How to Know What’s Most Likely to Save Lives

A window into the challenge of technical rigor

Some researchers insist that randomized controlled trials are the gold standard of evidence.

They're wrong.

RCTs are great, but for many health questions, an RCT isn't practical and isn't the most reliable source of knowledge. RCTs can't measure outcomes that require many years of follow-up. For rare diseases, there aren't enough patients. An RCT from years ago might not apply today.

Consider George Comstock, one of the best epidemiologists of the twentieth century. Comstock’s classic studies – including RCTs – showed that one in ten people infected with the tuberculosis bacteria develop active disease. Decades later, he wrote an article calling for studies on this very question. 

I puzzled about why he suggested that others repeat his studies. I was too embarrassed to ask him until many years later. 

His response: “That was then, this is now. The bacteria might have changed. The nutritional status of infected people might have changed. The inoculum [the number of bacteria that infect a person] might have changed. We won’t know unless we study it.” 

There’s no single, best approach to study health interventions. It’s not enough to understand the findings of a study; Comstock and specialists who really understand the nuance of issues see more deeply into methods and contexts and understand the need to interpret findings with caution. I discuss this more — and provide practical tools for technical rigor — in my forthcoming book, The Formula For Better Health: How to Save Millions of Lives — Including Your Own.

When Programs Provide Better Evidence Than Trials

RCTs are just one way to establish evidence-based programs. Public health often advances because of something very different: program-based evidence. 

One example of program-based evidence are studies led by Ed Mitchell at the University of Auckland School of Medicine. Many infants had died suddenly and unexpectedly — Sudden Infant Death Syndrome (SIDS). Many of these infants had been placed on their stomach to sleep — but maybe all babies were. To discover whether sleep position was the cause of the deaths, it was necessary to have a comparison group — infants who didn’t die. Mitchell’s group began a three-year case-control study to explore this and other possibilities.

A year into the study, the data wasn’t making sense to him. He told me when I spoke with him recently, “As a pediatrician, I couldn’t believe something as simple as placing babies prone could increase the risk of sudden death substantially. Basically thought it was rubbish.”

After he consulted with a prominent statistician a stunning conclusion became clear: Most SIDS deaths were attributable to putting babies on their stomach to sleep.

The evidence wasn’t definitive, but was strong enough for the New Zealand government to educate parents to put infants on their back to sleep. After New Zealand's back-to-sleep campaign, SIDS deaths then declined spectacularly, proving that back-sleeping prevents SIDS.

This is a great example of a situation in which an RCT would not have been ideal. An RCT would’ve taken years and been unlikely to result in a definitive outcome. SIDS is rare, adherence by parents to a randomized recommendation of back-sleeping would have been impractical, and it would have been unethical to randomize babies to prone-sleeping when this could increase the risk of SIDS. In this case, the program provided the evidence. New Zealand’s back-to-sleep campaign was implemented, rigorously evaluated, and proved what worked.

Data Analysis is an Art and a Science

Data analysis is an art and a science. Not all evidence is created equal. This relates to the levels of certainty I discussed in the post on why I take B12. But just because something is complicated doesn’t mean there isn’t a best way forward. Getting the right answer requires going beyond RCTs and beyond structured reviews; it requires assessment of each piece of evidence’s strengths, weaknesses, and relevance. 

One example: chlorthalidone, a medication to treat high blood pressure.

A few years ago, my organization, Resolve to Save Lives, recommended chlorthalidone as the preferred medicine in its class (diuretics) to treat hypertension. We made that call based on moderately strong evidence from a wide range of studies suggesting it might be more effective than the more common alternative, a medication called hydrochlorothiazide (HCTZ). 

Chlorthalidone costs a bit more, is not as available, and is less familiar to clinicians than HCTZ. Evidence leaned slightly in chlorthalidone’s favor, so we recommended it.

(Neither I nor my organization have ever taken, nor will ever take, any funding from drug companies. We recommend treatments and our independence has to be unquestionable. Both chlorthalidone and hydrochlorothiazide are low-cost, generic medications with many manufacturers.)

I was especially interested in a study launched by the U.S. Veterans Administration: a cluster-randomized trial comparing chlorthalidone to hydrochlorothiazide. It was clever, well-designed, and had the potential to settle the question as the first-ever direct comparison of the two drugs. I waited years for the results, fully expecting it would confirm our position.

But it didn’t.

The trial showed no difference—not in survival, not in rates of heart attack or stroke. I was glad we had always qualified our recommendation, noting that chlorthalidone might be better, but that the evidence wasn’t definitive.

Then I spoke to one of the best hypertension experts in the world. He dismissed the trial. He pointed out limitations—including trial duration, crossover design, maybe underpowering—and stood by his view that chlorthalidone was better. Our team looked at the data carefully. Another large study, involving hundreds of thousands of patients, also found no difference. Our conclusion? If chlorthalidone were much better, that difference likely would have shown up even with the study weaknesses.

Bottom line: We still don’t know for sure which is better, but we no longer recommend chlorthalidone over hydrochlorothiazide. The evidence just isn’t strong enough.

This isn’t just a story about two medicines. It’s a window into the challenge of technical rigor. You can do everything right—review the evidence, stay independent, wait for better data—and still find yourself unsure. Or wrong.

That’s why humility and clarity matter. Public health decisions almost always require action based on incomplete information.

We may never have perfect data, but we can still make excellent decisions.

This is not what you hear from politicians attacking vaccines or online influencers offering miracle cures. They make inaccurate statements with great certainty, and demand “proof”—but only when they want to block something. To promote misinformation, they cherry-pick factoids and cite poorly done or discredited studies. 

It’s easy to shout on social media. It’s much harder to look at the data carefully, acknowledge uncertainty, and still make the right call. But that’s what saves lives.

Whether it’s sudden infant death, high blood pressure, vaccinations, or the next health crisis, the best decisions come from careful analysis of the best available evidence—not ideology.

Because the cost of getting the science wrong isn’t a bad headline.

It’s a life lost.

Dr. Tom Frieden is author of The Formula for Better Health: How to Save Millions of Lives – Including Your Own.Dr. Tom Frieden is author of The Formula for Better Health: How to Save Millions of Lives – Including Your Own.

The book draws on Frieden's four decades leading life-saving programs in the U.S. and globally. Frieden led New York City's control of multidrug-resistant tuberculosis, supported India's efforts that prevented more than 3 million tuberculosis deaths, and led efforts that reduced smoking in NYC.

As Director of the CDC (2009-2017), he led the agency's response that ended the Ebola epidemic. Dr. Frieden is President and CEO of Resolve to Save Lives, partnering locally and globally to find and scale solutions to the world's deadliest health threats.

Named one of TIME's 100 Most Influential People, he has published more than 300 scientific articles on improving health. His experience is, for the first time, translated into practical approaches for community and personal health in The Formula for Better Health.