Feasibility of continuous physical activity monitoring: first-month recovery markers following joint replacement surgery

Introduction

Orthopedic procedures and especially joint replacement surgeries require careful postoperative observation to ensure optimal patient recovery. Adjusting treatment plans based on real-time patient data is crucial for facilitating this recovery process (Whiting et al., 2015). However, the conventional method of regular follow-up visits scheduled at fixed intervals and conducted in hospital outpatient clinics often proves inconvenient for patients and may fail to timely identify issues in the patient’s recovery process (Gualandi et al., 2019). To address these challenges, telemedicine has emerged as a promising solution, leveraging technological advancements to enable more frequent and remote monitoring of patients (Bahadori et al., 2020).

This innovative approach is particularly relevant in the field of orthopedics, where wearable technology has gained considerable attention. Wearable sensors allow continuous monitoring and enable objective data collection on various aspects of physical activity (PA), reducing reliance on potentially biased self-reports and providing a comprehensive view of a patient’s activity levels and recovery progress (Wainwright & Kehlet, 2022). Continuous monitoring facilitates the early detection of postoperative complications, allowing timely interventions (Master et al., 2022). Additionally, data from wearable devices can inform personalized rehabilitation programs, tailoring exercises to individual needs (Rodgers et al., 2019). Wearable technology also enhances patient engagement by providing real-time feedback and encouraging compliance with rehabilitation protocols  (Bahadori, Immins & Wainwright, 2018).

To effectively utilize PA as a functional indicator for postoperative recovery, it is crucial to have a comprehensive understanding of the dynamic nature of both the quantity and quality of PA following surgery. Unlike simple indicators such as heart rate or body temperature, PA provides a more holistic assessment of the rehabilitation and functional recovery process, integrating both the intensity and pattern of movements (Arnold, Walters & Ferrar, 2016). However, the interpretation and reliability of PA data are complex and influenced by factors including postoperative pain, medications, mood, and individual motivation (Sim et al., 2022). While subjective patient expectations can affect perceived recovery outcomes, the objective measurements provided by PA trackers offer a reliable and unbiased assessment of PA levels during the postoperative period.

Previous studies have explored the role of sensor-based PA monitoring in orthopaedic surgeries, particularly arthroplasties (Iovanel, Ayers & Zheng, 2023; Hammett et al., 2018). However, these investigations often lacked continuous assessment, focusing instead on discrete postoperative intervals (Gianzina et al., 2023; Sosdian et al., 2014; Boekesteijn et al., 2022). Feasibility studies highlight the potential of wearable sensors in various clinical settings, stressing the importance of evaluating practical aspects such as patient adherence, data accuracy, and overall acceptability (Lebleu et al., 2021; Luna et al., 2019). However, a significant research gap exists in understanding how wearable technology can be employed for continuous monitoring of postoperative recovery, especially during the immediate postoperative period when patients are most vulnerable to complications  (Natarajan et al., 2023). Therefore, further investigation is warranted to fully explore the capabilities of wearable technology in monitoring postoperative recovery and its potential as a valuable tool in assessing the nuanced changes in PA during this critical phase.

This study is an observational feasibility study aimed at assessing the practicality and efficacy of continuously measuring PA using wearable sensors in patients who underwent total knee arthroplasty (TKA) and total hip arthroplasty (THA). Our primary objective was to evaluate the feasibility of utilizing wearable sensors to capture continuous PA data during the perioperative period. Additionally, we examined the changes in PA patterns and levels following orthopedic surgeries in a detailed and comprehensive manner. We analyzed the variations in the quantity of different daily PAs during this early postoperative period, evaluated their patterns, and identified the most predictable, consistent, and valuable variables. Through this feasibility study, we aimed to provide insights into the utilization of wearable sensors for monitoring PA in orthopedic surgery patients, with a specific focus on the early postoperative period.

Materials & Methods

Results

The characteristics of the patients are demonstrated in Table 1. Thirteen out of the 16 patients in the study had previous lower limb surgeries performed more than six months before the current surgery, with previous surgeries including contralateral hip (n = 6) and knee arthroplasty (n = 3), osteotomies around the knee (n = 2), and other lower limb surgeries (n = 2). All included patients underwent primary arthroplasties in the knee or hip joints, and the surgical procedures and postoperative periods were without complications. A dataset comprising 660 days of data from 16 patients was used for analysis. Complete data consisting of 14 days of preoperative data and 28 days of postoperative data were available for 11 participants, and five patients had slightly deviated data availability. Specifically, two patients had 12 days of preoperative data, while another had 13 preoperative days of data. Additionally, one patient provided 26 days of postoperative data, and another provided 23 days. These minor variations in data availability were considered during subsequent analyses and interpretation of the study’s findings.

Feasibility outcomes

Data collection compliance: The overall compliance rate was high, with minimal gaps in data. Despite the missing 12 days of data, most patients provided complete data sets for both preoperative and postoperative periods.

Wearability and comfort: Patients reported no significant discomfort while wearing the PA trackers. No adverse events related to the use of the devices were reported, indicating good tolerability.

Data transmission and technical issues: Data transmission via Bluetooth to the web server was successful for most of the patients. Issues with data transmission were promptly addressed by contacting the patients, and there were no significant disruptions in data flow.

Figure 1 illustrates the variations in different PA categories throughout the perioperative period, providing a visual representation of how PA levels fluctuate before and after surgery.

Table 2 displays the median and interquartile range (IQR) of various PA categories during the preoperative period and four weeks after surgery, along with the corresponding p-values for comparing consecutive periods.

Additionally, Fig. 2 features a heatmap visualizing the average autocorrelation coefficients for various PA categories across all study subjects at different day lags. The heatmap highlights significant patterns, particularly showing that certain PA categories, such as continuous walking time, activity intensity count, and regular and slow continuous step counts, exhibit more consistent and stable changes over time compared to others.

Discussion

This study supports the practicality and effectiveness of wearable sensors in continuously monitoring PA in patients undergoing TKA and THA. The feasibility of using these sensors to collect continuous PA data throughout the perioperative period was demonstrated, with significant changes observed in key PA variables—such as step counts, continuous walking duration, and activity intensity count—during the critical early postoperative phase. These findings underscore the potential of wearable sensors to provide real-time, objective insights into patient recovery.

To ensure a robust baseline for postoperative comparisons, we collected wearable sensor data consistently over a two-week period before each subject’s surgery. Although the literature suggests that 3 to 7 days of PA monitoring is generally sufficient to provide reliable estimates of habitual activity levels, even in patients with reduced mobility (Ward et al., 2005; Hilden et al., 2023; Hart et al., 2011), we extended this period to 14 days. This extended monitoring period was specifically chosen to better capture daily variations and mitigate potential biases introduced by preoperative factors such as pain and reduced range of motion (ROM). We continued PA data collection for four weeks post-surgery to identify early recovery trends. This process confirmed the feasibility of collecting such data, revealing significant changes in PA, particularly in continuous walking metrics like step count and walking duration. A notable decrease in these variables was observed immediately after surgery, followed by a gradual increase over the first four weeks, suggesting their value as recovery indicators. Consistently, activity intensity counts corroborated these findings. Aligning with previous studies (Crizer et al., 2017), we noted changes in moderate PA and walking post-surgery, though these studies typically report outcomes months post-surgery (Harding et al., 2014; Lin et al., 2013; Lützner, Kirschner & Lützner, 2014). In a recent study, Christensen et al. (2023) examined a large cohort undergoing TKA and identified significant post-surgical improvements in various PA tracker variables, highlighting gait parameters as crucial recovery markers.

We employed autocorrelation to uncover temporal patterns in the time series data, which facilitates comprehending postoperative recovery dynamics and recognizing stable recovery markers. The strong short-term autocorrelation in categories like continuous walking time, activity intensity count, regular continuous steps, and slow continuous steps indicate stable and predictable PA patterns in the immediate postoperative period. This consistency reflects adherence to rehabilitation protocols, emphasizing regular, controlled walking to enhance recovery, build endurance, and prevent complications such as muscle atrophy and joint stiffness (Ward et al., 2021). However, it is important to recognize that while our study shows positive recovery trends, overall progress appears longer than anticipated. Full recovery often extends well beyond the first four weeks post-surgery, with many patients not returning to their baseline PA levels (Harding et al., 2014; Rivera et al., 2024; Bin sheeha et al., 2020). This highlights the challenges of early recovery, including persistent pain, reduced mobility, and the psychological effects of surgery—all of which can delay the return to normal activity levels. The discrepancies between patient-reported outcomes (PROMs) and objective PA measurements further underscore the need to integrate objective assessments, such as those provided by wearable sensors, with PROMs to achieve a more comprehensive understanding of recovery trajectories (Wainwright & Kehlet, 2022).

We observed no significant postoperative changes in daily sitting and standing times compared to the preoperative period. This lack of variation suggests that these variables might not effectively capture the nuances of early recovery. Interestingly, we also found no significant change in sporadic walking and the number of sporadic steps. This limitation could be attributed to the current sensor technology, which tends to be more accurate in detecting continuous walking patterns rather than sporadic movements. Wearable sensors may struggle with identifying irregular, brief periods of activity, potentially leading to underreporting or inaccuracies in these data points. Studies, particularly in populations with conditions like Parkinson’s disease, have similarly shown reduced accuracy in capturing sporadic walking (Salarian et al., 2004; Nguyen et al., 2017; Caldas et al., 2017). Addressing these limitations will be crucial for the development of future sensor technologies. Potential improvements might include refining the algorithms that detect short bursts of activity or enhancing the sensitivity of the devices to better differentiate between various types of movement (Yang et al., 2022).

While we noted no significant difference in sitting time during the initial postoperative week compared to the preoperative period, there was a notable increase in the second week, possibly attributed to increased bed rest, followed by a subsequent decrease. The time spent in a recumbent position was consistently higher post-surgery for the entire four-week period, albeit with some fluctuations, indicating areas for further exploration. Interestingly, sit-to-stand movements decreased post-surgery but then significantly increased in the first two weeks, eventually stabilizing. This trend was more consistent in comparison to sitting time. Sit-to-stand, a critical mobility metric, was observed in frequencies within the range reported in existing literature (Bohannon, 2015). Understanding the frequency of sit-to-stand movements can be crucial in identifying recovery challenges.

This study pioneers in assessing both the quantity and pattern of continuous PA data using a small wearable sensor as an early marker of recovery post-orthopedic surgery. However, it has several limitations. It is important to acknowledge the small sample size and the potential for generalization issues between TKA and THA patients. The rate of recovery can vary significantly between these groups, and drawing broad conclusions on detailed recovery is challenging with the limited number of participants in each subgroup. Notably, the inclusion of a 24-year-old woman with achondroplasia in the TKA group, diverging from the typical TKA patient demographic, was a deliberate choice due to the study’s focus on feasibility over comprehensive demographic representation. Future studies with larger sample sizes and separate analyses for TKA and THA patients are recommended to provide more definitive conclusions. Furthermore, the uneven distribution of participants between the TKA and THA groups precluded detailed statistical comparisons. While we did not validate the PA changes against established recovery assessment tools, our focus was to explore the feasibility of PA monitoring in the early postoperative phase. Despite the ongoing process of full validation, preliminary evidence supports the potential of wearable sensors in this context (Bartholdy et al., 2018; Christensen et al., 2014; Ghaffari et al., 2023b).

Our study demonstrates the feasibility of using small wearable sensors in collecting PA data during the perioperative period following orthopedic surgeries, marking a significant advancement in the realm of telemedicine for post-surgery care. The introduction of these sensors has transformed the collection and analysis of human motion data, offering a more convenient and precise approach compared to traditional in-person recovery monitoring methods. By tracking PA, these sensors provide essential insights into a patient’s health and well-being during recovery (Bohannon, 2015; Hendricks, Chong & Cusick, 2018; Warburton, Nicol & Bredin, 2006; Bowyer & Royse, 2016). However, while PA data serves as a promising recovery marker due to its significant, stable, and predictable changes post-surgery, it is crucial to recognize that recovery may extend well beyond the early postoperative period  (Wainwright & Kehlet, 2022). The full potential of this technology in broader, population-based studies remains to be explored. Future research should focus on larger-scale analyses to deepen our understanding of PA patterns in diverse patient groups and further establish PA as a reliable indicator of successful recovery.

Conclusions

This study demonstrates the feasibility of using small wearable sensors to continuously and objectively track PA during the critical perioperative phase of orthopedic recovery. Specifically, variables such as step counts, continuous walking duration, and activity intensity count emerged as promising markers for remotely monitoring early recovery after TKA and THA surgeries. However, given the small sample size and the exploratory nature of this study, these findings should be interpreted with caution.

While our results offer preliminary insights into the potential utility of wearable sensors in postoperative care, further research with larger, more diverse populations is essential to validate these findings and refine the use of these markers in clinical practice.

Supplemental Information