Other than age, anthropometric indices and one socio-demographic variable (p≤0.0144), there were no significant differences between BMI groups across basic descriptive characteristics. There was a non-significant tendency for a lower subsistence score in the OW/OB group (p=0.083) (Table 1).
[Place Table 1 here]
Walking volume indices (total daily steps and bouts, maximum bout length) were not significantly different between BMI groups (p≥0.0629) (Table 2). Compared with OW/OB participants, UW/NW participants displayed significantly higher peak 1-minute walking cadences, spent more time walking at cadences ≥100 steps.min-1, and accumulated more bouts and longer maximum bouts at ≥100 steps.min-1 (p≤0.0463). OW/OB participants spent less time in sedentary situations (0 steps.min-1), but more time in situations resulting in incidental steps (1-19 steps.min-1), than UW/NW participants (p≤0.0112). There was a non-significant tendency for OW/OB participants to choose a more varied bout length (higher S2) and longer walking bouts tended to contribute more to patterns of step accumulation (higher G), compared with UW/NW participants (p≥0.0861). In contrast, in UW/NW participants, walking parameters (S2, G) for bouted cadence were significantly higher compared with OW/OB participants (p≤0.005). In other words, UW/NW participants displayed a more varied bouted cadence choice, and higher bouted cadences contributed more to the accumulation of steps, compared with OW/OB participants (Table 2).
[Place Table 2 here]
For both BMI groups accumulated steps were relatively equally distributed in the Light and Moderate-2-to-Vigorous categories and ≈45% of daily steps were accumulated in the Moderate-1 accelerometer band (760-1951 cts.min-1) (Figure 1A). There were no significant differences between the BMI groups for steps accumulated in the four accelerometer count bands (p≥0.0911).
[Place Figure 1 here]
More than 85% of the participants were classified as active-to-very active, irrespective of BMI status (Figure 1B). There was no association between BMI category and step-defined activity categories (p=0.7958).
BMI groups did not differ for average daily steps (Figure 1C, p=0.2937) and there was no difference (p=0.5633) in the average daily steps between overweight (BMI 25-29.9 kg.m-2) and obese (BMI ≥30 kg.m-2) participants (13 489 steps.day-1 and 13 153 step.day-1, respectively).
For both BMI groups, a greater percentage of total steps were accumulated in longer walking bouts (75% of total steps at >15 minute bout length), than shorter, more frequent bouts (see Additional File 2: Figure S1 A).
The majority of the step accumulation occurs at the lower end of the bouted cadence spectrum (see Additional File 2: Figure S1 B). Noticeably, in the UW/NW group, in the range of 50-90% of the percentage of total steps accumulated, steps were accumulated at higher bouted cadences.
When ignoring walking bouts, and only considering the minute-by-minute accumulation of steps (unbouted), a greater number of steps are accumulated at higher walking cadences (see Additional File 1: Figure S1 C) compared with bouted cadence steps (see Additional File 1: Figure S1 B). The difference between the BMI groups was not as marked as for the bouted cadence step accumulation pattern in Figure S1 B (see Additional File 1).
There was a general similarity in the distribution of walking bout length and frequency patterns across cadence bands (see Additional File 3: Figure S2 A-C). Approximately 95% of walking bouts were accumulated in the lowest two walking cadences (1-39 steps.min-1), with approximately 82% of walking bouts accumulated in the lowest cadence (1-19 steps.min-1) primarily through fewer longer bout lengths. Walking cadences ≥60 steps.min-1 were associated with reductions in walking bout length and frequency, and the trend was consistent across BMI groups. Walking bouts were rare at cadences ≥100 steps.min-1 with a frequency of ≤5 bouts over the monitoring period (see Additional File 3: Figure S2 A-C).
In bivariate analyses, walking bout length S2 and G were not significantly associated with age, BMI, disease presence, PAI or any of the components of PAI (p≥0.063). Bouted cadence S2 and G were significantly and inversely associated only with BMI (p<0.001). Age, presence of disease, availability of electricity in the dwelling and PAI were significantly associated with unbouted cadence S2 and G (p≤0.038). Hence, forced multiple linear regression models were run with age, presence of disease and availability of electricity or PAI as independent variables, and unbouted cadence S2 and G as dependent variables. All model assumptions were met. Independent of age (p=0.140) and the presence of disease (p=0.054), availability of electricity in the dwelling was inversely and significantly associated with unbouted cadence S2 (p=0.040, β = -0.0392) (Model: p=0.0042, adjusted R2 = 0.0386). Similarly, independent of age (p=0.132), the presence of disease (p=0.049, β = +0.01149) and availability of electricity in the dwelling (p=0.042, β = -0.01344) were significantly associated with unbouted cadence G (Model: p=0.0037, adjusted R2 = 0.0395). PAI was not significant in any model (p≥0.061).