Despite research suggesting that multivitamins do little for reasonably well-fed Americans, the question is not settled say researchers from the National Institutes of Health.

A new analysis of deaths from heart disease over more than 20 years finds that women who took multivitamin-mineral supplements for three years or more were significantly less likely to die.

The authors urge caution because the benefit was not seen among men using the supplements long-term, among women taking them for less than three years or in anyone taking just multivitamins without minerals.

“It’s way too early to really know if use of multivitamin-minerals over time reduces the risk of cardiovascular mortality (or any other health problem or cause of death) in women,” Regan Bailey, who led the study, told Reuters Health in an email.

“Our study suggests the possibility that there is such a connection for women, which is an interesting finding but to know for sure would require a clinical trial,” said Bailey, a registered dietitian and researcher at the NIH Office of Dietary Supplements.

According to that office, multivitamin-mineral (MVM) products account for almost one-fifth of all purchases of dietary supplements and more than 40% of all sales of vitamin and mineral supplements. More than one-third of Americans take MVMs, spending about $ 5.5 billion each year on them.

Bailey said many health behaviors, such as eating nutritiously, exercising regularly and not smoking clearly reduce the risk of developing heart disease and dying from it.

“Taking a multivitamin-mineral should not be a replacement for any of these behaviors,” she said. “And it’s not clear whether taking a multivitamin-mineral added to a healthy lifestyle is going to provide an additional health advantage.”

For their study, online now in The Journal of Nutrition, Bailey’s team analyzed data from the National Health and Nutrition Examination Survey III, conducted between 1988 and 1994.

Information on multivitamin or multivitamin-mineral use was available for almost 9,000 healthy adults, who were 40 years and older when they answered the surveys. About 21% of those participants who took dietary supplements used MVMs - defined as three or more vitamins plus one or more minerals – and 14% used multivitamins without minerals.

The researchers followed the participants for an average of 18 years to find out how many died of cardiovascular disease.

Women who had been taking multivitamin-mineral dietary supplements for at least three years when they answered the survey had a 35% lower risk of dying from heart disease over the following decades when compared to women who did not use the supplements.

The results held after researchers adjusted for other factors that could influence the womens’ heart risks, including weight, blood pressure, cholesterol, alcohol or aspirin use, blood sugar control, education, age and race.

They acknowledge that it’s possible there were other aspects of health and behavior by women who took MVMs over the long-term that the study could not account for.

Another limitation, the study team writes, is only having responses about supplement use at the beginning of the study.

“We can’t be sure that the people continued their same supplementation practices over the average 18 years of follow-up,” Bailey said.

The U.S. Preventive Services Task Force, a government-backed independent panel, concluded in 2013 that there isn’t enough evidence showing a heart benefit from multivitamin products to recommend taking them. Bailey and her coauthors note that the panel only had two randomized controlled trials on which to base their conclusions.

Dr. Benjamin Baechler, a physician at the University of Minnesota in Minneapolis who practices integrative medicine, called the new study’s results thought provoking.

“When it comes to a daily supplement, this study suggests making a distinction between a product that only has vitamins and those that also have minerals,” he said.

Dr. Saverio Stranges, from the University of Warwick Medical School in Coventry, UK, said that stratifying the findings by gender was fine because men and women have different biology to some degree, but added that observational studies such as this one cannot prove cause and effect.

These findings are “not really in line with trial evidence which does not support the use of supplements for the purposes of cardiovascular disease prevention,” said Stranges, who has studied the use of MVMs.

Stranges said surveys show that supplement users in the U.S. tend to be better educated about healthy lifestyle habits so it’s possible that women who use MVM long term are just healthier in general.

“So this could also be a marker for other healthy behaviors which actually reduce the risk of cardiovascular mortality,” he said.