Abstract
Background
Non-alcoholic steatohepatitis (NASH) and type 2 diabetes (T2D) are both linked to substantial healthcare costs and are often co-occurring. We aim to quantify the incremental cost of NASH and T2D using real-world data.
Methods
Adults (≥18 years old) with ≥2 diagnosis codes for NASH and/or ≥2 diagnosis codes for T2D between 1/1/2016 and 12/31/2021 and ≥24 months of continuous claims enrollment (study period) were identified in electronic health records or claims in the Veradigm Integrated Dataset. Patients were stratified into 3 cohorts: NASH-only, T2D-only, and NASH + T2D. We calculated annualized costs for the 24-month study period and fit a generalized linear model (excluding the most expensive 1%) that controlled for disease cohort, age, sex, and modified Charlson comorbidity index to estimate the per year all-cause healthcare costs and incremental cost of adding T2D to a NASH diagnosis (or vice versa).
Results
We identified 23,111 patients diagnosed with NASH-only, 3,548,786 patients with T2D-only, and 30,339 patients with NASH + T2D. The model-predicted mean costs per year were $7,668 for patients with NASH-only, $11,226 for patients with T2D-only, and $16,812 for patients with NASH + T2D. The incremental increase in costs per year of adding T2D to NASH was 63% (+$4,846), and the incremental increase in costs per year of adding NASH to T2D was 42% (+$4,692).
Conclusions
Both NASH and T2D contribute to the high healthcare costs among patients with a dual diagnosis. Results from our analysis indicate that NASH comprises a high portion of total healthcare costs among patients with NASH and T2D.
Introduction
Non-alcoholic steatohepatitis (NASH) is the severe form of non-alcoholic fatty liver disease (NALFD) marked by progressive liver fibrosis and associated with hepatic complications, including cirrhosis and hepatocellular carcinomaCitation1. NASH is strongly associated with type 2 diabetes (T2D), and there is evidence suggesting NASH can be both a precursor to and complication of T2DCitation2. With the growing prevalence of T2D and associated metabolic conditionsCitation3, the prevalence of NASH is projected to increase by 63% between 2015 and 2030, from 16.5 million cases to 27.0 million casesCitation4.
While both NASH and T2D are associated with high annual healthcare costsCitation5,Citation6, the incremental cost of each condition is poorly understood. One study, utilizing a Markov model, estimated the incremental 20-year burden among patients with NASH and T2D was $667.9 billion, of which they attributed 75.5% to the management of T2D and 24.5% to the management of NASHCitation7. However, granular data quantifying the incremental cost of NASH and T2D using real-world data is lacking. The aim of this study was to quantify the incremental cost of NASH and T2D using real-world data.
Methods
Study design and data sources
We conducted a retrospective, observational cohort study using electronic health records (EHR) from the Veradigm Network EHR linked with closed insurance claims data from the Komodo Health Healthcare Map spanning January 1, 2016, to August 31, 2022Citation8. This linked dataset is certified as statistically de-identified under the HIPAA Privacy Rule and is therefore exempt from IRB approval. The dataset represents a large primary care population in the US with paid pharmacy and medical claims and has been used previously for cost analysis in NASHCitation9. Our study followed best practices for the study design and reporting of retrospective database studies as outlined by the Society for Pharmacoeconomics and Outcomes ResearchCitation10.
Study cohorts
We constructed three cohorts: NASH-only, NASH + T2D, and T2D-only. We first identified patients with ≥2 diagnosis codes for NASH between 1/1/2016–12/31/2021 and excluded patients with any diagnosis of viral hepatitis, alcohol-use disorder, alcohol-related liver disease, type 1 diabetes, or gestational diabetes. The first diagnosis of NASH was the index date provided: 1) the patient was ≥18 years old and 2) had ≥24 months of continuous enrolment in claims data after the index date. If the patient had multiple qualifying 24-month periods, then the most recent period was used. Patients meeting these criteria without a T2D diagnosis were assigned to the NASH-only cohort, whereas those with a T2D diagnosis at any time in their record were assigned to the NASH + T2D cohort. For the T2D-only cohort, we applied the same rules. However, in the final step, we excluded patients with a diagnosis of NASH at any time in their record to create the T2D-only cohort.
NASH was identified with the International Classification of Disease – 10th edition – Clinical Modification (ICD-10-CM) code K75.81. T2D was identified with the ICD-10-CM codes E11* and O24.11*, where * represents a wild card for all child codes. International Classification of Disease – 9th edition – Clinical Modification (ICD-9-CM) codes were used to identify patients with a history of T2D (250*). There is no ICD-9-CM code specific to NASH.
One key assumption of our analysis is that an ICD diagnosis code for NASH corresponds to a NASH diagnosis, which is only definitively diagnosed via a biopsy. However, a biopsy is rarely used for diagnosis in clinical practice, and an ICD-based diagnosis is considered acceptable for observational researchCitation11. Repeat diagnosis codes (≥2) were required for a confirmatory diagnosis, while a single diagnosis code was sufficient for excluding patients from the T2D-only cohort.
Baseline Characteristics
We recorded age, sex, race, ethnicity, and geographic region as of the index date. Using diagnosis codes from the 24-month study period, we calculated the Quan adaptation of the Charlson Comorbidity Index (CCI) Citation12, a modified version of the CCI (mCCI) that excluded points for diabetes or liver-related conditions, and the Diabetes Comorbidity Severity IndexCitation13. We captured the prevalence of several NASH-related conditions (autoimmune hepatitis, ascites, cirrhosis, gastroesophageal varices, hepatic encephalopathy, and hepatocellular carcinoma) during the study period. We also captured the results of hemoglobin A1c (HbA1c) testing as recorded in the EHR within 3 months of the index date.
Costs analysis
Using medical and drug claims data, we captured actual all-cause healthcare costs for products and services provided during the 24-month study period. All costs were annualized and reported per person per year.
To estimate the incremental cost of adding T2D to a NASH diagnosis (or vice versa), we first excluded the top 1% of spenders to reduce the data skew and then reduced computing complexity by selecting a random sample of 100,000 individuals from the T2D diabetes cohort. Then, we fit a generalized linear model with a gamma distribution and a log link function, controlling for disease cohort, age, sex, and mCCI to the cost data. Incremental costs were estimated by changing the disease indicator variable for patients in the NASH-only or T2D-only cohorts to NASH + T2D and observing the change in cost estimates. All analyses were conducted using SAS V9.4.
Results
Our final study cohorts included 23,111 individuals in the NASH-only cohort, 3,548,786 in the T2D-only cohort, and 30,339 in the NASH + T2D cohort (). Mean (standard deviation [SD]) age varied between cohorts and was 52.1 (14.2) years in the NASH-only cohort, 62.0 (14.0) years in the T2D-only cohort, and 59.0 (12.2) years in the NASH + T2D cohort (). The percentage of the cohort that was female was 57.4% in the NASH-only cohort, 56.4% in the T2D-only cohort, and 65.5% in the NASH + T2D cohort.
Figure 1. Patient selection. aAny diagnosis date after 1/1/2016 is considered a possible index date and evaluated on the criteria below. If multiple index dates meet al.l criteria, the earliest date will be used. bIf the patient had multiple qualifying 24-month continuous enrollment periods, the most recent period was used.
Table 1. Patient characteristics.
The mean (SD) mCCI was 0.7 (1.3) in the NASH-only cohort, 1.3 (1.9) in the T2D-only cohort, and 1.4 (1.8) in the NASH + T2D cohort (). The mean (SD) DCSI was 0.8 (1.3) in the NASH-only cohort, 2.0 (2.2) in the T2D-only cohort, and 2.1 (2.1) in the NASH + T2D cohort. Cirrhosis was documented in 7.2% of the NASH-only cohort, 0.6% of the T2D-only cohort, and 21.4% of the NASH + T2D cohort. Among the patients with an HbA1c result in the study period, mean (SD) HbA1c was 5.6 (0.4) in the NASH-only cohort, 7.0 (1.5) in the T2D-only cohort, and 7.0 (1.4) in the NASH + T2D cohort.
Among all qualifying individuals, mean (SD) all-cause healthcare costs were $10,919 ($57,785) for the NASH-only cohort, $16,728 ($154,614) for the T2D-only cohort, and $22,213 ($180,698) for the NASH + T2D cohort (). After excluding patients with costs in the top 1%, mean (SD) all-cause healthcare costs were $7,644 ($14,255) for the NASH-only cohort, $11,515 ($21,946) for the T2D-only cohort, and $16,120 ($26,144) for the NASH + T2D cohort. The model-predicted mean costs for each cohort were closely aligned with the actual costs: $7,668 for the NASH-only cohort, $11,226 for the T2D-only cohort, and $16,812 for the NASH + T2D cohort. Using this model, the incremental cost increase of adding a T2D diagnosis to the NASH-only population was $4,846 for a total mean cost of $12,513 (63.2% increase), and the incremental cost increase of adding a NASH diagnosis to the T2D-only population was $4,692 for a total mean cost of $15,917 (41.8% increase).
Table 2. All-Cause Healthcare costs for patients with nonalcoholic steatohepatitis (NASH) and/or type 2 diabetes (T2D).
Discussion
In this study of a large national claims database, we found that both NASH and T2D contribute substantially to the costs of a dual-diagnosed population. In a population with T2D, the addition of a NASH diagnosis was estimated to increase costs by 41.8%; by comparison, in a population with NASH, the addition of a T2D diagnosis was estimated to increase costs by 63.2%. Although there is limited analysis on the cost of comorbidities in NASH, our findings are consistent with a recent literature review, which concluded that costs were higher among NASH patients with comorbid diagnosesCitation14. Our findings contrast with a recent economic modeling study, which suggested that in a T2D population with comorbid NASH, 75.5% of healthcare costs could be attributed to diabetes careCitation7.
Previous estimates of the annual costs of uncomplicated NASH in the United States have ranged from $16,744 to $22,953Citation6,Citation9,Citation15. However, in these prior studies, 23.5% − 56.7% of NASH patients had a comorbid diagnosis of T2D. By comparison, in this study, the estimated unadjusted annual costs were $10,919 for patients with NASH-only and $22,213 for patients with NASH + T2D. The lower costs observed in patients with NASH-only are likely due both to not incurring costs from the management of diabetes and patients with comorbid diabetes having more severe NASH.
This real-world data study extends our knowledge of the economic burden of NASH, particularly among patients with comorbid diabetes. This study is significant because it attempts to disentangle the costs of NASH and T2D and assess the incremental costs of each. Future studies could explore the specific utilization categories contributing to costs among each population to determine if drug costs are driving the additional burden from a diabetes diagnosis.
Limitations
This retrospective study used routinely collected claims and EMR data and is subject to the typical limitations of using data not gathered for specific research purposes. This includes data entry errors, missing data, and coding specificity limitations, which could result in miscategorization of individuals. In particular, NASH can only be definitively diagnosed by liver biopsy; however, the rates of liver biopsy are low, and the results of testing are not available in this dataset [9]. Our approach of using the K75.81 ICD-10-CM code for identifying NASH patients is consistent with the recommendations for retrospective database studies but should be interpreted cautiously, since diagnosis is not confirmed with biopsyCitation16.
Costs may be underestimated as not all products and services are covered by insurance. In addition, this study included only insured individuals with 2 years of stable insurance enrollment and may not be representative of the US population. Furthermore, while our model incorporated age as a covariate, the base populations were not age-matched, and this may be contributing to the differences in costs.
Conclusions
In patients with a dual diagnosis of NASH and T2D, both conditions contribute to high healthcare costs. Our findings do not support diabetes being the primary driver of costs among NASH patients.
Transparency
Declaration of financial/other relationships
AB, DL, and MB are employees of Veradigm, which received fees from Madrigal Pharmaceuticals related to this work. JF, SD, and KM are employees of Madrigal Pharmaceuticals. ET is an employee at the University of Michigan, which received unrestricted financial support to the institution related to this work.
Peer reviewers on this manuscript have received an honorarium from CMRO for their review work but have no other relevant financial relationships to disclose.
Author contributions
All authors contributed to the conceptualization of the study. AB and DL were responsible for data handling and formal analysis of the data. All authors contributed to the development of the methodology implemented in the study, data visualization, interpretation of the results, and writing and revising of the manuscript. All authors approved the final submitted version of the manuscript.
Acknowledgements
Medical writing support was provided by Jessamine Winer-Jones, PhD an employee of Veradigm. This support was funded by Madrigal Pharmaceuticals.
Data availability statement
The data that support the findings of this study were used under license from Veradigm and Komodo Health. Due to data use agreements and its proprietary nature, restrictions apply regarding the availability of the data. Further information is available from the corresponding author.
Additional information
Funding
References
- Sheka AC, Adeyi O, Thompson J, et al. Nonalcoholic steatohepatitis: a review. JAMA. 2020;323(12):1175–1183. doi:10.1001/jama.2020.2298.
- Gastaldelli A, Cusi K. From NASH to diabetes and from diabetes to NASH: mechanisms and treatment options. JHEP Rep. 2019;1(4):312–328. doi:10.1016/j.jhepr.2019.07.002.
- Khan MAB, Hashim MJ, King JK, et al. Epidemiology of type 2 diabetes – global burden of disease and forecasted trends. J Epidemiol Glob Health. 2020;10(1):107–111. doi:10.2991/jegh.k.191028.001.
- Estes C, Razavi H, Loomba R, et al. Modeling the epidemic of nonalcoholic fatty liver disease demonstrates an exponential increase in burden of disease. Hepatology. 2018;67(1):123–133. doi:10.1002/hep.29466.
- Park J, Bigman E, Zhang P. Productivity loss and medical costs associated with type 2 diabetes among employees aged 18-64 years with large Employer-Sponsored insurance. Diabetes Care. 2022;45(11):2553–2560. doi:10.2337/dc22-0445.
- Wong RJ, Kachru N, Martinez DJ, et al. 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.
- Younossi ZM, Tampi RP, Racila A, et al. Economic and clinical burden of nonalcoholic steatohepatitis in patients with type 2 diabetes in the U.S. Diabetes Care. 2020;43(2):283–289. doi:10.2337/dc19-1113.
- Boikos C, Imran M, De Lusignan S, et al. Integrating electronic medical records and claims data for influenza vaccine research. Vaccines. 2022;10(5):727. doi:10.3390/vaccines10050727.
- 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):348–356.
- Motheral B, Brooks J, Clark MA, et al. A checklist for retrospective database studies—report of the ISPOR task force on retrospective databases. Value Health. 2003;6(2):90–97. doi:10.1046/j.1524-4733.2003.00242.x.
- Tapper E, Fishman J, Dodge S, et al. A higher fibrosis-4 (FIB-4) score is associated with higher healthcare costs and hospitalizations in patients with non-alcoholic steatohepatitis (NASH). J Hepatol. 2022;77:S167. https://www.journal-of-hepatology.eu/article/S0168-8278(22)00714-0/fulltext#relatedArticles. doi:10.1016/S0168-8278(22)00714-0.
- Quan H, Li B, Couris CM, et al. Updating and validating the charlson comorbidity index and score for risk adjustment in hospital discharge abstracts using data from 6 countries. Am J Epidemiol. 2011;173(6):676–682. doi:10.1093/aje/kwq433.
- Glasheen WP, Renda A, Dong Y. Diabetes complications severity index (DCSI)—update and ICD-10 translation. J Diabetes Complications. 2017;31(6):1007–1013. doi:10.1016/j.jdiacomp.2017.02.018.
- Witkowski M, Moreno SI, Fernandes J, et al. The economic burden of non-alcoholic steatohepatitis: a systematic review. Pharmacoeconomics. 2022;40(8):751–776. doi:10.1007/s40273-022-01140-y.
- Gordon SC, Fraysse J, Li S, et al. Disease severity is associated with higher healthcare utilization in nonalcoholic steatohepatitis medicare patients. Am J Gastroenterol. 2020;115(4):562–574. doi:10.14309/ajg.0000000000000484.
- 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):474–482. doi:10.1002/hep.31726.