Usually most studies come with many estimates of effects and sub-analyses. Which ones should you accept? If you are studying randomized control trials (RCTs), take only that estimate where the rule of randomization is strictly followed. All other effect estimates are not important. For studies that use non-randomized study designs, look for those estimates that are adjusted for age. For models that include more than one independent variable, take the estimate where age and additionally, effect of a well known confounders (on theoretical grounds) are used for adjustment. Alternatively, use the effect measure in a model where most variables are used for adjustment. A good rule of the thumb is to use the effect measure that is presented in the abstract.