Abstract
BACKGROUND:
Metabolic dysfunction–associated steatohepatitis (MASH), formerly nonalcoholic steatohepatitis, is characterized by fat accumulation and inflammation of the liver and may result in progression to cirrhosis and liver-related events.
OBJECTIVE:
To characterize the impact of cirrhosis and progression to liver-related events on costs and health care resource use (HCRU) among MASH patients in the United States.
METHODS:
The study cohort included patients with diagnosed nonalcoholic steatohepatitis (International Classification of Diseases, Tenth Revision, Clinical Modification code K75.81) in Optum’s deidentified Clinformatics Data Mart Database (October 2015 to December 2022) and were stratified by baseline cirrhosis status. Among those without cirrhosis at baseline, patients were further stratified by status of progression to cirrhosis during follow-up. Total HCRU and costs per-person per-year (PPPY) were estimated and compared descriptively between the cohorts. In addition, gamma generalized linear models were used to compare costs PPPY between those with vs without cirrhosis at baseline, as well as with vs without progression during follow-up, while adjusting for baseline patient and disease characteristics. Annual costs per person were also longitudinally modeled using gamma generalized linear mixed models to understand longitudinal changes in costs PPPY while accounting for time correlations within individual patients. Lastly, a series of sensitivity analyses were conducted to assess the impact of study design features and clinical variations of total costs PPPY.
RESULTS:
A total of 28,576 adults were included, and 9,157 (32.0%) had baseline cirrhosis; of the 19,419 without baseline cirrhosis, a total of 4,235 (21.8%) progressed over follow-up. Mean (SD) HCRU and costs PPPY were higher among patients with cirrhosis ($110,403 [$226,037]) than without ($28,340 [$61,472]; P < 0.01) and among those with progression ($58,128 [$102,626]) than without ($20,031 [$39,740]; P < 0.01). Costs remained significantly greater when adjusted for covariates, with a risk ratio (95% CI) of 1.99 (1.89-2.09) when comparing with vs without baseline cirrhosis and 2.28 (2.15-2.42) when comparing with vs without progression over follow-up. Costs increased with each subsequent year, to 21% by year 6 among those with cirrhosis at baseline and 49% among those without baseline cirrhosis who progressed.
CONCLUSIONS:
The financial burden of MASH is substantial and significantly greater among those with cirrhosis or disease progression. Although patients without cirrhosis incur lower burden, the increase over time is greater and associated with progression. Therapies that slow progression may help alleviate the financial burden, and strategies are needed to identify patients with MASH at risk of progressing to cirrhosis.
Plain language summary
This is a study of how expensive a liver disease called metabolic dysfunction–associated steatohepatitis (MASH) is in the United States. Using data from 28,576 adults with MASH, patients were grouped according to whether they had cirrhosis at the start of the study and whether they developed cirrhosis during the study. On average, costs were higher among adults with cirrhosis. Those who developed cirrhosis during the study also had higher costs than those who did not.
Implications for managed care pharmacy
This study of adults with MASH found that costs over the study period were higher for those with cirrhosis than without and higher when more severe liver issues developed. Given the magnitude of health care resource use and costs reported, in the context of increasing prevalence of MASH, strategies that slow or stop progression to cirrhosis and end-stage liver disease may help alleviate the financial burden of managing MASH in the United States in the long term.
Nonalcoholic steatohepatitis (NASH), the nomenclature of which was recently changed to metabolic dysfunction– associated steatohepatitis (MASH),1 is characterized by hepatic fat accumulation, inflammation, and hepatocyte injury.2 Prolonged inflammation and liver damage associated with MASH can lead to scar tissue formation, called fibrosis, in the liver.3 MASH is a progressive disease and, in an important minority of patients, significant fibrosis ultimately leads to cirrhosis.4 The risk of liver events due to MASH is influenced by a complex interplay of genetic and environmental factors, age, and metabolic comorbidities, including obesity, diabetes, dyslipidemia, and/or hypertension.5-7
MASH without cirrhosis is typically asymptomatic.8 MASH cirrhosis is frequently asymptomatic in its early stages (called compensated cirrhosis) and, when left untreated, can progress to ascites, gastroesophageal variceal bleeding, and hepatic encephalopathy, which are clinical consequences of portal hypertension (or clinically significant portal hypertension).7 Patients with cirrhosis who develop clinical sequelae of clinically significant portal hypertension are classified as having decompensated cirrhosis (DCC). Patients with MASH are at an increased risk of liver failure resulting from DCC and hepatocellular carcinoma (HCC), both of which may ultimately require liver transplantation (LT).9-11 As a result, individuals with MASH face a higher risk of morbidity and mortality than the general age-matched population.7,12 This elevated comorbidity burden and risk of mortality have a substantial impact on patients’ length and quality of life.7
Corresponding with rising clinical complications and prevalence of MASH, the burden of care attributable to advanced liver disease is also expected to grow.4,13 Earlier claims-based research in the United States estimated that annual health care costs of metabolic dysfunction–associated steatotic liver disease (MASLD) and/or MASH can range from $16,744 to $300,408 per person14,15; yielding approximately $15.7 billion annually in the United States.16 However, few claims-based cost estimates exist specifically for the US population with MASH. In addition, it is unclear how the burden of care differs among those with and without MASH cirrhosis, progresses over time among those without cirrhosis, and changes as the frequency of liver-related events in MASH increase. The aim of this study was to characterize the burden of care of patients with MASH in the United States. Specifically, it was of interest to determine the impact of cirrhosis and progression of liver disease as they relate to costs and health care resource use (HCRU).
Methods
DATA SOURCE, STUDY DESIGN, AND POPULATION
This retrospective cohort study was conducted according to best practices, including the reporting of studies conducted using observational routinely collected health data extension of the Strengthening the Reporting of Observational Studies in Epidemiology guidelines. This study used Optum’s deidentified Clinformatics Data Mart Database (CDM). As CDM is statistically deidentified under the Expert Determination method consistent with Health Insurance Portability and Accountability Act of 199617 and managed according to Optum customer data agreements, institutional review board approval was not required.
A cohort of adults in the United States diagnosed with MASH (≥1 inpatient claim of International Classification of Diseases, Tenth Revision, Clinical Modification [ICD-10-CM] code K75.81 [NASH] as the primary or secondary diagnosis or ≥2 outpatient claims of K75.81 in any position, occurring on different days) was identified from October 1, 2015, to December 31, 2022. Date of cohort entry was defined by the first eligible NASH diagnosis code (day 0) (Supplementary Figure 1 (1.6MB, pdf) , available in online article). Index date was day 30. This allowed for a latency window post-NASH diagnosis, to account for delays in the documentation of cirrhosis testing results, which was confirmed through initial exploratory analysis (Supplementary Figure 2 (1.6MB, pdf) ). The 30-day latency window, in addition to a 6-month period prior to first NASH diagnosis, was defined as the baseline window. ICD-10-CM codes for cirrhosis, DCC, LT, or HCC during the baseline window were used to define baseline cirrhosis status (Supplementary Table 1 (1.6MB, pdf) ). Timing and type of progression was assessed among those without cirrhosis at baseline and was defined by subsequent capture of advanced liver conditions, including cirrhosis, DCC, LT, or HCC at any point during the follow-up period. The follow-up period was defined as the day after index date to the end of follow-up. All covariates were assessed within the baseline window. The exclusion criteria window spanned the entire study period. Patients were excluded if they were younger than 18 years, did not have sufficient enrollment prior or following cohort entry, or had at least 1 diagnosis for conditions causing liver disease other than MASH, exposure to heavy metals during the study period, or insufficient continuous enrollment (Supplementary Figure 3 (1.6MB, pdf) ). All patients were followed for at least 12 months (unless death occurred), until the end of enrollment or the beginning of a gap of more than 30 days in enrollment, death, or end of the study period. Patients were classified into 2 cohorts based on those with and without cirrhosis at baseline. Among those without cirrhosis at baseline, 2 subgroups were created: those with and without progression during follow-up.
OUTCOMES
Annual HCRU and costs were estimated per person over each person’s available follow-up. Observed HCRU was considered within component categories: (1) hospitalizations, (2) emergency department (ED) admissions or observation unit stays, (3) outpatient visits, (4) diagnostic procedures, and (5) medication dispensations (Supplementary Table 2 (1.6MB, pdf) ). Annual all-cause health care costs per person were summarized per component (ED, hospitalization, outpatient, medication) and overall to estimate the total costs per-person per-year (PPPY). All procedures were recorded within the setting in which they occurred; therefore, costs were accounted for within that setting (eg, ED, hospitalization, outpatient) and not separately costed. All costs were standardized to 2022 US dollars (USD).18 All-cause HCRU and costs were selected as the focus because of the high degree of multimorbidity among many individuals with MASH, which would complicate accurate classification of liver-specific HCRU and costs.19 However, costs were modeled adjusting for patients baseline characteristics, including comorbidity profile, which will help elucidate the proportion of costs associated with comorbidities, an approach similar to prior studies in this field.14
COVARIATES
Cohort demographics and characteristics and baseline comorbidities summarized by the Elixhauser Comorbidity Index20 were considered as covariates. Aggregate metabolic-related comorbidity burden was summarized using the Diabetes Complications Severity Index.21 Specific comorbidities relevant to MASH (cardiovascular disease [CVD], obesity, and type 2 diabetes mellitus [T2DM]) were identified and patients were categorized, based on clinician input and prior publication, into metabolic comorbidity groups of interest15 (group 1: CVD and renal impairment; group 2: CVD and T2DM; group 3: renal impairment and T2DM; group 4: CVD, renal impairment, and T2DM; and group 5: hypertension, hyperlipidemia, T2DM, CVD, and obesity).
As clinical data were available, baseline fibrosis-4 index (FIB-4) score was calculated.22 Based on FIB-4 scores, risk of advanced fibrosis was grouped into low (FIB-4 < 1.0), intermediate (FIB-4 ≥1.0 to ≤3.25), and advanced (FIB-4 > 3.25) categories.23-25 Lastly, costs PPPY within the baseline window were summarized to gain insights into the baseline HCRU and health care–seeking behavior of patients.
STATISTICAL ANALYSIS
Cohort and baseline characteristics were summarized descriptively using summary statistics, including frequencies, proportions, and means. Outcomes of average HCRU and costs PPPY were descriptively summarized and compared within each set of cohorts/subgroups (with vs without cirrhosis and with vs without progression). All statistical comparisons of continuous measures were performed using Student’s t-test and categorical measures using chi-square test. To examine the impact of age, resulting data were stratified and compared within age categories (≤44 years, 45-64 years, and ≥65 years).
Gamma generalized linear models (GLMs) with a log-link, to account for right-skew distribution of costs, were used to compare costs PPPY, adjusting for age, sex, race, region, baseline CVD, T2DM, obesity, Diabetes Complications Severity Index, and weighted Elixhauser index, as well as baseline costs. Among those without baseline cirrhosis and progression during follow-up, the total PPPY costs were estimated for those with progression and by timing of progression, using interaction terms between these two variables. All patients with a positive cost estimate PPPY were included in these cost models. All results from the GLMs were presented as risk ratios (RRs) with 95% CIs to assess statistical significance and the magnitude of the relationship between clinical status or disease progression and total annual costs, as well as the effects of baseline covariates.
Lastly, annual costs per person, year-over-year, during the follow-up period were longitudinally modeled using gamma generalized linear mixed models (GLMMs), which accounted for time correlation within individual patients while adjusting for death during follow-up, in addition to the covariates used in the GLMs. The GLMMs were performed to compare differences in cost changes year-over-year between patients with and without cirrhosis, as well as between those with and without disease progression (among those without cirrhosis at baseline). To illustrate the changes in costs each year within cohorts, annual costs were predicted from the GLMMs.26
All results were presented separately for cohorts and subgroups. All analyses were conducted within the Optum Deidentified Data Workspace using Jupyter Notebook version 6.5.5 (Project Jupyter). Additional visualizations were created using R (v4.1.2; 2021).
SENSITIVITY ANALYSIS
A comprehensive series of sensitivity analyses were conducted to assess the impact of study design features and assumptions on total costs PPPY. First, alternative MASH case definitions were explored to provide additional clinical context for how total costs PPPY may vary for different subpopulations. These definitions included (1) a more sensitive definition: adding in individuals with nonalcoholic fatty liver disease ICD-10-CM diagnosis of K76; (2) a more specific definition: restricting to individuals with at least 1 inpatient diagnosis for NASH; and (3) adding in patients with alcohol-related diseases (Supplementary Figure 1 (1.6MB, pdf) ). Second, because of the uncertainty in timing of cirrhosis coding around index diagnosis for NASH, the window to classify baseline cirrhosis was varied (shorter [-90 to +30 days] and longer [-365 to +30 days]) to test for potential misclassification of progression, and a subset of patients with cirrhosis identified within the 30 days following first NASH diagnosis was identified. Third, the impact of continuous follow-up time periods was investigated by restricting to individuals with continuous follow-up of at least 18 months. Fourth, the impact of the COVID-19 pandemic was assessed by stratifying available follow-up into pre–COVID-19 and post–COVID-19 time frames (using March 1, 2020, as a cutoff). Finally, additional stratifications were explored to provide more specific clinical context: (1) by metabolic comorbidity profile groups to better understand the impact of comorbidity burden on costs; (2) by age <65 years at baseline to enhance precision by increasing the sample size within the younger age group and better understand the impact of age on costs; and (3) among a subset with laboratory values to allow characterization of the relationship between baseline FIB-4 levels and costs and to also understand potential misclassification of baseline cirrhosis. Total costs PPPY estimated from each sensitivity analysis were compared against the findings of the main analyses for the cohorts with and without cirrhosis, as well as for those with and without progression.
Results
BASELINE CHARACTERISTICS
Of approximately 20 million individuals within CDM during the study period, 131,328 individuals had at least 1 NASH diagnosis, and 28,576 patients met study eligibility criteria (Supplementary Figure 3 (1.6MB, pdf) ). Of these, 9,157 (32.0%) had cirrhosis at baseline, whereas 19,419 (68.0%) did not. Among those without cirrhosis, 4,235 (21.8%) experienced progression during follow-up and 15,184 (78.2%) did not. Overall, patients with cirrhosis had shorter follow-up and had significantly higher baseline comorbidity burden and costs than those without cirrhosis. Correspondingly, patients without cirrhosis who progressed also had higher baseline comorbidity and costs (Table 1). About 50% of those who progressed did so within the first year and DCC was the most common progression event followed by cirrhosis (Supplementary Table 3 (1.6MB, pdf) ).
TABLE 1.
Baseline Characteristics of the 4 Cohorts (With and Without Cirrhosis for Primary Objectives, With and Without Progression Among Those Without Cirrhosis for Secondary Objectives)
| With cirrhosis at baseline | Without cirrhosis at baseline | With progression during follow-up | Without progression during follow-up | |
|---|---|---|---|---|
| (n = 9,157) | (n = 19,419) | (n = 4,235) | (n = 15,184) | |
| Total follow-up, person-months | 269,616 | 752,478 | 177,280 | 575,198 |
| Total follow-up per person, years | ||||
| Mean (SD) | 2.5 (1.6) | 3.2 (1.5)a | 3.5 (1.6) | 3.2 (1.5)a |
| Age at index, years | – | – | – | – |
| Mean (SD) | 67.1 (10.8) | 59.8 (13.4)a | 63.8 (12.2) | 58.7 (13.4)a |
| Categorical age at index, n (%) | ||||
| ≤44 years | 350 (3.8) | 2,815 (14.5)b | 331 (7.8) | 2,484 (16.4)b |
| 45-64 years | 2,616 (28.6) | 7,813 (40.2) | 1,464 (34.6) | 6,349 (41.8) |
| ≥65 years | 6,191 (67.6) | 8,791 (45.3) | 2,440 (57.6) | 6,351 (41.8) |
| Sex, n (%) | – | – | – | – |
| Female | 5,999 (65.5) | 11,431 (58.9)a | 2,716 (64.1) | 8,715 (57.4)a |
| Male | 3,158 (34.5) | 7,982 (41.1)a | 1,518 (35.8) | 6,464 (42.6)a |
| Unknown | 0 (0.0) | 6 (0.0)a | 1 (0.0) | 5 (0.0)a |
| Race and ethnicity, n (%) | – | – | – | – |
| Asian | 171 (1.9) | 871 (4.5)a | 124 (2.9) | 747 (4.9)a |
| Black | 654 (7.1) | 1,380 (7.1)a | 348 (8.2) | 1,032 (6.8)a |
| Hispanic | 1,376 (15.0) | 3,648 (18.8)a | 713 (16.8) | 2,935 (19.3)a |
| White | 6,447 (70.4) | 12,647 (65.1)a | 2,873 (67.8) | 9,774 (64.4)a |
| Unknown | 509 (5.6) | 873 (4.5)a | 177 (4.2) | 696 (4.6)a |
| Insurance, n (%)c | ||||
| MAPD dual (Medicaid/Medicare) | 393 (11.1) | 418 (8.6)a | 133 (9.3) | 285 (8.3)a |
| MAPD LIS | 455 (12.9) | 523 (10.8)a | 178 (12.5) | 345 (10.0)a |
| Other | 2,379 (67.4) | 3,445 (70.8)a | 967 (67.9) | 2,478 (72.1)a |
| Unknown | 304 (8.6) | 477 (9.8)a | 147 (10.3) | 330 (9.6)a |
| Elixhauser Comorbidity Index (range = -19 to 89) | ||||
| Mean (SD) | 19.5 (33.9) | 0.9 (5.3)a | 1.8 (8.2) | 0.6 (4.2)a |
| Diabetes complications severity index (range = 0 to 2) | ||||
| Mean (SD) | 0.9 (1.7) | 0.1 (0.6)a | 0.2 (0.8) | 0.1 (0.5)a |
| Specific comorbidities of interest, n (%) | ||||
| Cardiovascular disease | 7,790 (85.1) | 13,108 (67.5)a | 3,187 (75.3) | 9,921 (65.3)a |
| Obesity | 3,725 (40.7) | 5,753 (29.6)a | 1,328 (31.4) | 4,425 (29.1)a |
| T2DM | 5,209 (56.9) | 5,899 (30.4)a | 1,534 (36.2) | 4,365 (28.7)a |
| Metabolic syndrome grouping, n (%) | ||||
| Group 1: CVD and renal impairment | 690 (7.5) | 797 (4.1)a | 256 (6.0) | 541 (3.6)a |
| Group 2: CVD and T2DM | 1,836 (20.1) | 3,010 (15.5)a | 714 (16.9) | 2,296 (15.1)a |
| Group 3: Renal impairment and T2DM | 87 (1.0) | 106 (0.5)a | 30 (0.7) | 76 (0.5)a |
| Group 4: CVD, renal impairment, and T2DM | 2,230 (24.4) | 1,069 (5.5)a | 384 (9.1) | 685 (4.5)a |
| Group 5: Hypertension, hyperlipidemia, T2DM, CVD, and obesity | 774 (8.5) | 1,051 (5.4)a | 285 (6.7) | 766 (5.0)a |
| FIB-4 measurement available, n (%)c | 3,357 (36.7) | 7,048 (36.3) | 1,546 (36.5) | 5,502 (36.2) |
| Low risk (<1.0) | 291 (8.7) | 2,210 (31.4)b | 328 (21.2) | 1,882 (34.2)b |
| Intermediate risk (≥1.0 to ≤3.25) | 1,579 (47.0) | 4,391 (62.3) | 996 (64.4) | 3,395 (61.7) |
| High risk (>3.25) | 1,487 (44.3) | 447 (6.3) | 222 (14.4) | 225 (4.1) |
| Baseline total costs of health care resource use (per-person per-year) | ||||
| Mean (SD) | $87,239 (121,782) | $26,581 (52,976)a | $38,638 (69,708) | $23,218 (46,716)a |
| Median (Q1; Q3) | $46,227 (17,845; 107,526) | $10,255 (3,895; 27,006) | $16,731 (6,516; 43,832) | $8,950 (3,417; 23,034) |
aP < 0.01 when using 2-tailed Student’s t-test to compare the means between the cohorts with vs without cirrhosis cohorts and with vs without progression.
bP < 0.01 when using chi-square test to compare the distribution of baseline FIB-4 between the cohorts with vs without cirrhosis and with vs without progression.
cProportion out of those with nonmissing FIB-4 values at baseline.
CVD = cardiovascular disease; FIB-4 = fibrosis-4 index; LIS = low-income subsidy; MAPD = Medicare Advantage Part D; Q1 = first quartile; Q3 = third quartile; TD2M = type 2 diabetes mellitus.
UNADJUSTED HCRU AND COSTS
Estimated HCRU PPPY was significantly higher among patients with cirrhosis at baseline than those without, across all components (P < 0.01), and was also higher among those with compared with those without progression during follow-up (P < 0.01) (Supplementary Table 4 (1.6MB, pdf) ). Estimates of HCRU PPPY remained significantly greater when stratified by age group at baseline (P < 0.01 for all except for liver biopsy and FibroScan [Supplementary Table 5 (1.6MB, pdf) and Supplementary Table 6 (1.6MB, pdf) ]). Costs PPPY were also substantially higher among those with cirrhosis at baseline, with mean (SD) costs PPPY of $110,403 ($226,037) among those with cirrhosis vs $28,340 ($61,472) among those without (P < 0.01 [Figure 1 and Supplementary Table 7 (1.6MB, pdf) ]). Hospitalizations were the largest contributor to annual costs for patients with cirrhosis.
FIGURE 1.
Annual Costs per Person Comparing With vs Without Baseline Cirrhosis, With Component Costs (Stacked Bars); Overall and Stratified by Age
When comparing those patients with vs without progression, costs PPPY among patients who progressed were $58,128 ($102,626) compared with $20,031 ($39,740) among those without progression (P < 0.01 [Figure 2 and Supplementary Table 7 (1.6MB, pdf) ]). Costs were largely attributable to outpatient visits (accounting for 41.9% of total costs in the progression subgroup and 51.5% in the subgroup without progression) and also generally increased with patient age (Supplementary Table 8 (1.6MB, pdf) and Supplementary Table 9 (1.6MB, pdf) ).
FIGURE 2.
Annual Costs per Person Comparing With vs Without Baseline Progression, With Component Costs (Stacked Bars); Overall and Stratified by Age
ADJUSTED TOTAL ANNUAL COSTS
When adjusted for baseline characteristics, including age and comorbidities, total PPPY costs among patients with cirrhosis were approximately 2 times higher (RR [95% CI] = 1.99 [1.89-2.09]) than among those without cirrhosis (Table 2). Increasing age and the presence of baseline CVD, T2DM, and obesity were associated with higher costs. Baseline CVD was associated with a 29% increase (1.29 [1.23-1.37]) in total costs PPPY; T2DM, a 30% increase (1.30 [1.24-1.37]); and obesity, a 12% increase (1.12 [1.07-1.18]).
TABLE 2.
Generalized Linear Model Estimates of Total Annual Costs per Person for Those With vs Without Baseline Cirrhosis and With vs Without Progression
| Variable | With vs without cirrhosis | With vs without progression | ||
|---|---|---|---|---|
| RR (95% CI) | P | RR (95% CI) | P | |
| Intercept | 7,059.58 (6,258.61-7,963.06) | <0.01 | 6,965.73 (6,115.99-7,933.52) | <0.01 |
| Cirrhosis (vs without cirrhosis) | 1.99 (1.88-2.09) | <0.01 | NA | NA |
| Progression (vs without progression) | NA | NA | 2.28 (2.15-2.42) | <0.01 |
| Age | 1.01 (1.01-1.02) | <0.01 | 1.01 (1.01-1.01) | <0.01 |
| Sex (male vs female) | 0.93 (0.89-0.97) | <0.01 | 0.9 (0.85-0.94) | <0.01 |
| Race and ethnicity (vs White) | ||||
| Asian | 0.75 (0.67-0.84) | <0.01 | 0.75 (0.67-0.84) | <0.01 |
| Black | 1 (0.92-1.09) | 0.98 | 1.05 (0.96-1.16) | 0.28 |
| Hispanic | 0.91 (0.86-0.96) | <0.01 | 0.94 (0.88-1.00) | 0.07 |
| Region (vs Midwest) | ||||
| Northeast | 0.87 (0.81-0.95) | <0.01 | 0.95 (0.87-1.04) | 0.27 |
| South | 0.95 (0.90-1.00) | 0.06 | 0.97 (0.91-1.04) | 0.38 |
| West | 0.88 (0.82-0.94) | <0.01 | 0.85 (0.78-0.91) | <0.01 |
| Baseline CVD | 1.29 (1.23-1.36) | <0.01 | 1.19 (1.12-1.26) | <0.01 |
| Baseline T2DM | 1.31 (1.24-1.37) | <0.01 | 1.24 (1.17-1.31) | <0.01 |
| Baseline obesity | 1.12 (1.07-1.18) | <0.01 | 1.05 (1.00-1.11) | 0.05 |
| Baseline weighted Elixhauser index | 1.00 (1.00-1.00) | <0.01 | 1.00 (0.99-1.00) | 0.15 |
| Baseline cost | 1.00 (1.00-1.00) | <0.01 | 1.00 (1.00-1.00) | <0.01 |
CVD = cardiovascular disease; NA = not applicable; RR = risk ratio; T2DM = type 2 diabetes mellitus.
Among those without cirrhosis at baseline, total PPPY costs among those with progression over their follow-up were more than 2 times higher (RR [95% CI] = 2.28 [2.15 = 2.42]) than among those without progression (Table 2). Increasing age, baseline CVD (1.19 [1.13-1.26]), T2DM (1.24 [1.18-1.31]), and obesity (1.06 [1.00-1.11]) were also associated with increased costs. Among those with progression, when adjusted for progression type and timing, total costs PPPY were generally higher among those who progressed sooner and were highest among those who progressed to DCC/HCC (Supplementary Figure 4 (1.6MB, pdf) and Supplementary Table 10 (1.6MB, pdf) ).
CHANGES IN TOTAL ANNUAL COSTS OVER TIME
Although costs were generally higher among those with cirrhosis at baseline (predicted total PPPY costs ranging from $55,742 in year 1 to $67,563 in year 6 compared with $23,766 in year 1 to $31,714 in year 6 among those without cirrhosis [Supplementary Figure 5 (1.6MB, pdf) ]), longitudinally, total costs per person increased by a larger magnitude among those without cirrhosis at baseline (Supplementary Table 11 (1.6MB, pdf) ).
Among those without cirrhosis at baseline, predicted total PPPY costs increased from $42,490 in year 1 to $63,110 in year 6 among those with progression and $17,039 in year 1 to $20,338 in year 6 among those without progression (Supplementary Figure 5 (1.6MB, pdf) ). Relative to year 1, this corresponds to a change in magnitude of 6% in year 2 to 49% in year 6 among those who progressed and up to 19% among those who did not progress, when adjusted for baseline characteristics (Supplementary Table 12 (1.6MB, pdf) ).
SENSITIVITY ANALYSES
Compared with the base case cohort definitions, the more sensitive definition (including patients with MASLD who may not have MASH) resulted in comparable total costs PPPY for those without cirrhosis and lower estimates for those with cirrhosis (Figure 3). Hospitalizations for MASH are infrequent at the time of first diagnosis; therefore, using a more specific definition to restrict to those with a NASH diagnosis in inpatient settings resulted in small sample sizes (n = 79 across cohorts) and increased cost estimates. When including those with alcohol-related diseases or varying the cirrhosis assessment window and continuous follow-up requirement, costs remained comparable to the base case estimates. Total costs PPPY were generally lower for the pre–COVID-19 pandemic period and higher for the period during/after, and the difference was most noticeable among those patients with cirrhosis at baseline. Among the additional clinical subgroups of interest, total costs PPPY generally increased with the complexity of metabolic comorbidity profile, age, and FIB-4 risk level.
FIGURE 3.
Forest Plot of Various Mean Annual Total Cost per Person From Various Sensitivity Analyses
Discussion
The epidemiologic burden of MASH is substantial. As prevalence continues to grow, understanding the economic burden is increasingly important; by 2025, MASH-related liver disease is predicted to be the most common indication for LT in the United States.27 The management of MASH is associated with substantial costs and increases with advancing levels of fibrosis14; the management of advanced liver diseases requiring LT can incur costs in excess of $300,000 PPPY among those with MASH or MASLD.15 Although a recent study demonstrated that costs of caring for patients with MASH with cirrhosis were approximately 3 times as high as for those without cirrhosis,28 how such costs are impacted by age and progression status and how they change over time remain unanswered. Our study, which set out to address some of these questions, provides a better understanding of how HCRU and costs vary by progression status and is an important step for understanding the contemporary economic burden of MASH.
A notable finding is the high prevalence of cirrhosis among patients with MASH (32%), which is much higher than reported in prospective, biopsy-based, cross-sectional studies (<5%).29 This high prevalence likely reflects the claims-based nature of the cohort. It should also be noted that the index date was established after meeting inclusion/exclusion criteria and, despite a washout window of 6 months, may not be sufficient to identify a patient at a certain point in their diagnosis or treatment journey. In addition, as screening for MASH is uncommon (reflected through low prevalence estimates),2 the high proportion of baseline cirrhosis in our study is potentially reflective of patients presenting late in the course of disease, when medical care for any indication is more likely to serendipitously identify MASH. Late presentation of disease is also at least partially likely to account for the high frequency of progression and decompensation. Our study showed that approximately 1 in 5 patients without cirrhosis at baseline progressed over follow-up; approximately 50% occurred within the first year. Together, the high prevalence of cirrhosis and frequent progression to cirrhosis and liver-related events highlight the need to identify patients at earlier stages of disease.
Our study findings emphasize that total costs PPPY with MASH are significantly higher for those with cirrhosis than those without, and among those without cirrhosis at baseline, total costs are significantly higher for those who progressed to cirrhosis compared with those without progression. Total costs for those who progressed were comparable to those of the baseline cirrhosis cohort. These findings are within the range of estimates published from several prior US commercial claims database studies among the MASLD/MASH population.15,28 Our study also explored changes in total costs per person over time and found that the magnitude of increase was higher among those without cirrhosis at baseline (vs those with) and higher among those with progression to cirrhosis (vs those without). These findings again highlight the opportunity for early and appropriate interventions to slow disease progression in MASH and decrease the economic burden.
Our study has extended on prior research by including a larger MASH-specific sample to compare costs among those with vs without cirrhosis by using claims-based data. Additionally, our study featured a variety of sensitivity analyses providing additional clinical context. For example, varying MASH definitions demonstrated higher costs associated with more specific MASH definitions, likely reflecting the inclusion of more severely affected patients due to conditioning on hospitalizations, whereas lower costs were observed for the broader definition that included MASLD, likely including those with relatively lower severity and is consistent with findings from prior studies reporting lower costs in the MASLD/MASH population.15 Restricting and stratifying results using FIB-4 value at baseline showed an increasing trend of total costs PPPY with baseline FIB-4, in similar trend as previously reported.14 In addition, we also considered the potential impact of COVID-19 and found higher costs and bigger differences between groups since the onset of the pandemic, particularly in the cirrhosis cohort. Although this finding may have been confounded by the post–COVID-19 time period also being later in calendar time (patients would have progressed further in severity), other potential clinical pathways should also be considered, such as obesity,30,31 increased alcohol consumption,32 or potentials of disruptions in care flow during the COVID-19 pandemic.33
Lastly, in addition to our sensitivity analyses providing additional clinical context, sensitivity analyses also explored potential limitations, and impact of uncertainties in study design demonstrated robustness of the main study definitions, of the cirrhosis assessment window, and of minimum continuous follow-up time. For example, varying the baseline cirrhosis window did not substantially affect estimates, and when isolating the subset who had a cirrhosis diagnosis within 30 days following the first MASH diagnosis, estimates were closer to those with cirrhosis at first MASH diagnosis (compared with those without cirrhosis at baseline/index but progressed during follow-up). These findings support the decision to include this group in the cohort with cirrhosis at baseline to account for reporting/diagnostic delays.
LIMITATIONS
The study relied on a large, well-validated database used extensively for studies of HCRU and costs.34-36 CDM contains laboratory results for a subset of patients, which provided additional insight on clinically relevant information to validate study definitions, such as cirrhosis status. Laboratory results were also used to calculate FIB-4 scores at baseline, which suggested a degree of misclassification (5%-10%) of baseline cirrhosis status affecting both groups similarly. A low-risk FIB-4 cutoff of 1.0 was used in this analysis to limit the degree of misclassification.25 Another strength of this study stems from the careful consideration of the clinical context in the study design, which led to the incorporation of reporting delay in cirrhosis diagnosis. This allowed for more accurate classification of patients with cirrhosis, as shown through the sensitivity analysis. Definitions of how to identify cirrhosis based on published literature were also validated with clinical input. Although the only way to definitively diagnose MASH and the presence of cirrhosis is via biopsy, in clinical practice biopsies are rarely performed37,38; less than 9% of patients in this study had a liver biopsy record during the baseline period. Therefore, this study relied on ICD-10-CM–based diagnoses, which followed similar algorithms previously used for observational research,12,39,40 and explored additional case definitions for MASH and cirrhosis assessment window through sensitivity analyses. In addition, sensitivity analyses were used to enhance interpretation and provide additional clinical context.
With regard to generalizability, CDM includes only those enrolled in large commercial and Medicare Advantage health plans, and thus findings may have limited relevance to those not covered by either of these plans. Previous studies have found inclusion within CDM to be associated with zip codes that had wealthier, older, more educated, and disproportionately White residents.41 Patients who died within 30 days of a MASH diagnosis would not have been included in the cohort. Given that these patients would likely have had more progressed disease (ie, would have been captured in the cirrhosis cohort) and would have had higher costs, by excluding these patients, cost estimates calculated here are likely conservative and underestimate differences in costs between those with vs without cirrhosis.
With regard to interpretation of the costs results, the cost estimates derived from this study reflect all-cause HCRU rather than MASH-specific HCRU and should be interpreted as costs incurred among the MASH population rather than costs attributable to MASH management. All-cause HCRU and costs were selected as the focus because individuals with MASH often have multiple comorbid conditions and their management can be interconnected, which would complicate accurate ascertainment of liver-specific costs.14,19 Because cost estimates are based on coding for the underlying HCRU, in conditions with multiple comorbid conditions, all-cause costs may provide a better indicator of the extent of the burden because of the potential for misclassification in the coding process.19 As such, attempting to estimate MASH-specific HCRU and costs would likely provide an underestimate and estimates would have greater uncertainty.19 It should also be noted that MASH liver disease progression was measured by a relevant event at any point during the follow-up; and patients with progression would need to remain in the dataset long enough to have a progression event identified for these events to be appropriately classified in the “progression cohort.” Therefore, estimates of HCRU and costs in this study may be underestimated for those with progression, given that any “preprogression” time is also contributing to the HCRU and cost estimates for a patient with progression. The results for the subgroup of patients with progression should be interpreted in this context. Lastly, because of there being relatively few board-certified hepatologists in the United States (n<600),42,43 the frequency and costs of hepatologist-specific visits were limited in CDM and only estimates for gastroenterologist/hepatologists were presented.
Conclusions
The burden of care for MASH in the United States is substantial and significantly greater among MASH patients with cirrhosis at baseline or among those who progress to advanced liver-related events. Future studies adopting similar designs and performed within different datasets, specifically focusing on the older, noncommercially insured patients, would help ensure generalizability of findings. Nonetheless, given the magnitude of HCRU and costs reported in this study, in the context of increasing prevalence of MASH, strategies that slow or stop progression to cirrhosis and end-stage liver disease may help alleviate the financial burden of managing MASH in the United States in the long term. Strategies are needed to identify patients with MASH at risk of progressing to cirrhosis.
ACKNOWLEDGMENTS
Some of the findings in this article were/will be presented in abstract form at AMCP Nexus 2023 and AMCP 2024.
Funding Statement
Drs Fishman and Kim are employees of Madrigal Pharmaceuticals, Inc. Ms Qian, Ms Rochon, Ms Sun, and Ms Szabo are employees of Broadstreet HEOR, which received funds from Madrigal Pharmaceuticals, Inc., for this work.
REFERENCES
- 1.Rinella ME, Lazarus JV, Ratziu V, et al. ; NAFLD Nomenclature consensus group . A multisociety Delphi consensus statement on new fatty liver disease nomenclature. Hepatology. 2023;78(6):1966-86. doi:10.1097/HEP.0000000000000520 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Sheka AC, Adeyi O, Thompson J, Hameed B, Crawford PA, Ikramuddin S. Nonalcoholic Steatohepatitis: A Review. JAMA. 2020;323(12):1175-83. doi:10.1001/jama.2020.2298 [DOI] [PubMed] [Google Scholar]
- 3.Koyama Y, Brenner DA. Liver inflammation and fibrosis. J Clin Invest. 2017;127(1):55-64. doi:10.1172/JCI88881 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 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 [DOI] [PubMed] [Google Scholar]
- 5.National Institute of Diabetes and Digestive and Kidney Diseases. Symptoms & causes of NAFLD & NASH. 2021. Accessed August 22, 2023. https://www.niddk.nih.gov/health-information/liver-disease/nafld-nash/symptoms-causes
- 6.Hamid O, Eltelbany A, Mohammed A, Alsabbagh Alchirazi K, Trakroo S, Asaad I. The epidemiology of non-alcoholic steatohepatitis (NASH) in the United States between 2010-2020: A population-based study. Ann Hepatol. 2022;27(5):100727. doi:10.1016/j.aohep.2022.100727 [DOI] [PubMed] [Google Scholar]
- 7.Povsic M, Wong OY, Perry R, Bottomley J. A structured literature review of the epidemiology and disease burden of non-alcoholic steatohepatitis (NASH). Adv Ther. 2019;36(7):1574-94. doi:10.1007/s12325-019-00960-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Rinella ME, Neuschwander-Tetri BA, Siddiqui MS, et al. AASLD Practice Guidance on the clinical assessment and management of nonalcoholic fatty liver disease. Hepatology. 2023;77(5):1797-835. doi:10.1097/HEP.0000000000000323 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Wong RJ, Singal AK. Trends in liver disease etiology among adults awaiting liver transplantation in the United States, 2014-2019. JAMA Netw Open. 2020;3(2):e1920294. doi:10.1001/jamanetworkopen.2019.20294 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Liu YB, Chen MK. Epidemiology of liver cirrhosis and associated complications: Current knowledge and future directions. World J Gastroenterol. 2022;28(41):5910-30. doi:10.3748/wjg.v28.i41.5910 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Sanyal AJ, Anstee QM, Trauner M, et al. Cirrhosis regression is associated with improved clinical outcomes in patients with nonalcoholic steatohepatitis. Hepatology. 2022;75(5):1235-46. doi:10.1002/hep.32204 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Tapper EB, Krieger N, Przybysz R, et al. The burden of nonalcoholic steatohepatitis (NASH) in the United States. BMC Gastroenterol. 2023;23(1):109. doi:10.1186/s12876-023-02726-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Estes C, Razavi H, Loomba R, Younossi Z, Sanyal AJ. Modeling the epidemic of nonalcoholic fatty liver disease demonstrates an exponential increase in burden of disease. Hepatology. 2018;67(1):123-33. doi:10.1002/hep.29466 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Tapper EB, Bonafede M, Fishman J, et al. Healthcare resource utilization and costs of care in the United States for patients with non-alcoholic steatohepatitis. J Med Econ. 2023;26(1):348-56. doi:10.1080/13696998.2023.2184967 [DOI] [PubMed] [Google Scholar]
- 15.Wong RJ, Kachru N, Martinez DJ, Moynihan M, Ozbay AB, Gordon SC. Real-world comorbidity burden, health care utilization, and costs of nonalcoholic steatohepatitis patients with advanced liver diseases. J Clin Gastroenterol. 2021;55(10):891-902. doi:10.1097/MCG.0000000000001409 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Younossi ZM, Blissett D, Blissett R, et al. The economic and clinical burden of nonalcoholic fatty liver disease in the United States and Europe. Hepatology. 2016;64(5):1577-86. doi:10.1002/hep.28785 [DOI] [PubMed] [Google Scholar]
- 17.US Department of Health and Human Services. Guidance regarding methods for de-identification of protected health information in accordance with the Health Information Insurance Portability and Accountability Act (HIPAA) Privacy Rule. Accessed May 28, 2024. https://www.hhs.gov/hipaa/for-professionals/privacy/special-topics/de-identification/index.html#:~:text=As%20discussed%20below%2C%20the%20Privacy,or%20in%20combination%20with%20other
- 18.US Bureau of Labor Statistics. Consumer Price Index (CPI) databases. Accessed November 20, 2023. https://www.bls.gov/cpi/data.htm
- 19.Lomas J, Asaria M, Bojke L, Gale CP, Richardson G, Walker S. Which costs matter? Costs included in economic evaluation and their impact on decision uncertainty for stable coronary artery disease. PharmacoEconom Open. 2018;2(4):403-13. doi:10.1007/s41669-018-0068-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care. 1998;36(1): 8-27. doi:10.1097/00005650-199801000-00004 [DOI] [PubMed] [Google Scholar]
- 21.Glasheen WP, Renda A, Dong Y. Diabetes Complications Severity Index (DCSI)-update and ICD-10 translation. J Diabetes Complications. 2017;31(6):1007-13. doi:10.1016/j.jdiacomp.2017.02.018 [DOI] [PubMed] [Google Scholar]
- 22.Sterling RK, Lissen E, Clumeck N, et al. ; APRICOT Clinical Investigators . Development of a simple noninvasive index to predict significant fibrosis in patients with HIV/HCV coinfection. Hepatology. 2006;43(6):1317-25. doi:10.1002/hep.21178 [DOI] [PubMed] [Google Scholar]
- 23.Blanco-Grau A, Gabriel-Medina P, Rodriguez-Algarra F, et al. Assessing liver fibrosis using the FIB4 Index in the community setting. Diagnostics (Basel). 2021;11(12). doi:10.3390/diagnostics11122236 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Berzigotti A, Tsochatzis E, Boursier J, et al. ; European Association for the Study of the Liver . Electronic address: easloffice@easloffice.eu; Clinical Practice Guideline Panel; Chair; EASL Governing Board representative; Panel members. EASL Clinical Practice Guidelines on non-invasive tests for evaluation of liver disease severity and prognosis - 2021 update. J Hepatol. 2021;75(3):659-89. doi:10.1016/j.jhep.2021.05.025 [DOI] [PubMed] [Google Scholar]
- 25.Shah S, Dhami-Shah H, Kamble S, Shukla A. FIB-4 cut-off of 1.3 may be inappropriate in a primary care referral pathway for patients with non-alcoholic fatty liver disease. J Hepatol. 2020;73(1):216-7. doi:10.1016/j.jhep.2019.12.025 [DOI] [PubMed] [Google Scholar]
- 26.Searle SR, Speed FM, Milliken GA. Population marginal means in the linear model: an alternative to least squares means. Am Stat. 1980;34(4):216-21. doi:10.1080/00031305.1980.10483031 [Google Scholar]
- 27.Witkowski M, Moreno SI, Fernandes J, Johansen P, Augusto M, Nair S. The economic burden of non-alcoholic steatohepatitis: A systematic review. PharmacoEconomics. 2022;40(8):751-76. doi:10.1007/s40273-022-01140-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Younossi ZM, Mangla K, Chandramouli AS, Lazarus JV. Total healthcare cost and characteristics associated with higher change in cost in patients with non-alcoholic steatohepatitis (NASH). Poster presented at: The International Liver Congress; 2022; London, UK. [Google Scholar]
- 29.Harrison SA, Gawrieh S, Roberts K, et al. Prospective evaluation of the prevalence of non-alcoholic fatty liver disease and steatohepatitis in a large middle-aged US cohort. J Hepatol. 2021;75(2):284-91. doi:10.1016/j.jhep.2021.02.034 [DOI] [PubMed] [Google Scholar]
- 30.Bramante C, Tignanelli CJ, Dutta N, et al. Non-alcoholic fatty liver disease (NAFLD) and risk of hospitalization for Covid-19. medRxiv. Preprint posted online September 2, 2020. [Google Scholar]
- 31.Dietrich CG, Geier A, Merle U. Non-alcoholic fatty liver disease and COVID-19: Harmless companions or disease intensifier? World J Gastroenterol. 2023;29(2):367-77. doi:10.3748/wjg.v29.i2.367 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Grossman ER, Benjamin-Neelon SE, Sonnenschein S. Alcohol consumption during the COVID-19 pandemic: A cross-sectional survey of US dults. Int J Environ Res Public Health. 2020;17(24):9189. doi:10.3390/ijerph17249189 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Kushner T, Cafardi J. Chronic liver disease and COVID-19: Alcohol use disorder/alcohol-associated liver disease, nonalcoholic fatty liver disease/nonalcoholic steatohepatitis, autoimmune liver disease, and compensated cirrhosis. Clin Liver Dis (Hoboken). 2020;15(5):195-9. doi:10.1002/cld.974 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Berger JS, Ashton V, Laliberté F, et al. Healthcare resource utilization and costs of rivaroxaban versus warfarin among non-valvular atrial fibrillation (NVAF) patients with diabetes in a US population. Adv Ther. 2023;40(3):1224-41. doi:10.1007/s12325-022-02422-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Welliver C, Feinstein L, Ward JB, et al. ; Urologic Diseases in America Project . Evolution of healthcare costs for lower urinary tract symptoms associated with benign prostatic hyperplasia. Int Urol Nephrol. 2022;54(11):2797-803. doi:10.1007/s11255-022-03296-0 [DOI] [PubMed] [Google Scholar]
- 36.Whitney D, Kamdar N, Hirth RA, Hurvitz EA, Peterson MD. Economic burden of paediatric-onset disabilities among young and middle-aged adults in the USA: A cohort study of privately insured beneficiaries. BMJ Open. 2019;9(9):e030490. doi:10.1136/bmjopen-2019-030490 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Anstee QM, Hallsworth K, Lynch N, et al. Real-world management of non-alcoholic steatohepatitis differs from clinical practice guideline recommendations and across regions. JHEP Rep Innov Hepatol. 2021;4(1):100411. doi:10.1016/j.jhepr.2021.100411 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.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):328-57. doi:10.1002/hep.29367 [DOI] [PubMed] [Google Scholar]
- 39.Hagström H, Adams LA, Allen AM, et al. Administrative coding in electronic health care record-based research of NAFLD: An expert panel consensus statement. Hepatology. 2021;74(1):474-82. doi:10.1002/hep.31726 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Romero-Gomez M, Kachru N, Zamorano MA, Darba J, Shreay S. Disease severity predicts higher healthcare costs among hospitalized nonalcoholic fatty liver disease/nonalcoholic steatohepatitis (NAFLD/NASH) patients in Spain. Medicine (Baltimore). 2020;99(50):e23506. doi:10.1097/MD.0000000000023506 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Dahlen A, Charu V. Analysis of Sampling Bias in Large Health Care Claims Databases. JAMA Netw Open. 2023;6(1):e2249804. doi:10.1001/jamanetworkopen.2022.49804 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Palmer WJ. Telemedicine for transplant hepatology. 2020. Accessed May 28, 2024. https://www.wolterskluwer.com/en/expert-insights/telemedicine-for-transplant-hepatology
- 43.Russo MW, Fix OK, Koteish AA, et al. Modeling the hepatology workforce in the United States: A predicted critical shortage. Hepatology. 2020;72(4):1444-54. doi:10.1002/hep.31425 [DOI] [PubMed] [Google Scholar]