Number Needed to Treat Calculator – NNT, NNH and ARR

On paper, these relative risk reductions look pretty good. You see a 30% reduction in heart attacks, and it seems like an advance; it’s something that should influences your judgment. But relative risk completely disregards how frequently that event happen as a baseline. In other words, if the baseline risk is low, then even a big improvement in relative risk doesn’t do much for individual patient.

The absolute numbers are far more important than the percentages doctors mention in press release, those are the real chances of something happening.

Why Real Numbers Are More Important Than Percentages

Number Needed to Treat Number Needed to Treat fills this void, bringing abstract numbers down to earth in human units. How many people need to undergo this intervention for exactly one person to be helped over the alternative? You can’t escape the treatment’s cost and side effects when you’re told that you’ll have to treat fifty people to save a single life. When it come to people instead of populations, the math is more difficult to sweep under rug.

The correct way to use it is with both the Experimental Event Rate and the Control Event Rate (the probabilities of the outcome occurring in the treated group versus the untreated group). The untreated). This is what the calculator subtracts/inverts for you; however, if your risk figures don’t match (i.e., same definition and time-frame), then you’ll end up with gibberish. For instance: You can’t compare a one year infection rate to a ten year mortality rate. That’s why the resulting NNT is meaningless.

The time frame anchors the number, an NNT of twenty over six months is not the same as an NNT of twenty over a decade. You might be tempted to interpret this as meaning there’s no change: if the range of plausible risk differences crosses zero, it could mean anything from a benefit to harm. But you need to check the confidence interval to see what it say. In this case, the believable range of risk differences includes both negative and positive values, which means that technicaly, the NNT extends from harm to benefit with infinity in between. That’s a statistical red flag. Either the study was small (so less precise) or the effect was modest, neither of which support a confident recommendation at this point.

As the table further up on the page illustrates, even slight changes in the risk difference move the NNT around. It illustrates importance of having good precision in your input estimates.

They’re used by public health planners as a way to best distribute resources. For a single patient, a high NNT (and therefore a small absolute risk reduction) appears inefficiently. But for a big population, those tiny fractions sum up to thousands of avoided events. That’s the paradox of preventive medicine: We accept high NNTs for important outcomes to spare everyone from the misery of widespread illness. Population benefit vs. Individual efficiency, that’s the tension in today’s preventive care.

Additionally, don’t mix up the “number needed to harm” with the “number needed to treat”. For instance, a medication that prevents a heart attack in every 50 people it’s administered to may lead to substantial bleeding in one out of every hundred. Those are different numbers… Pay attention to both. Ultimately, you’re trying to balance those different values: how bad would it be to prevent this versus how bad would it be to have this side effect? If the first is catastrophic and the second is mild, a high NNT is fine.

The ultimate decision relies on context. For one set of people, a statin may offer the same NNT as a different drug. But that doesn’t mean it offers the same value. Some patients will be more concerned about side effects from medication than others; some are worried about their family history, while others want to do everything they can to lower their genetic risk. Raw numbers comes from the calculator; meaning comes from the discussion with the patient.

When it comes to medicine, going from relative risk to an absolute measure clarifies medical marketing. The result is that you get rid of the hype and are left with the plain facts: what exactly will this intervention do for me? When you move away from percentage points and toward actual people, numbers becomes wise decisions. You should of looked at the absolute numbers first.