SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. DISCLOSURE
  7. References

The purpose of this study was to derive ActiGraph cut-points for sedentary (SED), light-intensity physical activity (LPA), and moderate-to-vigorous physical activity (MVPA) in toddlers and evaluate their validity in an independent sample. The predictive validity of established preschool cut-points were also evaluated and compared. Twenty-two toddlers (mean age = 2.1 years ± 0.4 years) wore an ActiGraph accelerometer during a videotaped 20-min play period. Videos were subsequently coded for physical activity (PA) intensity using the modified Children's Activity Rating Scale (CARS). Receiver operating characteristic (ROC) curve analyses were conducted to determine cut-points. Predictive validity was assessed in an independent sample of 18 toddlers (mean age = 2.3 ± 0.4 years). From the ROC curve analyses, the 15-s count ranges corresponding to SED, LPA, and MVPA were 0–48, 49–418, and >418 counts/15 s, respectively. Classification accuracy was fair for the SED threshold (ROC-AUC = 0.74, 95% confidence interval = 0.71–0.76) and excellent for MVPA threshold (ROC-AUC = 0.90, 95% confidence interval = 0.88–0.92). In the cross-validation sample, the toddler cut-point and established preschool cut-points significantly overestimated time spent in SED and underestimated time in spent in LPA. For MVPA, mean differences between observed and predicted values for the toddler and Pate cut-points were not significantly different from zero. In summary, the ActiGraph accelerometer can provide useful group-level estimates of MVPA in toddlers. The results support the use of the Pate cut-point of 420 counts/15 s for MVPA.

The prevalence of overweight and obesity among US children aged 5 years and under has more than tripled since the 1970's. Data from the 2007–2008 National Health and Nutrition Examination Survey (NHANES) indicate that 21% of children between the ages of 2 and 5 are overweight or obese (1). Low levels of physical activity (PA) are important contributing factor in the development and maintenance of obesity (2,3). However, methodological challenges related to the assessment of PA in young children have significantly hindered research efforts to quantify, understand, and promote PA in this population (4,5).

Given the limitations of child and parent self-report instruments and the high cost and participant burden associated with other objective assessment methods, accelerometry has become a popular method for measuring PA in children under five (5,6). Indeed, over the last decade, a considerable amount of research has been conducted to establish the validity of accelerometry in preschool-aged children and to identify intensity-related count thresholds for estimating time spent in sedentary (SED), light, and moderate-to-vigorous PA (5,6). To date, however, no cut-points have been developed for toddlers (<3 years) and the question of whether intensity-related thresholds established for preschool-aged children are valid for toddlers has not been adequately explored in the research literature.

Therefore, the aims of this study were to: (i) derive ActiGraph cut-points for SED, light, and moderate-to-vigorous PA (MVPA) in toddlers; and (ii) evaluate the predictive validity of the new toddler cut-points and cut-points developed for preschool-aged children in an independent sample of toddlers.

Methods

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. DISCLOSURE
  7. References

Participants and setting

Study participants were recruited in two waves from two licensed child care centers. The first wave of participants served as the validation and calibration sample and comprised 22 toddlers (14 girls, 8 boys) between the ages of 16 and 35 months (mean age = 2.1 years ± 0.4 years). The second wave of participants served as the cross-validation sample and comprised 18 toddlers (10 girls, 8 boys) between the ages of 16 and 35 months (mean age = 2.3 ± 0.4 years). None the participants had any limitations that restricted their participation in active play or structured PA. The protocol for this study was reviewed and approved by the Kansas State University institutional review board. Before participation, the parent or guardian of each child provided written informed consent.

Procedures

Participants were videotaped during a regularly scheduled 20-min play period. During the play period each child wore an ActiGraph GT1M accelerometer (Shalimar, FL) in a customized vest which secured the monitors snuggly against the child's trunk just at the level the right hip.

PA intensity during the free play session was coded using a modification of the Children's Activity Rating Scale (CARS) (7). In the original CARS scheme, PA intensity is coded as 1 = lying down or sitting, 2 = standing, 3 = walking slow/easy, 4 = walking moderate, 5 = running, strenuous activity. In the present study, a four-category scheme was implemented in which 1 = lying down or sitting, 2 = standing, 3 = walking, 4 = running. Activity intensity was coded directly from the video tape recording using a computerized direct observation software tool. The videotaped play sessions were coded by two trained research assistants. To assess interobserver agreement, five randomly selected participants were independently coded by the two observers. The intraclass correlation for the average weighted CARS score was 0.95.

A weighted average CARS score corresponding to each 15-s epoch of ActiGraph output was calculated by multiplying each numeric activity code by the percentage of the 15-s time interval in that code and summing the products (8). Scores of less than 2.0 were classified as SED, while scores greater than and equal to 3.0 were classified as MVPA. Scores between 2.0 and 2.99 were classified as light-intensity PA (LPA).

Statistical analyses

Receiver operating characteristic (ROC) curves were used to identify the count thresholds providing the highest sensitivity and specificity for differentiating: (i) SED from LPA and MVPA; and (ii) MVPA from SED and LPA. Classification accuracy was evaluated by calculating area under the ROC curve (ROC-AUC). An area of 1 represents perfect classification, whereas an area of 0.5 represents an absence of classification accuracy. ROC-AUC values of ≥0.90 are considered excellent, 0.80–0.89 good, 0.70–0.79 fair, and <0.70 poor (9).

To evaluate the performance of the toddler cut-points in the cross-validation sample, time spent in SED, LPA, and MVPA during the 20-min play session was estimated from the accelerometer data and compared with direct observation. Cut-points developed for preschool-aged children (10,11,12,13) were also applied to the accelerometer data and compared to direct observation. Differences between directly observed and accelerometer-based PA were evaluated for statistical significance using Wilcoxon rank sum test for paired samples.

Results

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. DISCLOSURE
  7. References

The optimal count thresholds for differentiating SED and MVPA from other intensity levels were 48 and 418, respectively. Thus, the 15-s count ranges corresponding to SED, LPA, and MVPA were 0–48, 49–418, and >418 counts/15 s, respectively. Classification accuracy was fair for the SED threshold (ROC-AUC = 0.74, 95% confidence interval = 0.71–0.76) and excellent for MVPA threshold (ROC-AUC = 0.90, 95% confidence interval = 0.88–0.92).

Table 1 displays the mean differences and 95% limits of agreement between directly observed and accelerometer predicted PA in the cross-validation sample. The toddler cut-point and the preschool cut-points significantly overestimated time spent in SED. Mean differences ranged from 6.3 min for the NHANES 100 cpm cut-point to 13.1 min for the Van Cauwenberghe cut-point. The toddler cut-point and the preschool cut-points significantly underestimated time in spent in LPA. Mean differences ranged from 6.7 min for the Pate cut-point to 11.7 min for the Van Cauwenberghe cut-point. For MVPA, all four cut-points provided underestimations; however, mean differences for the toddler and Pate cut-points were not statistically significantly different from zero. Across all levels of PA intensity, the limits of agreement associated with each cut-point were wide, indicating a large degree of individual variation in the difference between observed and predicted values.

Table 1.  Activity estimates and mean differences between observed and predicted time in each activity intensity
inline image

Discussion

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. DISCLOSURE
  7. References

The present study is the first to identify and cross-validate intensity-related ActiGraph cut-points for toddlers under the age of 3 years. The derived cut-point for MVPA (418 counts/ 15 s) exhibited excellent classification accuracy in the calibration study; and within the cross-validation sample, yielded group-level MVPA estimates that were not significantly different from directly observed MVPA levels. Importantly, this cut-point was almost identical to the preschool MVPA cut-point of 420 established by Pate and colleagues (12), suggesting that a cut-point of 420 counts/15 s may be useful for identifying episodes of MVPA in both toddlers and preschoolers.

With the exception of the Pate cut-points, the toddler cut-points were substantially lower in magnitude than those established for preschoolers. Predictably, application of these cut-points in the cross-validation sample resulted in substantial underestimations of LPA and MVPA, and large overestimations of SED activity. Our findings for the MVPA differ from those of Van Cauwenberghe et al. (14) who found the Pate cut-point to significantly underestimate time in MVPA in Belgian toddlers. This discrepancy may be related, in part, to the group's use of the OSRAC-P direct observation system which coded the highest activity level during each 15-s epoch regardless of its duration. In the present study, we employed a real-time direct observation system which calculated a duration-based weighted average of all PA intensity levels completed within each 15-s epoch.

The assessment of SED behavior via accelerometry has become a topic of considerable research interest (15). The present study derived a toddler-specific cut-point for SED behavior and compared it to those previously established for preschoolers. The widely applied cut-point of 100 counts/min (NHANES) was also compared. Within the cross-validation sample, all five cut-points significantly overestimated time in SED activity, with the NHANES cut-point providing the least bias. Notably, the preschool cut-points established by Reilly et al. (10), Sirard et al. (11), and Van Cauwenberghe et al. (13) overestimated SED time by over 10 min. These findings indicate that more research is needed to refine the assessment of SED behavior in young children. In the interim, the use of the 25 counts/15-s threshold for SED may be the most practical choice among toddlers and preschoolers.

Strengths of this study include the inclusion of a cross-validation group and the use of a validated real-time direct observation system as a criterion measure. Limitations include the relatively small sample, the limited duration of the play session, and the low levels of MVPA during each play session. Future studies should recruit larger samples and employ more structured activity trials with more opportunities to engage in MVPA. Because all cut-points exhibited large prediction errors at the individual level, future studies should explore the use of pattern recognition or other advanced data processing approaches to estimate activity type or intensity.

In summary, the ActiGraph can provide useful group-level estimates of MVPA in toddlers. The results support the use of NHANES cut-point for SED and the Pate cut-point of 420 counts/15 s for MVPA.

DISCLOSURE

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. DISCLOSURE
  7. References

The authors declared no conflict of interest.

References

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. DISCLOSURE
  7. References