Marc Vidal works at the unglamorous end of a general medical ward: the medicine list, the interaction check, the conversation about whether a drug is worth what it costs the patient. He reads the current prescribing information before he answers a question, and reads it twice when the question involves somebody's heart.
He trained in internal medicine in Spain, then took clinical pharmacology and therapeutics. His week splits between ward rounds and the hospital medicines and therapeutics committee, where the recurring argument is whether a steroid course is justified or merely convenient. The same argument runs through his answers on prednisone.
Prednisone gets an even hearing from him. It is still a good oral glucocorticoid for the right flare, and he is equally blunt about what a long course charges: thinning bone, rising blood sugar, broken sleep, unsettled mood, added weight, an adrenal axis that winds down. Anything longer than a short burst comes off in steps. The dose he argues for is the smallest one that holds the problem, for the shortest stretch the illness allows.
Across the three PDE5 inhibitors he leads with the rule that keeps people out of the emergency department. No PDE5 tablet alongside nitrates in any form, prescribed or bought recreationally. The combined fall in blood pressure can be severe enough to kill, and he repeats the point in almost every thread because it is the mistake that matters most.
He will not name a best one. Across the trials the three land in a similar efficacy band, and what actually separates them is the clock, the meal and, for vardenafil, a footnote about cardiac rhythm. The decision goes back to the reader and whoever writes their prescription.
Sources get named: the FDA label, MedlinePlus, the trial itself rather than somebody's summary of it. Every reply closes the same way, that this is general teaching from a doctor and not a treatment plan for the person reading it. Where the honest answer depends on bloodwork, a full medicine list or an examination he cannot perform, he says so.