A comprehensive geospatial epidemic model analyzing Brazil's COVID-19 pandemic demonstrates that pre-existing social inequalities were the primary driver of mortality disparities across municipalities, rather than unequal access to interventions. The modeling framework revealed that even without any interventions, deaths became increasingly concentrated in the most vulnerable communities as the epidemic spread. Non-pharmaceutical interventions following observed socially stratified patterns actually accelerated rather than created these gradients. However, the analysis uncovered a crucial policy insight: prioritizing vulnerable municipalities for vaccination could reverse mortality gradients entirely, with this protective effect amplified when combined with equitable adoption of preventive measures. This finding challenges assumptions that intervention access inequality is the main culprit in pandemic disparities. Instead, it points to deeper structural vulnerabilities that make certain populations inherently more susceptible to severe outcomes. The research provides a mechanistic framework for understanding how social determinants of health translate into epidemic mortality patterns. As a preprint awaiting peer review, these results require validation, but they offer valuable guidance for designing more equitable responses to future pandemic threats by addressing underlying social vulnerabilities alongside intervention distribution.