Medication name similarity–related look-alike, sound-alike (LASA) errors occur across prescribing, dispensing, and administration. These risks particularly increase when phonetic/orthographic similarity interacts with contextual factors, such as packaging, labeling, and computerized medication lists. This study decomposed the LASA risk factors, and proposed prevention policy measures through linkages between regulatory drug name reviews and trademark examination. We constructed a database of 1,425 error pairs from ISMP sources (2024) and other reports (e.g., FDA MedWatchand KOPS). The data were normalized by removing order-invariant duplicates (A–B = BA), standardizing basic name formatting, and tagging formulation/strength-driven confusions as contextual. Stages were classified as prescribing, dispensing, and administration; types were multi-coded as orthographic, phonetic, or contextual; and stage-conditional distributions (P(type|stage)) were estimated. The results showed higher phonetic/orthographic shares in prescribing, a dominant orthographic share in dispensing, and an increased contextual share in administration. Comparative analysis suggested that major jurisdictions combine quantitative pre-market name review(e.g., POCA) with post-market reporting and feedback. Meanwhile, Korea faces gaps due to the separation of approval-stage name reviews and trademark examinations. We proposed a Korean-adapted POCA (K-POCA) reflecting Hangul phonology and grapheme structure; standardized contextual review for high-risk candidates; trademark criteria aligned with patient safety (ingredient-confusion risk); and a linkage mechanism that feeds post-market error reports into review and post-market management. Overall, this study offers an integrated framework that links a case database, type–stage distributions, institutional comparisons, and policy recommendations at the intersection of patient safety, intellectual property, and regulatory science.