A feasibility study investigating the use of a thigh-worn accelerometer to measure physical activity in patients recovering from critical illness

INTRODUCTION

The consequences of critical illness are multifaceted, and can include significant physical deconditioning, acquired weakness and reduced functional capacity.¹ Physical activity is advocated for survivors of critical illness, initiated early in their critical care stay,² although currently there is little agreement regarding the optimum frequency and type of interventions.³ Outcome measures largely focus on specific functional milestones that are not necessarily representative of the patient’s consistent level of function.⁴ As an alternative, measurements of overall physical activity beyond rehabilitation sessions, could facilitate a more accurate representation of a patient’s recovery trajectory⁵ and provide more meaningful data in order to tailor rehabilitation interventions.⁶

Wearable technology provides a simple, unobtrusive and objective approach to continuous activity monitoring.⁷ Wearable devices for activity monitoring take many forms, measuring different variables to capture both physiological and activity data.⁸ They provide continuous data regarding activity, without any additional effort from the patient or staff. Wearable device application during critical illness poses unique challenges with various aspects of feasibility discussed in the literature such as wear time, comfort, adverse events and ease of data interpretation.^4,9,10 Wrist and ankle devices are common, however these may not be appropriate for patients during critical illness who are administered medications and invasively monitored using their limbs. Limb oedema, line insertions and dressings have been identified as complications associated with use in this population.^4,9,10 An alternative is a thigh-worn device, which has been examined within inpatient populations but not within critical care.¹¹ These devices can distinguish between lying/sitting and standing postures, which is of significance in the investigation of sedentary behaviours and low-level rehabilitation.¹² These devices provide objective activity data capturing acceleration and thigh position information, generating a near continuous picture of patient position and activity unable to be measured using subjective methods.

Investigation of feasible activity measurement tools in this population could provide valuable information for researchers and clinicians, and is currently lacking.⁷ To date, studies have only captured data over relatively short periods of time. Consequently, it is unclear if wearable devices are a feasible data collection tool throughout critical care admission. This should be examined due to the unique challenges posed in this environment. Although arm and ankle devices have been studied to some extent within critical care,^4,9,10,13 no study has examined the feasibility of a thigh-worn device as an alternative.

AIM

This study aimed to investigate the feasibility of using a thigh worn accelerometer to measure physical activity levels of patients recovering from critical illness whilst in critical care.

METHODS

APPROVALS

The study was approved by the Wales Research Ethics Committee 3 (reference: 18/WA/0086) (IRAS: 238464). Both University College London (UCL) Joint Research Office (ID:18IR06) and Cardiff and Vale Research and Development office (ID:7237) reviewed this study and gave their approval. The study was registered with UCL data protection (reference: Z6364106/2018/02/52) prior to data collection.

PARTICIPANTS

This feasibility study was a single-centre, prospective observational study. It was conducted within the 33-bed critical-care unit at University Hospital of Wales, Cardiff; a tertiary referral centre with a case mix of general medical, surgical, neurosurgical and trauma patients. Screening of potential participants was completed through routine daily reviews of new patients admitted to the unit. Inclusion and exclusion criteria are summarised in Table 1.

Table 1:  Inclusion and exclusion criteria

With the exception of age, pre-existing neuromuscular disease and unable to wear the activPAL^TM, the exclusion criteria were the same criteria used by the physiotherapy team to decide if a patient was appropriate for rehabilitation input. Determining if a patient was at risk of physical morbidity according to guidelines was completed using the short clinical assessment tool within the NICE¹⁴ guidelines. Patients ‘at risk’ were included as their length of admission was likely to be longer, therefore allowing comprehensive assessment of the device.

INTERVENTION

The activPAL4^TM (PAL Technologies Limited) was the wearable device used in the study. This tri-axial accelerometer attaches to the thigh and records data in 15 second epochs. Using position and acceleration information it can differentiate between sitting/lying, standing and stepping. Step count is also recorded. Figure 1 shows the device and its position on the thigh.

See Figure 1: activPAL4^TM device and device in situ 

The device was applied within 48 hours of admission to allow comprehensive evaluation of device feasibility at all stages of the critical care stay. Participants were not required to do anything different to their usual care apart from wearing the activPAL^TM device on their thigh for the duration of their critical care admission. Nursing staff completed standardised data collection sheets during the day shift at the same time as routine observations to document the participant’s activity during the preceding hour.

Raw data from the devices were downloaded into Microsoft Excel^TM [version 16.13.1] and exported to IBM SPSS^TM [version 22] for analysis. As there was no hypothesis testing, no statistical tests were performed. Feasibility included successful recruitment and retention of participants, wear time (downloaded from the device), number of adverse events (as reported in the medical notes), reasons for device removal and percentage agreement between device-recorded and nurse-observed activity. Feasibility outcomes were compared against a-priori criteria for success. They were based on data from previous studies and the recommended data needed for analysis of physical activity.^{15, 16} A wear time of ten hours per day and at least three days of data were considered as the criteria for feasibility success.

RESULTS

Data were collected between April and June 2018. Of 141 patients screened for participation, 12 were successfully recruited (Figure 2). One device was removed during preparation for discharge and lost, therefore no data were available. Analysis is based on 11 participants.

The demographic characteristics of the sample are displayed in Table 2. Due to the small sample size, non-parametric summaries were calculated.

Table 2:  Sample demographics 

WEAR TIME

The median (IQR) duration of device wear time was 268.34 (299.15) hours. This equated to a median (IQR) percentage wear time of 99.58% (5.69) of critical care admission. Adequate wear time was achieved for every participant (See Figure 3). In total, there were nine days in which sufficient wear time was not reached. Initial application of the device late in the day was responsible for six of these days.

No adverse events occurred. In total devices were removed on nine separate occasions, most commonly for recharging (n=3). Reasons for unplanned device removal (n=6) included imaging procedures, surgery and independent removal secondary to agitation. One device was removed from a participant before they were placed in a prone position. Additionally, it was noted that some activity was recorded in participants who were sedated and not moving.

DISCUSSION

This study aimed to investigate the feasibility of using a thigh worn accelerometer to measure physical activity levels of patients recovering from critical illness during critical care admission. To the author’s knowledge this was the first study to specifically investigate this, adding to the current literature regarding wearable technology use in this population. The key findings were that device wear time satisfied a-priori feasibility criteria, no adverse events occurred and device removal was required to allow recharging or a small range of procedures.

Results suggest that achieving a wear time of sufficient duration to estimate activity levels may be feasible in a select number of patients in critical care (8.5% of those screened for the study were recruited). Owing to time and resource constraints the inclusion and exclusion criteria allowed the researcher to focus the study on participants who would likely to be participating in active rehabilitation during the early phase of their critical care admission. Any future study design would need to allow inclusion of a wider range of participants for a more comprehensive evaluation in this population.

Wear time criteria were achieved for every participant, although there were days when ten hours of wear time was not reached. Unplanned removals were only responsible for three of these occasions. Circumstances demanding device removal are unavoidable in critically ill patients who may require scans and a mean (SD) wear time of 4.4 (0.8) days and 46.5 (2.3) hours respectively equating to 97% wear time. As wear time in the current study was measured throughout admission, a longer median (IQR) wear time of 268.34 (299.15) hours was achieved. As the devices remained in situ throughout admission, the effect of short periods of removal on overall wear time were less potent.

In agreement with previous studies, no adverse events were reported, supporting the safety of device use within critical care.^4,9,10,13 In critical illness survivors, removal of limb-worn devices due to oedema has been reported.⁵ This was not an issue for the activPAL^TM, which was applied to participants with different sized legs without needing to adjust the fit of the device. There were no requirements for the device to be moved for insertion of lines during the study as previously described.^4,9,10 The number of purposeful device removals was higher than previously reported for wrist and ankle devices in a critical care population. This is likely a reflection of the longer data collection phase for the current study. Participants were excluded from one study if they required any procedures necessitating device removal.⁴ The current study may provide a more representative evaluation of instances during critical illness that impact upon the ability to wear an activity monitor.

Although not directly related to the study’s primary aim, it was noted that as only a single device was used, distinguishing between sitting and lying positions was not possible, and on two occasions activity was detected in a participant who was sedated. Sitting may form a significant part of a patient’s rehabilitation and represent improvement in their recovery, therefore identification of both positions would be an important distinction to make. Occasions when activity was identified by the device while the patients were fully sedated require further investigation and raise concerns regarding its validity.

One important limitation of this study was the small sample size, which was primarily a result of time constraints. Reduced resources meant screening of 48 potential participants was not possible. Consequently, the sample was less likely to be representative of the population. As this was a single-centre study, external validity is reduced and results cannot be generalised to the wider critical care population.¹⁷ Only a single thigh-worn device was employed, therefore no robust conclusions can be made about the feasibility of similar devices. In addition, qualitative aspects of feasibility were not investigated. This would have provided a more comprehensive evaluation of the device in this unique population and is an area for further research.

CONCLUSION

This study offers some preliminary information to researchers and clinicians seeking to utilise wearable devices to monitor physical activity in patients recovering from critical illness. Results suggest the practical application of this device is feasible throughout critical care admission. Generalisability to the wider critical care population is reduced due to the small sample and single-centre design. Additional research is warranted to further investigate the validity of this device in these patients and explore the feasibility of such devices to inform the rehabilitation of patients recovering from critical illness.

Key points

1. The activPAL^TM accelerometer appears feasible to use with patients recovering from critical illness.
2. The lack of ability to distinguish between lying and sitting may limit its use within a critical care population.
3. Areas for further research, including validity of the activPAL^TM in this population have been highlighted.

FUNDING

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors. The project was undertaken as part of an MSc degree at UCL for which no external funding was received.

ACKNOWLEDGEMENTS

Grateful thanks to the patients and their families who participated in this study and to the nursing staff for kindly helping with observations. Additional thanks to the research nurses for their help with consenting. Thanks to Paul Twose and Erica Thornton for their help with data collection.

1. Kang J, Jeong YJ, Yun SY, et al. Post-Intensive care syndrome experience among critical care survivors: A meta-synthesis of qualitative research. Journal of Korean Critical Care Nursing. 2017;10(3):13-30. 
2. Needham DM, Feldman DR, Kho ME. The functional costs of ICU survivorship: collaborating to improve post-ICU disability. Am J Respir Crit Care Med. 2011;183(8):962-964. doi:10.1164/rccm.20101 2-2042ed
3. Connolly B. Describing and measuring recovery and rehabilitation after critical illness. Current Opinion in Critical Care. 2015;21(5):445-452. doi:10.1 097/mcc.0000000000000233
4. Kamdar BB, Kadden DJ, Vangala S, et al. Feasibility of continuous actigraphy in patients in a medical intensive care unit. Am J Crit Care. 2017;26(4):329-335. doi:10.4037/ajcc2017660
5. Denehy L, Berney S, Whitburn L, Edbrooke L. Quantifying physical activity levels of survivors of intensive care: a prospective observational study. Physical Therapy. 2012;92(12):1507-1517. doi:10.252 2/ptj.20110411
6. Connolly BA, Mortimore JL, Douiri A, Rose JW, Hart N, Berney SC. Low levels of physical activity during critical illness and weaning: The evidence–reality gap. J Intensive Care Med. 2017;34(10):818-827. doi:10.1177/0885066617716377
7. Verceles AC, Hager ER. Use of accelerometry to monitor physical activity in critically ill subjects: a systematic review. Respir Care. 2015;60(9):1330-1336. doi:10.4187/respcare.03677
8. Sylvia LG, Bernstein EE, Hubbard JL, Keating L, Anderson EJ. A practical guide to measuring physical activity. Journal of the Academy of Nutrition and Dietetics. 2014;114(2):199-208. doi:10.1016/j.jand.201
9. Beach LJ, Fetterplace K, Edbrooke L, et al. Measurement of physical activity levels in the Intensive Care Unit and functional outcomes: An observational study. Journal of Critical Care. 2017;40:189-196. doi:10.1016/j.jcrc.2017.04.006
10. Winkelman C, Higgins PA, Chen YJK. Activity in the chronically critically III. Dimensions of Critical Care Nursing. 2005;24(6):281-290. doi:10.1097/00003 465-200511000-00011
11. Baldwin C, van Kessel G, Phillips A, Johnston K. Accelerometry shows inpatients with acute medical or surgical conditions spend little time upright and are highly sedentary: Systematic review. Physical Therapy. 2017;97(11):1044-1065. doi:10.1093/ptj/pzx 076
12. Byrom B, Stratton G, Mc Carthy M, Muehlhausen W. Objective measurement of sedentary behaviour using accelerometers. Int J Obes. 2016;40(11):1809-1812. doi:10.1038/ijo.2016.136
13. Grap MJ, Borchers CT, Munro CL, Elswick RK Jr, Sessler CN. Actigraphy in the critically ill: correlation with activity, agitation, and sedation. American Journal of Critical Care. 2005;14(1):52-60. doi:10.403 7/ajcc2005.14.1.52
14. Rehabilitation After Critical Illness (Clinical Guideline 83). National Institute for Health and Care Excellence (NICE); 2009.
15. Edwardson CL, Winkler EAH, Bodicoat DH, et al. Considerations when using the activPAL monitor in field-based research with adult populations. Journal of Sport and Health Science. 2017;6(2):162-178.
16. Trost SG, McIver KL, Pate RR. Conducting accelerometer-based activity assessments in fieldbased research. Medicine and Science in Sports and Exercise. 2005;37(11):S531-S543. doi:10.1249/01.ms 0000185657.86065.98
17. Hammond F. Multicentre Trials. In: Hammond FM, Malec JF, Nick TG, Buschbacher RM, eds. Handbook for Clinical Research: Design, Statistics, and Implementation. Springer Publishing Company; 2014. doi:10.1891/9781617050992