AndroGel Real-World Evidence: What Registries and Observational Data Show

What kind of real-world data exists for testosterone gel?

The evidence falls into three groups: a randomized cardiovascular outcomes trial run in a high-risk older population (TRAVERSE), large observational cohorts built from Veterans Affairs records, and insurance-claims analyses of who keeps filling prescriptions. TRAVERSE used AndroGel 1.62% [5][4]; the claims analysis covered topical testosterone gels; the VA cohorts studied testosterone treatment in general and did not single out gel users.

TRAVERSE: the biggest controlled dataset in men using gel

TRAVERSE enrolled 5,246 men aged 45 to 80 with symptoms of hypogonadism, two fasting testosterone levels below 300 ng/dL, and existing cardiovascular disease or high risk of it. They were randomly assigned to daily 1.62% testosterone gel, dose-adjusted to keep testosterone between 350 and 750 ng/dL, or placebo gel [4]. Mean treatment lasted 21.7 months and mean follow-up 33 months [4].

The primary endpoint (cardiovascular death, nonfatal heart attack or nonfatal stroke) occurred in 182 men (7.0%) on testosterone and 190 (7.3%) on placebo, hazard ratio 0.96 (95% CI 0.78 to 1.17). Noninferiority required the upper end of that interval to stay below 1.5, and it did [4]. FDA reviewed the results and concluded TRAVERSE did not demonstrate a new safety signal or trend in cardiovascular outcomes [5].

The VA cohorts: lower death rates, and how much to trust them

Shores and colleagues followed 1,031 male veterans over 40 with total testosterone of 250 ng/dL or less and no history of prostate cancer; 398 started testosterone during routine care. Mortality was 10.3% in treated men versus 20.7% in untreated men, and after adjustment for age, body mass index, testosterone level, medical morbidity, diabetes and coronary disease, treatment was associated with a hazard ratio for death of 0.61 (95% CI 0.42 to 0.88) [2]. The authors cautioned that residual confounding may still be a source of bias [2].

The larger VA analysis examined 83,010 men with documented low testosterone and sorted them by whether testosterone replacement brought testosterone back to normal. Compared with men who received no treatment, normalizing testosterone was associated with lower all-cause mortality (HR 0.44, 95% CI 0.42 to 0.46), heart attack (HR 0.76, 0.63 to 0.93) and stroke (HR 0.64, 0.43 to 0.96) [3]. Men treated without reaching normal levels showed no difference in heart attack or stroke risk compared with untreated men [3].

Neither study assigned treatment at random or reported results for gel users alone [2][3]. TRAVERSE is the different design, with a randomized, blinded comparison [4].

How many men stay on a gel?

Schoenfeld and colleagues used the MarketScan claims database to follow 15,435 hypogonadal men who filled an initial topical testosterone prescription in 2009. By 6 months, 34.7% were still on therapy; by 12 months, 15.4%. Persistence ended at the first gap of more than 30 days between prescriptions [1]. Adherence looked similar for AndroGel and Testim, did not differ by age group, and about half of the men who stopped later restarted, mostly on the same product and dose [1].

Hematocrit: how gel compares with injections

Hematocrit is the safety measure most consistently tied to formulation. A network meta-analysis of 29 placebo-controlled trials covering 3,393 men found mean hematocrit rose with every formulation: 3.0% with gel (95% CI 1.8 to 4.3), 4.0% with intramuscular enanthate or cypionate (2.9 to 5.1), 4.3% with oral testosterone undecanoate, 1.6% with intramuscular undecanoate and 1.4% with patch [6]. Intramuscular enanthate or cypionate rose significantly more than patch; the other comparisons, including gel against intramuscular enanthate or cypionate, were not statistically significant [6].

A single-center retrospective review of 178 men treated for three years found erythrocytosis (hematocrit above 50%) in 12.8% of gel users, 35.1% of pellet users and 66.7% of injectable users. The injectable group was younger (mean 42.5 versus about 54 years) and started with higher hematocrit, so the groups were not comparable at baseline [8]. Reviews of the topic reach the same ordering, with short-acting injectables carrying the highest incidence [9], and one puts the risk of hematocrit above 0.52 at 315% higher than in controls [10].

PSA and prostate findings

A meta-analysis of placebo-controlled testosterone trials found the overall PSA difference after starting treatment was 0.10 ng/mL (95% CI -0.28 to 0.48), and the summary relative risk of prostate cancer across 11 trials was 0.87 (95% CI 0.30 to 2.50), an interval wide enough to include both harm and benefit [11]. Separately, prospective cohort studies showed no association between a man's own testosterone level and prostate cancer (relative risk 0.99 per 5 nmol/L increase) [11].

What FDA says about testosterone products

FDA's class-wide review found that ambulatory blood pressure studies confirmed a blood pressure increase with all testosterone products [7]. FDA also says none of the approved testosterone products are approved for men with low testosterone who lack an associated medical condition [5]. In June 2026 FDA requested that labels remove the limitation of use stating that safety and efficacy in age-related hypogonadism have not been established, and revise prostate cancer and benign prostatic hyperplasia safety information [5].