Can You Calculate “Number Needed to Treat” (NNT) from Ratio Data?
In the world of medical research and evidence-based practice, the Number Needed to Treat (NNT) is one of the most intuitive ways to communicate the benefit of a therapy. It tells us, simply, how many patients we need to treat with a specific intervention to prevent one additional bad outcome (or achieve one additional good outcome).
But a common question arises among students and clinicians: “I have the Odds Ratio (OR) or Relative Risk (RR) from a study. Can I just plug those numbers into a formula to get the NNT?”
The short answer is no—not directly. Let’s break down why, and how you can get there if you have the right pieces of the puzzle.
- * *
The Problem: Ratios vs. Absolute Differences
To understand why you can’t calculate NNT directly from a ratio (like an Odds Ratio or Relative Risk), you have to look at the definition of NNT.
NNT is the inverse of the Absolute Risk Reduction (ARR):
NNT = 1 / ARR
- ARR is an absolute measure (the difference between two percentages).
- Ratio data (OR or RR) are relative measures (the relationship between two percentages).
Example: Imagine a drug reduces the risk of a heart attack from 4% to 2%.
- The ARR is 2% (0.02).
- The NNT is 1 / 0.02 = 50.
Now, imagine a different drug reduces the risk from 0.4% to 0.2%.
- The ARR is 0.2% (0.002).
- The NNT is 1 / 0.002 = 500.
Notice that in both scenarios, the Relative Risk is the same: the risk was cut in half (RR = 0.5). However, the NNT is vastly different (50 vs 500). Because Ratios don't account for the “baseline risk” of the patient population, they cannot be used alone to calculate the NNT.
- * *
Can you calculate it if you have the baseline risk?
If you have the Relative Risk (RR) or Odds Ratio (OR) and you know the Control Event Rate (CER)—which represents the baseline risk of the population—then yes, you can calculate the NNT.
1. If you have the Relative Risk (RR):
First, calculate the Absolute Risk Reduction:
ARR = CER × (1 – RR)
Then calculate the NNT:
NNT = 1 / ARR
2. If you have the Odds Ratio (OR):
This is more complex because Odds are not the same as Risk. You must first convert the Odds back into a probability (Risk) for the intervention group, or use a conversion formula:
Experimental Event Rate (EER) = (OR × CER) / (1 – CER + (OR × CER)) ARR = CER – EER NNT = 1 / ARR
- * *
Why this matters for Evidence-Based Practice
The “Trap of Relative Risk” is a classic concept in medical statistics. A drug might be reported as having a “50% reduction in risk,” which sounds impressive. However, if that risk was only 0.1% to start with, the NNT is 2,000.
If you are reading a study and it only provides an Odds Ratio or Relative Risk, be cautious. Always look for the raw numbers (the contingency table) or the event rates for the control and treatment groups.
The Bottom Line
- Never assume you can derive NNT from a Ratio alone.
- Always look for the Absolute Risk Reduction (ARR) or the event rates for both groups.
- Remember: The NNT is highly dependent on the baseline risk of the patients in the study. https://everycalculators.com/ that works wonders for high-risk patients may have a uselessly high NNT for low-risk patients, even if the Relative Risk reduction remains the same.
Conclusion: When it comes to NNT, context is king. Ratios tell you how much a drug changes the math, but the NNT tells you if it’s worth the effort for your specific patient. Always prioritize absolute numbers over relative ones!