Association Between Direct-to-Consumer Advertising and Testosterone Testing and Initiation in the United States, 2009-2013

Key Points

Question  Is there an association between televised direct-to-consumer testosterone advertising and testosterone testing and initiation in the United States?

Findings  In this ecological study of 75 US designated market areas, each exposure to a testosterone advertisement was associated with monthly relative increases in rates of new testosterone testing of 0.6%, new initiation of 0.7%, and initiation without a recent baseline test of 0.8%.

Meaning  Regional exposure to televised direct-to-consumer advertising was associated with greater testosterone testing, new initiation, and initiation without recent serum testosterone tests.

Importance  Testosterone initiation increased substantially in the United States from 2000 to 2013, especially among men without clear indications. Direct-to-consumer advertising (DTCA) also increased during this time.

Objective  To investigate associations between televised DTCA and testosterone testing and initiation in the United States.

Design, Setting, and Population  Ecologic study conducted in designated market areas (DMAs) in the United States. Monthly testosterone advertising ratings were linked to DMA-level testosterone use data from 2009-2013 derived from commercial insurance claims. Associations between DTCA and testosterone testing, initiation, and initiation without recent baseline tests were estimated using Poisson generalized estimating equations.

Exposures  Monthly Nielsen ratings for testosterone DTCA in the 75 largest DMAs.

Main Outcomes and Measures  (1) Rates of new serum testosterone testing; (2) rates of testosterone initiation (in-office injection, surgical implant, or pharmacy dispensing) for all testosterone products combined and for specific brands; and (3) rates of testosterone initiation without recent serum testosterone testing.

Results  Of 17 228 599 commercially insured men in the 75 DMAs, 1 007 990 (mean age, 49.6 [SD, 11.5] years) had new serum testosterone tests and 283 317 (mean age, 51.8 [SD, 11.3] years) initiated testosterone treatment. Advertising intensity varied by geographic region and time, with the highest intensity seen in the southeastern United States and with months ranging from no ad exposures to a mean of 13.6 exposures per household. Nonbranded advertisements were common prior to 2012, with branded advertisements becoming more common during and after 2012. Each household advertisement exposure was associated with a monthly increase in rates of new testosterone testing (rate ratio [RR], 1.006; 95% CI, 1.004-1.008), initiation (RR, 1.007; 95% CI, 1.004-1.010), and initiation without a recent test (RR, 1.008; 95% CI, 1.002-1.013). Mean absolute rate increases were 0.14 tests (95% CI, 0.09-0.19), 0.05 new initiations (95% CI, 0.03-0.08), and 0.02 initiations without a recent test (95% CI, 0.01-0.03) per 10 000 men for each monthly ad exposure over the entire period.

Conclusions and Relevance  Among US men residing in the 75 designated market areas, regional exposure to televised direct-to-consumer advertising was associated with greater testosterone testing, new initiation, and initiation without recent testing.

Quiz Ref IDThe US Food and Drug Administration (FDA) permitted televised direct-to-consumer advertising (DTCA) for prescription medications in 1997. An argument for DTCA is that it may steer consumers to beneficial treatments for high-priority illnesses,¹ but it may also contribute to escalating drug prices and reduced patient safety.² A growing evidence base illustrates strong associations between televised DTCA and increased product prescribing for a variety of indications,^3-5 and the American Medical Association called for a ban on such marketing in 2015, attempting to reduce prescription costs and curb consumer demand for new, expensive, and potentially inappropriate treatments.⁶

Increased awareness of and demand for a medication may be beneficial if the risk-benefit ratio of the drug is favorable. For many men taking testosterone, that assumption is questionable. Testosterone therapies were originally approved to treat hypogonadism resulting from the disruption of the pituitary-hypothalamus-gonadal axis. Now many men take or are prescribed testosterone for age-related reduced testosterone levels or nonspecific symptoms without pathological hypogonadism. Despite considerable disagreement about the necessity and safety of treatment for these findings and symptoms, testosterone prescribing has increased substantially in the absence of proven new indications,^7-9 particularly in the United States, where the testosterone initiation rate rose from 20.2 to 75.7 per 10 000 person-years from 2000 to 2011.⁸

Although studies have yielded conflicting information on the risks and benefits of testosterone,^10-13 termination of a clinical trial because of excess adverse cardiovascular effects¹⁴ and observational studies suggesting increased cardiovascular risk^11,15 led the FDA to reevaluate the risks and revise product labeling requirements in 2014, and pharmaceutical manufacturers, who had previously heavily marketed testosterone products through televised branded, product-specific promotions and unbranded, “low T” awareness advertisements, voluntarily discontinued televised DTCA for testosterone products.

Given the controversies surrounding DTCA and testosterone, we examined the association between exposure to DTCA for testosterone and testosterone testing, initiation, and initiation without a preceding testosterone test.

This project was approved by the institutional review board of the University of North Carolina at Chapel Hill. Individual consent was not required for this analysis of deidentified data.

Quiz Ref IDWe licensed monthly television ratings data from Nielsen Media Research for televised testosterone advertisements in the largest 75 of 210 US designated market areas (DMAs), which are whole, adjoining counties centered around large cities whose populations receive similar television programming. We collected ratings between 2008 and 2013 for both brand-specific prescription testosterone advertisements, which are regulated by the FDA, and nonbranded, condition awareness (“low T”) ads, which are not FDA regulated; we measured DTCA using gross ratings points, a standard metric for quantifying advertising exposure reflecting the percentage of the target audience reached multiplied by frequency of exposures as measured by a combination of electronic metering and surveying across broadcast platforms.^16,17 We summed gross ratings points within the same DMA and month for all testosterone advertisements and for branded testosterone advertisements and condition awareness ads separately, and divided by 100 to obtain the mean number of times testosterone advertisements were viewed in television households in a given month; an ad with 100 gross ratings points per month can be interpreted as being seen 1 time per month by 100% of the audience, and each 1-unit increase in monthly exposures represented approximately 1 additional viewing of a television ad per household.^16,18

We used the MarketScan Commercial Claims and Encounters databases (Truven Health Analytics) to quantify population-level testosterone testing and initiation (in-office injection, surgical implant, or pharmacy dispensing) within DMAs. The MarketScan databases include individual-level clinical diagnoses, procedures, outpatient prescription dispensing (including from retail, mail-order, and specialty pharmacies), expenditures, and enrollment data paid for by employer-sponsored, fee-for-service, fully capitated or partially capitated plans. Claims from employees and their dependents and those with employer-based Medicare supplemental plans also are included. Claims are fully paid and adjudicated, and the research databases are constructed after sufficient time has elapsed to allow for submission, payment, and recording of all outstanding claims. Considerable standardization and quality control activities occur during database construction.¹⁹

Quiz Ref IDWe identified men aged 18 years or older residing within defined metropolitan statistical areas (MSAs); 1 or more MSAs make up a DMA. Geographic information was not available for those residing outside MSAs or in smaller micropolitan statistical areas, so our sample focused on those in more urban population centers. Using procedure codes from submitted insurance claims, we identified men newly tested for serum testosterone levels following 6 months without testosterone receipt or testing. Initiation of testosterone gels, patches, injections, or implants was defined as pharmacy dispensing or in-office receipt identified through procedure codes of testosterone following 6 months without prior testosterone receipt. Our main analysis identified initiators of all testosterone products, while subanalyses focused on initiators of individual advertised, branded products (AndroGel [Abbvie Inc] and Axiron [Eli Lilly Inc]). The denominator for monthly rate calculations was the number of adult men enrolled in the MarketScan databases within each MSA on the 15th of each month with at least 6 months of prior continuous enrollment.

We collapsed MSA-level information into DMA levels and linked the data with advertising information.

We collected county-level characteristics and aggregated them into DMAs to account for differences in population demographics, access to health care (measured by physician density), and socioeconomic status (median income). We derived county-level 5-year estimates of age and race/ethnicity distributions from the 2009-2013 American Community Survey, an ongoing yearly survey of US households conducted by the US Census Bureau; race/ethnicity designation was self-reported by choosing from fixed, standard categories. We derived annual estimates of county-level physician density and median income from the Health Resources and Services Administration’s Area Resource File.

We examined the overall pattern of testosterone testing and initiation between 2009 and 2013. We estimated monthly rates of testing and initiation per 10 000 men in each DMA and plotted national mean rates with their interquartile ranges. We assessed geographic variation by summing all advertising exposures throughout the study period within DMAs and divided these cumulative measures into 7 approximately equal strata using an equal-distribution (quantizing) algorithm to visualize regional variation in total DTCA; these DTCA levels were mapped across the United States by year to observe changes in geographic variation over time.

We describe the association between DTCA and testosterone testing and initiation using adjusted rate ratios (RRs) and 95% confidence intervals derived from multivariable Poisson models with generalized estimating equations²⁰ to account for clustering within DMAs over time; we used an autoregressive (AR-1) correlation structure. Direct-to-consumer advertising exposures were included as a continuous variable; thus, the resulting RRs indicate the association between a 1-unit increase in ad exposures and testosterone rates. Monthly advertising data (November 2008 to October 2013) were linked to monthly testosterone data 2 months in the future, allowing for the passage of time between viewing an ad, seeking care, and receiving treatment. To observe the effect of varied lag times, we performed a preplanned sensitivity analyses evaluating a 1-month lag and a post hoc analysis of a 3-month lag.

We adjusted the models for age (proportion of population in age groups of 35-44, 45-54, 55-64, 65-74, and ≥75 years, included in model as 5 variables), race (proportion of white and black population, included in model as 2 variables), physician density, socioeconomic status (included in model as a quadratic trend of median income because increases in income may be more meaningful for medication use at lower income values), seasonality (modeled with 4 variables, 1 for each season, each taking a 0 or 1 value), and overall increasing usage time trend⁸ to facilitate estimation of the increases in testosterone testing and initiation within a DMA not due to expected seasonal variation or overall national trends. We modeled time trends with cubic splines to allow for flexibility at nodes corresponding to key events: FDA warning of testosterone gel transfer from men to women and children²¹ (May 2009); release and approval of new testosterone gel formulations^22,23 (April 2011); and National Institutes of Health funding of observational studies of the cardiovascular safety of testosterone products²⁴ (August 2012).

Our primary exposure variable was total monthly testosterone advertising exposures, but we additionally investigated individual branded advertisements and unbranded condition awareness advertisements from 2012 to 2013; prior to 2012, there were primarily only unbranded “low T” condition awareness advertisements. We estimated the association of advertising with (1) new serum testosterone tests; (2) all new testosterone initiations, including branded and unbranded patch, gel, injection, and implanted formulations; (3) initiation without a recent test; and (4) product-specific initiation of branded testosterone gels. Model-based predicted rates were plotted against observed testosterone rates to evaluate model fit. As post hoc sensitivity analyses we investigated the associations in each year alone because there was substantial variation in advertising intensity by year and more diverse advertising, greater competition, and plateaued overall use during later years, and we explored nonlinear relationships by modeling DTCA with both a linear and a quadratic term.

We compared RR estimates from different periods and sensitivity analyses using ratios of RRs.²⁵ We considered a 2-sided threshold of P<.05 to be statistically significant. All analyses were conducted using SAS software, version 9.4 (SAS Institute Inc).

Of 17 228 599 unique men across the 75 DMAs (mean age, 42.8 [SD, 16.0] years), 1 007 990 (mean age, 49.6 [SD, 11.5] years) had new serum testosterone tests and 283 317 (mean age, 51.8 [SD, 11.3] years) initiated testosterone treatment during the study period (59% initiated testosterone gels, 36% injections, 3% patches, and 2% implants). Baseline DMA characteristics (Table 1) and testosterone use varied widely, with DMA-level testing rates ranging from 2.7 (95% CI, 0.7-7.3) to 30.5 (95% CI, 26.0-35.7) per 10 000 men and initiation rates ranging from 0.6 (95% CI, 0.1-1.9) to 23.2 (95% CI, 5.9-63.0) per 10 000 men (Table 1; see eTable 1 in the Supplement for DMA-level characteristics by year).

Monthly testosterone testing and initiation increased between 2009 and 2012 prior to decreasing through 2013 (Figure 1, A and B); initiation without a recent baseline test decreased slightly but consistently over the period. After introduction of Axiron in 2011, Androgel initiation rates increased and then decreased in 2012-2013 (Figure 1C). Mean monthly rates per 10 000 men were 24.6 (SD, 9.5) for testosterone testing, 7.6 (SD, 3.8) for testosterone initiation, and 2.4 (SD, 1.5) for testosterone initiation without prior testing.

Direct-to-consumer advertising exposure increased over the period with substantial variation, ranging from months without any advertising to a mean 13.6 (SD, 0.8) monthly household exposures in December 2012 (Figure 2A). Prior to 2012, Abbvie-funded, unbranded “low T” advertisements were the primary form of advertising. Following the introduction of Axiron and a more concentrated Androgel formulation in 2011, more branded, product-specific advertising emerged and unbranded condition awareness advertisements decreased (Figure 2B).

Television advertising intensity varied by DMA region, with cumulative exposures ranging from 154.9 to 203.5 advertisement exposures per DMA over the study period (mean, 179.1; SD, 10.0). The most concentrated advertising occurred in the Southeast and the southern Great Lakes regions (Figure 3), a geographic focus consistent through most of the study period (eFigure 1 in the Supplement).

Increases in advertisement exposure were associated with higher rates of testosterone testing, initiation, and initiation without testing (Table 2), except Axiron initiation, which was introduced later in the period and had lower overall use (Table 2; see eTable 2 and eFigure 2 in the Supplement for model goodness-of-fit estimates). Each additional ad exposure was associated with a relative 0.6% (95% CI, 0.4%-0.8%) increase in testing, 0.7% (95% CI, 0.4%-0.10%) increase in initiation, and 0.8% (95% CI, 0.2%-0.13%) increase in initiation of testosterone without baseline testing, corresponding to absolute monthly increases of 0.14 new tests (95% CI, 0.09-0.19), 0.05 initiations (95% CI, 0.03-0.08), and 0.02 initiations without baseline testing (95% CI, 0.01-0.03) per 10 000 men across the study period.

Prior to 2012, advertising—almost all of which was unbranded “low T” advertising—was associated with testosterone testing (RR, 1.012; 95% CI, 1.009-1.014) and initiation (RR, 1.012; 95% CI, 1.007-1.017); after 2012, the advertising was primarily branded, but associations with testing (RR, 0.998; 95% CI, 0.995-1.001) or initiation (RR, 0.998; 95% CI, 0.994-1.003) were no longer statistically significant (ratios of RRs in 2012-2013 vs 2009-2011, 0.986 [95% CI, 0.982-0.990]; P < .001 for testing and 0.986 [95% CI, 0.980-0.993]; P < .001 for initiation, suggesting a statistically significant difference in association between the periods). Androgel was the most widely used branded product during the entire study period and the most widely advertised branded product after 2012; its advertising was associated with increased rates of testing and initiation across all domains (Table 2). However, other advertising types showed a more attenuated or no association in 2012 and 2013.

When exploring nonlinear relationships between DTCA and testosterone by modeling with a quadratic term, we observed a “saturation effect” wherein the association between a 1-unit increase in DTCA exposure and testosterone became weaker at higher levels of DTCA (eTable 4 in the Supplement). For example, when a DMA’s monthly exposure increased from 1 to 2, the association between the increase in DTCA with new testing was an RR of 1.015 (95% CI, 1.012-1.019); from 5 to 6, an RR of 1.012 (95% CI, 1.009-1.014); and from 10 to 11, an RR of 1.007 (95% CI, 1.005-1.009).

There were no statistically significant associations between DTCA and testosterone testing and initiation in sensitivity analyses testing 1- and 3-month lag periods (eTable 5 in the Supplement).

Increasing televised DTCA for testosterone therapies was observed across US metropolitan areas between 2009 and 2013, and both branded and condition awareness ads were associated with increased testosterone testing and initiation. Although the average increase in testosterone rates associated with a single ad exposure was less than 1%, advertisements were widespread and frequent during the study period; with cumulative ad exposures of close to 200 in some DMAs, DTCA was associated with substantial overall increases in testosterone testing and initiation.

These associations were strongest when only a single brand was being advertised, and there may have been a saturation effect, as DTCA was more prevalent during 2012 and 2013 than earlier periods, yet with increasing safety concerns, introduction of new formulations, and competing advertising campaigns, testosterone use decreased somewhat and the associations of DTCA with testing or initiation became attenuated (eTable 3 and eTable 4 in the Supplement). Additionally, DTCA was associated with more initiation of testosterone without recent serum testing, contrary to treatment guidelines.²⁶

These findings illustrate the association between mass marketing efforts and demand for a prescription drug in a context in which it has been widely overused.^7,8 Ultimately, prescribing decisions rest with physicians, but patient preferences and requests for specific medications influence prescribing decisions,^27-29 and DTCA may be associated with increased patient demand. While other studies have demonstrated associations between DTCA and increasing medication use, this study demonstrates increases in potentially inappropriate use and increasing initiation during a time when most testosterone use was of questionable value for age-related testosterone decreases without strong evidence of benefit. Characterizing the role of DTCA in promoting testosterone initiation among a large segment of middle-aged and older men for nonspecific symptoms and age-related declines in testosterone levels is relevant to ongoing policy debates regarding DTCA. This study complements many others that suggest the contribution that DTCA may make in the early adoption of recently approved treatments whose risk-benefit profile may be quite unclear.²

Quiz Ref IDFurthermore, concerns remain regarding the safety and benefit of using testosterone to treat age-related declines in testosterone levels. Original FDA approval of testosterone formulations was based on trials demonstrating increased testosterone blood levels among men with specific disorders of the testes or pituitary, not amelioration of any clinical symptoms.³⁰ A recent study on the benefits of raising testosterone levels among older men with low testosterone but without pathological hypogonadism found moderate improvements in sexual function and mood but no benefit in vitality or walking distance³¹; this may reignite the cost-benefit discussion, yet the trials were too small to evaluate safety, and they offer little support for widespread use of testosterone outside the narrow approved indications. Although the current study lacks individual-level information on patients’ indications for treatment, much of the increase in testosterone use in the United States has been among men being treated for age-related decreases in testosterone or nonspecific symptoms.⁸

This study has limitations. As an ecologic study, it lacks information on individual-level DTCA exposures. Some individuals residing near DMA boundaries may view televised ads outside their assigned DMA. For pharmacy-dispensed products, the study assessed receipt from the pharmacy, not actual use by patients. Although there were no explicitly missing data from any data source, the findings may underestimate true testosterone testing and initiation if tests or prescriptions were not submitted to insurers for reimbursement. Additionally, the MarketScan databases contain information only on those with employer-sponsored health insurance in the United States; these results may not be generalizable to men covered by public insurance plans or uninsured men. This analysis was limited to investigating geographic variation in televised pharmaceutical advertising; there may be additional forms of influential promotion, such as physician-targeted promotion, internet advertising (including branded ads and unbranded disease awareness symptom quizzes), or advertisements by men’s health specialty clinics, thus potentially overattributing some testosterone use to televised DTCA. While televised DTCA is the most common and influential type of direct-to-consumer advertising medium,^32,33 internet advertising was also widespread throughout most of the study period, although it likely lacks the geographic variation of televised DTCA. However, televised DTCA was consistently associated with multiple dimensions of testosterone use. Finally, the statistically significant associations between DTCA and testosterone testing and initiation we report at 2 months were not apparent in sensitivity analyses testing of 1- and 3-month lag periods; the reasons for the dependence of the association on the 2-month lag period are not knowable from our data.

Quiz Ref IDAmong US men residing in 75 US DMAs, regional exposure to televised DTCA was associated with greater testosterone testing, new initiation, and initiation without recent testing.

Corresponding Author: J. Bradley Layton, PhD, Department of Epidemiology, University of North Carolina at Chapel Hill, 2104-I McGavran-Greenberg Hall, Campus Box 7435, Chapel Hill, NC 27599-7435 (blayton@unc.edu).

Author Contributions: Dr Layton had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Layton, Kim, Emery.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Layton, Kim, Emery.

Critical revision of the manuscript for important intellectual content: Kim, Alexander, Emery.

Statistical analysis: Layton, Kim, Emery.

Obtained funding: Emery.

Administrative, technical, or material support: Alexander, Emery.

Study supervision: Emery.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Alexander is chair of the FDA’s Peripheral and Central Nervous System Advisory Committee, serves as a consultant to a mobile startup to improve patient pain management, serves as a paid consultant to QuintilesIMS Health, and serves on a QuintilesIMS Health scientific advisory board. This arrangement has been reviewed and approved by Johns Hopkins University in accordance with its conflict of interest policies. No other disclosures were reported.

Funding/Support: This project was funded by grants from the National Institute on Aging (grant R01 AG042845) and the National Heart, Lung, and Blood Institute (grant R01 HL107345). The database infrastructure used for this project was funded by the Department of Epidemiology, University of North Carolina (UNC) Gillings School of Global Public Health; the Cecil G. Sheps Center for Health Services Research, UNC; the CER Strategic Initiative of UNC’s Clinical Translational Science Award (grant 1 ULI RR025747); and the UNC School of Medicine.

Role of the Funder/Sponsor: The sponsors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication.

Meeting Presentation: Portions of this work were presented as a poster at the 31st International Convention on Pharmacoepidemiology and Therapeutic Risk Management; August 2015; Boston, Massachusetts.

Additional Contributions: We thank Yaru Shi, MS, Division of Epidemiology and Biostatistics, School of Public Health, University of Illinois at Chicago, Glen Szczypka, NORC at the University of Chicago, and Dongmei Li, MS, UNC at Chapel Hill, for contributions to data acquisition and management, and Lisa Vera, BA, VeraCite Inc, for assistance with preparation of the manuscript. These contributors did not receive compensation for their roles.