Dose–response relationship of early out-of-bed mobilization with functional decline and adverse events in patients after cardiovascular surgery requiring prolonged intensive care: a retrospective cohort study Authors: Kotaro Hirakawa, Kentaro Kamiya, Hiromichi Ueki, Kentaro Hori, Tomohiro Iwakura, Atsuko Nakayama, Mitsuaki Isobe Published: 2026 ABSTRACT AbstractBackgroundPatients requiring prolonged ICU care after cardiovascular surgery are at high risk of functional decline, yet the optimal mobilization dose remains unclear. This study examined the associations between mobilization dose, functional decline at discharge, and adverse events during ICU rehabilitation.MethodsThis retrospective cohort study was conducted in a cardiovascular ICU and included 610 patients who underwent cardiovascular surgery and required ≥ 3 days of postoperative ICU care. The cumulative duration of out-of-bed activities at or above standing level (ICU Mobility Scale ≥ 4) was quantified, with 20 min defined as one unit. Patients were categorized into low-dose (< 2 units/day) and high-dose (≥ 2 units/day) groups. The primary outcome was postoperative functional decline, defined as a ≥ 5-point decrease in the Barthel Index at discharge compared with baseline. Adverse events during mobilization were also assessed. Propensity score matching was performed to adjust for baseline differences.ResultsAfter 1:1 propensity score matching, 196 patients were included in each group. Functional decline was more frequent in the low-dose group than in the high-dose group (31.6% vs. 19.9%; odds ratio 2.03; 95% confidence interval [CI] 1.19–3.45; p = 0.009). Adverse events were infrequent (1.2%) and did not differ significantly between groups (incidence rate ratio 0.88; 95% CI 0.43–1.74; p = 0.70). Restricted cubic spline analysis demonstrated a nonlinear dose–response relationship between mobilization dose and functional outcomes.ConclusionsLower mobilization doses were associated with functional decline, while higher doses were not associated with increased adverse events. Ensuring an adequate mobilization dose may be associated with better preservation of physical function without compromising safety.Supplementary InformationThe online version contains supplementary material available at 10.1186/s40560-026-00892-5.Keywords: Early mobilization, Out-of-bed exercise, Functional decline, Adverse events, Cardiovascular surgery Background Critically ill patients who require treatment in the intensive care unit (ICU) are known to be susceptible to ICU-acquired weakness (ICU-AW), which is associated with mortality, prolonged hospital stay, and delayed recovery following discharge [1]. Prolonged mechanical ventilation, particularly when the duration exceeds 48 h, leads to extended inactivity and accelerates structural and functional muscle deterioration in hospitalized patients [2]. Early mobilization is recognized as an effective strategy for enhancing patients’ physical function and reducing the duration of their mechanical ventilation and ICU stay, and it is also endorsed as a preventive measure in relation to post-intensive care syndrome [3–5]. Early mobilization interventions implemented in previous studies have included a wide variety of activities, ranging from passive movements performed in bed to walking [6]. However, the physical load and effects differ substantially between in-bed and out-of-bed exercises, with the latter comprising upright activities that require patients to overcome the force of gravity. Out-of-bed exercise is regarded as a crucial component of early rehabilitation, given that it is directly linked to meaningful functional recovery. Nevertheless, the association between the dose of out-of-bed exercise—defined as the duration or amount of time spent performing related activities—and the patient’s functional prognosis at discharge has not yet been sufficiently examined. In addition, extended durations of early mobilization have been suggested to increase the risk of adverse events such as arrhythmias and hypoxemia [7]. As a consequence, the optimal balance between the benefits and the potential risks of the out-of-bed exercise dose in relation to clinical outcomes remains unclear. To address this issue, the present study sought to examine the association between the dose of out-of-bed exercise and functional decline at discharge, as well as the relationship between the dose of out-of-bed exercise and adverse events in the ICU, among critically ill patients who require prolonged ICU management following cardiovascular surgery. Methods Participants This retrospective study was conducted at the Sakakibara Heart Institute, a cardiovascular center located in Fuchū, Japan. It included patients who underwent cardiovascular surgery between May 2018 and December 2022 and who required an ICU stay of three days or longer postoperatively. Both elective and emergency surgeries were included. The exclusion criteria were an ICU stay of two or fewer days, postoperative cerebral infarction or spinal cord infarction, transfer to another hospital within seven days of surgery, in-hospital death, and/or inability to ambulate preoperatively. Patients aged < 18 years were also excluded. The study was conducted in accordance with the requirements of the Declaration of Helsinki and the Ethical Guidelines for Life Sciences and Medical Research Involving Human Subjects, and it was approved by the Ethics Committee of Sakakibara Heart Institute (approval ID: 24–016). This observational study examined the association between different doses of early mobilization during ICU stay among patients requiring prolonged ICU management after cardiovascular surgery, and was conceptually framed as an emulation of a hypothetical target trial. Outcome measures The primary outcome was postoperative functional decline, which was defined as a decrease of at least five points in the Barthel Index (BI) score at discharge when compared with the preoperative (baseline) score. A 5-point decrease in the BI has been considered to represent a clinically meaningful deterioration in functional independence in hospitalized patients [8, 9]. The BI assesses independence in 10 activities of daily living, with total scores ranging from 0 to 100, where higher scores indicate greater functional independence. Each item is scored in increments of 0, 5, 10, or 15 points depending on the level of independence [10]. The preoperative BI was assessed prior to surgery to determine baseline functional status, either through direct evaluation or based on information obtained from the patient or their family when direct assessment was not feasible. The secondary outcomes were adverse events that occurred during mobilization in the ICU, including cardiac arrest; bleeding requiring hemostatic treatment; unplanned extubation; line removal requiring emergency intervention; falling to the floor; systolic blood pressure of < 80 mmHg; lethal arrhythmia; rapid atrial fibrillation, which was defined as a ventricular rate of > 150 bpm; and percutaneous arterial oxygen saturation (SpO₂) of < 80% lasting for more than 3 min [7]. Data collection For each patient, demographic and clinical data were collected from their medical records, physical examinations, physiological measurements, and interviews. Early mobilization was implemented in accordance with the 2012 Guidelines for Rehabilitation in Patients with Cardiovascular Disease [11]. The rehabilitation program for each patient was determined via multidisciplinary conferences involving intensive care physicians and/or cardiovascular surgeons, nurses, physical therapists, and registered dietitians. The intensity level and duration of mobilization were determined through daily multidisciplinary assessment based on patients’ hemodynamic, respiratory status, level of consciousness, and overall clinical condition. Mobilization was initiated and progressed in accordance with guideline-based safety criteria when patients met predefined institutional thresholds, and sessions were temporarily withheld in cases of hemodynamic instability, significant arrhythmia, severe hypoxemia, or other clinical concerns. In the absence of firmly established evidence regarding optimal mobilization dose, these decisions were not governed by a strictly standardized protocol, and variability in clinical judgment may have occurred even among patients with similar clinical profiles. Early mobilization was initiated at the bedside according to each patient’s clinical status and then progressed sequentially to sitting, standing, and walking. When clinically indicated—such as in patients with muscle weakness—bodyweight resistance exercises were also incorporated. All the sessions were delivered by physical therapists with more than five years of clinical experience, six times per week, with the aim of achieving the highest level of activity that could be tolerated by each patient. All the mobilization sessions were performed under continuous electrocardiogram (ECG), blood pressure, and oxygen saturation monitoring, and the attending intensivists were on standby to ensure patient safety. The dose of early mobilization was quantified as the cumulative time and number of days spent performing out-of-bed exercises with an ICU Mobility Scale score (IMS) ≥ four (standing or higher), as averaged per ICU day. The IMS ranges from 0 (lying in bed) to 10 (independent ambulation), with higher scores indicating greater mobility [12]. Detailed descriptions of each level are provided in Table S1. Moreover, one rehabilitation unit corresponded to 20 min of exercise according to the Japanese medical reimbursement system. This approach allowed standardized quantification of mobilization exposure based on both exercise duration and functional intensity during ICU rehabilitation. The duration of IMS ≥ 4 activities was not recorded minute-by-minute within each unit; thus, mobilization dose represents a pragmatic estimate rather than a precise measure of physical loading. Because mobilization dose accumulated over time during ICU stay and was influenced by patients’ evolving clinical status, it should be interpreted as a time-varying exposure rather than a baseline intervention assigned at a single time point. Statistical analysis To address potential confounding due to non-random treatment allocation, propensity score (PS) matching was performed. The continuous variables were summarized as medians (interquartile range) and the categorical variables as numbers (%). The patients were classified into the high-dose group (≥ two units/day) or the low-dose group (< two units/day) based on the average daily mobilization dose administered during their ICU stay. The cutoff of two units per day (approximately 40 min) was selected based on previous studies evaluating the dose of early mobilization and to represent a clinically meaningful threshold of mobilization intensity during ICU rehabilitation [13, 14]. The PS was estimated using clinically relevant baseline variables selected a priori based on clinical relevance and previous literature for their potential influence on mobilization feasibility and postoperative functional outcomes [3, 15], including patients’ age, sex, body mass index (BMI), cerebrovascular disease, chronic kidney disease (CKD), preoperative BI score, Acute Physiology and Chronic Health Evaluation (APACHE) II score, duration of mechanical ventilation, and postoperative days to the initiation of standing. The timing of standing initiation and early invasive mechanical ventilation duration (≤ 48 h) were included as a proxy indicator of early postoperative recovery and rehabilitation feasibility. The discriminative ability of the PS model was assessed by means of the area under the receiver operating characteristic curve (AUC). The patients were matched 1:1 between the high- and low-dose groups using a caliper width of 0.2 of the standard deviation of the logit of the PS. The covariate balance was evaluated using the standardized mean difference (SMD), with SMD values < 0.1 indicating adequate matching. The primary analysis used multivariable logistic regression within the PS-matched cohort to examine the association between mobilization dose and postoperative functional decline, adjusting for age, sex, BMI, cerebrovascular disease, CKD, preoperative BI score, APACHE II score, day of surgery (weekend), duration of mechanical ventilation, postoperative days to the initiation of standing, and use of continuous renal replacement therapy. To mitigate reverse causality and ensure appropriate temporal alignment, a sensitivity analysis was performed using mobilization dose during the first 72 h after ICU admission, adjusted for the same covariates as in the primary analysis. Subgroup analyses were additionally performed in clinically relevant subgroups. In addition, a worst-case sensitivity analysis was performed in the full cohort, in which patients who died during hospitalization or were transferred early were defined as having functional decline, and multivariable logistic regression analyses were conducted. As a complementary analysis, linear regression was performed to examine the association between mobilization dose and the continuous change in BI score. The secondary analysis evaluated the association between the mobilization dose and adverse events using Poisson regression to estimate incidence rate ratios, with the number of mobilization sessions used as the exposure (denominator), thereby reflecting the incidence of adverse events per session in the high- and low-dose groups. In addition, restricted cubic spline regression was also performed to visualize the dose–response relationship between the mobilization dose and patients’ postoperative functional decline and adverse events. In the analyses of adverse events, the number of mobilization sessions was included as an adjustment variable. A two-sided p value < 0.05 was considered statistically significant. All the statistical analyses were performed using IBM SPSS Statistics, version 22 (IBM Corp., Armonk, NY, USA) and R software, version 4.4.1 (R Foundation for Statistical Computing, Vienna, Austria). Results A total of 610 patients were included in the analysis. They had a median age of 73 years (interquartile range 65–79 years), and 38.7% of them were female (Figure S1). The AUC for the PS model was 0.68 (95% confidence interval [CI] 0.63–0.72). After 1:1 matching, 196 patients were included in each study group (Table 1). The patients’ baseline characteristics were well balanced, with SMDs < 0.1 for the preoperative and perioperative clinical variables. Table 1Baseline clinical characteristicsVariableBefore propensity score matchingSMDAfter propensity score matchingSMDEarly mobilizationEarly mobilizationHigh-doseLow-doseHigh-doseLow-dose(n = 393)(n = 217)(n = 196)(n = 196)Age, years73 (67, 79)73 (63, 79)0.1473 (67, 79)74 (63, 79)0.08Female, n (%)153 (38.9)83 (38.2)0.0272 (36.7)74 (37.8)0.02BMI, kg/m223.2 (20.9, 26.0)22.6 (19.9, 25.1)0.1622.6 (20.4, 25.0)22.6 (20.3, 25.0)0.02Comorbidity, n (%) Hypertension258 (65.6)146 (67.3)0.04125 (63.8)132 (67.3)0.07 Diabetes mellitus96 (24.4)50 (23.0)0.0355 (28.1)45 (23.0)0.11 Chronic kidney disease77 (19.6)70 (32.3)0.3062 (31.6)59 (30.1)0.03 Chronic heart failure45 (11.5)26 (12.0)0.0227 (13.8)22 (11.2)0.07Cerebrovascular disease51 (13.0)20 (9.2)0.1126 (13.3)20 (10.2)0.09Pre-Barthel Index, points100 (100, 100)100 (100, 100)0.04100 (100, 100)100 (100, 100) < 0.01Surgery type, n (%)0.190.17 CABG83 (21.1)39 (18.0)43 (21.9)35 (17.9) Valve36 (9.2)27 (12.4)23 (11.7)25 (12.8) Thoracic aortic94 (23.9)47 (21.7)35 (17.9)46 (23.5) Multiple171 (43.5)92 (42.4)88 (44.9)81 (41.3) Other9 (2.3)12 (5.5)7 (3.6)9 (4.6)Emergent, n (%)180 (45.8)89 (41.0)0.1084 (42.9)80 (40.8)0.04Day of surgery (weekend), n (%)42 (10.7)11 (5.1)0.2114 (7.1)10 (5.1)0.09Operation time, min258 (194, 345)264 (195, 340)0.06250 (189, 339)257 (196, 333)0.08Bleeding, ml275 (150, 433)270 (140, 480)0.01255 (135, 430)270 (143, 455)0.01IMV, h24 (18, 37)23 (18, 46)0.2023 (17, 34)23 (18, 41)0.08APACHE II score, points16 (13, 19)15 (12, 18)0.1115 (12, 19)15 (12, 19)0.01SOFA score, points6 (5, 8)7 (5, 9)0.197 (5, 8)7 (5, 9)0.05Modified NUTRIC score, points4 (3, 5)4 (3, 5)0.074 (3, 5)4 (3, 5)0.06IABP, n (%)37 (9.4)27 (12.4)0.1020 (10.2)21 (10.7)0.01NPPV, n (%)65 (16.5)32 (14.7)0.0432 (16.3)29 (14.8)0.04CRRT, n (%)29 (7.4)39 (18.0)0.3222 (11.2)35 (17.9)0.18BMI: body mass index; CABG: coronary artery bypass grafting; IMV: invasive mechanical ventilation; APACHE: Acute Physiology and Chronic Health Evaluation; SOFA: Sequential Organ Failure Assessment; NUTRIC: nutrition risk in critically ill; IABP: intra-aortic balloon pumping; NPPV: non-invasive positive pressure ventilation; CRRT: continuous renal replacement therapy; SMD: standardized mean difference Baseline clinical characteristics BMI: body mass index; CABG: coronary artery bypass grafting; IMV: invasive mechanical ventilation; APACHE: Acute Physiology and Chronic Health Evaluation; SOFA: Sequential Organ Failure Assessment; NUTRIC: nutrition risk in critically ill; IABP: intra-aortic balloon pumping; NPPV: non-invasive positive pressure ventilation; CRRT: continuous renal replacement therapy; SMD: standardized mean difference The exposure characteristics of early mobilization and patients’ postoperative outcomes are summarized in Tables 2 and 3, respectively. Functional decline occurred in 101 patients (25.8%), with a significantly higher incidence being observed in the low-dose group than in the high-dose group (31.6% vs. 19.9%, p = 0.008) (Fig. 1). Multivariable logistic regression analyses within the PS-matched cohort confirmed that low-dose mobilization remained significantly associated with functional decline (odds ratio [OR] 2.03; 95% CI 1.19–3.45; p = 0.009). In a sensitivity analysis using mobilization exposure defined within the first 72 h after ICU admission, low-dose mobilization remained significantly associated with postoperative functional decline (OR 2.65, 95% CI 1.58–4.46, p < 0.001) (Table S2). The subgroup analyses demonstrated consistent results across clinically relevant strata, without significant interactions (Figure S2). Furthermore, in a worst-case sensitivity analysis including all patients, in which in-hospital deaths and early transfers were classified as functional decline, the association remained statistically significant (OR 2.32; 95% CI 1.60–3.35; p < 0.001) (Table S3). As a complementary analysis using BI change as a continuous variable, the low-dose group was significantly associated with a lower BI change score (β − 5.38; 95% CI − 9.14 to − 1.62; p = 0.005). Table 2Exposure characteristics of early mobilizationVariableEarly mobilizationBetween-group difference (95% CI)p valueHigh-doseLow-dose(n = 196)(n = 196)Initiation of mobilization, days In-bed2 (1, 2)2 (1, 3)0 (− 1, 1)0.67 Out-of-bed2 (1, 3)2 (1, 3)1 (0, 1)0.09Out-of-bed mobilization, units/day ICU2.2 (2.0, 2.7)1.5 (1.3, 1.7)− 0.7 (− 0.8, − 0.5) < 0.001 Ward1.8 (1.5, 2.0)1.7 (1.4, 2.0)− 0.1 (− 0.2, 0.0)0.13 Total1.9 (1.7, 2.1)1.6 (1.4, 1.8)− 0.3 (− 0.3, − 0.2) < 0.001ICU: intensive care unit; CI: confidence interval Exposure characteristics of early mobilization ICU: intensive care unit; CI: confidence interval Table 3Postoperative outcomesVariableEarly mobilizationBetween-group difference (95% CI)p valueHigh-doseLow-dose(n = 196)(n = 196)Physical function at ICU discharge MRC sum score, points60 (56, 60)60 (55, 60)0 (− 2, 2)0.97 ICU mobility scale, points8 (8, 8)8 (7, 8)0 (0, 0)0.31Post-Barthel Index, points100 (100, 100)100 (90, 100)0 (0, 0) < 0.01Delirium, n (%)63 (32.1)72 (36.7)− 4.6 (− 14.0, 4.8)0.40ICU stay, days5 (4, 7)7 (5, 9)2 (1, 2) < 0.001Hospital stay, days16 (13, 24)18 (13, 27)2 (− 1, 4)0.11Home discharge, n (%)156 (79.6)139 (70.9)8.7 (0.2, 17.2)0.06Functional decline, n (%)39 (19.9)62 (31.6)− 11.7 (− 20.3, − 3.2) < 0.01ICU: intensive care unit; MRC: Medical Research Council; CI: confidence interval Postoperative outcomes ICU: intensive care unit; MRC: Medical Research Council; CI: confidence interval Incidence and risk of postoperative functional decline by early mobilization dose A total of 33 adverse events were observed in 2774 mobilization sessions (1.2% per session), including 24 episodes of systolic blood pressure < 80 mmHg, 7 cases of rapid atrial fibrillation, and one case each of lethal arrhythmia and bleeding (Table S4). The incidence of these adverse events per session did not differ between the low- and high-dose groups (1.3% vs. 1.1% per session; incidence rate ratio 0.88; 95% CI 0.43–1.74; p = 0.70) (Table 4). The restricted cubic spline analysis demonstrated a nonlinear dose–response relationship, whereby the risks of both postoperative functional decline and adverse events tended to decrease when the mobilization dose exceeded approximately 2 units/day (Fig. 2). Table 4Adverse events during early mobilization in the low- and high-dose groupEarly mobilizationSessions, unitsAdverse events, nIncidence rate, %IRR95% CIp valueLow-dose1421181.31.00High-dose1353151.10.880.43, 1.740.70IRR: incidence rate ratio; CI: confidence interval Adverse events during early mobilization in the low- and high-dose group IRR: incidence rate ratio; CI: confidence interval Association of early mobilization dose with postoperative functional decline (A) and adverse events (B) based on a cubic spline regression model Discussion This study examined the association between the early mobilization dose of out-of-bed exercise and the postoperative functional outcomes among critically ill patients who required prolonged ICU management following cardiovascular surgery. Patients with a lower dose of out-of-bed mobilization exhibited a significantly higher incidence of postoperative functional decline, but no increase in adverse events was observed based on dose-related differences. These findings suggest that an adequate dose of early mobilization may be associated with a lower risk of postoperative functional deterioration without compromising patient safety. Mobilization dose may partly reflect underlying patient severity and the feasibility of rehabilitation rather than representing a purely independent treatment effect. Patients with greater hemodynamic instability or postoperative complications may have had fewer opportunities to participate in out-of-bed mobilization. In addition, variability in clinical judgment and practice may have contributed to differences in mobilization dose even among patients with similar clinical profiles. Previous studies have demonstrated that early mobilization in the ICU improves patients’ physical function and reduces the duration of their mechanical ventilation and ICU stay [16–18]. However, most such studies have focused on the presence or timing of mobilization rather than on the dose of out-of-bed exercise. Hence, the quantitative relationship between the cumulative dose of out-of-bed exercise and patients’ postoperative functional outcomes has not been fully clarified. This study adds novel findings to the literature by demonstrating the nonlinear association between the mobilization dose and patients’ postoperative outcomes using cubic spline modeling and by highlighting the importance of maintaining an adequate out-of-bed exercise dose in relation to postoperative physical function. The implementation of early mobilization in the ICU is influenced by multiple factors, including the patient’s condition, perioperative management, and organizational characteristics such as the care provider’s profession, the presence of a dedicated mobilization team, and the existence of standardized protocols [14, 19]. In particular, the use of mechanical ventilation, mechanical circulatory support, and systemic management in patients at risk of multiple organ failure or bleeding may necessitate bed rest and limit the opportunities for early mobilization. To minimize confounding, this study applied PS matching to balance the preoperative and perioperative characteristics associated with mobilization practices. Furthermore, all the mobilization sessions were delivered by experienced physical therapists under multidisciplinary supervision, ensuring the consistency of the intervention and minimizing the bias. This methodological approach allowed for a more focused evaluation of the association between mobilization dose and outcomes, although residual and unmeasured confounding cannot be excluded. Interestingly, the timing of the initiation of mobilization and the maximum activity level achieved at ICU discharge did not differ between the two groups. This suggests that achieving early out-of-bed mobilization and maintaining an adequate out-of-bed exercise dose may be more strongly associated with postoperative functional decline than the final activity level attained. In other words, the timing of out-of-bed mobilization and the subsequent progression of the exercise dose are likely to have a substantial impact on patients’ functional status at discharge. Furthermore, in patients who undergo highly invasive procedures that render early out-of-bed mobilization challenging, ensuring an alternative mobilization dose via in-bed ergometer exercise or neuromuscular electrical stimulation may represent an effective strategy [20, 21]. This kind of multimodal approach, which aims to achieve out-of-bed mobilization as early as feasible, has important implications for clinical practice. Concerns regarding the safety of intensive early mobilization persist, given that large randomized trials have reported higher rates of adverse events [7]. Out-of-bed exercise carries inherent risks, including risks of unplanned extubation, hemodynamic instability, and falls. However, in this study, an increased mobilization dose did not elevate the risk of adverse events, suggesting that a higher mobilization dose was not associated with an increased risk of adverse events. The majority of observed adverse events were hemodynamic in nature, involving hypotension or tachyarrhythmia, and may be related to perioperative management practices specific to cardiovascular surgery, including catecholamine use, large-volume fluid administration, and circulatory assistance device support [22]. Although these factors do not necessarily preclude early mobilization, they may influence its safe implementation and could have affected both mobilization dose and clinical outcomes, representing potential sources of residual confounding. Therefore, in the context of cardiovascular surgery—where hemodynamic instability is inherently more likely to occur—strict adherence to guideline-based discontinuation criteria and standardized protocols, together with continuous multidisciplinary monitoring, may serve as practical quality indicators for ensuring the safe implementation of high-intensity mobilization. It must be acknowledged that this study has several limitations. First, because mobilization dose was influenced by patients’ evolving clinical status, reverse causality and time-varying confounding cannot be completely excluded. Patients with greater hemodynamic instability or postoperative complications may have had fewer opportunities to participate in higher-dose mobilization, and therefore the observed associations may partly reflect differences in clinical trajectories rather than a purely causal effect. Although mobilization dose was quantified as the average exposure over the entire ICU stay, sensitivity analyses using exposure defined within the first 72 h after ICU admission yielded consistent results. In addition, a directed acyclic graph was constructed to illustrate the presumed causal relationships, including time-varying clinical factors; however, these factors could not be fully adjusted for in the present analysis (Figure S3). Second, inclusion of early invasive mechanical ventilation duration in the propensity score model may have introduced overadjustment because this variable may partly reflect post-exposure processes rather than a pure baseline confounder, although it was included as a proxy for early postoperative recovery and rehabilitation feasibility. Third, as this was a single-center retrospective study, the generalizability of the findings remains limited and external validation is warranted. Fourth, exclusion of patients who died during ICU stay or were transferred early postoperatively may have introduced selection bias by preferentially including relatively stable patients. However, worst-case sensitivity analyses classifying these patients as having functional decline yielded consistent associations, suggesting that survivor bias alone is unlikely to fully explain the findings. Finally, physical activity occurring outside the physical therapist-led mobilization sessions was not captured, and the unit-based definition combined with the IMS threshold (≥ 4) may not have accurately reflected actual physical loading, potentially leading to exposure misclassification. In addition, although propensity score matching was applied, residual confounding due to unmeasured factors cannot be excluded. System-level factors, such as surgery timing, as well as fatigue or psychological distress, may have influenced mobilization dose and patient participation [23, 24]. Therefore, the present findings should be interpreted as exploratory associations rather than definitive evidence of a causal effect. Future studies incorporating comprehensive physical, psychological, and cognitive assessments are needed to optimize individualized early mobilization strategies. Conclusions In patients who require prolonged ICU management following cardiovascular surgery, a lower dose of out-of-bed exercise at a standing or higher level is associated with a greater incidence of postoperative functional decline but has no significant association with adverse events. Therefore, ensuring that patients receive an adequate dose of out-of-bed exercise while subject to careful risk management is important to preserve physical function in this high-risk population. Supplementary Information Supplementary Material 1. Acknowledgements The authors wish to thank the team at the Department of Cardiac Rehabilitation and Intensive Care Unit for their assistance in conducting the research detailed in this manuscript. We also sincerely thank Dr. Shuntaro Sato for his valuable statistical advice and insightful comments on the study design and analysis. Abbreviations Intensive care unit ICU-acquired weakness Barthel index Percutaneous arterial oxygen saturation Propensity score Body mass index Chronic kidney disease Acute Physiology and Chronic Health Evaluation Area under the receiver operating characteristic curve Standardized mean difference Confidence interval Author contributions KH, KK, HU, KHo, TI, AN, and MI conceptualized the study. KH and KK developed the methodology and performed the formal analysis. KH, HU, KHo, and TI contributed to the investigation and data curation. KH and KK contributed to software, validation, and visualization. KK, AN, and MI were responsible for project administration, supervision, funding acquisition, and resources. KH and KK drafted the original manuscript. All authors contributed to writing—review and editing and approved the final manuscript. Funding This work was supported by a grant from the Sakakibara Heart Foundation. Data availability The datasets generated and/or analysed during the current study are not publicly available due to privacy and ethical restrictions, but are available from the corresponding author on reasonable request. Declarations Ethics approval and consent to participate The study was conducted in accordance with the requirements of the Declaration of Helsinki and the Ethical Guidelines for Life Sciences and Medical Research Involving Human Subjects, and it was approved by the Ethics Committee of Sakakibara Heart Institute (approval ID: 24-016). Informed consent was obtained using an opt-out approach, whereby patients were provided the opportunity to decline participation through an ethics committee-approved public disclosure document. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement The datasets generated and/or analysed during the current study are not publicly available due to privacy and ethical restrictions, but are available from the corresponding author on reasonable request.