Schematic of the main (A) and exploratory (B) case-crossover study designs. VTE indicates venous thromboembolism.
eTable 1. ICD-9-CM and ICD-10-CM Codes Used to Define Incident VTE
eTable 2. Testosterone Therapy Use of Men Within Their Case and Control Periods Stratified by the Different Lengths of Periods, MarketScan, 2011-2017
eTable 3. Overall Association of TT Prescription With Risk of VTE Among Men With and Without Hypogonadism (Jasuja et al), MarketScan, 2011-2017
eTable 4. Overall Association of Consistent TT Prescription Users (>1 Fill) With Risk of VTE Among Men With and Without Hypogonadism, MarketScan, 2011-2017
eTable 5. Adjusted Exploratory Analyses (1 Month, 1-3 Month, 3-6 Month Case Periods) of TT Prescription With Risk of VTE Among Men of Differing Hypogonadism Status Stratified by Age and Route of Testosterone Exposure, MarketScan, 2011-2017
Customize your JAMA Network experience by selecting one or more topics from the list below.
Identify all potential conflicts of interest that might be relevant to your comment.
Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.
Err on the side of full disclosure.
If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.
Not all submitted comments are published. Please see our commenting policy for details.
Walker RF, Zakai NA, MacLehose RF, et al. Association of Testosterone Therapy With Risk of Venous Thromboembolism Among Men With and Without Hypogonadism. JAMA Intern Med. 2020;180(2):190–197. doi:10.1001/jamainternmed.2019.5135
Is clinical prescription of testosterone therapy associated with short-term risk of venous thromboembolism in men with and without hypogonadism?
In this case-crossover study comparing 6-month testosterone use for 39 622 men who had a venous thromboembolism with testosterone use 6 to 12 months before the venous thromboembolism, use of testosterone therapy in the 6-month case period was associated with an increased risk of venous thromboembolism among men with and without hypogonadism.
The findings suggest that testosterone therapy is associated with increased short-term risk of venous thromboembolism among all men prescribed the therapy.
Testosterone therapy is increasingly prescribed in patients without a diagnosis of hypogonadism. This therapy may be associated with increased risk of venous thromboembolism (VTE) through several mechanisms, including elevated hematocrit levels, which increase blood viscosity.
To assess whether short-term testosterone therapy exposure is associated with increased short-term risk of VTE in men with and without evidence of hypogonadism.
Design, Setting, and Participants
This case-crossover study analyzed data on 39 622 men from the IBM MarketScan Commercial Claims and Encounter Database and the Medicare Supplemental Database from January 1, 2011, to December 31, 2017, with 12 months of follow-up. Men with VTE cases who were free of cancer at baseline and had 12 months of continuous enrollment before the VTE event were identified by International Classification of Diseases codes. Men in the case period were matched with themselves in the control period. Case periods of 6 months, 3 months, and 1 month before the VTE events were defined, with equivalent control periods (6 months, 3 months, and 1 month) in the 6 months before the case period.
National drug codes were used to identify billed testosterone therapy prescriptions in the case period (0-6 months before the VTE) and the control period (6-12 months before the VTE).
Main Outcomes and Measures
The main outcome in this case-only experiment was first VTE event stratified by the presence or absence of hypogonadism.
A total of 39 622 men (mean [SD] age, 57.4 [14.2] years) were enrolled in the study, and 3110 men (7.8%) had evidence of hypogonadism. In age-adjusted models, testosterone therapy use in all case periods was associated with a higher risk of VTE in men with (odds ratio [OR], 2.32; 95% CI, 1.97-2.74) and without (OR, 2.02; 95% CI, 1.47-2.77) hypogonadism. Among men without hypogonadism, the point estimate for testosterone therapy and VTE risk in the 3-month case period was higher for men younger than 65 years (OR, 2.99; 95% CI, 1.91-4.68) than for older men (OR, 1.68; 95% CI, 0.90-3.14), although this interaction was not statistically significant (P = .14).
Conclusions and Relevance
Testosterone therapy was associated with an increase in short-term risk for VTE among men with and without hypogonadism, with some evidence that the association was more pronounced among younger men. These findings suggest that caution should be used when prescribing testosterone therapy.
The clinical indication for testosterone therapy is primarily to treat hypogonadism, a condition in which serum testosterone levels in men decrease below a specific threshold, resulting in sexual dysfunction, altered bone metabolism and body composition, and potential emotional dysregulation.1-4 Although testosterone levels may decrease with age, external causes of clinical hypogonadism include genetic diseases or complications from surgery, infection, and medications.4 Testosterone prescriptions among men increased more than 300% from 2001 to 20135,6; the increase is thought to be caused by testosterone therapy being prescribed for common symptoms, such as low libido and fat redistribution, associated with aging, obesity, and diabetes and not necessarily with clinical hypogonadism.7 This increase in prescription rate was more pronounced among men aged 18 to 45 years than among older men.5 In 2014, the US Food and Drug Administration released a warning about testosterone therapy and the potential risk of heart attack and stroke; since then, testosterone therapy prescriptions have decreased and eventually plateaued.8,9 Recent trends estimate that, in the United States, 2.3 million men older than 30 years (3.2%) were prescribed testosterone therapy in 2013 and that this trend decreased to approximately 1.15 million men (1.6%) in 2016.6,10 Evidence suggests that testosterone therapy is still being prescribed to men without hypogonadism.7
Venous thromboembolism (VTE), consisting of deep vein thrombosis and pulmonary embolism, is a common condition in the United States, with more than 1 million individuals experiencing a VTE annually.11 Baseline testosterone levels are not associated with increase in VTE risk.12 However, exogenous testosterone therapy may increase endogenous hematocrit levels, which can increase blood viscosity, platelet accumulation, and thromboxane A2 concentrations for up to 6 months and could subsequently increase risk of blood clot formation and subsequent VTE events.13-15 Testosterone therapy is most commonly administered via transdermal gels, patches, or intramuscular routes, each having their own rate of absorption and prescription strengths that potentially affect cardiovascular pathophysiologic factors.16 Pathophysiologic research suggests that exogenous testosterone therapy could increase VTE risk, but the 2 largest observational studies17,18 evaluating this association reached conflicting conclusions. Furthermore, these studies17,18 were underpowered to examine testosterone therapy use within important clinical subgroups, such as by clinical hypogonadism status,19 age, route of testosterone therapy exposure, and duration of testosterone therapy use.
Using a case-crossover design, this study tested our primary hypothesis that exposure to testosterone therapy is associated with increased risk of incident VTE among men stratified by clinical hypogonadism status. Secondary analyses stratifying by age group (<65 years vs ≥65 years), route of testosterone therapy exposure, and different case periods were performed to assess various risk profiles.
This case-crossover study included data obtained using inpatient and outpatient medical claims provided from the IBM MarketScan Commercial Claims and Encounter Database and the Medicare Supplemental Database (IBM Watson Health) from January 1, 2011, through December 31, 2017. The databases contain health care claims information from US employers, health plans, hospitals, and Medicare programs. Data on all enrollment records and inpatient, outpatient, ancillary, and drug claims are collated and linked via individual-level identifiers. The MarketScan databases are compliant with the Health Insurance Portability and Accountability Act, and all data are deidentified. As such, per MarketScan operational procedure, enrollee consent was not required or obtained. This study was deemed to be exempt according to the institutional review board process for the University of Minnesota.
The initial cohort included 93 205 men aged 18 to 99 years with at least 1 inpatient or 2 outpatient claims for VTE 7 to 184 days apart (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] and International Statistical Classification of Diseases, Tenth Revision, Clinical Modification [ICD-10-CM] codes provided in eTable 1 in the Supplement), with 1 corresponding anticoagulant prescription within 31 days of the initial VTE date. The positive predictive value was 91% in a validation study that used a similar definition, incorporating both inpatient and outpatient ICD-9-CM codes and requiring evidence of treatment.20-23
After restriction of the sample to men with at least 12 months of continuous enrollment before their VTE, 52 203 men remained. A total of 12 581 men with prevalent cancer were excluded, and the final analytic sample included 39 622 men with 12 months of follow-up before their VTE. For all men, only the incident VTE event was considered in these analyses.
The US Endocrine Society defines hypogonadism as “consistently low serum total testosterone and/or free testosterone concentrations”24(p 1716) with trademark symptoms (loss of body hair, small testis size, and delayed sexual development). MarketScan data do not contain specific laboratory values of tests or detailed information about patient symptoms. Therefore, hypogonadism for the main analysis was defined by the ICD-9-CM codes 257.xx and the ICD-10-CM codes E29.x and E89.5, which capture an array of hypogonadism diagnoses, including postirradiation and postsurgical diagnoses. These codes were based on a previous study17 that used claims data that controlled for hypogonadism status and a review of the diagnosis code descriptions. A sensitivity analysis was performed defining hypogonadism using a different set of ICD-9-CM and ICD-10-CM codes spanning various testicular, pituitary, and hypothalamic disorders used in an analysis performed by Jasuja et al.25
A case-crossover study design was used in which each man with VTE served as his own control. We defined exposure case periods of 6 months, 3 months, and 1 month before the incident VTE event, with equivalent exposure control periods (6 months, 3 months, and 1 month) starting 6 months before the incident VTE event (Figure, A). Various case period lengths were selected to assess the time frame in which VTE events could be triggered after testosterone prescription exposure. Because each case patient serves as his own control, the case-crossover design mitigates confounders that are time invariant during a patient’s observation period.
The main exposure of interest was a received drug claim for testosterone therapy 1 month, 3 months, or 6 months before an individual’s first VTE event. The case periods were defined based on the time that testosterone therapy is thought to affect the previously mentioned pathophysiologic factors known to increase VTE risk. We defined 612 different formulations for testosterone therapy, including a variety of prescription brands, strengths, and routes using a therapeutic detail code (6808010090) unique to MarketScan databases (RED BOOK, IBM Watson Health). Route of exposure was classified as transdermal (includes both gel and patch methods of application) or intramuscular injection. A sensitivity analysis was performed that defined testosterone therapy use as having more than 1 testosterone therapy prescription fill within the 3-month and 6-month case periods.
The presence of a testosterone prescription 1, 3, and 6 months before the incident VTE date was compared with the equivalent control periods that occurred 6 months before the VTE date. Conditional logistic regression was used to estimate the odds ratios (ORs) and corresponding 95% CIs for the risk of VTE associated with testosterone use in the case period compared with the control period. Hospitalization has been found to be associated with VTE,26,27 and the number of medical encounters may be an indicator of general health. As such, in multivariable models, we controlled for the number of inpatient hospitalizations, outpatient visits, and emergency department hospitalizations that occurred in the case and control periods.
To explore multiplicative interaction associations, we conducted the same overall analyses stratified by age category (<65 vs ≥65 years) and route of testosterone (transdermal vs intramuscular). Multiplicative interactive associations were formally tested by adding an interaction term in the original logistic regression models. An exploratory analysis was conducted that classified testosterone therapy use differently to look for lagged associations. Time windows of less than 1, 1 to 3, or 3 to 6 months before the first VTE event and the corresponding time windows for the control period 6 months before the VTE were used for all analyses (Figure, B). Use of testosterone therapy in the 3- to 6-month case period was used as a reference category (instead of no testosterone therapy use) to detect different risk profiles from the 1- to 3-month and 1-month case periods. All statistical analyses were performed with SAS statistical software, version 9.4 (SAS Institute Inc). A 2-sided test for P < .05 was used as the significance threshold.
A total of 39 622 men (mean [SD] age, 57.4 [14.2] years) were enrolled in the study, with 3110 men (7.8%) diagnosed with hypogonadism and 29 182 men (73.7%) younger than 65 years. Among the 36 512 men without hypogonadism, 374 (1.0%) were ever prescribed testosterone in the year before their VTE event. Among the 3110 men with hypogonadism, 1330 (42.8%) were prescribed testosterone in the 12 months before their VTE event. Compared with men without evidence of hypogonadism, men with hypogonadism had more outpatient encounters in the year before their VTE (Table 1). All analyses were stratified by hypogonadism status; primary results for men without hypogonadism are provided in Table 2 and for men with hypogonadism in Table 3.
Among the men without hypogonadism, testosterone prescription was more common in the 6 months immediately before the VTE event (case period; n = 294) than in the control period (n = 177) (Table 2). Use of testosterone therapy in the case period was associated with a doubling of VTE risk when using the 1-month case period (OR, 1.96; 95% CI, 1.24-3.10), 3-month case period (OR, 2.46; 95% CI, 1.71-3.53), and 6-month case period (OR, 2.02; 95% CI, 1.47-2.77) compared with testosterone therapy use in the equivalent control periods after adjusting for the number of inpatient hospitalizations, outpatient encounters, and emergency department encounters. eTable 2 in the Supplement provides additional information on the cross-classification of testosterone therapy use in the case and control periods for all period lengths.
Stratified analyses were also conducted to explore potential interactive associations by age category and route of testosterone therapy exposure (Table 4). Although the interactions were not statistically significant, men younger than 65 years without hypogonadism using testosterone therapy in the 6-month (OR, 2.33; 95% CI, 1.60-3.40) and 3-month case periods (OR, 2.99; 95% CI, 1.91-4.68) had greater VTE risk compared with men 65 years and older using testosterone therapy in the 6-month (OR, 1.41; 95% CI, 0.77-2.56) and 3-month case periods (OR, 1.68; 95% CI, 0.90-3.14). Estimates were similar when comparing the 1-month testosterone therapy use with the control period for both age groups. No significant interactions were apparent when analyses were stratified by route of testosterone therapy (transdermal vs intramuscular).
Sensitivity analyses using the alternative hypogonadism definition yielded similar association estimates (eTable 3 in the Supplement). Magnitudes of association were larger in the sensitivity analysis that required men to have more than 1 testosterone prescription fill (eTable 4 in the Supplement), with ORs of 2.46 (95% CI, 1.38-4.39) for the 6-month case period and 3.75 (95% CI, 1.95-7.22) for the 3-month case period.
In exploratory analyses, we considered exposure periods of less than 1 month, 1 to 3 months, and 3 to 6 months. Use of testosterone therapy in the less than 1-month and 1- to 3-month case periods was associated with approximately a doubling in the risk of VTE (<1-month OR, 1.96; 95% CI, 1.24-3.10; 1- to 3-month OR, 2.24; 95% CI, 1.42-3.54); however, no association was found in the 3- to 6-month period (OR, 0.99; 95% CI, 0.65-1.53). Risk in the less than 1-month and 1- to 3-month case periods (OR, 1.87; 95% CI, 1.01-3.46) was statistically significantly different from that in the 3- to 6-month case period (OR, 2.08; 95% CI, 1.17-3.70).
Among men with hypogonadism, testosterone therapy use was also higher in the total 6-month case period (n = 1069) compared with the control period (n = 697) (Table 3). Testosterone prescriptions in the 1-month (OR, 1.66; 95% CI, 1.34-2.04), 3-month (OR, 2.28; 95% CI, 1.91-2.72), and 6-month (OR, 2.32; 95% CI, 1.97-2.74) case periods were also associated with approximately double the risk of VTE compared with the control periods 6 months earlier after adjusting for confounders. There was no evidence of interactive associations by age group or route of testosterone therapy exposure (Table 4).
Sensitivity analyses yielded higher adjusted estimates when an alternate definition of hypogonadism was used (eTable 3 in the Supplement). In the other sensitivity analysis with testosterone therapy use defined as having filled more than 1 prescription, magnitudes of association were larger (eTable 4 in the Supplement). This sensitivity analysis yielded ORs of 2.95 (95% CI, 2.34-3.71) for the 6-month case period and 3.34 (95% CI, 2.42-4.60) for the 3-month case period.
In the exploratory analyses, we also analyzed the exposure periods of 1 to 3 months and 3 to 6 months. Among men with hypogonadism, the adjusted ORs were 1.96 (95% CI, 1.60-2.41) for the 1- to 3-month case period and 1.33 (95% CI, 1.06-1.66) for the 3- to 6-month case period. Risk in the 1- to 3-month case period was statistically significantly different from that in the 3- to 6-month case period (OR, 1.49; 95% CI, 1.12, 1.97), whereas risks in the less than 1-month vs 3- to 6-month case periods were not statistically significant (OR, 1.31; 95% CI, 0.98-1.76). Exploratory analysis results stratified by age category and testosterone therapy route are provided in eTable 5 in the Supplement.
Using a large medical claims database and a case-crossover approach, this study found that men without cancer prescribed testosterone therapy had approximately twice the risk of VTE within the 1-, 3-, and 6-month case periods compared with the equivalent control periods 6 months earlier. The association was stronger among men younger than 65 years compared with men who were older; however, these differences were not statistically significant. Associations between testosterone therapy and risk of VTE did not differ whether the route of testosterone therapy was transdermal or intramuscular. In addition, our study provides novel information about the time frame of exposure most pertinent to risk, with risk being greatest in the first 3 months after testosterone therapy initiation. These findings complement prior work8,28,29 suggesting that testosterone therapy use is associated with greater risk of stroke and myocardial infarction and suggest that practitioners should be cautious when prescribing testosterone therapy.
Prior observational studies about testosterone therapy and VTE are scarce and have been inconclusive about whether testosterone therapy affects VTE risk. A prior case-control study by Martinez et al18 also stratified by hypogonadism status and compared 19 000 men with VTE with 900 000 controls. Only 21 men without hypogonadism were current testosterone therapy users, and testosterone therapy use was associated with a 1.69-fold (95% CI, 1.09-2.63) increased risk of VTE. Among those with hypogonadism, 48 were receiving testosterone therapy, and current testosterone therapy use was associated with a relative risk of 1.08 (95% CI, 0.75-1.55). In addition, Baillargeon et al17 performed a case-control study of 30 000 men and 7500 VTE events identified from a large claims database in the United States (OptumInsight). The investigators found no significant association in that analysis, in which 158 VTE events among testosterone therapy users occurred and the focus was on testosterone use in the 15 to 60 days before the VTE event. That study used a traditional matched case-control design, and unmeasured time-invariant confounders, such as smoking status and physical activity, could not be directly addressed in the analyses. Our study corrects for this unmeasured confounding with the case-crossover design and appears to improve on the article by Baillargeon et al17 by having a larger number of exposed cases (374 testosterone therapy users without hypogonadism and 1330 testosterone therapy users with hypogonadism) to analyze different case periods, hypogonadism status, and age categories. Findings from smaller observational studies and randomized clinical trials15,30-33 have been inconsistent, with some reporting testosterone therapy to be a risk factor and others reporting no association. Other studies6,17,31 recruited only hypogonadal men or either controlled for or matched men by hypogonadism status but did not investigate the association of testosterone therapy on the large population of men without hypogonadism who receive it.
Furthermore, our study provides novel information about the time frame of exposure to testosterone therapy and risk of VTE. Our exploratory analysis found evidence that risk estimates were higher in the 1- to 3-month and less than 1-month case periods compared with the 3- to 6-month case periods. This observation is analogous to patterns seen for oral contraceptives and VTE risk, whereby women are at greatest risk shortly after beginning oral contraceptive use, and provides hypothesis-generating information for future research on testosterone therapy and VTE.34
Hypogonadism diagnosis, age, and route of testosterone exposure were identified a priori as subgroups of particular interest. Among men without hypogonadism, risk was greater among men younger than 65 years compared with those who were older, although the interaction was not statistically significant. This finding is of potential public health importance because of recent trends in increased testosterone therapy prescriptions among men younger than 65 years.5 Men younger than 65 years who begin testosterone therapy early to deter common health symptoms potentially associated with aging may be exposing themselves to greater risk of early-onset cardiovascular disease outcomes, especially if there is no clinical indication for starting therapy.
In the present analysis, VTE risk did not vary by route of exposure. The article by Baillargeon et al,17 which reported no overall association of testosterone therapy with VTE risk, also did not find differences when transdermal and intramuscular testosterone therapy were evaluated separately. Likewise, a meta-analysis of randomized clinical trials by Borst et al,33 with relatively small sample sizes and short follow-up times, also reported no overall association of testosterone therapy with cardiovascular risk and found no difference when transdermal and intramuscular testosterone therapies were evaluated. Venous thromboembolism was not specifically examined as an outcome. At present, there is no compelling evidence that the association of the risk of VTE with testosterone therapy varies according to route of exposure.
A strength of this study is the use of a large administrative claims data. To our knowledge, we had the largest number of cases included in a study evaluating testosterone therapy as a potential VTE risk factor. One of the strengths of the case-crossover design is its ability to account for time-invariant unmeasured confounding. With patients acting as their own controls, relatively static or slowly varying factors (eg, obesity, smoking status, and family history of VTE) are assumed to remain constant during the case and control periods.35 This design can be particularly useful in the context of administrative data with information on important potential covariates (eg, obesity or family history) not available in the data set. An assumption of the case-crossover design is that exposures are transient. Our exploratory analysis showed that the association of testosterone therapy with VTE was transient because the association was attenuated within the 3- to 6-month exposure category compared with the 1-month and 1- to 3-month categories.
Use of a large administrative claims database was also a limitation of this study. Other common weaknesses, such as misclassification and time-variant confounding found in claims databases, were minimized as best as possible given the data available. The VTE definition that we used had a good positive predictive value (91%) when validated in a different study population.23 We are not aware of any validation studies of hypogonadism diagnosis or testosterone therapy prescription claims. Because there is no consensus regarding how to define hypogonadism in administrative data, we used 2 different definitions: one similar to that used by Martinez et al18 and Baillargeon et al17 and another similar to that used by Jasuja et al.25 The estimates were similar regardless of which definition was used. Studies validating algorithms to define hypogonadism in administrative data are needed. The testosterone therapy codes used in this study were the same as those used by another claims-based study17 and resulted in prevalence that matched current trend analysis numbers.5 Pharmacy claims reflect whether patients fill their prescriptions; however, whether the patient takes the prescription cannot be assessed in administrative data. In addition, despite the large data set, precision of some stratified analyses was not optimal because of the sparsity of men prescribed testosterone therapy in subcategories.
Men without cancer who were prescribed testosterone therapy in our study’s case period had an increased risk of incident VTE compared with that in an equivalent control period 6 months before the VTE. This risk of VTE was present for men with and without hypogonadism who received a testosterone therapy prescription. These data combined with prior data suggest that future clinical trials of testosterone therapy, regardless of the indication, should capture VTE events as part of safety end points. Relative VTE risk may be exacerbated for men younger than 65 years using testosterone therapy with and without clinical indications. Men experiencing common symptoms that result from natural aging have considered testosterone therapy as a treatment; however, men without hypogonadism should assess cardiovascular disease risk with their physicians before prescription to minimize adverse cardiovascular outcomes.
Accepted for Publication: September 3, 2019.
Corresponding Author: Rob F. Walker, MPH, Division of Epidemiology and Community Health, School of Public Health, University of Minnesota, 1300 S Second St, Ste 300, Minneapolis, MN 55413 (email@example.com).
Published Online: November 11, 2019. doi:10.1001/jamainternmed.2019.5135
Author Contributions: Mr Walker and Dr Lutsey had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Walker, Zakai, MacLehose, Cowan, Lutsey.
Acquisition, analysis, or interpretation of data: Walker, Zakai, Cowan, Adam, Alonso, Lutsey.
Drafting of the manuscript: Walker, Cowan, Lutsey.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Walker, MacLehose, Cowan.
Obtained funding: Lutsey.
Administrative, technical, or material support: MacLehose, Cowan, Adam, Lutsey.
Supervision: Zakai, Lutsey.
Conflict of Interest Disclosures: Mr Walker and Drs Adam and Lutsey reported receiving grants from the National Heart, Lung, and Blood Institute, National Institutes of Health during the conduct of the study. Drs Zakai and Alonso reported receiving grants from National Institutes of Health during the conduct of the study. No other disclosures were reported.
Funding/Support: This study was funded and supported by grant R01-HL131579 (Mr Walker and Drs Adam and Lutsey) from the National Heart, Lung, and Blood Institute, National Institutes of Health.
Role of the Funder/Sponsor: The funding source 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; and decision to submit the manuscript for publication.