Table 5 Characteristics and results of studies that examined validity and accuracy of other accelerometers and inclinometers for measuring sedentary behaviour in older community-dwelling, healthy adults (mean age ≥ 60 years), ordered from largest to smallest sample size

Hutto, et al., 2013 [37]

n: 200

85 males, 115 females (58% female)

Mean age = 63.5 ± 8.3

USA: Data collected as subs-tudy of The Reasons for Geographic and Racial Differences in Stroke Study, a national cohort study of racial and regional disparities in stroke risk and mortality

Actical

Worn on waist

60-s epochs

Free-living

Activities: VA ≤100 cpm

Concurrent measure: wear-time logs

Observation period: 7 consecutive days during waking hours

Valid hours and days: ≥10 h; ≥4 days

Non-wear algorithms: 60-, 90-, 120-, 150-, and 180-min of consecutive zeros, no allowance for interruptions

Analysis: compared algorithms for estimating wear and non-wear time by computing min/day and % of total wear time in SB for each algorithm

ALGORITHMS USING THE FOLLOWING MINUTES OF CONSECUTIVE ZEROES TO MEASURE NON-WEAR TIME

Min/day were classified as SB

60-min: 618 ± 81

90-min: 649 ± 88

120-min: 667 ± 97

150-min: 675 ± 98

180-min: 679 ± 101

% of total wear time was SB

60-min: 75 ± 10

90-min: 77 ± 10

120-min: 77 ± 10

150-min: 77 ± 10

180-min: 77 ± 10

Klenk et al. (2016) [38]

n: 53

31 males and 22 females (41.5% female)

Mean age: 75.3 ± 4.6 y

Germany: Data collected as sub-study of ActiFE-Ulm study, a national cohort study of physical activity and health outcomes

ActivPAL3

Worn on left thigh

Laboratory-based

Activities: 2 bouts of sitting; 2 bouts of lying

Concurrent measure: ActivPAL worn on left thigh concurrently

Observation period: 10 s per bout of activity with a total observation period of mean 156.5 min ± 16.5

Analysis: computed agreement between the 2 monitors by (1) using Bland Altman methods and (2) computing for the activPAL sitting/lying category the degree to which activPAL3™ identified sitting/lying or different activities

Mean difference: − 2.00 s (±2 SD: 3.52)

Median agreement: 98.0% (IQR 95.9–99.0)

Expected difference in SB duration (95% CI) for 24-h measurement: − 44.5 min (− 69.9, − 20.0)

Wullems et al. (2017) [39]

n: 40

20 males, 20 females (50.0% female)

Mean age = 73.5 ± 6.3 y

UK: Convenience sample

2 GENEActiv

Worn on thigh

Laboratory-based

Activities: sitting in a chair; lying down

Observation period: 4 min per activity

Criterion: indirect calorimeter

Analysis: compared machine learning algorithm to three cut-point algorithms for classifying intensities of activities (if MET value ≤1.5 and position was not upright), computed sensitivity and specificity

3 methods using cut-point algorithms

Sensitivity: 99.3–99.9%

Specificity: 99.7%

Accuracy: 99.5–99.8%

Random Forest machine learning

Sensitivity: 99.9%

Specificity: 99.2%

Accuracy: 99.6%

Other findings: Participant-specific accuracies resulted in perfect score (100%) for all algorithms for SB.

Landry et al., 2015 [40]

n: 23

7 males and 16 females (69.6% female)

Mean age: 70.0 ± 6.6 y

Canada: Convenience sample recruited through newspaper advertisements, brochures distributed at community centres, and word of mouth

MotionWatch 8

Two worn on non-dominant wrist

60-s epochs

Laboratory-based

Activities: sitting in a chair; lying down

Criterion: portable calorimeter

Observation period: 5-min per activity

Analysis: used ROC to determine optimal cut-points with SB defined as < 1.5 MET, computed AUC, sensitivity, specificity, PPV, negative predicted value

AUC: 0.81 (95%CI: 0.78, 0.85)

Optimal cut-point for SB: ≤178.5 (d² = 0.14)

Sensitivity: 78%

Specificity: 70%

Accuracy: 70%

PPV: 30%

Negative predictive value: 94%