Measurement of Physical Activity by Actigraphy in Infants and Young Children with Pulmonary Arterial Hypertension

Abstract

Objective

To evaluate the feasibility, tolerability, and adherence with wearable actigraphy devices among infants and children with pulmonary arterial hypertension (PAH).

Study design

This multicenter, prospective, observational study included children ages 0–6 years with and without PAH. Participants wore the ActiGraph wGT3X-BT on the hip and FitBit Inspire on the wrist during waking hours for 14 days. Steps, vector magnitude counts per minute, activity intensity, heart rate, and heart rate variability were compared between groups.

Results

Forty-seven participants (18 PAH, 29 control) were enrolled from 10 North American sites. PAH patients were mostly functional class II (n = 16, 89%) and treated with oral medications at the time of enrollment. The number of wear days was not significantly different between the groups (ActiGraph: 10 [95% CI: 5.5, 12.2] in PAH vs 8 [4, 12] in control, P = .20; FitBit 13 [10, 13.8] in PAH vs 12 [8, 14] in control, P = .87). Complete data were obtained in 81% of eligible ActiGraph participants and 72% of FitBit participants. PAH participants demonstrated fewer steps, lower vector magnitude counts per minute, more sedentary activity, and less intense physical activity at all levels compared with control participants. No statistically significant differences in heart rate variability were demonstrated between the 2 groups.

Conclusions

Measurement of physical activity and other end points using wearable actigraphy devices was feasible in young children with PAH. Larger studies should determine associations between physical activity and disease severity in young patients with PAH to identify relevant end points for pediatric clinical trials.

Pediatric pulmonary arterial hypertension (PAH) is a rare but serious disorder with an estimated prevalence of 2–16 cases per million and about 75% 5-year survival from diagnosis.^1–4 At the time of this study, the only pulmonary vasodilators approved to treat pediatric patients with PAH in the US were inhaled nitric oxide for persistent pulmonary hypertension of the newborn and bosentan for children over age 3. Unfortunately, there is a lack of suitable clinical end points and measures of disease severity.⁵ The 6-minute walk distance (6MWD) is the most common primary end point in adult PAH clinical trials,⁶ however, infants and children with PAH are often too young and/or developmentally unable to perform 6MWD or standard cardiopulmonary exercise testing. Identification of novel measures that reproducibly and reliably reflect changes in hemodynamics or functional status in response to a therapeutic intervention is critical to advancing care in pediatric PAH.

We sought to use actigraphy to identify novel, developmentally appropriate, noninvasive end points in children with PAH. We aimed to explore the use of actigraphy in infants and children younger than 6 years, a population nearly universally unable to perform 6MWD. The aims of this study were to evaluate the feasibility, tolerability, and adherence with wearable actigraphy devices in young children with and without PAH and to explore the ability of these devices to detect differences in activity intensity and heart rate between the 2 groups.

Methods

This prospective observational study included children ages 0–6 years with and without PAH. Inclusion criteria for PAH participants were current diagnosis of pulmonary hypertension in World Symposium of Pulmonary Hypertension diagnostic group 1 as per established clinical criteria including prior cardiac catheterization meeting Group 1 criteria; Panama functional class II-IIIa⁷; treatment with an approved oral endothelin receptor antagonist, calcium channel blocker, phosphodiesterase 5 inhibitor, prostacyclin or prostacyclin analog, and/or soluble guanylate cyclase stimulator for at least 30 days prior to consent; no medication dose adjustments (other than weight-based adjustments) for at least 30 days prior to consent; and stable non-PAH medical therapy for 14 days prior to enrollment visit without dose adjustments, additions, or discontinuations (exceptions diuretics, anticoagulants, cold/seasonal allergy medications). Exclusion criteria included non-Group 1 pulmonary hypertension classification; any bone, neuromuscular, or other pathology that could limit activity; trisomy 21 or other genetic condition that prevented testing; use of any medication known to limit activity; active infection; other systemic condition that may have affected the safety of the participant or interfere with study assessments; active transplantation listing; or developmental or mental health concerns in the participant or guardian that could compromise safety with participation in the study. Control participants were identified by site-specific standard research recruitment practices, including email, letter campaigns, flyers, or identification of healthy siblings of PAH participants. Control participants were included if they were in good general health as evidenced by medical history reported by parent or guardian during screening and/or review of available medical records. Control participants were excluded for a diagnosis of asthma, inability to exercise due to intellectual or physical disability that impacted activities of daily living, or developmental or mental health concerns in the participant or guardian that could compromise safety with participation in the study. Patients were enrolled during the initial funding period from 9/01/2019 to 8/31/2021. Due to the COVID-19 pandemic hindering enrollment at nearly every clinical site, an extension was granted through 2/28/2022, but enrollment was required to stop at that time per the sponsor, the Food and Drug Administration.

At enrollment, the following variables were abstracted from the medical record of PAH participants: age, demographics (gender, race, ethnicity), and clinical data from physical examinations, echocardiograms, genetic and laboratory testing, and hemodynamic data from the most recent cardiac catheterization prior to enrollment. Age and demographics were collected in control participants. Control participants underwent a limited physical examination for height, weight, resting blood pressure, heart rate, and oxygen saturation at study enrollment.

PAH and control participants ≥1 year of age were instructed to wear a FitBit Inspire series device (Fitbit) and the ActiGraph wGT3X-BT device (Actigraph) as much as possible during waking hours for 14 days. Participants <1 year of age wore the FitBit only as infants were assumed to be nonambulatory and not suitable for activity measurement via the hip-worn ActiGraph. The FitBit Inspire is a triaxial accelerometer and heart rate monitor which participants wore on the non-dominant wrist. The ActiGraph is also a triaxial accelerometer and pedometer which participants wore on the right hip as this has been shown to be more feasible, better accepted by toddlers, and with a longer overall wear time compared with the wrist position.⁸ The device was programmed to record triaxial data at a frequency of 60 Hz. Participant families downloaded the Fitbit Dashboard and ActiGraph CentrePoint applications to their mobile devices to synchronize their data throughout the 2 weeks and upload at the end of the study period. Adverse events were assessed via telephone communication 15–21 days after the visit.

The study was approved the University of Colorado Multiple Institutional Review Board which acted as the central institutional review board under reliance agreements in place with the other sites. Informed consent was obtained from the participant’s parent or legal guardian prior to study enrollment.

Demographic and clinical variables were summarized using median and IQR for continuous variables and counts and percentages for categorical variables. Wilcoxon rank sum tests were used to compare continuous variables between the PAH participants and the healthy control participants. The sample size is small in this pilot study, limiting the power to demonstrate differences between the 2 groups.

A valid day for actigraphy was defined by a minimum of 4 hours of device wear time. Sleep or nonwear time was defined as any time period with zero counts for >90 minutes. According to the original protocol, participants with at least 3 valid weekdays and 1 valid weekend day in each of the 2 weeks of the 14-day recording period (≥8 days total) were included in the primary analyses. We also performed post hoc secondary analyses including all participants with at least 3 valid weekdays and 1 valid weekend day at any point during the 14-day study period (≥4 but <8 days).

ActiGraph data were processed using the ActiLife software (ActiGraph). Data were downloaded and integrated into 60-second epochs, per the manufacturer’s recommendations. Steps and vector magnitude counts per minute (CPM) were averaged over total wear time. Vector magnitude is calculated as the square root of the quadrate of the 3 separate dimensional axes $[{\left(x^2+y^2+z^2\right)}^{1/2}]$. Activity intensity was categorized according to Butte cut points for preschool children.⁹ These cut points were derived from 50 preschool children (mean 4.5 ± 0.8 years) participating in accelerometer counts by hip-worn ActiGraph and room calorimetry for minute-by-minute measurements of energy expenditure. Final ActiGraph accelerometer counts were determined as follows: sedentary 0–819 CPM, light activity 820–3907 CPM, moderate activity 3908–6111 CPM, and vigorous activity ≥6112 CPM.

FitBit activity intensity is determined by the vendor’s proprietary algorithm which may not be applicable to specific populations so only FitBit step counts were recorded. Heart rate variability (HRV) was defined as the mean squared error of minute-by-minute heart rate differences measured by FitBit over the study period.

Activity intensity levels were compared between groups using Wilcoxon rank sum tests. Clustered resampling was applied to compare steps and HRV.

Results

Forty-seven participants (18 PAH, 29 control) were enrolled from 10 sites during the funding period. No statistically significant differences in age, gender, or race were demonstrated between PAH and control participants (Table I). Three PAH participants were of Hispanic ethnicity while there were no Hispanic control participants. There were no statistically significant differences in body size or vital signs between the 2 groups except for lower resting diastolic blood pressure and higher oxygen saturation in the PAH participants.

Table I.

Demographic and physical examination characteristics of the participants

Characteristic	PAH n = 18	Controls n = 29	P value
Age, y	4.6 (3.8, 5.5)	3.5 (1.9, 5.6)	.18*
	Range 1.6–7.0	Range 0.9–6.9
Gender
Male	11 (61%)	18 (62%)	1.0
Female	7 (39%)	11 (38%)
Hispanic ethnicity	3 (17%)	0 (0%)	.05
Race
Asian	0 (0%)	2 (7%)	.15†
Black or African American	2 (11%)	0 (0%)
Caucasian	13 (72%)	24 (83%)
Other/multiple races	3 (17%)	3 (10%)
Physical exam data‡
Height, cm	104.9 (97.2, 109.3	100.0 (86.5, 113.9)	.71
Weight, kg	16.8 (13.7, 18.0)	15.6 (12.8, 19.4)	.94
BSA, m2	0.67 (0.58, 0.74)	0.64 (0.48, 0.74)	.49
Resting heart rate, bpm	102 (87, 108)	104 (98, 110)	.27
Systolic BP, mm Hg	93 (86, 98)	95 (90, 102)	.16
Diastolic BP, mm Hg	54 (48, 59)	61 (56, 66)	<.01
O2 saturation, %	99 (97, 100)	96 (95, 97)	.03

Data expressed as median (IQR) or n (%).

BP, blood pressure; BSA, body surface area.

^*Wilcoxon rank sum test.

^†Freeman Haltman extension of Fisher exact test.

^‡One participant had a video visit so there were no vitals performed.

PAH participants included those with idiopathic, heritable, and congenital heart disease-related PAH (Table II). Participants were more commonly functional class II. Twelve (67%) had a history of genetic testing with mutations identified in ACVRL1, ENG, TBX4, ATP13A3, and GDF2. The median N-terminal probrain natriuretic peptide level was 78.0 (12.5, 431.0) pg/mL at study enrollment in the 7 participants with those data. Oral PAH therapies were common, and 7 (39%) participants were on subcutaneous treprostinil during the study period. Echocardiograms were available at median 1.1 (0.0, 2.3) months prior to study visit. Median right ventricular pressure estimate by tricuspid valve regurgitant jet velocity on echocardiogram was 44 (21.7, 63.0) mm Hg. Median right ventricular fractional area change and tricuspid annular plane systolic excursion z score were 38.9 (35.5, 47.8) % and −1.69 (−3.50, 0.30), respectively. Thirteen (72%) participants had cardiac catheterization data at median 2.4 (0.8, 3.9) years prior to study visit (Supplemental Table I; available at www.jpeds.com). Data met diagnostic criteria for precapillary PAH with mildly elevated pulmonary vascular resistance and normal cardiac index.

Table II.

Characteristics of patients with PAH (n = 18)

Characteristic	n (%) or median (IQR)
Pulmonary arterial hypertension subtype
Idiopathic PAH	9 (50%)
Heritable PAH	3 (17%)
PAH associated with congenital heart disease	6 (33%)
Functional classification
Class II	16 (89%)
Class IIIa	2 (11%)
Genetic testing performed	12 (67%)
Positive genetic testing
ACVRL1	1 (8%)
ENG	1 (8%)
TBX4/small patella syndrome	1 (8%)
ATP13A3 mutations	1 (8%)
GDF2	1 (8%)
Labs*
Blood urea nitrogen, mg/dl	14 (12, 23)
Creatinine, mg/dl	0.30 (0.20, 0.35)
Aspartate aminotransferase	33 (25, 41)
Alanine aminotransferase	13 (13, 21)
Total bilirubin, mg/dl	0.40 (0.30, 0.50)
Hemoglobin, g/dl	12.3 (11.4, 13.6)
Hematocrit, %	37.0 (35.8, 37.9)
N-terminal proBNP, pg/ml†	78.0 (12.5, 431.0)
PAH medications
Sildenafil	4 (22%)
Tadalafil	13 (72%)
Bosentan	7 (39%)
Ambrisentan	7 (39%)
Selexipag	2 (11%)
Subcutaneous treprostinil	7 (39%)
Echocardiogram
Time from enrollment, m	1.1 (0.0–2.3)
Condition
Room Air	16 (89%)
Oxygen	2 (13%)
Septal flattening
None	4 (27%)
Mild	7 (47%)
Moderate/“D” shaped interventricular septum	3 (20%)
Severe with septal bowing	1 (6%)
Tricuspid valve jet velocity, m/s	3.47 (2.3, 4.0)
Right ventricular pressure estimate, mm Hg	44 (21.7, 63.0)
Tricuspid annular plane systolic excursion z score	−1.69 (−3.5, −0.3)
Right ventricular fractional area change, %	38.9 (35.6, 47.8)
Left ventricular internal dimension in diastole, mm	3.1 (2.9, 3.7)
Left ventricular internal dimension in systole, mm	1.90 (1.7, 2.2)
Left ventricular shortening fraction, %	40.0 (36, 45)
Left ventricular ejection fraction, %	65.5 (63.0, 69.0)

Data expressed as median (IQR) or n (%).

proBNP, N-terminal probrain natriuretic peptide.

^*n = 13.

^†n = 7.

Both devices were generally well tolerated. There were no adverse events with the ActiGraph. There were 2 reports of minor abrasion and skin rubbing with the FitBit.

For children between 1 and 6 years of age, there were no statistically significant differences in the number of device wear days between PAH and control participants [ActiGraph: 10 (95% CI 5.5, 12.2) in PAH vs 8 (4, 12) in control, P = .20; FitBit 13 (10, 13.8) in PAH vs 12 (8, 14) in control, P = .87]. Of those 42 participants (18 PAH, 24 controls), complete ActiGraph data were obtained in 34 (15 PAH, 19 control; 81% compliance) (Supplemental Table IIa; available at www.jpeds.com). Incomplete ActiGraph data (fewer than the required number of days over 14-day period) were obtained in 2 control participants. No ActiGraph data were recorded in 6 participants (3 PAH, 3 control) due to nonwear. Complete FitBit data were obtained in 34 of 47 participants (14 PAH, 20 control; 72% compliance). These were not the same 34 participants with complete ActiGraph data. Incomplete FitBit data were obtained in 8 participants (3 PAH, 5 control); 7 participants with ≥4 but <8 days and 1 participant with <4 days. No FitBit data were recorded in 5 participants (1 PAH, 4 control) due to nonwear. When participants with incomplete data (≥4 but <8 days; 2 ActiGraph, 7 FitBit) were included, adherence improved to 86% for the ActiGraph and 87% for FitBit (Supplemental Table IIb; available at www.jpeds.com).

Some reasons for nonadherence included: lack of data synch (FitBit) and upload (ActiGraph), failure to return the device to the study team, mechanical issues with the device (FitBit), language barrier resulting in difficulty with device use, and child’s refusal to wear the device in 1 participant (ActiGraph).

Analysis of findings from the 34 participants with complete FitBit data demonstrated trends toward lower total and weekday heart rate in PAH participants (Table III). Within PAH and control groups, there were no statistically significant differences between weekend and weekday heart rates (P = .91 for PAH and P = .49 for control). There were no statistically significant differences in HRV between the 2 groups. When the participants with incomplete data were included (41 total participants), lower total heart rate was observed in PAH participants [100.4 (94, 111) vs 110.9 (101.1, 117.2), P = .02] and there was a trend toward lower weekend heart rate in PAH participants (Table IV). Again, there were no statistically significant differences in HRV between the 2 groups or between weekend and weekday heart rates within the 2 groups.

Table III.

Primary analysis of FitBit heart rate data^*

Heart rate	PAH n = 14	Control n = 20	P value
HR total	103.0 (94, 110)	112.6 (106, 115)	.06
HR weekends	101.3 (94, 115)	111.1 (105, 115)	.12
HR weekdays	103.0 (93, 112)	112.9 (107, 121)	.05
Heart rate variability, ms	187 (132, 242)	246 (177, 340)	.312

HR, heart rate.

^*At least 3 weekday wear d and 1 weekend wear d during each of the 2 wk in the 14-d study period (≥8 d total).

Table IV.

Secondary analysis of FitBit heart rate data^*

Heart rate	PAH n = 16	Control n = 25	P value
HR total	100.4 (94, 111)	110.9 (101.1, 117.2)	.02
HR weekends	101.1 (94, 111)	112.1 (101.1, 121.8)	.06
HR weekdays	100.4 (93, 111)	110.9 (98, 113)	.20
Heart rate variability, ms	185 (132, 242)	260 (185, 398)	.25

Data expressed as median (IQR).

HR, heart rate.

^*At least 3 weekday wear d and 1 weekend wear d during the 14-d study period (≥4 d total).

Participants with PAH demonstrated more sedentary activity and less intense physical activity at all levels compared with control participants (Table V). The findings were similar on both primary and secondary analyses when 2 control participants with incomplete data were included. Participants with PAH also demonstrated fewer steps and vector magnitude CPM measured by ActiGraph compared with control participants (Supplemental Table III; available at www.jpeds.com). This was true overall, on weekends, and on weekdays. Within PAH participants, weekday vector magnitude CPM were lower compared with weekend counts [825 (497, 1001) vs 863 (554, 1027), P = .02), These findings were also demonstrated on secondary analysis. Within control participants, there were no statistically significant differences in weekday and weekend steps and vector magnitude CPM measured by ActiGraph. On FitBit, PAH participants demonstrated fewer total and weekday steps compared with control participants. There was a trend toward lower weekend steps in the PAH participants as well. There were no statistically significant differences between weekend and weekday steps in PAH participants.

Table V.

Analyses of physical activity intensity by ActiGraph using Butte intensity cut points in preschool children

Activity intensity	PAH n = 15	Control n = 19	P value
Primary analysis*
Sedentary activity	63.6 (56.4, 74.1)	49.6 (47.1, 54.3)	.002
Light intensity	33.7 (25.3, 39.2)	43.6 (40.0, 45.7)	.003
Moderate intensity	1.8 (1.3, 4.0)	4.5 (3.9, 5.6)	.002
Vigorous intensity	0.2 (0.1, 0.7)	1.1 (0.8, 1.7)	<.001
	PAH n = 15	Control n = 21	P value
Secondary analysis†
Sedentary activity	60.7 (55.1, 67.9)	49.6 (47.1, 54.3)	.014
Light intensity	34.9 (29.0, 43.1)	43.6 (40.0, 45.7)	.028
Moderate intensity	2.9 (1.4, 4.2)	4.5 (3.9, 5.6)	.023
Vigorous intensity	2.5 (0.1, 0.9)	1.1 (0.8, 1.6)	.013

Data expressed as median (IQR) of percent of time spent in each activity intensity.

^*At least 3 weekday wear d and 1 weekend wear d during each of the 2 wk in the 14-d study period (≥8 d total).

^†At least 3 weekday wear d and 1 weekend wear d during the 14-d study period (≥4 d total).

When step counts were compared between the 2 devices, ActiGraph step counts were lower compared with FitBit in control participants (Supplemental Table IV; available at www.jpeds.com). There were no statistically significant differences in step counts between the 2 devices in PAH participants.

Discussion

Despite challenges with compliance in both healthy children and children with PAH, these preliminary data support the utility of actigraphy to measure physical activity in observational and interventional trials in children as young as infancy.

The 6-minute walk test is the most common test of exercise capacity in PAH clinical care and adult medication trials. In this pilot study, we demonstrate that wearable actigraphy devices are tolerable in a young population, and it is feasible to measure activity in both patients with PAH and healthy children. While both research-grade and commercially available devices were tolerable and feasible, adherence proved challenging for both devices and both types of participants. Most nonadherence was related to lack of data synching/upload or minor device malfunctions, as opposed to unacceptability of the device by participants or parents/guardians. Telephone communication with families occurred at the end of the 2-week period, but this may not have provided sufficient reinforcement. Recurring communication with families during the study period might improve adherence. However, there were few differences in study findings from the planned, primary analyses and the post hoc secondary analyses which included participants who did wear the device, but for fewer than the recommended number of days in the protocol. These data may suggest that clinically relevant activity data can still be obtained in this population with a shorter wear time.

Our data demonstrate markedly lower levels of activity in very young PAH patients compared with healthy children, which has not previously been described. Zijlstra et al used the hip-worn ActiGraph wGT3X to demonstrate low activity levels in a slightly older cohort of preschool to school age children with PAH [3.1 (1.2–9.7) years] in the Dutch National Network of Pediatric Pulmonary Hypertension.¹⁰ In their study, less time spent in moderate to vigorous physical activity correlated with worse functional class and lower 6MWD in those who could perform the test. Additionally, lower vector magnitude CPM and less time spent in moderate to vigorous physical activity was associated with shorter time to PAH-related hospitalization or death. Activity may be an indicator of disease severity and a potential therapeutic end point for pediatric clinical trials in children as young as infants and toddlers. A current clinical trial of mono-vs dual-oral PAH therapy in incident pediatric patients with PAH is exploring changes in actigraphy parameters with treatment.¹¹

We also measured heart rate and HRV using the FitBit Inspire. We found few differences in heart rate and no statistically significant differences in HRV between PAH participants and healthy controls. This contrasts to preliminary data from our group in older children with PAH ages 7–17 years.¹² In that group, we found lower HRV in PAH participants compared with healthy children. We postulated that this might be due to more advanced disease, in which resting heart rate trends are higher and HRV is less. In the current study, lack of statistically significant differences in resting heart rates between younger PAH patents and healthy children may indicate less advanced disease earlier in life. Alternatively, the small sample size may have limited our ability to detect statistically significant differences between the 2 groups. Finally, electrocardiogram tracings are not available from FitBit, which would improve the quality of HRV data.

When analyzing actigraphy data, investigators are often limited by a vendor’s proprietary analytic algorithm or platform. Commercially available devices (such as FitBit) and research-grade devices (such ActiGraph) may produce different data as demonstrated by the different step counts measured by the 2 devices in this study.

There were some limitations to this study. The small sample size may have limited the power to demonstrate differences between the 2 groups in this pilot study. Selection bias may have occurred as the study included relatively well outpatient PAH patients who were motivated to participate in a research study. Findings may be different in more symptomatic patients with worse functional class. Additionally, recruitment was challenging during the COVID-19 pandemic. Some patients did not want to come to our centers for in-person care unless absolutely necessary, biasing enrollment in a research study. The small sample size did not allow us to adjust activity measures for age and sex. Published cut points are not specific to age and sex but are applicable to children of both sexes in a general age range. A larger sample size may have allowed us to explore the associations among age, sex, and activity levels in pediatric PAH. We did not correlate actigraphy data with echocardiographic or hemodynamic data because of the variable length of time between the activity data and clinical findings. Finally, the cross-sectional design of this study does not allow us to draw any conclusions about the effect of PAH treatments on physical activity in this population.

Future studies should explore the association between physical activity, disease severity, and disease progression in this younger age group in whom standard assessment of exercise performance is not possible. Further, the effect of PAH treatments on actigraphy should be studied for future trial design.

Glossary

CPM

Counts per minute

HRV

Heart rate variability

6MWD

6-minute walk test distance

PAH

Pulmonary arterial hypertension