P73: Utilizing Community-Based EHR Data Systems to Identify Liver Disease Risk in Female Patients
Chief Medical Officer
ObjectiveGI United States
Seeking to address gender disparities in healthcare diagnostics, we identify how EHR liver enzymes and other biomarkers can be used as gender specific proxies for disease advancement in NASH.
We analyzed 705 EHR records from 10 US centers for patients referred for NAFLD Transient Elastography (TE) screening. Demographics were as follows: 449 women and 256 men, with median age 57, 17% of men and 24% of women had BMI (Body Mass Index) > 40. About 24% had confirmed Type II Diabetes.
e analyzed TE liver stiffness (kPa), ALT, AST, AST/ALT ratio, by gender. An advanced fibrosis state was proxied by kPa>10. Binomial logistic regression was used to quantify likelihood of advanced fibrosis (AF) across dynamic thresholds of three biomarkers (AST, ALT and AST/ALT), by gender. Optimizing the sum of sensitivity (SEN) and specificity (SPE), we arrived at gender specific thresholds for TE referral recommendations. Risk ratios allowed us to quantify population disparities, which we evaluated with p-values, negative predictive values (NPV) and positive predictive values (PPV).
In our analysis women with AST/ALT ratios above 1.2 are more likely to have kPa>10 (p=0.0004). If we perform TE on all women above this threshold, 32% are expected to have kPa >10. A similar threshold could not be found for men (p-value 0.13 to 0.49). We found an AST threshold of 35 for both men and women to have relevance in this population. Above this threshold men and women are 2.69x and 3.85x more likely to have kPa >10 (p=0.00003 and p=1.76e-13). Performing TE above this threshold, we expect that 44% of women and 38% of men will have kPa>10. For ALT, we found a threshold of 59 for men and 44 for women (p= 0.046, p=6.4e-11). Above these thresholds, women and men are, respectively, 3.21x and 1.56x more likely to have kPa>10. Above these thresholds, we would expect 42% of women and 31% of men who undergo TE to have advanced fibrosis.
Literature shows women have a lower risk for non-alcoholic fatty liver disease (NAFLD) but a higher risk for advanced fibrosis once Non-alcoholic Steatohepatitis (NASH) is established1. Seeking to adapt our status quo and address gender disparities in healthcare diagnostics, we identified how EHR liver enzymes can be used as gender specific proxies for disease advancement in NASH.
Though AST and ALT values alone are not used as indicators of disease advancement, our analysis identifies thresholds that can assist in the decision for referring patients for Transient Elastography. Women with an ALT > 44 and/or an AST/ALT ratio > 1.2 are at high risk for Advanced Fibrosis and should be referred for TE NAFLD screening. Their predisposition for advanced fibrosis at the same lab thresholds is much higher than their male counterparts, as can be gleaned by comparing their risk ratios. Since women are more likely to receive later stage NASH diagnoses than men, and more often get diagnosed with acute liver failure2, we recommend these lab thresholds for earlier liver screening amongst women.