Association between preoperative COVID-19 and major postoperative pulmonary complications: a multicentre observational cohort study in China

Abstract

Objective

To evaluate the relationship between preoperative COVID-19 infection and major postoperative pulmonary complications (PPC) risk after major elective surgeries during the Omicron wave.

Design

A multicentre, prospective, observational cohort study.

Setting

Four tertiary medical centres in Beijing, China.

Participants

All adult patients who underwent major elective surgeries under general anaesthesia from 30 December 2022 to 18 May 2023 were screened for eligibility. A total of 3211 patients were included.

Primary and secondary measures

The primary outcome was 30-day major PPC, defined as pneumonia, acute respiratory distress syndrome or unexpected postoperative ventilation. The secondary outcomes included length of hospital stay (LOS), reoperation and mortality.

Results

Major PPC occurred in 3.5% of patients with preoperative COVID-19 and 3.3% of those without. Inverse probability of treatment weighting-adjusted analysis showed no significant association between preoperative COVID-19 within 12 weeks and PPC risk (adjusted OR, 0.89; 95% CI 0.69 to 1.13). However, multivariable analysis revealed that COVID-19 infection within 3 weeks was independently associated with an increased PPC risk (OR, 3.44; 95% CI 1.37 to 8.68). Cardiothoracic surgery (OR, 12.47; 95% CI 8.11 to 19.17) and longer duration of surgery (OR, 1.24 per hour; 95% CI 1.13 to 1.37) were significant risk factors. In the cardiothoracic subgroup, PPC risk was significantly elevated within 7 weeks of infection. No significant differences were observed in LOS, reoperation rates or mortality between patients with and without preoperative COVID-19 infection.

Conclusions

Preoperative COVID-19 infection within 12 weeks was not associated with an increased overall risk of major PPC during the Omicron wave. Although very short infection-to-surgery intervals and cardiothoracic surgery showed exploratory signals of higher risk, these findings should be interpreted cautiously and support an individualised approach to perioperative risk assessment.

Trial registration number

ChiCTR2200067250.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• This multicentre observational study included a large cohort of patients undergoing major elective surgery, enhancing the generalisability of the findings.

• Inverse probability of treatment weighting was used to balance baseline covariates between patients with and without preoperative COVID-19 infection, reducing confounding inherent to observational data.

• Sensitivity and exploratory subgroup analyses were conducted to assess the robustness of the results across different assumptions and patient subgroups.

• Residual confounding from unmeasured variables, such as baseline pulmonary function or perioperative medication use, cannot be completely excluded.

Introduction

The COVID-19 pandemic has had a profound impact on global healthcare systems, significantly influencing the management of elective surgeries.¹ Amid overwhelming numbers of COVID-19 cases, many elective procedures were postponed^{2 3} due to concerns about the adverse effects of preoperative infection on postoperative outcomes,^{4 5} particularly postoperative pulmonary complications (PPC).

Early in the pandemic, guidelines recommended postponing elective surgeries for at least 7 weeks after a COVID-19 diagnosis to reduce the risks of postoperative complications and mortality.⁶ This guidance was supported by studies showing that a longer interval between COVID-19 diagnosis and surgery significantly reduced the risk of adverse postoperative outcomes.^{7 8} However, some research suggests that the impact of COVID-19 on postoperative risks diminishes shortly after infection.^{9 10} As the pandemic progresses, especially with the emergence of new variants such as Omicron, the effects of SARS-CoV-2 infection on surgical outcomes remain insufficiently explored.

First identified in late 2021, the Omicron variant has demonstrated reduced virulence compared with earlier SARS-CoV-2 variants.¹¹ Its clinical presentation is less severe,¹² raising questions about whether previous guidelines and findings on postoperative risks are applicable. The Omicron BF.7 subvariant caused a rapid and widespread surge in Beijing^{13 14} from December 2022 to early 2023, following the relaxation of public health measures. This unique outbreak timeline offers a valuable opportunity to assess the impact of Omicron infections on postoperative complications in real-world settings.

This study aimed to explore the relationship between preoperative Omicron infection and PPC following major elective surgery. Using data from a large multicentre cohort, this research sought to provide valuable insights into the optimal timing of elective surgeries in the context of Omicron. Additionally, the findings aim to guide future surgical and public health strategies, as the global healthcare community continues to navigate the evolving challenges posed by the pandemic.

Methods

Study design and ethics

This multicentre, prospective, observational cohort study was conducted in four tertiary medical centres in Beijing, China: Peking University Third Hospital, Peking University First Hospital, Capital Medical University Xuanwu Hospital and Peking University People’s Hospital. The study protocol was approved by the Institutional Ethical Committee of the Peking University Third Hospital (IRB00006761-M2022852) in December 2022 and registered at www.chictr.org.cn (ChiCTR 2200067250). Written informed consent was obtained from all participants prior to enrolment. Data collection took place from 30 December 2022 to 18 May 2023. This report adhered to the Strengthening the Reporting of Observational Studies in Epidemiology guidelines for observational studies.

Participants

Eligible participants were adult inpatients scheduled for major elective surgeries⁸ during the study period. Surgical procedures included endoscopic or open hepatectomy, gastrectomy, colorectal resection, liver and pancreatic resections, mastectomy, prostatectomy, adrenalectomy, nephrectomy, urethrectomy, hysterectomy, lung resection, oesophagectomy, laryngectomy, hip replacement, knee replacement, laminectomy, spinal fusion, coronary artery bypass grafting, heart valve replacement, elective open or endovascular repair of abdominal aortic aneurysms, brain tumour resection and carotid endarterectomy (definitions are provided in online supplemental appendix). The exclusion criteria were an estimated survival of less than 24 hours or refusal to participate.

Data collection

Patient demographics (age, sex and body mass index) and medical information (preoperative laboratory results, imaging records, comorbidities and chronic medications) were extracted from the electronic health records. Comorbidities were identified using International Classification of Diseases, 10th Revision, Clinical Modification (ICD-10-CM) codes, and included diabetes, hypertension, coronary artery disease, congestive heart failure, asthma, chronic obstructive pulmonary disease, chronic bronchitis, chronic kidney disease, severe liver disease, stroke history and depression.

The COVID-19 vaccination status was retrieved from an online Health Kit system. Data on previous SARS-CoV-2 infection, including the date of a positive Reverse Transcription (RT)-PCR or rapid antigen test result and associated symptoms, were collected during the preoperative visit. COVID-19 severity was categorised as mild, moderate, severe or critical according to the 10th Chinese Diagnosis and Treatment Protocol for SARS-CoV-2 Infection (online supplemental appendix).¹⁵

Surgical details, including the American Society of Anesthesiologists (ASA) physical status classification, type of anaesthesia (general, regional or combined) and surgery duration, were recorded using the anaesthesia information management system. Postoperative outcomes were obtained from hospitalisation records, outpatient follow-ups and phone calls conducted 30 days after surgery. All data were securely uploaded to the Research Electronic Data Capture.

Handling of missing data

Missing data were handled using multiple imputation techniques, with 20 imputed datasets generated using chained equations to preserve the statistical power and minimise bias. The imputation models included all baseline covariates and outcome variables to ensure the accuracy of imputation.

Outcomes

Patients were categorised as having preoperative COVID-19 if they tested positive for SARS-CoV-2 via RT-PCR or rapid antigen test within 12 weeks before surgery. The interval between the COVID-19 diagnosis and surgery was calculated from the date of the first positive test to the day of surgery.

The primary outcome was 30-day major PPC, defined as pneumonia, acute respiratory distress syndrome (ARDS) or unexpected postoperative ventilation (online supplemental appendix).¹⁶ The secondary outcomes included length of hospital stay (LOS), 30-day reoperation rate and 30-day mortality.

Statistical analysis

Statistical analyses were conducted using SPSS (V.22.0, IBM Corp.) and R software (V.4.2.3, R Foundation for Statistical Computing). All tests were two-sided, and a P value<0.05 was considered statistically significant. The normality was assessed using the Kolmogorov–Smirnov test. Continuous variables are presented as mean±SD or median with IQR, and categorical variables are presented as counts (percentages). Categorical variables were analysed using the χ², continuity-corrected χ², or Fisher’s exact test. Continuous variables were analysed using t-tests or Mann-Whitney U tests depending on the data distribution.

The propensity score-based inverse probability of treatment weighting (IPTW) was applied to reduce confounding due to baseline differences between with and without COVID-19 infection. Propensity scores were estimated using a logistic regression model including pretreatment baseline covariates potentially related to COVID-19 exposure or postoperative outcomes, including age, sex, body mass index, ASA physical status classification, smoking history, major comorbidities, surgical category, anaesthetic technique and COVID-19 vaccination status. Covariate balance before and after weighting was assessed using standardised mean differences. Associations between preoperative COVID-19 infection and postoperative outcomes were estimated using IPTW-adjusted regression models, with results reported as adjusted ORs or differences with 95% CIs. To explore potential thresholds of the infection-to-surgery interval associated with differences in major PPC risk, maximally selected rank statistics were applied in an exploratory manner. Test statistics were calculated across candidate cut-off points, and the maximally selected statistic was determined while accounting for multiple testing using the method of Hothorn and Lausen.¹⁷ Given the absence of consistent cut-offs from prior studies or consensus and to align with the clinical data, the infection-to-surgery cut-offs were determined in a data-driven manner.

Prespecified sensitivity analyses were conducted to assess the robustness of the primary findings. These included analyses stratified by infection-to-surgery interval, COVID-19 severity and presence of COVID-19-related symptoms on the day of surgery. Additional sensitivity analyses addressing surgical case mix and anaesthetic technique were performed by excluding orthopaedic surgery, restricting analyses to procedures performed under general anaesthesia only, and excluding orthopaedic surgery performed under regional or combined anaesthesia. All sensitivity analyses were conducted using the same IPTW-adjusted regression framework as the primary analysis.

To identify factors independently associated with major PPC, multivariable logistic regression analyses were performed using the original, unweighted cohort. All clinically relevant preoperative comorbidities were included in the multivariable model a priori to control for potential confounding. Exploratory subgroup analyses were additionally conducted in patients undergoing cardiothoracic surgery to assess the association between preoperative COVID-19 infection and major PPC.

Sample size calculation

The sample size was calculated using PASS (NCSS Statistical Software, LLC, Kaysville, UT, USA). Based on an anticipated major PPC rate of 3.0% in non-COVID-19 patients¹⁶ and 6.1% in COVID-19 patients,⁷ corresponding to an expected absolute risk difference of approximately 3% and an OR of approximately 2.0, a total sample size of 3000 participants was estimated to achieve 90% power, with a two-sided α error of 0.05.

Results

Baseline patient characteristics

The flowchart of the study is shown in figure 1. Between December 2022 and May 2023, 4534 patients were assessed for eligibility. Of these, 481 were excluded: 1 patient had an estimated survival of less than 24 hours, 22 refused to participate and 458 did not meet the inclusion criteria. A total of 4030 patients were enrolled, of which 53 were lost to follow-up, leaving 3977 patients for the final analysis.

Figure 1. Flow chart for cohort recruitment including details on inclusion and non-inclusion criteria.

Preoperative COVID-19 infection was observed in 3211 patients (80.7%). Most cases were mild (2652, 82.6%), while 552 (17.2%) were categorised as ordinary and only 7 (0.2%) as severe or critical. The interval between COVID-19 diagnosis and surgery varied: 335 patients (10.4%) underwent surgery within <4 weeks, 1218 (37.9%) between 4 and 8 weeks and 1648 (51.3%) between 8 and 12 weeks. At the time of surgery, 178 patients (5.5%) with prior COVID-19 remained symptomatic. The majority (3579, 90.0%) were vaccinated against SARS-CoV-2, with 3035 (76.3%) receiving more than three doses. The baseline characteristics of the patients are summarised in table 1. Weighted baseline characteristics are shown in online supplemental table 1; most covariates achieved good balance after weighting.

Table 1. Baseline characteristics of the study patients (n=3977).

	No preoperative COVID-19 (n=766)	Preoperative COVID-19 (n=3211)	P value
Age (year), n (%)			<0.001*
<65	423 (55.2)	2235 (69.6)
≥65	343 (44.8)	976 (30.4)
BMI (kg/m2), median (IQR)	25.3 (22.9, 27.7)	24.9 (22.7, 27.3)	0.010*
Sex, n (%)			0.173†
Male	362 (47.2)	1425 (44.4)
Female	404 (52.8)	1786 (55.6)
Comorbidities, n (%)
Diabetes	143 (18.7)	502 (15.6)	0.046†
Hypertension	334 (43.6)	1174 (36.6)	<0.001†
Coronary artery disease	56 (7.3)	239 (8.3)	0.423†
Stroke	21 (2.7)	48 (1.5)	0.026†
Severe liver disease	4 (0.5)	20 (0.6)	0.949†
Chronic kidney disease	3 (0.4)	15 (0.5)	1.000†
Smoke, n (%)			0.002†
Never smoked	582 (76.0)	2623 (81.7)
Formerly smoked	92 (12.0)	299 (9.3)
Currently smoke	92 (12.0)	289 (9.0)
Vaccination against COVID-19, n (%)			<0.001†
0	94 (12.3)	306 (9.5)
1–2	97 (12.7)	443 (13.9)
≥3	575 (75.1)	2459 (76.6)
ASA, n (%)			0.035†
I	107 (14.0)	438 (13.7)
II	578 (75.9)	2317 (72.4)
III	75 (9.8)	422 (13.2)
IV	2 (0.3)	23 (0.7)
Type of anaesthesia, n (%)			<0.001†
Regional anaesthesia	124 (16.2)	275 (8.6)
General anaesthesia	642 (83.8)	2935 (91.4)
Duration of surgery (hour), median (IQR)	121.0 (89.0, 181.8)	120.0 (80.0, 179.0)	0.007*
Type of surgery, n (%)			<0.001†
Gynaecological surgery	61 (8.0)	393 (12.2)
Orthopaedic surgery	492 (64.2)	1317 (41.0)
Urological surgery	47 (6.1)	454 (14.1)
General surgery	65 (8.5)	424 (13.2)
Neurosurgery	34 (4.4)	211 (6.6)
Cardiothoracic surgery	67 (8.7)	344 (10.7)

Values are n (%) unless otherwise indicated.

^*Mann-Whitney U test.

^†χ2 test.

ASA, American Society of Anesthesiologists classification; BMI, body mass index.

Primary outcome

The overall incidence of 30-day major PPC was 3.4% (136/3977). Major PPC included pneumonia in 2.7% (106/3977), acute respiratory failure in 0.9% (37/3977) and unexpected ventilation in 0.6% (26/3977). Major PPC occurred in 111 patients (3.5%) with preoperative COVID-19 and in 25 (3.3%) without (OR, 0.89; 95% CI 0.69 to 1.13; p=0.327). After IPTW adjustment, preoperative COVID-19 within 12 weeks was not significantly associated with in-hospital and 30-day major PPC risk (in-hospital: adjusted OR, 0.92; 95% CI 0.70 to 1.21; p=0.543; 30-day: adjusted OR, 0.81; 95% CI 0.64 to 1.03; p=0.085; table 2). The incidence of unexpected ventilation was significantly higher in the COVID-19 group compared with the non-COVID group (adjusted OR, 2.33; 95% CI 1.22 to 4.78; p=0.014; table 2).

Table 2. Primary and secondary outcomes.

	No COVID-19 (n=731)	COVID-19 (n=3039)	Adjusted OR or difference	95% CI	P value
30-day major PPC, n (%)	24 (3.3)	103 (3.9)	0.81	0.64 to 1.03	0.085
Pneumonia	21 (2.8)	92 (3.0)	0.87	0.67 to 0.99	0.270
ARDS	5 (0.7)	30 (1.0)	0.70	0.45 to 1.08	0.106
Unexpected ventilation	3 (0.4)	22 (0.7)	2.33	1.22 to 4.78	0.014
In-hospital PPC, n (%)	18 (2.5)	82 (2.7)	0.92	0.70 to 1.21	0.543
In-hospital LOS (day), median (IQR)	7 (6, 9)	7 (5, 9)	−0.10	−0.54 to 0.33	0.645
Postoperative LOS (day), median (IQR)	5 (4, 6)	4 (3, 6)	−0.02	−0.45 to 0.41	0.926
30-day reoperation, n (%)	6 (0.8)	18 (0.6)	0.67	0.39 to 1.15	0.152
30-day mortality, n (%)	2 (0.27)	1 (0.03)	0.41	0.08 to 1.63	0.234

Values are shown as median (IQR) or number (%).

Adjusted ORs or differences with 95% CIs and P values were estimated using regression models weighted by inverse probability of treatment weighting (IPTW).

ARDS, acute respiratory distress syndrome; LOS, length of stay; PPC, postoperative pulmonary complications.

Identification of optimal infection-to-surgery intervals

Maximally selected rank statistics were applied to explore potential thresholds for the infection-to-surgery interval in relation to differences in major PPC risk. In the overall cohort, the standardised Wilcoxon statistic reached its global maximum at approximately 8–9 weeks after COVID-19 infection in exploratory analyses, while the curve also showed local increases around 3 and 6 weeks (online supplemental figure S1). Based on these exploratory patterns, infection-to-surgery intervals of 0–3 weeks and 3–6 weeks were subsequently examined for their impact on major PPC in sensitivity analyses. Similarly, in the cardiothoracic surgery subgroup, local increases were observed around 7 and 9 weeks (online supplemental figure S2). Accordingly, subgroup-specific intervals of 0–7 weeks and 7–9 weeks were examined in exploratory subgroup analyses.

Sensitivity analyses

Sensitivity analyses are summarised in online supplemental table 1. No significant association was observed between preoperative COVID-19 infection and major PPC when stratified by infection-to-surgery interval, including infection within 3 weeks (adjusted OR, 1.48; 95% CI 0.71 to 2.74; p=0.251) or within 6 weeks before surgery (adjusted OR, 1.22; 95% CI 0.88 to 1.67; p=0.224), compared with patients without COVID-19. Similarly, COVID-19 patients with or without symptoms on the day of surgery did not show an increased risk of major PPC compared with non-COVID-19 patients (asymptomatic: adjusted OR, 0.67; 95% CI 0.26 to 1.42; p=0.351; symptomatic: adjusted OR, 0.73; 95% CI 0.26 to 1.51; p=0.419). Mild or moderate infection was also not significantly associated with major PPC risk (adjusted OR, 0.95; 95% CI 0.75 to 1.21; p=0.689). In patients with preoperative COVID-19, receiving more than three doses of the vaccine was not significantly associated with a reduced risk of major PPC risk (adjusted OR, 0.61; 95% CI 0.35 to 1.15; p=0.100).

Additional sensitivity analyses addressing surgical case mix and anaesthetic technique showed consistent results. Excluding orthopaedic surgery (adjusted OR, 0.86; 95% CI 0.61 to 1.21; p=0.388), restricting analyses to procedures performed under general anaesthesia only (adjusted OR, 0.88; 95% CI 0.69 to 1.12; p=0.296), and excluding orthopaedic surgery performed under regional or combined anaesthesia (adjusted OR, 0.89; 95% CI 0.69 to 1.13; p=0.330) did not materially change the findings.

Secondary outcomes

Secondary outcomes are summarised in table 2. After IPTW adjustment, there were no significant differences in in-hospital LOS and postoperative LOS between patients with and without preoperative COVID-19 infection (in hospital LOS: adjusted difference, −0.10 days; 95% CI −0.54 to 0.33; p=0.645; in postoperative LOS: adjusted difference, −0.02 days; 95% CI −0.45 to 0.41; p=0.926). The 30-day reoperation rate did not differ significantly between the groups (0.8% vs 0.6%; adjusted OR, 0.67; 95% CI 0.39 to 1.15; p=0.152). Thirty-day mortality was rare in the study cohort, with two deaths in the COVID-19 group and one death in the non-COVID-19 group. The overall 30-day mortality was comparable between the two groups (adjusted OR, 0.41; 95% CI 0.08 to 1.63; p=0.234).

Factors associated with major PPC

Multivariable analysis identified several factors significantly associated with major PPC (online supplemental table 3; figure 2): preoperative COVID-19 within <3 weeks (OR, 3.44; 95% CI 1.37 to 8.68; p=0.009), longer surgery duration (OR, 1.24 per hour; 95% CI 1.13 to 1.37; p<0.001), and cardiothoracic surgery (OR, 12.47; 95% CI 8.11 to 19.17; p<0.001). Exploratory subgroup analyses were performed to examine the association between very short infection-to-surgery intervals and PPC risk, as well as to assess potential effect modification by surgical category. These analyses were based on smaller sample sizes and should be interpreted with caution due to the increased susceptibility to residual confounding. In the cardiothoracic subgroup, the major PPC risk was significantly higher in patients with preoperative COVID-19 <7 weeks than in those without COVID-19 (< 7 weeks: OR, 2.45; 95% CI 1.07 to 5.62, p=0.034). However, the risk was not elevated for surgery conducted 7–9 weeks or >9 weeks after infection (7–9 weeks: OR, 1.98; 95% CI 0.83 to 4.72, p=0.122; >9 weeks: OR, 0.47; 95% CI 0.19 to 1.17; p=0.103; figure 3). Given the limited sample size within this subgroup, these findings should be interpreted cautiously.

Figure 2. Multivariable logistic regression assessing the association between preoperative COVID-19 and major postoperative pulmonary complications. ASA, American Society of Anesthesiologists.

Figure 3. Risk of major postoperative pulmonary complications in patients with preoperative COVID-19 undergoing cardiothoracic surgery.

Discussion

In this multicentre observational study conducted during the Omicron wave of the COVID-19 pandemic, we evaluated the association between the interval of preoperative SARS-CoV-2 infection to major elective surgery and the risk of major PPC. Using an IPTW-adjusted analytical framework, we found that preoperative COVID-19 infection within 12 weeks was not associated with an increased risk of major PPC at the population level. The outcome definition of our study was intentionally restricted to major PPC, including pneumonia, ARDS and unexpected postoperative ventilation. By design, we did not adopt broader composite PPC definitions proposed by standardised frameworks such as the StEP-COMPAC Group (Standardised Endpoints in Perioperative Medicine - Core Outcome Measures for Perioperative and Anaesthetic Care),¹⁸ which include minor or transient respiratory events. Our findings suggest that, in the context of the Omicron variant, recent COVID-19 infection does not uniformly confer excess major PPC risk after surgery. Sensitivity analyses further supported the robustness of the primary findings. When stratified by infection-to-surgery intervals, including within 3 weeks or within 6 weeks before surgery, no statistically significant increase in major PPC risk was observed. A substantial proportion of patients underwent orthopaedic surgery, and some procedures were performed under regional or combined anaesthesia, which may be associated with a lower baseline risk of PPC. This case mix could have reduced the overall event rate and attenuated between-group differences. However, multiple sensitivity analyses excluding orthopaedic procedures, restricting the cohort to general anaesthesia only, and excluding orthopaedic surgery under regional/combined anaesthesia yielded consistent results, supporting the robustness of the primary findings. Similarly, analyses stratified by symptom status on the day of surgery, and COVID-19 severity yielded consistent results.

Studies conducted during the early outbreak of COVID-19, which was dominated by more virulent SARS-CoV-2 variants, reported substantially elevated rates of PPC^{7 8} and mortality.^{19 20} These observations formed the basis of early international recommendations advocating the postponement of elective surgery after SARS-CoV-2 infection.21,23 In contrast, accumulating evidence from studies conducted during the Omicron wave suggests that the perioperative risk associated with prior COVID-19 infection has reduced compared with the variant during early outbreak.^{9 24 25} In studies of major PPC, our study’s rate of 3.5% aligns with the 3.78% reported in a Chinese retrospective cohort²⁶ and the 3.4% reported in a French prospective cohort study²⁷ during the Omicron wave. This temporal shift in perioperative risk is likely multifactorial. The Omicron variant has been shown to exhibit lower intrinsic virulence and to cause milder clinical disease compared with earlier SARS-CoV-2 variants.^{11 12} In parallel, the widespread implementation of COVID-19 vaccination has changed the epidemiological context of SARS-CoV-2 infection.^{28 29}

In the present study, a high proportion of patients had received three or more vaccine doses; however, the study was not designed to evaluate the causal effect of vaccination on postoperative outcomes, and vaccination status should therefore be interpreted as a background characteristic of the study population rather than a direct explanatory factor. Furthermore, following the early easing period of COVID-19 control measures in China, elective surgical decision-making was relatively cautious. Patients of advanced age or with significant comorbidities may have preferentially deferred surgery during this period, resulting in a degree of self-selection toward a relatively healthier surgical cohort.

Notably, the IPTW-adjusted primary analysis estimated the average effect of preoperative COVID-19 at the population level and did not demonstrate an increased overall risk. In contrast, multivariable logistic regression performed in the original cohort identified surgery within 3 weeks of COVID-19 diagnosis as an independent risk factor for major PPC, alongside cardiothoracic surgery and longer operative duration. This discrepancy likely reflects a localised high-risk subgroup effect that may not substantially influence the population-average association. Therefore, the observed association for very short infection-to-surgery intervals should be interpreted as hypothesis-generating rather than definitive evidence mandating uniform surgical delay.

The cardiothoracic surgery subgroup further underscores the importance of contextual risk assessment. Cardiothoracic procedures are inherently associated with higher baseline pulmonary risk and greater susceptibility to PPC.^{2 30 31} In this subgroup, an increased risk of major PPC was observed when surgery was performed within 7 weeks of COVID-19 infection. Given the intrinsically high pulmonary risk of cardiothoracic surgery and the smaller sample size of this subgroup, these findings should be interpreted cautiously, but they suggest that infection-to-surgery interval may have differential implications in high-risk surgical populations.

Taken together, our findings support a more individualised approach to perioperative risk assessment in patients with prior COVID-19 infection, rather than a uniform postponement strategy. Although the overall population-level analysis did not demonstrate an increased risk of major PPC within 12 weeks after infection, the potential signal observed in very short infection-to-surgery intervals underscores the need for careful patient selection, particularly for high-risk surgical procedures. These results align with the direction of guideline updates that emphasise individualised risk assessment over fixed waiting periods.^{22 23} While the pandemic has officially ended, these insights may still inform perioperative decision-making in patients recovering from infectious diseases and offer methodological considerations for surgical planning in future epidemic settings.

This study has several limitations. First, as a multicentre observational study, it cannot establish definitive causal relationships between preoperative COVID-19 infection and PPC risk. Although extensive adjustment was performed using IPTW, the possibility of residual confounding should be acknowledged. Unmeasured factors, such as baseline pulmonary function, perioperative steroid use, nutritional status or other clinical variables, may still have influenced PPC risk. Second, the reliance on clinical records may have resulted in incomplete ascertainment of prior SARS-CoV-2 infection, particularly asymptomatic cases during periods of reduced mandatory testing, leading to potential misclassification bias. Third, heterogeneity in the interval between COVID-19 infection and surgery may have introduced additional bias that could not be fully addressed. Fourth, although vaccination status was considered as a baseline characteristic, variations in vaccine types, dosing schedules and individual immune responses were not captured, and the study was not designed to assess the causal effect of vaccination on postoperative outcomes. Fifth, the analysis focused on short-term outcomes, precluding insights into long-term recovery and quality of life. Sixth, differences in hospital criteria for scheduling elective surgeries during the Omicron wave may have introduced selection bias, as patients with milder symptoms were more likely prioritised. Finally, the study’s regional and temporal context, influenced by the characteristics of China’s healthcare system and public health policies during the Omicron period, may limit the generalisability of these findings to other settings, SARS-CoV-2 variants or future respiratory pandemics.

Conclusion

In this multicentre cohort study conducted during the Omicron wave, preoperative COVID-19 infection within 12 weeks was not associated with an increased overall risk of major PPC. This finding was consistent across primary and sensitivity analyses and supports the safety of proceeding with surgery after appropriate risk assessment in most patients. Exploratory subgroup analyses suggested a potential increase in risk among patients undergoing surgery within very short infection-to-surgery intervals and those undergoing cardiothoracic procedures; however, these findings were based on smaller subgroups and should be interpreted cautiously as hypothesis-generating rather than definitive. Overall, our results support an individualised approach to perioperative risk assessment rather than uniform surgical delay policies. Further prospective studies are warranted to confirm these subgroup findings, to refine risk stratification strategies and to better account for residual case-mix differences and unmeasured confounding.

Data availability statement

Data are available upon reasonable request.