Use of Number Needed to Treat in Cost-effectiveness Analyses

Vishvas Garg MBA BPharm, PhD; Xian Shen MS, PhD; Yan Cheng MS, PhD; James J Nawarskas PharmD BCPS; Dennis W Raisch PhD MS RPh


The Annals of Pharmacotherapy. 2013;47(3):380-387. 

In This Article

Abstract and Introduction


Objective: To review the use of number needed to treat (NNT) and/or number needed to harm (NNH) values to determine their relevance in helping clinicians evaluate cost-effectiveness analyses (CEAs).

Data sources: PubMed and EconLit were searched from 1966 to September 2012.

Study selection and data extraction: Reviews, editorials, non-English-language articles, and articles that did not report NNT/NNH or cost-effectiveness ratios were excluded. CEA studies reporting cost per life-year gained, per quality-adjusted life-year (QALY), or other cost per effectiveness measure were included. Full texts of all included articles were reviewed for study information, including type of journal, impact factor of the journal, focus of study, data source, publication year, how NNT/NNH values were reported, and outcome measures.

Data synthesis: A total of 188 studies were initially identified, with 69 meeting our inclusion criteria. Most were published in clinician-practice-focused journals (78.3%) while 5.8% were in policy-focused journals, and 15.9% in health-economics-focused journals. The majority (72.4%) of the articles were published in high-impact journals (impact factor >3.0). Many articles focused on either disease treatment (40.5%) or disease prevention (40.5%). Forty-eight percent reported NNT as a part of the CEA ratio per event. Most (53.6%) articles used data from literature reviews, while 24.6% used data from randomized clinical trials, and 20.3% used data from observational studies. In addition, 10% of the studies implemented modeling to perform CEA.

Conclusions: CEA studies sometimes include NNT ratios. Although it has several limitations, clinicians often use NNT for decision-making, so including NNT information alongside CEA findings may help clinicians better understand and apply CEA results. Further research is needed to assess how NNT/NNH might meaningfully be incorporated into CEA publications.


A concern often expressed by cost-effectiveness analysis (CEA) researchers is that the results are not consistently implemented in practice settings.[1–6] The International Society for Pharmacoeconomics and Outcomes Research (ISPOR) task force on use of pharmacoeconomics/health economic information in health care decision-making conducted a systematic literature review and found several reasons for lack of use of CEA evidence in clinical research.[2] Lack of expertise in interpreting health economic measures, inability to conceptualize gains in quality-adjusted life-years (QALYs), and lack of relevance to clinical settings were among the main reasons. If health economic evidence were presented in a format more understandable to clinical decision-makers, it may increase the likelihood of such findings being implemented in practice settings. One way to improve understanding of health economic evidence by clinical decision-makers is to provide clinical end points along with health economic end points in such studies.[7]

The number needed to treat (NNT) and number needed to harm (NNH) are commonly used outcome measures in clinical settings, providing a quick, short-hand approach to estimating relative efficacy of different treatments.[8–13] These measures are easily understood by clinicians and are widely used in clinical decision-making, despite several limitations.[14–16] For example, Stang and coworkers described how NNT may be difficult to interpret if comparators are not specifically stated, NNT is rounded unnecessarily, or the period for treatment effect is not specified.[15] Despite these shortcomings, NNT is often used as a tool in medical decision-making under the general rubric of evidence-based practice.[17] Thus, including NNT into CEA publications may increase understanding and relevance of CEA findings to clinical decision-makers.

When incorporated into CEA studies, NNT (or NNH) values may be compared with incremental cost-effectiveness ratios (ICERs). A recent example is a CEA of prostate cancer screening, based on the European Randomized Study of Screening for Prostate Cancer.[18] Investigators found that 413 persons would need to be screened for prostate-specific antigen to prevent 1 death from prostate cancer. Based on screening costs plus lifetime treatment costs for patients with prostate cancer, the ICER was $5,227,306 to prevent 1 death from prostate cancer or $262,758 per life-year saved. This article, published in a clinician-practice-focused journal, Urology, provides a well-designed CEA that incorporates NNT. Therefore, we sought to review other CEA studies that incorporated NNT to see how frequently and in which journals such studies are published.

Our purpose was to review the literature reporting results of CEA along with NNT and/or NNH. We summarized our findings by type and year of publication, impact factors of the journals, type of intervention assessed, method used to report NNT in a CEA study, conflicts of interest/funding sources, and data sources used. For this article, NNT and NNH include NNT to benefit (NNTB) and NNT to harm (NNTH), respectively.[13]