Symptom Clusters, Physical Activity, and Quality of Life; A Latent Class Analysis of Children During Maintenance Therapy for Leukemia

Abstract

Background:

Children undergoing treatment for acute lymphocytic leukemia (ALL) report co-occurring symptoms of fatigue, sleep disturbances, and depression as a symptom cluster. Physical activity (PA) may influence symptom severity and quality of life (QOL).

Objectives:

This study examined changes in symptoms and QOL during ALL maintenance in children categorized by symptom cluster and explored the influence of PA and symptoms on QOL.

Methods:

Self-report of fatigue, sleep disturbance, and depression, QOL, and PA were measured at the beginning and end of maintenance in 42 children aged 3 to 18 with ALL. Children were categorized into symptom cluster groups based on measurements at the beginning of maintenance.

Results:

Two latent classes of symptom clusters (low and high) were identified with significant differences between groups in symptoms at both the beginning and end maintenance (p < .01). Each group’s symptom levels did not change during maintenance. QOL was different between groups at both time points (p < .01) and did not improve. Children with low symptoms and high PA at the beginning of maintenance had better QOL as treatment ended compared to the physically active high symptom group and the inactive high symptom group (p < .01).

Conclusions:

Children with higher symptoms did not experience an improvement with time. Symptom and PA levels may influence QOL at the end of treatment.

Implications for Practice:

Maintenance therapy is a long time (1.5 years) in a child’s life. Symptom assessment is needed early in maintenance; interventions are needed for children with high levels.

INTRODUCTION

Advancements in the treatment of childhood acute lymphocytic leukemia (ALL) have resulted in a survival rate of 91% in children aged 14 and younger.¹ This progress has mostly been attributed to the optimization of risk-stratified chemotherapy regimens.¹ However, this chemotherapy also results in distressing symptoms that impact the child or adolescent’s quality of life (QOL) and interferes with normal activities that are important in ongoing development.² Fatigue, sleep disturbance, and sadness/depression are symptoms that are prevalent in children undergoing treatment for cancer and are known to contribute to symptom distress.^{2, 3}

Researchers have found differing patterns of change in symptoms over the trajectory of leukemia treatment. During the first year of treatment, fatigue measured by self-report decreases but its level can still be burdensome.^{4, 5, 6} Maintenance therapy, also called continuation therapy, is the final phase of treatment for ALL and is administered over two to three years.⁷ However, little is known about the prevalence and change in fatigue over the course of maintenance therapy.

Sleep disturbances are prevalent in children during ALL maintenance therapy.² More children in ALL maintenance have sleep scores that are above the clinical cutoff for diagnostic sleep disturbance compared to their healthy peers.^{8, 9} Parents have reported challenges with their child sleeping during maintenance therapy and some attribute the difficulties to corticosteroid medications.¹⁰ Sleep efficiency measured by actigraphy in children receiving maintenance therapy was found to be below acceptable levels during the time periods before and during a dexamethasone pulse.¹¹

Children with cancer experience sadness as a normal response to both psychological and physical pain associated with the cancer experience.³ Children with ALL have been identified as at risk for depression as they complete maintenance therapy.¹² Depressive symptoms have been associated with fatigue and sleep disturbances in adolescents and fatigue in school age children during cancer treatment.¹³

Symptom Clusters

Symptoms in children and adults with cancer rarely occur as single experiences. Two or more symptoms that co-occur and are related to each other are defined as a symptom cluster.¹⁴ Researchers have used multiple approaches to studying symptom clusters. When using an empirical approach, the researcher analyzes data from a multi-symptom measurement to identify a group of symptoms from a larger pool of symptoms; using an a priori approach, the researcher selects symptoms for the cluster before collecting the patient data.¹⁵

Most research on symptom clusters in the pediatric oncology population has included samples of mixed diagnoses.³ Using a multi-symptom inventory among 144 children and adolescents with cancer, Yeh and colleagues¹⁶ used an empirical approach to identify five separate clusters, one of which was fatigue, sleep disturbance, and depression. When selected a priori as potential symptoms that may cluster, fatigue, sleep disturbance, and depressive symptoms were shown to cluster in a sample of adolescents with cancer (n = 32), while children (n = 35) exhibited co-occurrence of fatigue and depressive symptoms.¹⁷

Multiple statistical methods can be used to analyze symptom clusters. One approach is to examine relationships between variables (variable-oriented); another is a person-oriented approach in which subgroups of people who have similar patterns of characteristics are identified.¹⁸ An example of a person-centered approach is the use of latent class analysis (LCA); LCA uses a categorical approach to classify people into groups whose symptom experience is similar.¹⁸ In our earlier study of 236 children in their first year of leukemia treatment, we measured fatigue, sleep disturbances, pain, nausea, and depression at four time points during the first year of ALL therapy. Three latent classes of children were identified; these groups had either mild, moderate, or severe symptom trajectories.²¹

Health Related Quality of Life.

During treatment for ALL, children experience a lower QOL compared to their healthy peers.^{19, 20} Although their QOL improves over the trajectory of treatment as chemotherapy becomes less intensive, some children experience impairments that persist in the months after treatment.^{19, 20} Cancer symptom severity negatively impacts a child’s QOL. In our earlier symptom clusters study of children undergoing leukemia treatment, those children with more severe symptoms over the first year of treatment had significantly lower QOL at the beginning of maintenance therapy.²²

Physical Activity

Physical activity involves any movement of the skeletal muscles; from a developmental perspective, physical activity is typically seen in play behaviors of children.²³ PA is important as it supports ongoing development and even small changes can positively impact health.²⁴ During the first year of ALL treatment children become significantly more inactive than their healthy peers.²⁵ Physical activity, however, can positively impact symptom distress and QOL in children during cancer treatment. Increased physical activity is associated with better sleep quality in children with cancer.²⁶ Children receiving ALL maintenance therapy who were more active had less fatigue during their steroid pulse.²⁷ Rehabilitation programs that increase physical activity have an immediate and sustainable positive impact on health-related QOL in children with cancer.²⁸

Purpose

Using a latent class analysis, we categorized children into symptom cluster groups based on measurements of fatigue, sleep disturbance, and depression taken at the beginning of maintenance therapy. The purpose of this study was to examine differences between the symptom cluster groups and explore changes within each symptom group in both individual symptoms and QOL over the trajectory of maintenance ALL therapy. We also evaluated if levels of physical activity at the beginning of maintenance therapy influenced symptoms and QOL at the end of maintenance.

METHODS

Design and Sample

This single group, repeated measures study was a follow up investigation to a larger study that evaluated symptoms clusters and functional outcomes in children during the intensive phases of treatment for newly diagnosed leukemia.²¹ This smaller companion study utilized the last set of measurements from a cohort of participants in the primary study; these measurements were performed at the beginning of maintenance therapy. A second set of measurements was then repeated at the last cycle of maintenance therapy. Children and families were invited to participate if they had completed the earlier study, were between the ages of 3 and 18 years when starting the primary study, were fluent English or Spanish, had not received cranio-spinal radiation, and were approaching the last cycle their maintenance chemotherapy. The two study sites were major cancer treatment centers in the Upper Midwestern and Southwestern regions of the United States. Children aged 7 to 17 provided assent to participate in the study and parents provided written consent. Children 18 years of age provided consent. The Institutional Review Board at each institutional site approved the study before implementation.

Study Measurements

Self-report measurements were completed on a data secure iPad; participants were directed to consider their symptoms and function during the past week. In the primary study, parents served as proxy reporters if their child was age 6 or younger. For consistency in reporting, if the parent reported in the first study, they continued in that proxy role in this follow up study. Measurements were completed at an outpatient clinic visit during the last cycle of maintenance therapy.

Fatigue.

Fatigue-related symptoms were measured using one of three validated fatigue scales. The 10-item Childhood Fatigue Scale (CFS) was used for children aged 7 to 12 when starting the primary study. Children rated how much they were bothered by fatigue on a 4-point Likert scale ranging from “Not at all” to “A lot” (Cronbach α = .76).²⁹ Adolescents aged 13 and older when starting the primary study completed the 13-item Adolescent Fatigue Scale (AFS) (Cronbach α = .87).³⁰ The 17-item Parent Fatigue Scale (PFS) was used to obtain proxy responses from parents for young children who were 6 years or younger when starting the primary study (Cronbach α = .88).³¹ Fatigue scores across the three groups were converted to a T-score of total fatigue score with a range of 20–80; a higher score indicating more severe fatigue.

Sleep Disturbance.

Sleep was measured using the Adolescent Sleep Wake Scale (ASWS) for subjects aged 13 to 18³² when starting the primary study and the Child Sleep Wake Scale (CSWS) for subjects aged 3 to 12 years when starting the primary study.³³ Both instruments included 5 sub-scales including going to bed, falling asleep, maintaining sleep, going back to sleep and returning to wakefulness; responses were scored on a Likert scale ranging from 1 to 6. Scores were then averaged across subscales. Cronbach α for the CSWS was .89 and for the ASWS was .86.^{32, 33} For this study, scores were reversed so that a higher score indicated a more severe symptom which was consistent with the other symptom measures.

Depressive Symptoms.

Measurements for depressive symptoms were completed with two separate scales; one was used at the beginning of maintenance and another at the end of maintenance. Scores were converted to a T-score of total score with a mean of 50, range of 20– 80, a standard deviation (SD) of 10, and a higher score indicating more severe depression. At the beginning of maintenance, The Child Depression Inventory (CDI-2)³⁴ was used as a self-report measure for all age groups. The instrument’s 27 questions have three response choices indicating the level of the symptom: 0 (absence of the symptom), 1 (mild symptom), or 2 (definite symptom).³⁴ The scale has established reliability and validity and requires a low level of reading. Depressive symptoms at the end of maintenance was measured using a shorter instrument, the 8-item PROMIS^® Pediatric Short Form for Depression, which is part of the National Institute of Health’s Patient Reported Outcome Measurement Information System (PROMIS^®) initiative. At the end of maintenance the measurement of depressive symptoms was changed to this PROMIS^® measure to decrease the response burden for children and parents³⁵; it has been used in children with cancer.³⁶ The child version was used for the child and adolescent participants and parent report version was used if the child was age 6 or younger when enrolled in the primary study.

Physical Activity.

The Godin-Leisure-Time Exercise Questionnaire (GLTEQ) is a self-report measurement and was used to assess the child or adolescent’s physical activity level. The participant recalls how many times on average they participated in strenuous, moderate, or mild exercise for more than 15 minutes during the previous week.^{37, 38} Each of the intensities includes descriptions and examples of activities. Scoring is calculated by multiplying each frequency by its metabolic measurement unit as: (3 × mild) + (5 × moderate) + (9 × strenuous) with a higher score indicates higher levels of physical activity.³⁷ The GLTEQ has demonstrated reliability and construct validity in children,³⁹ has a low response burden, and has been used in studies of pediatric cancer patients.⁴⁰ People with a score of 24 or higher are considered active.³⁸

Quality of Life.

The Peds Quality of Life Cancer Module (PedsQL CM) was designed to measure pediatric cancer-specific health-related QOL from the patient (7 and older) and parent perspectives for children who were age 6 years or younger when starting the primary study.⁴¹ The PedsQL CM has established reliability (Cronbach α = 0.72 child; α = 0.87 parent report) and is widely used in pediatric cancer research. The 27-item scale has 8 subscale domains specific to pediatric cancer; the scale uses a Likert response. Item scores are averaged for the total score; scores can range from 0 to 100 with a higher score indicating better QOL.⁴¹

Data Analysis

Latent class analysis (LCA) using SAS Proc LCA⁴² developed by The Methodology Center at Penn State⁴³ was used to create the symptom cluster model based on depression, fatigues and sleep scales at the beginning of maintenance. To determine the number of latent groups, the following model-fit statistics were used: log-likelihood, Bayesian information criterion (BIC), adjusted Bayesian information criterion (adjBIC), and entropy.

Changes in fatigue, sleep, and depression between beginning of maintenance and the end of maintenance by LCA groups were tested using t-tests. The change in QOL during maintenance was evaluated using a t-test. Differences in individual symptoms and in QOL between the LCA groups were also compared using a t-test. Physical activity measured at the beginning of maintenance therapy using GLTEQ was classified as low activity and normal activity, using scores of <24 and ≥24 respectively. Differences in fatigue, depression and sleep at the end of maintenance were also tested between low and normal activity PA groups using t-tests. PA and symptom cluster group memberships at the beginning of maintenance were evaluated together using ANCOVA to test for differences in QOL at the end of treatment.

RESULTS

Latent Class Analysis Groups

Three symptom scores (depression, fatigue, and sleep) at the beginning of maintenance were modeled using LCA. Models with 2 and 3 groups were evaluated. The fit statistics for 2 groups were: log-likelihood = −130.5, Bayesian information criterion (BIC) = 136.5, adjusted Bayesian information criterion (adjBIC) = 45.7, and entropy = 1.0 (indicating complete separation on groups) and for 3 groups: log-likelihood = −126.9, BIC = 185.3, adjBIC = 47.6, and entropy = 0.95 for comparison. The 2-group model was selected as superior to the 3-group model; the two groups in the model were labeled as low symptom and high symptom.

Demographics

A descriptive summary for demographics is presented in Table 1. Females and males were evenly distributed in the study sample Young children less than age 7 years and school-age children made up the largest age groups with only 3 adolescents participating. Thirty-eight percent of the sample was of Hispanic ethnicity with the larger of the two study sites located in the Southwest. The largest proportions of participants were in the average/standard risk and high-risk groups with low-risk and very-high risk patients. In the sample as a whole, the time between the measurements in the first and last cycles of maintenance was 89 weeks. As seen in Table 1, there was not a significant difference between the low symptom and high symptom groups in the breakdown of demographic characteristics.

Table 1.

Demographic Characteristics

Demographic Characteristics	Whole Group (n = 42) N (%)	Low Symptom Group (n = 34) N (%)	High Symptom Group (n = 8) N (%)	P
Gender				0.99
Male	22 (52)	18 (53)	4 (50)
Female	20 (48)	16 (47)	4 (50)
Age group				0.59
Young child (3–6 years old)	20 (48)	17 (50)	3 (37)
Child (7–12 years old)	19 (45)	15 (44)	4 50)
Adolescent (13–18 years old)	3 (7)	2 (6)	1 (13)
Race/ Ethnicity				0.51
Hispanic	16 (38)	13 (38)	3 (38)
Non-Hispanic white	15 (36)	13 (38)	2 (25)
Non-Hispanic black	3 (7)	3 (9)	0 (0)
Non-Hispanic other	8 (19)	5 (15)	3 (38)
Leukemia risk group				0.23
Low	5 (12)	4 (12)	1 (13)
Average/standard	15 (36)	14 (41)	1 (13)
High	14 (33)	9 (26)	5 (63)
Very high	8 (19)	7 (21)	1 (13)
	Mean (SD)	Mean (SD)	Mean (SD)	P
Age (years)	7.05 (3.89)	6.88 (3.86)	7.75 (4.20)	0.58
Time (weeks) between measurements in first and last cycles of maintenance	89.17 (28.86)	86.68 (26.92)	99.75 (36.12)	0.25

Symptoms During Maintenance Therapy

Individual symptoms scores for each LCA group (low and high symptoms) can be seen in Table 2. Between the two LCA groups at the beginning of maintenance, a significant difference was found in all three symptom scores consistent with their low or high group assignment. This difference continued to be evident at the end of leukemia therapy. When examining changes within each group, the low symptom group continued to have a low symptom levels without significant change during maintenance therapy. For the high symptom group, symptoms scores did not decrease significantly over the course of maintenance indicating that their symptom levels did not improve with time.

Table 2.

Differences Within LCA Groups and Between LCA Groups on Individual Scores on Symptoms, Quality of life, and Physical Activity

	Beginning of Maintenance		End of Maintenance		Differences Within LCA Group
Variable	M	SD	M	SD	P
LCA Low symptom group (n = 34)
Depression T score	40.6	6.3	43.2	9.5	.10
Fatigue T score	45.2	6.8	45.5	7.6	.87
Sleep Disturbance score	2.6	0.6	2.5	0.8	.92
QOL score	77.3	13.1	80.3	13.4	.16
Physical activity (GLTEQ)	37.3	21.0	55.8	38.3	.23
LCA High symptom group (n = 8)
Depression T score	51.3	12.8	54.2	10.2	.54
Fatigue T score	63.9	3.7	63.8	11.1	.98
Sleep Disturbance score	3.8	0.4	3.6	1.0	.38
QOL score	61.5	10.2	65.3	7.8	.44
Physical activity (GLTEQ)	27.7	27.0	33.8	15.5	.89
Differences Between LCA Groups	P		P
Depression T score	< .01		< .01
Fatigue T score	< .01		< .01
Sleep Disturbance score	< .01		< .01
QOL score	< .01		< .01
Physical activity (GLTEQ)	.31		.18

Abbreviations: LCA, latent class analysis; GLTEQ, Godin-Leisure-Time Exercise Questionnaire ; QOL, quality of life

Quality of Life (QOL)

QOL was significantly different between the low and high symptom groups at both time points (P < .01 for both) (Table 2). QOL scores improved within each symptom cluster group during the course of maintenance but the change was not significant in either group.

Physical Activity and Individual Symptoms

Self-report of physical activity measured by the GLTEQ showed no significant differences between the low and high symptom groups (Table 2). Physical activity scores on the GLTEQ in the low symptom group increased from 37.3 to 55.8 by the end of maintenance but the change was not significant; the high symptom group experienced little change in physical activity scores during maintenance therapy.

For the combined group, children who were categorized in the normal activity level group at the beginning of maintenance had significantly better sleep than the low activity level group at the end of maintenance (M = 2.62 and SD = 0.71; M = 3.01 and SD = 0.84 respectively; t = 2.3, P = .03). However, physical activity group membership at the beginning of maintenance did not influence levels fatigue or depression at the end of therapy.

Symptom Cluster, Physical Activity, and Quality of Life (QOL)

The influence of the combined membership, at the beginning of maintenance, in both a symptom cluster group (low or high) and a physical activity group (low activity or normal activity) on QOL was explored. Children in the low symptom group who were also in the normally active group at the beginning of maintenance had significantly better QOL at the end of treatment compared to the normally active, high symptom group and the inactive high symptom group (P < .01). The low symptom, active group trended towards a better QOL (P =.06) compared to the low symptom, inactive group. QOL scores for each group are seen in Table 3.

Table 3.

QOL Scores at End of Maintenance by LCA Symptom Group and Activity Group

	QOL Scores End of Maintenance
Group by Activity Level and Symptoms at Beginning of Maintenance	M	SD
Normally active – low symptom group (n = 21)	82.61	10.83
Normally active – high symptom group (n = 3)	66.70	7.13
Inactive – low symptom group (n = 9)	76.96	5.89
Inactive – high symptom group (n = 4)	64.71	11.20

Abbreviations: LCA, latent class analysis; QOL, quality of life

DISCUSSION

The latent class analysis approach to symptom cluster analysis supported the identification of groups of patients with similar symptom experiences rather than just groups of symptoms. The significant differences in symptom levels found between the low and high symptom groups was expected at the beginning of maintenance as this difference was what created the two groups. This difference between groups remained at the end of maintenance as each group’s symptom levels remained constant. Symptoms in the low group remained low, but the high symptom group did not experience a decrease in symptoms over their 1.91 years of maintenance therapy.

Maintenance therapy is a time that health care providers expect children and adolescents return to activities such as school and sports that are normal for their developmental stage.⁴¹ Health care providers may assume that disruptive and distressing symptoms will resolve on their own when therapy is less intensive. However, for children and adolescents with high symptoms, this does not appear to be the case. Study participants who were in the high symptom group continued to have persistent fatigue that was greater than 60, indicating a fatigue level higher than a standard deviation of 10 from a normative score of 50. While sleep scores for the low symptom group were similar to the mean score of 2.7 (reverse scored) of a community control group of children³³; sleep scores for the high symptom group were higher than the community control group indicating worse sleep which did not improve during maintenance therapy. This is consistent with the findings by Hinds and colleagues¹¹ who found that sleep efficiency was below healthy norms in children with ALL measured with actigraphy during a 5-day period before a maintenance dexamethasone pulse. Depressive symptom scores, although different between the low and high symptom groups, remained within 1 SD of a normative score of 50 indicating normal levels on average.

Children in the high symptom group had significantly lower QOL than children in the low symptom group. This finding was a continuation of the pattern in our earlier study with the larger sample of children that demonstrated that children participants with more severe symptoms over the first year of treatment had significantly lower QOL at the beginning of maintenance.²²

The mean GLTEQ score for physical activity for both symptom groups was greater than 24 indicating an active level; however the low symptom group had a non-significant trend of increasing physical activity during maintenance while the high symptom group’s physical activity levels remained static. Perhaps the high levels of symptoms interfered with their ability to participate in physical activity and while children who have fewer symptoms are better able to advance their physical activity over the course of chemotherapy. When the study sample was re-categorized into active and low active groups, the benefits of increased physical activity on sleep was evident with the active group having better sleep scores. The influence of the symptom cluster and physical activity on QOL were also explored together; children in the low symptom/normal activity group had significantly better QOL at the end of maintenance compared to the two high symptom groups (normal active and inactive). However, the two high symptom groups were small in size and so this outcome must be viewed with caution and further study is needed in a larger sample.

The sample size of 42 children limited further analysis of variables such as gender and risk group that may have influenced the symptom experience. However, in our earlier large study of symptoms during the first year of ALL therapy, the only demographic variable that influenced symptom severity was race/ethnicity; children of Hispanic ethnicity were less likely to report more severe symptoms.¹⁹ An additional limitation was the use of two different instruments for measuring depressive symptoms. In our earlier study, adolescents made up 19.5% of the total sample while in this follow up companion study, our sample only had 7% adolescent participants. When recruiting for this study which occurred at the end of therapy, we found that most adolescents were ready to “be done” with their ALL treatment and strongly communicated their lack of assent or consent.

These study findings recognize the inter-relationship of symptoms in children during ALL maintenance therapy. It highlights the need to carefully assess symptoms in children as they move into maintenance therapy and to not assume that children with high levels of distressing symptoms will have them resolve on their own with time. Moving forward, additional research is needed for developing and testing interventions for children with high symptom levels. Previous supportive care intervention studies on children with leukemia receiving maintenance chemotherapy have focused on physical activity and exercise in relation to cardiovascular fitness,⁴⁵motor function,⁴⁶ bone density,⁴⁷ and quality of life.⁴⁷ Another pilot study used a physical activity to target the symptom of fatigue in children during maintenance therapy but the intervention only occurred over a single cycle of chemotherapy.²⁷ In their pilot study, researcher evaluated a sleep hygiene and relaxation intervention for improving sleep and fatigue children during a single cycle of maintenance ALL therapy; the intervention was feasible and acceptable.⁴⁸ These pilot studies highlight potential interventions that merit further study in a larger, multi-site randomized controlled trial that considers children with high levels of symptoms. Maintenance therapy is lengthy and is a large portion of a child or adolescent’s life; recognizing and intervening in symptoms is important so that they are able to participate in the activities of childhood that are vital for their ongoing development.