The purpose of this study was to: 1) examine the physical activity levels and associated cardiorespiratory responses of individuals with stroke during normal daily activities which included their structured physical therapy sessions, and 2) examine the relationship between a previously reported activity level classification with measured physiological responses to daily activity (heart rate, ventilation rate). We used a commercially available wearable ambulatory physiological monitoring system. This study linked measured physiologic change with specific daily activities including activities associated with structured rehabilitation sessions, as well as the activities and times when the individuals were not in therapy.
Importantly, the findings of this study provide direct physiologic evidence to support the suggestion that individuals with stroke are generally inactive throughout the day, which is consistent with observational reports in the literature [10, 11]. Little information regarding the activity patterns of individuals with stroke throughout the day is available. Though Bernhardt et al.  demonstrated that individuals in the acute phase of recovery following stroke are generally inactive according to a subjective scale rating the therapeutic level of various activities from 0 (inactive) to 4 (highly therapeutic), the findings of the current study suggest that, among individuals in the subacute stage of recovery, even activities included in the categories of highest therapeutic relevance (e.g. walking) may not load the cardiorespiratory system sufficiently to elicit a training effect. Although both HR and VR generally increased with the subjectively rated AC, large individual differences in these relationships, as well as large ranges in the measures of HR and VR within each AC, for each individual. These findings suggest that physiological load cannot be assessed directly from AC. For example, S1 and S4 demonstrated marked differences in mean HR and VR across the AC levels (Figure 5). S2 demonstrated similar differences across AC levels 1, 3, and 4, though both HR and VR were greater for AC0 than for AC1. S2 spent only 2% of the testing session in the AC0 category (see Table 1), and this period may have been marked by an elevated HR and VR for reasons other than physical activity (i.e. anxiety or other stress). S3 demonstrated no apparent change in HR or VR across the AC categories. In addition to individual differences in the physiological response to activity across the individuals, each patient demonstrated large ranges in the measures of HR and VR within each AC, particularly for AC3 and AC4. For S1, HR ranged between 91 and 102 bpm during AC3 activities, and between 90 and 130 during AC4 activities. For S2, HR ranged between 74 and 95 bpm during AC3 activities, and between 81 and 101 bpm during AC4 activities. The other two participants demonstrated similar HR responses. The average HR range during AC3 activities across the four individuals was 18 bpm; during AC4 activities, the average HR range was 32 bpm. Furthermore, S1 demonstrated HR responses adequate to elicit a physiological training effect (i.e. HR greater than the minimum threshold for a training effect according to the American Heart Association scientific statement) for less than 50% of the time this individual spent in AC4 'highly therapeutic' activities. During AC3 'moderately therapeutic' activities, this same individual's HR did not enter the training zone at all. Clearly, an observational measure of activity level does not adequately describe the physiological load, or potential benefit, of individual activities, and addition of physiological parameters such as HR or VR are needed to assess the physiological load of activity for individuals. Ambulatory monitoring of physiological load during activity provides the capacity to assess the aerobic challenge associated with activity and adjust the intensity of activity on a person-to-person basis.
While previous work has inferred the therapeutic relevance of physical activity based on the expert opinion of experienced clinicians, the current study has added direct physiological measurement of the physiological load the activity to the understanding of the (potential) health benefits associated with the activity. This additional information available through the use of the physiologic monitoring has provided three important insights. First, and consistent with work by MacKay-Lyons and Makrides , the physiological load experienced by individuals during structured therapy sessions may not be sufficient to elicit a cardiovascular benefit or training effect. Second, tremendous individual differences exist in the individual's physiological response to physical activity during therapy and throughout the day. Third, even during activities which are deemed by expert opinion to be highly therapeutic, large ranges in measures of physiological response (i.e. heart rate, ventilation rate) suggest that these activities do not necessarily provide a cardiovascular training effect. These insights confirm that it is imperative that ambulatory physiological measurement systems (i.e. wearable heart rate monitors) be used during physical therapy sessions not only to ensure the safety of the patient, but also (and likely more commonly) to ensure that the patient is engaged with sufficient intensity to challenge the cardiovascular system to the point of training effect. Furthermore, the findings of the current study underscore the need to better understand the nature of the physical activities engaged in by individuals throughout the day, such as the type of activity, the duration that specific activities are performed, and the intensity of the activity in terms of the cardiorespiratory response. Ambulatory physiological monitoring of individuals with stroke throughout the day may provide a method of influencing individual activity profiles on a day-to-day basis and eventually via a method of real-time monitoring and prompting.
Activities engaged in by the individuals throughout the day were categorized according to a previously established method using observation techniques to infer therapeutic value of physiologic loads associated with activity . The results suggested that the four participants in the current study were engaged in activity that was deemed non-therapeutic for, on average, slightly more than 50% (range of 23 to 65%) of the day, which is consistent with the report of Bernhardt and colleagues . The individuals in the previous study spent 28% of the day engaged in minimally therapeutic activities (i.e. sitting supported out of bed). The individuals in the current study did not perform any activities that were considered to be in the minimally therapeutic category. Therefore, it seems that the individuals in the current study had a greater volume and extent of activity in categories of higher therapeutic relevance due, in part, to their higher functional capacity. For example, they were all capable of sitting unsupported, and therefore spent a larger percentage of the day, according to this scale, engaged in moderately and highly therapeutic activity (50% of the day, versus 12.8% in the previous study). The previous work by Bernhardt  examined individuals with stroke at an early stage of recovery while the current study explored activity profiles of in-patients who were later in their stage of recovery (four to eight weeks after stroke). It is unlikely that individuals able to ambulate independently (with aids), such as those who participated in the current study, would find sitting unsupported substantially challenging from a sensorimotor perspective or in terms of cardiovascular load, and therefore the recovery time differences may explain the increase in activities which, according to this scale, would be considered therapeutically-relevant if using the activity scale. These findings suggest that development of an alternate activity level scale designed specifically for individuals at later stages of recovery following stroke might be useful and more discriminative in assessing the physiological challenge of various daily activities.
A limitation of this study was sample size; a research assistant was required to spend 8 to 9 hours observing each participant, limiting the feasible number of participants, and limiting data collections to a single day. Therefore, the sample of participants included individuals who varied greatly in age and neurologic impairment, in order to explore in a case-study approach the level of activity among patients with stroke, and the relationship between activity level classification and continuously sampled physiological response. The development of movement assessment capability (e.g. accelerometers) and validation of the discriminative capacity of such measurements (to distinguish movement profiles) is essential to improve the practical application of this approach to remove time and cost constraints imposed by the necessity of a research assistant to manually document participant activities all day long. Such remote measurement of movement, as opposed to relying on observation, would also help to counter limitations associated with privacy and observation. In addition, it is possible that the participants may have altered their normal daily activities, or altered the level of effort provided during various tasks as a result of being observed throughout the day. It should be noted, however, that one might have anticipated an increase in relative activity under such a scenario and in the case of the present individuals they were characterized by relatively low levels of daily activity.
This study confirms and extends the results of previous research providing a detailed view of the activity patterns of individual patients with stroke and the associated physiological response throughout the day. First, the activity level of individuals with stroke during structured therapy sessions may not be of sufficient physiological challenge to elicit a cardiovascular training effect. Second, they appear to be relatively inactive throughout the day, and simple observation of their physical activity may not assess the therapeutic relevance of the activity. Third, we have associated a measure of physiological challenge with the individual's activities of daily living. Future research will examine methods of influencing the activity level of individuals with stroke in the rehabilitation hospital and community. Incumbent in that research will be the development of technology which will associate kinematic measurements with physiologic data. Such developments will facilitate inclusion of a larger sample size by autonomously providing a context of activity to the physiologic measure, reducing the cost of data gathering and enhancing feasibility. Data acquisition systems built on emerging sensor technologies will provide the understanding of the individual's activity necessary for meaningful interpretation of the physiologic response to activity throughout the day (and night). Information related to activity obtained at times outside of structured therapy sessions may serve to provide important insight regarding the individual's status not otherwise available to the therapist. In addition, these developments will allow precise measurement of function and intensity of activity in the community, promoting evidence-based therapeutic practice following discharge from the daily therapy program or rehabilitation hospital.