One number, and three reasons to be careful with it. In 689 men with cardiovascular disease, a total testosterone at or below 300 ng/dL came with 1.48 times the rate of death from any cause over a median of 55 months. The same analysis found nothing in women, it cannot show cause, and the low reading may be a sign of being unwell rather than the reason for it.
You are somewhere between forty and sixty. There has been a stent, or a scare, or a cardiologist who now knows your first name. Your energy is flat, your waist is thicker than it used to be, and your sex drive has gone quiet. Someone told you to check your testosterone. You want to know whether that is medicine or marketing.
A 2025 analysis of American national survey data gives part of an answer. It is worth reading properly, including the parts that sell nothing.
What the study did
Researchers pulled 1,177 adults with cardiovascular disease out of four cycles of the National Health and Nutrition Examination Survey: 1999 to 2000, 2003 to 2004, 2011 to 2012, and 2013 to 2014. Mean age was 66.01 years. 689 of them were men. Deaths were tracked for a median of 55 months, with an interquartile range of 44 to 71 months.
In men, low testosterone was defined as 300 ng/dL or below. After adjusting for age, SHBG, income ratio, ethnicity, education, hypertension and cardiovascular treatment, the men below that line died at 1.48 times the rate of the men above it. The 95 percent confidence interval ran from 1.08 to 2.02. The p value was 0.013.
Seven out of ten men were already below the line
487 of the 689 men, 70.68 percent, sat at or under 300 ng/dL. During follow up, 202 of the men died, 29.32 percent of them.
Stay with that prevalence for a moment. A marker present in seven men out of ten is not picking out some rare and unlucky subgroup. It is describing the ordinary condition of older men with heart disease.
The association held in younger men, but only just
Split by age, it survived in both halves. In men aged 65 and over the hazard ratio was 1.45, interval 1.09 to 1.93, p value 0.010. In men under 65 it was 1.39, interval 1.01 to 1.90, p value 0.041.
Look at that lower bound of 1.01. That is about as close to no effect at all as a statistically significant result can get. Split instead by blood pressure, the estimate was 1.35 in the men with hypertension, interval 1.06 to 1.71, and 1.93 in the men without it, interval 1.25 to 2.96. Hypertension was present in 485 of the 689 men, so the bigger looking number came from the smaller group.
In women it was not there at all
488 women were included, and 394 of them, 80.74 percent, fell below the threshold the authors used for women, 20 ng/dL. Before adjustment, low testosterone looked associated with death in women too: hazard ratio 1.53, interval 1.01 to 2.31, p value 0.042. After adjustment it disappeared: 1.23, interval 0.80 to 1.90, p value 0.341. Every female subgroup was null, by age and by hypertension.
The authors report that plainly. So should anyone who quotes them.
Low testosterone may be the smoke, not the fire
This is observational data. It can show that two things travel together. It cannot show that one causes the other. And there is a detail inside the paper that should slow anyone down before treating 1.48 as a verdict.
The men who died and the men who lived had almost the same testosterone. Median 343.06 ng/dL in those who died, 346.00 ng/dL in those who survived, p value 0.208. No meaningful difference in the raw number at all. What did differ was everything you would expect to differ. The men who died were 72.09 years old on average against 64.25. Their SHBG was higher, 53.75 against 46.61 nmol per liter. And 32.67 percent of them were on cardiovascular treatment, against 17.04 percent of the survivors.
That is a portrait of being older and sicker. Serious illness lowers testosterone on its own. A low reading in an unwell man of 72 can be a readout of his condition rather than a cause of what happens to him next. This design cannot separate the two, and it does not claim to.
Before you quote the 1.48. Almost as many men were dropped for missing testosterone data, 676 of them, as were kept in the analysis, 689. Cardiovascular disease was established by asking people whether a doctor had ever told them they had it. Testosterone was measured once, and levels move across the day. The authors also note that in a retrospective study it was difficult to adjust for confounders such as lifestyle changes.
What the paper does not license
The authors close by suggesting their results lend support to hormone replacement in cardiovascular disease. Their own data cannot carry that. Nobody in this analysis was treated as part of the study, and nobody was randomized. The largest randomized trial they cite enrolled 5,246 men with hypogonadism and found the therapy non-inferior to placebo. Non-inferior means not worse. It is a safety finding rather than a survival benefit, and those two get blurred together constantly.
What is actually useful here
A single total testosterone result carries less information than it appears to. In this study SHBG separated the men who died from the men who lived while total testosterone did not, and the paper cites work showing that total testosterone falls by around 0.8 percent a year in middle aged and older men while SHBG rises by 1.6 percent a year. The same figure on a lab slip does not mean the same thing at 60 as it did at 40. That is what a full panel is for.
If you already have cardiovascular disease, a testosterone number is one line in a much longer story that includes your age, your blood pressure and your treatment. Reading it inside that story, rather than on its own, is the work of a consult with a specialist.