Precise identification of respiratory system air leakage via dynamic ventilator waveform monitoring: a controlled trial to improve mechanical ventilation care efficiency Authors: Hong-lei Wu, Mei-juan Lan, Wang-Qin Shen, Li Chen, Jia-Ying Qian, Yan-hong Bian, Jia-hai Shi Published: 2025 ABSTRACT AbstractBackgroundMechanical ventilation is a crucial intervention for respiratory support in intensive care unit (ICU) patients. However, air leakage in the ventilator system often compromises patient outcomes. Conventional leakage detection methods are limited by low sensitivity and delayed response, necessitating more precise approaches. This study focuses on the systematic implementation of a nurse-led waveform analysis protocol for respiratory air leakage detection, optimized for mechanically ventilated patients in cardiothoracic ICUs.MethodsThis comparative study was conducted in the cardiothoracic intensive care unit of a hospital in Nantong, China, from June 2019 to May 2020. Three hundred and two patients requiring mechanical ventilation were categorized into the control group or experimental group depending on the time of admission: empirical leakage detection methods were used in the control group and a ventilator waveform-based detection method in the experimental group. The primary outcome was time to detect leakage in the ventilator pipeline, while secondary outcomes included time to detect airway leakage and nurse satisfaction with the detection methods.ResultsPipeline air leakage was detected within significantly shorter durations in the experimental group than in the control group (27.25 ± 18.42 vs. 7.21 ± 4.39 min; t = 13.86; P < 0.0001), even at a minimum cuff pressure of 30–50 cmH₂O (experimental group t = 17.48; P < 0.0001). Additionally, nurses in the experimental group expressed greater satisfaction with the leakage detection method (χ2 = 20.50, P < 0.001).ConclusionThe use of ventilator waveforms for detecting ventilator system leakage is more efficient and accurate than conventional empirical approaches. This innovative technique reduces the time to identify leaks while also enhancing nurse satisfaction, thereby potentially improving patient outcomes in ICU settings.Graphical Abstract Keywords: Ventilator waveform, Respiratory system leakage, Nursing technique, Clinical effect, Mechanical ventilation Introduction Mechanical ventilation is a core therapeutic intervention that directly influences the prognosis of intensive care unit (ICU) patients requiring respiratory support [1–3]. However, air leakage in the ventilator system is a critical challenge encountered in clinical practice that requires urgent solutions [4]. In the ICU, patients undergoing mechanical ventilation often develop complications due to undetected leaks within the ventilator circuit. Failure to promptly identify these leaks can substantially increase the patient’s risk of mortality [5, 6]. This reflects the high prevalence of air leakage in clinical practice and the ensuing threat to patient health outcomes. The etiology of air leakage is multifactorial and complex [7, 8]. The most common trigger is circuit disconnection at connection sites due to routine nursing procedures or patient repositioning [9, 10]. Another critical risk factor is suboptimal cuff management. Although clinical guidelines universally recommend maintaining cuff pressure at 20–30 cmH2O [11, 12], existing literature indicates that leakage may occur even when this pressure target is achieved [13]. Our studies based on clinical observations have shown that persistent leakage due to anatomical discrepancies between cuff dimensions and individual airway structures occurs in 19.37% of mechanically ventilated patients [14–16]. Yet another, albeit relatively rare, cause of ventilator system leakage, is broncho-pleural fistula, which has an insidious onset and rapid progression and can cause irreversible harm if not promptly addressed [17]. Current methods for the detection of leakage have several limitations: they are low in sensitivity, cause a delay in response, do not allow for real-time monitoring, and make rapid intervention difficult [18]. To overcome these drawbacks, we sought to develop a more precise and efficient leakage identification and management protocol by focusing on dynamic analysis of ventilator tidal volume–time waveforms. With this innovative approach, we aim to enhance the safety of mechanical ventilation therapy and reduce complication risks, thereby establishing a novel pathway for optimizing critical respiratory care. Methods In this comparative study, we investigated patients who were received mechanical ventilation at the cardiothoracic ICU of a Nantong, China hospital between June 2019 and May 2020. The study included patients who (1) were at least 18 years old; (2) required mechanical ventilation; (3) provided informed consent; (4) had up-to-date anthropometric data; and (5) showed stable vital signs on monitoring. Patients with significant pleural effusion or unstable vital signs were excluded from the study. The ventilator settings for the patients was determined by the physician according to their weight and clinical profile. Four nurses and doctors comprising an assessment and supervision team from the cardiothoracic surgery ICU performed all the interventions and recorded the basic information of the patients. All team members had over five years of experience in the cardiothoracic surgery ICU and had received training regarding the minimum leak technique (MLT), ventilator system leakage detection procedure, the detection time for ventilator system leakage, and degrees of satisfaction with methods for ventilator leakage detection. The electronic medical records of the patients were reviewed to extract data on several anthropometric variables: age (years), gender, height (m), weight (kg), and body mass index (BMI, kg/m2). Additionally, data on several ventilation parameters, including the peak inspiratory pressure (PIP) and tidal volume, were obtained from the ventilator display. The AVEA ventilator system (Vyaire Medical, Mettawa, IL, USA) was used for all the patients enrolled in this study. The mode of mechanical ventilation for each patient was set according to the discretion of the attending physician. Other equipment used were as follows: reinforced ETTs (Guangzhou Weili Medical Equipment Co., Ltd., Guangzhou, China, and Jiangxi Galanz Medical Equipment Co., Ltd., Jiangxi, China), cuff pressure measurement devices (Hangzhou Ranran Trade Co., Ltd., Hangzhou, China), stethoscopes (Beijing Hausheng Technology and Trade Co., Ltd., Beijing, China), and extendable tubes with a 1.5-m suction loop (Intesec Medical Devices Co., Ltd., Changzhou, China). All equipment was used in accordance with the manufacturers’ recommendations. Within one hour of the patient’s admission, the assessment and supervision team used a sputum suction tube to remove the patient’s nasal and oral secretions with negative pressure. The minimum cuff pressure was detected in a single attempt using the minimum leak technique (MLT), with all steps completed within 30 s. Then, the minimum cuff pressure was synchronously determined using the ventilator and stethoscope and the success/failure of achieving the minimum cuff pressure was recorded. As per the MLT, the endotracheal cuff is completely deflated, which allows the nurses to detect any audible sound of leakage and gradually reinflate the cuff until the leakage stops. This study focuses on two particularly critical and common types of leaks in clinical practice: circuit disconnection, which is triggered by the loosening of a specific connection point on the ventilator circuit, and airway leakage, which is confined to the endotracheal tube cuff. By limiting its scope to these two specific scenarios, this study uses a controlled framework for comparison between the empirical detection method and the novel waveform-based method, while minimizing interference from confounding variables, thereby enhancing the internal validity and reliability of the conclusions. For the control group, which included patients admitted between June and November 2019, the ventilator system leakage was detected via an empirical protocol guided by the Guidelines for Graded Nursing in Hospitals. All patients in the control group received 24 h of uninterrupted dedicated nursing care, which included continuous monitoring for changes in airway conditions or vital signs [19]. In addition, nurses were required to conduct a bedside round every 30 min and measure cuff pressure once every six hours [20]. Consequently, the detection of a leak in this group was a passive process, dependent on the routine checks according to the standard clinical assessment and triggered by the onset of clear clinical signs like inadequate ventilation, rather than by a specific, real-time protocol that allowed for an immediate, system-generated alert. In the experimental group, which included patients admitted between November 2019 and May 2020, a novel leakage detection system based on ventilator waveforms was employed, with patient rounds conducted every half hour, as per standard care. This novel technology is based solely on the systematic analysis and feature extraction of ventilator tidal volume-time waveforms, which allows the accurate detection of respiratory circuit leaks. The specific operational procedure for this detection method is detailed in Fig. 2. A detection operation process chart for resolving respiratory leakage problems in the experimental group Detection time for air leakage in the ventilator pipeline: To ensure standardized and reproducible assessment while simulating the unpredictability of leak locations in clinical practice, we created leaks at randomized components on the ventilator circuit (corresponding to the 1–9 o’clock positions on the flowchart). This randomized selection was designed to maintain the balance of the experiment and to cover potential common leak sources in the circuit. According to the MLT protocol, a standard procedure was followed for the documentation of minimal cuff pressure: First, a member of the evaluation and supervision team inflated the cuff for each patient, ensuring that the minimal effective airway seal is attained and confirming that circuit airtightness meets the required standards. After the nurse completed the prescribed observation period (approximately one minute), the evaluation and supervision team randomly selected a connection point in each of the two patient groups and created a controlled leak of 50–80 mL. Once the leak was initiated, video monitoring was activated simultaneously to record the total time required for the nurse responsible to identify the leak source and reestablish circuit seal. Throughout the process, the patient’s vital signs were maintained within stable limits to guarantee the safety and efficacy of the intervention. Detection time for air leakage in the airway on the premise that the patient has no ventilator pipeline leakage or broncho-pleural fistula: Days 1–3 after the operation, a member of the dedicated team adjusted the patients’ tracheal tube cuff pressure to 30–35 cmH2O and used a stethoscope to detect airway leakage. During monitoring, patients’ vital signs remained within the normal range. Time to detection of the leakage in the airway was from the onset of leakage in the airway to accurate leak localization, with the video surveillance tracking the assigned nurse’s actions until detection. The time to detection of the leakage was recorded as the primary outcome. Nurse Satisfaction: A questionnaire developed inhouse was used to investigate nurses’ satisfaction with the two methods for detecting ventilator system air leakage. The survey mainly covered three domains, namely the convenience, accuracy, and safety of the detection methods. The Likert 5-point scale was used to score the responses, with scores of 1 to 5 representing “very dissatisfied,” “dissatisfied,” “average,” “satisfied” and “very satisfied,” in that order [21]. The questionnaire was filled out by the nurses after completing their duties, and the effectiveness of the questionnaire was checked on the spot by the assessment and supervision team to ensure the quality of the responses. The study protocol of the present study was approved by the Institutional Review Board of Nantong University Affiliated Hospital (Ethics Committee 2015–108), and the study was registered in the Chinese Clinical Trial Registry (ChiCTR-COC-15006459). Informed consent was obtained from all patients prior to the initiation of any interventions. Statistical analysis All statistical analyses were performed using SPSS version 20 (IBM Corp., Armonk, NY, USA) and Prism 8 (GraphPad, San Diego, CA, USA). Data on the age, weight, height, BMI, PIP, and tidal volume were expressed as mean ± standard deviation, and intergroup comparisons were made using independent sample t-test. The gender, operation, and degrees of satisfaction were presented in counts, and intergroup comparisons were made using chi-square (X2) test. A p-value of < 0.05 was considered statistically significant. Results Characteristics of the study population Among the 305 patients assessed for eligibility, 3 were excluded from the analysis because they either regained consciousness from anesthesia or had missing or incomplete data. Thus, 302 patients were included in the final analysis. The two groups were separated temporally. The empirical method of detecting ventilator system leakage employed between June 2019 and November 2019 was used for the 127 patients. The ventilator waveform-based detection of ventilator system leakage, which was used for the period between November 2019 and May 2020, was employed in the case of 175 patients who received mechanical ventilation (Fig. 1). The two groups did not show any notable distinctions in the fundamental attributes of patients (Table 1). Flowchart for the study Table 1Baseline characteristics with control and experimental groups in mechanical ventilationParameterControl group(n = 127)Experimental group(n = 175)x2/tPGender (M/F)88/39115/600.4270.513Age (year)60.69 ± 12.6262.00 ± 12.730.8840.377Weight (Kg)64.74 ± 10.0965.29 ± 11.660.4280.669Height (cm)165.2 ± 7.16165.8 ± 8.260.6230.534BMI23.68 ± 3.1423.69 ± 3.480.0220.983PIP (cmH2O)19.90 ± 3.7820.31 ± 3.870.9320.352Tidal volume(ml)586.9 ± 79.22588.1 ± 98.90.1140.910M Male, F Female, BMI Body Mass Index, PIP Peak inspiratory pressure Baseline characteristics with control and experimental groups in mechanical ventilation M Male, F Female, BMI Body Mass Index, PIP Peak inspiratory pressure The demographic characteristics of the 14 nursing staff enrolled in this study, reflect a cohort of highly experienced, highly educated, and predominantly female mid-to-senior level nursing professionals (Table 2). Table 2Demographic and professional characteristics of nursing staffParameter(n = 14)Gender (M/F)2/12Levels of nursing staff N34 N29 N11Professional Titles Charge Nurse Practitioner10 Nurse Practitioner4Working years > 20Y1 15-20Y8 < 15Y5Educational attainment Bachelor’s degree13 Associate degree1 Demographic and professional characteristics of nursing staff Primary outcomes Ventilator system leakage detection procedure Figure 2 delineates the systematic protocol applied to the experimental group for the detection of the ventilator system leakage and its management. This algorithmic approach commences with circuit inspection and advances to cuff management and differential diagnosis, guided by waveform analysis and auscultatory findings. Time to detection of air leakage in the ventilator pipeline An overview of the outcome parameters is presented in Fig. 3. When compared with the control group, the experimental group had a significantly shorter time to the detection of air leakage in the ventilator pipeline. (27.25 ± 18.42 vs. 7.21 ± 4.39 min, t = 13.86, P < 0.0001). Comparison of the time to detection of air leakage in the ventilator pipeline in the control group and the experimental group Secondary outcomes: time to detection of air leakage in the airway Time to detect airway leakage in the airway at a cuff pressure of 30–35 cmH2O was recorded for 28 patients in the control group and 14 patients in the experimental group (Fig. 4A). A Scatter plot showing the time to detection of airway leakage (in minutes) when the airway is the minimum cuff pressure (MCP) for the control group and the experimental group (B) Bar chart comparing the time to detection of airway leakage (in minutes) between the control and experimental groups at different ranges of MCP For a MCP of 30–50 cmH2O, the time to detect airway leakage in the experimental group was significantly shorter than in the control group (t = 17.48, P < 0.0001). However, when the MCP was greater than 50 cmH2O, the two groups did not show any significant difference in the airway leakage (t = 1.26, P = 0.223; Fig. 4B). Tertiary outcomes Degrees of nurse satisfaction with method of ventilator leakage detection. The overview of the tertiary outcome parameters is presented in Table 3. Nurses expressed greater satisfaction with leakage detection method used for the experimental group than that used for the control group (x2 = 20.50, P < 0.001). Table 3Number of nurses reporting different degrees of satisfaction with the ventilator leakage detection methodsDegrees of satisfactionControl group (n = 14)Experimental group (n = 14)x2P19020.50P < 0.001253305404502 Number of nurses reporting different degrees of satisfaction with the ventilator leakage detection methods Discussion Our findings in this study reveal that our novel technology that is based on systematic analysis and feature extraction of ventilator tidal volume–time waveforms enables the accurate detection of respiratory circuit leaks through waveform features. This detection method efficiently identifies leak points, enhances the safety of mechanical ventilation, and reduces the workload of nurses, thereby decreasing the consumption of national healthcare resource. In terms of clinical practice, this detection method effectively helps nursing staff quickly identify and troubleshoot respiratory circuit leaks in mechanical ventilation patients, thus improving clinical care quality. Reducing ventilator pipeline leakage detection time: The study found that the experimental group significantly outperformed the control group in terms of the detection of air leakage in the ventilator pipeline. This indicates that the new method can significantly shorten leakage detection time and boost nursing efficiency. The clinical significance lies in enhancing the nurse’s ability to promptly identify and address leakage, ensuring effective ventilation support for patients and reducing possible risks like inadequate ventilation and poor oxygenation. This improves treatment safety and patient outcomes. Additionally, reducing the time to detection of airway leakage lightens nurses’ workload, allowing them more time to care for other patients and enhancing overall care quality [22]. The ventilator waveform feature analysis method used in the experimental group has several advantages over the traditional empirical detection methods. Waveform feature analysis offers higher accuracy and reliability by systematically analyzing tidal volume–time waveforms to identify leakage characteristics and locate leakage points [23]. The standardized process also minimizes human interference, ensuring more consistent and reproducible results [24, 25]. Reducing the time to detection of airway leakage: Airway leakage at cuff pressure of 30–35 cmH₂O was recorded for 42 patients. When the MCP was 30–50 cmH₂O, airway leakage was detected within a significantly shorter duration in the experimental group than in the control group. However, when the MCP was above 50 cmH₂O, there was no significant difference in the efficiency of leakage detection in the two groups. This shows that the waveform analysis method allows for a more timely detection than traditional methods at low leakage levels. It also identifies a specific MCP threshold range where the advantages of the new technology are maximized. The lack of significant intergroup difference in leakage detected at MCP of over 50 cm H₂O, might be attributed to the fact that at 30–35 cmH₂O, a large gas escape via the mouth is visible. The cuff’s coverage is much smaller than the patient’s airway cross-sectional area, leading to a distinct sound of gas escape that nurses can easily hear. Enhancing job satisfaction among nursing staff: The nursing staff expressed greater satisfaction with the air leakage detection system used for the experimental group than that used for the control group. This finding can serve as an important guide to optimize clinical nursing practice. The superior accuracy [26], stability, and timeliness of leak identification offered by the waveform analysis method significantly reduce the time cost and operational burden for nurses during leak troubleshooting, thereby effectively improving work experience quality and professional identity. Second, since the new system is based on quantitative analytical models, it substantially reduces the variability in the subjective judgment of nurses, unlike the traditional method. The nursing staff can make scientific decisions using visualized data based on objective quantitative indicators derived from waveform parameters. This evidence-based model not only enhances the credibility of clinical assessment but also tangibly elevates the perception of professional competence, thereby reinforcing job fulfillment [27]. Notably, improved leak detection efficacy benefits patient care quality through timely identification and precise management of ventilation abnormalities. Thus, nursing staff can ensure that patients continuously receive mechanical ventilation support in line with the therapeutic requirements. This closed-loop management mechanism improves clinical prognosis indicators while fostering an efficient and safe care ecosystem [28, 29]. This leads to a virtuous professional incentive cycle where nursing staff deliver high-quality care through enhanced professional capabilities and receive positive occupational feedback via improved patient outcomes [30, 31]. However, we must interpret this finding with caution. Given the non-blinded design of the study, the highly polarized satisfaction scores obtained in our study may have been significantly influenced by expectation bias. The positive feedback from the experimental group nurses, as users of an innovative technology, may be confounded by a preference for novelty, the sense of honor from participating in the study, or the sense of accomplishment from learning new skills. Therefore, we consider this result an exploratory indication rather than a conclusive finding. Representative successful case: In June 2019, a mechanically ventilated patient admitted to the cardiothoracic surgery ICU at our hospital presented with baseline vital signs of heart rate at 85 bpm, blood pressure of 132/77 mmHg, and SpO₂ of 100%. Twenty minutes after positioning the patient supine and completing the skin care handover, the patient developed sudden fluctuations in vital signs: heart rate increased to 130 bpm, blood pressure rose to 170/89 mmHg, and SpO2 decreased to 90%, accompanied by tachypnea and profuse sweating. Physical examination by the attending physician revealed no obvious abnormalities, and endotracheal intubation was performed. Clinical tests showed that the ventilator’s tidal volume–time waveform failed to return to baseline. In accordance with the preset detection protocols, a loosened water collection cup in the ventilator circuit (causing insufficient tidal volume) was identified within 30 s. After prompt intervention, the patient’s vital signs were restored to stable levels within 10 min. The study has a few limitations. First, the patients were recruited specifically from the cardiothoracic surgery ICU of a single center, which may affect the generalizability of the findings. Second, the use of a historical control group rather than a randomized or blinded design raises the possibility of subjective bias, particularly in the assessment of nurse satisfaction. Third, our focus was on short-term monitoring indicators, excluding factors associated with long-term prognosis and technical risks. Fourth, the study was conducted on a single brand of ventilators, which may limit the clinical applicability of our findings. Fifth, the “time to detection” included waiting time, which may not precisely evaluate the diagnostic efficiency of the waveform analysis itself. An important drawback is the protocol bias in the measurement of intervention time. While leakage detection in the control group relied on fixed-frequency bedside rounds and cuff pressure monitoring, real-time waveform analysis was used in the experimental group. This reflects as a discrepancy in the starting point of time measurement: a pre-scheduled observation point in the control group and a clinically triggered event (such as a standard ventilator low tidal volume alarm) in the experimental group. This discrepancy may have led to a falsely prolonged detection time in the control group, thereby overestimating the advantages of the experimental group. To ensure a fair comparison in future research, the intervention time in the control group should be measured from a relevant clinical event rather than a fixed observation round. Conclusions This study provides important, albeit preliminary, insights into an advanced waveform feature analysis method that can effectively reduce the time spent on ventilator pipeline leakage assessment compared to the conventional, low-frequency monitoring protocol. While this technology affords significant enhancement in process efficiency, it is crucial to acknowledge that our study was limited to short-term, process-based metrics and did not evaluate the impact on hard clinical outcomes, such as the rate of ventilator-associated pneumonia (VAP), duration of mechanical ventilation, or ICU length of stay. Therefore, while the implementation of this technology shows promise in strengthening the efficacy of nursing management for mechanically ventilated patients and potentially improving the safety of clinical treatment, we must be cautious not to overstate its proven benefits. The current evidence shows its value in optimizing specific intensive care procedures, but definitive conclusions about improved patient prognosis await validation from future research. Future studies with more rigorous designs, such as randomized controlled trials, and incorporate long-term clinical endpoints may help fully establish the technology’s impact on patient care and safety. Acknowledgements We thank Medjaden Inc. for scientific editing of this manuscript. Authors’ contributions HLW: conceptualization, methodology, validation, formal analysis, investigation, resources, writing of the original draft, and project administration. MJL:conceptualization, methodology, formal analysis, writing review and editing. WQS: methodology, resources, editing, and project administration. LC:funding acquisition and project administration. YHB: methodology, operation and Implementation, data Collection, data curation and funding acquisition. JYQ: methodology, operation and Implementation, data Collection and data curation. JHS: supervision, review and editing, correspondence. Funding The study received funding from the Nantong University Affiliated Hospital Nursing Special Project (Grant: Tfh 2415), National Health Commission Hospital Management Research Institute (Grant: 2025B2YH006), Jiangsu Provincial Hospital Association's Modern Hospital Management Research Center (Grant: JSYGY-3–2025-465), Self-Selected Clinical Research Project (LCYJ20252096) and Jiangsu Provincial Research Hospital (Grant: YJXYY202204-YSB19). Data availability No datasets were generated or analysed during the current study. Declarations Ethics approval and consent to participate The ethical protocol of the present study was approved by the Institutional Review Board of Nantong University Affiliated Hospital (Ethics Committee 2015–108), and the study was registered in the Chinese Clinical Trial Registry (ChiCTR-COC-15006459) with the date of May 29, 2015. An informed consent was obtained from all patients prior to the initiation of all interventions. 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. Hong-lei Wu and Mei-juan Lan contributed equally to this work. References Associated Data Data Availability Statement No datasets were generated or analysed during the current study.