The central idea of ELYNS is to make the scholarly
literature and scientific wisdom free and open to the
scientific as well as the common world.
more

Active Disease Surveillance in an Electronic Health Database Using a Common Data Model Based Tool

Published Date: January 18, 2018

Active Disease Surveillance in an Electronic Health Database Using a Common Data Model Based Tool

Denise M. Oleske*, and Raghava Danwada

Benefit-Risk, Innovative Platforms and Epidemiology, Safety Science, Pharmacovigilance & Patient Safety, AbbVie Inc., AP51-2, 1 North Waukegan Road, North Chicago, IL 60064, USA

 

*Corresponding author: Denise M. Oleske, Benefit-Risk, Innovative Platforms and Epidemiology, Safety Science, Pharmacovigilance & Patient Safety, AbbVie Inc., AP51-2, 1 North Waukegan Road North Chicago, IL 60064, USA, Tel: 847-937-2848; Fax: 847-938-8438; E-mail: denise.oleske@abbvie.com

 

Citation: Oleske DM, Danwada R (2018) Active Disease Surveillance in an Electronic Health Database Using a Common Data Model Based Tool. J Epid Prev Med 3(3): 138.

 

Abstract

 

Background: There is an increasing global interest for rapid, active disease surveillance with a population based context, especially for early or novel interventions. The use of electronic health databases with reproducible cohort identification tools with linkage capability offer promise for generating this efficiently.

Objective: The aim of this study was to evaluate the Cohort Identification and Descriptive Analysis (CIDA) tool developed through the FDA’s Sentinel Initiative for application in active disease surveillance by determining the incidence of a low frequency event hepatic decompensation (HD), a clinical outcome that can occur in the natural history of the disease or as an outcome during treatment for hepatitis C infection. 

Design: A prospective design, using the CIDA tool to construct a cohort treated with interferon and ribavirin for hepatitis C infection, was applied to a FDA Sentinel Common Data Model (CDM) formatted database covering the period of 1/1/2005 to 6/30/2015.

Main Outcome Measures: The incidence of HD in the time intervals: < = 2, > 2–4, > 4–12, and > 12–24 weeks after treatment initiation and by age and sex.

Results: The HD incidence rate increased since treatment start over the study intervals, increased with age, but there was no difference between the sexes. The HD rates obtained from the CIDA tool applied to an electronic health database were similar to that found from observational studies using primary data collection.

Conclusion: Cohort construction for active surveillance using large electronic claims database to descriptively characterize the incidence rate of a low frequency event whose magnitude cannot be able to be accurately characterized early from numerator only events in routine surveillance offers promise as a new population-based preventive medicine strategy.

 

Keywords: Surveillance epidemiology; Disease surveillance; Cohort identification; Hepatitis C; Infectious disease; Public health practice

 

Introduction

 

There is increasing global interest for identifying health outcomes in populations using widely available electronic health databases as a means of surveillance. The term “health outcome” refers to an event resulting from an intervention aimed at controlling or preventing disease. It can also refer to unintended consequences potentially resulting from the delivery of an intervention. Unintended unexpected events affecting the safety profile of an intervention may not be observed in the studies of the efficacy of the intervention as these studies may typically have fewer persons or are not powered to assess low frequency events other than efficacy. Fundamental to the new paradigm for disease or health outcome surveillance, regardless of setting, intervention, or population, is the need to understand the application of rapid cohort construction using “tools” applied to large electronic health databases that can be used in public health practice. The term “tools” used in this context refers to structured queries of a database which are comprised of SAS macros. There are a number of cohort identification or selection tools and approaches that have been described in the literature or found from internet searches [1–4]. Web based tools include those available from University of North Carolina’s The Carolina Data Warehouse for Health (CDW-H, https://tracs.unc.edu/index.php/services/biomedical-informatics/cdw-h), i2b2 distributed through a public portal based at Harvard Medical School (i2b2.org), Observational Health Data Sciences and Informatics (OHDSI, ohdsi.org), Northwestern University’s CAPRICORN (http//capricorncdrn. org and Stanford’s Cohort Discovery Tool (https://med.stanford.edu/researchit/tools/cohort-tool.html). These tools are designed to function on databases unique to their respective organizational configuration, may use open source or other specialized software, and special requirements and permissions may be in place to access the cohort construction tools. The main feature of all of these tools is the sole focus on the descriptive characterization of a cohort and typically restricted to a specific cohort of interest for evaluating a specific research question, such as one related to comparative effectiveness. However, persons interested in cohort construction from electronic health databases the following must be taking into consideration. First, critical to any cohort identification tool is the ability to reproduce the cohort for external validity using accepted programming code. Secondly, a particular need in cohort identification tools is robustness, namely, the ability to integrate its configuration to allow for active surveillance or longitudinal analysis of the health outcome of interest given various exposures or interventions. This latter point may require special tables and coding to allow start and stop dates of exposure, dosing, and formulations and confounding. And, lastly a further need is the evaluation of such tools in the ability to generate reproducible results characterizing health outcomes across treated or populations with an exposure from different care settings and organizations.

It is for the above reasons that the publicly available Cohort Identification and Descriptive Analysis (CIDA) tool, developed by the Federal Drug Administration (FDA) through its Sentinel Initiative, tools for active surveillance (https://www.sentinelinitiative.org) was chosen for evaluation [4]. Tools are derived in the Sentinel Initiative are based upon its legacy Mini-Sentinel Common Data Model (MSCDM) which is the standardization of data sources with different formats and variables such as electronic medical records maintained by health care providers, health insurance claims data and patient registries that are generated from patient interactions with the US healthcare system through their insurers, places of care, and providers. The Sentinel surveillance strategy, Active Risk Identification and Analysis (ARIA) offers a library of tools that have been developed to conduct surveillance and evaluation of health outcomes from treatment settings (Figure 1). These tools are publicly available and facilitate the statistical programmer in building reusable, flexible and scalable programs for the rapid construction cohorts and the analysis of exposure-outcome relationships rather than having to develop a series of one-of computer programs. Among the publicly available tools that have been developed include those for tabular summary of counts, cohort identification, descriptive analysis (prevalence and incidence rates), propensity score matching, Charlson/Elixhauser combined comorbidity score, medical utilization metrics, and self-controlled risk interval design. These publicly available tools, in addition to cohort identification, are fundamental to all study designs applying a prospective approach when evaluating the occurrence and magnitude of the risk of an event. The CIDA tool is categorized as a “Level 1” capability within the Sentinel ARIA. CIDA is one of pre-defined, parameterized, re-usable querying tools comprised of SAS coding so as to routinize safety surveillance. The tool may be applied in a variety settings which have unique patient volume data and where currently only an incident case (numerator) is assessed, such as in monitoring healthcare quality, infectious disease surveillance, and surveillance of adverse outcomes in marketed medical products by providing a more robust estimation and characterization of risk of an event through its cohort approach.

The CIDA tool was selected for this evaluation in consideration of its potential use in active disease surveillance with the focus on a cohort chronically infected with hepatitis C viral infection (HCV). The Centers for Disease Control and Prevention (CDC) estimated that 30,500 cases of acute hepatitis C occurred in 2014, a rate of 0.7 cases per 100,000 population and an increase from 2010–2012. Although the incidence of HCV in the US is declining largely due to improved screening of the blood supply and population, HCV is still a major public health problem. Of newly infected persons, 75%–85% develop chronic infection; 60%–70% of chronically infected develop chronic liver disease; and 5%–20% develop cirrhosis over a period of 20–30 years. The prevalence of chronic HCV infection in the US is estimated to range from 2.7 to 3.9 million persons. Sequellae of cirrhosis include hepatic decompensation and death (1%–5% will die from cirrhosis or liver cancer). The CDC reported there were 19,659 deaths in 2015 in the US associated with HCV infection. Approximately one-half of all these deaths occurred among persons aged 55–64 years. The number of deaths associated with HCV is likely underestimated in part due to lack of recording on death certificates despite evidence of substantial liver disease and unrecognized infections among the segments of the population at highest risk for HCV [5]. Although new directly acting antiviral (DAAs) medications offer shorter duration of treatment with a high cure rate, it is estimated that the prevalence of hepatic decompensation (HD) will continue to increase, with approximately 145,000 persons with HD in the HCV population projected by 2020 unless more persons with HCV are treated with an effective regimen [6]. And, persons with HCV infection developing HD are at increased risk of re-hospitalization as well as death [7,8].

The purpose of this study was to determine the ability of the CIDA tool for active surveillance by determining the magnitude and trends of hepatic decompensation, a low frequency health outcome, in a cohort with HCV infection treated with interferon (IFN) concomitant with ribavirin (RBV). To the authors’ knowledge, this is the first time a cohort was constructed from a common data model formatted database to perform active surveillance of a complex medical concept, hepatic decompensation (HD), in a population with a chronic disease (hepatitis C infection). HD may occur as part of progression of the natural course of disease or due to other factors, such as older age, male gender, duration of HCV infection, method of HCV transmission or treatment. For decades, IFN monotherapy regimens had been the primary treatment regimen for HCV infection [9]. Widely accepted criteria of persons for whom this therapy would be initiated included: HCV RNA positive in the serum, presence of significant fibrosis, acceptable hematological and biochemical values, and compensated liver disease but no evidence of hepatic decompensation [10]. Advances in antibody and molecular testing for characterizing the HCV virus along with the changing epidemiology of HCV, have resulted in regimes with IFN less commonly used for the treatment of chronic HCV infection in Western nations in favor of directly acting antiviral agents without IFN [5,9]. Professional organizations subsequently supported treatment for all HCV-infected persons in consideration of HCV genotype, except those with limited life expectancy [11]. Recommendations continue to evolve with accumulating evidence from clinical and laboratory studies. Nevertheless, the results from the evaluation of the CIDA tool can provide a strategy for constructing a historical background rate of HD that could be used for surveillance of the occurrence of early HD during treatment with the new HCV treatment regimens.

 

Methods

 

Dataset

A commercial medical claims database without personal identifiers (MarketScan®) was converted to the MSCDM (Ver. 4.0) format by EPHIR, Inc. (Boston, MA) for the time period 1/1/2005 to 06/30/2015. The MarketScan® database was used because it had the largest total number of unique persons averaging three years of continuous enrollment with enrollee representation from a wide range of US geographic areas.

Data Collection and Processing

The CIDA tool has the capability to flexibly identify and extract cohorts of patients from MSCDM formatted data based on variety of user specified options (e.g. study dates, exposure definition, outcome definition, incidence criteria, continuous enrollment requirements, age groups and so forth). The reusable code in CIDA has been constructed to include over 30 highly parameterized SAS macros, each performing a distinct function, making it flexible, transparent and easily maintainable system library. Thus, another reason for selecting this tool is that a highly regulated environment requires validation of programming code. The CIDA tool has been evaluated through the Sentinel Initiative for over 20 health outcomes using SAS code, statistical software accepted by governmental regulators.

The MSCDM consists of a suite of several tables: six of them are considered core for our use of the CIDA tool (enrollment, demographic, outpatient, dispensing, encounter, diagnosis, procedure). There are additional tables including death, cause of death, laboratory tests, and vital signs. However, it found that the CIDA tool in the publicly available version online at the time this study was conducted did not link to the death table. Thus, programming statements had to be created to link the CIDA tool to the death table for purposes of adjusting the denominator (“population at risk”) for the computation of the event probabilities over the study time periods. Options in the MSCDM tables for the number of digits used to extract diagnoses codes are available. The 5-digit option was used in the present study. Reference tables for assigning lab names, units of measures, and abbreviations are also available for the CIDA tool to link in a MSCDM formatted database. Summary tables are a useful feature of this cohort construction tool and serve both as a programming check and guide for sample size estimates. Table 1 lists the tables that could be used when applying the CIDA tool for cohort construction. Summarized in Table 2 are the features of the CIDA tool.

The study inclusion criteria were: aged 18+ years; both sexes; continuous enrollment for 365 days with both pharmacy and medical benefits allowing for an enrollment gap of 45 days; chronic hepatitis C infection determined by ICD-9 diagnoses codes as (one of 070.44 or 070.54, or two diagnosis codes of 070.70, 070.71, or V02.62 on separate days);and, received interferon (IFN) with ribavirin. Excluded were those who had an event of HD within the six months prior to the first prescription of IFN and who had received a directly acting antiviral agent (DAA) (boceprevir, telaprevir, sofosbuvir, simeprevir, ledipasvir /sofosbuvir, daclatasvir, ombitasvir, paritaprevir, or dasabuvir) for treatment of HCV infection. As there is no single ICD-9 diagnosis code for hepatic decompensation (HD), the medical concept was constructed from diagnoses codes found in the literature and with the consensus of physicians experienced in treating hepatitis C infection [8,12,13]. The presence of any one of following ICD-9 diagnoses codes and their corresponding labels were used to determine the occurrence of HD: 456.0 Esophageal varices with bleeding; 456.20 Esophageal varices in diseases classified elsewhere with bleeding; 567.23 Spontaneous bacterial peritonitis; 572.2 Hepatic encephalopathy; 572.3 Portal hypertension; 572.4Hepatorenal syndrome; 789.5 Ascites; 789.51 Malignant ascites; or 789.59  Other ascites. The occurrence of the initial ICD-9 diagnosis code for chronic HCV infection in the study time interval was defined as the “index diagnosis.”“Index prescription” was the date of the first prescription of IFN within 30 days after the index diagnosis and co-administered with ribavirin if it occurred within the first 7 days of the index prescription. An “early event” was the first HD diagnosis occurring within six months after the index prescription. The CIDA tool was used to extract the drugs and diagnoses of interest and apply the enrollment criteria to construct the cohort. The flow of the extraction process and resultant sample sizes are displayed in Figure 2.

The incidence of HD and 95% confidence intervals was performed linking the cohort to SAS ver. 9.4 for the study time intervals: < = 2, > 2–4, > 4–12, > 12–24 weeks. The 24 weeks of observation was selected to correspond to the shorter treatment durations of the newer interferon-free DAAs for HCV, thereby providing a historical background rate of HD in a treated HCV population. Six deaths occurred in the cohort within first 24 weeks, but 3 HD events occurred before the death and the denominators were adjusted accordingly in the calculation of the incidence rates according to study time intervals. 

As a sensitivity analysis, diabetes mellitus (DM) was substituted for HD in the CIDA tool because it has only one diagnosis code (ICD-9, 250), an expected higher frequency and greater risk in persons with HCV than those without HCV, higher in treated HCV persons than matched controls without HCV, and linked with faster HCV progression [14–19]. The same inclusion and exclusion criteria were applied as for the main analysis of HD. The study conduct adhered to ethical considerations and good pharmacovigilance processes as described in the documents from the European Medicines Agency, “Guideline on Good Pharmacovigilance Practices (GVP) – Module VIII (www.encepp.eu), from the European Network of Centers for Pharmacoepidemiology and Pharmacovigilance, and “Guide on Methodological Standards in Pharmacoepidemiology” (www.encepp.eu); and from the FDA Guidance, “Best Practices for Conducting and Reporting Pharmacoepidemiologic Safety Studies Using Electronic Healthcare Data” (https://www.fda.gov).

 

Results

 

Sample Population

A cohort of 2,883 unique patients with HCV infection was identified with the CIDA tool. The sample was predominantly males (62%), and middle-aged adults (mean age: males = 49.5 +/- 9.2 years, females = 48.0 +/- 10.1 years). The average duration between the diagnosis of chronic HCV infection and the initiation of IFN treatment was less than one month. There were 38 HD events within the six-month follow-up period, with the frequency of first diagnosis as follows: portal hypertension, 34.2%; ascites, 26.3%; hepatic encephalopathy, 26.3%; esophageal varices, 10.5%; and spontaneous bacterial peritonitis, 2.6% (Table 3).

Risk of Outcome

Although males were 1.7 times more likely to develop HD than females, the incidence rates were not significantly different (p = 0.13). The incidence rate of HD increased with age, with the overall incidence of 13.2 per 1000 (Table 4). Those developing HD were significantly older than those without HD (mean age: HD, 54.6 years; no HD 48.9 years, p < 0.001). The incidence of HD occurring after treatment initiation increased over the 24 weeks of observation and was a statistically significant increase (chi-square trend, p = 0.02) (Table 5).

Sensitivity analysis found the overall incidence of DM by 24 weeks of treatment initiation to be three-fold higher 40.9 per 1000, 95%CI: 34.3-48.8) than that of HD, and also exhibited an increasing trend after treatment initiation (data not shown).

 

Discussion

 

This study provides insight and evaluation of a new model for active disease surveillance based upon electronic health records which can be used in public health practice. While a number of cohort selection tools are available, there is no published peer-reviewed literature to the authors’ knowledge on the linkage of a cohort tool to the occurrence of health events in consideration of exposures of interest. This is a report of the functionality of a cohort construction tool for active surveillance of a low frequency health outcome over successive time intervals. Using the CIDA tool, a complex treated cohort was created to characterize the incidence of a complex health outcome, hepatic decompensation, over a pre-defined time period. The results generated from this tool were similar to that reported in the literature, in which the annualized incidence of HD in those with HCV treated with IFN-based therapy ranged from 1.2 to 3.9% [8,20,21]. The sensitivity analysis examination of the DM incidence rate was found to be higher than the HD rate, as expected, and also similar to the rate found among those with HCV in two community-based longitudinal studies [14,15]. The increasing incidence of HD and DM found over the study intervals may reflect the natural history of disease progressing or a treatment effect. The CIDA tool detected an increase in HD over time since initiation of IFN treatment which may be a reflection of the natural history of the disease and or may not be related to treatment. Whether or not the incidence of HD would continue to increase in patients receiving longer treatment duration with IFN and RBV (up to 48 weeks without DAAs) or some other intervention, remains a subject for further study. The data generated by the CIDA tool herein provides information which may be used in designing a comparator study using the newer treatment regimens for HCV and for determining the statistical power of a study required to detect potential differences among comparators for endpoints of interest.

The Sentinel Common Data Model is not the only CDM format available for electronic databases for health informatics researchers to consider. PCORnet and the OMOP approach have been discussed in the literature and their documentation is also available online [1,2]. The main feature of the structure of the common data model format of these two sources is the clinical granularity and availability of additional lexicons for defining medical concepts. PCORnet CDM (http://www.pcornet.org/pcornet-common-data-model/) has as the primary goal to provide a database to evaluate effectiveness of defined intervention, typically driven by a study protocol, whereby the evaluation ends with the end of the study. However, if the goal of a cohort extraction tool is for active safety surveillance, it is essential that the tool generates rapid and reproducible results tool for identifying and assessing a safety-related outcome longer than the conclusion of an effectiveness study. Standardized advanced tools, such as the propensity matching tool available in Sentinel’s ARIA, can explore the significance of a potential risk of a health event relative to some other intervention as a next step accounting for confounding. The standardization of such an approach allows public health practitioners and researchers to facilitate a common dialog about a potential risk of a safety event or other health outcomes using the same tool set.

There are limitations of the study. Selection bias may be operational if fewer patients with cirrhosis who are at higher risk for HD were treated. Cirrhosis status of the cohort was not determined as all relevant clinical and laboratory tests were not available. However, with the initiation of HCV treatment, it is assumed patients in the cohort had some level of disease severity which prompted intervention. Because of the low mortality in this sample, the study cohort as historical reference may then be only appropriate for those treated for HCV who have asymptomatic cirrhosis. Another limitation of the study is the small proportion of those aged 65+ years (0.76%) relative to the US population. Thus, the HD incidence may be underestimated as our data found a higher percentage of HD among elderly.

The execution of the CIDA tool was technically challenging as the publicly available SAS code required modifications in order to link to the CDM data tables. But, despite low HD incidence found in this study, the tool was able to successfully identify and characterize it over time intervals yielding rates similar to those found in observational studies. Refinement of the magnitude of the detected HD risk through use of the CIDA tool could be undertaken using additional advanced analytic and design tools in Sentinel as a next step, such as matching patients in the cohort created to patients in a comparator cohort based on propensity score computed from pre-defined covariates. Complex adjustment for confounding as part of prospective sequential analysis, when publicly available in ARIA, linked to a cohort created by the CIDA tool, could evaluate the significance of the risk observed over successive time periods. The cohort identification process, when intended for active surveillance, can be complicated. Table 6 is offered as a guide to selecting a cohort identification tool based upon our experience with the CIDA tool along with anticipating other requirements that may be necessary for its functionality. The Sentinel system continues to evolve. Thus, the use of a particular tool or set of tools requires attention to the version used in order to be appropriately applied to the corresponding version of the CDM.

 

Conclusion

 

The CIDA tool may be an efficient approach for creating cohorts in large electronic health databases to identify low frequency events, and thus, be useful for active surveillance of health outcomes in public health practice. The results can guide if further surveillance is warranted or a more refined analysis which includes comparators should be undertaken.

 

Acknowledgements

 

We thank Drs. Fabio Lievano, Mondira Bhattacharya, and Ryan D. Kilpatrick for their comments on the protocol development and the manuscript.

 

Disclaimer and Conflict of Interest

 

Dr. Oleske is an employee of AbbVie, Inc. Mr. Danwada is an employee of AbbVie Inc. EPHIR, Inc. received funding from AbbVie for the creation of the Common Data Model formatted database and for programming consultation. The content and conclusions stated in the manuscript are that of the authors’ not that of AbbVie nor of EPHIR.

 

References

 

  1. Bucur A, van Leeuwen J, Chen N-Z, Claerhout B, de Schepper K, Perez-Rey D, et al. Cohort Selection and Management Application Leveraging Standards-based Semantic Interoperability and a Groovy DSL. AMIA Jt Summits Transl Sci Proc. 2016;2016:25-32.
  2. Chakrabarti S, Sen A, Huser V, Hruby GW, Rusanov A, Albers DJ, et al. An interoperable similarity-based cohort identification method using the OMOP Common Data Model Version 5.0. J Health Inform Res. 2017;1:1–18.
  3. FitzHenry F, Resnic FS, Robbins SL, Denton J, Nookala L, Meeker D, et al. Creating a common data model for comparative effectiveness with the Observational Medical Outcomes Partnership. Appl Clin Inform. 2015;6:536-47. doi: 10.4338/ACI-2014-12-CR-0121.
  4. Food and Drug Administration (FDA). Sentinel Modular Programs. Querying Tools: Overview of Functionality and Technical Documentation, 2016. Available from: https://www.sentinelinitiative.org/sites/default/files/SurveillanceTools/RoutineQuerying/Sentinel-Routine_Querying_System-Documentation_0.pdf. [Accessed on February 15, 2017].
  5. Centers for Disease Control and Prevention (CDC). Viral Hepatitis Surveillance, United States, 2015. Available form: https://www.cdc.gov/hepatitis/statistics/2015surveillance/index.htm. Accessed on January 2, 2017.
  6. Davis GL, Alter MJ, El-Serag H, Poynard T, Jennings LW. Aging of hepatitis C virus (HCV)-infected persons in the United States: a multiple cohort model of HCV prevalence and disease progression. Gastroenterology. 2010;138(2):513-21, 521.e1-6. doi: 10.1053/j.gastro.2009.09.067.
  7. Planas R, Balleste B, Alvarez MA, Rivera M, Montoliu S, Galeras, JA, et al. Natural history of decompensated hepatitis C virus-related cirrhosis. A study of 200 patients. J Hepatol. 2004;40(5):823-30.
  8. Dienstag JL, Ghany MG, Morgan TR, Di Bisceglie AM, Bonkovsky HL, Kim HY, et al. A prospective study of the rate ofprogression in compensated, histologically advanced chronic hepatitis C. Hepatology. 2011;54(2):396-405. doi: 10.1002/hep.24370.
  9. Ghany MG, Strader DB, Thomas, DL, Seeff LB. Diagnosis, management and treatment of hepatitis C; an update. Hepatol. 2009;49(4):1335-74. doi: 10.1002/hep.22759. (American Association for the Study of Liver Diseases).
  10. Heim MH. 25 years of interferon-based treatment of chronic hepatitis C: an epoch coming to an end. Nat Rev Immunol. 2013;13:535–42. doi: 10.1038/nri3463.
  11. AASLD/IDSA HCV Guidance Panel. Hepatitis C guidance: AASLD-IDSA recommendations for testing, managing, and treating adults infected with hepatitis C virus. Hepatology. 2015;62(3):932-54. doi: 10.1002/hep.27950.
  12. Gomez EV, Rodriguez YS, Bertot LC, Gonzalez AT, Perez YM, Soler EA, et al. The natural history of compensated HCV-related cirrhosis: A prospective long-term study. J Hepatol. 2013;58(3):434-44. doi: 10.1016/j.jhep.2012.10.023.
  13. Gordon SC, Lamerato LE, Rupp LB, Holmberg SD, Moorman AC, Spradling PR, et al. Prevalence of cirrhosis in hepatitis C patient in the Chronic Hepatitis Cohort Study (CHeCS): a retrospective and prospective observational study. Am J Gasteroenterol. 2015;110(8):1169-77; quiz 1178. doi: 10.1038/ajg.2015.203.
  14. Antonelli A, Ferrari SM, Giuggioli D, Di Domenicantonio A, Ruffilli I, Corrado A, et al. Hepatitis C virus infection and type 1 and type 2 diabetes mellitus. World J Diabetes. 2014;5(5):586-600. doi: 10.4239/wjd.v5.i5.586.
  15. Louie KS, St Lauren S, Forssen UM, Mundy LM, Pimenta JM. The high comorbidity burden of the hepatitis C virus infected population in the United States. BMC Infect Dis. 2012;12:86. doi: 10.1186/1471-2334-12-86.
  16. Mehta SH, Brancati FL, Strathdee SA, Gonzalez AT, Perez YM, Soler EA, et al. Hepatitis C virus infection and incident Type 2 diabetes. Hepatol. 2003;38:50-6.
  17. White DL, Ratziu V, El-Serag HB. Hepatitis C infection and risk of diabetes: A systematic review and meta-analysis. J Hepatol. 2008;49(5):831-44. doi: 10.1016/j.jhep.2008.08.006.
  18. Lingala S, Ghany MG. Natural history of hepatitis C. Gastroenterol Clin North Am. 2015;44(4):717-34. doi: 10.1016/j.gtc.2015.07.003.
  19. Wang CS, Wang ST, Yao, WJ, Chang TT, Chou P. Hepatitis C virus infection and the development of type 2 diabetes in a community-based longitudinal study. Am J Epidemiol. 2007;166(2):196-203.
  20. Petta S, Di Marco V, Bruno S, Enea M, Calvaruso V, Boccaccio V, et al. Impact of virus eradication in patients with compensated hepatitis Cvirus-related cirrhosis: competing risks and multistate model. Liver Int. 2016;36(12):1765-1773. doi: 10.1111/liv.13156.
  21. Coverdale SA, Khan MH, Byth K, Lin R, Weltman M, George J, et al. Effects of interferon treatment response on liver complications of chronic hepatitis C: 9-year follow-up study. Am J Gastroenterol. 2004;99(4):636-44. 

 

Copyright: © 2018 Oleske DM, et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.