Table 1—

Hypothetical clinical trials of four new diets for weight loss

Statistical significanceClinical significance
YesNo
YesDiet A:Diet B:
n = 1,000n = 2,000
Mean difference −7.0 kgMean difference −0.5 kg
95% CI −7.6 to −6.4 kg95% CI −0.9 to −0.1 kg
p-value <0.0001p-value = 0.025
Conclusions:Conclusions:
Large studyLarge study
Large effectSmall effect
NoDiet C:Diet D:
n = 36n = 400
Mean difference −3.0 kgMean difference −0.2 kg
95% CI −6.3 to +0.3 kg95% CI −1.2 to +0.8 kg
p-value = 0.07p-value = 0.69
Conclusions:Conclusions:
Study not large enoughStudy probably large enough
Probably a real and moderate effect, but insufficient results to draw a reliable conclusionProbably a small effect
  • Effect size is the mean difference in weight (new diet minus control). CI: confidence interval. N represents the total number of subjects in a two-arm trial.