Diagnosis of Minimal Hepatic Encephalopathy Using Stroop EncephalApp

A Multicenter US-Based, Norm-Based Study

Sanath Allampati, MD; Andres Duarte-Rojo, MD; Leroy R. Thacker, PhD; Kavish R. Patidar, MD; Melanie B. White, RN; Jagpal S. Klair, MD; Binu John, MD; Douglas M. Heuman, MD; James B. Wade, PhD; Christopher Flud, MD; Robert O'Shea, MD; Edith A. Gavis, RN; Ariel B. Unser, BS; Jasmohan S. Bajaj, MD


Am J Gastroenterol. 2016;111(1):78-86. 

In This Article


Diagnosis for MHE is hampered by the lack of tools that can potentially be applied at point of care without psychological expertise.[13] Therefore, the use of relatively quicker tests such as the EncephalApp may be important to increase the rates of diagnosis and subsequent therapy of these patients. In the United States, given the lack of multicenter experience in MHE and of norms for PHES, which is the presumptive gold standard, it is even harder to test for MHE.[14,15] The current study results evaluate the applicability of EncephalApp for diagnosis independently against population norms and also in the context of two separate MHE tests, the ICT and PHES, for the diagnosis of MHE in multiple, non-geographically contiguous centers.

Because of the lack of norms available, we first recruited healthy controls for the diagnosis of MHE and adjusted them for age, education, and gender to increase standardization between sites. On the basis of these norms, we found, unsurprisingly, that the two gold standards tested were not perfectly correlated to each other.[16] We found that, similar to the single-center Virginia experience, that the psychomotor component of the EncephalApp, the OffTime+OnTime was significantly more impaired between those with and without MHE based on current gold standards compared with the outputs that require cognitive flexibility independent of psychomotor speed (OnTime minus OffTime) even though this output remained significantly more affected in cognitively impaired patients.[6,7] This is despite the variation in cirrhosis severity and proportion with prior OHE between the centers. Psychomotor speed-related determinants also remained significant in the validation data set and were ultimately predictive for OHE development.

The gold standards used evaluate different aspects of the cognitive function, whereas PHES is a paper–pencil test battery that evaluates psychomotor speed, visuo-motor coordination and attention using brief tests, the computerized and longer ICT tests working memory, response inhibition, and psychomotor speed.[8,9] The EncephalApp, which is relatively less intense compared with ICT, evaluates psychomotor speed and cognitive flexibility.[9] The proportion of subjects diagnosed with MHE using EncephalApp against direct norms, and MHE using EncephalApp cutoff points based on ICT, was similar between sites. There was a significant variation between MHE using EncephalApp based on PHES and direct diagnosis of MHE using ICT and PHES, respectively. Changes in PHES closely track liver disease severity, i.e., the MELD score, which was higher in the VA patients. The relative stability in MHE using EncephalApp norms between sites demonstrates the consistency of the EncephalApp in discriminating between groups and potentially increases its generalizability. The AUC of EncephalApp using PHES as the gold standard was overall higher than that using ICT as the gold standard. This may be due to the overlapping domains studied between both these EncephalApp and PHES (psychomotor speed underlies four of the five PHES components) compared with the ICT, in which response inhibition (i.e., lures) and not psychomotor speed-related outputs was used as the determinant of impairment.[8,9] When ICT or PHES was compared with EncephalApp, there was a high sensitivity but comparatively low specificity, which could encourage the use of EncephalApp as a high-sensitivity test in practice. Nevertheless, when each gold standard was analyzed as the basis of EncephalApp, cutoff points generated between sites were relatively consistent. The consistency was reiterated in the validation VA cohort, in which the proportion of MHE based on the EncephalApp alone, EncephalApp based on PHES, and EncephalApp based on ICT was statistically similar to the initial cohort.

The sites had different subject populations, with the highest number of those with OHE and the highest MELD score being seen in VA, whereas AR had patients with the lowest prevalence of alcohol-related liver disease, MHE diagnosed using PHES, and highest proportion of MHE diagnosed using ICT. Importantly, the cognitive performance of controls was similar between groups. There remained a moderate to good agreement between sites and between sites and combined values. As VA had the double the patients compared with AR or OH, it is not surprising that VA vs. combined had better phi coefficient compared with other sites. This difference between the sites reiterates the need for multicenter MHE studies, which have been lacking in the United States.

Our finding with poor agreement between individual "gold standards" is in alignment with prior single-center studies from Italy, Denmark, India, and Germany.[16–19] This highlights the lack of consensus between investigators regarding the optimal test for MHE diagnosis and also necessitated the use of two separate gold standards in this study. This experience, however, extends this beyond single-center studies and also evaluates agreements between different centers.

Although the diagnosis of MHE itself is important, ultimately it is the prediction of negative outcomes that is important to encourage testing.[2,4,18] Our results found that the prediction of OHE episodes was associated with EncephalApp results, whether it be related to independent norms or based on gold standard, especially in those without prior OHE. The PHES-based EncephalApp MHE results were likely not significant due to the inherent collinearity with individual PHES tests and MELD. The independent predictive ability of EncephalApp over liver disease severity and prior OHE underlines its potential use as a prognostic tool in this multicenter cohort.[8,12]

With the aging population of cirrhosis, impaired cognition in cirrhotics due to conditions other than MHE, i.e., dementia, remains a concern. To minimize this possibility, we excluded patients older than 65 years, those with signs or a history of dementia, and with a low mini-mental status examination. In addition, the cognitive tests used do not focus on memory-associated issues that predominate in most dementing conditions. The study is limited by controls that are different in demographics from the patients; however, this was controlled using regression of these values. This adjustment has resulted in formulae that are valid but may not be easy to impute into practice, which was the goal of EncephalApp. We have updated the site www.encephalapp.com to including age, gender and education-specific US-based norms for EncephalApp, ICT and PHES using the regression formulae generated by this study. This site is similar to www.redeh.org and will allow users to use any or all three of these tests as the basis for MHE diagnosis in their patients.

The current study shows that EncephalApp can be used as a method to diagnose MHE in a multicenter US population with acceptable inter-center agreement and prognostication value for the development of OHE. These results could increase the likelihood for MHE testing and potential treatment in the United States due to its multicenter norms for EncephalApp and other cognitive tests.