Physical Activity Monitoring and Acceptance of a Commercial Activity Tracker in Adult Patients with Haemophilia

Physical activity (PA) is highly beneficial for people with haemophilia (PWH), however, studies that objectively monitor the PA in this population are scarce. This study aimed to monitor the daily PA and analyse its evolution over time in a cohort of PWH using a commercial activity tracker. In addition, this work analyses the relationship between PA levels, demographics, and joint health status, as well as the acceptance and adherence to the activity tracker. Twenty-six PWH were asked to wear a Fitbit Charge HR for 13 weeks. According to the steps/day in the first week, data were divided into two groups: Active Group (AG; ≥10,000 steps/day) and Non-Active Group (NAG; <10,000 steps/day). Correlations between PA and patient characteristics were studied using the Pearson coefficient. Participants’ user experience was analysed with a questionnaire. The 10,000 steps/day was reached by 57.7% of participants, with 12,603 (1525) and 7495 (1626) being the mean steps/day of the AG and NAG, respectively. In general, no significant variations (p > 0.05) in PA levels or adherence to wristband were produced. Only the correlation between very active minutes and arthropathy was significant (r = −0.40, p = 0.045). Results of the questionnaire showed a high level of satisfaction. In summary, PWH are able to comply with the PA recommendations, and the Fitbit wristband is a valid tool for a continuous and long-term monitoring of PA. However, by itself, the use of a wristband is not enough motivation to increase PA levels.


Introduction
Haemophilia is a rare inherited haematological disorder that affects the blood coagulation process [1]. The clinical manifestations include an increased tendency for spontaneous bleeding into a joint, called haemarthrosis. Deficient levels of factor VIII (haemophilia A) or factor IX (haemophilia B) affect the severity of bleeding [1,2]. Bleeds in the elbows, knees, and ankles are the most common injuries in haemophilia and often occur during daily activities [3,4]. If repeated episodes are not treated, chronic damage evolves over time into chronic haemophilic arthropathy, which accounts for 75% of complications in haemophilia [4].
inhibitors (≥5 Bethesda units) and (2) having suffered joint or muscle bleeding in the last three months. During 2015, a total of 125 patients with haemophilia were attended to in routine visits at the hospital, 75 met the inclusion criteria, and finally, 28 agreed to participate in the study. Participation was voluntary and all participants signed a written informed consent. This study conformed to The Declaration of Helsinki and was approved by the Human Research Ethics Committee (H1406715601199) of the University of Valencia. This article adheres to the STROBE guidelines [33].

Procedures
Participants' data were extracted from medical health records and a personal interview. The following variables were recorded: Age, height, weight, BMI, educational level, computer skills, pharmacological treatment, history of bleeding, and joint health status according to the radiological Pettersson scale (maximum score 78 points, 13 points × 6 joints) [34] and the clinical Haemophilia Joint Health Score 2.1 (HJHS) (maximum score 124 points, 20 points × 6 joints, plus 4 points assigned to global gait) [35]. In both scales, higher values represent a worse outcome. The HJHS evaluation was performed by a senior physiotherapist widely experienced in haemophilic arthropathy.
In addition, on the first visit, a Fitbit Charge HR activity wristband (Fitbit, San Francisco, CA, USA) was delivered to each participant. Although this device records a large number of variables (distance travelled, energy consumed, continuous heart rate, stairs climb, sleep quality, etc.), this study focuses on analyzing the number of steps, active minutes, and wear time. Fitbit wristbands calculate the active minutes using metabolic equivalents (METs) for physical activities (walking, running, cycling, multi-sport, etc.) maintained at least 10 min in a row as follows: <3 METs, light active; 3-6 METs, fairly active; >6 METs, very active. One MET indicates the basal metabolic rate and is based on height, weight, age, and gender. Fitbit uses its own algorithm to obtain the equivalent METs from the number of counts provided by the accelerometer [36].
Participants were informed of the device's functions and were instructed to wear the wristband continuously from morning to night (except for water-related activities), follow their progress on the Fitbit website or mobile app, and synchronize and charge the wristband at least two times a week to minimize lost data.
During the first week of monitoring, participants were to continue with their usual daily routines. Average daily steps of this first week were considered the steps/day at baseline. After this week, participants were encouraged to try to comply with the recommendation of 10,000 steps/day or to increase the steps/day if they usually reached this recommendation already. However, they were not instructed to perform a new specific physical activity or athletic program. Using the baseline steps/day, participant data were divided into two groups for comparison and analysis: Active Group (AG), formed by participants with a number of steps/day greater than or equal to 10,000, and Non-Active Group (NAG), participants with less than 10,000 steps/day.
After monitoring PA during a 13-week follow-up, participants were called for a second visit to discuss their recorded data with them and to record weight and musculoskeletal bleedings. In all cases, less than 1000 steps/day corresponds to a non-typical day; for example, not having worn the wristband for most of the day. Therefore, a valid day of measurement was defined as a day with more than 1000 steps. In addition, to analyse the experience with the use of the Fitbit wristband, participants completed a technology acceptance questionnaire developed by Mercer et al. [37]. This 17-item questionnaire assesses the domains of external variables, perceived usefulness, perceived ease of use, attitude toward using, behavioural intention to use, and actual system use. Each item can be scored from 1 (strongly disagree) to 5 (strongly agree).

Data Processing and Statistical Analysis
In order to manage and download the data of patients' wristbands remotely, the researchers developed their own custom-made software. The downloaded data were processed later using Matlab (The MathWorks, Inc., Natick, MA, USA, version R2015a) software.
Normality of the data was verified using the Shapiro-Wilk test. Descriptive measures are shown with the mean and standard deviation. Anthropometric data and questionnaire results from the AG and NAG were compared using an unpaired t-test or Wilcoxon rank-sum test, depending on normality. The pre-post weight was compared using a paired t-test. Correlations between PA and the patients' characteristics were analysed using the Pearson coefficient. Results of correlations were interpreted as very weak (r < 0.20), weak (r ≥ 0.20 and r < 0.40), moderate (r ≥ 0.40 and r < 0.60), strong (r ≥ 0.60 and r < 0.80), or very strong (r ≥ 0.8).
A two-factor analysis of variance (ANOVA) [week range (4) × group (2)] was used to determine differences in physical activity variables over the course of the study period in both groups. The Bonferroni post hoc correction was applied to avoid type I error in the multiple comparisons when the ANOVA analysis indicated significant differences. Statistical significance was considered at p < 0.05. All statistical analyses were done in IBM SPSS Statistics for Windows (Version 22.0, IBM Corp., Armonk, NY, USA).

Sample Size
An a priori power analysis was conducted in G* power (Heinrich-Heine-Universität Düsseldorf, Düsseldorf, Germany, version 3.1.9.2) software to calculate the required sample size. With the present study design (ANOVA (4 × 2)), accepting a 5% alpha risk (α = 0.05) as well as a 20% beta risk (β = 0.2; power = 0.8), a total of 24 subjects were required to achieve at least a medium effect size (f = 0.25; Regarding the correlation analysis, with α = 0.05, power = 0.8, and the 26 final participants, the minimum effect size detectable is r = 0.52.

Participants
The group of 75 patients who met the inclusion criteria had a mean age of 37.24 (11.07) years old and presented a mean Pettersson score of 32.05 (21.34) points, while the study group had a mean age of 36.08 (9.54) years old and a mean Pettersson score of 29.40 (21.17) points. Since the between-group differences were not statistically significant (p > 0.05), the study group was representative with respect to the total sample in age and degree of arthropathy.
Two of the patients dropped out of the study due to personal reasons, with 26 participants constituting the final sample. All participants had computer skills and habitually used smartphones. According to the baseline steps/day, the AG and NAG were composed of 15 and 11 participants, respectively. Participant characteristics are shown in Table 1. The between-group analysis did not show significant differences at baseline.
After the 13-week follow-up, the AG presented a weight of 76.85 (15.33) kg and the NAG a weight of 79.46 (27.10) kg, with no significant differences with respect to baseline values (p > 0.05). During the monitoring period, four patients suffered bleeds in load joints (three patients from AG and one patient from NAG). The bleeds in the AG were all minor and not related to PA: One ankle haemarthrosis, provoked by ankle sprain while working the same day the patient received the prophylactic treatment; one spontaneous hip haemarthrosis, one day after the last FVIII infusion (mean steps/day: 13,220; steps on the day of bleeding: 12,925); and spontaneous bleeding in the knee in a patient who received daily prophylactic treatment (mean steps/day: 12,347; steps on the day of bleeding: 12,249 steps). In the NAG, one patient suffered two bleeds: One minor spontaneous bleed in the knee, 48 h post infusion of FVIII (mean steps/day: 7200; steps on the day of bleeding: 8176), and one ankle haemarthrosis related to PA caused by a long walk in a short time, 24 h after the last FVIII administration (steps on the day of bleeding: 12,407).

Physical Activity Monitoring
According to the Tudor-Locke et al. scale at baseline, one of the participants was sedentary, five were low-active, five were somewhat active, six were active, and nine were highly active, with 57.7% of the participants reaching the target of 10,000 daily steps. The average of the 26 participants was 10,441 (2997) steps/day (12,603 (1525) steps/day in the AG and 7495 (1626) steps/day in the NAG). For the AG, the mean number of steps/day in week one was 13,152 (1748) and remained similar in weeks 2-5 and 6-9, but decreased significantly (p = 0.027) in weeks 10-13. Regarding the NAG, the mean number of steps/day was 7578 (1979) at baseline, with no significant differences (p > 0.05) compared with the other periods. In both groups, the number of active minutes (light, fairly, and very active) remained constant over the weeks, with no significant differences ( Table 2). Throughout the study period, the 84.6% of participants reached the 150 min of PA per week. Figure 1 shows the daily mean steps (and 95% confidence interval) per week for both groups. Values are mean (standard deviation). p1: Within-group differences at baseline and weeks 2-5; p2: Differences at baseline and weeks 6-9; p3: Differences at baseline and weeks 10-13. ‡ : Indicate significant differences. *: Between-group significant differences (p < 0.05), **: Between-group significant differences (p < 0.001). significant differences. *: Between-group significant differences (p < 0.05), **: Between-group significant differences (p < 0.001).

Correlation Analysis
In general, the number of steps and minutes of activity (light, fairly, and very active) do not show significant correlations with age, BMI, and HJHS (Table 3). Only the correlation between very active minutes and HJHS score was significant (p = 0.045), but this correlation was weak (r = −0.40). Table 3. Results of correlation analysis.

Demographics
Steps

Technology Acceptance Questionnaire
Results are shown in Table 4. The total average was high (4.24 (0.51) for AG and 4.19 (0.42) for NAG), with no significant differences between groups. The items with the highest scores were: "I found it easy to learn to operate the activity tracker" and "Overall, the activity tracker was easy to use". The between-group analysis indicates that the AG presented higher values than the NAG in the number of steps (p < 0.001), distance (p < 0.001), and levels of activity (light, fairly, and very active min/day) (p < 0.05). These results were repeated in all periods analysed except for the fairly active minutes in weeks 10-13 (p > 0.05).
Of the 91 total recorded days, the mean valid days (>1000 steps/day) was 86.8 (6.5). At baseline, the average wear time (min/day) was 912.3 (107.6) for the AG and 870.9 (105.3) for the NAG, remaining high throughout the study period and comparable across groups and weeks with no significant differences (p > 0.05) ( Table 2).

Correlation Analysis
In general, the number of steps and minutes of activity (light, fairly, and very active) do not show significant correlations with age, BMI, and HJHS (Table 3). Only the correlation between very active minutes and HJHS score was significant (p = 0.045), but this correlation was weak (r = −0.40).

Technology Acceptance Questionnaire
Results are shown in Table 4. The total average was high (4.24 (0.51) for AG and 4.19 (0.42) for NAG), with no significant differences between groups. The items with the highest scores were: "I found it easy to learn to operate the activity tracker" and "Overall, the activity tracker was easy to use".

Discussion
In this work, a commercial activity tracker was used to perform a 13-week monitoring of daily PA in 26 adults with haemophilia. Our results showed that 57.7% of participants achieved the goal of 10,000 daily steps, and the 84.6% reported a mean number of active minutes greater than the recommended minimum to be healthy.
PA levels of the general population depend on multiple factors, such as country, sex, age, race/ethnicity, etc. [38,39]. Tudor-Locke et al. concluded that healthy adults typically perform a number of daily steps between 4000 and 18,000 [38].  [30][31][32]. Comparatively, the number of steps taken by the AG in our study is higher than those reported by Pérez-Alenda et al. and Goto et al., and even superior to the results of some studies conducted in the general population. Moreover, the mean number of steps of the NAG is higher than that showed by Goto et al. and Bassett et al., but lower than the rest of the studies analysed. In addition, our results showed that both groups comply with the recommendations of active minutes per week [14,15]. However, it should be noted that the exact algorithms used by Fitbit to classify PA levels using sensor data are unknown to consumers and researchers due to proprietary concerns, making it difficult to compare PA levels between the devices of different companies [21,36].
In addition to factors involved in the general population, levels of PA in PWH may decrease depending on age, BMI, and arthropathy [8,42]. In our study, the NAG presented a higher level of arthropathy than the AG, but this difference was not significant (p > 0.05). Furthermore, no strong correlations were found between the analysed variables. Only the relationship between very active minutes and joint health status was significant (p = 0.045), but this correlation was weak (r = −0.40).
Many of the studies performed with research accelerometers are carried out over a short period of seven-day monitoring [16]. This is due to the storage capacity and the need to return devices to researchers to extract the data. However, the data recorded with consumer trackers can be frequently transferred to a website [21]. Another important feature of these devices is that, according to some studies, they present good accuracy, validity, and reliability in the step count, and therefore, they are suitable for research purposes [19,20,43]. These characteristics have allowed us to obtain objective data remotely and to be able to analyse the evolution of PA levels over a long period. Our results show a high number of valid recorded days and minutes of wear time, allowing us to ensure that results are not biased by the amount of time the wristband was used. In addition, the wear time was maintained throughout the study period with no significant decline, demonstrating a very good adherence. Given the PA levels at baseline, no significant variations in the number of steps and active minutes were produced during the weeks monitored, with the exception of the steps/day in the AG. In this case, a significant decrease (p = 0.027) of the steps/day was observed in weeks 10-13. Regarding the usability and usefulness of the Fitbit Charge HR, both groups presented high scores on the technology acceptance questionnaire, indicating that participants were satisfied with the fitness tracker. These scores are higher than those obtained in the devices analysed in the work of Mercer et al. [37]. Given these results, the Fitbit Charge HR has been shown to be an adequate device for continuous and long-term monitoring of PA, but nevertheless, their use does not seem to be sufficient motivation to increase levels of PA in PWH, it being necessary to complement the use of the wristband with additional motivational strategies [30].
Finally, during the study period, three patients in the AG suffered minor bleeds not related to PA in load joints. In the NAG, one patient registered two bleeds, one of them due to a considerably higher than usual level of PA. Hence, despite arthropathy, PWH in prophylactic treatment are able to comply with PA recommendations for health with a minimal risk of bleeding.
Our study sample was limited to patients attended to in a single haemophilia centre, and, therefore, not all results can be extrapolated to the general haemophilia population. Furthermore, future studies are needed to investigate the use of the activity tracker in combination with additional strategies to encourage PA.

Conclusions
The results indicate that, despite arthropathy and thanks to the prophylactic treatment, adult patients with haemophilia are able to comply with PA recommendations with a minimal risk of bleeding. In addition, commercial activity trackers are suitable for continuous and long-term monitoring of PA in PWH, due to their technological characteristics as well as the high degree of wear adherence and satisfaction of use. Therefore, this type of device can help healthcare providers to optimize outcomes and make better use of available resources, allowing a tailored prophylaxis therapy based on the objective PA level of their patients. However, in addition to the wristband, it seems necessary to use additional motivational strategies to increase PA levels in non-active patients. Funding: This study was funded by Baxalta U.S. Inc., now part of Shire under an investigator-initiated research grant (H14-23641). Shire does not necessarily endorse, support, or agree with any or all the content.