What is the number needed to screen (NNS) and how is it calculated?

Number Needed to Screen (NNS): Definition and Calculation

The Number Needed to Screen (NNS) is calculated as the inverse of the absolute risk reduction (ARR), expressed mathematically as NNS = 1/ARR, where ARR represents the difference in event rates (typically mortality) between screened and unscreened populations. 1

Core Calculation Method

Basic Formula:

• NNS = 1 / [(Control event rate) − (Screened event rate)] 1
• The ARR is obtained by subtracting the event rate in the screened group from the event rate in the control (unscreened) group 1
• When expressed in mortality terms: NNS = 1 / [(Deaths per person in control) − (Deaths per person in screened)] 1

Alternative Calculation Method:

• NNS can be estimated by multiplying the Number Needed to Treat (NNT) by the number of people who must be screened to find one patient with the disease 2, 3
• This approach is useful when direct mortality data from screening trials are unavailable 3

Critical Adjustment for Follow-Up Duration

Time-adjusted NNS must be based on mortality rates per person-year rather than per person over the entire study period to avoid systematic overestimation. 1

Adjustment Formula:

• For a standardized comparison period of X years: Adjusted NNS = (Crude NNS) / X 1
• Example: A crude NNS of 2,399 over 13 years yields a 10-year adjusted NNS of approximately 1,845 (2,399 × 10/13) 1
• Shorter follow-up periods inherently produce higher NNS values, potentially making effective interventions appear less beneficial 4, 1

Common Pitfall:

• The USPSTF methodology for breast cancer screening omitted this adjustment, producing biased comparisons between age groups (women 40-49 years: reported NNS = 1,904 vs. adjusted ≈ 1,599 for 15 years; women 50-59 years: reported = 1,339 vs. adjusted ≈ 1,708 for 15 years) 4, 1
• NNS values should never be compared across studies unless the follow-up duration is standardized 1

Deriving NNS from Relative Risk

When only pooled relative risk (RR) data are available:

• Calculate ARR as: ARR = Baseline rate × (1 − RR) 1
• Then apply: NNS = 1 / ARR 1
• Example: For breast cancer screening with RR = 0.80 (20% risk reduction) and baseline mortality of 0.005, ARR = 0.001 and NNS = 1,000 1

Accounting for Participation and Selection Effects

The Number Needed to Be Screened (NNBS) adjusts for real-world participation rates and selection bias, providing a more accurate estimate than crude NNS. 5

• NNBS is derived from NNT adjusted for participation in screening and selection effects associated with participation 5
• For breast cancer screening, NNBS was 23% lower than crude NNS (NNBS = 601 vs. NNS = 781) 5
• For colorectal cancer screening, NNBS was 45% lower than crude NNS (NNBS = 688 vs. NNS = 1,250) 5
• This adjustment is especially important when comparing screening programs with disparate participation rates 5

Population-Specific Considerations

Baseline risk strongly influences NNS; therefore, age-stratified or risk-stratified calculations are essential and should never be pooled across heterogeneous groups. 1

• Higher-risk populations consistently exhibit lower NNS, indicating more efficient screening 1
• Example from tuberculosis screening in migrants: median NNS = 231 (IQR 1,022) to detect one case of active TB 4
• Example from prostate cancer screening (ERSPC): NNS decreased from 1,410 at 9 years to an estimated 503 at 12 years with continued follow-up 4, 6

Time-Dependent Nature of NNS

NNS is inherently time-specific and decreases with longer follow-up as cumulative mortality differences grow between screened and control groups. 6

• In the ERSPC prostate cancer trial, NNS decreased from 1,254 at year 9 to 837 at year 10 and 503 at year 12 6
• Reporting NNS at a single time point may lead to misinterpretation of screening benefits 6
• For chronic conditions requiring long-term follow-up, annualized NNS should be calculated to account for varying study durations 1

Confidence Intervals

• Confidence intervals for NNS are obtained by inverting and exchanging the confidence limits for the ARR 7
• The Wilson score method provides more accurate confidence intervals than the standard asymptotic method, which often yields intervals that are too narrow 7
• Always verify that the underlying ARR is statistically significant before interpreting NNS 1

Practical Examples from Current Screening Programs

Based on published data from the Netherlands:

• Cervical cancer screening: NNS ≈ 2,560 per year to prevent one death 2
• Breast cancer screening: NNS ≈ 1,000 per year to prevent one death 2
• Hypertension detection (ages 55-75): NNS ≈ 2,340 per year to prevent one death 2
• Colorectal cancer screening (hemoccult): NNS = 1,374 for 5 years to prevent one colon cancer death 3