Disease burden and economic impact of diagnosed non‐alcoholic steatohepatitis in five European countries in 2018: A cost‐of‐illness analysis

Abstract Background and aims Non‐alcoholic steatohepatitis (NASH) is a chronic disease that can progress to end‐stage liver disease (ESLD). A large proportion of early‐stage NASH patients remain undiagnosed compared to those with advanced fibrosis, who are more likely to receive disease management interventions. This study estimated the disease burden and economic impact of diagnosed NASH in the adult population of France, Germany, Italy, Spain and the United Kingdom in 2018. Methods The socioeconomic burden of diagnosed NASH was estimated using cost‐of‐illness methodology applying a prevalence approach to estimate the number of adults with NASH and the attributable economic and wellbeing costs. Given undiagnosed patients do not incur costs in the study, the probability of diagnosis is central to cost estimation. The analysis was based on a literature review, databases and consultation with clinical experts, economists and patient groups. Results The proportion of adult NASH patients with a diagnosis ranged from 11.9% to 12.7% across countries, which increased to 38.8%‐39.1% for advanced fibrosis (F3‐F4 compensated cirrhosis). Total economic costs were €8548‐19 546M. Of these, health system costs were €619‐1292M. Total wellbeing costs were €41 536‐90 379M. The majority of the undiagnosed population (87.3%‐88.2% of total prevalence) was found to have early‐stage NASH, which, left untreated, may progress to more resource consuming ESLD over time. Conclusions This study found that the majority of economic and wellbeing costs of NASH are experienced in late disease stages. Earlier diagnosis and care of NASH patients could reduce future healthcare costs.


| INTRODUC TI ON
Non-alcoholic fatty liver disease (NAFLD) is a chronic liver disease characterised by excessive fat deposition in the liver in the absence of competing liver disease aetiologies, such as alcoholrelated liver disease or chronic viral hepatitis. 1,2 The epidemic of obesity and type 2 diabetes in Europe and the United States (US) has led to an increasing prevalence of NAFLD, which is now one of the most frequent causes of chronic liver disease and one of the leading causes of cirrhosis and liver transplantation for end-stage liver disease (ESLD). 3 Non-alcoholic steatohepatitis (NASH)-the progressive, inflammatory form of NAFLD-is widely considered to be the hepatic manifestation of metabolic syndrome. 2 NASH is defined by changes observed on liver histology and includes the presence of ≥5% hepatic steatosis and inflammation with hepatocyte injury (eg ballooning), with varying degrees of fibrosis. Given that it is not feasible to conduct liver biopsies in studies of the general population (owing to practical, cost and ethical considerations), there is no direct assessment of the prevalence or incidence of NASH in the epidemiological literature. Patients with advanced fibrosis from NASH are at greater risk of progressing to ESLD and, thus, this population potentially exhibits the greatest disease burden and economic costs. 2 The costs associated with NASH are likely to increase overtime in parallel with increasing disease prevalence. [2][3][4][5][6][7][8] These costs can be classified as economic costs through their effects on healthcare, productivity and carers. Furthermore, administrative inefficiencies are associated with raising taxation revenue (which would otherwise be collected from NASH patients, their carers and employers) in order to fund government expenditures including healthcare and welfare benefits. These inefficiencies may be monetised and are referred to as deadweight losses.
NASH also negatively affects wellbeing and adds to premature morbidity and mortality. 8 To quantify these costs, a comprehensive exploration of economic and wellbeing-associated costs is warranted.
This study estimated the disease burden and economic impact of adults diagnosed with NASH in France, Germany, Italy, Spain and the United Kingdom (UK) in 2018. This type of cost-of-illness analysis may support measures to address an increasingly prevalent, yet mostly asymptomatic disease. 9

| MATERIAL S AND ME THODS
The socioeconomic burden of diagnosed NASH in the five European countries was estimated from a societal perspective using a cost-ofillness methodology applying a diagnosed prevalence approach. 10 This method has been described in detail previously and is briefly summarised in Figure 1. At a high-level, this involved estimating the number of people with diagnosed NASH in a base period (2018) and the economic and wellbeing costs attributable to this condition during the base year.
Only patients diagnosed with NASH incurred costs in the model. Costs attributable to comorbidities associated with NASH were excluded.
A review of the scientific literature was conducted to obtain inputs for modelling, supplemented with information obtained from existing databases and through consultation with 31 expert stakeholders including 18 clinical experts, 9 health economists and 4 patient group representatives via 5 separate in-person workshops and follow-up consultations. Table S1 outlines the details of experts on each panel for each country.
The data available to estimate the disease burden of NASH were limited and imperfect, requiring a considered methodological approach in the selection of inputs for the modelling. As such, the selection of inputs for this study followed a systematic hierarchical approach considering three factors: quality, generalisability and internal consistency. NASH and country-specific inputs were used where available. Otherwise, each data point was considered in terms of its associated quality, generalisability and internal consistency.
The inputs selected were validated with the aforementioned expert panel in each of the five European countries via a series of consultations leveraging principles of expert elicitation, as previously described. 11,12 Experts were tasked with validating estimates of the proportion of patients diagnosed with NASH at any stage following liver biopsy and/ or non-invasive diagnosis in their countries based on clinical practice and their expertise. These estimates were cross checked against literature where data were available. Furthermore, inputs relating to the probability of diagnosis and health system utilisation underwent stepwise validation given the lack of available peer-reviewed literature and published data on these topics. Where available, national estimated figures of hepatocellular carcinomas and liver transplantations formed the basis of estimates. The outcome of this provided consensus among the expert groups regarding the clinical practice (diagnosis and management) of NASH patients within each country.
F I G U R E 1 High-level materials and method flowchart (uploaded as File S1) Given that inputs were validated with experts within each country, we did not require all inputs to be consistently defined across countries. For example, the consensus led to some variation in how NASH was diagnosed in each country. NASH patients in the UK received a selection of diagnostic tests including liver function test, ultrasound and elastography and alpha-fetoprotein (AFP) in the context of hepatocellular carcinoma (HCC) screening (Table S12). In Germany, NASH patients also received a fibrosis score, and in selected cases computed tomography and magnetic resonance imaging as part of the diagnostic process including workup for suspected HCC. The limitations of non-standardised diagnostic approaches across European Union countries are described more fully in the discussion.  13 These derived prevalence rates were applied to population estimates by age and gender from country-specific sources. 14,15 The incidence of NASH by disease stage was used to derive the newly diagnosed and monitored populations, assuming a proportionate diagnosis rate. 16 Liver transplants attributable to NASH were applied to the age and sex distribution of NASH prevalence. 18 People aged 70 and over were assumed to be precluded from receiving a liver transplant, 17 an assumption that was validated with experts in each study country. Liver-related and excess cardiovascular disease mortality was based on Estes et al. 5 Excess cardiovascular disease mortality was estimated to comprise 60% of total attributable mortality according to the published data. 5

| Diagnosis of NASH
The diagnosis rates of NASH were determined based on consultation with clinical experts, highlighting that NASH diagnoses are primarily made incidentally during routine clinical investigations for incident liver function test abnormalities or steatosis on ultrasound. Expert opinion was elicited via a standardised process that was validated through other sources, where available. However, wide variation and uncertainty in diagnosis rates among the five countries persisted, hence the need to explore prevalence scenarios through sensitivity testing. It was suggested that a large proportion of the prevalent NASH population were undiagnosed prior to developing advanced fibrosis or cirrhosis. This is important as disease-specific intervention can only be offered after a NASH diagnosis has been established. In the absence of a NASH diagnosis, no disease-modifying interventions will be offered (refer to File S1). 11,12,17,19 The probability of diagnosis at each disease stage was initially derived from a UK economic evaluation of an alternative diagnostic pathway, which reported the probability of detection from a UK prospective cross-sectional feasibility study. 17,19 The These derived estimates are applied throughout the results section (derived probability of diagnosis scenario). Sensitivity analysis on the probability of diagnosis was conducted via lower and higher scenarios where these scenarios represent the application of the lowest/highest derived probability of diagnosis scenario across all included countries. This means that in the lower scenario, the derived probability of diagnosis scenario for Spain was applied to all countries, while, in the higher scenario, the derived probability of diagnosis scenario for Germany was applied to all countries. File S1 outlines the results of a lower and higher probability of diagnosis scenario (Tables S2-S11) applied to the lower and higher prevalence scenarios for each country. Variation in the revised probabilities of diagnosis may be as a result of a combination of factors including country-specific differences in clinical practice and perceptions of the number of 'silent' NASH patients in each country.

| Health system costs of diagnosed NASH
Health system costs include primary and secondary healthcare, diagnostic tests, pharmaceuticals and medical research. 17, 20 -5 4 Health system costs were estimated for patients who were 'newly diagnosed' and 'monitored' within each disease stage. Newly diagnosed refers to patients who, in 2018, are in their first year of care following diagnosis. Monitored refers to patients who, in 2018, are in their second or subsequent years of care, following a previous diagnosis. Health system costs by disease stage and type were calculated by applying the average, per-person health system cost for that disease stage to the respective diagnosed population. Per-person costs are slightly higher in the lower prevalence scenario as fixed costs are distributed over a smaller population. American Association for the Study of Liver Disease (AASLD). 17,20 These sources were supplemented with expert opinion sought from a panel of clinical experts in each of the five countries studied via a series of consultations leveraging principles of expert elicitation to accurately reflect how NASH patients are diagnosed and monitored in each country. 17 Please refer to File S1 for detailed information regarding the health system utilisations and unit costs derived for this study, including sources.

| Productivity costs of diagnosed NASH
Productivity costs were estimated via a human capital approach and include reduced workforce participation, lost productive time caused by absenteeism and presenteeism, forgone income because of premature mortality and search, hiring and training costs. [55][56][57][58][59][60] This approach involved estimating the number of hours of productivity that are lost owing to NASH. This was then converted into a monetary value by multiplying the number of hours by average weekly earnings adjusted for age, gender and general population employment rates. 55 It is recognised that the attribution of productivity costs to NASH in comparison to the associated comorbidity load of the patient cohort is complex, and literature sources, which controlled for comorbidities were targeted in the search strategy.
The impact of premature death on workforce participation is captured by forgone future income and does not impact absenteeism, presenteeism or reduced workforce participation. Productivity costs incurred through reduced workforce participation were estimated by applying reduced workforce participation attributable to NASH to the respective five European countries' general population employment rates and average weekly earnings by age and gender.
In lieu of country-specific estimates, reduced workforce participation attributable to ESLD comprising DCC to liver transplant was ob-  59 All patients provided written informed consent for use of their aggregated data. 59

| Other economic costs of diagnosed NASH
Other economic costs estimated include formal and informal care costs, deadweight losses and other financial costs such as funeral costs brought forward caused by premature mortality. [61][62][63][64][65][66][67][68][69][70][71][72][73][74][75][76][77][78] The average cost of formal care received by NASH patients with ESLD was based on a retrospective cost-of-illness study of patients with chronic liver disease conducted over 1 year in Italy, adjusted to 2018 euros. 63 The opportunity cost method was used to estimate the cost of informal care. This method measures the value of the alternative use of time spent caring, which is typically valued by productivity losses (or value of leisure time) associated with caring.
It assumes that time spent providing informal care could be alternatively used within the paid workforce or in leisure activities. The proportion of NASH patients with ESLD who received informal care was obtained from Scalone et al. 63 The average time spent on informal care was estimated using country-specific sources. 61,62 Informal care requirements were assumed to apply evenly across age and gender, varying only by disease stage. The age and gender adjusted average weekly earnings of primary carers was obtained from government sources. 61,62 Deadweight losses were estimated from inefficiencies associated with forgone taxation revenue and transfer payments. 64,65 Transfer payments estimated include government expenditure on healthcare and welfare. The number and value of claims each year attributable to NASH was calculated using government sources, adjusted for the proportion of total liver disease in the respective five European countries, which is owing to NASH. [66][67][68][69][70][71][72][73] Funeral costs brought forward caused by premature mortality were sourced from country-specific sources and adjusted to 2018. [74][75][76][77][78]

| Disease burden and wellbeing costs of diagnosed NASH
Wellbeing costs were estimated using the World Health Organization (WHO) burden of disease methodology and converted into euros using an estimate of the value of a statistical life year (VSLY). [79][80][81] This is a non-financial approach, where pain, suffering and premature mortality are measured in terms of disability-adjusted life years (DALYs).
DALYs are composed of premature mortality (years of life lost owing to premature death-YLL) and morbidity (years of healthy life lost owing to disability-YLD) components. DALYs are calculated by assigning disability weights to various health states, where zero represents a year of perfect health and one represents death.
Disability weights for DCC and HCC were sourced from the Institute for Health Metrics and Evaluation (IHME) Global Burden of Disease Study. 79 Several disability weights were available for HCC depending on the state of the disease (diagnosis compared with terminal disease). Disability weights for HCC were weighted according to the proportion of time spent in each state. DALYs are discounted at a rate of 3% consistent with WHO methodology. 80 The burden of disease as measured in DALYs was converted into euros using an estimate of the VSLY. The VSLY is an estimate of the value society places on an anonymous life. A per person VSLY for each of the five European countries was obtained from government published sources, or literature, and inflated where required. [81][82][83]

| Economic costs of diagnosed NASH
The following results are reported using the derived probabilities of diagnosis considering a lower-and higher-prevalence scenarios respectively. Results pertaining to the UK were converted from pounds to euros for comparison. 82 Table 4). The majority of health system costs were incurred in secondary healthcare, followed by diagnostic tests, primary healthcare and pharmaceuticals (   NASH such as HCC, ascites, variceal bleeding and encephalopathy (Tables S12-S16). However, variation was observed in the use of dieticians and exercise physiologists, as well as in the average utilisation of secondary healthcare services per diagnosed NASH patient.
Similarly, the care received by NASH patients in a primary healthcare setting was almost identical with most patients in the five European countries having seen their general practitioner once following diagnosis and once in monitoring, regardless of disease stage, for the purpose of managing their NASH. Some variation in the utilisation of general practitioner services was observed. Further detail regarding these differences is provided in File S1.

| Disease burden and wellbeing costs of diagnosed NASH
In 2018, people with diagnosed NASH in the five European coun-  (Table 5).

| D ISCUSS I ON
In order to estimate the disease burden and economic impact associated with NASH, a central input is the probability of diagnosis at each stage. NASH in non-advanced stages is usually asymptomatic and a large proportion of the prevalent population was not diagnosed in 2018. 84 One major finding of the analysis is a discordant assessment of the experts among the five countries with regards to the number of patients diagnosed with NASH at an advanced stage.
These differences are a major aspect behind the numeric difference observed in the cost-of-illness analysis. This is mostly impacted by The local clinical experts noted that primary health physicians in Spain conduct primary screening for NAFLD (Tables S11-S15).
This practice is significantly different in Italy, where significant heterogeneity in clinical practice exists owing to the lack of an established screening process for NASH. This means that, in any given centre, there will be patients who asked their general practitioner for a hepatologist referral, as well as patients who received a hepatologist referral from their general practitioner or other specialist in response to routine test results. This is further complicated by the fact that general practitioners in Italy do not refer patients based on their fibrosis stage, meaning hepatologists may see many F0 and F1 patients and refer them back to their general practitioner for monitoring without making a formal diagnosis. These country-specific differences underlie, in-part, the differing probabilities of diagnosis in this study.
Despite these differences, one consistent finding among the included countries was the greater economic burden imposed by ESLD. To this end, this study found that the proportion of diagnosed patients and health system costs were greater in patients with ESLD than in earlier disease stages within the five countries. Importantly, the majority of the undiagnosed population, which makes up approximately 88% of the total prevalence, had early-stage NASH. As such, patients with early-stage NASH can be thought of as dormant cases, meaning they incur low costs at present while having the potential to incur significant economic and wellbeing costs in the future after progressing to advanced fibrosis. This is important because the prevalence of NASH is forecasted to increase significantly over time. 6 Effective management of early-stage and advanced fibrosis because of NASH patients and preventive measures requires the establishment of a diagnosis. Increasing the proportion of diagnosis for patients with advanced fibrosis caused by NASH, so that they can be counselled for prevention and referred for treatment and active monitoring, will require enhanced awareness and understanding of NASH among general practitioners, specialists other than hepatologists (eg diabetologists) and the general public, in addition to the TA B L E 5 Disease burden and wellbeing costs results (total) (derived probability of diagnosis scenario) establishment of best-practice clinical management pathways within each country. 85 On the other hand, this is unlikely to be well implemented in clinical practice unless interventions or treatments are available for these patients.
In the literature, three estimates of the economic burden of NAFLD or NASH in the included countries have been published.  86 Balp et al found that despite NASH patients often having many comorbidities, NASH is independently associated with a significant disease burden and economic impact. 86 Furthermore, the authors confirm that data on the disease burden and economic impact of NASH are limited at this stage. 86  and countries with comparable demographics and validated via a series of consultations leveraging principles of expert elicitation. 10,11 This means that in many cases, the inputs underlying this study are uncertain, and changes in these inputs and parameters may have a significant impact upon the total estimate of the costs of NASH in the adult population in the five European countries in 2018. As discussed, NASH is defined by histological changes that can only be assessed by liver biopsy, which is the current 'reference standard'.
Consequently, the prevalence and incidence of NASH has required estimation as a result of a lack of epidemiological literature to use as an input into the modelling process. Furthermore, there are complexities in assigning attribution to NASH directly while excluding the effects of reduced productivity caused by comorbidities such as obesity and diabetes. Productivity inputs used in this study may be an underestimation of the productivity effects attributable to NASH in the five countries. These represent areas for future research.
There is a potential for this study to be updated once more data become available. This work also did not include long-term healthcare costs associated with the growing prevalence of NASH, which has been predicted to increase by 60% by 2030 without intervention. 89 Similarly, the projected costs associated with the benefits of hypothetical treatment were considered out of scope for this work; however, the reader is referred to a recent paper by Younossi 59 All patients provided written informed consent for use of their aggregated data. 59

ACK N OWLED G EM ENTS
The cost-of-illness model, literature review and expert consultations were conducted and coordinated by Deloitte, and supported and paid for by Intercept Pharmaceuticals. We are also grateful for the assistance provided by Professor Claude Le Pen in developing this manuscript.

AUTH O R S ' CO NTR I B UTI O N S
The following authors made substantial contributions to the con-