TreeScan uses event hierarchies and adjusts for multiple testing, but its performance in spontaneous reporting systems remains less established. Using US Food and Drug Administration Adverse Event Reporting System (FAERS) data (2015Q1-2025Q3), we compared TreeScan with conventional algorithms: proportional reporting ratio (PRR), reporting odds ratio (ROR), Bayesian confidence propagation neural network (BCPNN), and gamma Poisson shrinker (GPS). We assessed sensitivity, specificity and area under the curve (AUC) in 40 drug-event groups from the Observational Medical Outcomes Partnership reference set. Events were coded as Medical Dictionary for Regulatory Activities (MedDRA) Preferred Terms (PTs) and mapped to High Level Terms (HLTs). TreeScan Rule A required significant PT signals; Rule B also accepted PTs in significant HLT cuts. All methods were reapplied to cumulative quarterly data; time to detection (TTD) was measured from 2015Q1 to the first quarter when a reference-positive group met each method\'s signal criterion. PRR and ROR had the highest sensitivity 65%. TreeScan Rule A had 55% sensitivity and 100% specificity; Rule B had 60% and 80%. GPS had the highest AUC 0.860. Among reference-positive group detected by each method, median TTD was 0.5 quarters for PRR and ROR, 1.0 quarters for Rule B, 1.5 quarters for GPS and Rule A, and 2.5 quarters for BCPNN. TreeScan had no consistent TTD advantage over other methods. Rule B was more sensitive but less specific than Rule A.