Developing Markov Models From Real-World Data: A Case Study of Heart Failure Modeling Using Administrative Data.

Area of Science:

• Health economics
• Real-world evidence utilization
• Disease progression modeling

Background:

• Traditional Markov models rely on clinical trial data for disease state transitions.
• Real-world data (RWD) offers an alternative for defining health states and transitions outside controlled settings.
• Estimating the cost-effectiveness of interventions like telemonitoring (TM) requires accurate disease progression models.

Purpose of the Study:

• To present an alternative approach using RWD to define health states and model transitions for cost-effectiveness analysis.
• To estimate the cost-effectiveness of telemonitoring (TM) versus usual care for heart failure in a specific UK primary care trust (PCT).

Main Methods:

• Utilized UK hospital episode statistics (HES) data to estimate hospitalization incidence.
• Developed a monthly transition matrix with 5 health states based on prior hospitalizations and death.
• Employed probabilistic sensitivity analysis from a healthcare perspective to assess TM cost-effectiveness.

Main Results:

• Geographical variations in hospitalization rates necessitated localized transition matrices.
• In the evaluated PCT, TM resulted in mean additional costs of £3610 and 0.075 additional QALYs per patient.
• The mean incremental cost-effectiveness ratio (ICER) for TM was £48,172 per QALY.

Conclusions:

• Using administrative data to define health states and transition matrices is a feasible approach.
• Telemonitoring (TM) was found not to be cost-effective in this specific UK primary care setting.
• The study supports the increasing use of RWD for health economic evaluations and localized intervention assessments.