Analysis of 87,246 births across eight DC hospitals reveals that institutional characteristics account for nearly one-third of variance in hepatitis B vaccine refusal rates, with hospital-specific rates ranging from near-zero to over 50%. The multilevel regression identified that 22.1% of total refusal variance stems from stable hospital characteristics, while temporal trends contributed 9% of variance. This finding challenges the predominant focus on individual patient factors in vaccination disparities research. The institutional effect proves particularly significant given that safety-net hospitals achieved target refusal rates within one year of 2018 ACIP policy changes, while higher-baseline institutions required 2-4 years to respond. This differential response speed suggests that hospital culture, staff training protocols, and communication practices may be more modifiable drivers of vaccination outcomes than previously recognized. The reversal of typical racial vaccination patterns—where White infants showed lower refusal odds than Black infants—further underscores how institutional contexts can override broader population trends. For public health practitioners, these results indicate that system-level interventions targeting hospital practices, rather than solely patient education, may offer more efficient pathways to improving birth dose coverage and reducing disparities.