IDENTIFYINGWHETHER COST OR CLINICAL OUTCOMES DRIVE HETEROGENEITY: THE COST-EFFECTIVENESS VARIABILITY INDEX (CEVI)
Author(s)
Lívia Loamí Ruyz Jorge Paula, PhD1, Vitor Sain Vallio, MSc1, Fernanda Franco Munari, MSc1, Wellington dos Santos, PhD1, Moacyr Campos, MD1, Talita Garcia do Nascimento Castro, PhD1, Caio Sain Vallio, PhD2, Bruna Andrade Guedes, BSc1, Mateus Frederico de Paula, MSc1, Bruno Tirotti Saragiotto, PhD1, Felipe Ribeiro Cabral Fagundes, PhD1.
1Research and Data Science Division, Hi! Healthcare Intelligence, São José dos Campos, Brazil, 2Data Science and Machine Learning, Hapvida Health, Fortaleza, Brazil.
1Research and Data Science Division, Hi! Healthcare Intelligence, São José dos Campos, Brazil, 2Data Science and Machine Learning, Hapvida Health, Fortaleza, Brazil.
OBJECTIVES: Health systems frequently observe substantial variability in both costs and clinical outcomes, yet lack a simple tool to determine which dimension warrants priority investigation and intervention. Conventional cost-effectiveness analyses rely on averages that obscure clinically and economically relevant dispersion. This study introduces the Cost-Effectiveness Variability Index (CEVI), a single composite measure designed to identify whether variability is primarily driven by clinical response or resource consumption.
METHODS: CEVI is defined as the ratio of the Gini coefficient for cost to that for effectiveness, with cost as the numerator, consistent with standard practice. Values above 1 indicate greater cost variability; values below 1 indicate greater effectiveness variability; values around 1 indicate no clear predominance. Being scale-invariant, CEVI requires no normalization and remains robust to differences in units between monetary values and clinical outcomes. Validity was assessed through Monte Carlo simulation (n=200 per group) across three controlled scenarios: Scenario 1 modeled effectiveness instability with homogeneous costs; Scenario 2 modeled cost instability with stable effectiveness; Scenario 3 modeled balanced instability across both dimensions.
RESULTS: CEVI correctly identified the dominant source of variability across all simulated scenarios. In Scenario 1, CEVI was 0.52, indicating greater effectiveness variability. In Scenario 2, CEVI was 3.20, indicating greater cost variability. In Scenario 3, CEVI was 1.00, indicating balanced variability with no clear predominance. Results confirmed expected directional classification in all three conditions, validating CEVI's ability to distinguish between cost- and outcome-driven dispersion.
CONCLUSIONS: CEVI provides an objective, scale-invariant tool to determine whether analytical efforts should prioritize clinical heterogeneity or cost determinants. By identifying the dominant source of dispersion, CEVI enables health systems to more efficiently target quality-improvement initiatives, economic investigations, and value-based care interventions. Its simplicity and interpretability make it potentially applicable across diverse care pathways and health system contexts, supporting more rational and transparent allocation of analytical and operational resources.
METHODS: CEVI is defined as the ratio of the Gini coefficient for cost to that for effectiveness, with cost as the numerator, consistent with standard practice. Values above 1 indicate greater cost variability; values below 1 indicate greater effectiveness variability; values around 1 indicate no clear predominance. Being scale-invariant, CEVI requires no normalization and remains robust to differences in units between monetary values and clinical outcomes. Validity was assessed through Monte Carlo simulation (n=200 per group) across three controlled scenarios: Scenario 1 modeled effectiveness instability with homogeneous costs; Scenario 2 modeled cost instability with stable effectiveness; Scenario 3 modeled balanced instability across both dimensions.
RESULTS: CEVI correctly identified the dominant source of variability across all simulated scenarios. In Scenario 1, CEVI was 0.52, indicating greater effectiveness variability. In Scenario 2, CEVI was 3.20, indicating greater cost variability. In Scenario 3, CEVI was 1.00, indicating balanced variability with no clear predominance. Results confirmed expected directional classification in all three conditions, validating CEVI's ability to distinguish between cost- and outcome-driven dispersion.
CONCLUSIONS: CEVI provides an objective, scale-invariant tool to determine whether analytical efforts should prioritize clinical heterogeneity or cost determinants. By identifying the dominant source of dispersion, CEVI enables health systems to more efficiently target quality-improvement initiatives, economic investigations, and value-based care interventions. Its simplicity and interpretability make it potentially applicable across diverse care pathways and health system contexts, supporting more rational and transparent allocation of analytical and operational resources.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
MSR8
Topic
Clinical Outcomes, Economic Evaluation, Methodological & Statistical Research
Topic Subcategory
Artificial Intelligence, Machine Learning, Predictive Analytics
Disease
Generics