Skip to main content

Table 3 Cadence thresholds (steps/min) for absolutely-defined moderate and vigorous intensity based on regression and ROC curve analyses

From: Walking cadence (steps/min) and intensity in 61–85-year-old adults: the CADENCE-Adults study

Intensity METs

Measure

Regression thresholds

ROC thresholds

Heuristic thresholds

Value

95% PI

Value

95% CI

Value

3

Threshold (steps/min)

103.1

70.0–114.2

100.3

95.7–103.1

100

Se

81.6

86.8

86.8

Sp

86.6

84.5

83.3

PPV

80.5

79.2

78.0

NPV

87.4

90.4

90.3

Accuracy

84.6

85.5

84.8

AUC

0.93

0.91–0.95

4

Threshold (steps/min)

116.4

105.3–127.4

111.5

106.1–112.9

110

Se

68.2

88.2

89.4

Sp

93.5

87.3

84.1

PPV

65.2

55.1

50.0

NPV

94.3

97.7

97.8

Accuracy

89.7

87.4

84.9

AUC

0.94

0.92–0.96

5

Threshold (steps/min)

129.6

118.6–140.7

116.0

112.4–120.2

120

Se

44.4

96.3

81.5

Sp

98.5

87.9

91.6

PPV

60.0

28.6

32.8

NPV

97.2

99.8

99.0

Accuracy

95.9

88.3

91.1

AUC

0.96

0.94–0.98

6

Threshold (steps/min)

142.9

131.8–148.4

128.6

128.3–136.4

130

Se

33.3

100.0

83.3

Sp

99.8

97.1

97.5

PPV

66.7

27.3

26.3

NPV

99.3

100.0

99.8

Accuracy

99.1

97.2

97.3

AUC

0.99

0.98–1.00

  1. Segmented regression and Receiver Operating Characteristic (ROC) thresholds are represented as means (95% Prediction Intervals) for segmented regression and means (99% Confidence Intervals) for ROC. Trade-offs in terms of Sensitivity (Se), Specificity (Sp), Positive Predictive Value (PPV), Negative Predictive Value (NPV) and overall accuracy between the thresholds derived from the segmented regression and ROC analyses were considered to select heuristic thresholds. Selected heuristic thresholds reflect a purposely favored tolerance for false-negative versus false-positive classifications. AUC = Area under the curve, CI = Confidence Intervals, PI = Prediction Intervals