Table 3.
Optimal steps/day from ROC analysis and Colley et al. thresholdsa for censored and uncensored steps by activity cut point for children and adolescents
| |
|
Children |
|
|||
|---|---|---|---|---|---|---|
| |
Freedson 3 METS |
Freedson 4 METS |
Evenson |
|||
| Optimal | Colleya | Optimal | Colleya | Optimal | Colleya | |
|
Censored |
|
|
|
|
|
|
| Boys 6–11 yrs |
6,500 |
12,000 |
9,500 |
12,000 |
10,500 |
12,000 |
| FP/FNb |
(0%/6%) |
(0%/68%) |
(17%/21%) |
(0%/59%) |
(19%/20%) |
(5%/39%) |
| Accuracy |
.94 |
.35 |
.80 |
.56 |
.81 |
.80 |
| AUC |
.99 (.98–1.0) |
.91 (.88–.94) |
.90 (.87–.93) |
|||
| Girls 6–11 yrs |
6,500 |
12,000 |
9,000 |
12,000 |
10,500 |
12,000 |
| FP/FNb |
(0%/15%) |
(0%/83%) |
(15%/21%) |
(1%/74%) |
(21%/15%) |
(6%/42%) |
| Accuracy |
.85 |
.19 |
.81 |
.55 |
.80 |
.86 |
| AUC |
.88 (.85–.91) |
.90 (.88–.93) |
.88 (.85–.92) |
|||
|
Uncensored |
|
|
|
|
|
|
| Boys 6–11 yrs |
8,500 |
12,000 |
12,000 |
12,000 |
13,000 |
12,000 |
| FP/FNb |
(9%/6%) |
(0%/33%) |
(20%/21%) |
(20%/21%) |
(26%/21%) |
(43%/9%) |
| Accuracy |
.94 |
.68 |
.79 |
.79 |
.76 |
.72 |
| AUC |
.99 (.97–1.0) |
.88 (.85–.92) |
.86 (.83–.90) |
|||
| Girls 6–11 yrs |
8,500 |
12,000 |
11,500 |
12,000 |
13,500 |
12,000 |
| FP/FNb |
(0%/9%) |
(0%/48%) |
(24%/19%) |
(17%/27%) |
(22%/27%) |
(42%/11%) |
| Accuracy |
.91 |
.53 |
.79 |
.77 |
.77 |
.65 |
| AUC |
.97 (.95–1.0) |
.87 (.84–.90) |
.84 (.80–.89) |
|||
| |
|
Adolescents |
|
|||
| |
Freedson 3 METS |
Freedson 4 METS |
Evenson |
|||
| |
Optimal |
Colleya |
Optimal |
Colleya |
Optimal |
Colleya |
|
Censored |
|
|
|
|
|
|
| Boys 12–17 yrs |
8,500 |
12,000 |
11,500 |
12,000 |
10,500 |
12,000 |
| FP/FNb |
(21%/21%) |
(1%/68%) |
(11%/14%) |
(8%/22%) |
(15%/21%) |
(3%/34%) |
| Accuracy |
.79 |
.61 |
.88 |
.90 |
.83 |
.90 |
| AUC |
.99 (.97–1.0) |
.93 (.91–.96) |
.93 (.91–.95) |
|||
| Girls 12–17 yrs |
7,500 |
12,000 |
9,000 |
12,000 |
9,500 |
12,000 |
| FP/FNb |
(23%/16%) |
(0%/82%) |
(18%/19%) |
(4%/52%) |
(12%/13%) |
(2%/41%) |
| Accuracy |
.79 |
.78 |
.82 |
.95 |
.88 |
.95 |
| AUC |
.88 (.86–.91) |
.87 (.78–.96) |
.93 (.88–.98) |
|||
|
Uncensored |
|
|
|
|
|
|
| Boys 12–17 yrs |
11,000 |
12,000 |
14,000 |
12,000 |
13,000 |
12,000 |
| FP/FNb |
(28%/24%) |
(16%/38%) |
(15%/17%) |
(34%/6%) |
(21%/19%) |
(29%/13%) |
| Accuracy |
.74 |
.72 |
.84 |
.70 |
.80 |
.75 |
| AUC |
.84 (.82–.87) |
.91 (.88–.94) |
.89 (.87–.92) |
|||
| Girls 12–17 yrs |
10,500 |
12,000 |
11,500 |
12,000 |
12,000 |
12,000 |
| FP/FNb |
(24%/24%) |
(9%/52%) |
(23%/19%) |
(19%/24%) |
(17%/18%) |
(17%/18%) |
| Accuracy |
.76 |
.79 |
.77 |
.81 |
.83 |
.83 |
| AUC | .85 (.81–.88) | .84 (.73–.95) | .90 (.83–.96) | |||
a Colley et al. thresholds are 12,000 steps/day across gender and age groups.
bFP/FN is false positives and false negatives.
Accuracy = (true positives + true negatives)/(true positives + true negatives + false positives + false negatives).
AUC = area under the curve.