Date of Award

January 2020

Document Type


Degree Name

Master of Public Health (MPH)


School of Public Health

First Advisor

Joshua D. Wallach

Second Advisor

Joseph S. Ross


Background: Controlled clinical trials are used to guide the decisions made by patients, clinicians, and policy makers. In order to inform care, it is recommended that controlled trials with binary outcomes and/or survival data report both relative and absolute effect measures. Objectives: To evaluate the characteristics and trends of reporting of absolute risk reduction (ARR), number needed to treat (NNT), and number needed to harm (NNH) among controlled trials published in the highest impact factor medical journals. Methods: Among the six highest impact factor general and internal medicine journals according to Journal Citation Reports, we identified all controlled superiority trials reporting binary outcomes or survival data in 2001, 2007, 2013 and 2019. We abstracted key study characteristics and for each NNT, NNH, or ARR identified, we recorded whether 95% CIs, P-values, and effect estimates were reported, the location of reporting, and the corresponding outcome type (primary vs. secondary). Results: Among 685 eligible trials, 103 (15%) reported at least one NNT, NNH, and/or ARR. There was a slight increase in reporting over time (15.5% in 2001 vs. 34.0% in 2019; p <0.01). The 103 (103/685) trials with at least one NNT, NNH, and/or ARR were larger (1637 vs. 772 median enrollment) and more likely to be in the clinical area of cardiology (27.2% vs. 18.2%). Of the 142 NNT/NNH and 59 ARR values reported, the majority (92/142, 64.8% NNT/NNH; 50/59, 84.7% ARR) were for primary outcomes. Only one in ten NNT/NNH were reported with a 95% confidence interval and/or P-value (14/142, 9.9%). There were 60 NNT and NNH values reported with effect estimates, of which 52 (86.7%) were statistically significant. Conclusions: Our findings suggest that while there may be an increase in reporting of NNT, NNH, and/or ARR over time, reporting is still rare and often incomplete. Therefore, opportunities exist for improving the reporting quality and frequency of absolute measures in controlled trials.


This thesis is restricted to Yale network users only. This thesis is permanently embargoed from public release.