Lindsay P. Toth

Background/Objective: Instantaneous, accurate assessment of hydration and sweat loss has potential to optimize athlete performance and reduce risks associated with excessive exercise-related fluid loss. Reusable wearable hydration monitors, such as hDrop monitors, are an emerging non-invasive, quick, and easy to use alternative to existing methods. Purpose: To assess the accuracy of the hDrop wearable sweat monitors for estimating sweat loss and hydration status compared to established field-based measures, including body mass change and urine indices. Methods: Division I male collegiate soccer athletes contributed repeated training-session observations (38 total observations; 18 complete observations for sweat-loss analyses and 36 for hydration-score–USG analyses). They completed training sessions during which pre- and post-exercise body mass, fluid intake, urine output, and urine samples were collected. Sweat loss was calculated using changes in body mass adjusted for fluid balance. hDrop-derived estimates of sweat loss and hydration status were compared to criterion measures using mean bias, percent of criterion, mean absolute percent error (MAPE), repeated-measures Bland–Altman analysis, and generalized estimating equations accounting for observations clustered within athletes. Associations between hydration scores and urine-based markers were assessed using generalized estimating equations accounting for observations clustered within athletes. Results: The hDrop generated a sweat-loss estimate for 18 of 38 training-session observations (47.4%). Repeated-measures mean bias for sweat loss was −0.423 L (95% CI −1.106 to 0.259 L). The hDrop estimates averaged 106.2 ± 64.8% of criterion while MAPE was 50.3%, reflecting large individual error. In the repeated-measures analysis, criterion-measured sweat loss was positively associated with hDrop-estimated sweat loss (B = 0.204, 95% CI 0.018 to 0.391, p = 0.032). Significant proportional bias was present (slope = −0.790, 95% CI −1.438 to −0.143, p = 0.020), indicating greater underestimation at higher sweat-loss values. In repeated-measures GEE analyses, USG was not significantly associated with the hDrop hydration score either before practice (B = −407.9, 95% CI −1008.7 to 192.8, p = 0.183) or after practice (B = −135.0, 95% CI −803.8 to 533.7, p = 0.692). Conclusions: The hDrop generated sweat-loss estimates in fewer than half of training-session observations and demonstrated substantial individual-level error and significant proportional bias when estimates were available. Although the instantaneous hydration score decreased following exercise, its lack of association with USG leaves its physiological meaning and criterion validity uncertain. These findings do not support use of the hDrop as an interchangeable substitute for established methods of individualized sweat-loss or hydration assessment.

Was not provided for this article.

Background: Wearable sensor systems maximize visual and clinical feedback for physical therapists to enhance patient outcomes in rehabilitation medicine. However, physical therapists must adopt and accept new technologies for full integration into routine care to advance the use of technology in clinical care. Their role in technology design is critical in adopting and implementing technology. Interprofessional collaboration should be supported in the design of rehabilitation-assisted technologies. Objective: We used the established tenets of the Technology Acceptance Model to describe physical therapists’ expectations and experiences before and after using a novel wearable system in outpatient physical therapy. Methods: This multiple methods pilot study used a comparative pre-post survey and a qualitative semistructured focus group study design. Using purposive sampling, we recruited outpatient physical therapists to pilot the novel wearable technology, describe their expectations and experiences, and participate in a semistructured focus group discussion conducted to gather training and user experience information. Results: The study sample consisted of 5 physical therapists with an average age of 38.8 (SD 6.9) years and a work experience average of 12 (SD 7.7) years. Presurvey data show favorable expectations for usefulness and ease of use; however, favorability in both factors decreased after use. For perceived usefulness, all responses moved in the less favorable direction; mean difference −4.4 (SD 3.21); P=.04. All but 2 responses moved in the less favorable direction for overall perceived ease of use; mean difference −4.8 (SD 1.79); P=.04. Themed responses to open-ended questions in the postsurvey were feedback, setup time, accuracy, performance, and enhanced functional activities. Inductive content analysis of the focus group responses resulted in the following themes: system training, system benefits, system challenges, physical therapist perception of patients, and suggestions for improvement. The expectation for frequency of use decreased pre- to postexperience by 53% (mean −22, SD 14.40; P=.04). Conclusions: The Technology Acceptance Model–based survey responses and focus group themes outcomes demonstrated that physical therapists’ expectations for using new technology were not met. Engaging physical therapists in piloting novel wearable technology highlights the importance of physical therapist engagement in developing, refining, and implementing wearable devices for rehabilitation.

While previous research has linked physical activity (PA) with lower prescription medication consumption, limited evidence has investigated sedentary time (ST) as a major contributor to medication use, despite ST’s known association with chronic disease and mortality risk, even when PA volume is considered. This study aimed to examine the independent associations between objectively measured ST, patterns of sedentary bouts, and self-reported prescription medication use among adults ≥25 years of age. Thirty-two participants reported the number and type of medications they were currently prescribed and wore an accelerometer continuously on their hip for seven days to detect their ST. Poisson regression analysis was used to assess how average daily ST, sedentary bout frequency, and sedentary bout duration influenced medication use. The results revealed a significant association between greater ST and higher medication consumption. Specifically, each additional hour of ST per day, each sedentary bout, and each one-minute increase in bout duration were linked to a 66% (PR 1.66; 95% CI 1.25–2.19; p < 0.001), 36% (PR 1.36; 95% CI 1.12–1.64; p < 0.01), and 9% (PR 1.09; 95% CI 1.03–1.16; p < 0.01) higher prevalence of prescription medications, respectively. These findings suggest that higher ST is associated with a greater prevalence of using prescription medications in adults.

Background: ActiGraph activity counts are sensitive to sampling frequency (in hertz). However, it is unclear if ActiGraph’s ActiLife and Moving Average Vector Magnitude (MAVM) step-count algorithms are similarly affected. Methods: Forty-eight participants (22.1 ± 4.0 years) completed 4 × 3-min treadmill bouts at 2–5 mph while wearing three ActiGraph GT9X devices, initialized at 30, 80, and 100 Hz, on the hip in a randomized order. Criterion steps were hand-counted, and ActiGraph steps were obtained using the ActiLife and MAVM algorithms. Performance relative to the criterion was assessed using mean and SD of the error, mean absolute percent error, modified Bland–Altman analysis, and equivalence testing. Results: The ActiLife algorithm generally performed the best relative to the criterion at 30 Hz (range across speeds: mean error −0.8 to −21.7 steps/min; SD ±0.6 to ±20.2 steps/min; mean absolute percent error 0.7%–22.2%) versus 80 Hz (−1.2 to −22.0 steps/min; ±1.5 to ±19.1 steps/min; 1.1%–22.7%) versus 100 Hz (−2.4 to −23.6 steps/min; ±3.8 to ±17.6 steps/min; 2.1%–24.2%), respectively. For the modified Bland–Altman analysis, the 30-Hz data yielded the tightest 95% limits of agreement across speeds, followed by 80 Hz, and then 100 Hz. Similar trends were observed in equivalence testing. The MAVM algorithm did not appear to be affected by sampling frequency; however, it generally displayed less precision. Conclusion: The ActiLife algorithm is sensitive to sampling frequency, whereas the MAVM algorithm is not. When step count is the primary measurement outcome, a sampling frequency of 30 Hz is recommended.

Background: Prescription medication usage has been used as a predictor of disease prevalence and overall health status. Evidence suggests an inverse relationship exists between polypharmacy, which is the use of 5 or more medications, and physical activity participation. However, there is limited evidence examining the relationship between sedentary time and polypharmacy in adults. The aim of this study was to examine the associations between sedentary time and polypharmacy in a large nationally representative sample of US adults. Methods: Study sample (N = 2879) included nonpregnant adult (≥20 y old) participants from the 2017–2018 National Health and Nutrition Examination Survey. Self-reported minutes per day of sedentary time were converted to hours per day. The dependent variable was polypharmacy (≥5 medications). Results: Analysis revealed that for every hour of sedentary time, there was 4% greater odds of polypharmacy (odds ratio, 1.04; 95% confidence interval, 1.00–1.07, P = .04) after adjusting for age, race/ethnicity, education, waist circumference, and the interaction term between race/ethnicity and education. Conclusion: Our findings suggest increased sedentary time is associated with an increased risk of polypharmacy among a large nationally representative sample of US adults.

Purpose: The primary aim of this study was to compare steps per day across ActiGraph models, wear locations, and filtering methods. A secondary aim was to compare ActiGraph steps per day to those estimated by the ankle-worn StepWatch. Methods: We conducted a systematic literature review to identify studies of adults published before May 12, 2022, that compared free-living steps per day of ActiGraph step counting methods and studies that compared ActiGraph to StepWatch. Random-effects meta-analysis compared ActiGraph models, wear locations, filter mechanisms, and ActiGraph to StepWatch steps per day. A sensitivity analysis of wear location by younger and older age was included. Results: Twelve studies, with 46 comparisons, were identified. When worn on the hip, the AM-7164 recorded 123% of the GT series steps (no low-frequency extension (no LFE) or default filter). However, the AM-7164 recorded 72% of the GT series steps when the LFE was enabled. Independent of the filter used (i.e., LFE, no LFE), ActiGraph GT series monitors captured more steps on the wrist than on the hip, especially among older adults. Enabling the LFE on the GT series monitors consistently recorded more steps, regardless of wear location. When using the default filter (no LFE), ActiGraph recorded fewer steps than StepWatch (ActiGraph on hip 73% and ActiGraph on wrist 97% of StepWatch steps). When LFE was enabled, ActiGraph recorded more steps than StepWatch (ActiGraph on the hip, 132%; ActiGraph on the wrist, 178% of StepWatch steps). Conclusions: The choice of ActiGraph model, wear location, and filter all impacted steps per day in adults. These can markedly alter the steps recorded compared with a criterion method (StepWatch). This review provides critical insights for comparing studies using different ActiGraph step counting methods.

Previous studies suggest that the magnitude of morbidity/mortality reduction may differ between race-ethnic groups despite equated dose of physical activity (PA). The purpose of this study was to compare the potential racial-ethnic differences in cardiometabolic risk factors (CMRF) across quartiles of accelerometer-derived total activity counts/day (TAC/d) among US adults. The final sample (n=4144) included adults who participated in the 2003–2006 National Health and Nutrition Examination Survey (NHANES). CMRF included fasting glucose (FG), fasting insulin (FI), HOMA-IR, resting systolic (SBP) and diastolic blood pressure (DBP), waist circumference (WC), BMI, CRP, HDL-C, LDL-C, and triglycerides. Race-ethnic groups examined included non-Hispanic white (NHW), non-Hispanic black (NHB), and Mexican American (MA). In the highest quartile, NHW had significantly lower values of HOMA-IR, FI, SBP, BMI, WC, and HDL-C when compared to NHB. Compared to MA in the highest quartile, NHW had significantly lower values of HOMA-IR, FI, BMI, and triglycerides. Significant race-ethnic differences were found for several CMRF, especially among those who were in the top quartile of PA (e.g., the most active adults). It is probable that the protective effect of higher volumes of PA on CMRF is moderated by other non-PA factors distinct to NHB and MA.

Background: Ecological Momentary Assessment (EMA) allows input of affective, social, and behavioral factors by participants in real time, which reduces recall bias. While such data may help predict exercise behavior, one limitation is the need for additional devices (palm pilot) or the creation of an app for a smartphone. The use of an online survey distribution platform compatible with personal smartphones to conduct EMA of exercise behavior in free-living adults is still unclear. Methods: Across 14 days, texts with a survey link were distributed at 9:30 am, 1:30 pm, 5:30 pm, and 9:30 pm. The survey assessed hypothesized antecedents of exercise and allowed participants to log exercise behaviors. Feasibility outcomes were assessed using R. Results: From 6/13/16 to 10/26/16 recruitment yielded 29 (55% women, 24±6 years) participants. Out of 1624 prompts sent, 83.1% of the surveys were in compliance with research guidelines (i.e. completed within 60-min of receipt). On average it took 5.07±8.65-min to finish the surveys once they were started, with the average survey completed within 16.56±15.04-min from the time it was sent. Conclusions: Using Qualtrics for EMA has potential to minimize participant and administrative burden. Further feasibility testing in the general adult population is warranted.