Nonintubated thoracic anesthesia and postoperative cough after video-assisted thoracoscopic surgery: A propensity score-adjusted cohort study.

Save

BACKGROUND:

This study aimed to evaluate the association of nonintubated thoracic anesthesia with postoperative in-hospital antitussive medication use and postoperative complications after video-assisted thoracoscopic surgery.

METHODS:

This single-center retrospective cohort study included 223 patients who underwent video-assisted thoracoscopic surgery between January and December 2025. The exposure was nonintubated thoracic anesthesia (NITS) versus double-lumen tube intubation. The primary outcome was postoperative in-hospital antitussive medication use.

Secondary outcomes included chronic cough, sore throat, hoarseness, overall postoperative complications, postoperative pulmonary complications (PPCs), and antibiotic use beyond 48 h. Propensity score-based stabilized inverse probability of treatment weighting was applied.

RESULTS:

After weighting, NITS was associated with lower odds of postoperative antitussive medication use (odds ratio [OR]: 0.46; 95% confidence interval [CI]: 0.25–0.85; P = 0.013), chronic cough (OR: 0.36; 95% CI: 0.16–0.84; P = 0.019), and hoarseness (OR: 0.27; 95% CI: 0.08–0.87; P = 0.029). NITS was also associated with lower odds of overall postoperative complications (OR: 0.50; 95% CI: 0.27–0.94; P = 0.031). No significant association was observed for PPCs (OR: 1.03; 95% CI: 0.49–2.19; P = 0.929).

CONCLUSION:

In this propensity score-weighted retrospective cohort study, NITS was associated with lower odds of postoperative in-hospital antitussive medication use, chronic cough, hoarseness, and overall postoperative complications after video-assisted thoracoscopic surgery but not with PPCs. Prospective randomized studies are needed to confirm these findings.

Background

Postoperative cough following lung resection, commonly termed cough after pulmonary resection, is one of the most frequent respiratory symptoms in thoracic surgery, with reported incidence ranging from 21.1% to 55.8%.[1] Approximately 18% of patients may experience chronic cough lasting 1 year or longer,[2] suggesting that postoperative cough may follow a prolonged clinical course in a subset of individuals. Persistent cough can exacerbate incisional pain, disrupt sleep and social interactions, impair long-term quality of life,[3] and even contribute to anxiety or depressive symptoms.[4] The occurrence of postoperative cough has been associated with patient-related factors such as age, sex, and pre-existing cough, as well as surgical factors including the extent of pulmonary resection and the degree of lymph node dissection.[12356] However, these characteristics are largely nonmodifiable during the perioperative period, limiting their usefulness in preventive strategies. Anesthesia-related factors, including intubation duration and anesthesia time, have also been identified as independent risk factors for postoperative cough,[3578] indicating that perioperative management and airway strategies may represent key modifiable determinants of cough after pulmonary resection.

The conventional anesthetic technique for video-assisted thoracoscopic lung resection is double-lumen tube (DLT) intubation combined with one-lung ventilation (OLV), which provides a stable operative field.[9] However, DLT is inherently invasive and may increase the risk of airway-related complications, including mucosal injury, laryngeal edema, bronchospasm, and postoperative hoarseness.[10] In addition, the deep levels of anesthesia, neuromuscular blockade, and positive-pressure mechanical ventilation required to achieve OLV can contribute to diaphragmatic dysfunction, alveolar overdistension, ventilation–perfusion mismatch, and ventilator-associated lung injury.[111213] With the advancement of enhanced recovery after surgery (ERAS) principles, nonintubated thoracic anesthesia (NITS), which preserves spontaneous respiration, has regained interest as a minimally invasive anesthetic strategy aimed at facilitating postoperative recovery.[1415]

From a physiological perspective, the cough reflex can be triggered by airway inflammation, mechanical stimulation, or inhaled chemical or mechanical irritants.[16] The upper airway, particularly the larynx, carina, and proximal bronchial bifurcation, is densely populated with rapidly adapting receptors and C-fibers,[17] which are highly sensitive to mechanical and chemical stimuli and represent key sites for mechanically induced cough. By maintaining spontaneous breathing through a laryngeal mask airway, face mask, or nasal cannula, NITS avoids the mechanical friction of a DLT against the vocal cords and carina,[1819] and reduces the barotrauma and inflammatory response associated with mechanical ventilation, thereby potentially lowering the risk of mechanically induced cough. However, evidence specifically addressing the effect of NITS on perioperative cough remains limited.

A retrospective study and small randomized controlled trials suggest that maintaining spontaneous respiration may lower the incidence of postoperative cough.[2021] Nevertheless, a recent meta-analysis incorporating 19 randomized controlled trials reported no significant reduction in perioperative cough with NITS compared with intubated techniques.[22] Notably, multicenter randomized trials have shown that reducing mechanical stimulation of the airway, such as using a supraglottic airway device instead of endotracheal intubation, can significantly decrease postoperative cough and hoarseness, highlighting airway irritation as a potential key contributor to cough.[23] The heterogeneity observed across existing studies further indicates that different airway management strategies may exert varying impacts on cough-related outcomes.

In light of these limitations in the existing literature, we conducted a propensity score-adjusted retrospective cohort study to evaluate the effect of NITS on postoperative cough, hoarseness, and other perioperative complications. Our findings aim to provide additional evidence to inform thoracic anesthesia strategies and support the implementation of ERAS pathways.

Materials and Methods

Study design

This was a single-center, retrospective cohort study that screened 618 patients who underwent thoracic surgery between January 1, 2025, and December 1, 2025, at a tertiary teaching hospital. The study was designed and reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. The completed STROBE checklist is Supplementary File 1.

Eligibility criteria

Patients were eligible for inclusion if they were aged ≥18 years and underwent elective video-assisted thoracoscopic surgery (including wedge resection, segmentectomy, or lobectomy). Exclusion criteria included emergency surgery; complex procedures such as pneumonectomy or severe pleural adhesions; operations performed with anesthetic techniques other than DLT or NITS; the presence of significant airway disease (e.g., active hemoptysis or major airway stenosis); and cases with substantial missing clinical data.

Data collection

Two investigators reviewed the hospital electronic medical record system, anesthesia information system, and postoperative follow-up records to extract the following information for each patient: demographic characteristics, diagnostic and medical history data, perioperative anesthetic and surgical information, postoperative outcomes, and laboratory test results. All extracted data were entered into an electronic database using double data entry with independent verification, and a third investigator performed random audits to ensure accuracy. Definitions and sources of all recorded variables are provided in Supplementary Table 1, and details of missing data are presented in Supplementary Table 2.

Exposure

Patients were categorized into two groups according to the anesthetic technique documented in the anesthesia information system. The exposure group comprised patients who underwent nonintubated laryngeal mask airway-based anesthesia with preserved or intermittently supported spontaneous ventilation. Patients in the non-intubated group were managed according to the institutional anesthetic practice for NITS, as described in Supplementary eMethod 1.

The control group consisted of patients who received DLT intubation under general anesthesia with OLV.

Outcome measures

The primary outcome was the use of antitussive medications during postoperative hospitalization, defined as the administration of any predefined cough-suppressant medication recorded in the postoperative medication records.

Secondary outcomes included the incidence of postoperative chronic cough, postoperative sore throat and hoarseness, the use of therapeutic antibiotics beyond 48 h after surgery, overall postoperative complications, and postoperative pulmonary complications (PPCs). Overall postoperative complications were defined according to the European Perioperative Clinical Outcomes (EPCO) criteria and included any EPCO-defined postoperative complication with a severity of Clavien–Dindo Grade II or higher, i.e.,[24] complications requiring pharmacological treatment or more intensive intervention. PPCs were defined according to the EPCO criteria as a composite endpoint comprising respiratory infection, respiratory failure, pleural effusion, pneumothorax, atelectasis, bronchospasm, and aspiration pneumonia.[24] For this study, only EPCO-defined respiratory events with a severity of Clavien–Dindo Grade II or higher were counted as PPCs.

Detailed definitions of all outcome measures are provided in Supplementary Table 1.

Statistical analysis

All analyses were conducted according to a prespecified study protocol and statistical analysis plan.

The sample size was determined by the number of eligible cases available during the study period. Continuous variables were screened for outliers using boxplot inspection. Missing data were handled using multiple imputation by chained equations (mice package in R), generating five imputed datasets.

Imputation models included all variables used in the propensity score and outcome regression models, together with other auxiliary variables available in the study database.[25] To create a single completed dataset for subsequent analyses, imputed values for continuous variables were averaged across the five datasets. For dichotomous variables, values were coded as 1 if at least three of the five imputed datasets indicated the event and as 0 otherwise. Descriptive statistics and subsequent stabilized inverse probability of treatment weighting (IPTW) analyses were then performed using this completed dataset.

Normality was assessed using the Shapiro–Wilk test, histograms, and Q–Q plots (α =0.05). Continuous variables were summarized as mean ± SD or median (Q1, Q3), as appropriate, and categorical variables as counts (percentages). Statistical analyses were performed using RStudio and R version 4.5.0.

Propensity score estimation

Propensity scores were estimated using a logistic regression model to account for the nonrandomized nature of anesthetic technique selection, as anesthesiologists may have clinical preferences when choosing between double-lumen OLV and non-intubated general anesthesia. Accordingly, baseline variables that could influence anesthetic decision-making were incorporated into the propensity score model, including sex, age, body mass index (BMI), number of comorbidities, American Society of Anesthesiologists (ASA) physical status, preoperative FEV₁ percent predicted, preoperative cough requiring medication, planned type of lung resection (wedge resection, segmentectomy, or lobectomy), surgical laterality (right vs. left lung), upper lobe resection, and whether the lead anesthesiologist was a senior anesthesiologist.

Propensity score model diagnostics included assessment of multicollinearity among model covariates using the maximum generalized variance inflation factor and discrimination using the C-statistic, equivalent to the area under the receiver operating characteristic curve. These diagnostics are summarized in Supplementary Table 3.

Inverse probability of treatment weighting

Stabilized inverse probability of treatment weights for the average treatment effect (ATE) were derived from the estimated propensity scores and applied to construct a weighted pseudo-population. The distribution of stabilized weights was evaluated using the minimum, selected percentiles, median, mean, and maximum weight. Post-weighting precision and balance were summarized using the effective sample size [Supplementary Table 3].

Covariate balance after weighting was assessed using absolute standardized differences (ASDs), with an ASD ≤0.10 considered to indicate adequate balance. Binary outcomes were analyzed using IPTW-weighted logistic regression and reported as odds ratios (ORs) with 95% confidence intervals (CIs).

Sensitivity analyses

Sensitivity analyses were conducted to evaluate the robustness of the primary findings. In addition to the IPTW-based primary analysis, propensity score matching (PSM) was performed as a complementary approach to address confounding by indication. The same covariates used in the primary propensity score model were used to estimate the propensity scores for matching.

Patients receiving NITS were matched to patients receiving DLT anesthesia using 1:1 nearest-neighbor matching with a caliper width of 0.10 of the standard deviation of the logit-transformed propensity score. Covariate balance after matching was assessed using ASDs, with a value ≤0.10 considered to indicate adequate balance. The association between anesthetic technique and the primary outcome was then estimated in the matched cohort using conditional logistic regression.

As an additional sensitivity analysis, a conventional multivariable logistic regression model was fitted in the overall cohort, adjusting for clinically relevant baseline covariates. To further evaluate the potential impact of unmeasured confounding, E-values were calculated to quantify the minimum strength of association that an unmeasured confounder would need to have with both the exposure and the outcome, conditional on measured covariates, to fully explain the observed association [Supplementary eMethod 2].[26]

Exploratory analysis

As a prespecified exploratory analysis, perioperative C-reactive protein (CRP) levels measured at three time points (T0–T2) were analyzed using an IPTW-weighted generalized estimating equation (GEE) model with a gamma distribution and log link. The model included anesthetic strategy, time, and the interaction between anesthetic strategy and time to evaluate differences in longitudinal CRP trajectories between groups. Repeated measurements within the same patient were accounted for using GEE, and the patient-level stabilized IPTW weights derived from the primary propensity score model were applied to all repeated CRP observations from the corresponding patient.

Results

Patient characteristics

Figure 1 illustrates the patient selection process. Between 1 January 2025 and 1 December 2025, a total of 618 thoracic surgical procedures were screened. After applying the inclusion and exclusion criteria, 223 patients were included in the final cohort, comprising 111 patients in the NITS group and 112 patients in the double-lumen endotracheal intubation group.

Table 1 summarizes baseline demographic and clinical characteristics before and after stabilized IPTW. Before weighting, compared with the DLT group, patients in the NITS group were younger, less frequently male, had a lower BMI, fewer comorbidities, and a higher preoperative FEV₁ percent predicted. The NITS group also had lower proportions of ASA physical status III, preoperative cough requiring medication, and preoperative hypoxemia, and a higher proportion of procedures managed by a senior anesthesiologist.

The distribution of planned lung resection differed between groups, with fewer lobectomies and more wedge resections in the NITS group. After stabilized IPTW, all covariates included in the propensity score model achieved adequate balance, with absolute standardized mean differences below 0.10 [Table 1].

Inverse probability of treatment weighting results (primary and secondary outcomes)

In the stabilized IPTW-weighted analysis, NITS was associated with lower odds of postoperative in-hospital antitussive medication use compared with DLT anesthesia (OR, 0.46; 95% CI, 0.25–0.85; P = 0.013) [Figure 2]. NITS was also associated with lower odds of chronic cough (OR, 0.36; 95% CI, 0.16–0.84; P = 0.019), and postoperative hoarseness (OR, 0.27; 95% CI, 0.08–0.87; P = 0.029). In addition, NITS was associated with lower odds of overall postoperative complications (OR: 0.50; 95% CI: 0.27–0.94; P = 0.031).

No significant associations were observed for postoperative sore throat (OR: 0.63; 95% CI: 0.21–1.90; P = 0.411), antibiotic use beyond 48 h after surgery (OR: 0.67; 95% CI: 0.36–1.25; P = 0.214), or PPCs (OR: 1.03; 95% CI: 0.49–2.19; P = 0.929). Unadjusted incidences of perioperative and postoperative outcomes in the overall cohort and by anesthesia strategy are presented in Supplementary Table 4.

Sensitivity analysis results

After PSM, covariate balance was reassessed using ASDs. All matched covariates achieved adequate balance, with ASDs ≤0.10, and the postmatching balance is shown in Figure 3. In the matched cohort, NITS was associated with lower odds of postoperative in-hospital antitussive medication use compared with DLT anesthesia (OR: 0.44; 95% CI: 0.23–0.88; P = 0.019) [Table 2].

Conventional logistic regression yielded consistent findings for NITS versus DLT, with an OR of 0.41 (95% CI: 0.23–0.75; P = 0.004) in the univariable analysis and an adjusted OR of 0.46 (95% CI: 0.23–0.91; P = 0.025) after adjustment for the covariates listed in Supplementary Table 5.

In the E-value analysis, based on the risk ratio approximation of the observed odds ratio, the E-values were 2.32 for the point estimate and 1.39 for the 95% confidence limit closest to the null [Supplementary eMethod 2].

Intraoperative management and events

Unweighted descriptive comparisons showed that intraoperative and perioperative care pathways differed by anesthetic strategy [Supplementary Table 6]. Compared with the DLT group, the NITS group had a higher frequency of hypercapnia (78.4% vs. 39.3%), a lower median intraoperative intravenous morphine-equivalent dose (113.20 [Q1, 97.00; Q3, 141.65] mg vs.

150.00 [Q1, 118.92; Q3, 179.50] mg), and lower median doses of cisatracurium and mivacurium. Chest tube placement (71.2% vs. 94.6%) and urinary catheterization (20.7% vs.

79.5%) were also less frequent in the NITS group.

Exploratory analysis results

In this prespecified exploratory analysis, model-based marginal CRP estimates increased from T0 to the postoperative time points in both groups [Figure 4]. In the final stabilized IPTW-weighted gamma GEE model with a log link, the group-by-time interaction was not statistically significant (global Wald P = 0.087). The estimated NITS-to-DLT ratio of marginal means was 1.13 (95% CI: 0.79–1.62; P = 0.507) at T0, 0.71 (95% CI: 0.58–0.88; Holm-adjusted P = 0.005) at T1, and 0.78 (95% CI: 0.60–1.01; Holm-adjusted P = 0.119) at T2.

Discussion

In this retrospective cohort study of 223 patients, NITS was associated with lower odds of postoperative in-hospital antitussive medication use and a lower incidence of reported chronic cough at the routine 8-week follow-up. Antitussive medication use, the primary outcome, was selected as a treatment-based indicator of clinically relevant cough-related symptom management during hospitalization. The association with the primary outcome was consistent across sensitivity analyses.

For this outcome, the E-value was 2.32 for the point estimate, indicating that an unmeasured confounder would need to be associated with both anesthetic strategy and antitussive medication use by risk ratios of at least 2.32 each, beyond the measured covariates, to fully explain the observed association.[26] However, the E-value for the confidence limit was 1.39, indicating that residual confounding of more modest strength could shift the CI to include the null.

Postoperative cough is a common concern after thoracic surgery and may adversely affect postoperative recovery and quality of life.[2] A previous meta-analysis of 19 studies involving 4,755 patients identified young age, female sex, preoperative cough, right lobe operation, and lobectomy as independent risk factors for postoperative cough.[1] To reduce confounding by these and other observed preoperative factors related to anesthetic selection, we used stabilized IPTW to balance baseline characteristics between the two anesthetic groups.

Previous studies have suggested that NITS with nasal oxygen supplementation and preserved spontaneous ventilation may be associated with a lower incidence of postoperative cough.[20] In the present cohort, NITS using a supraglottic airway with preserved spontaneous breathing was associated with more favorable postoperative cough-related outcomes than DLT anesthesia, including lower use of in-hospital antitussive medication and a lower incidence of reported chronic cough at the routine 8-week follow-up. However, a previous meta-analysis including 19 randomized controlled trials reported that NITS did not reduce the incidence of perioperative cough.[22] Notably, that analysis primarily focused on intraoperative cough, which may be influenced by factors such as inadequate analgesia or insufficient suppression of local airway irritation. These methodological differences may partly account for the inconsistency between their findings and ours.

Beyond cough-related outcomes, postoperative sore throat and hoarseness are among the most common complications associated with endotracheal intubation. In the present study, NITS was associated with a significantly lower incidence of postoperative hoarseness compared with double-lumen endotracheal intubation (OR: 0.27; 95% CI: 0.08–0.87; P = 0.029) [Figure 2], whereas no significant reduction in postoperative sore throat was observed (P > 0.05). One possible explanation is that, although tracheal intubation is avoided, the continuous presence of a laryngeal mask in the pharyngeal region may itself cause localized mechanical irritation.[27]

The effect of NITS on overall PPCs remains controversial.[21222829] In the present study, NITS was not associated with a statistically significant reduction in the composite incidence of PPCs. This finding should be interpreted in light of the heterogeneous nature of the PPC composite. Descriptively, several individual pulmonary outcomes, including respiratory infection, pneumothorax, and bronchospasm, occurred less frequently in the NITS group, whereas pleural effusion and atelectasis occurred more frequently.

This mixed pattern may have attenuated any apparent difference in the overall PPC composite. By contrast, NITS was also associated with lower odds of the composite overall postoperative complications, a finding broadly consistent with previous reports.[1930] Potential contributors may include reduced airway instrumentation, less airway irritation, avoidance or minimization of controlled mechanical ventilation, lower intraoperative opioid and neuromuscular blocker exposure, and improved perioperative analgesic management. Nevertheless, given the retrospective design and limited sample size, these findings should be considered hypothesis-generating and warrant confirmation in larger prospective studies.

NITS with preserved spontaneous breathing provides an alternative anesthetic strategy to conventional DLT anesthesia. It reflects a less instrumented perioperative approach that aims to reduce anesthetic interventions through optimized analgesia and preservation of spontaneous ventilation, thereby maintaining a more physiologic perioperative state.[31] Importantly, NITS should not be regarded simply as the avoidance of tracheal intubation, but rather as a distinct anesthetic care pathway.

Consistent with this concept, unadjusted descriptive data showed that the NITS group received lower intraoperative intravenous morphine-equivalent doses and lower doses of neuromuscular blocking agents and underwent chest tube placement and urinary catheterization less frequently than the DLT group [Supplementary Table 6]. These differences characterize the overall perioperative care pathways associated with the two anesthetic strategies. In the prespecified exploratory analysis, CRP increased postoperatively in both groups.

Although the overall group-by-time interaction was not statistically significant, the NITS group had a lower model-based marginal CRP level at T1 (postoperative days 1–2) than the DLT group (ratio of marginal means, 0.71; 95% CI: 0.58–0.88; Holm-adjusted P = 0.005); this difference was not statistically significant at T2 (postoperative days 3–5) after adjustment for multiple comparisons. These exploratory observations are hypothesis-generating and require confirmation in prospective studies.[32]

NITS also involves inherent technical challenges. In particular, spontaneous ventilation with relatively low tidal volumes may predispose patients to intraoperative hypercapnia.[27283133] Hypercapnia was observed more frequently in the NITS group than in the DLT group (78.4% vs. 39.3%).

The median time to airway device removal was 10.00 min (Q1, Q3: 8.00, 22.50) in the NITS group and 16.50 min (Q1, Q3: 10.00, 25.00) in the DLT group. These unadjusted descriptive comparisons do not establish the safety or clinical consequences of intraoperative hypercapnia. Hypercapnia may cause cerebral vasodilation and increased cerebral blood flow, and sustained carbon dioxide retention may contribute to hypercapnic encephalopathy.[34] Because this retrospective study did not include standardized measures of arterial carbon dioxide tension, acid-base status, emergence quality, or agitation, the severity of hypercapnia and its effects on postoperative recovery could not be determined.

These issues should be evaluated in future prospective studies. Moreover, inadequate ventilation or other foreseeable factors may necessitate intraoperative conversion to endotracheal intubation. Consequently, the application of this technique should be approached cautiously, with careful patient selection and consideration of surgical characteristics and anesthetic expertise.

From another perspective, although NITS may reduce invasiveness from the patient’s standpoint, it does not diminish the complexity of anesthetic management. The implementation of multimodal analgesia and finely titrated sedation to preserve spontaneous breathing places higher technical demands on anesthesiologists and requires sustained vigilance, representing an important practical consideration in the broader adoption of this technique.

Several limitations of this study should be acknowledged. Although IPTW, PSM, and E-value analysis were used to reduce and evaluate confounding, residual confounding cannot be fully excluded. The primary outcome, postoperative in-hospital antitussive medication use, was identified from clinical orders and electronic medical records.

This outcome is relatively objective but remains a surrogate for cough burden and may be affected by prescribing preferences, institutional practice patterns, patient reporting, clinical thresholds for antitussive therapy, and discharge timing. Thus, medication use may not fully capture the incidence or severity of postoperative cough. Moreover, cough-related secondary outcomes were based on follow-up reports, which may have underestimated transient symptoms and limited standardized assessment of cough duration.

In addition, given the single-center design, extrapolation of these findings should take into account institutional expertise and perioperative practice patterns, which may affect the implementation and reproducibility of nonintubated techniques.

Conclusion

Overall, compared with DLT anesthesia, NITS with preserved spontaneous breathing was associated with lower in-hospital antitussive medication use and fewer reports of chronic cough at 8 weeks. Prospective multicenter studies with standardized cough assessments may help clarify these associations.

Authors’ contributions

All authors contributed to the conception and design of the study. SYD, QYJ, and YQZ contributed to defining the intellectual content, conducted the literature search, and prepared the manuscript. QYJ, XXZ, and QYH acquired the data.

SYD, QYJ, and SXY analyzed the data. All authors participated in editing and critically revising the manuscript. SYD and YQZ served as guarantors.

All authors reviewed and approved the final manuscript and take responsibility for the integrity of the work.

Ethical statement

The study protocol was approved by the Institutional Ethics Committee of University-Town Hospital of Chongqing Medical University (approval No. IIT 2025-173; approval date: January 29, 2026).

Declaration of Helsinki (for human research)

The study was conducted in accordance with the Declaration of Helsinki.

Participation consent

Patient consent was waived due to the retrospective design and anonymized data.

Conflicts of interest

There are no conflicts of interest.

Data availability

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Conversion factors for intravenous opioids to intravenous morphine-equivalent dose

The morphine-equivalent dose indicates the equivalent amount (in mg) of intravenous morphine for one unit of the specified opioid administered intravenously.

Supplementary eMethods Contents

eMethod 1: Standardized protocol for non-intubated thoracic anesthesia with preserved spontaneous ventilation

Patients in the nonintubated thoracic anesthesia group were generally managed according to our institution’s standardized protocol, which aimed to preserve spontaneous ventilation whenever clinically feasible. Further details of this anesthetic approach have been described previously.[12345]

Institutional protocol for nonintubated thoracic anesthesia

Adequate analgesia was maintained while minimizing intravenous analgesics and avoiding or limiting neuromuscular blocking agents whenever feasible. After entering the thoracic cavity, spontaneous ventilation was gradually encouraged. Anesthetic agents were titrated intraoperatively to regulate the rate and amplitude of spontaneous breathing in accordance with surgical needs, thereby maintaining an optimal operative field.

If significant mediastinal movement or deep, forceful respirations interfered with surgical exposure, small supplemental doses of muscle relaxants – rocuronium (3-20 mg) or cisatracurium (1-5 mg) – were administered to suppress spontaneous breathing and facilitate low–tidal-volume controlled ventilation.

1. Indications

2. Contraindications

3. Anesthetic induction and intraoperative management

eMethod 2: E-value for unmeasured confounding

E-value calculation for protective associations (RR < 1)

The E-value quantifies the minimum strength of association, on the risk ratio (RR) scale, that an unmeasured confounder would need to have with both the exposure and the outcome, conditional on measured covariates, to fully explain away an observed association.[1] For associations expressed on the RR scale, the E-value is defined as Equation (S1):

Equation (S1) (for RR > 1):

When the observed association is protective (RR < 1), the reciprocal of the RR is first taken (RR∗=1/RR), and the E-value is then calculated using the same formula:

This transformation ensures that the E-value is always greater than or equal to 1 and allows a symmetric interpretation of robustness to unmeasured confounding for both harmful and protective associations.

In addition to the point estimate, the E-value was also calculated for the limit of the 95% confidence interval closest to the null (RR = 1), which reflects the minimum confounding strength required to attenuate the association to nonsignificance. For protective associations, this corresponds to the upper bound of the confidence interval. If this bound includes or exceeds 1, the E-value for the confidence limit is defined as 1.

Approximation of the risk ratio from the odds ratio for common outcomes

The primary analysis estimated the association using odds ratios (ORs) derived from logistic regression models. Because the primary outcome was common (incidence approximately 30%), the odds ratio cannot be directly interpreted as a risk ratio. Therefore, following the recommendation of VanderWeele and Ding for common outcomes, we approximated the risk ratio using the square-root transformation, as Equation (S2):[12]

This approximation provides a conservative and model-independent estimate of the RR that does not rely on assumptions regarding the baseline risk in the unexposed group and has been proposed specifically for sensitivity analyses using the E-value framework. The same transformation was applied to the confidence interval limits of the OR to obtain corresponding approximate RR bounds, which were then used to calculate the E-value for the confidence limit closest to the null.

Application to the present study

For the primary outcome, nonintubated thoracic anesthesia was associated with reduced odds of postoperative in-hospital antitussive prescription use (OR: 0.46; 95% CI: 0.25–0.85). Using the square-root approximation, the corresponding risk ratio was estimated as RR ≈ 0.677. After converting the protective effect to the risk ratio scale (RR* =1/RR), the E-value for the point estimate was 2.32, and that for the confidence limit closest to the null was 1.39.

Related articles

PubMed

Sex-based disparities in whooping cough prevalence in Sierra Leone: Evaluating World Health Organization Health Equity Assessment Toolkit data.

Abstract Background This study examines the trends and sex-based inequality in the prevalence of whooping cough in Sierra Leone, from 2000 to 2019. Methods We utilized national health records extracted from the WHO Health Equity Assessment Toolkit database to analysed age-standardized prevalence per 100 000 population and employed inequality measures: Difference (D), Population Attributable Fraction (PAF), Population Attributable Risk (PAR) and Ratio (R) to examine sex-based inequalities. R...

PubMed

Prevalence of Suspected Obstructive Sleep Apnea Among Patients With Chronic Cough and Relatively Low BMI.

Background/Aim Recent studies have suggested an association between chronic cough and obstructive sleep apnea (OSA). This study aimed to estimate the prevalence of suspected OSA and describe associated clinical characteristics among Japanese adults with chronic cough. Patients and Methods Outpatients with chronic cough were recruited from three Japanese institutions between March 2023 and February 2025. Eligible participants were ≥20 years old with a cough lasting >8 weeks. Overnight home sle...