Intra-Alveolar Gelatin Sponge Delivery of Dexamethasone vs. Methylprednisolone After Mandibular Third-Molar Surgery: A Randomized Controlled Trial

Impacted mandibular third-molar surgery commonly causes early postoperative pain, swelling, and trismus. This randomized, controlled, three-arm parallel trial evaluated whether intra-alveolar corticosteroid delivery via an absorbable gelatin sponge improves postoperative recovery compared with a saline control. Fifty-five patients were assessed for eligibility; 37 healthy adults (18–35 years) undergoing standardized mandibular third-molar extraction were randomized to dexamethasone 8 mg (Decort®), methylprednisolone 40 mg (Prednol®), or control (saline), all applied intra-alveolarly using a gelatin sponge carrier. Doses were selected using standard systemic glucocorticoid equivalence tables as a pragmatic potency reference, acknowledging unknown intra-alveolar pharmacokinetics/bioavailability. The prespecified primary endpoint (used for sample size planning) was postoperative Day 1 VAS pain; key secondary endpoints were Day 1 analgesic consumption and Day 3 facial swelling. Pain (VAS), analgesic use, trismus, and facial swelling (tragus–pogonion, tragus–labial commissure, and angulus–canthus distances) were assessed on postoperative Days 1, 2, 3, and 7 by a blinded evaluator. Two participants in the methylprednisolone group did not attend postoperative visits. To address potential attrition bias, an Intention-to-Treat (ITT) sensitivity analysis using conservative control-median imputation was performed alongside the available-case analyses. A global False Discovery Rate (FDR) correction was also applied to control for multiplicity. In both analyses, the steroid groups showed lower Day 1 pain scores than the control group. Methylprednisolone was associated with lower Day 3 swelling values than control for the tragus–pogonion and angulus–canthus measurements. These findings should be interpreted as preliminary, given the small sample size, linear swelling measurements, and lack of blinding verification.