If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
To validate parent-reported child habitual total physical activity against accelerometry and three existing step-count thresholds for classifying 3 h/day of total physical activity in pre-schoolers from 13 culturally and geographically diverse countries.
Design
Cross-sectional validation study.
Methods
We used data involving 3- and 4-year-olds from 13 middle- and high-income countries who participated in the SUNRISE study. We used Spearman's rank-order correlation, Bland–Altman plots, and Kappa statistics to validate parent-reported child habitual total physical activity against activPAL™-measured total physical activity over 3 days. Additionally, we used Receiver Operating Characteristic Area Under the Curve analysis to validate existing step-count thresholds (Gabel, Vale, and De Craemer) using step-counts derived from activPAL™.
Results
Of the 352 pre-schoolers, 49.1 % were girls. There was a very weak but significant positive correlation and slight agreement between parent-reported total physical activity and accelerometer-measured total physical activity (r: 0.140; p = 0.009; Kappa: 0.030). Parents overestimated their child's total physical activity compared to accelerometry (mean bias: 69 min/day; standard deviation: 126; 95 % limits of agreement: −179, 316). Of the three step-count thresholds tested, the De Craemer threshold of 11,500 steps/day provided excellent classification of meeting the total physical activity guideline as measured by accelerometry (area under the ROC curve: 0.945; 95 % confidence interval: 0.928, 0.961; sensitivity: 100.0 %; specificity: 88.9 %).
Conclusions
Parent reports may have limited validity for assessing pre-schoolers' level of total physical activity. Step-counting is a promising alternative – low-cost global surveillance initiatives could potentially use pedometers for assessing compliance with the physical activity guideline in early childhood.
Parent reports of their child's level of physical activity are not accurate.
•
An alternative simple objective monitoring method is required, especially for low- and-middle-income countries.
•
Step-counting (e.g., using pedometers) provides an accurate low-cost option and may be suitable internationally for population monitoring of physical activity in early childhood to improve children's health and prevent obesity and related diseases, such as diabetes, high blood pressure and some cancers.
1. Introduction
In 2016, the World Health Organization (WHO) recommended promotion of physical activity in early childhood as a critical component of the global obesity prevention agenda.
Consequently in 2019, the WHO developed the first global guidelines for physical activity, sedentary behaviours and sleep for under 5 s to tackle the obesity pandemic and improve children's health and development.
Among pre-schoolers (3–4 years), the WHO recommends daily total physical activity (TPA) of at least 180 min including 60 min of moderate- to vigorous-intensity physical activity (MVPA). Despite the publication of these guidelines in 2019,
partly due to practical issues like budget limitations and uncertainty over the validity and cultural appropriateness of physical activity surveillance measurements globally. As such, there is a need for a relatively simple, low-cost and sustainable method of surveillance which has criterion validity (against accelerometry) to assess compliance with the guidelines and allow comparisons across studies, cultures, populations, and countries. One obvious option on cost grounds is proxy reports from parents,
Cross-sectional examination of 24-hour movement behaviours among 3- and 4-year-old children in urban and rural settings in low-income, middle-income and high-income countries: the SUNRISE study protocol.
but there is no geographically and culturally validated questionnaire for global surveillance of physical activity in the early years (under 5 s) to date.
have validated parent-reported physical activity against the commonly used criterion measure of accelerometry in children under 5 years: Bacardi-Gascón et al.
are limited to a single geographical location and in high income countries (HICs). To the best of our knowledge, there have been no attempts to validate parent-reported TPA against activPAL™ measured TPA (the criterion method) in 3- to 4-year-olds across various countries. activPAL™ is one of the most commonly used research-grade accelerometers for objective measurement of physical activity, and has been validated for measurement of TPA against direct observation in this age group.
Another simple low-cost global alternative to parent reports for global public health surveillance of TPA is step-counting. Over the last decade several studies have developed step-counting thresholds to classify 3 h of TPA in pre-schoolers.
However, previous studies proposed widely varying step-count thresholds derived from the ActiGraph accelerometers which were equivalent to 3 hour daily TPA
suggested 11,500 steps/day. To date, it remains unclear which of these step-count thresholds provides the most accurate measure which might be suitable to use for global surveillance of TPA in pre-school-aged children and whether these thresholds are valid when using different step-count devices and placements.
Therefore, in order to help develop relatively simple methods of physical activity assessment suitable for global public health surveillance in pre-schoolers in future, the aims of our study were to (a) validate parent-reported habitual TPA against activPAL™ measured habitual TPA (calculated as total time spent stepping min/day) in pre-schoolers from geographically and culturally diverse countries, and (b) cross-validate existing step-count thresholds for determining habitual TPA against activPAL™ measured TPA. Given the diversity of our sample, we also explored differences in validation outcomes between parent-reported TPA and accelerometer-measured TPA across the socio-demographic characteristics of study participants.
2. Methods
This study was a secondary analysis of activPAL™ (PAL Technologies Ltd, Glasgow, UK) data collected as part of the first and second pilot phases of the SUNRISE study (https://sunrise-study.com/), an international cross-sectional study of movement behaviours in the early years.
Cross-sectional examination of 24-hour movement behaviours among 3- and 4-year-old children in urban and rural settings in low-income, middle-income and high-income countries: the SUNRISE study protocol.
The SUNRISE study is being conducted in 43 high-, middle- and low-income countries. Over 2500 children aged 2–6 years from 23 countries have completed the pilot phase of the study, with activPAL™ data available for 955 children from 17 countries. Data are de-identified and available on request from the SUNRISE Coordinating Centre based at the University of Wollongong (UOW), Australia. The SUNRISE study protocol was reviewed and approved by Human Research Ethics Committee at the UOW (2018/044) and ethics committees in each participating country; all parents of participating children gave informed consent.
A total of 352 pre-schoolers aged 3–4 years from 13 countries who participated in the pilot phases 1 and 2 of the SUNRISE study comprised the sample for our validation study. Participants were included in the current study if: (i) they had both activPAL™ and parent-reported habitual TPA data; (ii) they had three valid days of activPAL™ measurement (i.e., a valid day was defined as having 24-h of data), which is appropriate to measure usual level of TPA
; and (iii) they were aged 3.0 to <5.0 years. One participant was excluded because the parent-reported level of child physical activity was recorded as zero.
No significant differences in participant characteristics were found between those included and excluded in our study, except for a higher percentage of urban children that have been included in this validation study (59 % vs 44 %). Since future global surveillance of physical activity in children will need to take place in diverse settings, a range of countries (Australia, Bangladesh, Brazil, China, Hong Kong, Indonesia, Japan, Malaysia, South Africa, South Korea, Sri Lanka, Sweden, and Vietnam) and income levels (lower-middle, upper-middle, and high-income countries) were represented in our study.
Habitual TPA by accelerometry was assessed using activPAL™, an activity monitor worn on the thigh with an accelerometer to record time spent sitting/lying, moving/stepping and standing in 15-second epochs.
TPA was calculated as the total time spent stepping per day (min/day). activPAL™ has been validated against direct observation of physical activity for measurement of TPA in 3–4 year-olds, with high sensitivity and specificity for measurement of TPA relative to direct observation and no significant bias in measurement of TPA.
Cross-sectional examination of 24-hour movement behaviours among 3- and 4-year-old children in urban and rural settings in low-income, middle-income and high-income countries: the SUNRISE study protocol.
This allowed collection of three full days of data (3 × 24-hour period) on TPA. Based on the 2019 WHO Global Guidelines for TPA in children aged 3–4 years,
participants were classified as meeting the guideline if they spent an average of at least 180 min/day in TPA.
Habitual TPA by parent reports was assessed using a parent questionnaire completed by self-administration, or interviewer-administered when necessary, for example, where literacy posed challenges.
Cross-sectional examination of 24-hour movement behaviours among 3- and 4-year-old children in urban and rural settings in low-income, middle-income and high-income countries: the SUNRISE study protocol.
A collaborative approach to adopting/adapting guidelines - the Australian 24-Hour Movement Guidelines for the early years (birth to 5 years): an integration of physical activity, sedentary behavior, and sleep.
Parents were asked: “On a 24-hour period in the past week, how much time did the 3- to 4-year-old child who is participating in this study spend in a variety of physical activities, spread throughout the day? For example: active play, running, playing with balls, moving to music/dancing, swimming, riding a scooter/tricycle/bike.” Parent reports were recorded in hours and minutes and were converted to min/day to calculate parent-reported habitual TPA. This was used to classify participants as meeting the physical activity guideline if they spent an average of at least 180 min/day in TPA.
The habitual number of steps taken was assessed using activPAL™ accelerometers. As noted earlier, activPAL™ records time spent sitting/lying, moving/stepping and standing in 15-second epochs,
Since children were asked to wear the device continuously for 3–5 days, its stepping function allowed collection of three full days of data (3 × 24-hour period) on total step-counts. These were used to classify participants as either meeting or not meeting the 180 min/day of TPA in three ways based on the three step-count thresholds in the literature
Parents reported their child's date of birth (or age in complete years if date of birth was unknown) and this was used to determine the child's age in years and months. Parents reported their child's sex as either boy or girl. The highest level of education completed by the parent or other members of the household was recorded based on each participating country's educational classification and this was then grouped into two categories due to varying educational classifications between countries: low (secondary/high school or below) or high (tertiary education or above) education. Country income level was classified based on the World Bank classification (https://datahelpdesk.worldbank.org/knowledgebase/articles/906519-world-bank-country-and-lending-groups): lower-middle, upper-middle, and high-income country. However, we treated the two categories for middle-income country (MIC) as one category (i.e., MIC) in our analyses due to fewer children (n = 47) from lower-middle income countries. A child's residential area was recorded as either urban or rural based on the location of the Early Childhood Education and Care (ECEC) centre or community where children were recruited to participate in the SUNRISE study.
Cross-sectional examination of 24-hour movement behaviours among 3- and 4-year-old children in urban and rural settings in low-income, middle-income and high-income countries: the SUNRISE study protocol.
Descriptive analyses were performed to describe characteristics of participants and presented as means, standard deviations (SDs), frequency, and percentage (%). We assessed the validity of the parent-reported level of child TPA in meeting the WHO TPA guideline for pre-schoolers in four ways. First, we used Spearman's rank-order correlation (r) to determine the ability of the questionnaire to correctly rank order children by time spent in TPA (min/day) measured by activPAL™. The strength of the correlations was classed as: 0.00–0.19 ‘very weak’; 0.20–0.39 ‘weak’; 0.40–0.59 ‘moderate’, 0.60–0.79 ‘strong’; and 0.80–1.0 ‘very strong’.
Our sample was powered to detect a correlation of r: 0.1489 at 80 % power and 0.05 significance level. Secondly, we used Kappa statistics to assess the ability of parent reports to place individuals in tertiles of habitual TPA measured by accelerometry: strength of agreement was classified using Landis and Koch: <0.00 ‘poor agreement’; 0.00–0.20 ‘slight agreement’; 0.21–0.40 ‘fair agreement’; 0.41–0.60 ‘moderate agreement’; 0.61–0.80 ‘substantial agreement’; and 0.81–1.00 ‘almost perfect agreement’.
We also assessed the ability of the questionnaire results to classify participants as meeting (sensitivity) or not meeting (specificity) the TPA guideline by reporting percentage agreement. Sub-group analyses were conducted by the various socio-demographic characteristics (i.e., sex, education class, residential area, and country income level) to explore differences in correlation and classification accuracy between parent-reported TPA and accelerometer-measured TPA. Finally, we used Bland–Altman plots to evaluate bias between parent-reported habitual TPA measure and accelerometry by plotting the difference between the two methods against accelerometry (the criterion method).
We also used Bland–Altman plots to calculate ‘limits of agreement’ (LOA, i.e., mean bias ±1.96 SD). Additionally, we used Pearson's correlation to test for systematic bias between the difference and the criterion method.
We validated the three proposed step-count thresholds by calculating sensitivity, specificity, and area under the receiver operating characteristic curve (ROC-AUC) for Gabel, Vale & De Craemer step-count thresholds using steps derived from activPAL™.
The ROC-AUC provides a measure of classification accuracy by graphically plotting the y-axis as true positive rate (sensitivity) and x-axis as false positive rate (1 − specificity). ROC-AUC values were defined as excellent (0.9–1.0), good (0.8–0.9), fair (0.7–0.8), or poor (<0.7).
Statistical significance was determined at 5 %. All analyses were performed in Stata/IC v.16.1 for Mac (StataCorp, College Station, Texas, USA) except for ROC-AUC which was performed using SPSS v.27 for Mac (IBM Corp, Armonk, NY, USA).
3. Results
We included 352 pre-schoolers aged 3.0–4.9 years from 13 countries: 3 lower-middle, 5 upper-middle, and 5 high-income countries (Supplementary Table A online). Descriptive characteristics of participants are presented in Table 1. The proportions of boys and girls included in the current study were similar. A slight majority (59.1 %) lived in urban areas and over two-thirds (67.0 %) were from lower- and upper-MICs. The mean age was 4.4 (SD: 0.3) years. Children accumulated an average of 119 (SD: 32) min/day of TPA and accumulated an average of 8784 steps (SD: 2548) as measured by activPAL™. Using the parent reports, children achieved an average of 188 (SD: 127) min/day of TPA.
Table 2 shows rank-order correlations between parent-reported TPA and accelerometer-measured TPA across various socio-demographic characteristics of participants. There was a very weak but statistically significant positive correlation between parent-reported TPA and accelerometer-measured TPA (r: 0.140; p = 0.009). When stratified by various socio-demographic characteristics, correlations ranged from very weak to weak (r: 0.034–0.233). Correlations were statistically significant for boys, participants from highly educated families, and those from MICs.
Table 1Descriptive characteristics of study participants (as frequency and percentage unless specified).
Derived variable based on World Bank classification.
HIC
0.112
0.233
67.2
66.7
67.3
0.050
MIC
0.156
0.016
50.4
80.0
49.8
0.024
Note: HIC = high-income countries; MICs = middle-income countries.
Agreement means the proportion of children who were accurately classified by both the parent reports and accelerometry (the criterion method) as meeting or not meeting the TPA guidelines.
Sensitivity means proportion of children who are accurately classified as meeting the TPA guidelines by parent reports.
Sensitivity means proportion of children who are accurately classified as not meeting the TPA guidelines by parent reports.
Overall, there was slight agreement between accelerometer-measured TPA and parent-reported TPA (κ: 0.030) (Table 2). With the various socio-demographic characteristics considered, there remained slight agreement between accelerometer-measured TPA and parent-reported TPA, except among girls where there was disagreement between the two methods (κ: −0.012). Parent reports showed an overall sensitivity of 75.0 % and specificity of 55.2 % for meeting 180 min of TPA guideline per day. When stratified by the various socio-demographic groups, parent reports showed sensitivity of 0.0 %–100.0 % and specificity of 49.6 %–67.3 % for meeting the TPA guideline.
The Bland–Altman plots (Fig. 1) demonstrated an over-estimation of habitual child TPA time from parent reports compared to the activPAL™ measurement (mean bias: 69 min/day; SD: 126; 95 % limits of agreement [LOA]: −179, 316). As shown in Fig. 1, most parents with less active children over-reported their child's habitual TPA. There was also systematic bias in the measurement of child TPA by parent reports compared to accelerometer data, as parents tended to over-report their child's habitual TPA to a larger extent in less active children (r: −0.106; p = 0.047).
Fig. 1A ‘modified’ Bland–Altman plot between accelerometer-measured child habitual TPA and parent-reported child habitual TPA. The figure shows mean bias (middle solid line) of 69 min/day in (over-)estimation of a child's TPA by parent reports compared to accelerometer measurement and its associated lower and upper limits of agreement (below and above the mean bias line, respectively). The dots indicate that over-reporting of a child's TPA by parent reports was higher among less active children.
step-count cut-point with an AUC of 0.945 (95 % CI: 0.928, 0.961), 100.0 % sensitivity and 88.9 % specificity (Supplementary Fig. A online). The Vale et al.
Our findings suggest that simple parent-reporting of child TPA is not likely to be adequate for global surveillance of the WHO physical activity guideline for pre-schoolers. However, we found that step-counting, using the De Craemer et al.
found a weak correlation of 0.39 (95 % CI: 0.19, 0.56) between parent-reported child habitual TPA and accelerometer-measured child habitual TPA, which is higher than the correlation in our study (r: 0.14; p = 0.009). Unlike Sarker et al.
which used Actical accelerometers, our study used activPAL™ accelerometers which have been validated for measurement of TPA against direct observation in this age group.
had a mixed sample, involving infants, toddlers, pre-school and schoolchildren aged 4–70 months and did not report correlation/agreement specifically for each age group. The present study only included pre-schoolers aged 3–4 years that may have different activity patterns as well as spend less time with their parents compared to younger children in the Sarker et al.
; whereas our study included participants from lower- and upper-middle-income countries and with diverse culture and lifestyles. Nevertheless, the results of the current study are consistent with a previous review of validation studies of physical activity measures in children, even though most validation studies included in the review involved children older than those participating in our study.
The poor/weak correlation in validation studies of parent-reported physical activity could be caused by parents over-reporting their child's physical activity due to social desirability
or because they do not know how much activity their child participates in during weekdays when they are at an ECEC centre for example.
Despite the importance of TPA in early childhood for current child health and development, and future health according to the WHO Ending Childhood Obesity (ECHO) Report
Whilst proxy reports from parents are simple and cheap for monitoring physical activity in young children, our results suggest that they are not likely to be valid for global public health surveillance of physical activity in early childhood. Consequently, parent questionnaires may not be suitable for monitoring compliance with the WHO physical activity guidelines in early years. Given the need for a globally validated physical activity measurement for surveillance purposes to monitor progress towards the global targets, and based on our results, step-counting may be a more accurate alternative to parent reports. At the moment, there is currently no consensus on culturally and geographically valid step-count targets for classifying 3 h of TPA in pre-schoolers.
be used for assessing compliance with meeting the TPA guideline in early years because it is geographically and culturally valid against TPA measured by activPAL™. There are barriers to using accelerometers in population-based studies as they are intrusive, require complicated data reduction and analysis, produce huge data sets, and are expensive,
at about a minimum of $254 USD per device up to >$1000 USD, depending on the device. As such, surveillance studies and national surveys in future could potentially use much simpler, cheaper devices like pedometers to assess the prevalence of compliance with WHO guidelines. Pedometers are highly correlated to accelerometer step-counts in young children
; however, exact step-count thresholds should be established as being accurate in the population that they are being used before being included as surveillance measures due to potential differences between methods of measuring step-counts (including differences due to the placement of the device, e.g. with activPAL™ worn on the thigh and pedometers usually worn on the hip).
A strength of our study was the relatively large sample of young children compared to previous validation studies among this age group.
Moreover, to our knowledge, this is the first study to assess validity of parent reports of their children's physical activity based on a sample from vastly differing contexts, including lower- and upper-MICs as well as HICs. This is also the first study to cross-validate existing step-count thresholds in such a varied sample of pre-schoolers.
Our study had some limitations. Recruitment of participants in the SUNRISE study was determined independently in each country due to the varying contexts in which the study was conducted, including use of convenience cluster sampling.
Cross-sectional examination of 24-hour movement behaviours among 3- and 4-year-old children in urban and rural settings in low-income, middle-income and high-income countries: the SUNRISE study protocol.
The sample was not representative. However for a methodological study the main requirements are adequate sample size, wide range of settings, and a range of levels of TPA from low to high, and all these requirements were met in our study. We did not have any low-income country (LIC) study participants as classified by the World Bank. However, our study included participants from Bangladesh which is classified as a LIC according to the Organisation for Economic Co-operation and Development's (OECD) Development Assistant Committee (DAC; https://www.oecd.org/dac/financing-sustainable-development/development-finance-standards/daclist.htm) and also included three lower-MICs. The parent-reported questions in our study were based on available physical activity, sedentary behaviour, and sleep guidelines for the early years,
A collaborative approach to adopting/adapting guidelines - the Australian 24-Hour Movement Guidelines for the early years (birth to 5 years): an integration of physical activity, sedentary behavior, and sleep.
and alternative questions might have higher validity. In addition, it is possible that the observed bias in over-reporting of a child's TPA by parents varied between countries due to cultural differences; however, with our sample size we were not adequately powered to explore differences in biases by country of origin. As such, a further study focusing on potential differences in the biases in physical activity reporting by parents across countries may be useful. We identified a gender difference in accuracy of parent reporting of TPA (more accurate in the boys than girls) – the reasons for this are not clear and may be worth investigating further in future. However, correlations between parent-reported TPA and accelerometer-measured TPA, though statistically significant, were very low in the boys in the present study and so the practical significance of this gender difference is probably quite limited – validity of parent reporting was low in both boys and girls. Further, we used a research-grade accelerometer to test the three existing step-count thresholds and therefore cheaper pedometers need to be validated before use. Lastly, there was a suggestion of a possible gender difference in the classification accuracy as there was disagreement between the two methods in measurement of habitual TPA among girls which also requires further research.
5. Conclusions
Despite the importance of TPA in early childhood for current child health and development, and future health according to the WHO ECHO Report
there is currently no global surveillance of TPA in early childhood, and one major barrier to a surveillance system is cost and complexity of the measurement method. The present study provides evidence that parent reports may have limited validity for this purpose as parents cannot recall their child's physical activity adequately, at least not using fairly simple questions. However, our study also shows that step-counts may be an accurate and relatively simple, potentially low-cost, alternative to assessment of progress towards meeting the global physical activity targets in this age group.
The following are the supplementary data related to this article.
ROC curve for the three existing step-count thresholds. A ROC-AUC value of 0.9–1.0 is considered excellent classification whereas a value <0.7 is considered poor classification. Ideally, a low value on the x-axis (1 − specificity) and a high value on the y-axis (sensitivity) means a ROC-AUC value closer to 1.0. Compared to the other two step-count thresholds, the De Craemer et al. threshold of 11,500 steps/day shows excellent classification as it maximised the sensitivity (true positives) and minimised the false positives (1 − specificity) in classifying children as meeting or not meeting the 2019 WHO guideline of 180 min/day of TPA in pre-schoolers.
List of countries where participants for the SUNRISE study were recruited, by income level.
CRediT authorship contribution statement
TM-V, XJ, and JJR contributed to the conceptualisation, data analysis and interpretation, drafting, review, and final approval of the manuscript. ADO, MST, CED, AAF, CT, DK, GH, HKT, KHC, ML, MSH, PC, and PWPC contributed to data acquisition as well as reviewed and approved the manuscript.
Funding information
This work was supported by the Sir Halley Stewart Trust (grant number 2674); the University of Wollongong Australia and a National Health and Medical Research Council of Australia Investigator Grant (grant number APP1176858); the Canadian Institutes of Health Research Planning and Dissemination (grant number 392396); the Universiti Kebangsaan Malaysia Research University Grant (grant number GUP-2018-142); Brazilian National Council for Scientific and Technological Development (CNPq) (grant number 309301/2020-3); Pham Ngoc Thach University of Medicine's Fund for Science (grant number # 1320/HĐ-TĐHYKPNT); the Sasakawa Sports Research Grant from Sasakawa Sports Foundation (grant number 190A2-004); the Biomedical Research Foundation, Bangladesh; the Faculty of Health Sciences at the University of the Witwatersrand; and the Dr Stella de Silva Research Grant at the Sri Lanka College of Paediatricians. The views presented in this work are solely the responsibility of the author(s) and do not necessarily represent the views of the funding sources.
Confirmation of ethical compliance
The SUNRISE study protocol was reviewed and approved by Human Research Ethics Committee at the UOW (2018/044) and ethics committees in each participating country; all parents of participating children gave informed consent.
Declaration of interest statement
None.
Acknowledgment
The SUNRISE study data were collected and managed using REDCap electronic data capture tools hosted at University of Wollongong. Our thanks also go to PAL Technologies (Glasgow, Scotland) for support for the purchasing of activPAL™ and the analysis of the data. Finally, we wish to thank the SUNRISE Coordinating Centre staff at Early Start, UOW and participated countries' research teams for their support. The authors have no financial or other interest in the products or distributors of the products used in this study.
References
World Health Organization
Report of the Commission on Ending Childhood Obesity.
Cross-sectional examination of 24-hour movement behaviours among 3- and 4-year-old children in urban and rural settings in low-income, middle-income and high-income countries: the SUNRISE study protocol.
A collaborative approach to adopting/adapting guidelines - the Australian 24-Hour Movement Guidelines for the early years (birth to 5 years): an integration of physical activity, sedentary behavior, and sleep.
00437-6&id=gr1.jpg){kind=link}