Cost-effectiveness Thresholds Used by Study Authors, 1990-2021

Abstract

This study examines what study authors consider to be appropriate cost-effectiveness analysis thresholds as reflected in the referenced thresholds in their published cost-effectiveness analyses.

For decades, many US authors of cost-effectiveness analyses (CEAs) referenced a common benchmark or threshold—$50 000 per quality-adjusted life-year (QALY)—to reflect how much health gains were “worth,” although this standard had little theoretical or empirical justification.¹ The benchmark is important for health policy discussions because it reflects how much people are willing to pay for health gains or risk reductions or how individuals and society balance the purchase of health care against the purchases of other goods and services.

There are different approaches for inferring an appropriate threshold.² However, the different methods can produce widely different estimates. An alternative way to understand the matter is to examine prevailing views of experts, thus capturing conventional wisdom on the topic in the face of the disparate approaches and estimates. Building on previous work,^1,3 this study examined what study authors believe the appropriate threshold should be as reflected in the referenced thresholds in their published CEAs and how these thresholds have changed over time.

Methods

We analyzed data in the Tufts Medical Center CEA Registry, a database of CEAs presenting incremental cost-per-QALY ratios.⁴ The registry contains a variable on the cost-effectiveness threshold cited by authors as the appropriate standard by which to compare the study’s own results. Although the registry contains CEAs from many countries, we focused on US studies applying US-based cost data and standards for clinical practice and context, published from 1990 through 2021. We examined whether and how references to key thresholds ($50 000 per QALY, $100 000 per QALY, $150 000 per QALY, others [eg, gross domestic product–based or uncommon thresholds], and no threshold) changed over 4 periods (1990-1999, 2000-2009, 2010-2019, and 2020-2021) with a Cochran-Armitage trend test, which also allowed us to test for departure from the linear trend. Using multivariable logistic regression models with robust standard error, we also estimated whether the likelihood of authors citing $150 000 per QALY as a threshold varied according to whether the analysis investigated an intervention for cancer and whether the study was funded by the drug or medical device industry over the 4 periods. We also evaluated an alternative model specification by adding 2- and 3-way interaction terms between study funders, cancer-related intervention, and time, but the addition of the interaction terms did not significantly improve the model fit based on the likelihood ratio test (P = .22). Statistical analyses were conducted in Stata version 17 (StataCorp) using a 2-sided statistical significance level of P < .05.

Results

Of 3276 studies in the sample, the proportion referencing a $50 000-per-QALY benchmark initially increased, but then declined, from 41 of 209 (19.6%) in 1990-1999 to 289 of 861 (33.6%) in 2000-2009, 505 of 1761 (28.7%) in 2010-2019, and 56 of 445 (12.6%) in 2020-2021. US-based authors increasingly referenced $100 000 per QALY—19 of 209 (9.1%) in 1990-1999, 199 of 861 (23.1%) in 2000-2009, 691 of 1761 (39.2%) in 2010-2019, and 209 of 445 (47.0%) in 2020-2021)—or $150 000 per QALY—13 of 861 (1.5%) in 2000-2009, 158 of 1761 (9.0%) in 2010-2019, and 105 of 445 (23.6%) in 2020-2021. The trends were statistically significant for all 3 thresholds (Table 1). Results of the regression model indicated that the odds of referencing the $150 000-per-QALY benchmark in cancer-related CEAs (117/658 [17.8%]) were 2.22 times (95% CI, 1.70-2.90) greater than in non–cancer-related CEAs (159/2618 [6.1%]), while industry-funded CEAs (70/723 [9.7%]) reported 1.44 times (95% CI, 1.07-1.95) greater odds of citing the $150 000-per-QALY threshold than non–industry-funded CEAs (206/2553 [8.1%]) (Table 2).

Table 1. Cost-effectiveness Thresholds Referenced by Authors of US-Based CEAs, 1990-2021^a.

Threshold	No. (%)
	1990-1999 (n = 209)	2000-2009 (n = 861)	2010-2019 (n = 1761)	2020-2021 (n = 445)	All periods (N = 3276)b
$50 000 per QALY	41 (19.6)	289 (33.6)	505 (28.7)	56 (12.6)	891 (27.2)
$100 000 per QALY	19 (9.1)	199 (23.1)	691 (39.2)	209 (47.0)	1118 (34.1)
$150 000 per QALY	0	13 (1.5)	158 (9.0)	105 (23.6)	276 (8.4)
Otherc	34 (16.3)	117 (13.6)	183 (10.4)	35 (7.8)	369 (11.3)
No threshold	115 (55.0)	243 (28.2)	224 (12.7)	40 (9.0)	622 (19.0)

Abbreviations: CEA, cost-effectiveness analysis; QALY, quality-adjusted life-year.

^a Categories are mutually exclusive. When studies cited multiple thresholds, they were classified by the highest threshold referenced. For example, if a CEA referenced both $50 000 per QALY and $100 000 per QALY, it was classified into $100 000 per QALY.

^b Based on Cochran-Armitage trend tests. All trends were statistically significant at a P < .001. Except for the trend in the “others” category, all of the categories had a statistically significant nonlinear trend.

^c Other thresholds referenced include the gross domestic product–based thresholds (eg, 3 times gross domestic product) or uncommon thresholds (eg, $20 000 per QALY or $200 000 per QALY).

Table 2. Association Between Referencing a $150 000-per-QALY Threshold and Study Characteristics (N = 3276)^a.

Characteristics	No.	No. (%) referencing threshold	Odds ratio (95% CI)	P value
Publication year			1.27 (1.22-1.32)	<.001
Nonindustry funder	2553	206 (8.1)	1 [Reference]
Industry funder	723	70 (9.7)	1.44 (1.07-1.95)	.02
Non–cancer-related CEAs	2618	159 (6.1)	1 [Reference]
Cancer-related CEAs	658	117 (17.8)	2.22 (1.70-2.90)	<.001

Abbreviations: CEA, cost-effectiveness analyses; QALY, quality-adjusted life-years.

^a The multivariable logistic regression model was applied with robust standard error using a reference of the $150 000-per-QALY threshold (yes/no) as a dependent variable. Publication year (as a continuous variable) and indicators for study funder and cancer-related CEAs were used as independent variables. The entire sample (N = 3276) was used in the analysis. For sensitivity analyses, samples were restricted by excluding “other” thresholds cited (n = 2907) and excluding “other” and “no threshold” cited (n = 2285), but results remained unchanged.

Discussion

Between 1990 and 2021, authors of CEAs increasingly cited $100 000 or $150 000 per QALY as the benchmark by which to judge whether an intervention was cost-effective. The trend is generally consistent with recent theoretical and empirical work in the US.^5,6 The Institute for Clinical and Economic Review, a US-based nonprofit organization, uses a $100 000- to $150 000-per-QALY threshold as the basis for recommending value-based drug prices.

Cancer-related CEAs referenced higher thresholds than non–cancer-related CEAs, which may suggest a view that interventions for diseases associated with greater mortality and morbidity warrant higher thresholds. Authors of industry-funded analyses referenced higher thresholds, perhaps suggesting a desire by those authors to justify higher value-based prices for drug therapies. Study limitations include that not all CEA authors stated a threshold for comparison, that the search strategy for the CEA registry may have omitted some published analyses, and that cancer was used as a proxy for disease severity.

Section Editors: Jody W. Zylke, MD, Deputy Editor; Kristin Walter, MD, Senior Editor.

Supplement.

Data Sharing Statement