duchenne muscular dystrophy, cardiovascular disease, cystic fibrosis (first report).

CAR-T therapies, hemophilia A, asthma (second report), and psoriasis (second report).

ICER = The Institute for Clinical and Economic Review; QALY = quality-adjusted life-year.

Discussion

ICER has suggested a “grand bargain,” or an implicit agreement in which plans should refrain from imposing access restrictions on products priced in accordance with ICER’s value-based price.^10,11,12 We found health plan drug coverage restrictiveness to be generally associated with ICER’s conclusions on cost-effectiveness. Plans tended to cover drugs with lower (more favorable) CERs more generously than products with higher (less favorable) CERs. Nevertheless, plans often restricted drugs with favorable ratios. For example, ICER determined in its 2018 assessment of targeted therapies for moderate to severe plaque psoriasis that many treatments represent reasonable value.¹³ However, most health plans still require that enrollees first try and experience treatment failure with at least 1 less expensive alternative—topical therapies (eg, corticosteroids), phototherapy, oral systemic treatments (eg, methotrexate), or even older biologic drugs shown to be less effective than newer agents (eg, adalimumab)—before gaining access to a targeted therapy.

Although we found drug coverage restrictiveness to be associated with ICER’s estimated CERs, we did not find a similar association between drug coverage restrictiveness and ICER’s clinical evidence rating or ICER’s estimated discount from a drug’s WAC necessary for a drug’s CER to be $100,000 or less per QALY. These findings may suggest that estimated CERs may be the most influential aspect of ICER’s value assessments.

Our findings suggest that other factors, apart from ICER reports and a drug’s cost-effectiveness, may also influence health plan coverage restrictiveness. These factors include whether the drug is indicated for a cancer or for a pediatric population, the time since a drug’s FDA approval, whether FDA included the drug in an expedited review program or has orphan drug status, whether the drug is self-administered, and the drug’s annual cost. These findings are consistent with previous research.^3,5,6

Previous research has found that health technology assessments represent roughly 10% of the evidence plans cite.¹⁴ Not all plans cited ICER’s value assessments. Plans that did cite ICER’s value assessments infrequently did so, although citation frequency increased somewhat over time. Importantly, our analysis only addressed plans’ citation of ICER’s value assessments in their drug coverage policies. Other research suggests that plans may also use ICER’s assessments when negotiating drug price rebates and discounts,^15,16,17 although net drug prices (ie, after rebates or discounts) still tend, on average, to exceed the ICER’s estimated value-based price.¹⁸ Notably, although the Veterans Administration uses ICER’s value assessments when negotiating drug prices with product manufacturers, it does not use unfavorable ICER value assessments to restrict patient access to care.¹⁹

Although we found coverage restrictiveness to be associated with a product’s cost-effectiveness, we do not know whether plans’ decisions would have been different in the absence of ICER’s value assessment. To more directly evaluate the influence of ICER’s value assessment we examined changes in plans coverage decisions in the year following an ICER assessment. We found limited evidence that plans more often narrowed coverage criteria for therapies ICER automatically deemed low value compared with therapies that ICER did not automatically deem low value. An important limitation of this analysis is that there may be other factors beyond those considered here that may affect plans’ adjustments of coverage criteria (eg, changes in clinical guidelines, or the quality of the clinical evidence supporting a product). Moreover, plans may be unable to adjust their coverage criteria in the 12 months following ICER’s assessment because of existing contracting arrangements.

LIMITATIONS

Our study has a number of other limitations. With respect to ICER value assessment citation frequency, we accounted only for citations in publicly available plan coverage policies. We cannot account for instances in which plans reviewed an ICER value assessment but did not cite it in their decision. Research has found that plans vary widely with respect to evidence they cite in their specialty drug coverage policies.¹⁴ Further, plans may have different committees (eg, pharmacy and therapeutics committees vs value assessment committees) that consider different types of evidence (eg, clinical or economic assessment), which may affect the evidence cited in coverage policies.

Second, the drug prices manufacturers negotiate with health plans are not publicly disclosed. Therefore, we relied on dosing information from the drug’s label and a proprietary source of drug pricing to estimate drugs’ annual costs, which were included as an independent variable in our model.⁸ Third, our findings may not generalize to drugs not in the SPEC Database, to other commercial health plans, or to public payers (eg, Medicare and Medicaid).

ICER has modified its processes in an effort to make its assessments easier for plans to implement. For example, since December 2019, ICER’s reports have typically included a set of proposed coverage criteria among their policy recommendations. However, critics have suggested that ICER’s assessments may have limited utility because they do not adequately reflect the distinct patient populations that different health plans serve, or that they do not align with health plan decision criteria, or because plans believe that QALYs have limited real-world relevance.²⁰

Future research should explore how ICER’s value assessments could be more helpful, for example, by identifying the circumstances in which health plans have found ICER’s reports to be most valuable. Notably, ICER frequently evaluates drugs prior to their FDA approval, hence providing timely information for adjudicating coverage of nascent technologies.

Conclusions

The restrictiveness of US commercial health plans’ specialty drug coverage policies is associated with ICER’s estimate of a drug’s cost-effectiveness. Health plans seldom cite ICER’s value assessments in their coverage decisions, although citation frequency is increasing. We did not observe a trend for plans more often narrowing coverage criteria for therapies ICER automatically deemed low value (CER >$175,000 per QALY) compared with therapies that ICER did not automatically deem low value.