This study was conducted this single-center retrospective cohort study at a tertiary care university hospital, where approximately 250 adult cardiovascular surgeries are performed annually. Patient data were collected using electronic medical records.

We included patients admitted to the intensive care unit (ICU) under tracheal intubation after adult cardiovascular surgery. The use of cardiopulmonary bypass (CPB) during surgery was not distinguished. We excluded patients with a preoperative elevated diaphragm, cases involving a diaphragmatic incision—such as descending aortic or thoracoabdominal arterial replacement, cases of reoperation, and cases in which were difficult in assessing the diaphragm due to non-cooperation with the examination or inability to perform ultrasound examinations. The inclusion period for the patient registry was July 2013 to December 2014.

We performed spontaneous breathing trial (SBT) when patients had no uncontrolled bleeding; controlled pain; good communication; and stable hemodynamics, fluid balance, and electrolytes. Parameters measured were arterial blood gas analysis in spontaneous breathing, minute volume of ventilation, respiratory rate, tidal volume, rapid shallow breathing index (RSBI, respiratory frequency/tidal volume(L)), forced expiratory pressure, and pressure at which the tracheal tube leaks. Patients were assessed for bilateral diaphragmatic motion and pleural effusion using two-dimensional ultrasonography at the SBT routinely.

Extubation criteria were PaO2/FiO2 > 250 mmHg, PaCO2 35–50 mmHg, respiratory rate < 25 breaths/min, RSBI < 100, minutes volume of ventilation (mL)/preoperative body weight (kg) < 200 mL/kg, tidal volume (mL)/preoperative body weight (kg) > 5 mL/kg, vital capacity (mL)/body weight at preoperative > 10 mL/kg, forced expiratory pressure > 20 cmH2O, and pressure at which the tracheal tube cuff leaks < 20 cmH2O. If some criteria were not achieved, the final decision to extubate was made by experienced intensivists based on findings of SBT and diaphragmatic dysfunction.

Exposure was defined as the presence of diaphragmatic dysfunction. This study assessed the presence of diaphragmatic dysfunction using ultrasound at the time of the first SBT on bilateral sides. The evaluation was performed with the upper body as upright as possible. Scanning was performed by placing the probe between the ribs, which were not disturbed by drains. As far as possible in B-mode, the head–caudal movement was observed at the apex of the diaphragmatic dome.

Diaphragmatic dysfunction were classified as normal, incomplete dysfunction, or complete dysfunction. Diaphragm movements have been reported to be 2.6–30 mm at rest and 16.7–110 mm during deep breathing [4]. We determined whether diaphragm movement was enough according to the criteria of 20 mm at rest and 100 mm during deep breathing. In accordance with the movement of the diaphragm, we have classified three levels of diaphragmatic movement: normal, incomplete, and complete. Incomplete and complete dysfunction were defined as the diaphragmatic dysfunction group and normal was defined as enough movement toward the caudal side both during rest and deep breathing. Incomplete dysfunction was defined by the movement of the diaphragm to the caudal side on inspiration during rest, without sufficient movement on inspiration during deep breathing. Complete dysfunction was defined by the movement of the diaphragm toward the head side during inspiration at rest (paradoxical movement). Complete dysfunction was also defined as the absence of diaphragmatic movement both during rest and deep breathing. In patients who underwent multiple SBTs, the more severe stage of dysfunction was defined as the final diagnosis.

The primary outcome was weaning off in mechanical ventilation. The duration of mechanical ventilation was defined as duration from the date of ICU admission to the date of weaning off in mechanical ventilation. For patients who used noninvasive positive pressure ventilation (NPPV) after extubation, the date of weaning off in NPPV was defined as the date of weaning off in mechanical ventilation. For patients who were reintubated, the date of weaning off in mechanical ventilation was defined as the date of weaning off in mechanical ventilation after reintubation. For patients who had a tracheostomy, the date of weaning off in mechanical ventilation was defined as the date of weaning off in mechanical ventilation after tracheostomy.

The secondary outcomes were reintubation, death from all causes, and improvement of diaphragm position assessed by chest radiographs. We defined reintubation as intubation within 72 h after extubation. Death and improvement of diaphragm position were observed for up to December 2019. To evaluate the long-term prognosis, the post-operative observation period was set at least 5 years. We compared chest radiographs performed before surgery and after extubation, and investigated the date of improvement of the diaphragm position to the preoperative position. Chest radiographs were performed in the supine or sitting position, while the patient was in the ICU, and basically in the standing position after leaving the ICU. Diaphragmatic elevation was defined as the location of the diaphragm within the anterior fourth rib on chest radiograph or elevation of the diaphragmatic dome apex > 1 rib interval (3 cm) compared to the preoperative level. Chest radiographs were evaluated by one physician with 10 year experience. The assessment of chest radiographs was not blinded.

This study was collected participants’ baseline characteristics data, including age, sex, body mass index (BMI), surgical urgency, operative time, blood loss, operative procedure, American society of anesthesiologists—physical status (ASA-PS) classification, EuroSCORE II [5, 6], Charlson comorbidity index (CCI) [7, 8], sequential organ failure assessment (SOFA) score [9] at first SBT, and history of thoracic surgery. The operative methods were differentiated into coronary artery bypass grafting (CABG), valve surgery, aortic surgery, congenital heart disease surgery, and others, and categorized as either single or combined surgery. Surgical urgency was differentiated into four levels: elective, urgent, emergency, and salvage. We set the following as potential confounders; age, sex, BMI, surgical urgency, operation time, blood loss, procedure (aortic surgery or not), and EuroSCORE II.

The results were expressed as mean and standard deviation, median and interquartile range, or count (%). We performed survival time analysis for participants without missing covariates. The event was defined as incidence of weaning off in mechanical ventilation, death from all causes, and improvement of chest radiographs, respectively. The date of ICU admission for each participant was used as the starting point for survival time analysis.

We compared event rates by the presence of diaphragmatic dysfunction. For weaning off in mechanical ventilation and improvement of chest radiograph, we used the cumulative incidence function to calculate event rates. For the death from all causes, we used the Kaplan–Meier methods to calculate event rates.

To estimate risk-adjusted association between diaphragm dysfunction and outcomes, we used Fine-Gray models or Cox proportional hazard models. For nonfatal outcomes (incidence of weaning off in mechanical ventilation and improvement of chest radiographs), Fine-Gray models were used to account for the competing risk of death, and subdistribution hazard ratio (sHR), and 95% confidential intervals (95% CIs) were reported. For death from all causes, Cox proportional hazard models were used, and hazard ratio (HR) and 95% CIs were reported.

We calculated sHR and HR adjusted by potential confounders. Association between diaphragmatic dysfunction and reintubation was analyzed by univariate logistic regression. All analyses were conducted using Stata version 17.0 (StataCorp, Texas, USA). Statistical significance was set at p < 0.05 (two-tailed).