Efficiency Analysis of Public Health Spending in Brazilian Capitals Using Network Data Envelopment Analysis (DEA)

Aim: In 1988, Brazil implemented profound changes in the organization and financing of its public health system, with the creation of the Unified Health System (Sistema Unico de Saúde - SUS), establishing universal health coverage. The gradual expansion of the health system and entitlements to services has been accompanied by the debate about the appropriate level of government spending and health system efficiency. Design / Research methods: The study uses VRS - output oriented, Dynamic Network SBM DEA model, period 2008-2013, to depict the relationships that take place between diverse levels of care (primary health care/PHC and secondary-tertiary health care/STC). DMUs are Brazilian state capitals, which implement key health policies and assist patients from smaller surrounding municipalities, especially for STC. Inputs are PHC and STC budgets; outputs are their respective services provided and avoidable deaths. The link variable is PHC medical consultation, entrance door to the system and gatekeeper for more complex levels of care. Dynamic model evaluates efficiency across time. Conclusions / findings:Overall performance was 0.86; for PHC, 0.90; for STC, 0.85 (SD=0.15). 8 out of 27 capitals were fully efficient. Capitals increased average scores in both levels of care, but only STC had a positive technological change (frontier shift >1). Link variable behavior denotes a bottleneck between levels of care. Projections onto the frontier enable establish own management diagnosis and goals for financing and development. Originality / value of the article: Network models mimic hierarchically organized health systems. The appliance of results aids health policy.