Impact of Sustained Virologic Response on Risk of Type 2 Diabetes Among Hepatitis C Patients in the United States

J. Li; T. Zhang; S. C. Gordon; L. B. Rupp; S. Trudeau; S. D. Holmberg; A. C. Moorman; P. R. Spradling; E. H. Teshale; J. A. Boscarino; M. A. Schmidt; Y. G. Daida; M. Lu


J Viral Hepat. 2018;25(8):952-958. 

In This Article


Patient Population

CHeCS is an observational multicenter study that includes adult (≥18 years) chronic hepatitis C patients from four large health systems. The study follows all guidelines of the US Department of Health and Human Services regarding the protection of human subjects. Protocols are reviewed annually by the institutional review board at each of the study sites: Geisinger Clinic, Danville, Pennsylvania; Henry Ford Health System, Detroit, Michigan; Kaiser-Permanente Hawai'i, Honolulu, Hawai'i; and Kaiser-Permanente Northwest, Portland, Oregon. CHeCS study methods have been previously described.[6] Briefly, electronic administrative data and electronic health records for patients ≥18 years who received health services at any study site between 1 January 2006 and 31 December 2015 were used to identify study candidates; eligibility was confirmed during medical chart abstraction.

For this analysis, patients were included if they had received HCV treatment prior to 31 December 2015. Patients were excluded if they had hepatitis B co-infection or pre-existing diabetes, which was defined as the presence of an International Classification of Diseases, version 9 or 10 (ICD9/10) diagnosis code for type 1 diabetes (ICD-9-CM code 250.x1/x3 and ICD-10-CM code E10.xxxx) or type 2 diabetes (ICD-9-CM code 250.x0/x2 and ICD-10-CM code E11.xxxx) in their electronic health record. "Index date" was defined as 12 weeks after the end of a patient's most recent course of HCV therapy. Prior HCV treatment history was captured as a baseline covariate.

Antiviral HCV Therapy and Response

Detailed antiviral medication data (drug name and start/stop dates) were collected via chart abstraction for the patient's most recent course of HCV therapy. Combination therapy was identified when the multiple hepatitis drugs were administered concomitantly. Data on routine HCV RNA quantification tests were obtained via the electronic health record. Patients were classified as having achieved SVR if laboratory tests collected ≥12 weeks post-therapy showed undetectable viral RNA loads; otherwise, the patients were classified as having treatment failure (TF).

Baseline Covariates and Possible Risk Factors

Baseline data (at the start of a patient's most recent course of HCV therapy) included the following: demographic information (age, sex, race/ethnicity, estimated median annual household income and insurance status); HCV genotype; Charlson/Deyo comorbidity score (calculated from inpatient, outpatient and claims data for 12 months prior to the index date); laboratory test results for imputation of the Fibrosis-4 (FIB4) score (based on our validated classification categories of ≤1.21, 1.21–5.88, >5.88 or "unknown"); alanine aminotransferase (ALT) levels (<45 U/L, ≥45 U/L or "unknown"); and HIV co-infection.

Baseline cirrhosis status was determined using data for 2 years prior to the start date of a patient's most recent course of HCV therapy. Due to the observational nature of this study, the availability of cirrhosis data varied. Roughly 20% of our sample had liver biopsy or Fibroscan data collection, and 60–70% had laboratory data for the calculation of FIB4. Cirrhosis data were sometimes inconsistent between the various sources for an individual patient. To overcome this variation, we implemented the following hierarchical classification algorithm to identify cirrhosis: (i) decompensated cirrhosis identified using our validated Classification and Regression Tree model;[7] (ii) F4 liver biopsy or Fibroscan >12.5; (iii) FIB4 > 5.88;[8] and (iv) the presence of ICD9/10 diagnosis codes for cirrhosis.

Statistical Analysis

Time-to-event outcomes included the incidence of T2D (defined using ICD9/10 codes) or death. Follow-up began after the achievement of SVR or, for treatment-failure patients, the date 12 weeks after HCV treatment initiation. Patients were followed until the outcome of interest or 10 years of follow-up. Cox proportional hazards regression was used to estimate the impact of SVR and other factors on the risk of T2D; death was included using a competing risk approach. To assess the influence of additional risk factors on any potential SVR effect, we began the analysis by testing each individual risk factor effect, considering any possible risk factor-by-SVR interactions. Variables demonstrating either individual effects or interactions with SVR (P < .05) were included in the multivariable model; those with significant effects (P < .05) or interactions with SVR after adjustment for other covariates were retained in the final model using a forward model selection approach. Study site was included in all multivariable analyses as an adjustment variable. Because FIB4, ALT and cirrhosis are highly correlated, they were fitted into multivariable models separately to avoid confounding effects.