Symptom Classes in Decompensated Liver Disease

Abstract

Background & Aims:

Patients with decompensated liver disease have been categorized by disease severity. This analysis sought to classify patients with end-stage liver disease (ESLD) based on symptoms rather than disease state and to identify distinct severity classes of physical and psychological symptoms.

Methods:

Patients with a Sodium Model for ESLD (MELD-Na) score ≥15 were recruited from liver clinics in two healthcare organizations. They completed the Condensed Memorial Symptom Assessment Scale (CMSAS), Revised Ways of Coping Checklist (RWCCL), Patient Health Questionnaire (PHQ-9), Life Orientation Test-Revised (LOT-R), and the Short Form Health Survey (SF-36). Cross-sectional data were analyzed using latent class mixture modeling.

Results:

The sample (N=191, age 56.6±11.1 years, 33.5% ETOH, 28.3% NASH, 13.1% autoimmune/PBC/PSC) was predominantly male (64.2%), Child-Turcotte-Pugh (CTP) class C (49.5%), with an average MELD-Na score of 18.7±4.9. Three distinct classes of symptoms were identified: mild (26.7%), moderate (41.4%), or severe (31.9%) symptoms. Symptom classes were independent of disease severity and demographic characteristics, except age. All CMSAS symptoms and PHQ-9 scores were significantly different across the three classes (p-values <0.05). The symptom classes also differed significantly in physical and mental quality of life, optimism and avoidance coping behaviors (all p<0.001).

Conclusions:

Patient reported symptom severity occurred independent of disease severity, contrary to common assumptions. Focusing on the moderate and severe symptom classes as well as patients’ history of ESLD complications may enhance providers’ ability to improve symptom management for this population.

In the United States, chronic liver disease (CLD) and cirrhosis is the 11^th most common cause of death and affects almost 4.5 million adults.¹ Death rates related to CLD and cirrhosis increased 31% from 2000–2015.¹ The clinical course of CLD leads to high healthcare utilization, with 150,000 inpatient stays annually.²

As people with CLD progress to decompensated liver disease common symptoms include lack of energy, pain, and difficulty sleeping, all of which are highly prevalent, severe, and distressing.³ Other symptoms of decompensated liver disease include loss of appetite, nausea, vomiting, weight loss, and jaundice.⁴ Changes in level of consciousness and depression may also occur. Individuals with decompensated liver disease experience an average of 8.5 symptoms, making symptom management challenging for health care providers.⁵ Although symptoms occur together,⁶ little is known about potential symptom clusters or symptom classes in patients with ESLD and their ability to cope with symptoms. A symptom cluster can be defined as three or more symptoms that occur concurrently and may or may not share the same etiology.^7,p.39 Symptoms within a cluster can be analyzed in two ways: the identification of specific symptom clusters (e.g., pain cluster) or of different class or subgroup scores (e.g., low or high scores) based on patient responses on symptom specific instruments.⁷ Gaining knowledge about the impact of different symptom classes in patients with ESLD could allow for a more thorough symptom assessment, creating the possibility for more targeted and efficient symptom management to improve patient outcomes and physical and mental health. Specific symptom classes have been linked to poorer quality of life and increased risk of death in other chronic diseases.^8–10 The purpose of this study was to identify distinct severity classes of physical and psychological symptoms among patients with ESLD and examine key demographic, coping style, physical and mental health (quality of life), and clinical differences among the observed classes.

Methods

Study design

A joint institutional review board (IRB) approved the longitudinal study of ESLD patient-caregiver dyads (NINR: 1R01NR016017–01) from which the cross-sectional survey baseline data for this paper came.¹¹

Study sample

Patients and caregivers were recruited from liver clinics in two urban study sites, a Veterans Affairs medical center and an academic center. Each site provides liver transplant services and is the primary referral center for advanced liver disease care in the region. Patients were recruited whether or not they were liver transplant candidates and were eligible to enroll in the study if they were 21 years or older, had a Sodium Model for End-Stage Liver Disease (MELD-Na) score of 15 or greater, and decision-making capacity. They were excluded if they had a diagnosis of liver cancer, had undergone a liver transplantation, were being treated with antiviral medications for hepatitis C, were diagnosed with a psychiatric illness (e.g., schizophrenia) that was not well controlled, or had a non-liver terminal illness with life expectancy less than 12 months.

All health care providers (N = 14) at the two study sites participated in screening of patients during routine appointments in the liver clinics where study investigators were present for referrals. Patients and caregivers who decided to participate signed an informed consent form during the clinic visit. Initial survey packets were completed in the clinic or at home. A few patients and caregivers were referred by providers and contacted remotely; a separate survey packet including a consent form was mailed to each of them with a stamped, self-addressed envelope to return the survey after completion. Upon completion of the survey, each patient and caregiver received $30 each for their time and inconvenience. No participants were recruited and consented via telephone.

Measurement

The 14-item Condensed Memorial Symptom Assessment Scale (CMSAS) was used to capture patient reported symptom prevalence and distress.¹² Patients ranked each of 11 physical symptoms from 1 (rarely) to 5 (very much) based on how much it bothered or distressed them. Each of three psychological symptoms was ranked from 1 (rarely) to 4 (almost constant) based on frequency. If patients did not experience a specific symptom, a score of zero was listed. The psychometrically validated CMSAS was selected because it has been used with patients with cirrhosis.^12,13 It is shorter and less burdensome than the Memorial Symptom Assessment Scale (MSAS), which we had used successfully in a pilot study.³

The 10-item Patient Health Questionnaire (PHQ-9) was used to obtain patient reported severity of depressive symptoms.^14,15 The valid and reliable questionnaire provides a single score which is the sum of nine items rated on a 4-point scale (0 = not at all to 3 = nearly every day). PHQ-9 scores of 5, 10, 15, and 20 represent mild, moderate, moderately severe, and severe levels of depression, respectively.¹⁴

The reliable and valid Revised Ways of Coping Checklist’s (RWCCL) 15-item problem-focused coping subscale and the 10-item avoidance subscale were used to measure coping style.^16,17 For each item, the respondents were asked to record the degree to which they used that strategy to deal with their problem (0 = never used, 1 = rarely used, 2 = sometimes used, 3 = regularly used).

The reliable and valid Life Orientation Test-Revised (LOT-R) was used to measure optimism and pessimism.^18,19 Participants were asked the extent to which they agreed to 10 statements from 0 (strongly disagree) to 4 (strongly agree). The higher the score the more optimistic or pessimistic one is.

The Short Form Health Survey (SF-36) is a reliable and validated multi-purpose, health survey with 36 questions. Its psychometrically-based physical and mental health summary measures were used to calculate participant physical and mental health/quality of life.²⁰ Higher scores demonstrate better health.

Comorbidities were assessed using the Charlson Comorbidity Index.²¹ Patients’ clinical information was collected from their medical records.

Statistical analysis

Prior to the statistical analysis missing data were addressed. Skipped questions were treated as missing at the individual item level. Questions with two circled answers were handled by averaging the response unless otherwise indicated by the survey instrument instructions.

Descriptive statistics of central tendency and dispersion were used to describe the sample. Latent class mixture modeling was used to identify distinct severity classes based on the 14 CMSAS scores (range 0–5 for physical items [distress] and 0–4 for psychological symptoms [frequency]) and PHQ-9 score (performed in Mplus version 8).²² Model fit of mixture model solutions (i.e., 3 vs. 2 clusters) were compared using entropy (closest to 1), classification probabilities (above 0.95), class proportions (no class with <5% of the sample) and the parametric bootstrapped likelihood ratio test (p<0.05).²³ Full information maximum likelihood estimation was used to account for the < 1% of data that were missing at random. Symptoms and clinical factors were compared across classes using ANOVA (with eta² to calculate effect sizes for symptom data) and chi-square testing where appropriate.

Results

Of 336 eligible ESLD patients, 78 were excluded for various reasons (patient not feeling well, unable to contact, died or received a liver transplantation before returning baseline data). Two patients were excluded after providing consent. Of the 256 patients who enrolled in the study, 64 were excluded. Of these 27 were lost to follow-up, 14 died, 14 withdrew (too sick or too busy) and 9 were withdrawn by the investigator (received a transplant prior or had cancer). Data from 191 were available for analysis (Flow Diagram, Supplemental Material).

The sample (N=191, mean age 56.6±11.1 years) was predominantly male (64.2%), white (89.5%), and had had cirrhosis for a median of 2 [IQR = 1–5] years (Table 1). Participants’ CLD included a mix of etiologies. They had an average MELD-Na score of 18.7 and 49.5% of them had Child-Turcotte-Pugh (CTP) Class C liver disease.

Table 1.

Characteristics of the Sample by Latent Symptom Class

	Full Sample N=191	Mild n=51	Moderate n=79	Severe n=60	p-value
% female	68 (35.8%)	13 (25.5%)	29 (36.7%)	26 (43.3%)	0.144
Age, years	56.6±11.1	61.4±11.2	54.7±10.2	55.5±10.2	0.001
Duration of cirrhosis, years	4.7±5.9	3.5±4.7	4.6±6.7	5.8±5.6	0.129
Charlson Comorbidity Index Score	3.7±1.9	4.1±2.1	3.4±1.7	3.8±2.0	0.124
MELD-Na score	18.7±4.9	18.0±4.8	19.2±4.9	18.8±4.9	0.356
Primary Etiology
NASH/Cryptogenic	54 (28.3%)	16 (31.4%)	18 (22.8%)	20 (32.8%)
Viral hepatitis	22 (11.5%)	5 (9.8%)	8 (10.1%)	9 (14.8%)
ETOH	64 (33.5%)	17 (33.3%)	30 (38.0%)	17 (27.9%)	0.684
Autoimmune/PBC/PSC	25 (13.1%)	5 (9.8%)	12 (15.2%)	8 (13.1%)
ETOH & Hep C	21 (11.0%)	6 (11.8%)	8 (10.3%)	7 (11.5%)
Other	5 (15.0%)	2 (3.8%)	3 (3.8%)	0 (0.0%)
Child-Turcotte-Pugh class
Class A	7 (3.7%)	3 (6.0%)	3 (3.8%)	1 (3.7%)
Class B	89 (46.8%)	27 (54.0%)	37 (46.8%)	25 (41.0%)	0.393
Class C	94 (49.5%)	20 (40.0%)	39 (49.4%)	35 (57.4%)

Abbreviations: MELD-Na, Sodium Model for End-Stage Liver Disease; NASH, Nonalcoholic Fatty Liver Disease; ETOH, Ethyl Alcohol; Hep C, Hepatitis C; PBC, Primary Biliary Cholangitis; PSC, Primary Sclerosing Cholangitis.

Three distinct classes of symptoms (in a gradient from mild to severe) were identified (entropy = 0.87, average posterior probabilities exceeded 0.94, parametric bootstrapped likelihood ratio test p<0.001). Of the patients, 26.7% experienced mild symptoms (Class 1), 41.4% had moderate symptoms (Class 2), and 31.9% had severe symptoms (Class 3; Table 2).

Table 2.

Symptom by Latent Symptom Class

	Mild n=51	Moderate n=79	Severe n=61	F-value: p-value	eta-square
Lack of Energy	1.6±1.4	3.6±0.9	4.3±0.7	105.0: <.0001	0.527
Lack of Appetite	0.7±1.1	2.0±1.4	3.3±1.3	50.9: <.0001	0.351
Pain	0.9±1.4	2.3±1.7	3.9±1.1	57.0: <.0001	0.377
Dry Mouth	0.6±1.1	1.8±1.7	3.1±1.5	37.1: <.0001	0.283
Weight Loss	0.5±1.0	1.2±1.5	2.4±1.7	23.8: <.0001	0.202
Drowsy	0.8±1.2	2.7±1.3	3.8±1.2	75.2: <.0001	0.445
Shortness of Breath	0.9±1.3	1.6±1.6	2.7±1.6	21.1: <.0001	0.183
Constipation	0.5±1.0	0.9±1.4	1.5±1.6	8.9: =.0002	0.086
Difficulty Sleeping	1.5±1.6	3.0±1.5	4.0±1.5	36.8: <.0001	0.281
Difficulty Concentrating	0.6±1.2	2.1±1.4	3.4±1.4	60.3: <.0001	0.391
Nausea	0.4±0.8	1.2±1.4	2.7±1.6	47.0: <.0001	0.333
Worrying*	0.7±1.1	1.8±1.1	2.8±0.9	56.2: <.0001	0.374
Feeling Sad*	0.4±0.8	1.3±1.1	2.5±1.0	57.4: <.0001	0.379
Feeling Nervous*	0.4±0.8	1.1±1.1	2.1±1.2	36.3: <.0001	0.278
Depression (PHQ-9)	4.1±3.9	8.5±3.5	15.3±4.7	113.1: <.0001	0.546

^*Psychological symptoms

All 14 CMSAS symptoms differed significantly across the three classes, as did PHQ-9 scores (Table 2 and Figure 1).

Figure 1:

The green line indicates the mild symptom class, the yellow line indicates the moderate symptom class, and the red line the severe symptom class.

The largest effect sizes in symptoms across the three classes were depression (eta²=0.546), distress from lack of energy (eta²=0.527), distress from feeling drowsy (eta²=0.445), distress from difficulty concentrating (eta²=0.391), frequency of feeling sad (eta²=0.379), distress from pain (eta²=0.377), and frequency of worrying (eta²=0.374).

The three classes of symptoms occurred independently of MELD-Na, CTP score, gender proportion, duration of cirrhosis, etiology, and Charlson comorbidity score. We did not detect significant associations between disease severity and symptom classes either independently (Table 1) or in combination with Charlson comorbidity score (multivariate data not shown) in this sample.

Patients in the mild symptom class were older than patients in the other symptom classes (p=0.001) (Table 1). There was no difference among classes in problem-focused coping, but there was a gradient of avoidance coping from severe to mild symptom classes (p<0.0001) (Table 3), with the severe symptom class demonstrating the most avoidance coping. Optimism also demonstrated a gradient with the most optimistic in the mild symptom class (p<0.0001). Both physical and mental health were significantly worse across the three symptom classes (p<0.0001).

Table 3.

Other Factors by Latent Symptom Class

	Mild n=51	Moderate n=79	Severe n=61	F-value: p-value	eta-square
Problem-Focused Coping	29.5±10.6	30.2±8.3	30.8±9.5	0.25: =.7760	0.003
Avoidance Coping	10.0±5.7	12.5±5.7	16.2±5.9	16.4: <.0001	0.150
Optimism	17.6±4.4	16.1±5.1	13.1±5.8	11.3: <.0001	0.107
SF-36 Physical Summary	39.8±9.0	33.0±9.8	27.7±7.4	25.5: <.0001	0.215
SF-36 Mental Summary	51.7±8.5	42.9±10.1	32.8±10.1	51.9: <.0001	0.358

Discussion

This study of 191 adults with ESLD identified distinct classes of physical and psychological symptoms. The study has significant findings. First, the three classes had a gradient from mild to severe symptoms and were independent of MELD-Na, CTP score, gender proportion, duration of cirrhosis, etiology, and Charlson comorbidity score. Second, the largest difference in symptoms among the three classes were depression, distress from lack of energy, distress from feeling drowsy, distress from difficulty concentrating, frequency of feeling sad, distress from pain, and frequency of worrying. Third, in the mild symptom class, patients were slightly older compared with patients in the severe symptom class. Fourth, as expected, both physical and mental health were progressively worse across the three symptom classes from mild to severe.

This study offers counterintuitive evidence that symptom class is disassociated with the severity of underlying liver disease and comorbidities. For health care providers, the recognition that symptom experience in ESLD may not be linked with disease severity is an important finding. The dissociation of symptom experience and disease severity has also been noted in the advanced heart failure (HF) literature.¹⁰ In this study, the commonly used MELD-Na score was not associated with the symptom classes. This argues that either we are missing some measurable portion of the disease, or how patients experience symptoms apart from the disease severity is an independent outcome in need of proper assessment tools. In looking at other severe illness states, Miaskowski et al. identified three distinct symptom classes among outpatients with breast, gastrointestinal, gynecological, or lung cancer who completed the MSAS: 1) low, 2) moderate, and 3) high symptom occurrence.²⁴ Only performance status and comorbidity scores differed among the classes; none of the clinical characteristics they measured (i.e., time since diagnosis, cancer diagnosis, types and number of prior treatments, reason for current therapy, presence or number of metastatic sites) differed among classes.

A gradient from mild to severe was observed in both physical and psychological symptoms in our study based on the CMSAS scale. This finding speaks to the importance of psychological symptoms as being at least equal to physical symptoms in influencing patients with ESLD experiences. Symptoms across the three classes included three psychological symptoms and four physical symptoms. This is clinically relevant because physical and psychological symptoms occur at the same time and should be addressed by health care providers concurrently. In clinical practice, the focus is often on patients’ physical symptom experience with less attention to the co-occurrence of concurrent psychological and physical symptoms, despite the association of psychological distress with increased liver disease mortality.²⁵

Studies in advanced HF corroborate our early findings in ESLD. Among patients with class III/IV HF measuring physical and psychological symptoms, patients in the moderate symptom class were 80% more likely than patients in the mild symptom class to experience a clinical event requiring hospitalization in the coming year.¹⁰ Patients in the severe HF symptom class were twice as likely as patients in the mild symptom class to have a clinical event in the coming year, independent of prognostication using objective data. These finding indicate the importance for providers to obtain a detailed physical and psychological symptom assessment of patients with ESLD.

In addition to HF,¹⁰ a limited but increasing body of evidence suggests that symptom classes occur in patients with other chronic diseases (e.g., COPD,²⁶ dementia,²⁷ and cancer).²⁸ In HF, mismatches between integrated data on physical and psychological symptoms and hemodynamic values have been identified.²⁹ Lee et al. characterized three symptom-hemodynamic profiles: 1) mild symptoms and poor hemodynamics, 2) moderate symptoms and matched good hemodynamics, and 3) severe symptoms and average hemodynamics. The two mismatched profiles including mild symptoms and severe symptoms were associated with increased event risk (emergency room visits, hospitalization, and death). For health care providers to complete a comprehensive assessment is critical to understand the relationship, or lack of, between symptom classes and objective clinical data in patients with ESLD. Adding the additional dimensions of psychological and physical distress to the usual focus on objective disease findings could enhance providers’ ability to improve symptom management for patients and identify patients who may be at risk for upcoming clinical events, similar to findings in HF.²⁹ We may not know exactly how symptom classes fit into the physiologic landscape of cause and effect and patient outcomes, but it is clear that psychologic symptoms play a role in patient outcomes and need greater attention in the clinical assessment of patients with ESLD.

By assessing for patterns of symptoms, providers may identify appropriate resources to improve overall symptom management of patients with ESLD such as referral to social work, psychiatry, physical therapy or palliative care. For patients with chronic diseases engaged with palliative care, the most common aspects of care received were symptom management followed by psychological support/counseling, and disease education.³⁰ Research on the effectiveness of palliative care has primarily targeted patients with cancer. Findings have shown that patients achieve better symptom control,³¹ quality of life,³² lower depression,³² and longer survival.³³

This study has several limitations. The design was cross-sectional, which prohibits an exploration of symptom class membership over time. Second, although 15 symptoms were included in the analysis, they may not capture the complex range of symptoms that patients with ESLD may experience. Third, of the 15 symptoms measured, only 4 were psychological. Lastly, the study excluded patients with a MELD-Na<15 which resulted in few CTP A patients in the cohort. Therefore, observations about symptom class and patients with decompensated liver disease may not be generalizable to those with well compensated disease. Strengths of the study include a large sample size with a uniform diagnosis, which may enhance generalizability of study findings.

Conclusion

Three distinct classes of physical and psychological symptoms were identified in patients with ESLD and found to be independent of MELD-Na, CTP score, gender proportion, duration of cirrhosis, etiology, and Charlson comorbidity score. Assessing the complex effect of simultaneously occurring symptoms in ESLD is essential for improving symptom management of the disease. Future researchers should examine interactions of symptoms within each class, the stability of classes over time, and the relationship between symptom classes, clinical events, and mortality. Such knowledge could lead to more targeted and efficient symptom management strategies that would improve class symptoms and quality of life in patients with ESLD.

What You Need to Know.

Background:

Chronic liver disease and cirrhosis is the 11^th most common cause of death in the United States, with a rising incidence over the last two decades. The symptoms associated with decompensated liver disease affect both mental and physical quality of life. To date, little research has focused on the patient experience and examined how symptoms cluster in individuals and whether this clustering offers clinical insights into management and prognosis.

Findings:

This study identified 3 distinct classes of symptoms in patients with predominantly Child’s B and C cirrhosis with a mean MELD-Na of 18.7. These 3 classes describe the patient symptom experience as mild, moderate, or severe, and counterintuitively showed no correlation to disease severity. Symptom classes were correlated with physical and mental quality of life.

Implications for patient care:

Clinicians should be aware that symptom experience may not mirror disease severity. Examining how symptoms vary over time and with disease progression may lead to more targeted symptom management strategies.

Abbreviations:

CLD

Chronic Liver Disease

CTP

Child-Turcotte-Pugh

CMSAS

Condensed Memorial Symptom Assessment Scale

ESLD

End-stage liver disease

ETOH

Ethyl Alcohol

LOT-R

Life Orientation Test-Revised

MELD-Na

Model for End-Stage Liver Disease including sodium

NASH

Nonalcoholic Fatty Liver Disease

PHQ-9

Patient Health Questionnaire

PBC

Primary Biliary Cholangitis

PSC

Primary Sclerosing Cholangitis

QOL

Quality of Life

RWCCL

Revised Ways of Coping Checklist

SF-36

Short Form Health Survey