Study design, approval and registration

The planned study is a parallel-group, double-blinded, randomised controlled, single-centred trial with 1:1 allocation ratio. The study has been approved by the institutional review board (IRB) at Xijing Hospital (KY20232058-F1) and has been registered on ClinicalTrials.gov as Inhaled Nitric Oxide ReDuce postoperatIve pulmoNAry complicaTions in patiEnts (INORDINATE) trial with the identifier NCT05721144. The study protocol is reported in accordance with the Standard Protocol Items Recommendations for Interventional Trials checklist.54

Study aim

The aim of our study is to investigate and test the hypothesis that intraoperative NO inhalation can reduce PPCs from 30% to 20% in participants planning to undergo surgery with general anaesthesia and who have recently (within 7 weeks) been infected with SARS-CoV-2.

Participants

Our study aims to include a total of 660 participants who have been recently infected with SARS-CoV-2 within a period of 7 weeks. The eligibility criteria for participation in the study include being 18 years of age or older and scheduled for surgery under general anaesthesia. However, participants who meet any of the exclusion criteria listed in table 1 will not be included in the trial.

Randomisation, allocation and concealment, and blinding

Eligible participants will be randomly allocated at 1:1 ratio using a web-based randomisation service (Research Electronic Data Capture, REDCap, developed by Vanderbilt University). The randomisation process will use permuted-block randomisation, with stratification based on the predicted risks of PPCs using the Assess Respiratory Risk in Surgical Patients in Catalonia (ARISCAT) index (<45 (low and medium risk), ≥45 (high risk); as specified in table 2), as well as the time interval from infection to surgery(≤10 days, 11–28 days and 29–49 days). Random blocks of 4 or 6 will be used for randomisation.

The allocation of participants into groups will be carried out in a manner that ensured blinding. This means that the participants, caregivers and investigators responsible for analysing data and assessing outcomes were unaware of the group assignments. Only one investigator will be unblinded and tasked with preparing the appropriate test gas. Subsequently, the test gases are to be then covered and kept blinded to the other individuals involved in the study. Both NO and placebo gas (air and O2) will be delivered to the breathing circuit through an NO delivery system. The entire device would be covered with a surgical drape. The unblinded investigator, who has no involvement in any aspect of the surgery, anaesthesia, data recording, outcome assessment or data analysis, would adjust the concentration and gas flow as necessary.

Interventions

Participants will be randomised to receive either NO or placebo gas during general anaesthesia (figure 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Study flow chart. ARISCAT, Assess Respiratory Risk in Surgical Patients in Catalonia; ICU, intensive care unit; NO, nitric oxide; DVT, deep venous thrombosis

The NO group will receive NO inhalation at a concentration of 80 ppm through the respiratory circuit of a mechanical ventilator. The intervention will commence immediately after tracheal intubation and will persist until either extubation or the patient exits the operation room, whichever happens first.48–50 The treatment gas will be delivered to the breathing circuit through an NO delivery system (iNO will portable NO therapy device, Nanjing Novlead Biotechnology, China; or SLE 3600 INOSYS iNO System, SLE, UK). The concentration of NO delivered to each participant will be closely monitored by the unblinded investigator. NO and the nitrogen dioxide (NO₂) levels will be monitored through in-line NO/NO₂ sensors incorporated in the NO delivery system. Met haemoglobin levels will be monitored via blood gas or periopheral pulse oximeter (Masimo Rainbow Pulse CO-Oximetry, Masimo, California, USA).

To ensure safety, the NO₂ levels will be monitored and maintained below 5 ppm. Previous tests have demonstrated that our delivery circuits do not exceed 1.5 ppm of NO₂ when delivering 80 ppm of NO gas at 90% FiO₂.55 Methaemoglobin levels will be monitored and kept below 5%. If the safety threshold is exceeded, the concentration of NO delivered will be halved from 80 ppm to 40 ppm and closely monitored until a reduction is observed. If the levels of NO2 or methaemoglobin persist above 5 ppm or 5%, respectively, the NO concentration will be progressively halved until reaching a safe level. The reason and frequency of adjustments made to the delivery of NO will be carefully documented and recorded. Any modifications or changes to the NO delivery, such as concentration adjustments or reductions, will be noted in the study records. The reasons for these adjustments, such as safety concerns or specific participant requirements, will also be documented. The frequency of these adjustments, including how often and when they occur, will be recorded as well. This detailed documentation will help ensure accurate monitoring and assessment of the NO delivery throughout the study.

After the intervention, the weaning and discontinuation of NO will be conducted while closely monitoring the participant’s haemodynamics. Abrupt discontinuation of inhaled NO can potentially lead to a rebound of PAP, resulting in worsened oxygenation, hypoxaemia, systemic hypotension, bradycardia and right ventricular failure.56

Therefore, the weaning process for NO at our institution involves gradually decreasing the concentration. If any of the following adverse events occurs during the NO weaning process and is considered to be related to the discontinuation, after ruling out other potential causes, the concentration of NO will be reset to the lowest level at which the participant stabilises. This occurrence will be documented as a failed attempt at NO weaning

• Steeply worsening hypotension: This is defined as a decrease in mean arterial pressure by more than 20 mm Hg. In such cases, intravenous vasopressors will be initiated. For patients who were already receiving vasopressors before the weaning process, a greater than 50% increase in the vasopressor dose will be required to maintain haemodynamics.

• Sharp deterioration of hypoxaemia: If there is a significant worsening of hypoxaemia, characterised by the need for an increase in the FiO2 by more than 20%, or an elevation of positive end-expiratory pressure by more than 5 cm H2O, the intervention will be adjusted accordingly.

By carefully monitoring and adjusting the weaning process, we aim to minimise any potential adverse effects associated with the discontinuation of NO.

The control group will receive a mixture of air and O₂ inhalation through the respiratory circuit of a mechanical ventilator, with the settings identical to those in the NO.

Participants will be allowed to discontinue the study at any time, either voluntarily or based on the clinician’s or investigator’s judgement for safety reasons. These reasons may include but are not limited to, adverse reactions to the intervention, worsening health conditions or any other factors that could jeopardise the participant’s well-being.

• Sudden onset of hypotension attributed to ventricular (left and/or) dysfunction, characterised by a drop in MAP exceeding 20 mm Hg, not attributable to hypovolaemia, haemorrhage or vasoplegia.

• Severe hypoxia defined as decrease in oxygen saturation (SpO₂) to below 88% (equivalent to a PaO₂ level below 54 mm Hg) while on 100% oxygen, not attributable to pulmonary oedema, ARDS or pulmonary embolism.

• Elevated NO₂ levels exceeding 5 ppm and methaemoglobin levels exceeding 5%, which prove irreversible even after reducing NO delivery to below 5 ppm.

All participants will receive perioperative management based on our institutional standard practice, which includes various components such as preoperative evaluation and preparation, general anaesthesia with or without neuraxial anaesthesia and/or nerve blocks, mechanical ventilation, inotropic drugs and vasopressors, postoperative sedation and analgesia, diuretics, intravenous fluids, antibiotics and invasive monitoring. This management regimen will encompass a range of interventions and monitorings, including but not limited to non-invasive or invasive arterial pressure, ECG, central venous pressure, cardiac output, pulse oximetry, temperature, urine output, arterial blood gases, coagulation monitoring and frequent routine laboratory examinations. No additional interventions or laboratory tests beyond the standard perioperative management will be conducted on participants.

Data collection and management

Data will be collected prospectively at various time points throughout the study. The baseline evaluation will capture information about patients’ characteristics, including demographics, comorbidies, medications, preoperative lab tests, symptoms, vaccination and treatment history of SARS-CoV-2 infection. During the surgical procedure, data will be collected on the administration of anaesthesia, including dosage, timing, mode of administration of all drugs, as well as the mode and parameters for mechanical ventilation. Surgical procedure-related characteristics, such as the incision position, duration of procedure and any intraoperative adverse events, will also be recorded. Postoperative management, including analgesia regimen, transfer location from the operation room (postoperative anaesthesia care unit, ICU or ward) and any complications, will be documented.

Regarding the intervention treatment, specific data will be collected concerning its implementation. This will encompass whether the intervention adhered to the randomisation results, the dosage range of NO inhalation, maximum levels of NO2 and methaemoglobin, any adjustments in NO concentration, the intervention’s duration and any unsuccessful attempts to discontinue NO. These intervention-related characteristics will be collected and recorded by the unblinded investigator and will remain inaccessible to other study stakeholders during the implementation period. Follow-up data will be gathered 30 days postsurgery, focusing on adverse pulmonary events, hospital readmissions and survival.

All data collected for the trial will be entered into the REDCap application,57 a secure data management system. Access to the data will be restricted and granted by the principal investigator (PI) to authorised investigators within the study team. The PI will assign specific privileges for data import and export to study investigators. To ensure the quality and integrity of collected data, various measures are implemented. Software properties, such as logic checks and validation of data fields (eg, reference range, valid or invalid values), are used to optimise data quality. Additionally, study investigators will conduct weekly checks to ensure data consistency. If any errors related to data collection or entry are identified, the study team will correct them by referring to the original data source. These corrections will be summarised in data quality reports to maintain accuracy and reliability.

Outcomes

The primary endpoint is a composite measure of PPCs occurring within the first 7 days after surgery. The definitions of these complications are based on the European Perioperative Clinical Outcome definitions58 as well as other recognised definitions from relevant literature.59 The specific components of the composite PPCs include respiratory infection, respiratory failure, pleural effusion, atelectasis, pneumothorax, bronchospasm, aspiration pneumonitis and pneumonia. Detailed definitions for each of these components can be found in table 3. Outcomes will be measured daily until postoperative day 7 or hospital discharge by blinded investigators. Chest X-rays and routine laboratory tests will be assessed when any pulmonary complications are suspected.

Secondary endpoints will include postoperative 30-day all-cause mortality; severity of PPCs scaled by Clavien-Dindo classification60 (online supplemental file etable 1); unplanned ICU admission; postoperative length of hospital stay; thrombotic events (including deep venous thrombosis and pulmonary embolism) and postoperative Comprehensive Complication Index. Other measurements recorded include non-pulmonary complications including stroke, cardiac infarction, ventricular dysfunction, delirium, acute kidney injury, surgical revision for bleeding, hypotension/hypertension, arrhythmia, hyper-responsiveness. We also recorded the number of times that the concentration of NO was adjusted. To perform 30-day follow-up, telephone contact (patient and/or relatives) will be used. In case loss to follow-up by telephone, the following methods will be used: contacting the patient’s general practitioner or other relatives (during enrolment at least three contact information from close relatives will be collected).

Safety outcomes include hypotension, defined as SBP<90 mm Hg or a drop ≥30% of baseline lasting for 5 min; hypertension, defined as SBP>180 mm Hg or a rise ≥30% of baseline lasting for 5 min; arrhythmia consisting of bradycardia (heart rate (HR) <40 beats per minute (bpm)), tachycardia (HR>100 bpm), or new onset of arrhythmia requiring antiarrhythmic drugs; airway hyper-responsiveness, defined as an airway peak pressure >40 cm H2O; the number of NO concentration adjustments; and massive bleeding (haemorrhage exceeding 1000 mL).

Sample size estimates and statistical analysis

An epidemiologist with extensive experience in designing, conducting and analysing clinical trials, not involved in patient management and blinded to the assigned intervention will be responsible for the statistical analyses.

Data will be stored electronically via a web-based CRF, and analysis will be conducted using the R program (R Core Team, 2016, Vienna, Austria61). We will not apply any imputation for missing primary outcomes, while other important missing covariates will be imputed appropriately. All analyses will be performed on the cohort of randomised patients according to their assigned randomisation groups, except for those whose planned surgeries are delayed or cancelled after randomisation.

The incidence of PPCs is reported to be approximately 20% among all surgical populations,15 62 and this rate can increase to as high as 39.5% after COVID-19 infections.5 We hypothesise a PPC incidence of 30% in the control group. Based on recent studies demonstrating the efficacy of inhaled NO in improving oxygenation during COVID-19,47 50 we anticipate a reduction in PPC incidence from 30% to 20% (a 30% relative reduction). The sample size calculation is performed using Pearson’s χ² test with a two-sided significance level of 0.05 and 80% power. To account for a 10% drop-out rate with continuity correction, a sample size of 330 per group is needed.

Demographic and baseline disease characteristics will be summarised with the use of descriptive statistics. Categorical variables will be reported as absolute numbers and percentages. Unadjusted univariate analyses, to compare the two treatment groups, will be based on χ² or Fisher’s exact test. Relative risks and 95% CIs will be calculated by means of the two-by-two table method with the use of log-normal approximation. Continuous variables will be reported as mean±SD or median and IQR. Normality will be evaluated using visual histogram evaluation and a Q-Q plot. Between-group differences will be evaluated using the t-test or Wilcoxon signed rank test, in accordance with normality of the distribution. The full analysis R code will be uploaded on GitHub before completion of enrolment.

For the primary analysis, the χ² test will be employed for an unadjusted analysis, and risk ratios with 95% CIs will be estimated using Wald’s likelihood ratio approximation test. The effect size of the intervention will be expressed as risk differences along with their corresponding 95% CIs. The number needed to treat will be calculated as the reciprocal of the risk difference, and the result will be rounded to the nearest whole number. The outcome will be further analysed as a binary variable through generalised linear models (GLM) with a binomial distribution family and a logit link function. This analysis will include adjustments for stratification variables (ARISCAT index and time interval from infection to surgery) as fixed effects.

Several prespecified secondary analyses were conducted. First, for preplanned secondary outcome measures, GLM regression analyses were employed for binary outcomes, with distribution choices capable of handling right-tailed positive data (eg, truncated Poisson, gamma distribution or inverse Gaussian). Cox proportional hazard models were used for length-of-stay outcomes, incorporating a shared frailty model. The proportionality assumption was visually inspected using Kaplan-Meier plots, log-log plots and Schoenfield residuals, with no clear evidence of divergence. Non-parametric comparisons were made using Kaplan-Meier curves and between-group log-rank tests. For continuous outcomes, linear or quartile regression was chosen based on the skewness of the distribution. Models were adjusted for the ARISCAT index and the time interval from infection to surgery as per randomisation strata, treated as fixed effects. Between-group risk differences were estimated and reported with 95% CIs. Adverse reactions were documented and analysed, with differences in incidence tested using Fisher’s exact tests. No type I error corrections were conducted for multiple tests in this context.

Prespecified subgroup analyses of the primary outcomes will be performed using logistic regression models for subgroups defined by age (<65 or ≥65), sex (male or female), types of surgery (cardiac or non-cardiac), body mass index (≤35, >35 kg/m²), surgical complexity grades (minor, intermediate and major), duration of anaesthesia (<2 hours, 2–3 hours or ≥3 hours), a history of smoking, one-lung ventilation, ARISCAT (<26, 26–44 or ≥45), and time intervals between infection and surgery (≤10, 11–28, 29–49 days). The effect modifications or interactions will be assessed through the inclusion of interaction terms between the stratification variable and randomisation group in the logistic regression model.

In a prespecified sensitivity analysis, the effect of the intervention on the primary outcome will be re-estimated with GLM using a binomial distribution. Additional adjustment will be made for the aforementioned subgroups and any variables showing substantial imbalance across treatment arms at baseline. To achieve this, a logistic regression model with stepwise selection will be employed to estimate the treatment effect and predictors of PPCs. Prerandomisation clinical data will be included in the model if their univariable p<0.1, and there is no correlation between them. Collinearity and overfitting will be assessed using a stepwise regression model and the Pearson correlation test. The treatment group (NO or control) will be forced into the multivariate model. If the outcome event proves to be rare, a Poisson regression model will be used. Classic logistic regression will be performed with a consistent number of events, and the number of covariates in the model will be decided based on the number of outcomes. The ARISCAT score and operation duration score used as one randomisation stratification will be treated as continuous variables, exploring their continuous impacts on the outcome. Time until pulmonary complications within seven postoperative days was assessed using Kaplan-Meier survival curves and log-rank test. Per-protocol and as-treated analyses of PPCs will be performed as sensitivity analyses. We anticipate a low missing rate of the primary outcome. If more than 5% data are missing for the primary outcome, multiple imputations will be carried out as a sensitivity analysis.

Since the primary outcome of the present study is a composite one, the choice of the statistical method is an important part of design because various methods provide different power, depending on the situation. In addition to the standard analysis described above, the following analyses will be performed to test the robustness of the trial findings: (1) the number of positive component events (ie, ‘count’) across the composite will be assessed. The groups will be compared on the count using a Wilcoxon rank-sum test, and the OR with the 95% CI will be assessed with a proportional odds logistic regression model; (2) the effect of the intervention in each component will be analysed using a GLM using a Bonferroni correction for multiple comparisons. The 99.37% Bonferroni-corrected CIs will be reported (1–0.05/8=0.9937); (3) a multivariate (ie, multiple outcomes per subject) generalised estimating equations (GEE) model will be used to estimate a common effect OR across the components; (4) the average relative effect test will be assessed by averaging the component-specific treatment effect from the distinct effects model, and testing whether the average is equal to 0. In the GEE distinct effect model, a distinct treatment effect is estimated for each component; (5) heterogeneity of treatment effect across components will be assessed by a treatment-by-component interaction test in the distinct effects GEE model and (6) patients in the treatment and control groups will be formed into matched pairs based on their risk profiles. The win ratio will be calculated and reported to address the inherent limitation of composite outcomes.63

To enhance the interpretability of the results and provide valuable information for decision-making, the INORDINATE trial will employ Acceptability Curve Estimation using Probability above Threshold (ACCEPT) analysis.64 This analysis method involves calculating and plotting the probabilities of the true difference between treatments being above various acceptability thresholds, based on the data obtained from the trial. By using this approach, the limitations of traditional binary trial conclusions, which categorise results as either positive (meeting the trial aim) or negative (not meeting the trial aim), can be overcome. ACCEPT analysis provides a more nuanced understanding of the data and allows for a range of possible threshold values to be considered, thereby facilitating more informed decision-making.

Statistical significance for the primary outcome will be indicated by a p value of 0.05 with use of a two-sided hypothesis test. No correction for multiple comparisons was applied in the evaluation of secondary or other outcomes. Thus, such results are exploratory and are reported as point estimates with 95% CIs.

Monitoring of the study

Independent auditors will be responsible for ensuring compliance with the clinical trial protocol and verifying the accuracy of data collection in accordance with the Good Clinical Practice (GCP) guidelines. The study will be monitored and followed up according to the protocols established during the initial set-up, and all reporting will adhere to current national and international requirements. This rigorous monitoring and adherence to guidelines aim to maintain the integrity and quality of the study.

Ethical considerations

The administration of NO is deemed clinically safe in this trial and will be conducted by a skilled professional with experience in respiratory therapy. To mitigate the potential inflammatory response in the respiratory tract or lung tissue injuries resulting from the generation of NO₂, the entire process will be continuously monitored to ensure concentrations remain below 5 ppm. Although there is a recognised risk of harm to healthy individuals when methaemoglobin (metHb) content exceeds 15%–20%,65 adverse effects of inhaling NO are rarely reported in the literature. Furthermore, it is noteworthy that the binding of NO to haemoglobin is reversible, with a half-life of 15–20 min after stopping NO administration.66

This protocol strictly follows the principles of the ‘Declaration of Helsinki’ in full. The study will be governed by the implementation of current International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use - Good Clinical Practice (ICH-GCP) and China National Drug Administration (NMPA) related regulations. The study data planned to be collected are essential for assessing the efficacy, safety, quality and application of drugs. The collection and utilisation of this data ensure confidentiality and compliance with relevant laws and regulations, thereby protecting subject privacy. Written informed consent (online supplemental file 2) will be obtained in accordance with the Declaration of Helsinki,67 laws and regulations for conducting clinical trials, the General Data Protection Regulation, the data protection directive and other rules applying to the protection of personal data. Patients are allowed to decline consent at any point during the study; in such cases, all relevant information shall be removed. All private information will be kept confidential and removed before data analyses.

Patient and public involvement

No patients or members of the public were involved in the design and development of the protocol. Furthermore, they will not participate in the analysis or presentation phases of the study on its completion.

Ethics and dissemination

Ethical approval has been obtained on 2 February 2023 from the Xijing Hospital Ethical Committee (KY20232058-F-1, protocol version: 1.0, date: 2 February 2023). The study is registered on ClinicalTrials.gov (NCT05721144) and available online (https://clinicaltrials.gov/ct2/show/NCT05721144?term=NCT05721144&draw=2&rank=1). Any protocol changes will be approved by relevant ethical review boards. Final results will be submitted to a peer-reviewed journal and presented at relevant academic conferences.

Study initiation, timing and source of funding

The study commenced after receiving approval from the institutional ethical committee and subsequent registration on clinicaltrials.gov. Enrolment includes consecutive participants who have provided written informed consent. The first patient was enrolled on 17 February 2023, and the first 60 participants were randomised by 8 June 2023. The study team is exclusively responsible for designing, conducting the trial, as well as analysing, drafting and editing the final reports.

Funding for the trial is provided by the National Natural Science Foundation of China (81970448), the university and the hospital foundation for patient outcome improvement (2021LC2202 and XJZT21L17).