Can Steps per Day Reflect Symptoms in Children and Adolescents Undergoing Cancer Treatment?

Abstract

Background:

Multiple symptoms occur in children receiving cancer therapy. Decreased steps-per-day may be associated with burdensome symptoms.

Objective:

To evaluate associations between self-reported symptoms (pain interference, anxiety, depressive symptoms, psychological stress, and fatigue) and function (physical function-mobility and physical activity) and cumulative symptom count with steps-per-day.

Methods:

Five sites enrolled English-speaking children, 8–17 years, receiving treatment for a first cancer diagnosis. Patient-reported outcome (PRO) surveys were administered before (T1) and after (T2) a course of chemotherapy. Garmin VivoFit ^® 3 accelerometers were worn 7 days prior to each data point. Univariate change in scores over time were evaluated with dependent sample t-tests. Pearson correlations examined associations between PRO domains and step count. Multivariable mixed effect models examined associations between steps and PROs.

Results:

Participants’ (n=65) steps-per-day decreased during treatment (4099 [T1] and 3135 [T2]; p < .01) with larger reductions observed during hospitalization and in younger children compared to adolescents. Steps significantly correlated with Patient-Reported Outcome Measurement Information System (PROMIS) Pediatric Physical Activity and Physical Function-Mobility. Decreased steps-per-day were associated with increased fatigue and cumulative symptom count.

Conclusions:

In children and adolescents with cancer, steps-per-day can serve as an indicator of fatigue, cumulative symptom count, physical activity and physical functioning-mobility.

Implications for Practice:

Child self-reports of physical activity and physical function are valid during cancer therapy and should be captured. In the absence of self-report, decreasing step count may prompt additional assessments related to fatigue or cumulative symptom count and trigger early interventions to support physical activity and physical function-mobility.

Introduction

Children and adolescents undergoing cancer treatment report multiple, concurrent symptoms such as fatigue, nausea, pain, sleep disturbance and anxiety.^1,2 Many of these symptoms are subjective, often under-reported by clinicians, and are best understood through child self-report.³ When a child is too young, too sick, or otherwise unable to provide self-report, proxy reports may be used.⁴ Although considered an acceptable substitute when needed, proxy reports are not direct replacements for the child’s own report.⁵ Additionally, we know that children may not self-disclose symptoms unless directly asked about them.³ Identifying other metrics that correlate with a child’s symptom experience holds value as this may alert clinicians and parents to ask additional questions regarding symptom burden.

In children less than 16 years of age, the Lansky Play-Performance Scale (LPPS) is commonly used to monitor physical function during cancer therapy.⁶ A recent study demonstrated that clinician and parent proxy report of LPPS scores often differ from a child’s own report of physical function, yet moderate, positive correlations were present between parent/clinician LPPS scores and child self-reported measures for fatigue and pain interference.⁷ Other studies have found that a child’s physical function is decreased in the presence of symptoms.^8,9 It is unknown if other objective measures, such as steps-per-day, can be useful indicators of a child’s symptoms and functioning during therapy.

National guidelines recommend 60 minutes of moderate-to-vigorous activity per day in children ages 6–17 years,¹⁰ to promote health (bone strength, muscle development and overall physical fitness); decrease risks for chronic diseases such as obesity, cardiovascular disease, and diabetes; and promote normal growth and development during childhood.¹¹ Preliminary research equates sixty minutes of daily activity to 10,000 to 11,700 steps-per-day for adolescents, 13,000–15,000 steps-per-day in elementary age boys, and 11,000–12,000 steps-per-day in elementary age girls.¹² Although there are recommendations for adult oncology patients related to physical activity during therapy and after therapy,¹³ guidance during childhood cancer is more generic advocating for children to “move more” while noting the safety of engaging in physical activity and/or exercise during therapy.¹⁴ In the absence of pediatric oncology specific guidance, activity levels are currently benchmarked against recommendations for the healthy pediatric/adolescent population. As such, children with cancer often fall very short of meeting these recommendations, with well-documented decreases in physical activity during and after treatment.^8,15,16 Decreases are hypothesized to occur secondary to muscle atrophy, decreased muscle strength, obesity, decreased ankle dorsiflexion, and/or anthracycline-related cardio toxicities associated with cancer therapy.¹⁷ Decreased physical activity and physical function-mobility have also been reported to vary with the treatment setting (in-patient vs. out-patient).^9,15

Tracking steps-per-day in oncology has been proposed as an objective way to capture unbiased reports of activity which can provide another indicator of patient health.¹⁸ Currently, it is unknown to what degree steps-per-day may correlate with pediatric self-reports of symptoms and functioning during therapy. The purpose of this study was to evaluate associations between pediatric self-reported outcomes during cancer therapy and accelerometer-based step data by 1) examining correlations between steps-per-day and patient-reported outcomes (PROs; pain interference, anxiety, depressive symptoms, psychological stress, fatigue, physical function-mobility, and physical activity) and 2) exploring associations between steps and cumulative symptom count (number of co-occurring symptoms). It was hypothesized that decreased steps-per-day would be associated with higher symptom burden. This work is based upon the conceptualization that symptoms and physical function impact a person’s overall quality of life. As such, the study was guided by the Conceptual Model of Health-Related Quality of Life.¹⁹

Methods

This study was a sub-study of a larger project to validate PRO measures in children with cancer and other diseases.^20,21 Five sites participated in the sub-study after receiving Institutional Review Board approval: Children’s Healthcare of Atlanta/Emory University, St. Jude Children’s Research Hospital, University of North Carolina at Chapel Hill, Duke University, and UPMC Children’s Hospital of Pittsburgh.

PARTICIPANTS

Eligibility for participation included children ages 8–17 years, with a first cancer diagnosis, receiving active treatment (chemotherapy, radiation or bone marrow transplant). Enrollment generally occurred during the first six months of cancer therapy, but at least 4 weeks after diagnosis and at least 3+ weeks post-cancer definitive surgery (if applicable). Additionally, children/adolescents needed to be English speaking, willing to wear the accelerometer for two one-week periods and cognitively able (per the clinician’s judgement) to independently complete surveys. Parental consent and child assent (per each facility’s policies) were obtained for all participants.

DATA COLLECTION

Data collection occurred from July 2017 through October 2018. Data were collected at two time points, during regularly scheduled health care visits and in connection with the larger study.²⁰ Survey data collection for time point 1 (T1) occurred within the 72 hours prior to beginning a treatment cycle, when relatively low symptom burden was anticipated. Time point 2 (T2) occurred 7–17 days after T1 (for those receiving chemotherapy or bone marrow transplant) or 4+ weeks after T1 (for those receiving radiation), when a higher symptom burden was expected. Secondary to variances in the literature regarding symptom trajectories during childhood cancer treatment, T2 was selected to coincide with anticipated nadir in bone marrow suppression after beginning a cycle of chemotherapy.^22,23 Participants were provided Garmin VivoFit^® 3 monitors and instructed to wear for 7 days prior to each survey time point (T1 and T2). Recruitment occurred around pre-determined cycles, by cancer diagnosis, and coincided with a cycle of high intensity treatment. As such, the time from diagnosis to enrollment varied among participants but generally occurred early in therapy, normally within the first 3–6 months of starting treatment.

For data collection, study personnel provided tablets during in-person visits for participants to electronically complete the questionnaires (paper-backup available). Electronic links to the questionnaires were emailed to families if no face-to-face healthcare visit was scheduled during the T2 window timeframe. At each study visit, data from the Garmin VivoFit^® 3 monitor were downloaded from the device. If the families did not have a regularly scheduled visit within the T2 timeframe, then participants mailed back their monitor to the study staff or brought it to their next scheduled visit. Submitted questionnaires and physical activity data were securely housed in PRO Core, a survey system developed by the University of North Carolina Lineberger Comprehensive Cancer Center.

Measures

ECOLOGICAL SURVEY

A 9-question ecological survey was developed, based upon available evidence in the literature, to identify external factors that may have influenced the collected step data.^24,25 Questions pertained to participation in sports, weather conditions interfering with activity, parental/peer encouragement, parental/peer participating in physical activities with child, availability of sports equipment, safety of walking in neighborhood, and access to playgrounds, parks or gyms. Five response options were available for selection, ranging from “Never” to “Almost Always”. An additional question asked about overnight hospital stays in the past 7 days and requested a “Yes/No” response. The descriptive ecological survey was reviewed by 3 content area experts and determined to have high face validity. The survey was completed by the child, with parental assistance as needed, at each time point. The child was asked to provide a reason for any non-wear time of the monitor. The recall period was “In the past 7 days”.

PATIENT-REPORTED OUTCOME MEASUREMENT INORMATION SYSTEM^® (PROMIS^®)

PROMIS pediatric instruments were administered electronically (through tablets or electronic links) via the PRO Core platform. Written, 8-item, domain-specific, short forms were used for backup data collection (i.e. internet outage). Assessed domains included: anxiety, depressive symptoms, fatigue, pain interference, physical activity, physical function-mobility, and psychological stress. PROMIS Pediatric measures include a 7-day recall period and have been validated for use in children and adolescents with cancer.^9,20,26,27 The T-score metric scale was utilized to standardize scoring of the PROMIS measures with a mean of 50 (standard deviation 10). For symptom-based measures, such as anxiety or fatigue, higher scores represent worse symptom experiences. Higher scores in domains related to physical activity and physical function-mobility reflect better functioning. Prior research has identified a minimally important difference (MID) of 3 points for many of the PROMIS Pediatric measures.²⁸ Changes in MID of 3 or more points are therefore considered clinically meaningful.

PEDIATRIC PRO-CTCAE MEASUREMENT SYSTEM

The Pediatric Patient-Reported Outcomes version of the Common Terminology Criteria for Adverse Events (Ped-PRO-CTCAE) measurement system was used to assess co-occurring symptoms at T1 and T2 as a measure of cumulative symptom count. The Ped-PRO-CTCAE includes a library of items for children with cancer to report up to 62 symptoms. The Ped-PRO-CTCAE is valid and reliable for use in children with cancer.^3,29–32 The Ped-PRO-CTCAE includes questions on symptom presence, frequency, severity and/or interference with daily activities, using a 7-day recall period. Participants answered questions related to 15 “core” symptoms that commonly occur across all disease groups and cancer treatment protocols. Core symptoms included abdominal pain, anorexia, anxiety, constipation, cough, depression, diarrhea, fatigue, headache, insomnia, oral mucositis, nausea, pain, peripheral sensory neuropathy, and vomiting.³¹ These 15 symptoms were scored dichotomously (1 if the symptom was reported, or 0 if the symptom was absent) and summed to represent individual symptom count (range 0–15).

ACCELEROMETER

The Garmin VivoFit^® 3 is a commercially available, tri-axial, accelerometer with reliable and valid data related to step count.^33,34 This monitor was selected for its ease of use, long battery life (one year), robust memory (30 days), water resistance, small size of the monitor unit, availability of both adult/adolescent and child-sized wrist bands, and for its low cost (< $90 per monitor). Based on available evidence for best practices in handling accelerometer output,³⁵ data were included if available for at least 4 days during a defined 7-day period prior to each PRO assessment point (to estimate steps for the full week prior to T1 and T2). Eligible days were defined as those with a minimum of 10 hours of wear time between the hours of 6am and 10pm. Wear time was determined by the study team through examining inactivity (no recorded movement) in each 15-minute time interval during wear. As there was a high likelihood that the missing data did not occur at random (i.e. may have occurred with increased frequency during in-patient hospitalizations), the a priori decision was made not to impute missing step data. The algorithm for calculating activity intensity via the Garmin monitors is proprietary information and not readily available to researchers which limits the ability to compare findings to other published data related to activity intensity. Additionally, at the time of concept development, no publications were located to validate the accuracy of intensity level activity measurements for the selected monitor. For these reasons, the decision was made to only utilize step counts in the analysis.

Statistical Analyses

Descriptive statistics were used to summarize participant characteristics. For descriptive analyses, age was categorized as children (8–12 years) and adolescents (13–17 years). Univariate change in scores over time were evaluated with dependent sample t-tests. Pearson correlations were used to examine the association between PROMIS Pediatric domains and step count. We used multivariable mixed effect models to examine the association of PROMIS domains or Ped-PRO-CTCAE symptom AE count (total binary count of symptom presence or not for 15 core symptoms) with number of average daily steps-per-week as the outcome. A separate model was used for each PROMIS domain (centered in the regression model at 50) and for Ped-PRO-CTCAE symptom AE count, and a random intercept was included in the model accounting for within-patient data dependency. Models were adjusted for time period (reference: week 1), child age (centered on 8 years old), gender (reference: female), stayed at least 1 night in hospital in past week (reference: no nights), cancer type (reference: leukemia or lymphoma), race (reference: white) and parent education (reference: college degree). To examine the responsiveness of PROMIS measures compared to step counts, we calculated effect sizes (mean change from T1 to T2, divided by the SD of the change) for each measure. One participant had greater than 15,000 steps-per-day at T1. No explanation was available to explain the step count, which may have been accurate. Out of caution and concern that the data point would skew results in our small sample size, we removed the child’s data for T1. Significant associations were based on an alpha level of .05. Analyses were conducted with SAS software, Version 9.4 of the SAS System for UNIX, copyright © 2016 SAS Institute Inc., Cary, NC, USA.

Results

Of the 75 children and adolescents enrolled in the study, 6 withdrew before participation began and 4 had accelerometer malfunctions leading to missing step data. Of the remaining 65 participants, the majority self-identified as white (61%), female (53%), and had leukemia/lymphoma (58%) (Table 1). At T1, 53 children had accelerometer data meeting pre-set criteria for inclusion in analyses and patient-reported outcome data; at T2, 51 children and adolescents were included based on a priori criteria. Because some participants only had data at one time point, the number of participants with data at both time points was 46.

Table 1.

Participant Characteristics (N= 65)

Variable	Frequency (%)
Child age (years)
Mean (SD)	13.4 (2.8)
Gender
Male	30 (46.9)
Female	34 (53.1)
Race
White	39 (61.9)
Black	18 (28.5)
Asian	4 (6.3)
Other	2 (3.1)
Cancer Type
Leukemia/Lymphoma	38 (58.4)
Solid Tumor	16 (24.6)
Neuro-oncology	11 (16.9)
Cancer Treatment
Chemotherapy	61 (93.8)
Radiation	4 (6.2)
Educational Level of Caregiver
High School or lower	14 (22.2)
Some College	17 (26.9)
College Degree	25 (39.6)
Postgraduate Degree	7 (11.1)
Overnight Hospitalization
Within 7 days prior to T1	15 (24.1)
Within 7 days prior to T2	23 (39.6)
Average Accelerometer Wear Time
Hours/day (up to 16) prior to T1 (SD)	14.8 (.7)
Hours/day (up to 16) prior to T2 (SD)	14.8 (.7)

All percentages do not equal 100 secondary to missing data.

Abbreviations: SD, standard deviation; T1, time point 1; T2, time point 2.

STEP DATA

Average steps-per-day were 4099 at T1 and 3135 at T2 (Table 2). For children with complete data at both T1 and T2 (n=46), their average steps-per-day reduced significantly (p< .01) by 1040 steps between time points. The standardized effect size for change in average steps-per-day from T1 to T2 was −.79; considered as “large” by Cohen.^36,37 No significant difference in step counts were observed by sex. Decrease in steps over time was much larger for children (1280 steps-per-day) than adolescents (885 steps-per-day; p<.01). Steps were lower at both T1 (p<.01) and T2 (p=.01) in those who reported at least 1 overnight hospital stay during the data collection time frames compared to children who were not hospitalized.

Table 2.

Average Daily Steps by Age Group and Time Point

	Time Point 1		Time Point 2
Age (years)	n	Average steps/day (SD)	n	Average steps/day (SD)
8–12	19	5033 (2275)	20	3443 (2166)
13–17	34	3578 (1661)	31	2936 (1738)
Total	53	4099 (2009)	51	3135 (1912)

Abbreviations: SD, Standard Deviation.

ECOLOGICAL SURVEY

Participants somewhat or strongly agreed that sports equipment (e.g., balls, bicycles) was available for use at home (70%) and that playgrounds, parks or gyms were easily accessible (66%). Other responses included: “often” or “almost always” receiving encouragement to engage in physical activity (19%); participating in organized sports (e.g., basketball, gym class) for at least 4 days within the past 7 days (7%); and feeling that it was safe to walk or jog alone in their neighborhood during the day (61%). Only 9% reported that weather (e.g., rain, cold, heat) interfered “quite a bit” or “very much” with their ability to be physically active during the study time points. Participants self-reported overnight hospitalizations of 24.1% (T1) and 39.6% (T2) during the weeks in which the monitors were worn. Participants reported “forgot to wear” as the top reason for non-wear time of accelerometers.

STEPS AND PATIENT-REPORTED OUTCOMES

Table 3 displays PROMIS Pediatric mean scores at both assessment points, stratified by age group, for participants. Overall, mean scores for the symptom domains (anxiety, depressive symptoms, pain interference, fatigue, and psychological stress) ranged from 42.2 to 50.6 at T1 and remained relatively unchanged at T2. PROMIS Pediatric Physical Function-Mobility scores were similar at both time points as well.

Table 3.

PROMIS Pediatric Mean Scores (SD) of Participants by Age Group and Time

	Time Point 1 (n=53)			Time Point 2 (n=51)
PROMIS Domain mean (SD)	8–12 Yrs (n=19)	13–17 Yrs (n=34)	Total	8–12 Yrs (n=20)	13–17 Yrs (n=31)	Total
Physical Activity	43.5 (12.1)	40.5 (7.6)	41.6 (9.4)	37.8 (8.0)	38.4 (10.2)	38.2 (9.3)
Physical Function - Mobility	47.2 (8.5)	41.6 (8.2)	43.6 (8.6)	44.1 (11.1)	41.1 (9.3)	42.3 (10.1)
Pain Interference	43.7 (8.0)	46.6 (9.4)	45.6 (9.0)	45.8 (12.3)	47.3 (8.9)	46.7 (10.3)
Fatigue	42.2 (11.5)	48.1 (11.0)	46.0 (11.5)	46.0 (16.7)	49.1 (11.0)	47.9 (13.5)
Depressive Symptoms	43.7 (9.5)	46.6 (10.1)	45.5 (9.9)	45.5 (12.2)	45.8 (10.6)	45.7 (11.1)
Anxiety	43.9 (9.4)	44.8 (10.2)	44.5 (9.8)	45.9 (12.0)	43.6 (10.0)	44.5 (10.8)
Psychological Stress	49.0 (10.5)	50.6 (8.6)	50.0 (9.3)	50.0 (13.0)	50.1 (9.8)	50.1 (11.1)

Abbreviations: PROMIS, Patient-Reported Outcome Measurement Information System; SD, standard deviation; Yrs, years.

The standardized effect size for change in average steps-per-day from T1 to T2 for the PROMIS Physical Function-Mobility (−.15) and Fatigue (.17) are considered a “small” effect by Cohen.^36,37 Other symptom effect sizes were negligible over time (ranged from .01 to .12 in absolute value). In contrast, PROMIS Pediatric Physical Activity means decreased significantly (p =.01) over time, for all participants with PRO data at both time points, by 3.7 points. The change in physical activity was experienced more by children (5.6 points decrease; p=.04) than adolescents (2.51 points decrease; p=.14). While not statistically significant (p=.09), the average physical activity change scores over time was higher for patients hospitalized at T2 (6.8 points, SD = 12.7) than patients not hospitalized (1.9 points, SD = 8.9). The standardized effect size for change in the PROMIS Physical Activity was −.35; which is between a small to moderate effect.^36,37

CUMMULATIVE SYMPTOM COUNT

At T1, children/adolescents experienced an average of 7.2 co-occurring symptoms (SD=3.4) and at T2, the children/adolescents reported an average of 7.7 symptoms (SD=3.5). There were no statistically significant differences by age group or over time.

ASSOCIATION OF STEP COUNT AND PATIENT-REPORTED MEASURES

Table 4 provides correlations between average daily step count-per-week and the PROMIS data by age group and time point. PROMIS Pediatric Physical Activity and Physical Function-Mobility had significant correlations with step counts at T1 and T2 for the whole sample. PROMIS Physical Function-Mobility displayed a moderate level of positive correlation with steps in both age groups at T1 (r=.51 in children; r=.60 in adolescents). At T2, Physical Function-Mobility maintained a significant correlation in the adolescents. PROMIS Physical Activity was also found to have an overall positive correlation with steps at both time points (r=.30 at T1; r=.49 at T2) but displayed more variability by time point and age group. At T2, average-daily-steps in the adolescent group were associated with PROMIS domains of Pain Interference, Fatigue and Anxiety.

Table 4.

Correlation of PROMIS Mean Scores and Average Steps per Week

	Time Point 1			Time Point 2
PROMIS Domain	8–12 Yrs	13–17 Yrs	Total	8–12 Yrs	13–17 Yrs	Total
Physical Activity	.48a	.03	.30a	.35	.60b	.49b
Physical Function-Mobility	.51b	.60b	.60b	.22	.54b	.39b
Pain Interference	−.12	−.33	−.28a	−.05	−.36a	−.20
Fatigue	.16	−.33	−.19	−.06	−.50b	−.26
Depressive Symptoms	.07	−.09	−.07	.27	−.32	−.03
Anxiety	.13	−.28	−.12	.34	−.41a	−.02
Psychological Stress	.28	−.22	−.01	.09	−.31	−.10

^ap <.05;

^bp <.01.

Abbreviations: Yrs=years.

Table 5 presents results from the multivariable mixed effect model including the estimates of the association between the patient-reported data and average daily steps-per-week adjusting for time point, demographics, and clinical characteristics including hospitalization. PROMIS Pediatric domains of Physical Activity, Physical Function-Mobility, and Fatigue were statistically associated with steps-per-day. Using the PROMIS MID, a 3-point improvement in PROMIS Pediatric scores of Physical Activity and Physical Function-Mobility were associated with increases in step counts of 115.8 steps-per-day and 238.8 steps-per-day, respectively. A 3-point worsening in PROMIS Pediatric Fatigue scores was associated with decreases of 94.8 steps-per-day. In addition, the average symptom count was negatively associated with average steps-per-day. A one unit increase in the average number of symptoms was associated with a decrease of 114 steps-per-day. Both T2 (relative to T1) and hospitalization overnight (compared to no hospitalization) were significantly associated with lower steps-per-day.

Table 5.

Average Daily Step Count Associated with Patient-Reported Outcome, Time Point and Hospitalization

	Patient-Reported Outcome (PRO) Included in Regression Model with Average Daily Step Count as Outcome
	Physical Activity	Physical Function-Mobility	Pain Interference	Fatigue	Depressive Symptoms	Anxiety	Psychological Stress	Symptom Count
	Steps (SE)	Steps (SE)	Steps (SE)	Steps (SE)	Steps (SE)	Steps (SE)	Steps (SE)	Steps (SE)
PRO	38.6 (14.6)a	79.6 (21.0)b	−30.0 (17.5)	−31.6 (13.5)a	−14.5 (16.9)	12.8 (16.9)	−21.5 (16.0)	−114.1 (49.7)a
Time 2 (Ref: time 1)	−590.4 (195.8)b	−706.9 (184.6)b	−714.7 (188.9)b	−701.8 (185.6)b	−715.9 (192.7)b	−702.9 (187.6)b	−710.4 (190.7)b	−709.6 (189.9)b
Hospitalization (Ref: no)	−1160.2 (294.6)b	−949.3 (294.8)b	−1160.7 (298.3)b	−1095.6 (296.3)b	−1204.4 (305.6)b	−1298.5 (298.4)b	−1260.4 (294.7)b	−1217.5 (292.7)b

Abbreviations: PRO, patient-reported outcome; Ref, reference; SE, standard error.

^ap<.05;

^bp<.01.

Note: The model also included the child’s age, gender, race, and cancer type (neuro-oncology, solid tumor, and leukemia/lymphoma) but were not statistically related to average daily steps per week.

Discussion

Children and adolescents with cancer experience multiple symptoms during a course of chemotherapy. Self-report of symptoms from children is the gold standard but not always possible. This study demonstrates that steps-per-day can provide information which can help clinicians gage a child’s level of physical activity, physical function-mobility, fatigue and cumulative symptom count. In this study, children reported over 7 co-occurring symptoms at each assessment point. An important finding of this study was the negative impact of cumulative symptom count (multiple co-occurring symptoms) on steps-per-day. Studies in adult oncology have shown that higher symptom burden (measured as the number of concurrent symptoms) is associated with decreased physical function.^38,39 A prior latent profile analysis study in children with cancer reported that those with higher symptom count had decreased functional status.⁸ This study extends these findings by demonstrating an inverse association between cumulative symptom count and step data in children with cancer. This finding suggests that lower step counts may be an indicator for additional assessment of co-occurring symptoms.

Our work provides additional construct validity of the PROMIS Pediatric domains of Physical Activity and Physical Function-Mobility by demonstrating moderate associations with a clinical anchor (step count). This association remained significant after adjusting for hospitalization of the child during the assessment period. In addition, the PROMIS Physical Activity measure and the accelerometer picked up significant decreases in activity from T1 (just prior to a round of treatment) to T2 (just post a round of treatment). We have no reasonable explanation for the observed .03 correlation between steps and physical activity at T1 in adolescents, especially given the strong correlation at T2. Our findings highlight that child self-report using PROMIS is a reliable indicator of physical activity and/or physical function-mobility during cancer therapy. This is an important finding as others have identified inconsistency, across all ages, in the validity of other self-report physical activity measures.⁴⁰

This study demonstrated an inverse correlation between self-reported fatigue and average daily step count. This relationship is in agreement with prior publications.^41,42 Increased physical activity has proven benefits for combating cancer related fatigue in adults. Less evidence is available in children, but preliminary studies suggest similar benefits.^43,44 This is important as fatigue may moderate other symptoms and predict overall symptom burden during therapy.⁴⁵ Inpatient settings, where fatigue is highest and steps-per-day lowest, may hold potential for future intervention studies related to increased physical activity.

Our findings confirm that children receiving treatment for cancer, especially those hospitalized, have drastically reduced step counts. Overall steps-per-day across all age groups were highest at T1 with 4099 steps-per-day. Even at the highest peak, these step numbers are significantly below the recommended 10,000+ steps-per-day for healthy children and adolescents.⁴⁶ Steps additionally decreased between T1 and T2 in the entire cohort suggesting that the burden of treatment may unduly influence a child’s ability to carry out normal play and activities of daily living. The finding of decreased physical activity during childhood cancer therapy is in line with other publications^47,48 and has implications for the long-term health of childhood cancer survivors by further increasing risk factors for chronic disease and mortality.^49,50 As children undergo treatment for months to years, maintaining physical activity is also critically important to continued progression along their developmental trajectory.⁵¹ The safety of increased physical activity in children receiving active cancer therapy has been demonstrated, yet effective and feasible interventions which impart lasting change are currently lacking.⁵² Over the past decade, evidence has become available to strongly support the integration of early oncology rehabilitation programs to maintain function and physical activity during childhood cancer therapy, yet the majority of pediatric oncology treatment centers have yet to operationalize this practice.^53,54

Implications for Practice

The association between average daily step count and pediatric self-reported physical activity and physical function-mobility have important implications for nurses, and other clinicians, working with children undergoing treatment for cancer. For example, younger children demonstrated larger declines in daily steps and physical activity, over the course of therapy, as compared to adolescents. Additionally, larger declines in daily steps and physical activity were observed in hospitalized children/adolescents. This finding suggests that all children/adolescents, but especially those younger in age and hospitalized, need intervention to assist with maintenance of normal physical activity during therapy. During in-patient stays, nurses can provide motivation for children to be physically activity through interaction with psychosocial teams, empowering parental assistance, and encouraging participation in events that may be offered (such as in-patient yoga or exercise classes, or walking). Referrals to oncology rehabilitation programs, physical therapy or other experts in the field of exercise science are recommended during therapy.¹⁴ Further, our findings suggest that future studies capturing self-reported changes in physical activity or physical function, during cancer therapy, may also wish to collect data related to hospitalization stays as this may aid in data interpretation.

As child self-reports of physical activity and physical function are valid during therapy, nurses and other clinicians should, at a minimum, be routinely asking children/adolescents about their activity or capturing this information through formalized self-reports. For children/adolescents noted to be wearing a physical activity monitor, it is recommended that clinicians consider asking about step count and recording this information in the medical record for use in monitoring changes and to trigger deeper conversations regarding physical function- mobility, physical activity and fatigue. Children 5–6 years of age are not typically asked to self-report but could potentially wear child activity monitors to provide measurable data. Using information from these sources could be used to support early referrals for supportive care (i.e. Physical Therapy) and to guide nursing education to families aimed at promoting physical activity.

Limitations

Our findings, limited by their representation of English-speaking children, may not be generalizable to all children with cancer. Although 61.9% of the sample self-identified as White, there was diversity with 37.8% non-white participants. Accelerometer data were retained in the analyses only if they demonstrated rigorous adherence to monitor wear time. Missing data was higher than anticipated primarily related to non-wear time and difficulty retrieving (syncing) data from a commercial device. Additionally, the small sample size may have limited our ability to detect significant correlations between individual symptoms and steps. This study focused on the presence of symptoms or cumulative symptom count as an indication of a child’s steps-per-day. The severity and interference of specific symptoms which may be relevant to physical activity and physical function-mobility were not explored. Lastly, this study cannot be used to imply causal relationships between symptoms and function and steps-per-day.

Conclusions

This study provides evidence related to correlations between symptoms and steps-per-day in children with cancer. Children undergoing therapy had decreased step counts which correlated with child self-report of PROMIS Pediatric Physical Activity, Physical Function-Mobility, Fatigue, and cumulative symptom count. These findings support the use of child self-report to accurately monitor changes in physical activity and physical function during therapy. Clinical implications from this study support routine monitoring of child self-report during therapy, when possible, to accurately capture symptom burden and to trigger early referrals related to supportive care for maintenance of normal physical activity and physical function. Step monitoring may also serve as an objective indicator for overall symptom count, fatigue, physical activity and physical function. If reduced steps are noted, this finding could be used to trigger a broader assessment of co-occurring symptoms which has been shown to produce a cumulative symptom burden. Research implications include further exploration of the interaction between symptoms and physical activity or physical function-mobility, especially in an in-patient setting, and creation of effective intervention studies to improve physical activity while measuring the associated short and long-term outcomes.