Feasibility and Efficacy of Lung Ultrasound to Diagnose Postoperative Hypoxemia-A Prospective Study

Background Postoperative hypoxemia is associated with morbidity and mortality. We aim to evaluate the feasibility and efficacy of lung ultrasound (LUS) to diagnose pulmonary complications in patients suffering from hypoxemia after general anesthesia, and compare to thoracic computed tomography (CT). Methods Adult patients received general anesthesia and suffered from hypoxemia in the PACU, were analyzed. Hypoxemia was defined as a SPO 2 less than 92% for greater than 30 seconds on room air. LUS was performed by a trained anesthesiologist once hypoxemia occurred. After LUS examination, each patient was transported to radiology department for thoracic CT scan within 1 hour before returning to the ward. Results From January 2019 to May 2019, 113 patients (61 men) undergoing abdominal surgery (45 patients, 39.8%), video-assisted thoracic surgery (31 patients, 27.4%), major orthopedics surgery (17 patients, 15.0%), neurosurgery (10 patients, 8.8%) and other surgery (10 patients, 8.8%) were included. CT diagnosed 327 of 1356 lung zones as atelectasis while LUS revealed atelectasis in 311 of the CT-confirmed zones. Pneumothorax was detected by CT scan in 75 quadrants, 72 of which were detected by LUS. Pleural effusion was diagnosed in 144 zones on CT scan and LUS detected 131 of these zones. LUS was reliable in diagnosing atelectasis (sensitivity 98.0%, specificity 96.7% and diagnostic accuracy 97.2%), pneumothorax (sensitivity 90.0%, specificity 98.9% and diagnostic accuracy 96.7%) and pleural effusion (sensitivity 92.9%, specificity 96.0% and diagnostic accuracy 95.1%). Conclusions Lung ultrasound is feasible, efficient and accurate in diagnosing different etiologies of postoperative hypoxia in the PACU.


Background
Hypoxemia is caused by mismatching of pulmonary ventilation and perfusion, intrapulmonary shunting, reduced functional residual capacity, loss of hypoxic pulmonary vasoconstriction, recumbent body position, injured pulmonary secretion clearance, sedatives and muscle relaxants induced transient loss of respiratory muscle tone, and pain. 1 Hypoxemia occurs frequently in the immediate postoperative recovery course in both pediatric and adult patients. [2][3][4] Prolonged and severe hypoxemia is associated with nausea, vomiting, postoperative cognitive dysfunction, delayed wound healing, surgical site infection, arrhythmias, prolonged hospital stay and death. [5][6][7] Rapid diagnosis and appropriate management must be made by the anesthesiologist once hypoxia occurs postoperatively. Chest x-rays (CRX) had been restricted due to the disadvantage of poor quality 8 . Although thoracic computed tomography (CT) is considered the gold standard to elucidate causes of hypoxia, radiation exposure and the need to transfer unstable patients makes CT a less than ideal tool. Bedside lung ultrasound (LUS) has the advantages on sensitivity, accuracy, nonradiation, non-invasiveness, reproducibility and convenience. It has been validated to diagnose atelectasis, pneumonia, pleural effusion and pneumothorax. [9][10][11][12][13][14] The primary aim of this study is to evaluate the feasibility and efficacy of lung ultrasound to diagnose pulmonary complications in patients suffering from hypoxemia after general anesthesia in the postanesthesia care unit (PACU), and compare lung ultrasound results with thoracic CT.

Patients
The study was approved by the review committee of Second Affiliated Hospital of Zhejiang University (IR2018001133, 2018/12/05) and registered at ClinicalTrials.gov (NCT03802175) before patient enrollment. Informed consents were obtained from all patients. Adult patients who received general anesthesia and suffered hypoxemia in the PACU were included in this study. Postoperative hypoxemia was defined as a decreased oxygen saturation measured by pulse oximetry (SPO 2 ) less than 92% for greater than 30 seconds while on room air 20 minutes after extubation. 15 Exclusion criteria included: covered surgical dressings from opening thoracic or breast surgery preventing ultrasound examination; body mass index (BMI) greater than 40 kg/m 2 ; lack of cooperation due to cognitive dysfunction; residual muscle relaxants resulted in incomplete recovery of muscle strength (Train of four stimulation, TOF < 0.9); respiratory forgetfulness from residual opioid; hemodynamic instability; anemia; significant bleeding, fever or hypothermia. Besides, patients were withdrawn if SPO 2 decline to 85% or less or admission of intensive care unit (ICU) happened.
Continuous intra-venous propofol, remifentanil with inhalational sevoflurane was utilized for anesthesia maintenance after intubation. Supplemental cisatracurium was provided for adequate muscle relaxation when needed. Volume-controlled ventilation with tidal volume of 5-8 mL/kg (5-6 mL/kg for OLV and 6-8 mL/kg for TLV), respiratory rate (RR) of 12-15 breaths/min, FiO 2 of 0.5-0.6 and positive end-expiratory pressure (PEEP) of 5 cm H 2 O was utilized to maintain an end-tidal carbon dioxide pressure (P ET CO 2 ) between 35 and 45 mmHg and a peak airway pressure of less than 30 cmH 2 O (specific parameter was adjusted according to the type of surgery and patient's condition).
Depth of anesthesia monitoring was completed by bispectral index (BIS) with an appropriate value of 40-60. Before closing chest, each patient undergoing VATS received a recruitment maneuver (RM) by forcing sustaining inspiration at the level of 30-40 cm H 2 O airway pressure for 10-20 seconds, then OLV was converted to TLV until extubation. Besides, a chest tube was connected to a watersealed bottle to provide drainage of any leaked air or fluid. Those undergoing non-VATS did not receive RM. All patients were transported to the PACU after operation. Before extubation, the set of mechanical ventilation in the PACU was same with that in the operating room. Extubation was performed when the following criteria were met: VT > 5 mL/kg; and minimal RR of 11 breaths/min; hemodynamic stability (a maximum variation of mean arterial pressure and heart rate was 20% around the baseline value); normothermia. Neostigmine (0.02mg/kg) was used for reversal of neuromuscular blocking before extubation. After extubation, the patient inhaled oxygen through a face mask at 3-6L/min for about 15 minutes then the face masks were removed. During the next time, patients were supplemented with oxygen again through masks as temporary treatment if the SPO 2 declined to less than 92%.

Lung Ultrasound Examination
With a 2 to 5 MHz convex probe in an ultrasound device (Mindray, Guangdong, China), LUS imaging was performed by two trained anesthesiologists (Chen X, Kai S, both with more than 1 year of ultrasound learning) once hypoxemia occurred. The anterior and posterior axillary lines divided each hemithorax into three regions (anterior, lateral and posterior), each region was further divided into two quadrants (superior and inferior) ( Figure 1). The anesthesiologist performed LUS examination from the left lung to the right in the above order. Atelectasis was diagnosed as a tissue-like pattern or hypoechoic juxta-pleural consolidations with hyperechoic static air bronchograms. 10 A juxta-pleural consolidations or tissue-like structure may also indicate pneumonia. However, the visualization of dynamic air-bronchogram helps exclude atelectasis. 16 With a negative predictive value of 100%, presence of lung sliding excluded the diagnosis of pneumothorax. 17 Meanwhile the diagnosis of pneumothorax should combine with the lung point, barcode sign on M mode and absence of lung sliding. 13,[18][19][20] On this basis, the absence of pleural sliding in the anterior, lateral or posterior chest on LUS was defined as small, medium or large size of pneumothorax. 21 Presence of anechoic area fluctuating with respiration identified pleural effusion. 22 Examination of pleural effusion was performed with the patient in the semi-recumbent position. A large pleural effusion was diagnosed when the maximal interpleural distance was more than 25 mm on ultrasonography and effusion must be visible on at least three intercostal spaces. Less than 15 mm of maximal interpleural distance was defined as small effusion. 23 Combined with symptoms such as dyspnea, a minimum of 3 B-lines in at least two anterior or lateral quadrants in each thorax may benefit for the consideration of pulmonary edema. 24 LUS scores (0-36, calculated by adding up all the 12 individual quadrant scores) assess aeration changes and a higher grade represents more serious aeration loss but inapplicable for pneumothorax ( Figure 2). [25][26][27] Score 0, healthy lung, equidistant A-lines parallel to the sliding pleura; score 1, moderate aeration loss, no fewer than 3 dispersive B lines originated from the pleural; score 2, serious aeration loss, presence of coalescent B lines with irregular pleural; scoring 3, absolute aeration loss, subpleural consolidation. The stored video of the worst irregularity was analyzed off-line by Chen X and Kai S. In case of disagreement, a third anesthesiologist (Lina Y, with 5 years of ultrasound learning) reviewed the uncertain images and made the final diagnosis.

Computed Tomography Scan
After LUS examination, every patient with stable hemodynamic and spontaneous respiration was transported to radiology department by a nurse anesthetist for thoracic CT scan within 1 hour after LUS examination. During transport, all patients received oxygen through face masks. Scanning from apex to diaphragm with the patient in supine position, the examination was performed with a 128slice spiral CT device (Siemens, Amberg, Germany). With a window width of 1500 Hounsfield Units and a section thickness of 0.5 mm, all CT sections were stored for reconstruction and computerized analysis. Blinded to our study, a trained radiologist reported the CT findings by judging negative (-) or positive (+) for absence or presence of consolidation, effusion or pneumothorax in the same anatomic quadrant.

Data Collection
Demographic data including gender, age, height, weight, American Society of Anesthetist (ASA) score, BMI, vital signs and smoking habit were recorded. Medical history, pulmonary function test and physical examinations were extracted from the Electronic Medical Record. At bedside, we collected surgical information, duration of mechanical ventilation and PACU stay, time needed for LUS examination and time needed for CT scan (transportation plus CT scan plus oral report). Cumulative opioid dose (calculated by duration and weight), volume of fluid administration (sum of crystalloid and colloid) as well as blood products, arterial blood gas at the end of operation including hemoglobin, arterial partial pressure of oxygen (PaO 2 ), arterial partial pressure of carbon dioxide (PaCO 2 ) were also recorded.

Statistical Analysis
PASS software (version16.0) was used to calculate the sample size. Estimated the sensitivity and specificity of LUS are based on the previous study (sensitivity 87.7%, specificity 92.1%) 28 , assumed the allowable error is 10% and α error of 0.05 (bilateral). The calculated sample size for sensitivity and specificity was 50 cases and 38 cases, respectively. Considering the same sample size was adopted for both LUS examination and CT scan, 100 cases were taken from each group of 50 cases.
The total sample size was 110 cases when combined with a dropout rate of 10% at last. A total of 110 patients were needed with previous study and following assumptions: an α error of 0.05, a β value equal to 0.15 and a dropout rate of 10%. With after testing normality distribution, mean ± standard deviation or median (interquartile range) were used to describe continuous variables and comparison of them were performed with a paired-t test or Mann-Whitney U-test as appropriate. Categorical variables were expressed as frequency and percentage, and compared with Chi-squared test or

Results
From January to May 2019, 138 adult patients were evaluated for eligibility. Twenty-five patients were excluded and 113 patients were ultimately enrolled (Figure 3). During the study, all the LUS examinations and CT scans were performed successfully and a total of 1356 pairs of ultrasound cineloops and CT images were stored for all patients. Table 1  One patient was diagnosed with diffuse interstitial syndrome due to multiple B-lines in all the 12 lung quadrants and CT scan made the same conclusion. Both LUS examination and CT scan showed no abnormalities in 12 patients.
The time needed for LUS examination was significantly shorter than CT scan (10.8 ± 1.8 minutes versus 26.8 ± 4.2 minutes, P < 0.001). Kappa for agreement between the first two observers of atelectasis, pneumothorax and pleural effusion respectively was 0.951 (P < 0.001), 0.858 (P < 0.001) and 0.964 (P < 0.001). To solve the disagreement, the third reviewer mainly devoted to evaluate the diagnosis of pneumothorax. Table 2  Among the data we collected, post hoc analyses revealed no correlative factor which influences LUS scores significantly (Table 3). Postoperative typical LUS and corresponding thoracic CT images of atelectasis, pneumothorax, pleural effusion were displayed in Figure 4.

Discussion
Our study showed high accuracy of LUS in diagnosing pulmonary complications such as atelectasis, pneumothorax and pleural effusion, with a high degree of sensitivity and specificity. Consistent with previous publications in both children and adults, 29,30 bedside LUS, is reliable, portable, radiationless and fast in investigating pulmonary pathologic abnormalities. Previous publications on LUS were mostly from emergency departments and ICU, to our knowledge, this is the first study to advocate the application of LUS to investigate hypoxia in PACU. In addition, our study population included various types of surgery and patients with COPD or cardiovascular symptoms were not excluded. This may better reflect the real world experience. Since postoperative thoracic CT is not routinely in clinical practice, LUS in the PACU may help differentiate unexpected respiratory pathologies. Probably, our study could provide clinical significance for timely and appropriate treatment of postoperative hypoxemia in future.
Hypoxemia is primary triggered by atelectasis from compression, gas absorption and loss -ofsurfactant. 31 Postoperative atelectasis was associated with pneumonia and could result in delayed discharge. 32 Early detection and treatment of atelectasis was essential for improving prognosis. Due to the advantages like simple, convenient, time-consuming and non-radiation, LUS can be repeated at the bedside. It has been confirmed the sensitivity and specificity of the diagnosis of atelectasis by lung pulse in ultrasound were 93% and 100%, respectively. 33,34 When compared with Magnetic Resonance Imaging (MRI), LUS showed a sensitivity of 88%, specificity of 89% and accuracy of 88% in diagnosing pulmonary atelectasis. 10 LUS demonstrated excellent diagnostic accuracy (97.2%) in our study, higher than reported (90.7%) by Yu X et al. 28 In Yu's study, only patients undergoing elective intracranial surgery and without pre-operative pulmonary comorbidities were enrolled, whereas our study included a heterogeneous patient population for diversity. To eliminate the interference of adipose layer in the ultrasonic image, obese patients (BMI > 40 kg/m 2 ) were excluded. Considering the safety of patients transferring to CT scan, those with hemodynamic instability were also exclude.
Though hypoxemia were more likely occurred in these patients, but the whole study only excluded 3 relevant patients (Fig. 3) and it exert almost no effect on the result. The incidence of atelectasis 72.6% in our study was lower than previous reported 90% 35 which partly was due to routine recruitment maneuvers at the end of VATS group. Though lung-protective strategies such as low TV, a lower FiO 2 , higher RR RM and PEEP has been reported to decreased postoperative respiratory complications significantly 36-38 and applied in our anesthesia protocol. However, atelectasis is still occurred frequently in our study. PEEP has been reported as a successful method for improving oxygenation and respiratory function during general anesthesia but the optimal level is still inconclusive. 39-41 Though a PEEP of 5 cmH 2 O in our study has referred to previous study, but a higher PEEP may be much more beneficial for reducing atelectasis formation as it had been recommend by some researchers. 42 RM combined with PEEP were also beneficial for reducing atelec- Second, the presence of consolidation on LUS alone was insufficient for diagnose pneumonia 60 . In recent study by Zhou et al., 61 the combination of LUS and procalcitonin had a better diagnostic value for pneumonia. Timely diagnosis of suspected aspiration pneumonia by LUS intraoperatively may beneficial for patients but still need more researches in our future work. Last, patients undergoing cardiac surgery were not included whose atelectasis and hypoxemia were more prominent after cardiopulmonary bypass than other surgery.

Conclusions
In conclusion, we showed that application of LUS to diagnose etiologies of hypoxemia in the PACU is feasible and prompt. LUS was sensitive and specific to diagnose pulmonary complications when compared to thoracic CT scan.

Consent for publication: Not applicable
Availability of data and materials: The datasets generated and/or analysed during the current study are not publicly available due to the manuscript has not been received yet but are available from the corresponding author on reasonable request. Data were described as mean ± standard deviation or median and inter-quartile range as appropriate.
LUS score was described in patients without pneumothorax(N=85).
Abbreviations: SD, standard deviation; IQR, inter-quartile range; M, male; F, female; BMI, Body Mass Index; ASA, American Society of Anesthesiologists classification; SPO 2 , oxygen saturation measured by pulse oximetry; PACU, postanesthesia care unit; LUS, lung ultrasound Table2. Agreement between LUS and CT of pulmonary complications for accumulated quadrants