Analysis of a Simulation Model to Estimate Long-term Outcomes in Patients with Nonalcoholic Fatty Liver Disease

IMPORTANCE Quantitative assessment of disease progression in patients with nonalcoholic fatty liver disease (NAFLD) has not been systematically examined using competing liver-related and non-liver-related mortality. OBJECTIVE To estimate long-term outcomes in NAFLD, accounting for competing liver-related and non-liver-related mortality associated with the different fibrosis stages of NAFLD using a simulated patient population. DESIGN, SETTING, AND PARTICIPANTS This decision analytical modeling study used individual-level state-transition simulation analysis and was conducted from September 1, 2017, to September 1, 2021. A publicly available interactive tool, dubbed NAFLD Simulator, was developed that simulates the natural history of NAFLD by age and fibrosis stage at the time of (hypothetical) diagnosis defined by liver biopsy. Model health states were defined by fibrosis states F0 to F4, decompensated cirrhosis, hepatocellular carcinoma (HCC), and liver transplant. Simulated patients could experience nonalcoholic steatohepatitis resolution, and their fibrosis stage could progress or regress. Transition probabilities between states were estimated from the literature as well as calibration, and the model reproduced the outcomes of a large observational study. EXPOSURE Simulated natural history of NAFLD. MAIN OUTCOMES AND MEASURES Main outcomes were life expectancy; all cause, liver-related, and non-liver-related mortality; and cumulative incidence of decompensated cirrhosis and/or HCC. RESULTS The model included 1 000 000 simulated patients with a mean (range) age of 49 (18-75) years at baseline, including 6 6% women. The life expectancy of patients aged 49 years was 25.3 (95% CI, 20.1-29.8) years for those with FO, 25.1(95% CI, 20.1-29.4) years for those with F1, 23.6 (95% CI, 18.3-28.2) years for those with F2, 21.1(95% CI, 15.6-26.3) years for those with F3, and 13.8 (95% CI, 10.3-17.6) years for those with F4 at the time of diagnosis. The estimated 10-year liver-related mortality was 0.1% (95% uncertainty interval [UI], <0.1%-0.2%) in F0, 02% (95% UI, 0.1%-0.4%) in F1, 1.0% (95% UI, 0.6%-1.7%) in F2, 4.0% (95% UI, 2.5%-5.9%) in F3, and 29.3% (95% UI, 21.8%-35.9%) in F4. The corresponding10-year non-liver-related mortality was 1.8% (95% UI, 0.6%-5.0%) in F0, 2.4% (95% UI, 0.8%-6.3%) in Fl, 5.2% (95% UI, 2.0%11.9%) in F2, 9.7%(95% UI, 4.3%-18.1%) in F3, and 15.6% (95% UI, 10.1%-21.7%) in F4. Among patients aged 65 years, estimated 10-year non-liver-related mortality was higher than liver-related mortality in all fibrosis stages (eg, F2: 16.7% vs 0.8%; F3: 28.8% vs 3.0%; F4: 40.8% vs 21.9%). CONCLUSIONS AND RELEVANCE This decision analytic model study simulated stage-specific long-term outcomes, including liver- and non-liver-related mortality in patients with NAFLD. Depending on age and fibrosis stage, non-liver-related mortality was higher than liver-related mortality in patients with NAFLD. By translating surrogate markers into clinical outcomes, the NAFLD Simulator could be used as an educational tool among patients and clinicians to increase awareness of the health consequences of NAFLD.