Journal Logo


Impact of Acute Dietary Manipulations on DXA and BIA Body Composition Estimates


Author Information
Medicine & Science in Sports & Exercise: April 2017 - Volume 49 - Issue 4 - p 823-832
doi: 10.1249/MSS.0000000000001148


Body composition assessment is a critical part of medical, health, and fitness research (2,23,36). Because each method of body composition assessment provides only an estimate of an individual's actual body composition, it is important to minimize the errors inherent to estimation. Furthermore, inadequate reporting of preassessment procedures in reports of research leaves the reader to question whether relevant potential errors were accounted for or simply ignored. Dual-energy x-ray absorptiometry (DXA) and bioelectrical impedance analysis (BIA) are two methods of body composition assessment that are frequently used in research settings. It has previously been demonstrated that DXA body composition estimates are influenced by body fluid alteration (15,30) and food or fluid ingestion (15,25,38). These studies indicate that the lean soft tissue compartment may be particularly impacted by these acute changes. Likewise, BIA estimates are altered in response to food and fluid ingestion (10,32). These acute alterations in body composition estimates may be due to increases in total body fluid or alterations in its distribution, gastrointestinal content, or changes in electrolyte concentrations (14,20). Ultimately, the current evidence indicates that ingested matter is largely “seen” as lean body mass, potentially due to the water content of foods and fluids (10,15,25,32,38). However, it is unclear whether the specific foods consumed play a role in the observed effects.

The International Society for Clinical Densitometry’s Official Positions warn that information derived from DXA scans is dependent on a number of factors, including hydration status and prior food intake (14). A common recommendation to combat this unnecessary variability is to conduct body composition assessment after an overnight fast (25). Although this is likely preferable to measuring body composition later in the day after activities of daily living have occurred, this recommendation makes the assumption that an overnight period without food is sufficient to negate nutritional differences from the previous day(s). However, the habitual diet, as well as acute intake, before body composition assessment could potentially impact the values obtained, even after an overnight fast. For DXA, stores of muscle glycogen and associated water are included in the lean soft tissue compartment. While stores of glycogen vary with body size, dietary intake and fitness status, a 70-kg man may store approximately 500 g of glycogen in his body, although greater amounts can be stored in response to glycogen loading (1). When the additional 3 to 4 g of water per gram of glycogen are considered (27), there are several kilograms of body weight that can be acutely modified (i.e., depleted or bolstered), in addition to alterations due to overall hydration.

BIA can be influenced by hydration and variation in body fluids (21,37). Several researchers have reported that bioelectrical impedance values are decreased by acute food and fluid intake, which leads to underestimations of body fat (10,32). The observed decline in impedance values is likely due to increased total body water as a result of the water content of fluids and food items. Different electrode arrangements could potentially impact the relationship between body water and impedance changes. Hand-to-hand, foot-to-foot, and hand-to-foot electrode arrangements are commonly used for BIA. The arms and legs disproportionately affect whole-body impedance values due to their small cross-sectional area relative to the trunk (20,29). This, combined with the anatomical contact points of the electrodes, indicates that hand-to-hand devices are excessively influenced by the composition of the arms, whereas foot-to-foot devices are excessively influenced by the composition of the legs. If body water of the arms and legs are differentially impacted by acute food and fluid intake due to gravitational differences in relation to the body position during measurement or differences in the amount of ingested fluid accommodated by each compartment, BIA body composition estimates obtained through different electrode arrangements may also be differentially affected. In addition, regional body composition differences between the arms and legs could potentially impact the obtained estimates.

A recent study by Rouillier et al. (31) examined the effect of following a nonstandardized high-carbohydrate diet for 3 d on measures of body composition using DXA. Total body weight and lean body mass increased (0.8% and 1.5%, respectively), percent body fat decreased, and a 4.5% increase in appendicular lean mass was reported. Since these very short-term changes do not likely represent a true change in body composition, it seems apparent that the acute diet led to the observed alterations, even after an overnight fast (5). However, there were several noteworthy limitations in Rouillier et al.'s study: only male subjects were included, there were no measures of body water content during the dietary program, only a single method of body composition assessment was used, a single diet was used (i.e., a high-carbohydrate diet), and the dietary intervention was unstandardized. Specifically, participants were given a list of high-carbohydrate foods that they were encouraged to eat, but no food items were provided and no restrictions were given. The present study was designed to provide more detailed information about the impact of acute dietary manipulations on body composition estimates without these limitations.

Although a small number of previous studies have reported the impact of food and fluid ingestion on individual body composition assessment technologies, to our knowledge there have been no direct comparisons of the impact on estimates obtained by multiple assessment techniques. In addition, prior studies contained a single and sometimes unstandardized dietary condition, which eliminates the ability to determine whether the content of the diet plays a role in acutely altering body composition estimates. Therefore, the purpose of this study was to examine the effects of standardized high-carbohydrate (HC) and very low carbohydrate (VLC) diets on body composition estimates obtained by DXA, tetrapolar foot-to-foot BIA, and bipolar hand-to-hand BIA in recreationally active male and female adults.



This experiment utilized a counterbalanced design consisting of two 2-d conditions to examine the effects of acute dietary manipulations on body composition estimates provided by DXA and BIA (Fig. 1). Participants underwent two dietary conditions: HC diet and VLC diet. The first day of each condition consisted of baseline measurements in the morning and the consumption of a standardized diet throughout the day. In the afternoon of the first day, participants returned to the laboratory to repeat the assessment procedures. The second morning of each condition consisted of identical experimental procedures as the first morning, and the condition was completed after these assessments. After a washout period, each participant completed the second dietary condition. This design allowed an examination of the effects of two drastically different diets, both soon after consumption of food items and after an overnight fast, on body composition estimates.

Study Design. Each participant underwent body composition assessment (BCA) a total of six times. BCA assessment consisted of DXA, tetrapolar BIA, and bipolar BIA. For all morning visits (four total), each volunteer reported to the lab after a 12-h overnight fast (F). The two dietary conditions were very low-carbohydrate (VLC: 1 to 1.5 g·kg−1 carbohydrate) and high-carbohydrate (HC: 9 g·kg−1 carbohydrate). Half of the subjects began the study with the HC diet, while half began with the VLC diet. A washout period of 5 d for males and approximately 1 month for females was placed between dietary conditions.


The target population of this study was recreationally active and active adult males and females. Inclusion criteria for the study were that participants must be between the ages of 18 and 30, generally healthy (i.e., no known disease or chronic health problem), premenopausal with regularly occurring menstrual cycles over the past 6 months for female participants, and recreationally active or active (i.e., > 120 min·wk−1 of self-reported moderate or vigorous physical activity). There was no specific body mass index requirement for participants. Potential participants were excluded from the study if they failed to meet any of the inclusion criteria, were currently pregnant or may have become pregnant during the study, or had started or stopped taking the dietary supplement creatine within the past month. Due to the provision of food items as part of the study, potential participants were also excluded from participation in the study if they had a known or suspected food allergy to any of the food items utilized. In addition, individuals were not allowed to participate if they had any medical condition that could reasonably be negatively affected by participation in the study (e.g., diabetes, hyperlipidemia, and so on).

This study was approved by the university’s institutional review board. Before any experimental procedures, participants were screened for potential eligibility and read and signed a university-approved written informed consent document. A medical history form was also completed to screen for additional potential contraindications to participation in the study.

Formulation of diets

Both dietary conditions were formulated based on carbohydrate content relative to body weight. The HC condition consisted of approximately 9 g·kg−1 body weight of carbohydrate. Similar high levels of carbohydrate consumption are frequently used in carbohydrate loading programs (3,8,11,12), and this level of intake has been reported to significantly increase muscle glycogen content after a single day of consumption (8). The VLC condition consisted of approximately 1 to 1.5 g·kg−1 body weight of carbohydrate. It is questionable whether the short-term absence of dietary carbohydrate reduces glycogen content, as some have reported no decline (19,40) while others have reported modest reductions (16). Because of this, the VLC condition served as an energy-matched comparison diet, without the expectation that glycogen content would be reduced by this short period of VLC intake.

While dietary carbohydrate was the focus of the diet formulation, the total caloric content of each diet condition was kept within 4% of the caloric content of the other condition. The percentage of energy from carbohydrate, fat, and protein was approximately 82%, 10%, and 8% for the HC diet and 13%, 67%, and 20% for the VLC diet. Each participant was categorized by baseline body weight to prescribe the appropriate carbohydrate and energy content of the diet. The following weight categories were utilized (values in kilograms): <45, 45–55, 55–65, 65–75, 75–85, 85–95, 95–105, 105–115, and >115.

Order of dietary conditions

Before the commencement of the first experimental condition, the initial participants were randomly assigned by coin flip to undergo either the VLC or HC condition first. Once an equal number of participants had been randomly assigned to undergo the VLC or HC condition first, a permuted block method was used to assign the remaining participants to ensure an equal number of participants began each condition first.

Scheduling of visits relative to menstrual cycle

Female subjects were interviewed concerning their menstrual cycle to schedule research visits within the follicular phase. Both experimental conditions were completed within 5 d after cessation of menstrual bleeding. One condition was completed per month in two consecutive months.

Dietary considerations

A 1-d diet and fluid intake record was collected before the first day of each experimental condition. Before the second condition, participants were provided with their diet and fluid intake record from before the first experimental condition and asked to follow this pattern of intake as closely as possible. Participants were asked to fast (i.e. consume no food or fluids) for 12 h before each morning research visit, as well as refrain from consuming alcohol, nicotine, or caffeine from 12 h before the first visit of a particular condition until the completion of the final visit of that same condition. Consumption of these items was allowed during the washout period, but was restricted again beginning 12 h before the first visit of the second condition. Participants were asked not to consume any dietary supplements during the experimental conditions, beginning 1 d before the first research visit.

Physical activity

Participants were asked to refrain from moderate to vigorous physical activity starting the evening before the first day of a given condition. Participants were also asked to not engage in any exercise (i.e. minimize physical activity beyond activities of daily living, such as walking) throughout the first day and the morning of the second day until after the dietary condition was completed. Participants were encouraged to continue normal physical activity during the washout period, provided that they refrain from moderate to vigorous physical activity beginning the evening before the second condition.

Reporting to the laboratory

For each visit, participants reported to the laboratory between 6 AM and 10 AM after a 12-h overnight food and fluid fast. Participants provided researchers with their diet record from the previous day upon reporting to the laboratory. Each participant was required to report to the laboratory at approximately the same time for each morning visit of a particular condition, and efforts were made to conduct all morning visits throughout both conditions at approximately the same time of day. Participants were asked to wear minimal clothing (e.g., athletic shorts and a T-shirt) and to wear equivalent clothing items during all visits. Participants were asked to refrain from wearing any items that contained metal.

For the second visit of each condition, each participant reported to the laboratory in the afternoon between 2 and 5 PM after consuming the midafternoon snack (see Provision of Food Items). The time since the most recent consumption of the provided food items was recorded. The experimental procedures were conducted as in the morning session, with the exception that the participants were not reporting to the laboratory in the fasted state.

Body weight and composition assessment

Upon arrival, each participant was asked to void his or her bladder before obtaining body weight. Body weight and height were determined on an electronic scale and stadiometer (Seca 703). Participants underwent a DXA scan on a calibrated Hologic Discovery W machine. Participants were positioned according the National Health and Nutrition Examination Survey (NHANES) recommendations: body supine, head straight, small amount of space between the arms and torso, hands flat on the table, and feet together (26). All participants were able to fit within the scanning dimensions of the device, both breadthwise and lengthwise. The same researcher analyzed all DXA scans using the Hologic APEX software (version 3.3) to promote consistency between measurements. To demarcate body regions, the analysis lines were placed as depicted in the NHANES Body Composition Procedures Manual: horizontal lines inferior to the skull and at the level of the iliac crest, vertical lines adjacent to the vertebral column and between the legs, and diagonal lines through both glenohumeral joints and femoral necks (26). The following within-subject coefficients of variation for a fan-bean DXA have been reported: 0.1% for total mass, 0.5% for total lean mass, and 1.6% for total fat mass when two scans are conducted in the same assessment session, and 0.5% for total mass, 0.8% for total lean mass, and 1.7% for total fat mass when scans are conducted on consecutive mornings (25).

The effective dose of radiation reported for a whole body scan on the Discovery W system is 8.4 μSv in adults (4). According to the NHANES Body Composition Procedures Manual, the average effective dose of background radiation that an individual in the United States receives each year is 3600 μSv, and the dose of radiation an individual receives from a DXA scan falls within the range of background radiation (26). In addition, an individual would receive approximately 1100 μSv from a standard diagnostic x-ray of the spine (26). The NHANES Body Composition Procedures Manual reasons that the risk from DXA scans is low. Therefore, even when six DXA scans were conducted over the course of a week to a month, the amount of effective radiation imposed on each participant was low.

BIA estimates of body composition were collected using tetrapolar foot-to-foot (Tanita SC 331S) and bipolar hand-to-hand (Omron HBF-306) devices. Participants stood upright for both measurements, but the arms were at the side of the body for foot-to-foot measurements and were held out from the body, forming a 90° angle between the torso and arms, for the hand-to-hand measurements. Measurements were collected in triplicate for each BIA device, and the values were averaged. While absolute differences between BIA and DXA have been described, BIA devices similar to those used in the present study have been reported to be able to detect body composition changes over time (24,39).

Provision of food items

At the end of the morning research visit on the first day of either condition, participants were given commercially packaged food items for consumption that day in accordance with their body weight. For the HC condition, food items included individually packaged fruit juices, pasta, fruit snacks, granola bars, and fruit bars. For the VLC condition, food items consisted primarily of individually packaged nuts, cheeses, hard-boiled eggs, dried meats, and low-carbohydrate shakes. The provision of commercially prepackaged food items minimized the risk of contamination and allowed for more precise quantification of nutrient content within each dietary condition. The nutritional content of the experimental diet for each weight class is depicted in Table 1.

Nutritional content of experimental diets.

Participants were asked not to consume any other food items other than those provided and to consume all of the food items provided to them by research personnel. The food items were assigned to be consumed in five occasions: morning meal (7–10 AM; immediately after the first research visit), midday meal (11 AM–1 PM), midafternoon snack (2–4 PM), evening meal (5–7 PM), and nighttime snack (7–9 PM). The timing of the nighttime snack was altered as needed to allow for a 12-h fast before the participant returned to the laboratory for assessment the following morning.

Tracking adherence to dietary protocol

Participants were given a dietary checklist of the food items provided to them to mark each food item they consumed. Although participants were instructed not to consume any food in addition to the food items provided to them, they were also asked to record any additional items they consumed if this did occur. Participants were also required to return uneaten food items. In addition, although participants were allowed ad libitum noncaloric, noncaffeinated fluid intake, they were asked to record all fluid intake on a provided form. This form included images to serve as a guide in estimating quantities of fluids consumed.

Analysis of diet records

Diet records for the day before each experimental condition were analyzed using the freely available United States Department of Agriculture SuperTracker tool (35). The total amounts of calories, carbohydrate, fat, protein, sodium, and fluid for each diet record were obtained. One trained researcher performed all dietary analysis to promote consistency of results.

Washout period

It has been demonstrated that glycogen levels can remain elevated for 5 d after carbohydrate loading (3). Therefore, the washout period in male subjects was 5 d. To ensure that both conditions were completed at the same time of the menstrual cycle, female participants completed one condition immediately after the cessation of menstrual bleeding in two consecutive months (i.e., two consecutive menstrual cycles). Both male and female washout periods were as short as possible in an attempt to minimize real changes in body composition. Participants were instructed to attempt to continue their regular dietary intake and physical activity habits to maintain a stable body weight during the washout period.

Statistical procedures

Power analyses were conducted using reported changes in lean body mass from previous research (7,13,31). These analyses indicated a required sample size of at least 18 participants of each sex. This sample size is also similar to a previous study examining changes in DXA estimates of body composition induced by acute daily activities, including food intake (25).

Descriptive characteristics of study participants were generated. Paired samples t tests were performed to compare anthropometric characteristics of participants at the baseline visit for each condition, dietary intake in participants the day before beginning each condition, total fluid intake during each experimental condition, and the time between last food consumption and experimental visits. Independent samples t-tests were used to compare baseline characteristics of male and female participants.

Three-factor repeated-measures analysis of variance was performed for dependent variables using time and condition as within-subjects factors and sex as the between-subjects factor. An alpha level of 0.05 was used to determine statistical significance. Significant two-way interactions were followed up via simple main effects. For two-way interactions between two within-subjects factors (i.e., time and condition), repeated measures analysis of variance was utilized to compare each time point in each condition and to compare each condition at each time point. The resultant pairwise comparisons were examined for significant differences. For two-way interactions with one within-subjects factor and one between-subjects factor (i.e., sex), the simple main effect for the within-subjects factor was determined via repeated measures analysis of variance and examination of the resultant pairwise comparisons. For the between-subjects factor, independent samples t tests were used to compare sexes. Significant time main effects were examined using post hoc pairwise comparisons.

Partial η2 effect sizes and 90% confidence intervals for effect sizes were calculated using scripts generated by Smithson (33) in IBM SPSS Statistics (version 20.0.0). Partial η2 represents the proportion of the effect plus error variance that is attributable to the effect for a given dependent variable. To ensure that the confidence interval for a proportion of variance effect size, such as partial η2, does not include zero when there is a significant difference in an F-test using an alpha level of 0.05, it is necessary to use a 90% confidence interval (22,34). Statistical analysis was conducted using IBM SPSS Statistics (version 20.0.0).



Fifty-four individuals (27 males and 27 females) were screened for eligibility. Two individuals did not meet entry criteria due to food allergies or lack of menstrual cycle. Four individuals were eligible, but never commenced the study. Forty-eight individuals (24 males and 24 females) initiated and completed all aspects of the study. The completion rate was 100%. Baseline characteristics of study participants are presented by sex in Table 2. There were no differences in baseline body weight, lean soft tissue, fat mass, body fat percentage measured by DXA, or total body water between the HC and VLC conditions (see Table, Supplemental Digital Content 1, Baseline Visit Comparison,

Participant characteristics.

Dietary intake

Dietary intakes on the day before each experimental condition were not significantly different (P > 0.05; see Table, Supplemental Digital Content 2, Participant Characteristics, There was a 94.8% adherence rate for following the food intake plan exactly as prescribed. Noncompliance included consuming very small amounts of additional food (n = 1) and failing to finish a very small amount of the provided food items (n = 4). Uneaten food items were returned to study investigators. Deviations from the protocol were deemed small by study investigators, and the decision was made not to exclude these individuals from the analysis. There was no difference between conditions for the duration between last food or fluid consumption and experimental visits (see Table, Supplemental Digital Content 3, Comparison of Duration since Last Food or Fluid Consumption Prior to Each Study Visit,

Body weight and body composition

No three-way interactions (i.e., sex by dietary condition by time) were present for any dependent variables. A time by dietary condition interaction was present for body weight as measured by electronic scale (P = 0.031, ηp2 = 0.136). Body weight increased in both conditions in response to feeding (+0.7 kg in HC and +0.5 kg in VLC) and returned to baseline values in HC, but not VLC. On the second morning of the VLC condition, body weight was 0.2 kg below the initial visit. No other interactions or main effects involving the dietary conditions were present.

Time main effects were present for body mass, total lean soft tissue, and total fat mass as measured by DXA (Table 3). Estimates of total mass, total lean soft tissue mass, and trunk lean soft tissue mass increased in response to feeding by 0.6, 0.8, and 0.6 kg, respectively, whereas total fat mass decreased by 0.2 kg. Each of these parameters returned to baseline values by the final research visit within each condition. Based on ηp2 values, variation in total mass, total lean soft tissue, and trunk lean soft tissue due to time accounted for 79%, 76%, and 74% of the variation due to time plus error. Although time main effects were also present for leg lean soft tissue and trunk fat mass, no effects were observed for arm lean soft tissue, nor leg or arm fat mass. As a result of feeding, estimates of leg lean soft tissue increased by 0.2 kg and trunk fat mass decreased by 0.1 kg.

DXA body composition results.

A time by sex interaction was present for body fat percentage measured by DXA (P = 0.02, ηp2 = 0.138). In males, body fat percentage decreased by 0.3 percentage points in response to feeding, but returned to baseline values after the second overnight fast. In females, body fat percentage decreased by 0.6 percentage points in response to feeding, and then increased 0.9 percentage points after the second overnight fast so that body fat percentage was higher after the second overnight fast as compared to after the first. However, the variation in body fat percentage due to the sex by time interaction only accounted for 14% of the variation in the interaction plus error.

Time main effects were present for impedance, total body water mass, fat-free mass, fat mass, and body fat percentage as measured by tetrapolar BIA (Table 4). Impedance dropped by 25 Ω in response to feeding, but increased 35 Ω after the second overnight fast so that impedance was higher on the second morning of the study than on the first. Total body water increased in response to feeding, but returned to baseline values after the second overnight fast in each condition. In accordance with the impedance changes, fat-free mass estimates increased in response to feeding, but fell below baseline values at the second morning visit in each condition. In addition, fat mass and body fat percentage decreased in response to feeding but were increased after the second overnight fast so that values were higher on the second morning of the study as compared to the first. Body fat percentage as measured by bipolar BIA increased in response to feeding and fell after the second overnight fast, but remained higher than the baseline visit. ηp2 values indicate that the variation in BIA parameters due to time accounts for 37% to 79% of the variation due to time plus error.

BIA results.

Average percent changes for dependent variables, representing changes at the afternoon visit relative to the baseline morning visit, are displayed in Figure 2. The numerical values for the percent change between the baseline visit and the afternoon visit, as well as the range of individual responses, which indicates the most extreme individual responders for each dependent variable, are displayed in the table of Supplemental Digital Content 4 (see Table, Supplemental Content 4, Percent Changes for Dependent Variables, Although statistical interactions were not present, several of the average percent changes appear dissimilar. For example, in males and females combined, trunk fat mass decreased by 1.2% on average in response to HC, but decreased by 3.1% in response to VLC; in males, trunk lean soft tissue increased by 3.1% in response to HC, but only 1.6% in response to VLC.

Body Composition Percent Changes. Average percent changes in body composition measurements relative to baseline visit. A, For DXA, an increase in most lean soft tissue regions at visit 2 (first afternoon; fed), but a return to baseline values at visit 3 (second morning; fasted) was observed. B, For DXA, a decrease in total fat mass at visit 2 due to the decline in estimates of trunk fat mass, but a return to baseline values at visit 3. C, For BIA, a decrease in impedance, fat mass, and body fat percentage assessed by tetrapolar BIA, as well as increases in total body water, fat free mass, and body fat percentage assessed by bipolar BIA at visit 2; differences relative to baseline were still present at visit 3, although the direction of these changes varied.a indicates significant difference from Morning 1 (P < 0.01). b indicates significant difference from Morning 1 (P < 0.001). c indicates significant difference from Morning 1 (P < 0.0001). d all BIA variables except TBW were significantly different at all three time points. Data compared by three-way repeated-measures ANOVA using condition and time as within subjects factors and sex as the between subjects factor. No condition or sex differences were detected.


This is the first study to examine the impact of acute diets of different macronutrient content on estimates of body composition obtained by multiple assessment technologies. The major result of our study was that acute food ingestion, regardless of macronutrient content, altered a number of body composition estimates obtained from DXA, tetrapolar BIA, and bipolar BIA. When compared directly, fat mass was affected to a greater extent in response to acute food ingestion when assessed by tetrapolar BIA (−1.8%) as compared to DXA (−0.7%), although changes in fat-free mass and lean soft tissue were equivalent (+1.7% for both). The concomitant relationship observed for BIA is likely due to the highly interrelated nature of BIA estimates, which are each largely influenced by the same impedance value. For DXA, the trunk region was particularly affected by food intake, as lean soft tissue estimates increased by 2.8% and fat mass decreased by 2.2%. Feeding impacted all parameters assessed by tetrapolar BIA. Impedance, fat mass, and body fat percentage decreased by approximately 4.4%, 1.8%, and 2.5% on average, and total body water increased by approximately 2% on average. Statistical interactions were also present that indicated that body fat percentage measured by DXA responded differently to the nutritional intervention in males and females and that body weight measurements were altered differently by the high-carbohydrate and low-carbohydrate diets. However, the magnitude of these differences was very small. Contrary to DXA and tetrapolar BIA, bipolar BIA produced higher estimates of body fat percentage in the fed state as compared to the fasted state. This seems to indicate that the measured impedance of the arms increased in response to acute food and fluid ingestion, whereas the impedance of legs decreased. One possible explanation is that the influence of gravity as participants stood upright for BIA assessments may have contributed to a greater quantity of fluid residing in the legs, thereby decreasing impedance for the tetrapolar measurement.

Some of the observed statistically significant results have questionable clinical significance, particularly total fat mass as assessed by DXA. It has previously been reported that the coefficient of variation for total fat mass assessed by DXA when scans are conducted back-to-back is approximately 1.6% (25). This is far greater than the observed 0.7% average change in response to food intake, which only equates to approximately 200 g of fat mass. It appears that average total fat mass assessed by DXA is relatively stable, even after food consumption of varying nutritional content. The marginal influence of food intake on total fat mass is unlikely to appreciably alter the evaluation of body composition or health status at the group level, although the wide range of observed changes could impact the assessment of an individual.

The variability in responses to feeding indicate that some individuals may be more resistant to alterations in body composition estimates due to acute feeding, while others exhibited noticeably large changes that could easily compromise the accurate assessment of their body composition over time. The present study supports that males and females are impacted similarly by food and fluid ingestion before body composition assessment. Contrary to our expectations, the macronutrient profile of the diets had little impact on the estimates of body composition obtained. Similar caloric and fluid intake led to similar responses, regardless of the drastically different macronutrient content of the experimental diets. However, there were some cases in which the average change appeared dissimilar between conditions, despite no statistical interaction (see Table, Supplemental Digital Content 4, Percent Changes for Dependent Variables,

The altered estimates of DXA lean soft tissue seen in the present study were similar in type and magnitude to those of Nana et al. (25). They reported an increase in average total and regional lean mass of up to 1.3% and 1.9% in response to an unstandardized diet and an increase in average total and regional lean soft tissue of up to 1.5% and 3.2% in response to a single standardized test meal, with the largest change occurring in the trunk region. The findings of Horber et al. (15) provide additional support that the trunk region is most responsible for the increase in lean soft tissue in response to feeding. Unlike Rouillier et al., (5,31) we found no group differences for any body composition estimates after an overnight fast when following an acute high-carbohydrate diet. The duration of the diet (i.e., 1 d vs 3 d) and the utilization of a standardized diet in the present study could have contributed to the different findings.

BIA measures of impedance decreased by 4.4% within an hour of food ingestion, which is consistent with previous reports (10,32). In accordance with the decrease in impedance, declines in calculated fat mass and body fat percentage were seen in response to feeding. However, unlike previous reports indicating that both impedance and body fat percentage returned to baseline fasting values after an additional overnight fast (32), impedance, fat-free mass, fat mass, and body fat percentage were all significantly different on two consecutive mornings in the present study. The magnitude of changes from one morning to the next were relatively small, and depending on the level of accuracy desired, may or may not be of clinical significance.

The practical importance of the present study may depend on the context in question. In research settings, when all known sources of error should be minimized, it seems clear that it would be ideal to perform body composition after a standardized set of procedures, such as an overnight food and fluid fast and voiding of the bladder. The present study indicates that an overnight fast is likely sufficient dietary control before body composition assessment, particularly for DXA, even after consuming short-term diets of drastically different nutritional content. In the clinical setting, if a one-time body composition assessment is being conducted to provide the patient, client, or health care professional with a general estimate of an individual's body composition, it may not be necessary to employ meticulous preparatory procedures. However, if serial body composition measurements will be taken over the course of time, it is likely very important to ensure that the patient or client follows a standard preassessment procedure. Errors due to food and fluid intake, hydration, and physical activity could either produce artificial changes in body composition estimates or mask real changes.

An additional consideration is the potential importance of the observed heterogeneous alterations of different body regions. The distribution of lean soft tissue has been reported to play a role in resting energy expenditure (17), adipose tissue volume (6), and mortality (18). It is also well known that body fat distribution plays a role in the risk for developing cardiovascular and metabolic disease (9,28). Based on the results of the present study, different body regions may be affected to different degrees by acute food ingestion. For example, the trunk region exhibited a concomitant increase in lean soft tissue and decrease in fat mass, but other body regions did not experience the same magnitude of change. If the food and fluid intake of a research participant or clinical patient is not standardized before assessment, it could compromise the accuracy of measurements related to regional body composition and disease risk.

The present study had several limitations: 1) Only a young population with no known diseases was studied; 2) the acute changes in diet may have elicited different effects than habitual diets; 3) differences in activities of individuals before each condition were not fully controlled; 4) because no objective measure of energy expenditure was obtained, the dietary prescriptions may have lacked optimal precision; 5) this study relied on self-report of food consumption rather than direct observation; 6) this study relied on self-report of menstrual bleeding in female participants; 7) muscle glycogen content was not measured; (8) urinary volumes and specific gravity of urine were not assessed.

In summary, body composition assessment by DXA and single-frequency BIA is affected by acute food and fluid ingestion on the day of assessment. The magnitude of these changes varies between individuals, but is sufficient to potentially obscure true changes in body composition or produce artificial changes. Total lean soft tissue measurements made by DXA are increased approximately 1.7% on average in response to feeding, with some participants demonstrating an increase of over 4.5%. Regional lean soft tissue measurements are increased up to 3% on average by feeding, and individual participants exhibited changes up to ±9%. DXA total and trunk fat mass estimates are decreased by up to 3% on average in response to feeding. Impedance measured by tetrapolar BIA decreased by 4.4% in response to feeding, and this subsequently affected measures of total body water, fat-free mass, and fat mass by approximately 1.5% to 2%. However, impedance measured by bipolar BIA increased, leading to higher estimates of body fat percentage.

Based on the results of the present study, it appears that when body composition is assessed by DXA, an overnight fast may be sufficient dietary control even when acute diets of drastically different macronutrient content are consumed the day before assessment. However, when body composition is assessed by BIA, the context of assessment may dictate whether an overnight fast is sufficient dietary control. Although statistically significant differences in BIA estimates were found when consecutive mornings were compared, the magnitude of these differences was small. Lean soft tissue and fat-free mass estimates of DXA and BIA are similarly affected by acute food intake, but DXA fat mass estimates are more robust than BIA estimates and exhibit minimal changes. The use and reporting of standardized procedures before body composition assessment will enhance the accuracy of measurements and allow researchers and clinicians to minimize errors that could be detrimental to accurate tracking of body composition changes.

The authors would like to acknowledge Dr. Suzy Weems for her assistance in evaluating the experimental diets, as well as Dr. Paul Gordon, Dr. Rodney Bowden, and Dr. Grant Morgan for their contributions to this study. Finally, the authors thank the research participants for volunteering to take part in this project. This research was supported by Baylor University.

The authors declare no conflicts of interest. The results of the present study do not constitute endorsement by ACSM. The results of the study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation.

Author contributions were as follows: G. M. T. and P. W. G. designed the research, P.W.G. performed the power analysis, G. M. T., F. E. M., and J. S. F. conducted the research, G. M. T. analyzed DXA scans, F. E. M. analyzed dietary records, G. M. T. performed statistical analysis, and G. M. T. wrote the article. All authors read and approved the final manuscript.


1. Acheson KJ, Schutz Y, Bessard T, Anantharaman K, Flatt JP, Jéquier E. Glycogen storage capacity and de novo lipogenesis during massive carbohydrate overfeeding in man. Am J Clin Nutr. 1988;48(2):240–7.
2. Rodriguez NR, Di Marco NM, Langley S, American Dietetic Association; Dietitians of Canada; American College of Sports Medicine. American College of Sports Medicine position stand. Nutrition and athletic performance. Med Sci Sports Exerc 2009;41(3):709–31.
3. Arnall DA, Nelson AG, Quigley J, Lex S, Dehart T, Fortune P. Supercompensated glycogen loads persist 5 days in resting trained cyclists. Eur J Appl Physiol. 2006;99(3):251–6.
4. Blake GM, Naeem M, Boutros M. Comparison of effective dose to children and adults from dual X-ray absorptiometry examinations. Bone. 2006;38(6):935–42.
5. Bone J, Burke LM. DXA estimates of body composition and carbohydrate loading. Ann Nutr Metab. 2016;68(3):228–9.
6. Bosy-Westphal A, Müller MJ. Identification of skeletal muscle mass depletion across age and BMI groups in health and disease—there is need for a unified definition. Int J Obes (Lond). 2015;39(3):379–86.
7. Brehm BJ, Seeley RJ, Daniels SR, D'Alessio DA. A randomized trial comparing a very low carbohydrate diet and a calorie-restricted low fat diet on body weight and cardiovascular risk factors in healthy women. J Clin Endocrinol Metab. 2003;88(4):1617–23.
8. Bussau VA, Fairchild TJ, Rao A, Steele P, Fournier PA. Carbohydrate loading in human muscle: an improved 1 day protocol. Eur J Appl Physiol. 2002;87(3):290–5.
9. Després JP. Body fat distribution and risk of cardiovascular disease an update. Circulation. 2012;126(10):1301–13.
10. Gallagher M, Walker KZ, O'Dea K. The influence of a breakfast meal on the assessment of body composition using bioelectrical impedance. Eur J Clin Nutr. 1998;52(2):94–7.
11. Goforth HW Jr, Arnall DA, Bennett BL, Law PG. Persistence of supercompensated muscle glycogen in trained subjects after carbohydrate loading. J Appl Physiol (1985). 1997;82(1):342–7.
12. Goforth HW Jr, Laurent D, Prusaczyk WK, Schneider KE, Petersen KF, Shulman GI. Effects of depletion exercise and light training on muscle glycogen supercompensation in men. Am J Physiol Endocrinol Metab. 2003;285(6):E1304–11.
13. Haderslev KV, Haderslev PH, Staun M. Accuracy of body composition measurements by dual energy x-ray absorptiometry in underweight patients with chronic intestinal disease and in lean subjects. Dyn Med. 2005;4(1):1.
14. Hangartner TN, Warner S, Braillon P, Jankowski L, Shepherd J. The Official Positions of the International Society for Clinical Densitometry: acquisition of dual-energy X-ray absorptiometry body composition and considerations regarding analysis and repeatability of measures. J Clin Densitom. 2013;16(4):520–36.
15. Horber FF, Thomi F, Casez JP, Fonteille J, Jaeger P. Impact of hydration status on body composition as measured by dual energy X-ray absorptiometry in normal volunteers and patients on haemodialysis. Br J Radiol. 1992;65(778):895–900.
16. Hultman E, Bergström J. Muscle glycogen synthesis in relation to diet studied in normal subjects. Acta Med Scand. 1967;182(1):109–17.
17. Hunter GR, Weinsier RL, Gower BA, Wetzstein C. Age-related decrease in resting energy expenditure in sedentary white women: effects of regional differences in lean and fat mass. Am J Clin Nutr. 2001;73(2):333–7.
18. Kang SH, Park JW, Yoon KW, Do JY. Limb/trunk lean mass ratio as a risk factor for mortality in peritoneal dialysis patients. J Ren Nutr. 2013;23(4):315–23.
19. Knapik JJ, Meredith CN, Jones BH, Suek L, Young VR, Evans WJ. Influence of fasting on carbohydrate and fat metabolism during rest and exercise in men. J Appl Physiol (1985. 1988;64(5):1923–9.
20. Kyle UG, Bosaeus I, De Lorenzo AD, et al. Bioelectrical impedance analysis—part I: review of principles and methods. Clin Nutr. 2004;23(5):1226–43.
21. Kyle UG, Bosaeus I, De Lorenzo AD, et al. Bioelectrical impedance analysis-part II: utilization in clinical practice. Clin Nutr 2004;23(6):1430–53.
22. Lakens D. Calculating and reporting effect sizes to facilitate cumulative science: a practical primer for t-tests and ANOVAs. Front Psychol. 2013;4:863.
23. Lee CD, Blair SN, Jackson AS. Cardiorespiratory fitness, body composition, and all-cause and cardiovascular disease mortality in men. Am J Clin Nutr. 1999;69(3):373–80.
24. Minderico CS, Silva AM, Keller K, et al. Usefulness of different techniques for measuring body composition changes during weight loss in overweight and obese women. Br J Nutr. 2008;99(2):432–41.
25. Nana A, Slater GJ, Hopkins WG, Burke LM. Effects of daily activities on dual-energy X-ray absorptiometry measurements of body composition in active people. Med Sci Sports Exerc. 2012;44(1):180–9.
26. NHANES Body Composition Procedures Manual [Internet] 2013. [cited 2015 Aug 14] Available from:
27. Olsson KE, Saltin B. Variation in total body water with muscle glycogen changes in man. Acta Physiol Scand. 1970;80(1):11–8.
28. Patel P, Abate N. Body fat distribution and insulin resistance. Nutrients. 2013;5(6):2019–27.
29. Roche AF, Heymsfield S, Lohman TG. Human body composition. Human Kinetics. 1996;384.
30. Rodriguez-Sanchez N, Galloway SD. Errors in dual energy x-ray absorptiometry estimation of body composition induced by hypohydration. Int J Sport Nutr Exerc Metab. 2015;25(1):60–8.
31. Rouillier MA, David-Riel S, Brazeau AS, St-Pierre DH, Karelis AD. Effect of an acute high carbohydrate diet on body composition using DXA in young men. Ann Nutr Metab. 2015;66(4):233–6.
32. Slinde F, Rossander-Hulthén L. Bioelectrical impedance: effect of 3 identical meals on diurnal impedance variation and calculation of body composition. Am J Clin Nutr. 2001;74(4):474–8.
33. Smithson M. Correct confidence intervals for various regression effect sizes and parameters: the importance of noncentral distributions in computing intervals. Educ Psychol Meas. 2001;61(4):605–32.
34. Steiger JH. Beyond the F test: effect size confidence intervals and tests of close fit in the analysis of variance and contrast analysis. Psychol Methods. 2004;9(2):164–82.
35. SuperTracker [Internet]United States Department of Agriculture; [date unknown]. [cited 2016 May 3] Available from:
36. Swain DP. ACSM’s Resource Manual for Guidelines for Exercise Testing and Prescription. Philadelphia: Wolters Kluwer Health/Lippincott Williams & Wilkins; 2014.
37. Thompson DL, Thompson WR, Prestridge TJ, et al. Effects of hydration and dehydration on body composition analysis: a comparative study of bioelectric impedance analysis and hydrodensitometry. J Sports Med Phys Fitness. 1991;31(4):565–70.
38. Thomsen TK, Jensen VJ, Henriksen MG. In vivo measurement of human body composition by dual-energy X-ray absorptiometry (DXA). Eur J Surg. 1998;164(2):133–7.
39. Utter AC, Nieman DC, Ward AN, Butterworth DE. Use of the leg-to-leg bioelectrical impedance method in assessing body-composition change in obese women. Am J Clin Nutr. 1999;69(4):603–7.
40. Vendelbo MH, Clasen BF, Treebak JT, et al. Insulin resistance after a 72-h fast is associated with impaired AS160 phosphorylation and accumulation of lipid and glycogen in human skeletal muscle. Am J Physiol Endocrinol Metab. 2012;302(2):E190–200.


Supplemental Digital Content

© 2017 American College of Sports Medicine