Nonalcoholic fatty liver disease (NAFLD) has become the most common cause of chronic liver disease in children and adults not only in the United States, but in many parts of the world. The term NAFLD encompasses histological changes ranging from simple steatosis (ie, fat in the hepatocytes without inflammation); steatosis and inflammation with or without fibrosis, the latter being designated nonalcoholic steatohepatitis (NASH); to cirrhosis with or without hepatocellular carcinoma. The rising prevalence of NAFLD is not surprising given that obesity has become epidemic in the last few decades. The most common indication for liver transplantation by the year 2025 has been projected to be NASH (1). The disease burden of NAFLD has often been neglected in the past, but during the last decade, it has come to the forefront because there has been a necessary increase in awareness of the natural history of disease and the long-term risk of developing cirrhosis and hepatocellular carcinoma. Health-related quality of life (QOL) is a key component in assessing disease burden and is particularly of value in those with chronic disease.
Several studies looking at QOL in adults and children with NAFLD have been published (2–4). The NASH Clinical Research Network (NASH CRN), a network established by the National Institute of Diabetes and Digestive and Kidney Disease, has been involved in analyzing data from individuals with NAFLD enrolled in a large US database. An initial study involving 237 adult patients with NAFLD showed significantly lower QOL scores compared with those with hepatitis B or hepatitis C (2). A more recent study from the NASH CRN looking at 713 adults showed that those with NAFLD had worse physical and mental health scores compared with the general population in the United States with and without chronic illness, and also reported that the poorer physical scores were associated with cirrhosis (3). In the pediatric NASH CRN study, children with NAFLD showed impaired QOL compared with healthy children (4).
Obese children have poor self-esteem and are at increased risk for symptoms of depression and interpersonal difficulties (5–7). The fact that children with NAFLD have impaired QOL is not surprising. Although there have been studies to look at QOL in children with NAFLD in comparison to those without, it is not clear whether the QOL is impaired secondary to NAFLD per se or because of obesity. Additionally, although QOL has been examined in children with NAFLD, other psychological aspects relevant to obese children, including depression and self-esteem, have not. Finally, there does not appear to be any information on how present management of NAFLD affects QOL in children.
In the present study of children with a diagnosis of NAFLD, our objectives were to examine the association between NAFLD and QOL as well as other psychosocial outcomes, to compare psychosocial outcomes to obese children without known NAFLD; and to determine whether present standard care for NAFLD results in weight loss and improvement in psychosocial outcomes longitudinally.
Patients and Procedures
The study received approval from the Mount Sinai School of Medicine institutional review board. All of the eligible patients were given an informed consent procedure before participating in the study. As per institutional requirements, additional written assent forms were completed by patients between ages 14 and 17; verbal assent was sought for all of the patients younger than 14 years.
Children between 8 and 17 years old with NAFLD were enrolled prospectively in the pediatric liver clinic. A diagnosis of NAFLD was made if at least 3 of the following criteria were met: body mass index (BMI) >97th percentile, alanine aminotransferase (ALT) >50 IU/L, ultrasound consistent with NAFLD, and/or liver biopsy histology consistent with NAFLD (8). A total of 14 children had a liver biopsy. All of the children had screening tests performed to rule out other common causes of liver disease including viral hepatitis, autoimmune hepatitis, α1-antitrypsin deficiency, and Wilson disease. After informed consent procedures, an investigator distributed questionnaires to children participating in the study and remained available to assist if necessary. Questionnaire completion took approximately 30 minutes and typically occurred at the end of a clinic visit. After being given standard care treatment recommendations, participants were scheduled for a clinical follow-up visit approximately 6 months later. At the time of their follow-up visit, participants were then given the set of questionnaires again. After the baseline visit, participants’ age, BMI, and ethnicity were recorded from their medical charts. After the follow-up visit, only BMI was recorded.
Comparison Cohort: Obese/Overweight Controls
The comparison group of obese/overweight control (OC) children (n = 40) was recruited during the first day of a summer camp for obese/overweight children at Mount Sinai Medical Center and during patients’ routine visits to the hospital's general pediatrics clinic. An investigator approached the pediatric staff to determine whether any patients who presented for a routine outpatient visit were obese/overweight based on their medical growth charts, and had not been diagnosed as having NAFLD or any other chronic illness. A member of the pediatric staff then asked these selected patients and their parents whether they would be interested in participating in a research project. None of the 40 had been diagnosed as having NAFLD and 21 of the 40 children had documented ALT in normal range (ULN <53 U/L). Children at the summer camp had been screened medically and NAFLD had been ruled out on the basis of normal ALT. Although additional sonograms would have been ideal, these had not been done when associated with normal ALT. Children attending the general pediatric clinics met the criteria based on BMI and had no medical condition deemed to be chronic, but not all had screening for ALT. It was believed that given they did not carry a diagnosis of NAFLD and the medical implications associated with it, they could serve as controls.
After informed consent procedures, participants were asked to complete the same questionnaire packet as the NAFLD cohort. This occurred immediately after the end of the camp session or routine appointment in either the examination room or in the waiting room based on families’ preferences. The investigator remained available in the event that a child requested help completing the measures. Lastly, the child's, age, BMI, and ethnicity were recorded from their medical charts.
The standard care given to participants included both dietary and physical activity recommendations. Children had regular clinic visits where their liver chemistries were checked. They had liver imaging and liver biopsy as required. Those enrolled in research protocols or with advanced histological changes were prescribed antioxidant therapy such as vitamin E. Families were given the option to consult with a nutritionist for assistance with implementing dietary changes and an endocrinologist. No enrolled child received any novel experimental drug/therapy. Overall, the process was not standardized but was consistent with present recommendations for NAFLD management.
Children's Depression Inventory
The CDI is a widely used scale that measures symptoms of depression within the last 2 weeks in children and adolescents ages 7 to 17 years (9). It consists of 27 items, which assess negative mood, anhedonia, ineffectiveness, indecisiveness, and negative self-esteem. Higher scores indicate greater endorsement of depressive symptoms. This scale has demonstrated internal consistency reliability ranging from α = 0.83 to α = 0.94 across studies of children and adolescents. It has also been shown to have test-retest reliability, and construct convergent and predictive validity.
Pediatric Quality of Life Inventory 4.0
The Pediatric Quality of Life Inventory (PedsQL) is a commonly used measure of health-related QOL in healthy and medically ill children through 4 subscales (10). The subscales included are physical functioning, emotional functioning, social functioning, and school functioning. There is a child (ages 8–12) and adolescent version (ages 13–18). This measure consists of 23 items that are applicable for healthy school and community populations as well as those representing medically ill children, and it is scored using a 5-point Likert scale. From 0 to 100, higher scale scores indicate better health-related QOL. When administered to a group of 1677 healthy and medically ill subjects from pediatric health care settings, it was found to wield both reliability and internal validity because it was also related to indicators of injury and illness burden (11).
Body-Esteem Scale for Adolescents and Adults
The Body-Esteem Scale for Adolescents and Adults (BESAA) is a brief measure that is used to assess the effect of weight on self-esteem (12). It consists of 23 items and addresses general physical appearance, physical comparison to one's peers, personal self-reflection of physical attributes, and weight satisfaction and/or preoccupation that are divided into the 3 subscales of appearance, attribution, and weight. The degree of agreement for each statement is indicated on a 6-point Likert scale ranging from 0 (never) to 5 (always). Higher scores suggest greater self-esteem related to one's appearance. Shown to have discriminant and construct validity, high test-retest correlation, internal consistency, and thus reliability across age ranges have been demonstrated for all of the 3 subscales when administered to 1334 participants between the ages of 12 and 25 years. Specifically, reliability for BESAA appearance was α = 0.92, for BESAA attribution α = 0.81, for BESAA weight α = 0.94. Test-retest reliability 3 months after the initial administration was high as well with BESAA appearance r(95) = 0.89, P < 0.01, BESAA attribution r(95) = 0.83, P < 0.01, and BESAA weight r(95) = 0.92, P < 0.01. Based on earlier research, before including in statistical analyses, we examined the reliability of the BESAA scales separately for children younger than age 12. Internal consistency for BESAA appearance was α = 0.86 (α = 0.88 for children older than 12), for BESAA attribution α = 0.84 (α = 0.79 for older children), and for BESAA weight α = 0.87 (α = 0.84 for older children). Because the BESAA scales appeared to be reliable for all of the enrolled participants, data were included for all of the children.
All analyses were conducted using SPSS 17.0 (SPSS Inc, Chicago, IL). A P value (α level) ≤0.05, 2-tailed was chosen as the level of statistical significance. Before recruitment, there were approximately 60 patients with NAFLD on the clinic roster within the specified age range. The study was powered with the assumption that we may be able to recruit 40 to 50 of these patients, with roughly the same number of OCs because our typical recruitment rate for psychosocial research is approximately 80%. To determine the range of effect sizes that our target sample could yield, a power analysis was conducted using the G*Power 3 statistical power analysis program (13) with a desired power of 0.80 at α = 0.05. Using analysis of variance (ANOVA) with 2 groups, the necessary sample size to detect large effect sizes was 52; for medium effect sizes it was 128. Therefore, the present study, n = 98, is powered to detect between large and medium effect sizes.
Independent samples t tests and χ2 analyses were used to compare demographic characteristics between groups (eg, NAFLD, OC). All of the analyses of “weight” used BMI. For comparisons between the NAFLD and OC groups on the CDI and PedsQL total scores, ANOVA was used. Multivariate analyses of variance (MANOVA) were conducted to compare scores between these groups on the CDI, PedsQL, and BESAA subscales. To examine change over time within the NAFLD group, paired samples t tests were used for BMI, whereas ANOVA and MANOVA were conducted on the total score or subscales of each study measure with weight loss status entered as a separate factor.
Preliminary Analyses Between NAFLD and OC Groups
The 2 groups were first compared on demographic variables. Independent sample t tests were used to compare the ages and BMI scores for both the NAFLD and OC group. In both instances, there were no significant differences between the 2. The mean age in the NAFLD group was 12.27 years (standard deviation [SD] 2.34) and that in the OC was 11.75 years (SD 2.78). Furthermore, the mean BMI in the NAFLD group was 29.78 (SD 5.60), whereas the mean BMI for the OC group was 27.65 (SD 5.18); however, sex composition did differ between the groups with the OC group having significantly more women, χ2 = 11.01, P < 0.01. Of particular relevance to NAFLD, there was no difference in the frequency of Hispanic youth in each group (79.2% in the NAFLD cohort, 85% in the OC). Because the OC sample was recruited at 2 different settings, preliminary comparisons on demographic variables were conducted. There were no differences on any of the demographic variables between patients recruited at the camp and in clinic.
Comparisons on Primary Outcomes Between NAFLD and OC
Table 1 depicts means and SDs for each cohort (the baseline assessment for the NAFLD group). First, total CDI score was examined. An ANOVA was conducted controlling for sex. The mean score for the NAFLD group was 10.32 (SD 8.21) and that for the OC group was 6.34 (SD 6.06), F(1,83) = 8.03, P < 0.01, with the NAFLD group scoring higher on the CDI than the OC group. Next, a MANOVA was run, again controlling for sex, consisting of the 5 CDI subscale scores. Again, the NAFLD group scored significantly higher, F(5,79) = 2.83, P < 0.01. When examining the individual subscales, negative mood, ineffectiveness, and negative self-esteem were all significantly higher in the NAFLD group. Within the OC group, there were no differences between children recruited at the camp and in the clinic. There was no significant correlation between BMI and CDI total and subscale scores for the OC group; however, for the NAFLD group, BMI was significantly correlated with negative mood (r = 0.40, P = 0.01) and interpersonal problems (r = 0.31, P = 0.04).
Table 2 displays mean scores by group for the PedsQL total and subscale scores. An ANOVA controlling for sex was also used to compare the results of the PedsQL total score in the NAFLD and OC groups. This result was a nonsignificant trend, F(1,85) = 2.83, P = 0.10. Next, a MANOVA was conducted on the PedsQL subscales by group, which was also nonsignificant, F(4,81) = 1.14, P = 0.34. Within the OC group, patients recruited from the camp had significantly higher scores on the physical functioning and social functioning subscales. There were no significant correlations within the OC group between BMI and PedsQL total and subscale scores. Interestingly, BMI correlated with PedsQL school, r = −0.33, P = 0.02.
Table 3 displays means and SDs for each group. A significant difference was found between the NAFLD and OC cohorts’ responses on the BESAA, when the 3 subscales of appearance, attribution, and weight were entered into a MANOVA, F(3,80) = 3.10, P = 0.03. All 3 subscales significantly differed between the groups, with lower scores in the NAFLD group. Furthermore, there was also an interaction with sex, F(3,80) = 3.35, P < 0.03, with the lowest scores among women with NAFLD. Within the OC group, there were no differences between children recruited at the camp and in the clinic. For children with NAFLD, BMI correlated with all of the BSEAA subscales: appearance (r = −0.44, P < 0.01), attribution (r = −0.57, P < 0.01), and weight (r = −0.41, P = 0.01).
Change Over Time
Of the 48 participants with NAFLD, 33 (68.8%) completed follow-up appointments. The PedsQL scores did not change after standard care in the 33 patients with NAFLD who completed the follow-up evaluations, but the CDI score differed between those whose BMI improved or not. The baseline BMI for those who completed a follow-up appointment (mean 29.10, SD 4.57) and those who did not (mean 31.27, SD 7.35) was first compared. This result was nonsignificant, t(46) = 1.25, P = 0.22; however, those who did not come back for a follow-up visit had significantly higher baseline scores on the CDI total, negative mood, interpersonal relationships, and anhedonia scales, as well as all of the BSEAA scales. Among those who completed follow-up, to examine changes in BMI over time, a paired samples t test was performed. The difference in BMI before (mean 29.10, SD 4.57) and after the standard care treatment (mean 28.95, SD 4.41) was not significant, t(32) = .37, P = 0.71. A total of 18 children (54.5%) had stable or improved BMI (mean 1.59, SD 2.10), whereas 15 children (45.5%) had a worse BMI (mean 1.58, SD 1.14). Further analyses controlled for whether children experienced improved BMI outcomes.
Table 4 displays mean and SDs on all of the psychosocial measures before and after standard care for the NAFLD cohort. To assess changes on the measures before and after treatment a repeated measures factorial ANOVA was conducted to examine whether there were differences over time overall and between those whose BMI improved or not. CDI total decreased overall, F(1,27) = 6.30, P = 0.02, and there was also a significant interaction between time and BMI status with a significant decrease in depression surprisingly more so among those whose BMI worsened, F(1,27) = 5.37, P = 0.03. There were no other significant changes over time in both overall and between groups on the other outcome measures (including the CDI scales as well as all of the PedsQL and BSEAA scales).
The present study is the first to compare QOL in children with NAFLD with obese controls and also to assess psychological status by using standardized measures of depression (CDI) and the effect of weight on BSEAA. In our study, QOL in children with NAFLD was not significantly different from obese children without NAFLD. This finding was surprising to us because other larger studies have shown poor QOL in children with NAFLD compared with healthy controls (4); however, the fact that psychosocial parameters including depression and self-esteem are worse in children with NAFLD than in obese children without NAFLD is an important new message to health care providers and parents. It seems that assessing QOL alone may be too broad when trying to determine specific psychological manifestations of NAFLD exclusive of BMI. Without the inclusion of the more comprehensive measures, clinically significant differences could have been masked.
One of the reasons why others have not compared with obese controls may be related to the fact that specialists often do not have access to children outside their given specialty. Our group collaborated with a general pediatrician who took responsibility for selecting appropriate non-NAFLD obese controls. Thereafter, a research assistant under the supervision of a pediatrician and study psychologist attended the general pediatric clinic and summer camp to consent eligible children. The mean CDI score, as well as relevant subscales, were worse in children with NAFLD than in obese controls. Specifically, the individual subscales, negative mood, ineffectiveness, and negative self-esteem were all significantly higher in the NAFLD group. A significant difference was also found between the NAFLD and OC cohorts’ responses on the BESAA, with lower scores in the NAFLD group in all 3 subscales of appearance, attribution, and weight. Furthermore, there was also an interaction with sex, with lowest scores among women with NAFLD. This is consistent with the study performed in adults in which female sex was one of the factors independently associated with lower QOL (3). Ethnicity did not influence scores in our study, and the proportion of Hispanic ethnicity was not significantly different in the obese control and NAFLD groups.
The decision to recruit a child to the obese control group was based on the fact that the child did not carry a label of NAFLD and therefore did not experience the same hardships that children diagnosed as having NAFLD are suddenly faced with (ie, the diagnosis and implications of having a chronic illness in addition to being obese or overweight). After being diagnosed, children are required to regularly attend additional medical appointments, undergo blood draws and other procedures, including ultrasound and liver biopsy, and, in certain cases, take medications. Nineteen children of 40 in the control group had no documentation of blood work done, including ALT. Most would agree that a serum ALT alone is not sufficient to make a diagnosis of NAFLD, and a normal ALT certainly does not exclude a child from having the disease. The cutoff for “normal” ALT in the present study is the cutoff of 53 U/L used in the laboratory at our institution. There are published data to show that ULN of ALT is as low as 25.8 U/L in boys and 22.1 U/L in girls (14), in line with ULN used in adults, but this has not gained widespread acceptance in pediatrics yet. Whether the control group had ALT values for review was thought irrelevant in light of the fact that most children with NAFLD at time of diagnosis are asymptomatic and do not have physical stigmata related to liver disease. Unlike adults who have physical problems associated with their liver disease, the children with NAFLD in our cohort did not appear to have any physical effect secondary to NAFLD per se given none had significant liver disease in the form of cirrhosis/portal hypertension. Therefore, it is not the clinical manifestations of NAFLD that directly affect a pediatric patient's QOL. Rather, it is the aftermath of the diagnosis of NAFLD as well as the frequency of medical visits and interventions associated with it that, we postulate, affects the psychosocial well being for these children and adolescents—explaining why the CDI and BESAA scores were worse in the NAFLD cohort.
There have been several studies looking at the QOL both in adults and in children with a diagnosis of NAFLD. In the large multicenter pediatric study of the NASH CRN (4), children with NAFLD reported lower QOL scores than a reference sample of healthy children, with nearly 40% in the NAFLD cohort having impaired QOL, worse scores in girls and in those of non-Hispanic ethnicity. The most common symptoms reported were fatigue, trouble sleeping, and sadness. Possible limitations of the study include its cross-sectional nature and lack of comparison with obese controls (2–4). In the large multicenter adult NASH CRN study, subjects reported a worse physical and mental health score compared with the US population with and without chronic illness (3). QOL was particularly low in individuals with cirrhosis. There were no children with cirrhosis or portal hypertension included in the present study. Factors associated with lower QOL in the adult NAFLD study included older age, female sex, presence of type 2 diabetes mellitus, and poverty. NAFLD is considered to be the hepatic manifestation of the metabolic syndrome and poor QOL reported in adults may be related to comorbidities such as diabetes mellitus, which have a well-established association with compromised mental health (15). Children with NAFLD commonly have insulin resistance and no child in the present study had diabetes mellitus.
NAFLD is the most common cause of elevated liver chemistries in children today both in the United States and in many parts of the world. After the diagnosis is established, standard care involves counseling regarding lifestyle and dietary changes. These children are then monitored at regular intervals in the outpatient clinics where their anthropometry is recorded, results of blood tests ± imaging followed, and counseling continued. The present study has demonstrated that standard care, even though for a short duration, did not improve the psychosocial measures in our pediatric NAFLD cohort when they were followed longitudinally. What is even more concerning is that the 15 of 48 enrolled children with NAFLD, who did not come for a follow-up visit, had significantly higher baseline scores on the CDI total, negative mood, interpersonal relationships, and anhedonia scales as well as all of the BSEAA scales than those who did. Thus, at least in our cohort it appeared that those who may benefit most from intervention stayed away. Having said that, given the fact that standard care did not improve psychosocial outcome in those that did follow-up regularly, the present study highlights the need for earlier and different interventions for this vulnerable group of children. Another unexpected outcome was that children who gained weight were less depressed. It is possible that children who were not forced to diet or exercise and continued to gain weight were less “stressed” because they were not being deprived of foods they enjoyed, nor “forced” to engage in exercise, which they did not want to do. Although speculative, it could also be that these children were responsive to the attention from health professionals, which could have translational implications when considering whether to apply more aggressive interventions.
The limitations of the present study include the relatively low number, which may not have been able to detect small effect sizes compared with the large multicenter studies and the inability to have ALT in all obese control children. The strengths of the present study are that this is the first study to compare more comprehensive psychosocial parameters linked to obesity including QOL not only in NAFLD but also in obese controls without known NAFLD. Additionally, the longitudinal nature of the present study allows us to assess the effect of present standard care management. Our findings need to be confirmed in larger studies, for a longer period of time. Our study reveals that children as young as 12 years have poor psychosocial outcomes that do not change over time with standard care and highlights the impetus for clinicians and mental health professionals to urgently come up with more innovative approaches. With the epidemic of obesity and NAFLD, the psychosocial arena should not be neglected in future research studies.
The authors thank Dr J.W. Varni, who designed the QOL questionnaires that were used in the present study.