BACKGROUND AND PURPOSE
In individuals with heart failure (HF) who also have implantable cardioverter defibrillator (ICD) or cardiac resynchronization therapy (CRT) devices, low daily activity is associated with greater 2-, 3-, and 5-year mortality1,2 and lower 11-month survival.3 Although daily activity is highly associated with various indices of functional status and exercise tolerance,4 changes in indices of functional and physical performance do not seem to be accompanied by changes in daily activity.5–7 Furthermore, daily activity patterns do not seem to change over time, especially among those with low daily activity levels.2,3,8
In patients with HF, traditional rehabilitation approaches are based on the cycle of inactivity and deconditioning paradigm, where interventions aimed at reversing the deconditioning component of the cycle are expected to result in changes in the inactivity component.9–12 However, this may be a faulty assumption. Interventions for improving daily activity in individuals with HF have been investigated in only 8 studies.7,9–15 Exercise-based interventions alone do not seem to have an impact on daily activity, despite increases in physical perfomance.6,7 Therefore, there is a need for an effective method to measure daily activity over time to accurately track and monitor clinically meaningful change in daily activity. External uniaxial and triaxial accelerometers have the capability to be worn and measure data over extended periods. Triaxial accelerometers, which measure acceleration in the 3 cardinal axes, may provide greater accuracy in measuring energy expenditure compared with uniaxial accelerometers, which typically measure acceleration in the vertical axis.16 However, Kelly et al.17 demonstrated comparable performance between uniaxial and triaxial accelerometers compared with oxygen consumption. It is important to note that external accelerometer use is dependent on patient compliance, and therefore consistency and completeness of data collection can be compromised and are difficult to apply in routine clinical practice. In contrast, activity data from an implanted Medtronic ICD or CRT device are readily and consistently available over long periods of time and are feasible for use in clinical practice. The minimal clinically important difference (MCID) for the Medtronic patient activity measure, which represents the smallest amount of change that a patient would identify as clinically important, is 1.08 hours/64.8 minutes per day.1
Regarding the concurrent validity of the patient activity measure, the accelerometer-based daily activity data from Medtronic ICD/CRT devices has been investigated in 2 previous studies. Shoemaker et al.3 observed a moderate-to-strong linear correlation between Medtronic ICD/CRT daily activity data and RT3 triaxial accelerometer data, although these results were not included in the final publication because of too few data points. A study conducted by Pressler et al.18 examined the concurrent validity of daily physical activity assessed by ICD/CRT devices compared with a validated external accelerometer, the AiperMotion, in adult patients with HF. Their results demonstrated moderate correlations (r = 0.64, P < .001) but high variability with differences of up to several hours.18 Furthermore, Pressler et al.18 did not examine measurement of change in daily activity over time.
Therefore, additional investigation into the concurrent validity of ICD/CRT-based daily activity data is warranted using a different, well-established, and validated external accelerometer such as the Actigraph GT3X (Actigraph LLC, Pensacola, FL) accelerometer, which has been validated and used in low-activity populations such as those with chronic pulmonary disease and HF.2,3,19–21 The purpose of this pilot study was to examine the concurrent validity of Medtronic ICD or CRT–based daily activity data using the Actigraph GT3X accelerometer.
This is a preliminary analysis of the first 16 subjects completing the first phase of a controlled trial involving exercise-based and psychosocial-based interventions to improve daily activity in HF between October 2014 and December 2015 (clinicalTrials.gov ID NCT02331524). The primary purpose of the controlled trial is to investigate the effect of 2 different intervention approaches ( daily activity feedback with associated encouragement and  health coaching with an associated individualized home exercise program) on daily activity, exercise tolerance, HF-related health status, and lower extremity functional strength. The secondary purpose was to examine the concurrent validity of the daily activity data of Medtronic ICD or CRT devices versus the activity count, step count, and activity hours per day measures of the Actigraph GT3X triaxial accelerometer in patients with HF. Subjects were recruited from the Spectrum Health Heart and Lung Specialized Care Clinic, a regional, tertiary care clinic for patients with advanced HF, and from the Spectrum Health Cardiac Device Clinic. The inclusion criteria for the study included adults aged 40 and over, diagnosis of HF with subsequent implantation of a Medtronic ICD or CRT device at least 6 months before study enrollment, and a New York Heart Association Functional Class (NYHA-FC) II to III symptoms on optimal medical therapy. The exclusion criteria included any comorbid medical disease that would prevent safe participation in an individualized exercise program such as severe osteo-, rheumatoid-, and gout-related arthritis, unstable angina, exercise-induced arrhythmias, uncontrollable diabetes, or atrial fibrillation with rapid ventricular rate in the preceding 30 days; recent (<6 weeks) or planned (<6 months) major cardiovascular events or procedures; current participation in a regular exercise training program; and HF due to severe, uncorrected primary valvular disease, congenital heart disease, or obstructive cardiomyopathy. The study protocol was approved by the institutional review boards of Grand Valley State University and Spectrum Health. Informed consent was obtained from each participant and all study procedures ensured that the rights of the subjects were protected.
Daily Activity Measurement
Devices: The GT3X
The external accelerometer selected for this study was the Actigraph GT3X triaxial accelerometer, which measures acceleration in 3 planes over a variety of epochs that can be selected by the investigator. In older versions of Actigraph triaxial accelerometers that used data algorithms compatible with the GT3X,22 intramonitor and intermonitor reliability were high, as well as concurrent validity with indirect calorimetry.21,23 The Actigraph GT3X has also been shown to have high adherence by older adult patients with HF.2 The Actigraph GT3X was worn on the dominant hip by all subjects for 7 days at baseline and at 3-month follow-up. A 7-day monitoring period more accurately depicts weekly routines and reduces the likelihood of a significant “Hawthorne effect.”24,25 The same Actigraph GT3X monitor was used for baseline and follow-up for each subject to minimize error that might result from minor differences in individual monitors.20
Devices: Medtronic ICD/CRT Devices
Medtronic ICD or CRT devices contain a single-axis accelerometer that records the number of daily minutes in which the patient exceeded an activity level equivalent to 70–80 steps per minute. The data are stored in the device for a rolling 14-month period and can be extracted during a “save-to-disk” device interrogation, are also presented as “Patient Activity” on clinical reports as a weekly average, and are graphically portrayed over a 14-month rolling period.
Actigraph GT3X measurements used 60-second epochs using a sampling rate of 30 Hz. Data were processed using the Actilife software (Actigraph Corp, Pensacola, FL) to include wear time validation using the Troiano 2007 algorithm and the Troiano adult 2008 cut points26 for activity intensity categorization, total daily steps, and total daily vector magnitude units (VMUs).
Medtronic device data were extracted using the Medtronic Paceart software and converted to an Excel spreadsheet using a Medtronic proprietary conversion tool or through Medtronic proprietary conversion of Carelink interrogation data.
Computed GT3X parameters included an activity count (an overall VMU of movement in all 3 planes), a step count based on vertical (y axis) data, and hours or activity per day, both overall and based on intensity level of activity. Patient activity, measured in hours per day, was the only variable used from the Medtronic ICD/CRT devices. Daily activity data measured by both devices over 7 days at baseline and 3-month follow-up and were used for analysis.
Intraindividual correlations between measurements were examined using Pearson correlation coefficients for Medtronic patient activity (hours per day) and the following Actigraph GT3X parameters: daily activity count (VMU); x, y, and z axis counts; steps per day; and total hours of light, moderate, and vigorous physical activity. Correlation coefficients were calculated for the 7-day averages and for all daily observations for all subjects at both baseline and 3-month follow-up. Correlation coefficients for change in each variable from baseline to follow-up were calculated using the 7-day averages. Bland–Altman plots were used to examine measurement agreement between the 2 devices using all individual daily observations at baseline and 3-month follow-up for overall agreement and the 7-day average for change between baseline and 3-month follow-up. The Bland–Altman plot compares a known highly accurate method (ie, the GT3X) with a new method (Medtronic ICD/CRT) to assess the degree of agreement between methods.27
Multivariate regression analysis was used to explore potential demographic variables that may explain differences in daily activity between devices. Owing to a small sample size, separate multivariate regression analyses with stepwise entry were conducted for predicting Actigraph GT3X 7-day average activity as the independent variable using the 7-day average ICD/CRT activity and a demographic variable. Demographic variables included age, gender, body mass index, NYHA-FC, baseline 6-minute walk test distance, and baseline 30-second timed chair rise repetitions. The regression equation was then used to correct the ICD/CRT daily activity, which was then included in the aforementioned Bland–Altman analyses. A level of significance of P < .01 was used for all statistical tests. Analyses were completed using the IBM SPSS Statistics version 20 (Armonk, NY).
A total of 16 subjects were included in the analysis. Descriptive characteristics are shown in Table 1. Moderate-to-strong, statistically significant correlations (P < .001) were found between the Medtronic ICD/CRT and GT3X for hours of activity per day, VMU, and steps per day for the 7-day averages and for all individual daily observations over the 7-day monitoring period at baseline and 3-month follow-up (Fig. 1). Similarly, moderate-to-strong, statistically significant correlations (P < .001) were found between devices for change in 7-day averages of hours of activity per day, VMU, and steps per day between baseline and follow-up (Fig. 2). Correlation coefficients between ICD/CRT patient activity and GT3X parameters can be found in Table 2.
Regarding the regression analyses, gender was the only demographic variable that was a statistically significant contributor to a model seeking to explain a greater amount of variance in Actigraph GT3X activity. Alone, the ICD/CRT 7-day average activity explained 85.5% of the variance in Actigraph GT3X activity. Including gender in the model explained an additional 6.8% (Table 3).
Regarding a more detailed description of the agreement analyses, the Bland–Altman plots comparing the daily activity measures from the Medtronic ICD/CRTs and the Actigraph GT3X reveal 2 key findings. First, the Medtronic devices generally accounted for approximately 0.80 hours less activity per day compared with the Actigraph GT3X evidenced by the mean differences between measurements of −0.77 (the middle line of the top panel of Fig. 2). Second, the Bland–Altman plots reveal generally good agreement between the 2 devices with all but 5 data points falling outside the 95% confidence interval limits, although this confidence interval is relatively large (1.17 to −2.71). The magnitude of ICD/CRT activity underestimation (−0.39 hours) and the size of the 95% confidence interval is less when adjusting ICD/CRT activity for gender. The Bland–Altman plots for change in activity between baseline and 3-month follow-up as measured by the ICD/CRT and Actigraph GT3X also demonstrate generally good agreement for the change in daily activity (Fig. 2, bottom panel) with the ICD/CRT devices accounting for approximately 0.20 more hours of change in activity but with a relatively large confidence interval spanning 1.96 hours.
This study demonstrated moderate-to-strong correlations between the Medtronic ICD/CRT patient activity measure and Actigraph GT3X hours of activity, VMU, and steps per day measures, with the strongest correlations between ICD/CRT patient activity and GT3X hours of activity and steps per day that accounted for 69% and 65% of the variance, respectively. Using the weekly averages for hours of activity and steps per day improves the explained variance to 85.5% and 71.2%, respectively. Moderate-to-strong correlations were also found between change in ICD/CRT patient activity and GT3X hours of activity and steps per day, explaining 58% and 74% of the variance, respectively. However, despite the strength of these correlations, Bland–Altman plot analysis demonstrated relatively large confidence intervals for individual measurements (3.88 hours) and change over time (1.96 hours).
Previous work performed by Pressler et al.,18 using the AiperMotion triaxial accelerometer, showed only moderate intraindividual correlations and low agreement (evidenced by a 95% confidence interval of 6.2 hours in the Bland–Altman plots) between device activity measurements. This study demonstrated higher intraindividual correlations and better agreement between devices, suggesting that the Medtronic ICD/CRT may be a valid measure of daily activity and change in daily activity, when compared with the Actigraph GT3X triaxial accelerometer.
Although agreement between devices was generally good, the Medtronic ICD/CRT seems to consistently underestimate daily activity by 0.80 hours both in this study and that of Pressler et al.18 Although the confidence interval for agreement in this study was narrower than that of Pressler et al.18 by several hours, 3.88 hours is still large. A possible explanation is that the ICD/CRT measures activity using a single, sagittal/anterior–posterior (z axis) axis accelerometer with an activity threshold for determining whether the patient was active/walking, whereas the Actigraph GT3X uses a triaxial sensor to measure any activity or body movement. Indeed, higher correlations were found between ICD/CRT hours of activity per day and Actigraph GT3X steps per day (based on vertical axis movement) and y axis counts as compared with VMU. Therefore, the ICD/CRT underestimation of 0.80 hours/day and the relatively large confidence interval is likely due to the Actigraph GT3X capturing other, nonambulatory movement. It is interesting to note that although the ICD/CRT devices use a z axis accelerometer, the correlations between ICD/CRT activity and the x and z axis counts were lower than for the y axis counts. A possible explanation is that subjects likely did not keep the Actigraph GT3X perfectly centered over the lateral hip the entire 7-day monitoring period, introducing variability in how ambulatory activity was captured by these axes.
The mechanism by which gender contributes to error in ICD/CRT daily activity measurement is unclear. Adjusting for gender clearly improves ICD/CRT accuracy, but the 95% confidence interval on the Bland–Altman plot is still 2.5 hours. We hypothesize that females may perform a larger amount of nonambulatory movement/activity, which is captured by the GT3X but not by the ICD/CRT devices. Future studies should further explore these differences in ambulatory and nonambulatory activity, and whether increases in nonambulatory activity are associated with improved clinical outcomes.
Regarding measurement of change in daily activity over time, the Medtronic ICD/CRT seems to be able to detect change (Fig. 2), which captured approximately 0.20 more hours of activity per day and demonstrated a strong correlation with change in steps per day measured by the Actigraph GT3X.
An important observation of the analysis of change is both the magnitude and direction of differences between devices. Examination of the scatter plots of change in ICD/CRT patient activity versus GT3X activity hours per day and steps per day (Fig. 2, bottom panel) reveals that 2 subjects were “misclassified” by the CRT/ICD devices as having increased activity when the GT3X demonstrated a decline in activity. Conversely, another subject was “misclassified” by the CRT/ICD device as having decreased activity when the GT3X demonstrated an increase in activity. However, the magnitude of difference in these 3 cases was less than 0.71 hours. The estimated MCID for ICD/CRT devices is 1.08 hours, and therefore these 3 misclassified subjects do not exceed this MCID threshold. The remaining 13 subjects demonstrated agreement in direction of change measured by both devices, with the magnitude of disagreement being less than 0.89 hours for any given subject across a wide range of activity levels (Fig. 3, bottom panel).
An important corollary finding of this study is revealed in the descriptive statistics stratified by NYHA-FC (Table 4). First, there is greater variability in measured daily activity between devices for those who are Functional Class II. This group also had fewer hours of sedentary activity and a greater number of steps per day.
This is evident in the Bland–Altman plot (Fig. 3, top panel) with a slightly greater variability in agreement between devices for those who are more active. These findings conflict with those found by Pressler et al.18 who found a nonsignificant trend toward lower variations in daily activity measurements between devices in those with Functional Class II HF. As previously discussed, the Actigraph GT3X may capture other, nonambulatory movements not captured by the ICD/CRT devices. This could account for the difference in activity levels between those with Class II and Class III HF, where those with Class II HF have a lower symptom severity and are able to participate in more nonambulatory daily activity.
It is also important to note the extremely low activity of the patients included in this sample. Most subjects in this sample had less than 15 minutes of moderately intense activity in a week as measured by the Actigraph GT3X and none with more than 50 minutes, so it is not clear whether the ICD/CRT devices would be more accurate in a sample that was moderately vigorously active where such activity might be better detected. The sample used by Pressler et al.18 was similarly inactive. Therefore, it is not known to what extent the use of ICD/CRT devices could be problematic for determining whether a patient has met the recommended activity guidelines of at least 150 minutes of activity per week in those who are more active.
To our knowledge, this is the first study to examine the accuracy of the Medtronic ICD/CRT activity measurement to measure change. Without the supporting empiric evidence of this study, several previous studies demonstrated changes in ICD/CRT daily activity associated with several other meaningful clinical outcomes. Shoemaker et al.1 previously observed changes in ICD/CRT daily activity that were associated with changes in clinical status. The PARTNERS HF study28 and Adamson et al.29 demonstrated the utility of Medtronic ICD/CRT patient activity parameter to predict clinical worsening of HF and hospitalization for acute decompensated HF, respectively. Another study by Braunschweig et al.30 demonstrated an increase in mean daily activity after the implantation of the Medtronic InSync II device and improvement in NYHA-FC. The results of this study provide empiric evidence supporting the use of the Medtronic ICD/CRT devices to measure change in daily activity over time.
A primary limitation of this study is the small sample size. However, as a pilot study, it is intended only as a preliminary analysis of the first group of subjects to reach the 3-month follow-up as part of the controlled trial involving exercise-based and psychosocial-based interventions to improve daily activity in HF. Regardless of the argument for a small sample being detrimental, the statistical significance remains high.
Medtronic ICD/CRT devices may provide a sufficiently accurate, responsive measure of daily activity in patients with HF when compared with the Actigraph GT3X in this small preliminary data set.
The authors wish to thank Brenda Boer, RN, of the Spectrum Health Cardiac Device Clinic for her help with subject recruitment and data collection.
1. Shoemaker MJ, Curtis A, Vangsnes E, et al. Analysis of daily activity
from implanted cardiac defibrillators: the minimum clinically important difference and relationship to mortality/life expectancy. World J Cardiovasc Dis. 2012;2(3):129–135.
2. Walsh JT, Charlesworth A, Andrews R, et al. Relation of daily activity
levels in patients with chronic heart failure
to long-term prognosis. Am J Cardiol. 1997;79(10):1364–1369. doi: .
3. Howell J, Strong BM, Weisenberg J, et al. Maximum daily 6 minutes of activity: an index of functional capacity derived from actigraphy and its application to older adults with heart failure
. J Am Geriatr Soc. 2010;58(5):931–936.
4. da Silva VZ, Lima AC, Vargas FT, et al. Association between physical activity
measurements and key parameters of cardiopulmonary exercise testing in patients with heart failure
. J Card Fail. 2013;19(9):635–640. doi: .
5. Shoemaker MJ, Curtis A, Vansgnes E, et al. Clinically meaningful change estimates for the six-minute walk test and daily activity
in individuals with chronic heart failure
. Cardiopulm Phys Ther J. 2013;24(3):21–30.
6. Shoemaker MJ, Keenoy B, Smith B, et al. Interventions to improve daily activity
in individuals with COPD and CHF: a systematic review. World J Cardiovasc Dis. 2013;3:275–286.
7. Savage PA, Shaw AO, Miller MS, et al. Effect of resistance training on physical disability in chronic heart failure
. Med Sci Sports Exerc. 2011;43(8):1379–1386. doi: .
8. Shoemaker MJ, Curtis A, Paul R, et al. Longitudinal daily activity
patterns in individuals with chronic heart failure
: derivation and association with prognosis and clinical characteristics. Cardiopulm Phys Ther J. 2013;24(4):25–33.
9. Gottlieb SS, Fisher ML, Freudenberger R, et al. Effects of exercise training on peak performance and quality of life in congestive heart failure
patients. J Card Fail. 1999;5(3):188–194. doi: .
10. Mueller L, Myers J, Kottman W, et al. Exercise capacity, physical activity
patterns and outcomes six years after cardiac rehabilitation in patients with heart failure
. Clin Rehabil. 2007;21(10):923–931. doi: .
11. Van den Berg-Emons R, Balk A, Bussmann H, et al. Does aerobic training lead to a more active lifestyle and improved quality of life in patients with chronic heart failure
? Eur J Heart Fail. 2004;6(1):95–100.
12. Brodie DA, Inoue A. Motivational interviewing to promote physical activity
for people with chronic heart failure
. J Adv Nurs. 2005;50(5):518–527. doi: .
13. Willenheimer R, Rydberg E, Cline C, et al. Effects on quality of life, symptoms and daily activity
6 months after termination of an exercise training programme in heart failure
patients. Int J Cardiol. 2001;77(1):25–31. doi: .
14. Witham MD, Gray JM, Argo IS, et al. Effect of a seated exercise program to improve physical function and health status in frail patients ≥70 years of age with heart failure
. Am J Cardiol. 2005;95(9):1120–1124. doi: .
15. Witham MD, Fulton RL, Greig CA, et al. Efficacy and cost of an exercise program for functionally impaired older patients with heart failure
: a randomized controlled trial. Circ Heart Fail. 2012;5(2):209–216. doi: .
16. Ayen TG, Montoye HJ. Estimation of energy expenditure with a simulated three-dimensional accelerometer. J Ambul Monit. 1988;1:293–301.
17. Kelly LA, McMillan DGE, Anderson A, et al. Validity of actigraphs uniaxial and triaxial accelerometers for assessment of physical activity
in adults in laboratory conditions. BMC Med Phys. 2013;13:5. doi: .
18. Pressler A, Danner M, Esefeld K, et al. Validity of cardiac implantable electronic devices in assessing daily physical activity
. Int J Cardiol. 2013;168(2):1127–1130.
19. Maurer MS, Cuddihy P, Weisenberg MS, et al. The prevalence and impact of anergia (lack of energy) in subjects with heart failure
and its associations with actigraphy. J Card Fail. 2009;15(2):145–151.
20. Howell J, Delisle S, Jones M, et al. Actigraphy in heart failure
subjects with a reduced and normal ejection fraction. J Card Fail. 2008;14:S110.
21. Van Remoortel H, Raste Y, Louvaris Z, et al. Validity of six activity monitors in chronic obstructive pulmonary disease: a comparison with indirect calorimetry. PLoS One. 2012;7(6):e39198. doi: .
22. GT3X+ and wGT3X+ Device Manual. Pensacola, FL: Actigraph Corp. http://http://www.actigraphcorp.com
/wp-content/themes/ActiGraphFE2012/media/2013/12/MOS13DOC09-Moses-Users-Manual-1.pdf. Accessed February 10, 2014.
23. Brage S. Reliability and validity of the computer science and applications accelerometer in a mechanical setting. Meas Phys Educ Exerc Sci. 2003;7(2):101–119.
24. Steele BG, Holt L, Belza B, et al. Quantitating physical activity
in COPD using a triaxial acceleromter. Chest. 2000;117:1359–1367.
25. Matthews CE, Ainsworth BE, Thompson RW, et al. Sources of variance in daily physical activity
levels as measured by an accelerometer. Med Sci Sports Exerc. 2002;34:1376–1381.
26. Troiano RP, Berrigan D, Dodd KW, et al. Physical activity
in the United States measured by accelerometer. Med Sci Sports Exerc. 2008;40:181–188.
27. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1:307–310.
28. Whellan DJ, Ousdigian KT, Al-Khatib SM, et al. Combined heart failure
device diagnostics identify patients at higher risk of subsequent heart failure
hospitalizations: results from PARTNERS HF (Program to Access and Review Trending Information and Evaluate Correlation to Symptoms in Patients with Heart Failure
) study. J Am Coll Cardiol. 2010;55:1803–1810.
29. Adamson PB, Smith AL, Abraham WT, et al. Continuous autonomic assessment in patients with symptomatic heart failure
: prognostic value of heart rate variability measured by an implanted cardiac resynchronization device. Circulation. 2004;110:2389–2394.
30. Braunschweig F, Mortensen PT, Gras D, et al. Monitoring of physical activity
and heart rate variability in patients with chronic heart failure
using cardiac resynchronization devices. Am J Cardiol. 2005;95:1104–1107.
Keywords:© 2017 Cardiovascular and Pulmonary Section, APTA
daily activity; heart failure; physical activity