https://orcid.org/0000-0002-9576-3072
ABSTRACT
Background
Physical activity (PA) is important in Cystic Fibrosis (CF) management. Fitness wearables are becoming increasingly popular as measurement tools of PA; however, the accuracy of these devices should first be evaluated.
Objective
The purpose of this study was to assess the accuracy of the ActivPAL and Fitbit Charge 2 as a measure of step count in Cystic Fibrosis.
Methods
Twenty-one participants were recruited from an adult CF Center in Ireland for a single session of testing. Participants walked for 5 min at five pre-determined speeds in a controlled testing environment (2, 2.5, 3, 3.5 and 4 miles per hour on a treadmill) and at three self-selected speeds in a corridor (slow, medium, and fast). They concurrently wore an accelerometer (ActivPAL) and fitness wearable (Fitbit Charge 2), and both were compared to visual observations. Step count is the outcome being assessed.
Results
The ActivPAL under-estimated step count by 0.63% across treadmill speeds and 1.1% across self-selected walking speeds. The Fitbit Charge 2 underestimated the step count by 2.97% across treadmill speeds and by 6.3% across self-selected walking speeds. Very strong correlations were found between the ActivPAL and visual observations (r: 0.93 to 0.99), while the Fitbit Charge 2 ranged from weak to very strong correlations when compared to visual observations (r: 0.34 to 0.84).
Conclusion
The ActivPAL and Fitbit Charge 2 demonstrated acceptable validity for step count measurement in CF. These devices can be used for tracking PA during interventions in people with CF.
Introduction
Cystic Fibrosis (CF) is a life limiting and complex multisystem disease (Quon and Aitken, Citation2012). CF is primarily characterized by dysfunction of the respiratory and digestive systems including chronic respiratory infections, progressive respiratory failure, pancreatic insufficiency, liver failure, male infertility, and CF-related diabetes (Stoltz, Meyerholz, Welsh, and Longo, Citation2015). Physical activity (PA) is regarded as an important component in the management of CF (Castellani et al., Citation2018). Higher levels of PA and exercise have been shown to improve: quality of life (Hebestreit et al., Citation2010); bone density (Gupta et al., Citation2019); and sputum expectoration (Kriemler et al., Citation2013). PA can reduce the rate of decline of lung function (Schneiderman et al., Citation2014) and can improve aerobic capacity (Hebestreit et al., Citation2010), both of which are linked to improved survival in CF (Nixon, Orenstein, Kelsey, and Doershuk, Citation1992).
Subjective measurements, such as questionnaires, are a relatively inexpensive and easy way to evaluate PA levels (Bradley, Kent, Elborn, and O’Neill, Citation2010). However, they are often prone to over-reporting PA and under-reporting sedentary behavior across the general population (Dyrstad, Hansen, Holme, and Anderssen, Citation2014). Therefore, objective measurements, such as pedometers, accelerometers, and fitness wearables may provide value in assessing and increasing PA.
One device used to assess PA is the ActivPAL (AP). The AP (PAL Technologies Ltd) is an accelerometer that records sitting, standing, and walking in a single unit device and can estimate energy expenditure. It is small and slim and is worn on the thigh. One study completed among healthy adults reported that the AP has appropriate validity for step count, with <1.1% step count error, regardless of walking speed (Ryan, Grant, Tigbe, and Granat, Citation2006). Similarly, another study among community dwelling older adults reported that the AP can accurately record step number and is valid, also with <1% error (Grant, Dall, Mitchell, and Granat, Citation2008).
In recent years, there has been an increase in the number of fitness wearables available, providing healthcare providers with objective measures of PA levels and exercise behaviors (Chiauzzi, Rodarte, and DasMahapatra, Citation2015). As a result, independent evaluation of the accuracy of the metrics from fitness wearables has increased (Wright, Hall Brown, Collier, and Sandberg, Citation2017). However, our understanding of the performance of these devices is still largely unknown. Fitbit is the most studied brand of fitness wearables (Fuller et al., Citation2020). The Fitbit Charge 2 (FC2) fitness device (Fitbit Inc) is worn on the wrist and aims to provide the wearer with real-time feedback on PA parameters such as step count and time spent on various levels of activity. Additional features of the device include estimating energy expenditure and time spent asleep, although the accuracy of these parameters has been found to be poor (Feehan et al., Citation2018). The Fitbit allows for PA levels to be tracked longitudinally, which enables the user to monitor their activity levels and could allow healthcare professionals to tailor PA advice and recommendations, and as a result encourage maintenance or an increase in PA behavior (Paul et al., Citation2015). However, prior to using the FC2 in a research setting or in PA promotion more information on accuracy is required.
Several studies have assessed the accuracy of the wrist worn FC2 in different populations. One study in older adults showed that the FC2 overestimated steps by 12.3% against previously validated technology over a 24-h protocol (Tedesco et al., Citation2019). While a study among college students found that the FC2 underestimated steps by 4.54% in a controlled walking setting (Keating et al., Citation2018).
While previous research has assessed these devices among other populations and under varying test conditions, we are unable to assume that AP and FC2 are accurate methods of recording step counts in people with CF (PWCF). A systematic review evaluating motion sensors in CF suggests that research should aim to identify the accuracy of such devices to evaluate their effectiveness, prior to investigating these devices in interventions to increase PA (Bradley, Kent, Elborn, and O’Neill, Citation2010). Technology informing fitness wearables is constantly evolving; however, establishing accuracy prior to using such devices in clinical trials is warranted but also crucial for clinical trials. This study intends to inform a subsequent research study to optimize PA in PWCF through the use of a fitness wearable (Curran et al., Citation2020). A recent systematic review aimed to synthesize the literature in this area and concluded that Fitbit devices had a tendency to underestimate steps in a controlled testing environment and overestimate steps in free-living conditions (Feehan et al., Citation2018). However, studies in this review included a variety of Fitbit devices that were evaluated under diverse testing conditions in numerous populations. Most studies included healthy individuals, and some evaluated their accuracy in individuals suffering from chronic conditions such as multiple sclerosis, post stroke, and cardiac conditions, thereby limiting the application and transferability of such research to people with PWCF.
Both the AP and FC2 devices have been widely used in previous research and measure important determinants of PA (Edwardson et al., Citation2017; Feehan et al., Citation2018). They have become increasingly popular in recent years with technological advancements as measurement tools in PA and also for health promotion (Bunn, Navalta, Fountaine, and Reece, Citation2018; Mercer et al., Citation2016). Determining the accuracy of these devices would assist health professionals in determining if such devices are appropriate for use in PWCF. This study aimed to determine the criterion validity (Mokkink et al., Citation2012) (accuracy) of the AP and FC2 to record step count compared to visual observations.
Methods
This study involved a single session of testing, as per previous validation studies investigating activity monitors (Grant, Dall, Mitchell, and Granat, Citation2008; Ryan, Grant, Tigbe, and Granat, Citation2006; Takacs et al., Citation2014). This study was conducted in accordance with Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist. This study was divided into two parts. Part one involved a controlled laboratory test that required participants to walk at five pre-determined speeds on a treadmill. The second part investigated three self-selected speeds (i.e. slow, medium, and fast) in a corridor. Observer step count was considered the criterion measure for steps, similar to previous validation studies (Dwyer et al., Citation2009; Paul et al., Citation2015; Sushames et al., Citation2016; Swartz et al., Citation2003). A step was defined as the initial point of contact of one foot to the initial point of contact of the opposite foot (Levine, Richards, and Whittle, Citation2012). This was recorded by the observer using a clicker to determine the total number of steps. Walking trials were video recorded, for both parts using one video camera (JVC Everio Model GZ-MS210BEK) that was placed directly behind the participant. A randomly selected subset (50%) of these were reviewed to verify the step count. This video footage was reviewed by a second independent assessor not involved in the study analysis in order to verify step count. If the difference in the direct observations were greater than 10 steps, the footage was viewed a second time by both reviewers and this revised estimate was used (Sushames et al., Citation2016). The rationale behind the video was as an added layer of confirmation to further enhance the robustness of this study.
Participants
Twenty-one participants were recruited from the Adult CF Center at University Hospital Limerick, Ireland. In order to be eligible for inclusion, participants were required to have a confirmed diagnosis of CF (based on CF-causing mutations and/or a sweat chloride concentration during two tests of >60 mmol/l) and be able to walk unaided for at least 40 min. Exclusion criteria included any neurological disorder, cognitive disorder, or musculoskeletal injury that would impair walking. Informed consent was obtained from all participants, and ethical approval was obtained from the University Hospital Limerick Research Ethics board. Sample size was calculated based on the level of agreement between the two measures. An ICC > 0.8 would detect at least a moderate relationship between the step count data of the devices (Evenson, Goto, and Furberg, Citation2015). To achieve 95% power, at a 0.05 level test to reject the null hypothesis that the ICC is less than or equal to 0.8, a minimum of 14 participants would be required.
Measurement of demographic data
Sex, age, height, weight, and lung function were recorded at baseline. Spirometry was conducted as per the American Thoracic Society (ATS) standard techniques (Miller, Citation2005). Values are expressed as a percentage of the predicted values for height, sex, and age for adults (Hankinson, Odencrantz, and Fedan, Citation1999).
Devices
The AP model is triaxial (ActivPAL3) () and the sampling frequency was 20 Hz with a dynamic range of ± 2 gravitational (g) units. The device weighs 20 g (5 cm x 3.5 cm x 0.7 cm) and measures sitting, standing, and walking and can provide step count data based on lower limb movement. We derived the step count to be compared to our criterion measure (observation). Prior to each walking speed, the AP monitors were connected to the software and programmed to begin recording straight away. This was completed on a flat surface next to the participant. There was one AP device for each walking speed. The device was placed one-third of the distance between the patella and inguinal crease at the midline of the anterior surface of the participant’s right thigh by the same researcher and secured with adhesive dressing (tegaderm), similar to previous research (Aguilar-Farías, Brown, and Peeters, Citation2014; Montoye et al., Citation2017). Data is downloaded and analyzed using proprietary algorithms.
Figure 1. ActivPAL and Fitbit Charge 2.
The FC2 () weighs 23 g (14 cm x 17.6 cm x 1.4 cm) and measures step count, distance, active minutes, and sleep. The FC2 collects data in 60 second epochs. FC2 is a microelectromechanical triaxial accelerometer that converts raw acceleration to step counts using proprietary algorithms (Dinesh and Freedson, Citation2012). We derived the step count to be compared to our criterion measure (observation). The FC2 was applied by the same researcher to the non-dominant wrist of the participant using a fitted band. Step count was taken from a cumulative figure; therefore, the step count was recorded at baseline and after each walking trial. The display on the Fitbit illustrates step number, time, and date with options to view heart rate, distance, active minutes, and calories.
Protocol
Treadmill walking
Participants were instructed and familiarized with the use of the treadmill prior to starting the test as part of their routine physiotherapy review appointments. They were asked not to use the handrails on the treadmill. The AP and FC2 were applied to the participants as described previously. Prior to the treadmill trials, the participant sat on a chair that was placed on the treadmill. The AP was kept in a horizontal position and placed directly onto the right thigh and secured with tegaderm. Participants then walked at five different speeds (2, 2.5, 3, 3.5 and 4 miles per hour (mph)) for 5 min each as per previous studies (Le Masurier, Lee, and Tudor-Locke, Citation2004; Le Masurier and Tudor-Locke, Citation2003; Ryan, Grant, Tigbe, and Granat, Citation2006; Takacs et al., Citation2014). The rationale for walking for 5 min was replicating methods of other researcher groups. Furthermore, we were cognizant of participant fatigue, particularly with varying levels of lung function and so we felt 5 min would be sufficient at each speed. At the end of the trial, the treadmill was stopped, and a chair was placed on the treadmill, directly behind the participant. Once the participant sat down, the AP was removed, again kept in a horizontal position to avoid additional steps being incorrectly counted and was placed onto the table and immediately downloaded using the AP software. This was repeated for each speed. The acceleration and deceleration phases of walking were included as it was not possible to distinguish these from each of the trials. Each speed was picked at random. Speed order was established by the participants selecting folded pieces of paper randomly from an envelope. They were provided with a rest period between each speed. Participants rested on a chair that was placed directly behind them on the treadmill itself at the end of the trial once the treadmill had come to a stop. Step count was recorded with a clicker by an independent assessor (physiotherapist) not otherwise involved in the study. Step count was recorded from the cumulative figure on the FC2 and deducted from the previous figure to attain a step count for each walking trial. At the start of the next trial the FC2 steps were documented in case there were additional steps/arm swings during the rest period.
Self-selected walking
Participants were then asked to walk in the outpatient corridor of the Adult CF Center, at a self-selected walking speed for slow, medium, and fast pace, for 5 min each. Slow pace was described as easy relaxed walking pace, medium was described as able to maintain a conversation while walking, and fast was described as unable to maintain a conversation while walking. A similar setup was applied for the self-selected walking trials whereby a chair was placed at the starting position. The AP was programmed and applied directly onto the participant’s right thigh by the researcher and secured with tegaderm. Participants were asked to walk for 5 min around two cones set 10 m apart, while step count was again recorded by an independent observer using a clicker. At the end of the five-minute trial, the participants were asked to stop where they were. A chair was placed directly behind them and they were asked to sit down while the AP was removed. The AP was connected to the software and data was immediately downloaded. Step count data from the observer and the FC2 were recorded. At the start of the next trial, the FC2 steps were documented (in case there were additional steps/arm swings during the rest period).
Statistical analyses
Data were visually inspected by histogram plots, and the data were determined to be normally distributed and therefore descriptive data are presented as mean ± standard deviation (SD). The software package SPSS (version 24) was used for all analysis. Criterion validity was assessed in this study compared the step count results from both devices to the observer count. In order to assess device validity, the following tests were conducted: Pearson correlation coefficients; Bland–Altman plots; and absolute mean percentage difference.
Pearson correlation coefficients were calculated to determine the correlation between both devices to observer count. Correlation coefficients can be used to rank the order of the devices at each speed according to their correlation with the observer (Salkind, Citation2006). Correlations for the absolute value of r were: 0.00–0.19 “very weak”; 0.20–0.39 “weak”; 0.40–0.59 “moderate”; 0.60–0.79 “strong”; 0.80–1.00 “very strong” (Evans, Citation1996). In constructing the Bland–Altman plot, the y-axis represents the difference between the observer and the AP/FC2 estimates, and the x-axis represents the mean of the two estimates. The narrower the limits of agreement, the more accurate the measurement device.
Results
Participant age, baseline anthropometry, and lung function were collected at the start of testing. Twenty-one participants (15 F, 6 M) completed this study with a mean age of 25.3 years (±5.98), mean BMI of 21.7 kg/m2 (±3.3), mean FEV1 of 64.8% (±29) and mean FVC of 83% (±24.8). CFTR genotypes were: homozygote delF508 (n = 12); delF508/G55ID (n = 3); delF508/1717 GA (n = 2); delF508/3007delg (n = 2); and DelF508/C2052delA (n = 2). Video recordings of randomly selected walking speeds at both treadmill and self-selected speeds were reviewed, and the difference between measures was not greater than 10 steps for any of the trials. This confirms the validity of the observer count.
Treadmill walking
Step count data for overall results for the AP and FC2 are presented in . presents a summary of AP, FC2, and observer count for steps across all speeds. Bland–Altman plots were used to assess the level of agreement between the AP and the observer, and similarly between the FC2 and the observer.
Table 1. Overall % difference for ActivPAL and Fitbit Charge 2 across treadmill walking speeds and self-selected walking speeds.
Table 2. Summary of ActivPAL, Fitbit Charge 2 and Observer Count for steps across all speeds.
AP against observer
Overall, the AP underestimated the number of steps by 0.63% across all treadmill speeds (). When investigating individual speeds, the AP had a 2.6 step count difference at 2mph (0.5% under-estimation) when compared to observer step count. The largest difference was 6.6 steps at 4mph (1% under-estimation) (). The AP correlated very strongly with observer step count (r = 0.95 to 0.99, p = .0005) (). The Bland–Altman plots demonstrated narrow limits of agreement ().
Table 3. Mean Absolute Difference and % difference in steps between ActivPAL and Fitbit Charge 2 when compared to observer step count – treadmill walking and self-selected speeds.
Table 4. Pearson correlation co-efficients for the AP with the FC2 when compared to observer step count for both treadmill walking and self-selected speeds (r and p values).
Figure 2. Bland Altman plot comparing average steps for the Observer step count and ActivPAL at 2mph. Solid line indicates the mean difference between the two measures, dashed lines indicate the limits of agreement (1.96 SDs of the mean difference).
FC2 against observer
The FC2 under-estimated step count by 2.97% across all treadmill speeds (). The FC2 was most accurate at 2 mph where a difference of 13.7 steps was calculated, equating to an underestimation of steps by 2.8%. The most significant difference in step count during treadmill testing was at 3.5 mph, where a mean absolute difference of 31.2 steps was noted, highlighting that the FC2 underestimated steps by 5.3%, when compared to the observer count (). The FC2 demonstrated weak to very strong correlations with the observer step count (r ranging from 0.40 to 0.84, p = .024 to 0.0005) () and wider limits of agreement when compared to the AP ().
Figure 3. Bland Altman plot comparing average steps for the Observer step count and Fitbit at 2mph. Solid line indicates the mean difference between the two measures, dashed lines indicate the limits of agreement (1.96 SDs of the mean difference).
Self-selected walking speed
Walking speeds for self-selected pace were 1.03 m/s for slow speed, 1.18 m/s for medium speed, and 1.41 m/s for fast walking.
AP against observer
Across all self-selected walking speeds, the AP under-estimated step count by 1.1%. Small differences in step count during each specific self-selected walking speed is presented in . The largest variance was at slow-walking speed with a mean absolute difference of 8.4 steps (1.5% under-estimation) when compared to the observer count, while at a medium pace this was just 4 steps (0.7% under-estimation). In addition, the AP was very strongly correlated with the observer count (r = 0.93 to 0.96, p = .0005) () with narrow limits of agreement as illustrated by the Bland–Altman plots.
FC2 against observer
The FC2 under-estimated the step count by 6.3% across all self-selected walking speeds. The absolute mean difference of 27.8 steps (4.9% under-estimation) was noted during this self-selected “medium” speed (). In contrast, the greatest disparity in step count was noted in “fast” walking. There was a difference of 53.7 steps between both absolute means (FC2 and observer) equating to a step count under-estimation of 9.7%. Correlations varied from weak to strong (r = 0.34 to 0.72) and was not significant only at the “medium” speed () with wider limits of agreement as illustrated by Bland–Altman plots.
Discussion
PA is highly recommended in CF as a key component in the management of this disease (Castellani et al., Citation2018). Recent technological advances have resulted in the popular use of fitness wearables in the evaluation of PA, particularly walking. As such, robust evaluation of the properties of such devices is required. This study compared the accuracy of the AP and the FC2 against observer step count in a CF population in a controlled treadmill condition and in self-selected free walking. The results indicate that the AP under-estimates step count by 0.63% during controlled walking speeds (2 mph, 2.5 mph, 3 mph, 3.5 mph and 4 mph) and by 1.1% in self-selected (i.e. slow, medium, and fast) walking speeds. The AP is accurate across all walking speeds. The FC2 is more accurate with treadmill walking speeds with a 2.97% underestimation of steps while at self-selected walking speeds it under-estimates step count by 6.3% overall.
ActivPAL
The AP had a low absolute mean percentage step difference (<1.1%) and correlated very strongly with the observer, which suggests that the AP is a valid measure to assess step count in PWCF. This is similar to a study evaluating the validity of the AP in a healthy population (Ryan, Grant, Tigbe, and Granat, Citation2006). Ryan, Grant, Tigbe, and Granat (Citation2006) also found an absolute percentage error of <1.11% for step count. A similar methodological approach was used across our study and in the study by Ryan, Grant, Tigbe, and Granat (Citation2006) thereby endorsing the applicability and extractability of these results. In another respiratory population, the validity of the AP was not accurate at slow speeds in patients with chronic obstructive pulmonary disease (mean 0.56 m/s) (Cindy Ng, Jenkins, and Hill, Citation2012). However, the speed tested was substantially slower than the slow walking speed observed in our study (1.03 m/s). In a chronic low back pain population the AP also demonstrated validity for measuring step count with limits of agreement less than 1% (Ryan et al., Citation2008), similar to the findings of our study.
Fitbit charge 2
Overall, the FC2 demonstrated moderate validity when compared to the observer count. The FC2 had a 2.97% under-estimation of steps in treadmill walking and a 6.3% under-estimation in self-selected walking speeds with a weak to strong correlation with the observer. Globally, there is no consensus on what is an acceptable degree of error for PA wearables. Some research suggests that mean errors of 20% or less have acceptable validity for clinical purposes (Schneider, Crouter, Lukajic, and Bassett, Citation2003). While other research suggests that step count monitors that are within 3% of the manual count are considered to be “extremely accurate,” while those within 10% are considered “acceptable” (Schneider, Crouter, and Bassett, Citation2004; Tudor-Locke et al., Citation2006). Taking the latter and more conservative figures, into consideration, the FC2 could be considered accurate for measuring step count during controlled walking and acceptable for PA promotion purposes. Furthermore, a recently published systematic review evaluating the effectiveness of several Fitbit devices reported that such devices may acceptably underestimate steps within ± 3% for controlled settings and ± 10% in self-selected conditions (Feehan et al., Citation2018), which is similar to our study’s findings. Feehan et al. (Citation2018) found that step count accuracy varied with speed of ambulation and placement of the device. This is consistent with other reviews on PA wearables (Evenson, Goto, and Furberg, Citation2015; Fuller et al., Citation2020). Furthermore, it should be noted that the accuracy in free living varied according to the reference grade criterion measure. For example, the Fitbit step count was within ± 10% in 3/5 wrist worn Fitbit studies when compared to the AP or Actigraph. However, in one study a 35% step count discrepancy was found between a Fitbit device worn on the torso compared to an Omron pedometer worn on the ankle (Feehan et al., Citation2018).
Fitbits are widely available and have become increasingly popular in recent years as a means to monitor and promote PA. However, in research settings, accuracy is variable due to factors including differences between devices used, placement of such devices and variability in the study protocols. The results from our study are similar to previous research conducted in a healthy population investigating the Fitbit Flex (Sushames et al., Citation2016) where it underestimated step count, although not to the same extent (2.97% in our study vs. 15% in a controlled walking task). In the same study, a 21.2% step count difference was noted in free living over 7 days (Sushames et al., Citation2016). If one is to consider this 6.2% step count difference found by Sushames et al. (Citation2016) between controlled and free living environments, by the same logic, this may infer that a step-count underestimation of the FC2 up to 9.17% could be expected in a free living environment in PWCF. However, it must be noted that Dierker and Smith (Citation2014) found no significant difference in step count using a seven-day trial. This difference may be attributed to variability in wear time. Sushames et al. (Citation2016) asked participants to remove the device during water-based activities, while Dierker and Smith (Citation2014) requested participants to remove the devices during exercise. Similarly, previous research in the Fitbit One found no significant difference between the device and the observer count and also had <1.3% error (Takacs et al., Citation2014). Fuller et al. (Citation2020) summarized that the Fitbit was the most extensively researched PA wearable and it can accurately measure steps in a controlled setting.
Implications for practice
While both the AP and FC2 demonstrated appropriate levels of accuracy, it should be noted that PA research must additionally consider the practicality of such devices for use in research. Some of the key considerations include the ability to blind participants from their PA data and also acceptability of the participants to wear a PA device. While the AP was more precise in terms of both treadmill and self-selected walking, it does not provide the wearer with direct and continuous feedback on ongoing step count like the FC2 provides. However, this can be advantageous in research, which aims to blind the participants from their data. Using the AP for initial assessment of PA behavior may be valuable as the participant is blinded to their results. Conversely, the use of the FC2 may hold further potential to impact the PA levels of the wearer when used as part of an intervention. A continuous wear protocol can be easily achieved with both devices, which may enhance compliance. Finally, previous research suggests that fashion/appearance is an important factor in the acceptability of such devices, as a thigh worn device was poorly accepted by the CF population, particularly in females (Dias et al., Citation2012). This is further supported by another study in which participants deemed wrist worn devices, which are unobtrusive, to be acceptable for PA monitoring (Shelley et al., Citation2018). Both devices demonstrate acceptable levels of accuracy and either would be appropriate for use in clinical practice in PWCF.
Strengths of this study include that it had a powered sample size and assessed step count in both controlled and self-selected walking speeds, as per previous validation studies. Treadmill walking and overground self-selected walking speeds were chosen as kinematics can vary between the two, with increases in hip range of motion on a treadmill combined with a significant decrease in stance time (Alton, Baldey, Caplan, and Morrissey, Citation1998; Warabi et al., Citation2005). This may explain the difference observed in percentage error between both walking conditions. We feel that testing participants in both walking conditions enhances the robustness of this study, as many of our participants would use a treadmill for exercise. The participants in this study are largely representative of the broader Irish adult CF population (mean age: 31.6 years, mean BMI: 23.3 kg/m2, mean FEV1 of 70%) (Cystic Fibrosis Registry of Ireland, Citation2018) thus enhancing the applicability of such results.
Limitations
For translation or implementation purposes, limitations of the study need to be considered. This study did not avail of the full measurement facilities of these devices, and future research should investigate the accuracy of other parameters, such as energy expenditure, sleep, and activity minutes. Furthermore, this study did not consider free living tasks or simulated activities of daily living.
Conclusion
This study has established the criterion validity of the AP and FC2 for measuring PA in PWCF and would be suitable for the assessment of step count. Future research contemplating these devices for use in PA interventions should consider the features, acceptability, and practicality of both devices to enhance compliance with the CF population.
Acknowledgments
The authors would like to acknowledge the Health Research Institute, University of Limerick for funding this research and Truck Run 4 Katie, a charitable organization for funding Fitbits.
Disclosure statement
The authors declare that they have no competing interests.
Additional information
Funding
References
- Aguilar-Farías N, Brown WJ, Peeters GG 2014 ActiGraph GT3X+ cut-points for identifying sedentary behaviour in older adults in free-living environments. Journal of Science and Medicine in Sport 17(3):293–299. 10.1016/j.jsams.2013.07.002.
- Alton F, Baldey L, Caplan S, Morrissey M 1998 A kinematic comparison of overground and treadmill walking. Clinical Biomechanics 13(6):434–440. 10.1016/S0268-0033(98)00012-6.
- Bradley JM, Kent L, Elborn JS, O’Neill B 2010 Motion sensors for monitoring physical activity in cystic fibrosis: What is the next step? Physical Therapy Reviews 15(3):197–203. 10.1179/1743288X10Y.0000000005.
- Bunn J, Navalta JW, Fountaine CJ, Reece JD 2018 Current state of commercial wearable technology in physical activity monitoring 2015–2017. International Journal of Exercise Science 11:503.
- Castellani C, Duff AJ, Bell SC, Heijerman HG, Munck A, Ratjen F, Sermet-Gaudelus I, Southern KW, Barben J, Flume PA, et al. 2018 ECFS best practice guidelines: The 2018 revision. Journal of Cystic Fibrosis 17(2):153–178. 10.1016/j.jcf.2018.02.006.
- Chiauzzi E, Rodarte C, DasMahapatra P 2015 Patient-centered activity monitoring in the self-management of chronic health conditions. BMC Medicine 13(1):77. 10.1186/s12916-015-0319-2.
- Cindy Ng LW, Jenkins S, Hill K 2012 Accuracy and responsiveness of the stepwatch activity monitor and ActivPAL in patients with COPD when walking with and without a rollator. Disability and Rehabilitation 34(15):1317–1322. 10.3109/09638288.2011.641666.
- Curran M, Tierney AC, Collins L, Kennedy L, McDonnell C, Jurascheck AJ, Sheikhi A, Walsh C, Button B, Galvin R, et al. 2020 Steps ahead: Optimising physical activity in adults with cystic fibrosis: Study protocol for a pilot randomised trial using wearable technology. HRB Open Research 3:21.
- Cystic Fibrosis Registry of Ireland 2018 Cystic fibrosis registry of Ireland: Annual report. https://cfri.ie/annual-reports/.
- Dias A, Gorzelniak L, Jörres RA, Fischer R, Hartvigsen G, Horsch A 2012 Assessing physical activity in the daily life of cystic fibrosis patients. Pervasive and Mobile Computing 8(6):837–844. 10.1016/j.pmcj.2012.08.001.
- Dierker KE, Smith BK 2014 Comparison between four personal activity monitors and the actigraph GT3X+ to measure daily steps. Medicine and Science in Sports and Exercise 46:S792. 10.1249/01.mss.0000495877.54448.6c.
- Dinesh, J. and Freedson, P., 2012. ActiGraph and Actical physical activity monitors: a peek under the hood. Medicine and science in sports and exercise, 44(1 Suppl 1), p.S86
- Dwyer TJ, Alison JA, McKeough ZJ, Elkins MR, Bye PTP 2009 Evaluation of the SenseWear activity monitor during exercise in cystic fibrosis and in health. Respiratory Medicine 103(10):1511–1517. 10.1016/j.rmed.2009.04.013.
- Dyrstad SM, Hansen BH, Holme IM, Anderssen SA 2014 Comparison of self-reported versus accelerometer-measured physical activity. Medicine and Science in Sports and Exercise 46(1):99–106. 10.1249/MSS.0b013e3182a0595f.
- Edwardson CL, Winkler EA, Bodicoat DH, Yates T, Davies MJ, Dunstan DW, Healy GN 2017 Considerations when using the activPAL monitor in field-based research with adult populations. Journal of Sport and Health Science 6(2):162–178. 10.1016/j.jshs.2016.02.002.
- Evans JD 1996 Straightforward Statistics for the Behavioral Sciences. United States: Thomson Brooks/Cole Publishing Co.
- Evenson KR, Goto MM, Furberg RD 2015 Systematic review of the validity and reliability of consumer-wearable activity trackers. International Journal of Behavioral Nutrition and Physical Activity 12(1):159. 10.1186/s12966-015-0314-1.
- Feehan LM, Geldman J, Sayre EC, Park C, Ezzat AM, Yoo JY, Hamilton CB, Li LC 2018 Accuracy of Fitbit devices: Systematic review and narrative syntheses of quantitative data. JMIR mHealth and uHealth 6(8):e10527. 10.2196/10527.
- Fuller D, Colwell E, Low J, Orychock K, Tobin MA, Simango B, Buote R, Van Heerden D, Luan H, Cullen K 2020 Reliability and validity of commercially available wearable devices for measuring steps, energy expenditure, and heart rate: Systematic review. JMIR mHealth and uHealth 8(9):e18694. 10.2196/18694.
- Grant PM, Dall PM, Mitchell SL, Granat MH 2008 Activity-monitor accuracy in measuring step number and cadence in community-dwelling older adults. Journal of Aging and Physical Activity 16(2):201–214. 10.1123/japa.16.2.201.
- Gupta S, Mukherjee A, Lodha R, Kabra M, Deepak KK, Khadgawat R, Talwar A, Kabra SK 2019 Effects of exercise intervention program on bone mineral accretion in children and adolescents with cystic fibrosis: A randomized controlled trial. Indian Journal of Pediatrics 86(11):987–994. 10.1007/s12098-019-03019-x.
- Hankinson JL, Odencrantz JR, Fedan KB 1999 Spirometric reference values from a sample of the general U.S. population. American Journal of Respiratory and Critical Care Medicine 159(1):179–187. 10.1164/ajrccm.159.1.9712108.
- Hebestreit H, Kieser S, Junge S, Ballmann M, Hebestreit A, Schindler C, Schenk T, Posselt H, Kriemler S 2010 Long-term effects of a partially supervised conditioning programme in cystic fibrosis. European Respiratory Journal 35(3):578–583. 10.1183/09031936.00062409.
- John D, Freedson P 2012 ActiGraph and actical physical activity monitors: A peek under the hood. Medicine and Science in Sports and Exercise 44(1S):S86. 10.1249/MSS.0b013e3182399f5e.
- Keating XD, Liu J, Liu X, Shangguan R, Guan J, Chen L 2018 Validity of fitbit charge 2 in controlled college physical education settings. ICHPER-SD Journal of Research 9:28–35.
- Kriemler S, Kieser S, Junge S, Ballmann M, Hebestreit A, Schindler C, Stüssi C, Hebestreit H 2013 Effect of supervised training on FEV1 in cystic fibrosis: A randomised controlled trial. Journal of Cystic Fibrosis 12(6):714–720. 10.1016/j.jcf.2013.03.003.
- Le Masurier GC, Lee SM, Tudor-Locke C 2004 Motion sensor accuracy under controlled and free-living conditions. Medicine and Science in Sports and Exercise 36:905–910. 10.1249/01.MSS.0000126777.50188.73.
- Le Masurier GC, Tudor-Locke C 2003 Comparison of pedometer and accelerometer accuracy under controlled conditions. Medicine and Science in Sports and Exercise 35(5):867–871. 10.1249/01.MSS.0000064996.63632.10.
- Levine D, Richards J, Whittle MW 2012 Whittle’s Gait Analysis-E-Book. UK: Elsevier Health Sciences.
- Mercer K, Li M, Giangregorio L, Burns C, Grindrod K 2016 Behavior change techniques present in wearable activity trackers: A critical analysis. JMIR mHealth and uHealth 4(2):e40. 10.2196/mhealth.4461.
- Miller MR 2005 Standardisation of spirometry. European Respiratory Journal 26(2):319–338. 10.1183/09031936.05.00034805.
- Mokkink LB, Terwee CB, Patrick DL, Alonso J, Stratford PW, Knol DL, Bouter LM, De Vet HC 2012 COSMIN checklist manual. Amsterdam: University Medical Center. https://fac.ksu.edu.sa/sites/default/files/cosmin_checklist_manual_v9.pdf.
- Montoye AHK, Pivarnik JM, Mudd LM, Biswas S, Pfeiffer KA 2017 Evaluation of the activPAL accelerometer for physical activity and energy expenditure estimation in a semi-structured setting. Journal of Science and Medicine in Sport 20(11):1003–1007. 10.1016/j.jsams.2017.04.011.
- Nixon PA, Orenstein DM, Kelsey SF, Doershuk CF 1992 The prognostic value of exercise testing in patients with cystic fibrosis. New England Journal of Medicine 327(25):1785–1788. 10.1056/NEJM199212173272504.
- Paul SS, Tiedemann A, Hassett LM, Ramsay E, Kirkham C, Chagpar S, Sherrington C 2015 Validity of the Fitbit activity tracker for measuring steps in community-dwelling older adults. BMJ Open Sport and Exercise Medicine 1(1):e000013. 10.1136/bmjsem-2015-000013.
- Quon BS, Aitken ML 2012 Cystic fibrosis: What to expect now in the early adult years. Paediatric Respiratory Reviews 13(4):206–214. 10.1016/j.prrv.2012.03.005.
- Ryan CG, Grant P, Gray H, Newton M, Granat M 2008 Measuring postural physical activity in people with chronic low back pain. Journal of Back and Musculoskeletal Rehabilitation 21(1):43–50. 10.3233/BMR-2008-21106.
- Ryan CG, Grant PM, Tigbe WW, Granat MH 2006 The validity and reliability of a novel activity monitor as a measure of walking. British Journal of Sports Medicine 40(9):779–784. 10.1136/bjsm.2006.027276.
- Salkind NJ 2006 Encyclopedia of Measurement and Statistics. California: SAGE Publications.
- Schneider PL, Crouter SE, Bassett DR 2004 Pedometer measures of free-living physical activity: Comparison of 13 models. Medicine and Science in Sports and Exercise 36(2):331–335. 10.1249/01.MSS.0000113486.60548.E9.
- Schneider PL, Crouter SE, Lukajic O, Bassett JDR 2003 Accuracy and reliability of 10 pedometers for measuring steps over a 400-m walk. Medicine and Science in Sports and Exercise 35(10):1779–1784. 10.1249/01.MSS.0000089342.96098.C4.
- Schneiderman JE, Wilkes DL, Atenafu EG, Nguyen T, Wells GD, Alarie N, Tullis E, Lands LC, Coates AL, Corey M, et al. 2014 Longitudinal relationship between physical activity and lung health in patients with cystic fibrosis. European Respiratory Journal 43(3):817–823. 10.1183/09031936.00055513.
- Shelley J, Fairclough SJ, Knowles ZR, Southern KW, McCormack P, Dawson EA, Graves LE, Hanlon C 2018 A formative study exploring perceptions of physical activity and physical activity monitoring among children and young people with cystic fibrosis and health care professionals. BMC Pediatrics 18:335.
- Stoltz DA, Meyerholz DK, Welsh MJ, Longo DL 2015 Origins of cystic fibrosis lung disease. New England Journal of Medicine 372(4):351–362. 10.1056/NEJMra1300109.
- Sushames A, Edwards A, Thompson F, McDermott R, Gebel K, Buchowski M 2016 Validity and reliability of Fitbit Flex for step count, moderate to vigorous physical activity and activity energy expenditure. PLoS One 11(9):e0161224. 10.1371/journal.pone.0161224.
- Swartz A, Bassett D, Moore J, Thompson D, Strath S 2003 Effects of body mass index on the accuracy of an electronic pedometer. International Journal of Sports Medicine 24:588–592.
- Takacs J, Pollock CL, Guenther JR, Bahar M, Napier C, Hunt MA 2014 Validation of the Fitbit one activity monitor device during treadmill walking. Journal of Science and Medicine in Sport 17(5):496–500. 10.1016/j.jsams.2013.10.241.
- Tedesco S, Sica M, Ancillao A, Timmons S, Barton J, O’Flynn B 2019 Validity evaluation of the Fitbit charge2 and the Garmin vivosmart HR+ in free-living environments in an older adult cohort. JMIR mHealth and uHealth 7(6):e13084. 10.2196/13084.
- Tudor-Locke C, Sisson SB, Lee SM, Craig CL, Plotnikoff RC, Bauman A 2006 Evaluation of quality of commercial pedometers. Canadian Journal of Public Health 97(Suppl S1):S10–15. 10.1007/BF03405359.
- Warabi T, Kato M, Kiriyama K, Yoshida T, Kobayashi N 2005 Treadmill walking and overground walking of human subjects compared by recording sole-floor reaction force. Neuroscience Research 53(3):343–348. 10.1016/j.neures.2005.08.005.
- Wright SP, Hall Brown TS, Collier SR, Sandberg K 2017 How consumer physical activity monitors could transform human physiology research. American Journal of Physiology-Regulatory, Integrative and Comparative Physiology 312(3):R358–R367. 10.1152/ajpregu.00349.2016.