We performed a prospective, observational cohort study of people living with COPD admitted to hospital for an AECOPD (AECOPD group) and people with stable COPD symptoms (Stable group). This single-center study was undertaken between January 2018 and December 2019 at the University Hospitals of Leicester, the United Kingdom, where individuals were recruited from hospital wards or the pulmonary rehabilitation (PR) service.

People with an AECOPD were screened by COPD specialist nurses and recruited when medically stable and close to being discharged. People with stable disease were screened by the PR team at their initial PR assessment and enrolled prior to starting their PR program. Inclusion criteria were as follows: being ≥18 years old; having a confirmed clinical diagnosis of COPD from spirometry data in medical records; and for the AECOPD group, an admission with a primary diagnosis of exacerbation of COPD was required. Participants were excluded if they had a physical or visual impairment or comorbidities that prevent participation, required palliative care, were participating in another study, or were unable or unwilling to provide written informed consent.

All participants were asked to wear an Equivital EQ02+ LifeMonitor device (Equivital) (hereby, “vest”) during waking hours for 6 consecutive weeks. During their baseline visit, participants practiced putting on and removing the vest with a researcher first, and then independently while supervised, and they were also given written and visual instructions to take home. Participants were asked to remove the vest during water-based activities and to charge the sensor electronics module (SEM) overnight, at least every other night.

Our patient and public involvement (PPI) group contributed to the design of the study. Specifically, members selected the Equivital LifeMonitor, from a choice of three wearable devices, and provided feedback on the duration of wear period. They also provided feedback on study documentation, wording, and verbal description of the study for recruitment purposes.

The AECOPD group were asked to start wearing the vest following discharge from hospital, making the day after discharge the first day of wear. The Stable group were asked to wear the vest following their baseline study assessment (after initial PR assessment). The day after the baseline visit was the first day of wear (Multimedia Appendix 1).

Participants were contacted by a researcher via telephone 1-3 days after their baseline visit to evaluate acceptability of the vest. Participants had further follow-up telephone calls on a fortnightly basis for troubleshooting purposes or could contact the study team on an ad hoc basis.

The vest measured heart rate (HR), respiratory rate (RR), skin temperature (ST), and physical activity (PA) [31]. PA was classified as stationary or ambulatory using an inbuilt triaxial accelerometer. HR was obtained using built-in ECG electrodes, RR was recorded with a built-in expansion belt, and ST was measured using a thermometer in the SEM (15-second epoch).

Definitions and criteria for feasibility indicators are specified in Table 1. Adherence was defined by the number of days the vest was worn during the 6-week study period and the daily wear time. Missing data were examined and classified as either battery depletion (failing to charge the SEM) or nonwear.

The signal quality of HR and RR were based on the proprietary confidence algorithm, accounting for activity and connection artefacts. Based on manufacturer’s recommendations, a signal quality threshold of 85% was used for HR and 80% for RR to indicate whether each 15-second value was deemed acceptable. Data quality for HR and RR was defined as the percentage of daily wear time with acceptable signal quality. Signal quality and data quality were used to identify the confidence of the data generated by the vest.

Field notes from the follow-up phone calls were analyzed to ascertain any common problems that participants had with the technology or other aspects of the study participation. Acceptability was defined as reporting no problems with the technology.

Vital signs examined were HR, RR, ST, and PA. HR and RR were calculated as the average HR and RR, respectively, during wear time. PA was calculated as the proportion of daily wear time when the patient was ambulatory.

Demographics, clinical histories, comorbidities, and spirometry data were obtained from medical records or information provided by participants. Height and weight were obtained from medical records or measured. Chest circumference and waist circumference were measured.

The Medical Research Council dyspnea scale [32] was used to measure breathlessness.

No formal sample size calculation was undertaken for this feasibility study. A sample size of 50 participants per group was decided on the basis of potential suitable participants, logistics, and resources available.

Data were analyzed using R (version 4.0.0). Continuous variables distributions were tested for normality. Data are reported as mean (SD) or median (IQR) and differences between groups were assessed using a 2-sample unpaired t test or Mann–Whitney U test, respectively. Frequency comparisons between groups were assessed using the Fisher test. The Spearman rank correlation coefficient (r_s) was used to analyze associations between variables (Cronbach α=.05).

All participants provided written informed consent (Research Ethics Committee 15/LO/2055) and the study was prospectively registered (ISRCTN12855961).

Figure 1 outlines recruitment details, reasons for withdrawal, and completion rate for the AECOPD and Stable groups. The AECOPD group had a lower BMI, more severe dyspnea, more hospital admissions, and more frequent exacerbations, but they were otherwise similar to the Stable group (Table 2).

For all 84 participants, the vest was worn for a median of 37.0 (IQR 27.8-40.0) days and the median daily wear time was 12.0 (IQR 10.8-13.1) hours. The median HR signal quality was 90% (IQR 80%-94%), and the median RR signal quality was 93% (IQR 92%-95%; Figure 2A). The median HR data quality was 81% (IQR 58%-91%), and the median RR data quality was 85% (IQR 77%-91%; Figure 2B).

There were no significant between-group differences in the number of days the vest was worn, the longest number of consecutive days worn, or the average daily wear time (Table 3). The AECOPD group spent a significantly lesser median time ambulatory (10.1%, IQR 8.6%-15.0% vs 13.4%, IQR 9.5%-19.3%; P=.03) and showed a lower median HR signal quality (88.5%, IQR 75.8%-92.6% vs 92.3%, IQR 81.0%-96.4%; P=.04) than the Stable group.

From follow-up phone calls, 21 participants (60%) in the AECOPD group and 36 participants (74%) in the Stable group found the vest acceptable. Five (14%) participants in the AECOPD group reported that they did not wear the vest while unsettled or feeling unwell after returning home. Five (10%) participants in the Stable group and 5 (14%) participants in the AECOPD group experienced some discomfort wearing the vest. Three (9%) participants and 5 (10%) participants in the AECOPD and Stable groups, respectively, reported that they did not wear the vest on days on which they felt unwell. Four (11%) participants in the AECOPD group had problems removing the SEM from the cradle of the vest, compared to one participant (2%) in the Stable group.

For the whole sample, 33 (40%) participants were allocated a larger vest size, which was comparable between groups. Compared to those who completed the study, a larger proportion of participants who withdrew required a larger vest size than the manufacturer’s guidance (71% vs 34%, P=.005). There were no associations between participants vest fitting and other feasibility measures (Multimedia Appendix 1).

For the Stable group, HR signal quality and HR data quality were positively correlated with BMI (r_s=0.45, P=.008; r_s=0.44, P=.008; Figures 3A and 3B). For the AECOPD group, RR data quality was negatively correlated with waist circumference and BMI (r_s=–0.49, P=.009; r_s=–0.44, P=.02; Figures 3C and 3D).

In this study, it was possible to continuously measure vital signs (RR, HR, ST, and PA) in free-living conditions using multiparameter wearable technology for people living with COPD after hospitalization and during stable symptoms. Measurement of vital signs using our technology was more challenging post AECOPD. The Equivital LifeMonitor was acceptable to some people with stable COPD symptoms and post AECOPD and produced data of sufficient quality. It should be noted that some participants felt uncomfortable with a device around their chest, and data quality was influenced by body composition. Overall, continuous VSM during daily life is possible, and its potential utility for supporting patient post AECOPD should be explored.

In this study, some participants stated that they did not want to wear a device around their chest. As well as the physical implications of a vest-like device, people with COPD may be influenced by the psychological impact of a device around the chest, as reported previously [26]. While most participants in our study reported no discomfort wearing the vest, similar to a previous study measuring respiratory rate with a chest belt [26], some participants reported that the vest felt restrictive at times or made them feel breathless. Perceived breathlessness could have influenced vest acceptability, as COPD populations tend to prefer watch-like devices [33]. However, measuring respiratory rate from such devices is challenging [34]. The use of a wearable, wireless patch has been successful in continuously monitoring vital signs of inpatients [35], and although this would reduce the perceived breathlessness with a vest-like device, Rubio et al [34] reported that a chest band could measure the RR more reliably than patch-like devices.

Recruitment was more challenging in the AECOPD group, with 15.4% of eligible patients recruited for the AECOPD group, compared to 41.3% in the Stable group, as seen in previous similar studies [9,17]. In a study where patients were asked to record oxygen saturation (SpO₂) and RR using 2 separate devices for 2 months, 79% measured SpO₂ and 60% measured RR three times per day, while 98% and 83% measured SpO₂ and RR, respectively, once per day [26]. In this study, we observed a comparable adherence to findings when patients took daily measures (84% of participants provided vest data) but were able to capture a vast amount of data (on average 12 hours of data). Compared to the AECOPD group, a greater proportion of the Stable group wore the vest for >5 weeks (46% vs 71%). Technology continues to develop multiparametric wearable devices [36-39] to reduce patient burden, but it is possible that this remains a significant barrier in an AECOPD population with lower digital literacy.

Participant feedback from telephone calls suggested that the vest was acceptable overall, with 68% of participants reporting no problems. Some participants reported that they chose not to wear the vest on the days that they felt unwell. Vitacca et al [18] asked participants to complete a weekly 12-item Respicard (recording symptoms, SpO₂, and HR) for 6 months and identified that participants with worse respiratory values had poorer adherence. In this study, 10% of the AECOPD group withdrew because they experienced problems wearing the vest, and 14% of participants in the AECOPD group struggled to engage with the device once returning home. Following discharge from hospital post AECOPD, symptoms remain elevated, and it takes time for patients to recover to their normal symptoms and daily activities [7,40]. Despite our single piece of technology reducing the need for patients to measure multiple vital signs and the observational nature of the study, the greater symptom burden in an AECOPD population reduced adherence to wearable technology.

Our results show that some patients were unable to participate as their chest size exceeded the maximum vest size, or they felt that the maximum vest size was not a suitable fitting. The Equivital LifeMonitor used in this study was originally designed to monitor vital signs in a military population [41]. A greater proportion of those who withdrew required a larger vest size than the completers (71% vs 34%). Existing wearable technology is more broadly marketed toward a healthy population and is typically not tailored for people living with COPD. While the form of the technology used in this study was generally acceptable, advancements in more discrete technologies are needed.

Similar to previous reports in healthy men [31], this study shows that HR and RR measurements obtained from the vest are of sufficient quality in a COPD population. Evidence suggests accurate vital signs measurements of clothing monitors such as the Zephyr BioHarness and Hexoskin [42,43], but our PPI members found such devices challenging to put on and remove owing to their tight-fitting nature. Despite the technology used in this study being tested by our PPI group, physical impairments affected participants’ ability to wear the vest and charge the SEM. Compared to 2% of the Stable group, 11% of the AECOPD group reported problems putting the SEM in the cradle, with some needing help from a cohabitant. The HR signal and data quality were also worse for people with a lower BMI, which is seen more often in an AECOPD population [44-46]. This may be owing to lower body composition and lower conductance [47]; therefore, a weaker connection between the skin and the electrodes embedded in the vest. These problems have been observed elsewhere [26] and must be considered by manufacturers, researchers, and clinicians when selecting digital health technologies.

Although the number of patients assessed in this study was low as a proportion of patients screened, introducing the possibility of selection bias, recruitment is often challenging post AECOPD [17,28,29]. Our single piece of technology aimed to passively capture multiple vital signs; however, some participants may prefer active participation to obtain recordings. It was not possible to measure SpO₂ and blood pressure in the continuous and unobtrusive manner in line with this study. The lack of an age-matched healthy control group prevented us from identifying the unique difficulties with the use of multiparameter technology in a COPD population. This study may have benefitted from measuring vital signs overnight, to obtain individualized “baseline” vital sign values. A more rigorous qualitative exploration of participants’ experiences would have provided greater insights than telephone call field notes.

Following hospitalization for AECOPD and during stable symptoms, it was possible to continuously measure RR, HR, ST, and PA using multiparameter wearable technology during free-living conditions. The Equivital LifeMonitor was acceptable to participants and produced data of sufficient quality, despite some reports of discomfort with wearing a device around the chest and data quality influenced by body composition. Overall, continuous VSM during daily life is possible for people living with COPD and its potential utility for supporting patients post AECOPD should be further explored.