Impact of a non-contributory pension for older adults on hospitalisations and mortality: a protocol for a natural experimental study of the Brazilian continuous cash benefit (BPC).

introductionThe continuous cash benefit (Benefício de Prestação Continuada, BPC) is Brazil's main non-contributory social pension. It guarantees a monthly income equivalent to one minimum wage to individuals aged ≥65 years and to people with disabilities living in poverty, but evidence on its health effects remains limited. This protocol describes the design and methods for evaluating the effects of BPC on hospitalisations and mortality among older adults and their household members. METHODS AND ANALYSIS: This natural experiment study uses linked administrative records from the 100 Million Brazilian Cohort (2001-2021), BPC payments (2016-2021), mortality (2001-2020) and hospitalisation data (2008-2018). Eligibility for elders is defined by age (≥65 years) and monthly family per capita income (<¼ of the minimum wage or up to 1/2 the minimum wage for specified groups). We defined as treated individuals who initiate BPC receipt from 2016 onwards. Untreated individuals are income or age-eligible or near-eligible adults who do not receive the benefit. The study population comprised people who were registered at any time in the cohort but were still alive on 1 January 2016. We triangulate two approaches. First, we will employ fuzzy regression discontinuity design using age and income eligibility thresholds to estimate local average treatment effects for different bandwidths. To assess the continuity assumption, we will analyse individuals aged 60 years or older to test outcome smoothness, while eligibility remains defined at age 65. Second, we will employ survival models with inverse probability of treatment weighting to estimate effects under conditional exchangeability assumptions. Outcomes include all-cause and cause-specific mortality and hospitalisations among older adults and their co-inhabitants. Analyses account for household clustering using cluster-robust SEs. Sensitivity analyses will assess robustness to data quality, municipal characteristics and potential overlap with Bolsa Família receipt. ETHICS AND DISSEMINATION: This study was approved by the Brazilian National Research Ethics Commission (CONEP) (approval number: 85981225.6.0000.0040), in accordance with the Brazilian National Health Council Resolution No. 466/2012. Findings will be disseminated through peer-reviewed publications, conference presentations, and policy briefs targeting stakeholders in social protection and health in Brazil. Results will also be shared with relevant governmental and non-governmental actors to support evidence-informed decision-making.