Making Sense of Medical Statistics: What Patients Should Do

  • Share
  • Read Later

Seung Kang died in a Philadelphia hospital in 2005. He was only 59, and just a week before his death he’d been feeling quite healthy. But a heart catheterization showed blocked vessels to his heart, and a cardiothoracic surgeon recommended immediate bypass surgery — Kang’s second open-heart surgery in two years.

Further tests revealed a possible snag. The right ventricle of Kang’s heart had attached to his breastbone, the sternum. “I think [the surgeon] quoted a risk of death around 5%,” says Bon Ku, Kang’s son-in-law. Sure enough, when the surgeons cut into the sternum they also cut into the ventricle that was attached to the bone. Kang died two days later.

“It’s crazy. A 5% risk is high,” Ku says now. But his wife and their family needed to make a decision about surgery quickly. “It wasn’t that big of a deal when the surgeon was telling us, and I just don’t think we asked those questions that we should have asked.”

Whether 5% seems like a big number or a small number, of course, depends a lot on the situation. But a growing body of research shows that presenting the same information in different ways, with different statistics, can lead people to very different decisions. We know that we don’t always fathom risk easily. More surprising, perhaps, is that — at least when it comes to some very commonly used risk statistics in medicine — there is no evidence that doctors process the numbers any better than patients do. Ku and his wife, Kang’s daughter, are both MDs. Yet under pressure and pressed for time, they too relied on just one or two numbers they had available and on the expert opinion of their surgeon.

“Informed decision-making assumes that you’ve informed your patients about benefits and risks. We do this by presenting them with numbers,” says Elie Akl, an assistant professor of medicine at both the University of Buffalo and McMaster University. This week, Akl and colleagues are publishing a new report about how exactly patients perceive those numbers.

The report is written for the Cochrane Collaboration, an organization that synthesizes existing research findings. The authors set out to compare three risk statistics in particular: the absolute risk reduction, the relative risk reduction and the number needed to treat. Studies to date suggest already that people may be more persuaded by some stats than others. Akl and colleagues scoured the medical literature for every published study on their topic and compiled results from all scientific articles that met their criteria, 35 studies in total.

The review finds that, in general, both health consumers and health professionals have better comprehension of absolute risk reductions than relative risk reductions. For example, if a new hypothetical drug reduces annual heart-attack incidence in a population from 2% to 1%, then the absolute risk reduction is 1 percentage point (0.02 – 0.01 = 0.01). However, the relative risk reduction would be 50%, since the risk has been cut in half (0.01/0.02 = 0.5).

Although people usually understand the absolute measures better (yes, that includes doctors, who are just as susceptible to confusion as the rest of us), the relative measures tend to be more persuasive. That 50% figure just seems much more impressive than the 1%, so we may be more likely to act on it — for better or worse. (The “number needed to treat” in this case would be 100. To avert 1 heart attack case, we would need to treat 100 patients.)

The math is not too complex, but the implications still may be. If Seung Kang’s family had been given relative risks rather than absolute risks — if, for example, they’d been told that his death risk was 100% higher than was normal for the surgery because his right ventricle was attached to the sternum — then might they have responded differently than they did to the overall 5% death risk? And if so, would that have been a good thing?

“The studies [in the new Cochrane review] looked mostly at whether people understand information, how they perceive the medication being presented, and how persuaded they are,” Akl says. “What we would be more interested to know is whether this format or the other is consistent with values and preferences.”

In other words, we want to know which presentation leads people to the decisions they would actually like to make. With today’s big push for evidence-based medicine, we may one day find out, though for now it’s an open question. With so many facts to process, how can we know which really matter and which are most relevant for our health decisions? Here are a few tips patients should keep in mind:

Don’t be shy. Ask for the kind of information you’d find useful — especially if you tend to find numbers confusing or frustrating. What kind of complications does your doctor expect? Is this procedure common? Knowing that your doctor performs a procedure every day and hasn’t had some particular complication in years may be easier for you to process than the overall complication rate.

Do the math. As you do think through numbers, pay attention to which stats you’ve received. Are they relative measures or absolute measures? A procedure that cuts your risk by one third will have a much bigger impact if you’re at high risk to begin with. E.g., dropping down from 30% to 20% is a bigger absolute change than dropping from 3% to 2%.

Weigh the options. Ask about alternatives to the procedure you’re considering. What would happen if you didn’t get surgery, or if you opted against a new medication? If you can, get a second opinion from another health professional.

Know thyself. Medical staff can give you pros and cons, but only you know how much you value different outcomes, or how keen you are to avoid side effects. That’s why you get to make the final decision.