Giving heart attack survivors vitamin D3 and adjusting the dose until their blood level cleared a target produced fewer major heart events, but not by enough for the trial to call the difference real. In the TARGET-D trial published September 7 in the European Heart Journal, researchers at Intermountain Medical Center Heart Institute randomized 630 patients to usual care or targeted vitamin D and followed them an average of 4.2 years. Major adverse cardiovascular events, a combined count of death, heart attack, heart failure hospitalization, and stroke, occurred in 15.7% of the vitamin D group and 18.4% of the usual care group. That is 2.7 fewer events per 100 patients treated, or about 1 event avoided for every 37 people treated, and roughly a 15% lower relative risk (hazard ratio 0.85). The trial missed statistical significance (P = 0.40). That means it failed to prove the benefit, not that it disproved one.
Key takeaways
- The main endpoint missed significance, but the trial did not rule benefit out. The true effect very likely sits anywhere from 42% lower to 24% higher (95% confidence interval 0.58 to 1.24), a range that includes a benefit worth having.
- With 630 patients and about 104 total events, the trial could only confirm a large effect. A 15% reduction was never within its reach.
- Repeat heart attacks were lower with vitamin D: 3.8% versus 7.9%, or 4.1 fewer per 100 patients, about 1 avoided for every 24 treated. That was one of four secondary endpoints, so it is a lead worth testing, not a result to act on.
- Deaths were the same in both groups: 8.9% with vitamin D and 9.2% with usual care (hazard ratio 0.98, P = 0.95).
What the study found
TARGET-D enrolled patients between April 2017 and May 2023 and closed follow-up on March 17, 2025, once at least 104 primary events had occurred. Participants had a median age of 63 and 78.1% were men. Their vitamin D levels at the start were low: a median 25-hydroxyvitamin D of 25 ng/mL, with 87.0% at or below 40 ng/mL. More than half, 52.4%, started on 5,000 IU of vitamin D3 daily, far above the usual over-the-counter dose.
The design was the point. The treatment arm did not get a single fixed pill. Doses were titrated on an algorithm to reach and hold a blood level above 40 ng/mL, up to 80 ng/mL.
On the four secondary endpoints, the results split. Repeat heart attack was lower with vitamin D, 3.8% versus 7.9%, about 52% lower in relative terms (hazard ratio 0.48) with about a 3 in 100 chance of coincidence (P = 0.03). In absolute terms that is 4.1 fewer heart attacks per 100 patients over the average 4.2 years, roughly 41 per 1,000, or about 24 people treated for one avoided heart attack.
The other three moved the wrong way or nowhere. Heart failure hospitalization was 4.2% versus 3.5%, about 21% higher (hazard ratio 1.21, P = 0.65). Stroke was 1.6% versus 0.9%, about 69% higher (hazard ratio 1.69, P = 0.47). Neither difference was statistically meaningful, and both were built on very few events.
Why the main result missed statistical significance
A P value of 0.40 is easy to read as “vitamin D did nothing.” That is not what it says. It says this particular trial could not separate the difference it saw from chance, and there are specific, unglamorous reasons why.
The trial was small, and significance runs on events, not patients. TARGET-D stopped once about 104 primary events had accumulated across both arms. An event count that size gives a trial a good chance of catching a very large effect and very little chance of catching a modest one. To confirm a 15% reduction with any confidence, you need thousands of patients and several hundred events. The confidence interval shows the consequence directly: the data are compatible with anything from a 42% reduction to a 24% increase. That is not a precise measurement of zero. It is an imprecise measurement of something.
The composite endpoint diluted the one signal that showed up. The primary endpoint bundled death, heart attack, heart failure hospitalization, and stroke into a single count. Vitamin D moved one of those four, repeat heart attack, and left the other three flat or slightly worse on a handful of events each. Composite endpoints exist to accumulate events faster in a small trial, and the price is that a real effect on one component gets averaged against three components where nothing is happening. A trial with recurrent heart attack as its primary endpoint would have been judged on a much cleaner number.
Titration takes time, and the clock started before the levels did. Nobody in the treatment arm hit the target blood level on day one. Reaching and holding a 25-hydroxyvitamin D above 40 ng/mL takes repeat testing and dose escalation over months. Follow-up was counted from randomization, so an early stretch of each patient’s time on trial was time not yet at the intended level. Any effect that depends on sustained exposure gets watered down by that lead-in.
The comparison arm was usual care, not placebo. Vitamin D sells over the counter in every pharmacy in the country. Nothing stopped patients assigned to usual care from taking it on their own, and in a population where 87% started at or below 40 ng/mL, some almost certainly did. Every control patient who supplements narrows the real gap between the groups and pulls the measured result toward no difference.
Modern care after a heart attack is already good. Patients in both arms received guideline treatment, and that keeps baseline event rates low. Fewer events means a supplement has less room to demonstrate anything, and it means the trial needed to be larger, not smaller, to detect the effect it was looking for.
Could vitamin D still help, and how would we find out?
It could. The honest reading of TARGET-D is that the question is still open, and the trial that would close it looks different from this one in five ways.
- Placebo controlled and blinded, so the comparison group is truly unsupplemented rather than free to buy the intervention at the drugstore.
- Recurrent heart attack as the primary endpoint, since that is the outcome that moved here and the one with a plausible mechanism, instead of a four-part composite that buries it.
- Enrollment restricted to people with a documented deficiency, because correcting a level that is already adequate has nothing to correct, and diluting the trial with those patients costs power.
- Everyone brought to target before the follow-up clock starts, so what is measured is time at the intended blood level rather than time spent titrating toward it.
- Powered for a realistic effect size. A trial hunting a 15% to 20% reduction needs thousands of patients and years of follow-up. TARGET-D had 630 and about 104 events.
The authors have said they plan larger trials. That is the right response to this result, and it is a different response from either declaring vitamin D useless or declaring it proven.
Dr. Kumar’s take
For years, the defense of every null vitamin D trial has been the same: the dose was fixed, nobody measured blood levels, so of course it failed. TARGET-D removed that excuse. It measured, it titrated, it pushed levels high. What it could not do was enroll enough patients to answer the question it asked.
I want to be careful about how this gets read in both directions. A 15% reduction that misses significance in a 630-patient trial is exactly what a real but modest benefit looks like when you go hunting for it with too small a net. It is also exactly what nothing looks like. Both readings fit the data, and admitting that is more useful than picking the one you prefer.
The press release picked one. Intermountain led with the second heart attack result and the phrase “cut in half.” That number is real, but it is one of four secondary endpoints in a trial that stopped at about 104 total events, while stroke and heart failure both drifted the other way on a handful of events each. Run four comparisons and one landing at P = 0.03 is not remarkable on its own. Reporting the endpoint that went your way and skipping the two that did not is how an unproven trial gets sold as a win.
The accompanying European Heart Journal editorial by John W. Davis and JoAnn E. Manson is titled “More is not always better: vitamin D supplementation post-myocardial infarction in the Target-D trial.” Manson led VITAL, which randomized 25,871 adults to 2,000 IU of vitamin D3 daily or placebo and found no reduction in major cardiovascular events over a median 5.3 years (hazard ratio 0.97, P = 0.69). VITAL is the main reason I do not expect a large heart benefit from vitamin D. It was big enough to find a modest effect and it found nothing. The counterargument for TARGET-D is that VITAL gave a fixed dose to a mostly replete population, and that is a fair objection. It also narrows the open question to something specific: does correcting a genuine deficiency after a heart attack reduce a second one? Nobody has yet run a trial large enough to answer that.
What it means for you
If your vitamin D level is low, correcting it is reasonable for bone and muscle health. What this trial does not support is treating vitamin D as cardiac protection after a heart attack. What it also does not support is the opposite claim, that it definitely does nothing. Unproven is where this sits, and unproven is not the same as disproven.
Nothing here suggests harm from correcting a deficiency, and lead author Heidi May said the team observed no adverse outcomes at the higher doses. If a documented deficiency is worth correcting for other reasons, this trial gives no reason to avoid doing so. It just does not give a cardiac reason to do it.
The levers with real evidence behind them are less interesting and more effective. Home blood pressure monitoring cut heart attacks by a third in one large study, and nearly 9 in 10 US heart disease deaths are preventable through modifiable risk factors. A supplement is easier to swallow than that, which is exactly why the headline traveled further than the result.