In clinical medicine you follow correlations. A positive HIV test and a running nose mean red alarm. If you know that from sole past observations or by reading about immunosuppression plays no role in giving the indicative treatment. In research mechanisms are, of course, mandatory everywhere.