OBJECTIVES: Clinicians are told to use the number needed to treat (NNT) to compare the benefits of therapeutic strategies, and researchers are asked to report results this way, generally without considering differences among the studies from which these were derived. METHODS: The crude NNT currently advocated is compared to the NNT standardized for a common outcome, follow-up time, study population and comparator. An NNT model for cardiovascular disease is described as an example that addresses differences among studies of secondary prevention of cardiovascular disease. Crude NNTs are compared to those obtained from the model. RESULTS: Follow-up in the 18 trials identified varied from 1.0 to 6.2 years; rates of cardiovascular events in the untreated subgroups ranged from 4.8% to 45.9%. The crude NNTs were more variable (9.1-163.7) than those obtained from the model (9.1-75.2). The effect of standardization was substantial in some cases, with proportional changes ranging from a 91% decrease to a 223% increase. CONCLUSION: Using an NNT model to account for differences in study design allows for more meaningful comparisons.
OBJECTIVES: Clinicians are told to use the number needed to treat (NNT) to compare the benefits of therapeutic strategies, and researchers are asked to report results this way, generally without considering differences among the studies from which these were derived. METHODS: The crude NNT currently advocated is compared to the NNT standardized for a common outcome, follow-up time, study population and comparator. An NNT model for cardiovascular disease is described as an example that addresses differences among studies of secondary prevention of cardiovascular disease. Crude NNTs are compared to those obtained from the model. RESULTS: Follow-up in the 18 trials identified varied from 1.0 to 6.2 years; rates of cardiovascular events in the untreated subgroups ranged from 4.8% to 45.9%. The crude NNTs were more variable (9.1-163.7) than those obtained from the model (9.1-75.2). The effect of standardization was substantial in some cases, with proportional changes ranging from a 91% decrease to a 223% increase. CONCLUSION: Using an NNT model to account for differences in study design allows for more meaningful comparisons.
Authors: Yola Moride; Robert A Hegele; Anatoly Langer; Ruth McPherson; David B Miller; Stéphane Rinfret Journal: Can J Cardiol Date: 2008-04 Impact factor: 5.223
Authors: Theodore K Marras; Mariam Hassan; Kevin C Mange; Monika Ciesielska; Shilpa Dhar Murthy; Zhanna Jumadilova; Anjan Chatterjee Journal: ERJ Open Res Date: 2022-07-11