The first thing a clinician needs to know to determine whether an article is providing valid, clinically useful results has virtually nothing to do with statistics. What is important is the basic architecture of the study, and doctors can diagnose an article in much the same way they diagnose patients.
If you want to decide whether a treatment does more good than harm, the issues you need to consider have nothing to do with such things as multivariate analysis, chi-squares or P values. They have to do with whether the people who have had the treatments and who are being compared really are comparable in every way that might materially effect their clinical outcome, other than their treatment itself, of course. There is only one way of being sure of this, and that is to randomize. That's why the randomized controlled trial (RCT) has been so crucial in terms of sorting out therapy.