Population-Based Return on Investment of Deploying Transient Elastography

Mazen Noureddin, MD

,
Andrew Mackenzie, BS

,
Elaine Zhao, BS

,
Scott C. Howell, DO

,
Michael Tunkelrott, MBA

,
Ian Duncan, PhD

The American Journal of Managed Care, September 2021, Volume 27, Issue 9

Deploying vibration-controlled transient elastography/controlled attenuation parameter devices at the population level is a financially advantageous solution to address the epidemic of fatty liver disease.

ABSTRACT

Objectives: To evaluate the cost savings outcomes, from the payer’s perspective, of deploying vibration-controlled transient elastography/controlled attenuation parameter (VCTE/CAP) machines for detecting and monitoring fatty liver disease (FLD).

Study Design: We modeled disease transitions and costs under the current observed pathway and under an alternative pathway in which VCTE/CAP devices are adopted. Marginal savings (or costs) due to implementing the device are derived by comparing the aggregate costs between the 2 pathways. Sources of potential savings are 2-fold. First, VCTE/CAP tests result in early identification of patients with FLD (the majority are currently undiagnosed), allowing for proactive intervention and behavior change to slow the progression of disease in these patients. Second, VCTE/CAP tests can reduce the aggregate volume of some current diagnosis methods, such as liver biopsy, imaging, and laboratory work.

Methods: Our model relied on administrative claims data consisting of 5 million commercial members and 3 million Medicare members to inform baseline statistics on disease prevalence, health care cost and utilization, and disease progression associated with different severities of liver disease. We consulted expert clinical opinion and medical literature to inform our assumptions related to device adoption and use.

Results: Scenario testing demonstrated positive net savings within 2 to 3 years after device deployment. Across a 5-year time span, we estimate net savings up to $2.64 per member per month (PMPM) for Medicare payers and up to $1.91 PMPM for commercial payers.

Conclusions: We conclude that deploying VCTE/CAP devices is a financially advantageous solution to address the epidemic of FLD.

Am J Manag Care. 2021;27(9):In Press. https://doi.org/10.37765/ajmc.2021.88645

_____

Takeaway Points

Chronic liver disease and cirrhosis are leading causes of death in the United States. Given the high prevalence and cost of fatty liver disease (FLD), it is paramount to implement accurate, easy-to-use tests to identify individuals who are at risk for advanced liver disease. We constructed an economic model to assess the cost savings outcomes of deploying transient elastography to identify and monitor patients with FLD in a payer population. Outcomes are evaluated across multiple dimensions and scenarios, and we estimate up to $2.64 per member per month net savings over 5 years.

  • Although several papers have outlined cost-effectiveness of noninvasive diagnosis models, we are not aware of any analysis that incorporates the cost of the testing device, which is an important consideration for executives when making investment decisions.
  • This is the first analysis to quantify potential cost savings at the population level as opposed to the individual level, making it more informative for business decision makers.
  • We relied on longitudinal, concrete claims data to calculate disease transition probabilities, frequencies, and costs, whereas most existing literature gathers such data points through meta-analysis.

_____

Chronic liver disease and cirrhosis are leading causes of death in the United States,1 and the spectrum of fatty liver disease (FLD) is among the most common liver conditions. FLD is characterized by the accumulation of fat in liver cells and further distinguishes between nonalcoholic fatty liver disease (NAFLD) and alcoholic fatty liver disease (AFLD) depending on whether heavy alcohol use is involved. Besides the distinction of alcohol use, both types of diseases have similar diagnostic pathways.

NAFLD can progress to a more advanced form known as nonalcoholic steatohepatitis (NASH).2 NASH consists of inflammation and cellular injury in addition to fat accumulation; it may progress to cirrhosis and hepatocellular carcinoma.2,3 A US prevalence study estimated that 85.3 million Americans had NAFLD and 17.3 million had NASH in 2016.4 These conditions contribute billions of dollars to the country’s health care costs.5 In addition, NAFLD is an independent risk factor for cardiovascular disease. Hence, NAFLD and NASH are emerging as important disease states for payers to manage.

AFLD has been estimated to affect about 5% of the US population.6 Although it is lower in prevalence compared with NAFLD, the burden of AFLD is still significant. AFLD surpassed hepatitis C virus (HCV) as the leading indication for liver transplants in 2016,7 and a 2018 US population-based study demonstrated that individuals aged 25 to 34 years have experienced the fastest increasing cirrhosis-related mortality in recent years, driven entirely by alcohol-related liver disease.8

Despite the high prevalence and costs, most patients with FLD are asymptomatic and undiagnosed. This has led to delayed identification of advancing liver disease, which can have significant human and economic costs. In fact, the American Diabetes Association recently called for assessment of NAFLD in patients with prediabetes and diabetes who have elevated liver enzymes to address the lack of timely identification of NAFLD.9

With the large population at risk for these adverse outcomes due to FLD, increasing payer focus, and looming US guidelines, cost-effective and easily implemented liver assessment strategies are needed at the point of care. Vibration-controlled transient elastography (VCTE) and controlled attenuation parameter (CAP) represent a potential solution. VCTE/CAP is a noninvasive screening procedure used to detect liver stiffness and fat, the 2 key identifying characteristics of FLD. In this study, we constructed an economic model to determine the potential return on investment (ROI) from deploying a VCTE/CAP–based care model to identify people with FLD in both Medicare and commercial payer settings.

METHODS

Data

Our economic model relied on administrative claims data consisting of 5 million commercial members in 2015 to 2017 from IBM’s MarketScan data set and 3 million Medicare members in CMS’ Limited Data Set from 2012 to 2016. These data sets were used to inform baseline statistics on disease prevalence rates, as well as health care cost and utilization associated with various levels of liver disease. Key data elements contained in each data source include member enrollment, diagnosis codes, procedure codes, and claim allowed amounts (insurance and member paid amounts).

Definitions and Member Identification

For the purpose of this model, we segmented the spectrum of FLD-related conditions into multiple severity levels: simple fatty liver, steatohepatitis and fibrosis, cirrhosis, end-stage liver disease (ESLD), and liver cancer. Disease levels were defined using the International Classification of Diseases, Tenth Revision diagnosis codes (see eAppendix A [eAppendices available at ajmc.com]). We identified and assigned members to disease levels based on this code list.10 Members with certain severe conditions that are likely to result in unavoidable high costs, such as cancer (non–liver-related), severe injuries or burns, and end-stage renal disease, were subsequently excluded from the analysis. eAppendix B contains the full list of exclusion conditions.

Disease Transition Probabilities

We modeled disease transition probabilities between each disease level using multistate Markov models. Claims data were summarized on a half-year basis, so the cycle length of the disease transition probabilities is half a year. We limited remission from members in levels 3, 4, and 5 because this is clinically unlikely to happen. In other words, once a patient acquires cirrhosis, ESLD, or liver cancer, the member can only remain in that level, progress to the next level, or leave the health plan.5 Furthermore, we require a patient to demonstrate remission of their condition in 2 consecutive periods to move down into a less severe level. These adjustments are necessary to counteract inconsistencies in medical claims coding and to reflect real-world experience more accurately. The resulting transition tables can be found in eAppendices C and D.

Economic Model Development

The model was built to project medical cost savings driven by the device, relative to the net cost of the device. Two sources of value generation are considered. First, VCTE/CAP tests result in earlier identification of patients with FLD, allowing for proactive intervention to slow down disease progression in these patients. Second, use of the VCTE/CAP test reduces the volume of current diagnosis methods, such as liver biopsy, imaging, and laboratory work (definitions of utilization categories can be found in eAppendix E). The aggregate effect of VCTE/CAP on utilization and cost is estimated by comparing the current-state diagnosis and treatment (usual care) pathways with future-state pathways (after the introduction of the device). The economic model includes the following parameters:

  • The population in the model is held constant over time (ie, the number of new entrants into the health plan is equal to the number of members who left the plan in the prior period either through death or change in coverage).
  • For the purpose of trending, we adopt an inflation rate of 2% and medical trend rate of 6%.11,12
  • Rental cost of the VCTE/CAP device is $2800 per month per unit.
  • Assuming that device adoption is limited to specialists and that it will ramp up gradually over time, one-third of eligible members are assumed to have access to the test by year 3. If the device is assumed to be deployed at point-of-care settings, we assume final penetration rates will be higher and that half of the total membership will have access by year 3. The number of devices assumed to be installed in each period is based on the penetration rates described above, a member-to–primary care physician ratio of 1500:1, and a member-to-specialist ratio of 3750:1.13,14 An implication for smaller payers is that as the plan population gets smaller, the concentration of members with access to specific physicians may become more diluted. As such, we have modeled various penetration assumptions that scale member access.

Clinical Assumptions

To quantify the downstream benefits of early identification of liver disease, we imputed the prevalence of undiagnosed liver disease at each disease level by subtracting the observed prevalence rates in claims data from the total estimated US prevalence rates reported in Estes et al.4 Because this study focuses solely on NAFLD, we scaled up the estimates according to the ratios observed in claims data to reflect the total prevalence of all FLD conditions. The resulting prevalence rates are presented in Table 1. Unsurprisingly, the diagnosis prevalence rates are much higher in the Medicare population compared with commercial, and much more so in the more severe disease levels.

Research findings suggest a high correlation between FLD and diabetes.2,15 To remain conservative with the impact and savings estimates, we assume that increases in screening of patients who unknowingly carry FLD are limited to the diabetes population. The number of incrementally diagnosed patients with liver disease is then calculated to be the portion of the population who have a liver condition, do not currently have a liver diagnosis, have diabetes, visit their doctor, and have access to a VCTE/CAP device at their doctor’s office. A formulaic representation of the logic is included in eAppendix F.

Although identifying more patients who unknowingly carry liver disease is beneficial for early intervention and outcomes improvement, it likely drives up health care costs in the short term as providers address the condition. Based on clinical expert input, we modeled increases in health care costs when patients with undiagnosed liver disease are made aware of their liver condition by assuming they receive a 1-time imaging test, additional annual office visits, and additional annual laboratory tests.

For the earlier stages of FLD, no effective treatment is demonstrated beyond weight loss and lifestyle changes. However, research has suggested that when patients are led to focus on their disease, they are more likely to engage in positive lifestyle changes that encourage recovery even without medical treatment. For example, in a meta-analysis of randomized controlled trials of NASH, around 21% to 33% of patients given placebo demonstrated significant improvements in various measures of liver health.16 Therefore, we model a 10% to 25% decrease in disease progression and/or increase in regression (smaller change for patients with less severe levels and vice versa) for those who are incrementally diagnosed. In addition, many payers offer chronic condition management (CCM) programs that drive positive lifestyle changes. The model incorporates the impact of CCM programs on the faster regression for patients with newly diagnosed FLD by referencing the impact of weight loss on fatty liver resolution previously studied through research.17,18 If a CCM program is available to induce positive lifestyle changes and weight loss, the model calculates an increased likelihood of improving steatosis and/or fibrosis when undiagnosed liver patients are identified and enroll in a CCM program. This increase ranges from 10% to 20% for the commercial population and 5% to 10% for the Medicare population.

The second component through which we quantify the impact of VCTE/CAP relates to its medical substitution effects. Using VCTE/CAP to diagnose liver conditions can replace certain procedures that are traditionally used for testing. The utilization reduction assumptions employed in the model are presented in Table 2, and the resulting utilization and cost impact are displayed in eAppendices G and H.

Negative VCTE/CAP results for fibrosis generally eliminate the need for liver biopsies and the need to continue prescribing imaging tests after an initial imaging. A large portion of complex laboratory tests may also be avoided. If VCTE/CAP results indicate fibrosis, however, confirmatory testing via biopsy may be prescribed. Alternatively, confirmatory testing can also shift to laboratory tests or imaging, which, given their noninvasiveness, may be preferred by patients.19 In addition, VCTE/CAP better distinguishes the precirrhotic patients from cirrhotic patients than the traditional Fibrosis-4 (FIB-4) score method and can thus avoid cancer screening for the precirrhotic patients, as well as reduce monitoring from the typical semiannual imaging and lab tests prescribed for patients with cirrhosis to once per year or less.20

Finally, if VCTE/CAP were to be made available at point-of-care settings, we also assume that this results in a 5% reduction in office visits for lower-severity patients (levels 1 and 2) to acknowledge potentially reduced referrals to liver specialists for the initial testing or monitoring of FLD.

RESULTS

Real-world outcomes vary based on the specific payer scenario. In this subsection, results are reported for representative combinations of input scenarios. The key levers that influence outcomes include projection time frame, market segment, device penetration expectations, device place-of-service setting, and availability of a CCM program. All savings are reported on a cumulative basis over the entire projection time frame. We also report savings on a per-member per-month (PMPM) basis over the entire population (not just those with FLD).

Table 3 illustrates the various outcomes scenarios for the Medicare model. Under scenario E, for example, a high device penetration level is modeled for a Medicare payer with 100,000 members deploying devices at specialists, where high device penetration corresponds to gradually ramping up device placement to give one-third of members access at their specialists over 2 years. This scenario produces an estimated 5-year cumulative gross savings of $10.1 million and a net savings (subtracting device rental) of $8.8 million. This corresponds to a 5-year 6.5:1 ROI and a net savings of $1.46 PMPM.

Compared with scenario E, scenario C produces a less robust financial bottom line at $1.41 net savings PMPM and a 1.7:1 ROI, driven by the disproportionate increase in device rental cost in comparison with the increase in medical savings. However, it is important to note that a point-of-care strategy will more effectively drive positive outcomes in a much broader patient base, while still producing significant savings compared with a specialist-only deployment strategy.

When factoring in the impact that CCM programs may have in terms of inducing positive lifestyle changes and curbing disease progression in those incrementally diagnosed patients with liver disease, we see the strongest financial outlooks of $15.9 million total net savings or $2.64 PMPM over 5 years in scenario A. This emphasizes the importance of pairing effective intervention strategies together with the diagnosis and monitoring solutions to drive the best outcomes.

For small payers (with a population of 10,000) we model more conservative penetration assumptions, and in practice it may be important for payers with diluted member-to-provider distribution to take steps in managing the concentration of membership per device to maintain high ROI, potentially via dedicated testing centers or different processes compared with deploying the devices at provider offices directly. Scenarios G and H produced smaller savings compared with their larger-payer counterparts in scenarios A and B, but the results are still positive and represent significant savings potential.

The Figure displays annual net savings results for scenario A and illustrates the time interplay of the ROI. Because the majority of savings are caused by a decrease in the speed of the progression of the disease, year 1 net savings tend to be negative as the projected aggregate cost and increased frequency of FLD testing outweigh the current state of FLD testing costs. Furthermore, the magnitude of net savings tends to increase over time, where there remain residual financial savings from members identified with FLD in earlier years.

Similar scenarios are constructed for the commercial model; results are provided in Table 4. The relationships among the various commercial scenarios are consistent with those among the Medicare scenarios described above. Overall, each Medicare scenario produces a stronger financial outcome compared with its commercial counterpart, due to the higher disease prevalence rates in the older population. The commercial scenario outcomes range from $11.4 million net savings ($1.91 PMPM, 2.2:1 ROI) over 5 years under scenario I to $0.3 million net savings ($0.83 PMPM, 1.6:1 ROI) over 3 years under scenario P.

DISCUSSION

The importance of early identification of liver disease is often overlooked, and the clinical pathways to stage liver disease have not been incorporated into the mainstream practice flows of primary care physicians. Most health plans have experience managing the high cost of treating HCV. Diagnosis and management of HCV can be performed through blood tests alone, but there is considerable variability in diagnostic workup for FLD, involving many testing modalities. High diagnostic variability combined with the lack of a clear diagnostic pathway expose health plans to unnecessary and preventable excessive costs across a large segment of their patient populations. Moreover, current testing patterns have left a vast majority of FLD cases currently undiagnosed and progressing at worrisome rates. By identifying disease early, when it is more easily mitigated, health deterioration and associated preventable future costs can be proactively managed or avoided. This economic model demonstrates that deploying VCTE/CAP devices to screen for and monitor liver stiffness and liver fat in members with diabetes can yield net savings to the payer, directly affecting bottom-line performance. Although short-term costs increase due to implementation of the testing devices and the identification and management of additional patients with active liver disease, the incremental cost is outweighed by the downstream savings from the avoidance of, or delay of progression to, advanced liver disease. The reductions in unnecessary referrals, biopsies, and imaging further increase cost savings.

The model also demonstrates that cost savings are increased when expanding access to a broader patient base. High penetration of the VCTE/CAP device in primary care, combined with more intensive behavioral engagement through CCM programs, yield the highest returns in both Medicare and commercial settings.

Although our model regards only the population with diabetes as being at high risk for FLD, it is important to note that many other indicators are also relevant to FLD, including obesity, metabolic syndrome, and cardiovascular disease. Applying the device beyond the diabetes population is likely to expand the clinical benefits to more patients who unknowingly have FLD. At the same time, marginal identification rates may decrease as service utilization climbs when more low-risk patients are included. This competing impact between savings and cost leaves opportunities for future analyses on this subject.

Specific to NAFLD, the VCTE/CAP technique can more accurately assess NAFLD over time than can blood test scores, such as the FIB-4 and NAFLD fibrosis scores, that rely on biomarkers found in blood.2 These blood test scores cannot determine the severity of NAFLD in one-third of the population, for whom a quantitative test, such as VCTE/CAP, is required. As health plans and clinical societies continue to emphasize the management of liver health, it is important to focus on implementing tests that are accurate and precise—yet easy to use—in the management of high-prevalence diseases.

Limitations

This analysis relies on a set of assumptions that are informed by clinical experts and the literature. We performed due diligence in verifying the reasonability of all assumptions but acknowledge the possibility for certain assumptions to deviate from reality. Our results should not be viewed as an exact prediction of real-life outcomes, but rather should be used to inform directionality and relative magnitudes by decision makers. For example, the actual prescribing patterns of physicians, once they are equipped with the VCTE/CAP device, may significantly vary depending on administration guidance, reimbursement structure, local clinical practice, and other factors. If physicians do not follow the screening guidelines for patients who are at risk of having FLD, for example, they would not realize the benefits of incrementally identifying a significant portion of patients with FLD who are currently unaware of their disease.

In addition, although granular clinical differences may exist between AFLD and NAFLD, the model relied on the assumption that AFLD and NAFLD generally follow similar diagnostic pathways and responses to lifestyle interventions. We deem this an appropriate assumption because VCTE/CAP applies equally to both types of FLD by detecting liver stiffness and fat, and the main treatment intervention recommended for FLD is healthy lifestyle changes and weight loss.

Finally, the pharmaceutical industry is aggressively developing treatments for this silent pandemic.21,22 The effect of such potential drug launches is not included in this model because of uncertainties around these launches.

CONCLUSIONS

Through modeling care pathways and expenditure, we demonstrated that deploying VCTE/CAP devices to screen for and monitor liver stiffness and liver fat for members with diabetes is economically beneficial to the payer. The short-term incremental costs of device implementation and managing additional patients are outweighed by the downstream savings from avoidance or delay of advanced liver disease.

Acknowledgments

Certain data used in this study were supplied by International Business Machines Corporation. Any analysis, interpretation, or conclusion based on these data is solely that of the authors and not International Business Machines Corporation.

Author Affiliations: Cedars-Sinai Medical Center (MN), Los Angeles, CA; Santa Barbara Actuaries (AM, EZ), Santa Barbara, CA; AIDS Healthcare Foundation (SCH), Los Angeles, CA; Echosens North America (MT), Waltham, MA; University of California, Santa Barbara (ID), Santa Barbara, CA.

Source of Funding: Echosens.

Author Disclosures: Dr Noureddin reports consultancies or paid advisory boards for Abbott, Allergan, Blade, Echosens North America, Fractyl, Gilead, Intercept, Novartis, OWL, Pfizer, Roche Diagnostics, and Siemens; grants received from Allergan, Bristol Myers Squibb, Conatus, Enanta, Galectin, Galmed, Genfit, Gilead, Novartis, Shire, and Zydus; and stock ownership in Anaetos and Viking. Mr Mackenzie, Ms Zhao, and Dr Howell report that funding for this paper was provided by Echosens. Mr Tunkelrott is employed by Echosens. Dr Duncan reports receiving payment for involvement in the preparation of this manuscript through Santa Barbara Actuaries Inc.

Authorship Information: Concept and design (MN, AM, EZ, SCH, MT); acquisition of data (MN, AM, EZ, ID); analysis and interpretation of data (MN, AM, EZ, SCH, ID); drafting of the manuscript (MN, AM, EZ, SCH, MT); critical revision of the manuscript for important intellectual content (MN, AM); statistical analysis (AM, EZ, ID); obtaining funding (MT); and administrative, technical, or logistic support (AM, EZ).

Address Correspondence to: Andrew Mackenzie, BS, Santa Barbara Actuaries, 3221 Calle Mariposa, Santa Barbara, CA 93105. Email: amackenzie@sbactuaries.com.

REFERENCES

1. Kochanek KD, Murphy SL, Xu J, Arias E. Deaths: final data for 2017. Natl Vital Stat Rep.2019;68(9):1-77.

2. Chalasani N, Younossi Z, Lavine JE, et al. The diagnosis and management of nonalcoholic fatty liver disease: practice guidance from the American Association for the Study of Liver Diseases. Hepatology. 2018;67(1):328-357. doi:10.1002/hep.29367

3. Noureddin M, Sanyal AJ. Pathogenesis of NASH: the impact of multiple pathways. Curr Hepatol Rep. 2018;17(4):350-360. doi:10.1007/s11901-018-0425-7

4. Estes C, Anstee QM, Arias-Loste MT, et al. Modeling NAFLD disease burden in China, France, Germany, Italy, Japan, Spain, United Kingdom, and United States for the period 2016-2030. J Hepatol. 2018;69(4):896-904. doi:10.1016/j.jhep.2018.05.036

5. Younossi ZM, Tampi R, Priyadarshini M, Nader F, Younossi IM, Racila A. Burden of illness and economic model for patients with nonalcoholic steatohepatitis in the United States. Hepatology. 2019;69(2):564-572. doi:10.1002/hep.30254

6. Wong T, Dang K, Ladhani S, Singal AK, Wong RJ. Prevalence of alcoholic fatty liver disease among adults in the United States, 2001-2016. JAMA. 2019;321(17):1723-1725. doi:10.1001/jama.2019.2276

7. Cholankeril C, Ahmed A. Alcoholic liver disease replaces hepatitis C virus infection as the leading indication for liver transplantation in the United States. Clin Gastroenterol Hepatol. 2018;16(8):1356-1358. doi:10.1016/j.cgh.2017.11.045

8. Tapper EB, Parikh ND. Mortality due to cirrhosis and liver cancer in the United States, 1999-2016: observational study. BMJ. 2018;362:k2817. doi:10.1136/bmj.k2817

9. American Diabetes Association. 4. comprehensive medical evaluation and assessment of comorbidities: Standards of Medical Care in Diabetes—2019. Diabetes Care. 2019;42(suppl 1):S34-S45. doi:10.2337/dc19-S004

10. ICD-10. CMS. Updated March 4, 2021. Accessed August 25, 2020. https://www.cms.gov/Medicare/Coding/ICD10/index

11. Current US inflation rates: 2000-2021. US Inflation Calculator. Accessed August 25, 2020. https://www.usinflationcalculator.com/inflation/current-inflation-rates/

12. Medical cost trend: behind the numbers 2020. PricewaterhouseCoopers. June 2019. Accessed August 25, 2020. https://www.pwc.com/us/en/industries/health-industries/assets/pwc-hri-behind-the-numbers-2020.pdf

13. Weber E. Fifty-state survey: network adequacy quantitative standards. Maryland Insurance Administration. October 18, 2016. Accessed August 25, 2020. https://insurance.maryland.gov/Consumer/Documents/agencyhearings/UMD-LawSchool-DrugPolicyClinic-NetAdqSurvey10182016.pdf

14. Finarelli HJ Jr. The Hospital Executive’s Guide to Physician Staffing. HealthLeaders Media; 2009.

15. Younossi ZM, Golabi P, de Avila L, et al. The global epidemiology of NAFLD and NASH in patients with type 2 diabetes: a systematic review and meta-analysis. J Hepatol. 2019;71(4):793-801. doi:10.1016/j.jhep.2019.06.021

16. Han MAT, Altayar O, Hamdeh S, et al. Rates of and factors associated with placebo response in trials of pharmacotherapies for nonalcoholic steatohepatitis: systematic review and meta-analysis. Clin Gastroenterol Hepatol. 2019;17(4):616-629.e26. doi:10.1016/j.cgh.2018.06.011

17. Vilar-Gomez E, Martinez-Perez Y, Calzadilla-Bertot L, et al. Weight loss through lifestyle modification significantly reduces features of nonalcoholic steatohepatitis. Gastroenterology. 2015;149(2):367-378.e5, quiz e14-15. doi:10.1053/j.gastro.2015.04.005

18. Hannah WN Jr, Harrison SA. Effect of weight loss, diet, exercise, and bariatric surgery on nonalcoholic fatty liver disease. Clin Liver Dis. 2016;20(2):339-350. doi:10.1016/j.cld.2015.10.008

19. Vijay P, Gish RG, Alkhouri N, Noureddin M. Screening for nonalcoholic fatty liver disease in the primary care clinic. Gastroenterol Hepatol (N Y). 2019;15(7):357-365.

20. European Association for the Study of the Liver. EASL clinical practice guidelines: management of hepatocellular carcinoma. J Hepatol. 2018;69(1):182-236. doi:10.1016/j.jhep.2018.03.019

21. Noureddin M, Anstee QM, Loomba R. Review article: emerging anti-fibrotic therapies in the treatment of non-alcoholic steatohepatitis. Aliment Pharmacol Ther. 2016;43(11):1109-1123. doi:10.1111/apt.13620

22. Noureddin M, Zhang A, Loomba R. Promising therapies for treatment of nonalcoholic steatohepatitis. Expert Opin Emerg Drugs. 2016;21(3):343-357. doi:10.1080/14728214.2016.1220533