What are poor, acceptable, and great Number Needed to Treat (NNT) figures?

Understanding Number Needed to Treat (NNT) Values

An NNT of 6 is considered excellent, representing a highly effective intervention where treating 6 patients prevents one additional adverse outcome. 1

What NNT Actually Measures

• NNT is calculated as the inverse of absolute risk reduction (ARR): NNT = 1/ARR × 100 2, 1
• For every 6 patients treated when NNT = 6, only 1 patient benefits from the intervention while 5 would have the same outcome regardless of treatment 1
• The measure represents the expected number of patients requiring treatment to prevent one additional adverse event compared to control therapy 3, 4

Categorizing NNT Values

Great NNT Values (Highly Effective)

• NNT of 6 or lower represents exceptional therapeutic benefit 1
• An NNT of 16 (as seen with tamoxifen for breast cancer prevention) is considered clinically meaningful, preventing one additional breast cancer per 16 women treated 2
• NNT values between 10-20 generally indicate strong clinical benefit, as demonstrated in cardiovascular prevention where NNT of 22 for composite CV events and 43 for heart failure were considered acceptable benchmarks 2

Acceptable NNT Values (Moderate Benefit)

• NNT values of 20-50 represent acceptable therapeutic benefit depending on the severity of the condition being prevented 2
• Ezetimibe plus simvastatin with NNT of 50 over 6 years demonstrates acceptable benefit for cardiovascular prevention 1
• For serious outcomes like stroke or myocardial infarction, NNT values up to 50 may justify treatment 2

Poor NNT Values (Marginal Benefit)

• NNT values exceeding 50-70 represent marginal therapeutic benefit requiring careful consideration of treatment burden 1
• Evolocumab (Repatha) with NNT of 67 over 2.2 years exemplifies a poor NNT, where substantial effort yields minimal benefit 1
• When NNT exceeds 100, the intervention requires treating large numbers of patients for minimal population-level benefit 2

Critical Contextual Factors

Time Period Dependency

• Always consider the time period over which NNT applies—this is the most common pitfall in interpretation 1
• An NNT of 39 for preventing headaches over 8 years (tamoxifen trial) differs substantially from the same NNT over 1 year 2
• Shorter time periods with low NNT values indicate more powerful interventions 1

Population Specificity

• NNT values are specific to the studied population and cannot be generalized across different patient groups 1
• Biomarker stratification can dramatically alter NNT values—for example, cardiovascular prevention NNT ranged from 16 to over 300 depending on biomarker status 2
• The intervention's specific formulation and dosing affect NNT calculations 1

Statistical Significance

• Always verify that the NNT is statistically significant before making clinical decisions 1
• Non-significant NNT values may reflect chance findings rather than true treatment effects 1

Balancing NNT with Number Needed to Harm (NNH)

• When NNH approaches or falls below NNT, the intervention may cause more harm than benefit 2
• Tamoxifen demonstrates this balance: NNT of 16 for breast cancer prevention versus NNH of 115 for venous thromboembolism and NNH of 73 for endometrial cancer 2
• The therapeutic index (NNH/NNT ratio) should exceed 2-3 for favorable risk-benefit profiles 2

Practical Application Algorithm

1. Identify the NNT value and associated time period 1
2. Verify statistical significance of the finding 1
3. Assess if NNT < 20 (great), 20-50 (acceptable), or > 50 (poor) 2, 1
4. Compare NNT to NNH values for the same intervention 2
5. Consider disease severity—lower NNT thresholds for less serious conditions 2
6. Evaluate population applicability to your specific patient 1