Trends in Testosterone Prescribing for Age-related Hypogonadism in Men With and Without Heart Disease

Abstract

This cohort study examines the trend in prescription testosterone use among male Medicare beneficiaries with and without coronary artery disease aged 50 years or older.

Signals of cardiovascular risk from off-label prescription testosterone use began with the early termination of a placebo-controlled trial in frail, older men. Basaria et al¹ revealed more cardiovascular events among men randomized to receive testosterone than among men randomized to receive placebo. Some observational studies reinforced these concerns for cardiovascular risk.^2,3 In 2014, the US Food and Drug Administration (FDA) issued a safety communication about testosterone drugs, and the Endocrine Society expressed particular concern about testosterone use by men with heart disease.^4,5 In 2015, the FDA revised testosterone labels, highlighting the lack of efficacy evidence for age-related hypogonadism and adding a warning about possible increased cardiovascular risk.⁶

In this study, we explored trends in testosterone use among Medicare-insured men with and without coronary artery disease (CAD) to see how prescribing changed in response to concerns.

Methods

This study was approved by the institutional review board at Dartmouth College. Patient consent was waived by this institutional review board because the research could not be practicably conducted without a waiver and because the risks to individuals were deemed reasonable in relation to the anticipated advantages.

Using a 40% random sample of Medicare fee-for-service administrative data from January 1, 2007, to December 31, 2016, we created annual cohorts of men aged 50 years or older, with and without a same-year CAD diagnosis, and measured testosterone prescription fills under Medicare Parts B and D. Use was considered on-label if patients had a same-year diagnosis for any indicated condition (eg, Klinefelter syndrome) or off-label if patients did not have a same-year diagnosis.

We calculated the percentage of men who received prescription testosterone, using direct adjustment for national rates and logistic regression for hospital referral region rates (adjusting for age, race, Part D low-income subsidy, and Medicaid eligibility). We limited regional analyses to the 273 hospital referral regions with 500 or more beneficiaries in each annual cohort. Testosterone expenditures were estimated by multiplying the observed reimbursement rate (from the 40% sample) by 2.5. Secondary and sensitivity analyses were included for Part D recipients subject to prior authorization, age-stratified models (<65 years vs ≥65 years), and cohorts in which a CAD diagnosis or a testosterone-approved indication was determined with a 3-year (rather than same-year) look-back analysis.

Results

The annual cohorts ranged from 1.8 to 3.1 million, representing 10% to 12% of all fee-for-service male Medicare enrollees older than 50 years. Annual CAD cohorts ranged from 629 823 (33.9% of the full cohort) in 2007 to 1 017 484 (32.3% of the full cohort) in 2016. The mean (SD) age was 71.2 (9.0) years. The adjusted percentage of all men who received on-label prescription testosterone ranged from 0.2% (3 681) to 0.3% (8 861); off-label prescription testosterone use peaked in 2013 at 3.2% of the cohort (29 144) with CAD and 2.4% of the cohort (45 029) without CAD.

Testosterone use was consistently higher for men with CAD than for men without CAD, 2.0% and 1.6% of the cohorts, respectively, in 2016, for example (Figure 1). Sensitivity analyses replicated the main analyses. Annual testosterone spending was $108 million in 2007 and $402 million in 2016. The proportion of men with Part D prescription testosterone fills subject to prior authorization was 21% (13 685) in 2012 and 49% (31 462) in 2016. Regional testosterone use varied substantially. Use was consistently twice the national rate in 4 regions but was less than half the national rate in 14 regions. For example, the national rate in 2012 was 2.0% in 4 regions with high use, and in 2016, use ranged from 4.4% to 7.0% of the full cohort; in 14 regions with low use in 2012, the use ranged from 0.5% to 1.0% in 2016 (Figure 2).

Figure 1. On-label and Off-label Prescription Testosterone Use and Medicare Testosterone Expenditure.

Use measurements are crude annual rates. Medicare spending includes Part D records and Part B claims. Measures from the 40% sample are multiplied by 2.5 for an estimated total, adjusted to 2016 dollars using the Consumer Price Index inflation calculator. Total cost figures likely underestimate Medicare expenditures as only use among fee-for-service parts A, B, and D enrolled beneficiaries was calculated. CAD indicates coronary artery disease.

Figure 2. Regional Variation in Prescription Testosterone Use in 2012 and 2016, by Heart Disease Status.

A, Hospital referral region rates of off-label testosterone use in 2012 before warnings. B, Hospital referral region rates of off-label testosterone use in 2016 after warnings. Each dot represents 1 of 273 regions. The horizontal line with percentage in each graph indicates the national rate. In orange regions, the proportion of men with coronary artery disease (CAD) who used testosterone was 2 or more times the national rate in both 2012 and 2016. In green coded regions, the proportion of men with CAD who used testosterone was less than half the national rate in 2012 and 2016. Plot includes only regions with 500 or more Medicare beneficiaries who met the inclusion criteria for the CAD or non-CAD cohort.

Discussion

After the publication of safety warnings from the FDA and observational studies that found testosterone harmful, use among Medicare beneficiaries declined after 2013 but remained higher than in 2007, and was largely off-label. In 2016, testosterone use was still higher for men with CAD, those at the highest risk, demonstrating a nonselective response to possible cardiovascular risk.

What prompted the observed trends? Off-label testosterone advertising (which used the term low T) targeting men at higher cardiovascular risk (diabetes, hypertension, and high cholesterol) voluntarily stopped in 2014. Regional use variation suggests that local prescribing cultures also matter.

Although prior authorizations for prescription testosterone increased between 2012 and 2016 (and were associated with less off-label use), they do not appear to have differentially affected men with CAD. Persistently higher testosterone use among patients with CAD suggests the need for more effective dissemination of label changes. A robust prior authorization process could also help, as could rigorous evidence-based safety-review processes, richer shared-risk contracts, and patient education.

This study is limited by the imperfect nature of claims data and the relatively short time frame of the observation. Similar results from sensitivity analyses, using 3-year reviews rather than same-year diagnoses, reinforce our conclusions and suggest that a misclassification is minimal.

In 2015, the FDA mandated that testosterone manufacturers conduct a randomized clinical trial assessing cardiovascular risk; the completion of such trials is projected for 2022. Until these results emerge, caution should prevail.