UCSF

Background and aims

Primary care providers need strategies to identify NAFLD patients and select for specialty referral, but proposed algorithms have only been studied in established NAFLD patients.

Methods

We implemented an algorithm for all adults with diabetes mellitus in a large primary care practice and excluded hepatitis B and C or alcohol use. Applying annual Fibrosis-4 Index and NAFLD Fibrosis Score for 5 years, we categorized patients as low-risk, indeterminate-risk, or high-risk for advanced fibrosis. We targeted all high-risk and messaged each primary care provider, recommending hepatology linkage. We collected final diagnosis and fibrosis (F0-4) outcomes. Using multivariable logistic regression, we assessed risk factors for advanced fibrosis stage (F3-4).

Results

Of 3028 patients, 1018 were low-risk, 577 indeterminate-risk, and 611 high-risk. There were 264 target patients; their 89 primary care providers received a message per patient suggesting hepatology referral. The majority (n=149) were referred; at triage, 118 were deemed likely NAFLD. Of these, 90 completed visits, 78/90 were diagnosed as NAFLD, and 69/78 underwent fibrosis staging, with F3 to 4 in 25/69. In multivariable analysis, hemoglobin A1c ≥8% (OR=7.02, 95% CI: 1.29-38.18) and Fibrosis-4 Index (OR=1.79, 95% CI: 1.07-2.99) were associated with increased risk of F3 to 4.

Conclusions

This is the first prospective study testing a case-finding strategy in primary care and almost 1/3 of diabetes mellitus were high-risk for advanced fibrosis. When prompted, 73% of primary care providers placed referrals and 76% of patients completed visits, revealing 86% NAFLD and 36% F3 to 4. This study demonstrates the readiness for such a strategy in primary care; integrating hemoglobin A1c into this algorithm may further improve the performance of Fibrosis-4 Index in this setting.