How to Calculate Number Needed to Screen (NNS)
The number needed to screen (NNS) is calculated as the inverse of the absolute risk reduction (ARR): NNS = 1/ARR, where ARR is the difference in event rates between screened and unscreened populations. 1
Basic Calculation Method
The fundamental formula is straightforward:
- NNS = 1 / Absolute Risk Reduction 2, 3
- ARR = (Control group event rate) - (Screened group event rate) 1
- Express as: NNS = 1 / [(Deaths in control group per person) - (Deaths in screening group per person)] 1
For example, if the control group has a mortality rate of 0.005 (5 deaths per 1000) and the screening group has 0.004 (4 deaths per 1000), the ARR = 0.001, making NNS = 1/0.001 = 1000 3, 4
Alternative Calculation Using NNT
When screening trials measure treatment benefit rather than direct screening effect, NNS can be estimated by multiplying the number needed to treat (NNT) by the prevalence of unrecognized disease: 3, 4
- NNS = NNT × (1/Prevalence of undetected disease) 4
- This method requires knowing both the treatment efficacy and the detection rate from screening 4
Critical Adjustment for Follow-Up Period
A major pitfall is failing to adjust NNS for different follow-up periods, which leads to systematic overestimation when comparing age groups or studies with varying durations. 1
Time-Adjusted Calculation:
- Calculate mortality rate per person-year rather than per person over the entire follow-up 1
- For a standardized comparison period of X years: Adjusted NNS = (Crude NNS) / X 1
- Example: If crude NNS is 2399 over 13 years, the 10-year adjusted NNS = 2399 × (10/13) = approximately 1845 1
The USPSTF methodology failed to account for this, resulting in biased comparisons showing women aged 40-49 had NNS of 1904 versus 1339 for ages 50-59, when properly adjusted values were much closer (1599 vs 1708 for 15-year follow-up). 1
Calculating from Relative Risk Data
When you have relative risk (RR) from meta-analysis:
- Calculate pooled RR using random-effects model 1
- Determine baseline mortality rate in control group 1
- Calculate ARR = Baseline rate × (1 - RR) 1
- NNS = 1/ARR 1
For breast cancer screening with RR = 0.80 (20% reduction), if baseline mortality is 0.005, then ARR = 0.005 × 0.20 = 0.001, making NNS = 1000 1
Confidence Intervals for NNS
Always calculate confidence intervals by inverting and exchanging the confidence limits of the ARR, accounting for the NNS scale that ranges from 1 through infinity to -1. 5
- Use Wilson score method rather than simple Wald method, as Wald produces inappropriately narrow confidence intervals 5
- If ARR confidence interval crosses zero, NNS becomes undefined or includes both positive and negative values 5
Time-Dependency Considerations
NNS values are inherently time-specific in survival studies, and reporting at a single time point leads to misinterpretation. 6
- NNS decreases with longer follow-up as cumulative mortality differences grow 6
- For prostate cancer screening, NNS decreased from 1,254 at 9 years to 503 at 12 years with continued follow-up 6
- Always specify the time period when reporting NNS 2, 6
Common Pitfalls to Avoid
Shorter follow-up periods necessarily produce higher NNS values, which can make interventions appear less effective than they actually are. 1
- Never compare NNS values across studies without standardizing for follow-up duration 1
- Account for differences in screening intervals (annual vs biennial) when calculating cumulative NNS 1
- Consider drop-out rates and compliance, as these affect the denominator 6
- Distinguish between "number needed to invite" (intention-to-screen) versus "number needed to actually screen" (per-protocol) 1
Population-Specific Adjustments
NNS varies substantially by baseline risk, so age-stratified calculations are essential. 1