Our findings suggest that acutely violating the preliminary measurement BIA assumptions does not significantly impact the derived %FM and impedance values. In general, women appear to have more variability in their BIA measures than men, particularly with dehydration using the hand-to-foot device. However, these minor differences in the measurements were similar among all participants regardless of their obesity status. The magnitude of the differences between trials are less than what is expected with day-to-day variation (< 5%).
The use of BIA devices to assess body composition is common in health and fitness facilities and research studies. Although the preliminary measurement BIA assumptions are well known, they are often not optimally followed in practice, particularly in the general public. It is recommended that BIA measurements be avoided when participants are dehydrated, within 4-h of food and beverage consumption, and within several hours of moderate-to-strenuous exercise [6] as these factors may influence the derived body composition assessment.
In terms of water and food intake, the literature is inconsistent on the magnitude and even direction of change (14–19). Similar to some studies [14, 15], we report non-statistically significant differences in %FM of ~ 1%. However, even in studies that report statistically significant differences with water and/or food intake, the magnitude of these differences are within ranges expected with day-to-day variation (generally < 2%) [16, 17][18]. Further, there were no consistent differences between studies that do or do not report significant differences in %FM in terms of diet composition, with high fat, high carbohydrate and ad libitum food intakes most commonly examined. The composition of the diet theoretically may influence body impedance and the rate of gastric emptying, however, one study reports that impedance values are similar immediately after consuming a meal and many hours later [19]. We extend these observations by demonstrating that impedance measures after food and water intake are not significantly different using either hand-to-hand, hand-to-foot or foot-to-foot BIA machines. Together, this suggests that food and/or water intake is unlikely to have a meaningful impact on impedance measured body fat assessment.
The non-voided bladder condition did not significantly change the impedance or %FM values when compared to the control trial. Although, the consumption of 1L of water did increase body mass it was not enough to statistically increase %FM. In this study, 1 kg difference in body mass is theoretically associated with a 0.68% higher FM which is in line with a previous study that suggests a non-voided bladder could affect BIA measurements by up to 1% [20]. Thus, non-voided bladder is likely to have minimal effects on %FM estimates.
There are several changes that occur with exercise such as changes in skin blood flow, temperature, heat production and fluid loss [19], that may increase or decrease impedance. The literature on the effects of exercise on estimated %FM and impedance is mixed with studies showing decreased impedance by 28–40 Ω [21], or no change in impedance following moderate intensity aerobic exercise [21–23] as observed in this study. In the literature, the largest differences observed are less than 1% FM even with exercise intensity of 60 to 83% HRmax for as long as 45 minutes. These minimal differences suggest that moderate intensity exercise is unlikely be associated with large differences in predicted %FM.
For dehydration, theoretically one would expect low fluid status would result in an increase in impedance and thus increase in predicted %FM. In this study, impedance was not significantly increased in the dehydrated condition, and in fact trended in the opposite direction (Women: -23.2%; Men: -9.8%) and %FM (Women: -1.9%; Men: 1.2%). The lower %FM is likely due to the reduction of average body mass of -0.74 kg among participants. A study conducted by Thompson et al 1991 also report a significant decrease in %FM in the dehydrated state after exercising for 30 minutes and sitting in a steam room when compared with control, though the exact %FM difference was not reported [24]. However, that study had a much larger decrease in body mass (average of 2.81%) than was observed in our study (< 1%). Further, we report that the differences in BIA measured %FM may be large in women than men. Reasons for this are unclear but may be due to differences in %FM and fat distribution. That said, it is important to consider that we did not observe differences by BMI category. Despite our large range in BMI (20.2 to 37.8 kg/m2) the difference in %FM that resulted by violating the preliminary BIA assumptions are similar among all participants regardless of their BMI categories. Thus, future work may consider the potential sex differences in how these factors, and in particular dehydration may influence body composition assessment. Nevertheless, it is important to remember that the vast majority of values observed in this study were within 5% of control. Further, the < 2% differences we observe in this study is far lower than the 15 to 19.5% reduction in FM that are reported in exercise intervention even with minimal weight loss [25].
Further, these measures were generally comparable to DXA and SOS assessments. The exception was the Omron HBF (hand-to-hand model) in women where the %FM values were significantly lower than DXA and SOS. This reinforces the notion that %FM obtained cannot be directly compared between the various devices, but also suggests that the acute violation of the core BIA assumptions may not have a large influence on the %FM obtained regardless of the measurement site used. Further, these variations in %FM are far smaller than what one would expect with clinical weight loss interventions [26].
Some strengths and limitations of this study are worth mentioning. We are one of the few studies to examine the effect of violating the core BIA assumptions on the estimation of body composition among multiple BIA devices. In the current study, three BIA devices with different measurement sites were used. Although there are several different devices available on the market, they all use measures of impedance and body weight to assess body composition. That we also observe no differences in impedance suggest that these observations likely hold true for other BIA devices using different algorithms. However, we are unsure if the differences in body composition would be larger if more than one core BIA assumption was violated at the same time. We are also unsure if our results extend to older individuals or populations with chronic conditions. Finally, we have a relatively small sample of 40 adults, and retrospective power analyses suggest that 182 participants are needed for the largest difference (-1.9%FM) to be significant. Nevertheless, the clinical relevance of these differences of this magnitude even if significant are questionable as they are comparable to be what would be expected with the 2 to 5% day-to-day variation (18,33).
It can be concluded that preliminary measurement BIA assumptions have a very small effect (< 2%) on the derived %FM and impedance values. Women tend to have larger variability in %FM measures than men. Nevertheless, these differences associated with acutely violating the core BIA assumptions are far smaller than what would be expected with weight loss interventions and is within what is expected with day-to-day variation.