Influence of nonalcoholic fatty liver disease on inflammatory bowel disease hospitalizations in the United States

Introduction

The spectrum of nonalcoholic fatty liver disease (NAFLD) encompasses a wide range of liver diseases, ranging from milder early hepatic steatosis to late-stage steatohepatitis [1]. Worldwide, NAFLD is believed to be the most common form of chronic liver disease, with a prevalence of 25% [2]. In the United States (US), about 33% of the adult American population have some degree of NAFLD [3]. Common risk factors implicated in developing NAFLD include obesity, metabolic syndrome, and insulin resistance [4]. The pathogenesis of NAFLD stems from insulin resistance and hyperinsulinemia. Hyperinsulinemia also promotes a proinflammatory environment with elevated cytokines, tumor necrosis factor-ɑ, and endocrine disruption with higher leptin and lower adiponectin levels. This complex multifactorial process leads to steatotic deposits in the liver and promotes fibrosis of the hepatocytes and the surrounding extracellular matrix [5]. The diagnosis and grading of NAFLD are based on the absence of significant alcohol consumption, coexisting causes of chronic liver disease, and the extent of steatosis and fibrotic changes within the liver [6]. NAFLD places a significant burden on the US healthcare system. The annual cost of care for a NAFLD patient with private insurance was estimated to be $7804 on average for a new diagnosis. In contrast, the total annual cost for a matched control with similar comorbidities but without NAFLD was $2298 [7].

Inflammatory bowel disease (IBD) results from an immune dysregulation of the gastrointestinal mucosa in response to intestinal microbes in genetically susceptible hosts [8,9]. It consists of two clinically distinct entities, Crohn’s disease (CD) and ulcerative colitis (UC). The current literature reports that NAFLD may be seen in about one third of IBD patients [10]. Alterations in the gut microbiota, immunosuppressive therapy, colonic surgeries and total disease duration are considered possible predictors of NAFLD development in IBD patients [11,12]. However, there is a significant paucity of data on NAFLD and its influence on IBD hospitalizations. Hence, we utilized the National Inpatient Sample (NIS) to highlight and compare hospitalization characteristics and clinical outcomes of IBD hospitalizations with NAFLD and without NAFLD in the US.

Materials and methods

Study population

The NIS database was searched for all adult (≥18 years) hospitalizations with a primary/principal discharge diagnosis of IBD in the US between 2016 and 2019. As the NIS database does not have any identifiers, it is safe to assume that the total IBD hospitalizations may include patients who were hospitalized either once or multiple times with the diagnosis of IBD. The study population was further subdivided based on the presence or absence of NAFLD. Patients with other causes leading to liver damage as mentioned in Supplementary Table 1 were excluded.

Table 1 Comparison of hospital characteristics, comorbidities, outcomes, and hospitalization characteristics in cohorts of patients with IBD, with or without NAFLD

Results

Clinical outcomes

IBD hospitalizations with NAFLD had significantly higher all-cause inpatient mortality than IBD hospitalizations without NAFLD (3.1% vs. 1.4%, P<0.001). They also had higher race-specific mortality for Whites (3.12% vs. 1.49%, P<0.001), Blacks (2.59% vs. 1.13%, P<0.001), and Hispanics (3.44% vs. 1.27%, P<0.001). The cohort with NAFLD also had higher mortality for both males (3.36% vs. 1.55%, P<0.001) and females (2.94% vs. 1.36%, P<0.001). For all age groups, mortality was higher in the IBD hospitalizations with NAFLD than in those without NAFLD (Table 2).

Table 2 Comparison of clinical outcomes in IBD with NAFLD and IBD without NAFLD cohorts

Predictors of inpatient mortality

Increasing age (adjusted odds ratio [aOR] 1.041, 95% confidence interval [CI] 1.038-1.045, P<0.001) and higher CCI (for CCI=2, aOR 3.742, 95%CI 3.186-4.394, P<0.001; and for CCI>3, aOR 6.587, 95%CI 5.613-7.730, P<0.001) were identified to be independent predictors of all-cause inpatient mortality for IBD hospitalizations with NAFLD (Table 3).

Table 3 Multivariable logistic regression of IBD patients with NAFLD that died

Discussion

To our knowledge, this is the first national study to examine the influence of NAFLD on inpatient outcomes in IBD patients. We compared 1,208,170 IBD hospitalizations without NAFLD to 64,090 IBD hospitalizations with NAFLD. We noted a higher mean age, male predominance, and a greater proportion of Whites and Hispanics in the IBD hospitalizations with NAFLD. We noted higher values for all-cause mortality (3.14% vs. 1.44%, P<0.001), length of stay by 0.83 days, healthcare burden by $14,069, and complications such as acute kidney injury, acute respiratory failure, and shock for IBD hospitalizations with NAFLD. Those patients also had higher age-specific, race-specific, and sex-specific mortality. Increasing age and a higher comorbidity burden were independent predictors of mortality in hospitalized IBD patients with NAFLD.

NAFLD is a global pandemic in the modern world. Numerous studies have confirmed the epidemiological burden of NAFLD in IBD [16,17]. Nevertheless, there are no screening guidelines for NAFLD in the general population or those with IBD [15]. Our study provides gastroenterologists with real-world information about the influence of NAFLD on IBD hospitalizations and identifies individuals at higher risk of adverse clinical outcomes.

The risk of combined NAFLD and IBD has been associated with inflammation, a prolonged disease course and immunosuppressive therapy [18]. Advancing age, male sex, higher body mass index and prior surgery have also been associated with developing NAFLD in IBD patients [18,19]. The peak prevalence of NAFLD is reported in individuals between the ages of 50 and 65 years, mostly attributed to the increased risk of metabolic syndromes with aging [20,21]. Our study also shows a higher mean age for IBD hospitalizations with NAFLD, correlating to increased risk with prolonged inflammation and a longer disease course. Numerous studies have also shown an increased prevalence of NAFLD in males [22]. One of the most widely accepted explanations is the lower risk from gynoid gluteal-femoral subcutaneous fat distribution compared to the higher risk seen with android visceral adiposity. This is because upper body subcutaneous and visceral fat deposition, especially the abdominal, is associated with a higher risk for type 2 diabetes and cardiovascular disease. Lower body fat deposits, such as gluteal or femoral, are related to metabolic health and a lower risk for abnormalities of glucose and lipid homeostasis [23]. However, with increasing age, females also have an equally high risk of developing NAFLD, mostly due to changes in adiposity patterns [23]. Similarly, in our study there was a higher proportion of males in the IBD hospitalizations with NAFLD compared to females.

A recent study by Kubiliun et al assessed the impact of known genetic variants. It revealed that African American individuals generally carried fewer risk-increasing alleles for the development of NAFLD, compared to Whites and Hispanics [24]. Although most studies agree that the prevalence of NAFLD is lowest in African-Americans and highest in Hispanics, few demonstrate variability in the prevalence of NAFLD when comparing Whites with Hispanics. It has been theorized that this variation is due to differences in geography and diagnostic modalities [25]. Studies by Lazo et al and Trico et al showed a higher prevalence of NAFLD in Hispanics, followed by whites [26,27]. Kasarala et al noted similar racial patterns, with White (odds ratio [OR] 1.13, 95%CI 1.04-1.22; P<0.001) and Hispanic (OR 1.52, 95%CI 1.34-1.73; P<0.001) populations showing a higher percentage of NAFLD, whereas the African American population (OR 0.592, 95%CI 0.518-0.678; P<0.001) had a lower proportion of NAFLD [28]. We also noted racial disparities in our study, with the NAFLD cohort having a greater proportion of Hispanics and Whites, while the non-NAFLD cohort had a higher percentage of African Americans (Table 1). This disparity has been mainly attributed to variables such as lifestyle, dietary habits, genetic differences and socioeconomic factors [29]. The Western diet has more processed foods and includes more red meat, and high amounts of saturated fats and high-sugars [24].

Studies have demonstrated that T2DM, hypertension, dyslipidemia, and prior surgery were associated with a greater risk of NAFLD in IBD patients [17]. Emerging evidence shows that patients with chronic kidney disease, with or without dialysis, are more likely to develop hepatic steatosis due to inflammation, oxidative stress, insulin resistance, atherogenic dyslipidemia and various metabolic defects [30]. In a retrospective study of 80,000 patients, IBD was also associated with a greater risk of chronic kidney disease, especially in the younger patient population (hazard ratio 7.88, 95%CI 2.56-24.19; P<0.05). However, the pathophysiology is not completely understood [31]. Hypertension and cardiovascular disease are strongly linked with the aging of the liver via various molecular and cellular processes, which can enhance the development of NAFLD [32]. IBD is also independently associated with a greater risk of atherosclerotic cardiovascular disease, heart failure and atrial fibrillation. This is linked to shared risk factors, such as obesity, T2DM, gut microbiome dysfunction, and IBD disease progression or its treatment [33]. Similarly, in our study, there was a higher proportion of patients with hypertension, obesity, T2DM and chronic kidney disease in the NAFLD cohort compared to the non-NAFLD cohort (Table 1).

Prior bowel resection has been associated with a higher incidence of NAFLD and IBD concurrently. This association has been noted in multiple studies [10,34]. The pathophysiology was theorized to be small intestinal bacterial overgrowth, activation of toll-like receptor-4 leading to increased intestinal inflammation [35,36]. In our study, ileostomy as a complication in the IBD hospitalizations with NAFLD was statistically significant (OR 0.68, 95%CI 0.56-0.83; P=0.001). This is a unique finding that warrants further investigation.

With the recent advances in clinical practice and the evolution of biological and non-biological IBD treatments, the previously noted higher mortality risk in the IBD population has seen an overall improvement in the last decade [37]. In IBD populations, studies suggest a higher mortality rate in patients with UC compared to CD [38]. A study by Abomhya et al compared CD and NAFLD patients to NAFLD alone and found that patients with CD and concomitant NAFLD did not have higher mortality [39]. According to a study by Nooriyan et al, patients with CD and NAFLD were at a higher risk of death compared to those with CD alone (adjusted hazard ratio [aHR] 2.72, 95%CI 1.68-4.42; P<0.001) however the same was not true for patients with UC and NAFLD (aHR 1.35, 95%CI 0.79-2.32; P=0.27) [40]. In our study, we noted greater inpatient mortality among IBD hospitalizations with NAFLD compared to the non-NAFLD cohort (3.1% vs. 1.4%, P<0.001). Age-specific, sex-specific and race-specific mortality was also higher in patients with NAFLD.

A multivariate logistic regression analysis of risk factors in IBD hospitalizations with and without NAFLD showed that increasing age and comorbidities are major contributors to inpatient mortality. Although cardiovascular disease is reported to be most the common cause of death overall, our study interestingly did not show this association with inpatient mortality [41-43]. Our study also showed a higher risk of acute kidney injury including need for hemodialysis, lung injury with associated acute liver failure, and poor respiratory outcomes needing mechanical ventilator support, in patients with NAFLD. The underlying pathogenesis of NAFLD with increased hepatic factors, hepatokines (angiopoietin-like protein 4, fetal globulins and fibroblast growth factor), along with impaired reticuloendothelial system activity and inflammatory mediators contribute to the subsequent kidney, liver and lung injury patterns [44,45].

Based on a study by Peery et al, the annual health care expenditure estimated for gastrointestinal disease was $135.9 billion, with IBD accounting for $7.2 billion [46]. It is clear that IBD makes a substantial contribution to the increased healthcare burden in terms of total healthcare cost and increased use of healthcare services. NAFLD further amplifies this cost. In a study conducted by Allen et al, using a large national administrative claims database, it was found that the annual healthcare costs for NAFLD were almost double beyond the initial peri-diagnosis period ($3789 vs. $2298 per subject) [7]. Our study noted a higher mean total healthcare cost ($69,536 vs. $55,467, P<0.001) and a longer mean length of stay in the NAFLD cohort as compared to the non-NAFLD cohort. This is most likely due to more complications of disease courses resulting from different comorbidities, medical and surgical management, and other acute care needs. Higher admissions were noted in urban teaching hospitals (75.43%), constituting two-thirds of total inpatient admissions for NAFLD patients, followed by urban non-teaching (18.58 %) and rural hospitals (5.99%). Additionally, larger hospitals were found to have more than half of the admissions for NAFLD with IBD, with the rest going to medium and small hospitals. This may be mostly because large academic centers are usually tertiary care referral centers and accept complex and advanced patients. They also have highly trained and specialized physicians, along with the resources necessary for the management of complications, when compared to non-teaching urban and rural centers.

Our study had several strengths. The data for this study were gathered from the NIS, the largest multi-ethnic national database encompassing IBD hospitalizations. Additionally, our study was designed to analyze many items of outcome-oriented and biodemographic information and provides a comprehensive detailing of the magnitude of this disease. It will also help physicians understand the magnitude and burden of this disease. This positive association of hospital outcomes in patients of IBD with NAFLD warrants the benefits of a prospective study that can help physicians with resource allocation, predicting hospitalization courses and in turn improving prognostication.

However, our study is not without its limitations. There were inherent limitations due to its retrospective nature. The usage of ICD codes means that coding errors and misclassification cannot be excluded. The data analyzed were not from individual patients, but rather hospitalizations of IBD patients. Another limitation of the NIS data is an appropriate ICD-10-based billing of the disease and risk factors by providers. This might have introduced record bias by underreporting and potential misclassification of NAFLD patients into the non-NAFLD group. However, given the nature of the study and the database that was used, it is not possible to estimate the patient population that had NAFLD but were not tested or previously diagnosed. Hence, a more prospective study with both test and control arms will give a better understanding of these patterns and eliminate this record bias altogether.

Nevertheless, our sample size and analysis provide more information and a unique perspective on the impact of NAFLD in IBD patients. Large, multicenter prospective studies are needed to further investigate these associations and the associated risk of morbidity and mortality.

In conclusion, NAFLD is an important consequence of IBD. Our study found that IBD hospitalizations with NAFLD had higher all-cause inpatient mortality than non-NAFLD IBD patients (3.1% vs. 1.4%, P<0.001). NAFLD-IBD cohort patients also had a longer mean length of stay and greater total healthcare costs. This cohort of patients was also predisposed to a greater extent to severe complications, such as acute kidney injury, acute respiratory failure, acute liver failure and C. difficile infection, among others. More research into the association between the duration of IBD, IBD medication history and the development of NAFLD may provide a more thorough understanding of the correlation between NAFLD and IBD.