BACKGROUND & AIM: Renal impairment is associated with chronic hepatitis B (CHB). To overcome prior study design differences, we used propensity score matching to balance the non-CHB and CHB cohorts and generalized linear modelling (GLM, models using probit and logit linking functions for complex models) to evaluate the effect of CHB, treatment and cirrhosis on renal function. METHODS: A retrospective cohort (1996-2017) from one U.S. university medical centre. Included patients had ≥12 months of serial creatinine laboratories and a baseline estimated glomerular filtration rate (eGFR, by the Modification of Diet in Renal Disease Study equation) ≥60 mL/min/1.73 m2 . Propensity score matching was performed using age, sex, ethnicity, diabetes, hypertension and baseline eGFR. GLM was performed to generate adjusted mean eGFR over time. RESULTS: Adjusted mean eGFR was significantly higher for non-CHB vs. untreated CHB patients (eGFR 87.4 vs. 85.6, P= 0.004, n = 580, median follow-up = 82 months). A significant difference in adjusted mean eGFR between untreated vs. entecavir (ETV)-treated CHB patients (eGFR 85.1 vs. 83.5, P= 0.02, n = 340, median follow-up = 70 months) was found among non-cirrhotic CHB. Among treated CHB, there was no difference in adjusted mean eGFR between non-cirrhotic vs. cirrhotic patients (eGFR 77.0 vs. 76.5; P= 0.66, n = 112, median follow-up = 58 months). CONCLUSION: After PSM and GLM, the significant predictors for worsening renal function were age, hypertension and diabetes mellitus but not CHB, ETV or cirrhosis. However, given small sample size, data regarding the use of ETV in patients with cirrhosis should be interpreted with caution and requires additional investigation.
BACKGROUND & AIM: Renal impairment is associated with chronic hepatitis B (CHB). To overcome prior study design differences, we used propensity score matching to balance the non-CHB and CHB cohorts and generalized linear modelling (GLM, models using probit and logit linking functions for complex models) to evaluate the effect of CHB, treatment and cirrhosis on renal function. METHODS: A retrospective cohort (1996-2017) from one U.S. university medical centre. Included patients had ≥12 months of serial creatinine laboratories and a baseline estimated glomerular filtration rate (eGFR, by the Modification of Diet in Renal Disease Study equation) ≥60 mL/min/1.73 m2 . Propensity score matching was performed using age, sex, ethnicity, diabetes, hypertension and baseline eGFR. GLM was performed to generate adjusted mean eGFR over time. RESULTS: Adjusted mean eGFR was significantly higher for non-CHB vs. untreated CHB patients (eGFR 87.4 vs. 85.6, P= 0.004, n = 580, median follow-up = 82 months). A significant difference in adjusted mean eGFR between untreated vs. entecavir (ETV)-treated CHB patients (eGFR 85.1 vs. 83.5, P= 0.02, n = 340, median follow-up = 70 months) was found among non-cirrhotic CHB. Among treated CHB, there was no difference in adjusted mean eGFR between non-cirrhotic vs. cirrhotic patients (eGFR 77.0 vs. 76.5; P= 0.66, n = 112, median follow-up = 58 months). CONCLUSION: After PSM and GLM, the significant predictors for worsening renal function were age, hypertension and diabetes mellitus but not CHB, ETV or cirrhosis. However, given small sample size, data regarding the use of ETV in patients with cirrhosis should be interpreted with caution and requires additional investigation.