Intention-to-treat analysis keeps all trial data, showing true drug effects, even with dropouts. It contrasts with per-protocol analysis, where dropout data is excluded, possibly skewing results.