This prospective cohort study was conducted in an 18-bed surgery ICU at Renji Hospital, Shanghai, China. Patients aged 18 years or older who underwent gastrointestinal, biliary and pancreatic, urology, or orthopedic surgery and were transferred to the ICU from June 2022 to September 2022 were screened. Patients with 1 or more of the following conditions were excluded: neurological or mental disorders, dementia, chronic sleep disorders, failure to achieve postanesthesia resuscitation and to be extubated before 6:00 PM, use of remifentanil exceeding 0.03 μg/kg/min, hemodynamic instability (defined as an irreversible systolic blood pressure below 90 mm Hg), or at a high risk of reoperation during the study period as determined by the attending physician.

This study was approved by the Ethics Committee of Renji Hospital (KY2022-145-A). All data have been anonymized to protect patient privacy, and participants benefit from long-term follow-up care as compensation. Written informed consent was obtained from patients or their legal representatives.

The sleep monitoring period was from 9:00 PM on the night of the surgery to 7:00 AM on the following day. A wearable wristband (HONOR BAND 6, Huawei Technologies Co., Ltd) was used to collect sleep data. At 7:00 AM the next day, the bedside nurse removed the wristband and transferred the encrypted sleep data to a customized application. The nurse-RCSQ was then completed by the bedside nurse, while a nonbedside nurse interviewed the patient and filled out the self-RCSQ.

Patient demographic data, medication history, operation information, ICU treatments, and sleep parameters were collected. Both patient- and nurse-reported RCSQ scores were rated from 0 (Bad) to 100 (Good) on a visual analog scale [15] (Multimedia Appendix 1). Sleep data from wearable wristbands were automatically translated into sleep parameters, including Huawei sleep score, light sleep, deep sleep, REM, and wakefulness, with customized, built-in software that translates cardiopulmonary coupling and accelerometry data into sleep-wake periods [14].

Patients were divided into two study groups based on the surgical approach: (1) MIS group—patients who underwent laparoscopic or robotic-assisted techniques via small skin incisions [5] and (2) TOS group—patients who underwent traditional, open surgical techniques.

The primary outcome was the sleep scores of patients who underwent MIS or TOS. The secondary outcomes included the consistency of sleep status assessed by wearable wristbands and RCSQs, as well as quantitative sleep parameters from the wristbands.

Based on preliminary data, and based on a .05 α level and 80% power, we calculated a sample size of 25 per group to observe an average increase of 10 points in Huawei and RCSQ sleep scores in patients who underwent MIS compared to patients who underwent TOS [16]. All values are presented as mean (SD), median (IQR), number, and percentage (%) as appropriate. Continuous data were analyzed using 2-tailed t tests or the Mann-Whitney U test, as appropriate. Bland-Altman concordance analysis was performed to assess the agreement between the 2 assessments. Multivariable linear regression was conducted to estimate the effects of surgical mode on sleep-related indexes. A 2-sided P value <.05 was considered statistically significant. Anthropometric data and measurements were analyzed using R software (version 4.10; R Foundation for Statistical Computing).

A total of 61 patients received sleep assessments; 28 (46%) patients underwent MIS, while 33 (54%) patients underwent TOS. Table 1 presents the demographic and perioperative characteristics between the MIS and TOS groups. The MIS group had a lower proportion of upper abdominal surgeries and a higher proportion of lower abdominal and pelvic surgeries compared to the TOS group.

An average increase of 10 points was observed in the nurse-RCSQ score (mean 60.9, SD 16.9 vs mean 51.2, SD 17.3; P=.03), self-RCSQ score (mean 58.6, SD 16.2 vs mean 49.5, SD 14.8; P=.03), and Huawei sleep score (mean 77.9, SD 4.5 vs mean 68.6, SD 11.1; P<.001) in patients who underwent MIS compared to patients who underwent TOS (Figure 1). After adjusting for baseline information including the surgical site, the effect of surgical mode (MIS or TOS) was significant for the self-RCSQ score (P=.03) and Huawei sleep score (P<.001), as shown in Table 2. Furthermore, the Bland-Altman analysis demonstrated a high level of consistency, with over 95% agreement between the Huawei sleep score and self-RCSQ (58/61, 95%) and nurse-RCSQ scores (59/61, 97%; Figure 2).

Quantitative analysis on the duration and depth of sleep using Huawei software is presented in Table 3. The MIS group had a 1-hour longer total sleep time (mean 503.0, SD 91.4 vs mean 437.9, SD 144.0 min; P=.04) and a 25-minute longer REM sleep time (mean 81.0, SD 52.1 vs mean 55.8, SD 44.5 min; P=.047) compared to the TOS group. Similarly, patients who underwent MIS had a higher deep sleep continuity score (mean 56.4, SD 7.0 vs mean 47.5, SD 12.1; P=.001) compared to patients who underwent TOS. However, the deep sleep time, light sleep time, deep sleep ratio, light sleep ratio, REM sleep ratio, and awakening frequency showed no significant differences in patients who underwent MIS versus TOS(all P>.05). Multiple linear regression showed that the effects of the surgical mode (MIS or TOS) on Huawei sleep score, total sleep time (P=.03), REM sleep time (P=.02), and deep sleep continuity score (P=.02) were all significant after adjustment for age, sex, Acute Physiology and Chronic Health Evaluation (APACHE) score, use of sedation and analgesia, number of drainage tubes and medical interventions, and surgical site (Table 4).

This study reveals that patients who underwent MIS exhibit better sleep quality, characterized by longer total sleep time, increased REM sleep duration, and higher deep sleep continuity compared to those who underwent TOS. Furthermore, wearable sleep monitoring wristbands provided quantitative sleep data that align with RCSQ assessments. These findings provide initial support for the use of modern sleep monitoring technology in managing sleep in patients in surgical ICUs.

Disruptions to regular sleep-wake cycles can lead to a range of complications, including anxiety, immune disorders, prolonged hospitalization, and social reintegration difficulties [17]. MIS techniques reduced invasiveness and promoted faster recovery, even in critically ill patients. Our results show a significant 10-point improvement in sleep scores for patients who underwent MIS compared to TOS, consistent with previous findings indicating that major surgeries can exacerbate sleep disturbances compared to minor ones [18]. Our findings can be attributed to a combination of physical and psychological factors. Laparoscopic repair is associated with reduced postoperative pain and faster recovery while minimizing psychological stress, as shown by lower anxiety and stress scores [19,20]. Consequently, patients who underwent MIS require fewer narcotics, including opioids, on postoperative days 0 and 1, which reduces sleep disturbances caused by anesthesia use [18,21]. At the molecular level, MIS minimizes mechanically induced damages and immune responses, leading to reduced systemic release of acute phase proteins, leukocytosis, and interleukin-6 [22]. This reduction in inflammatory responses may affect circadian rhythm and sleep quality by impacting vagal projections, sympathetic ganglia, and the blood-brain barrier [23,24]. Our local quality control report in 2021 also found that patients who underwent MIS had higher scores for sleep and overall in-hospital satisfaction compared to patients who underwent TOS.

Our Huawei data show that patients who underwent MIS experienced a significant reduction in sleep disruption, with an additional hour of total sleep time, an extra 25 minutes of REM sleep, and improved deep sleep continuity. REM sleep is a critical component of sleep quality, characterized by active brain activity, vivid dreaming, and relaxed skeletal muscles [25]. Major surgeries often disrupt REM sleep, with reports of significant suppression following open heart surgery [26] and a decline of REM sleep from 18% to 0% after open cholecystectomy on the night of the surgery [27]. Sleep continuity is a key indicator of sleep quality, as disruptions can lead to fragmented sleep, impaired memory consolidation, and cognitive dysfunction [28]. In our study, both groups had relatively low sleep continuity, consistent with previous research showing that even sufficient sleep duration can be fragmented in postoperative patients in the ICU [29]. Assessing sleep continuity may be a clinically relevant and reproducible method for evaluating sleep disruption in ICU settings [30].

Quantifying the sleep status of patients in the ICU is challenging due to the difficulties in implementing PSG measurement. Laboratory studies have shown that wristbands exhibit high sensitivity (0.965) and accuracy (0.863) compared to PSG [31]. In critically ill patients, wristbands have been validated in correlating with PSG to identify total sleep time and wakefulness [12,32]. However, patient selection in previous studies had affected monitoring accuracy, as wristbands monitoring has much less specificity for patients who are ventilated compared with those who are not (51% vs 83.7%), possibly due to sedation and constraints [33]. Our study focuses on patients who have regained consciousness after anesthesia and found that Huawei wristband sleep scores show high consistency (>95%) with both self-RCSQ and nurse-RCSQ scores. Our results suggest that this technology is suitable for monitoring sleep patterns in patients after they wake up in the ICU.

Wearable sleep monitoring devices offer a distinct advantage in clinical and research settings for patients in the ICU as they provide both objective, quantitative information on sleep states and continuous monitoring capabilities for future analysis. For instance, these devices can facilitate the investigation of sleep disorders and their correlation with adverse outcomes, such as delirium [13,34]. They offer a valuable early warning system for sleep-related complications and serve as a tool to assess the impact of ICU interventions and medications on sleep quality and patient outcomes. Additionally, long-term sleep monitoring, spanning over several days, and its variability (night-to-night variability) can have profound implications on patient outcomes and the well-being of health care associates [35]. The collaboration between medical and engineering professionals enables the utilization of extensive clinical data to guide the development of hardware and software for these devices, evolving them from wearable health monitors into professional medical devices.

This study, conducted at a single center, has several limitations. First, our sleep monitoring period (between 9 PM and 7 AM) was limited by local sleep and clinical work habits, which may introduce bias. In future research, it would be desirable to conduct continuous sleep monitoring in the ICU to examine sleep patterns over a longer period. Second, the higher absolute value of the Huawei sleep score in our study suggests that if we aim for large-scale clinical use, a validated wearable device scoring system with clear and simple standards is necessary. Finally, a larger, multicenter study is necessary to validate the findings through continuous recruitment.

Compared to TOS, MIS contributed to a higher sleep quality for patients in the ICU after surgery, manifested as longer sleep time, longer REM sleep time, and better continuity of deep sleep. Wearable monitoring wristbands hold the potential for quantified sleep assessment in ICU settings.