Regular physical activity (PA) supports healthy development and overall well-being in early childhood.^1–5 The World Health Organization (WHO) recommends that children aged 1–5 years engage in at least 180 minutes of total PA per day, including moderate- to vigorous-intensity activities.⁵ Although accelerometers are widely recognized as valid tools for assessing PA in children,^6–10 accelerometer-based assessments are costly and often impractical due to participant burden and logistical constraints in large-scale surveillance or clinical settings.^11,12 Therefore, there is a growing need for a brief, practical tool to evaluate young children’s PA that can be implemented efficiently in population monitoring and clinical practice.^13–15

Several caregiver-reported PA questionnaires for young children exist;¹² however, few have been validated specifically for toddlers aged 1–2 years. For example, the Preschool-age Physical Activity Questionnaire (Pre-PAQ)¹⁶ was developed and validated for preschool-aged children (3–5 years); however, it was not designed for toddlers. The Patient-Reported Outcome Measurement Information System (PROMIS) Early Childhood PA scale was more recently developed to target children aged 1–5 years;¹⁷ yet, our prior work found low convergent validity and notable floor effects among toddlers.¹⁸

The Movement Behavior Questionnaire-Child (MBQ-C)¹⁹ is a newly developed caregiver-reported tool designed to rapidly assess 24-hour movement behaviors (i.e., PA, sleep, and screen time) among children under 5 years who are able to walk independently. The MBQ-C PA domain consists of only 4 items, offering a brief and practical approach well-suited for large-scale studies or clinical settings. Initial validation of the MBQ-C among Australian children demonstrated good test-retest reliability and acceptable convergent validity for the PA domain.¹⁹ In a subsequent validation study among Chinese kindergarteners, test-retest reliability was only moderate, and convergent validity was acceptable;²⁰ however, findings were largely limited to children aged 4–5 years. Given the substantial developmental differences in movement patterns between toddlers and preschool-aged children, further validation in toddlers is essential before broader application in this age group. To fill this research gap, the present study examined the test-retest reliability and convergent validity of the MBQ-C PA domain among toddlers.

Movement Behavior Questionnaire-Child (MBQ-C) Physical Activity Domain. The MBQ-C PA domain includes 4 items asking caregivers to think about the past week and report the amount of time their child engaged in active play on a typical weekday (Q1A) and a typical weekend day (Q2A), and in energetic play (worded as “vigorous activities”) on a typical weekday (Q1B) and weekend day (Q2B). Vigorous activities were described as running, jumping, dancing, and riding bikes or scooters.

Of the two available MBQ-C versions (open-ended and closed-ended), we used the closed-ended version, because its rapid scoring allows for immediate feedback, making it a practical option for clinical and primary care settings.¹⁹ Response categories for active play were: 0 minutes, 1–30 minutes, 30–60 minutes, 1–2 hours, 2–3 hours, 3–4 hours, and more than 4 hours per day. Response categories for energetic play included: 0 minutes, 1–15 minutes, 15–30 minutes, 30–60 minutes, 1-1.5 hours, 1.5-2 hours, and more than 2 hours per day.

To assess test-retest reliability, mothers were instructed to complete the MBQ-C PA domain twice: first via an online survey link sent before their child’s accelerometer wear period, and again using a paper version on the final day of wear. Responses were converted to minutes per day using the midpoint of each categorical range (e.g., 0 minutes for “0 minutes”; 15 minutes for “between 1 and 30 minutes”; 45 minutes for “between 30 and 60 minutes”; 90 minutes for “between 1 and 2 hours”).¹⁹ Daily active play and energetic play minutes were computed as (weekday minutes x 5 + weekend day minutes x 2) ÷ 7.¹⁹

Accelerometer assessment. To examine convergent validity, we used ActiGraph wGT3X-BT accelerometers (100 Hz; Ametris LLC; Pensacola, Florida, USA). Children wore the device on their hip for 7 consecutive days. Accelerometer data were downloaded and integrated into a 15-second epoch using the ActiLife software (version 6.14). Two data processing approaches were applied. First, the ActiLife software was used to estimate cut-point based PA levels. Data recorded between 6:00 AM and 10:00 PM were extracted.^23–25 Non-wear time, defined as ≥ 20 consecutive minutes of zero counts,^26–29 was excluded. A valid day required ≥ 480 minutes of wear time within the 16-hour window.^26,30,31 Participants with ≥ 4 valid days were included in analysis.³¹ For intensity classification, we applied toddler-specific cut-points⁶ rather than the random forest algorithm³² used in the original MBQ-C validation,¹⁹ because that algorithm has not been validated for toddlers. Light-intensity PA (LPA) was defined as 25–417 vertical counts per 15 seconds, and moderate- and vigorous-intensity PA (MVPA) as > 417 counts per 15 seconds.^6,28 Daily minutes of MVPA and total PA (LPA + MVPA)^27,28 were computed for each participant.

Second, data were processed using the GGIR R package^33,34 to compute Euclidean Norm Minus 1g (ENMO).^35,36 ENMO is a device-independent metric that provides a robust, comparable measure of movement intensity that is independent of device cut-points, allowing for cross-study comparisons. Non-wear periods were identified using GGIR’s default settings, which detects sustained periods of low acceleration variability. Missing values during non-wear periods were imputed using GGIR’s built-in imputation procedure.^33,34 Mean ENMO values were averaged across valid days to compute mean acceleration (mg), representing overall PA volume.³⁶

Other measurements. Mothers reported demographic and household information, including the child’s sex, age, race, ethnicity, maternal education, and childcare attendance. Residential addresses were used to assign a Child Opportunity Index (COI) score, a validated indicator of neighborhood resources and conditions relevant to a child’s health and development.³⁷ COI scores were categorized into five levels: very low, low, moderate, high, and very high. WHO weight-for-length percentiles were calculated³⁸ using anthropometric data obtained from the child’s most recent clinic visit summary.

A total of 72 mother-child dyads were enrolled in this ancillary study. One dyad was excluded due to insufficient accelerometer data and no retest MBQ-C data, resulting in 71 dyads for analysis. Children had a mean age of 24 (standard deviation [SD] = 4) months (range: 16–34 months), and all were able to walk independently. Mothers had a mean age of 35 (SD = 4) years (range: 23–41 years). The average accelerometer wear time was 14.9 (SD = 1.3) hours/day within the 6 AM-10 PM window. Accelerometer-derived mean acceleration was 21.0 (SD = 4.5) mg (Table 1). Means of total PA and MVPA were 245 (SD = 42) and 79 (SD = 23) minutes/day, respectively. As shown in Table 2, accelerometer-derived PA estimates tended to be lower among 1-year-olds than 2-year-olds (p = 0.03–0.13). In contrast, MBQ-C PA estimates tended to be higher among 1-year-olds than 2-year-olds (p = 0.07–0.19).

Test-retest reliability. Test-retest reliability was evaluated among 64 participants, excluding 7 whose test-retest interval exceeded 14 days. Among the 64 included participants, the test-retest interval ranged from 4 to 14 days (median of 7 days; interquartile range of 4–9 days). Among those who completed the retest within 7 days, the ICC was 0.45 (95% confidence interval [CI] = 0.23, 0.67) for active play minutes and 0.58 (95% CI = 0.40, 0.77) for energetic play minutes (Table 3). Among those who completed it within 14 days, the ICC was 0.32 (95% CI = 0.13, 0.52) for active play minutes and 0.58 (95% CI = 0.43, 0.73) for energetic play minutes.

Validity. The proportions of respondents endorsing “0 minutes” for Q1A, Q1B, Q2A, and Q2B were 1%, 9%, 0%, and 10%, respectively. No respondents selected the highest category (“more than 4 hours” for active play or “more than 2 hours” for energetic play). In the convergent validity analysis, MBQ-C-derived active and energetic play estimates were not significantly correlated with accelerometer-derived mean acceleration, total PA, or MVPA (p > 0.05; Fig. 1).

This study examined the test-retest reliability and convergent validity of the MBQ-C PA domain among toddlers aged 1–2 years. Energetic play demonstrated moderate test-retest reliability (ICC = 0.58), whereas active play showed poor test-retest reliability (ICC = 0.45). No to minimal floor and ceiling effects support the measure’s feasibility for use in this age group. However, correlations between MBQ-C-derived PA estimates and accelerometer-derived PA metrics were negligible, indicating limited convergent validity of the MBQ-C PA domain for toddlers.

Comparison of PA levels with prior MBQ-C validation studies. In this toddler sample, MBQ-C-derived active play estimates (107 minutes/day at retest) were lower than those reported in the previous MBQ-C validation studies: 193 minutes/day in an Australian sample (n = 54)¹⁹ and 128 minutes/day in a Chinese kindergarten sample (n = 211).²⁰ Similarly, MBQ-C-derived energetic play estimates (35 minutes/day) were lower than 75 minutes/day¹⁹ and 69 minutes/day²⁰ observed in those respective samples. Accelerometer-derived PA levels in our study (245 minutes/day of total PA; 79 minutes/day of MVPA) were also substantially lower than the Chinese sample (approximately 500 minutes/day of total PA; 250 minutes/day of MVPA), despite our use of lower cut-points applied to hip-worn accelerometer data. Direct comparisons with the Australian sample are limited due to different wear locations (hip vs. wrist). While the CAMPAS cohort is considered relatively active compared with toddler samples in the US³¹ and Australia²³ and with estimates from meta-analysis,⁴¹ its PA levels remained far below those observed in prior MBQ-C validation studies of older children, highlighting age-related differences in daily movement patterns between toddlers and preschool-aged children.

Comparison of test-retest reliability with prior MBQ-C validation studies. Our test-retest reliability estimates (ICC = 0.45–0.58) were markedly lower than those reported in the Australian sample by Trost et al. (ICC = 0.86 to 0.88)¹⁹ and somewhat lower than those reported in the Chinese sample by Song et al. (ICC = 0.52 to 0.66).²⁰ The lower ICCs in our study could partly be explained by a longer test-retest interval. Trost et al.¹⁹ administered both the test and retest after accelerometer wear within a short 3-day interval, whereas both our study and Song et al.²⁰ conducted the initial test before accelerometer wear and the retest after a 7-day or longer interval. Additionally, the lower ICCs could also be explained by significantly lower PA levels reported at retest as compared to the initial test. Because the retest occurred approximately 7 days after the initial test, it is possible that children were genuinely less active during the retest week. Another potential explanation is that mothers may have reduced over-reporting at retest due to increased awareness that their child’s PA was monitored with an accelerometer during that period. These design differences likely contributed to the lower test-retest reliability observed in our study compared with Trost et al.¹⁹

Convergent validity. This study found that correlations between MBQ-C-derived and accelerometer-derived PA were negligible, in contrast to prior findings from Trost et al. (ρ = 0.25–0.39)¹⁹ and Song et al. (ρ = 0.35).²⁰ The discrepancy may stem from the challenges conceptualizing toddlers’ active play and energetic play and estimating their play durations. At this age, PA often consists of brief, intermittent bursts rather than discrete and sustained activity bouts.^42,43 This activity pattern makes it challenging to estimate intensity and duration of PA.^42,43 Consistent with developmental expectations, accelerometer-derived PA levels were higher among 2-year-olds than 1-year-olds; however, MBQ-C-derived PA estimates showed the opposite pattern, resulting in small, but nonsignificant, negative correlations (e.g., ρ = -0.18 between MBQ-C-derived active play and accelerometer-derived MVPA). This mismatch illustrates inherent difficulties caregivers face in estimating intensity and duration of toddler movement.

Challenges in caregiver-reported PA assessment for toddlers. In response to the growing need for brief, practical PA assessment tools in young children, new caregiver-reported instruments have been recently developed, including PROMIS Early Childhood PA scale.¹⁷ The PROMIS PA scale is distinctive in that it captures children’s physiological responses to PA and quantifies the number of active days per week (e.g., “In the past 7 days, how many days did your child so physically active that he/she sweated?”; “How many days did your child play so hard that he/she felt tired?”¹⁷) rather than the amount of daily PA. Our validation of the PROMIS Early Childhood PA scale among toddlers found moderate test-retest reliability and poor convergent validity with accelerometer measures.¹⁸ Unlike PROMIS EC, the MBQ-C quantifies daily duration of active and energetic play, aligning more directly with WHO 24-hour movement guidelines⁵ and the emerging 24-hour movement behavior paradigm,⁴⁴ thereby offering conceptual advantages for surveillance. Nonetheless, the present findings indicate that among 1- to 2-year-olds, the MBQ-C PA domain demonstrates only modest reliability and poor correspondence with accelerometer-derived measures. Subjective recall of intensity and duration is prone to estimation bias without clear benchmarks for what constitutes “active” and “vigorous” play among toddlers.⁴² Rapid PA assessment among toddlers may need to incorporate alternative proxy indicators, such as PA environments and parenting practices.⁴⁵

Limitations. This ancillary study leveraged the existing CAMPAS protocol, in which the initial MBQ-C administration occurred before accelerometer deployment. In addition, the test-retest interval was longer than optimal for reliability evaluation. These study design aspects may have contributed to the lower ICCs. Second, we evaluated only the PA domain of the MBQ-C; because 24-hour time-use diary data were unavailable we could not evaluate the validity of the screen time or sleep domains. Finally, the study relied exclusively on maternal reports and fathers were not included, which may limit the generalizability of findings to broader caregiver contexts.

Among children aged 1–2 years, the MBQ-C PA domain demonstrated poor-to-moderate test-retest reliability and very low convergent validity relative to accelerometer-derived PA metrics. These findings underscore the inherent challenges of caregiver-reported PA assessment in toddlers whose movement patterns are highly sporadic and unstructured. Rapid PA assessment among toddlers may need to incorporate alternative proxy indicators, such as PA environments and parenting practices. While the MBQ-C PA domain has shown acceptable performance in preschool-aged children, its use for toddlers needs further validation in population-based samples across diverse counties and cultural contexts. Adaptation or modification of the MBQ-C PA domain may be necessary before it can be applied for PA assessment among children aged 1–2 years.