OBJECTIVES: To inform drug formulary decision-making, US reimbursement authorities commonly require budget impact analyses (BIA), typically reported as budget impact (BI) per member per month (PMPM). However, proper interpretation of BI PMPM is hampered by the lack of accepted pre-specified PMPM thresholds defining a financially acceptable BI. We updated a prior systematic review of published US BIA to establish PMPM benchmarks and assess how BIA authors qualitatively interpreted their own results.

METHODS: Systematic PubMed/Embase searches (01/2003-11/2022) were conducted to identify full-text, peer-reviewed, US-based BIAs that reported pharmacotherapy BI PMPMs. Base case BI PMPM (inflated to 2022 value) and authors’ interpretations were analyzed descriptively by PMPM quartile.

RESULTS: We identified 78 BIA reporting BI PMPM. The median (interquartile range [IQR]) PMPM was USD cent (¢) 1.4 (-0.2, 4.6) across all 78 estimates and ¢2.3 (1.3, 7.0) across 57 non-negative estimates (i.e., PMPM>¢0). Among PMPM>¢0, 86% were reported with interpretations. These were more frequently provided for lower versus higher PMPM quartiles: Q1 (i.e., lowest quartile), 100%; Q2, 86%; Q3, 86%; Q4 (i.e., highest quartile), 73%. Among the most common terms used to describe financial acceptability for PMPM>¢0, interpretation patterns by quartile (ordered here from lowest to highest) were ambiguous (“minimal”: 21%/29%/7%/13%; “small”: 29%/36%/21%/0%; “modest”: 14%/0%/21%/13%). When stratified by interpretation term, the median (IQR) PMPM was lowest for “minimal” (n=10, ¢1.4 [0.5, 2.4]), followed by “limited/low/negligible/neutral” (n=7, ¢1.6 [1.2, 8.8]), “small” (n=12, ¢1.6 [1.2, 2.2]), “moderate/manageable/affordable/justifiable” (n=7, ¢3.3 [1.9, 5.4]), “modest” (n=12, ¢5.4 [1.7, 9.8]), no interpretation (n=8, ¢6.6 [2.1, 19.7]), and “considerable” (n=1, ¢209.3).

CONCLUSIONS: We provide updated benchmarks for BI PMPM. Additional research is needed to establish benchmarks to guide BI interpretation that account for various factors like therapeutic area, disease burden, and disease rarity. Until then, authors might want to refrain from providing potentially misleading qualitative judgment regarding BI acceptability.