Comparison of Markov model and discrete-event simulation techniques for HIV

Area of Science:

• Health economics and outcomes research
• Mathematical modeling in medicine
• HIV/AIDS clinical research

Background:

• Markov models are standard for predicting long-term outcomes but are complex and data-intensive.
• Discrete-event simulation (DES) is an emerging alternative for modeling HIV outcomes.
• Comparison of Markov and DES models using 48-week clinical trial data is needed.

Purpose of the Study:

• To compare the clinical and cost-effectiveness predictions of Markov and DES models for HIV.
• To evaluate the long-term (5-year and lifetime) validity of both modeling approaches.
• To assess the utility of DES in capturing detailed HIV disease progression and treatment response.

Main Methods:

• A cohort of 100 antiretroviral-naive HIV patients treated with lopinavir/ritonavir was used.
• Parameter estimates populated both Markov and DES models for long-term outcome comparison.
• Models were modified using relative risk data from a separate study comparing atazanavir and lopinavir/ritonavir.

Main Results:

• DES showed a slight predictive advantage, capturing detailed CD4+ T-cell counts and viral load categories.
• Both models yielded similar 1-year clinical estimates, but DES offered better 5-year predictive validity.
• Both models predicted cost savings for lopinavir/ritonavir over atazanavir, with similar cost estimates derived.

Conclusions:

• DES models offer superior face validity for decision-makers due to natural disease progression depiction.
• DES models facilitate probabilistic sensitivity analysis and allow inclusion of more variables without aggregation.
• DES demonstrates better long-term predictive capacity, isolating implications of crucial input data differences for HIV management.