Abstract
Background
The recent resurgence of Mycoplasma pneumoniae (MP) pneumonia has increased the number of children with lobar pneumonia who may receive flexible bronchoscopy (FB), yet the decision to perform FB remains subjective, and no quantitative tool exists for this specific high-risk subgroup. This study aimed to develop and internally validate a nomogram to predict FB utilization in children with MP-related lobar pneumonia using routine admission indicators to support real-time bedside decision-making.
Methods
This single-center retrospective study included 345 children (aged 1 month–18 years) with serology-confirmed MP-related lobar pneumonia (January 2022–March 2025); those with congenital heart disease, immunodeficiency, or plastic bronchitis were excluded. Patients were divided into FB (n=171) and non-FB (n=174) groups based on whether FB was performed. Laboratory predictors were restricted to values obtained within 72 hours of admission and before the FB decision. FB was recommended if symptoms or signs persisted after ≥72 hours of standard therapy. Multivariate logistic regression was used to identify predictors, and a nomogram was constructed and internally validated by 10-fold cross-validation. Model performance was assessed by area under the curve (AUC), calibration, and decision curve analysis (DCA).
Results
Viral coinfection [odds ratio (OR) =23.271, 95% confidence interval (CI): 9.299–64.900, P<0.001] was the strongest independent risk factor for FB utilization, while bacterial coinfection (OR =0.215, 95% CI: 0.088–0.507, P<0.001) and male sex (OR =0.146, 95% CI: 0.061–0.313, P<0.001) were independent protective factors. The nomogram incorporating these six variables demonstrated moderate discrimination (apparent AUC =0.741, 95% CI: 0.689–0.792; cross-validated AUC =0.705) and good calibration (Hosmer-Lemeshow P=0.56). At the optimal cutoff value of 0.56, the model had a sensitivity of 51.5% and a specificity of 85.1%. DCA demonstrated net clinical benefit over a “treat-all” strategy within a threshold probability range of 0.1–0.3.
Conclusions
The nomogram showed modest ability to predict FB use in this single-center cohort, but its clinical utility for identifying children with lobar pneumonia caused by MP who truly require FB remains uncertain.
Keywords: Mycoplasma pneumoniae (MP), children, flexible bronchoscopy (FB), nomogram, predictive model
Highlight box.
Key findings
• In children with lobar pneumonia caused by Mycoplasma pneumoniae (MP), viral coinfection was a strong risk factor for flexible bronchoscopy (FB) intervention [odds ratio (OR) =23.271], while bacterial coinfection (OR =0.215) and male sex (OR =0.146) were protective factors. A nomogram including viral/bacterial coinfection, sex, age, fever duration, and cough duration showed moderate discrimination (area under the curve =0.741) and calibration (P=0.56), with net benefit over “treat-all” at threshold probabilities 0.1–0.3.
What is known and what is new?
• MP pneumonia (MPP) can be severe. FB helps clear airway mucus plugs and improves outcomes in refractory patients. Recent studies have developed prediction models for bronchoscopy-related outcomes in MPP, but none targeted FB utilization in lobar MPP using routine indicators.
• We developed and validated a nomogram predicting FB need. Viral coinfection was the strongest predictor (OR >23); bacterial coinfection and male sex showed unexpected protective effects.
What is the implication, and what should change now?
• The nomogram offers a quantitative decision tool to identify children at high risk of FB during MP epidemics, supporting resource allocation and reducing unnecessary procedures. In this cohort, viral coinfection was strongly associated with FB use; however, these findings should be interpreted cautiously and require external validation before influencing bronchoscopy decisions. Early antibiotics may reduce FB in bacterial coinfections; viral testing may aid risk stratification but requires validation in prospective studies.
Introduction
Mycoplasma pneumoniae (MP) is one of the most common causes of community-acquired pneumonia in children aged ≥5 years (1). The MP pneumonia (MPP) epidemic has spread every few years worldwide (2). There has been a sharp increase in the number of MPP cases globally since 2023 due to the relaxation of coronavirus disease 2019 (COVID-19) pandemic restrictions (2,3). During the epidemic, many children with MPP experienced rapid progression, characterized by persistent high fever, intensive coughing, and imaging findings of lobar pneumonia or even “white lung”. These patients often receive flexible bronchoscopy (FB) intervention (4). The outbreak of MPP in children has led to a relative shortage of medical resources in hospitals. To meet the medical needs of children, those with severe conditions and those likely to receive FB intervention should be given priority for hospitalization. Children with lobar pneumonia are among the potential candidates for FB intervention. Lobar pneumonia caused by MP in children is characterized by extensive inflammation involving entire or multiple lung lobes (5). Timely intervention is critical to prevent these patients from progressing to severe MPP, necrotizing pneumonia, or plastic bronchitis (PB). In addition to appropriate antibiotics, steroids and FB are also used in severely ill children with MPP (6). FB and bronchoalveolar lavage (BAL) play important roles in the diagnostic and therapeutic benefits of severe or refractory MPP. They facilitate the clearance of airway mucus plugs and enable the collection of BAL fluid for pathogen identification (7). However, FB is an invasive procedure associated with potential complications, prolonged hospitalization, and increased costs (8). Therefore, careful selection of patients for FB is needed to avoid overtreatment. Current guidelines recommend that FB and BAL are warranted in patients with severe MPP associated with mucus plug obstruction or PB to reduce complications and sequelae (7). However, the decision to perform FB often remains subjective and experience-based, especially in children who present with lobar pneumonia without overt respiratory failure. Several recent studies have developed prediction models for BAL in children with MPP, focusing on refractory MPP, pulmonary consolidation, and PB (9-11). These models offer valuable insights but none directly predict FB use in our target population. Most focus on complications, require specialized biomarkers, or estimate treatment benefit rather than FB likelihood. Given the absence of a universal benchmark for FB referral in this subgroup, a model predicting FB utilization under standardized institutional protocols offers practical value. To our knowledge, no model has yet been developed specifically to predict FB use in children with lobar MPP using routine clinical indicators for real-time bedside decision-making. Therefore, this study aimed to develop and internally validate a nomogram model based on clinical indicators to predict FB utilization in children with lobar pneumonia caused by MP, providing a practical tool to assist in bedside decision-making. We present this article in accordance with the TRIPOD reporting checklist (available at https://tp.amegroups.com/article/view/10.21037/tp-2026-0453/rc).
Methods
Study design and participants
We conducted a single-center, retrospective cohort study of children with lobar pneumonia caused by MP who were hospitalized at Guangdong Women and Children Hospital from January 1, 2022 to March 1, 2025. The study was registered under Clinical Trial Number MR-44-25-044601. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of Guangdong Women and Children Hospital (No. 20251037). Informed consent was waived, as the data were analyzed anonymously.
Inclusion and exclusion criteria
Inclusion criteria: (I) age 1 month to 18 years; (II) manifestations of fever, cough and pulmonary lobe consolidation on chest radiographs or low-dose computed tomography (CT) scans; and (III) MP DNA detected in nasopharyngeal secretions or BAL fluid by polymerase chain reaction (PCR), next-generation sequencing (NGS), or an MP antibody titer of ≥1:160 (12).
Exclusion criteria: congenital heart disease, motor developmental delay, immunodeficiency, MP-induced PB or necrotic pneumonia, and incomplete medical records.
BAL under FB
Written informed consent for the FB procedure was obtained from the parents. After routine preparation, patients received oxygen (1–2 L/min) and intravenous midazolam (0.2 mg/kg). A flexible bronchoscope (Olympus BF-XP290 or BF-P290) was inserted nasally, with topical 1% lidocaine anesthesia. BAL was performed in the most affected subsegmental bronchus, and fluid was collected for microbiological testing. All procedures were performed by experienced pediatric endoscopists. All the children received at least 3 days of standard anti-infective therapy; FB was recommended if symptoms or pulmonary signs did not improve.
Variables and data collection
Data extracted from electronic records included: demographics (sex, age), clinical presentation (fever/cough duration before admission), radiological findings (number of consolidated lobes, pleural effusion, atelectasis), laboratory indices [white blood cell count (WBC), platelet count (PLT), C-reactive protein (CRP), procalcitonin (PCT), alanine aminotransferase (ALT), aspartic acid transferase (AST), lactate dehydrogenase (LDH)], treatment (azithromycin, doxycycline, steroids), coinfections (viral/bacterial by nucleic acid test or culture), and comorbidities (allergic rhinitis/sinusitis). The decision to perform FB was made by the attending pediatric pulmonologist in consultation with the bronchoscopy team, following a standardized institutional protocol. Indications for FB included: (I) refractory hypoxemia [peripheral oxygen saturation (SpO2) <92% despite oxygen therapy]; (II) persistent lobar consolidation on chest imaging after ≥72 hours of standard anti-infective therapy; or (III) clinical deterioration (worsening respiratory distress or fever curve) despite treatment. Imaging findings (chest radiographs or low-dose CT) were reviewed jointly by pediatric radiologists and pulmonologists to assess the extent of consolidation, atelectasis, and potential mucus blockage. FB was deferred if parental consent was withheld or if resources (bronchoscopy slot, endoscopy team availability) were unavailable at the time of the decision, which may have introduced nonclinical selection bias. To minimize the risk of information leakage, all laboratory predictor values were restricted to those obtained within 72 hours of admission and before the FB decision—typically after at least 72 hours of standard anti-infective therapy. For each patient, we used the most abnormal values within this pre-decision window (highest for CRP, PCT, and WBC; lowest for PLT).
Selection of candidate predictors
Candidate predictors were selected based on clinical plausibility and availability at the time of the FB decision, informed by causal considerations from directed acyclic graph (DAG) analysis. Variables measured after the FB decision or lying on the treatment pathway between baseline disease characteristics and FB were excluded. Steroid use exemplifies such an excluded variable for two reasons. First, our DAG analysis identified steroid use as a mediator of the causal pathway from baseline inflammatory markers to FB intervention, including blocking the true effects of upstream baseline predictors and yielding biased coefficients. Second, steroid use typically occurs after clinical severity assessment, making it unsuitable as a predictor for early decision-making.
Statistical analysis
The statistical analysis was conducted via SPSS software 22.0 (SPSS Inc., Chicago, IL, USA) and R software (version 4.5.3; R Foundation for Statistical Computing, Vienna, Austria). Continuous variables with non-normal distribution were presented as median [interquartile range (IQR)] and compared between groups using the Mann-Whitney U test. Categorical variables were expressed as numbers and percentages (n, %) and analyzed with the Chi-squared test or Fisher’s exact test where appropriate. DAGs were used as a complementary tool to guide variable selection. All the DAG-identified variables were entered into the multivariate model regardless of statistical significance, to preserve clinical interpretability. While DAG analysis is more commonly used in causal inference, we adopted it here to enhance the clinical interpretability of the model and to avoid adjusting for mediators that would bias coefficient estimates. Multicollinearity was assessed by the variance inflation factor (VIF) (VIF <5). A nomogram was constructed using independent predictors from the multivariate model. Model performance was assessed by discrimination [area under the curve (AUC) with 10-fold cross-validation], calibration (Hosmer-Lemeshow test and calibration curves), and clinical utility [decision curve analysis (DCA)]. The events-per-variable (EPV) ratio was calculated as the number of FB events divided by the number of candidate predictors in the multivariate model. Patients with incomplete medical records were excluded (complete-case analysis). The proportion of missing data was low (12/345, 3.5%), making complete-case analysis appropriate and unlikely to introduce substantial bias. The sample size was determined by the number of eligible patients treated at Guangdong Women and Children Hospital during the study period. Post-hoc analysis confirmed adequate statistical power, with an EPV ratio of 28.5, which exceeded the recommended minimum of 10. A two-sided P<0.05 was considered statistically significant.
Results
Baseline characteristics of the study population
Among the 3,396 children with MPP, 851 received FB. Of these, 368 children were diagnosed with lobar pneumonia caused by MP. After patients with PB (n=5), necrotic pneumonia (n=2), congenital diseases (n=1), immunodeficiency (n=1), motor development retardation (n=2), or incomplete data (n=12) were excluded, 345 children with lobar pneumonia were enrolled. They were divided into FB group (n=171) and non-FB group (n=174) based on whether FB was performed (Figure 1). All patients recovered without major complications. Compared with the non-FB group, the FB group had a significantly shorter cough duration before admission (6.00 vs. 7.00 days, P<0.001), higher rates of pleural effusion (7.02% vs. 1.72%, P=0.02), steroid use (52.05% vs. 39.08%, P=0.02), viral coinfection (40.94% vs. 20.11%, P<0.001), and allergic rhinitis/sinusitis (9.94% vs. 4.02%, P=0.03). Laboratory findings showed lower PLT (273.00×109/L vs. 318.00×109/L, P<0.001), higher CRP (18.48 vs. 10.48 mg/L, P<0.001), and higher PCT (0.15 vs. 0.09 ng/mL, P<0.001) in the FB group. No significant differences were observed in sex, age, fever duration, number of consolidated lobes, atelectasis, WBC, ALT, AST, LDH, antibiotic regimens, or bacterial coinfection (P>0.05). Detailed baseline data are shown in Table 1.
Figure 1.
Flow chart for patient selection. FB, flexible bronchoscopy; MP, Mycoplasma pneumoniae; PB, plastic bronchitis.
Table 1. Clinical and laboratory differences between the FB and non-FB groups.
| Characteristics | FB (n=171) | Non-FB (n=174) | Pvalue |
|---|---|---|---|
| Sex (males) | 78.00 (45.61) | 87.00 (50.00) | 0.42 |
| Age (years) | 6.00 (4.00–8.00) | 6.00 (4.00–8.00) | 0.39 |
| Clinical presentation | |||
| Fever duration before admission (days) | 6.00 (5.00–7.00) | 6.00 (2.75–8.00) | 0.89 |
| Cough duration before admission (days) | 6.00 (4.00–7.00) | 7.00 (5.00–10.00) | <0.001*** |
| Pulmonary radiology | |||
| Numbers of lobes consolidated | 1.00 (1.00–1.00) | 1.00 (1.00–2.00) | 0.22 |
| Pleural effusion | 12.00 (7.02) | 3.00 (1.72) | 0.02* |
| Atelectasis | 1.00 (0.58) | 2.00 (1.15) | >0.99 |
| WBC (×109/L) (ref: 5×109/L–14×109/L) | 7.53 (6.02–9.43) | 6.94 (5.59–9.71) | 0.17 |
| PLT (×109/L) (ref: 190×109/L–520×109/L) | 273.00 (227.00–342.00) | 318.00 (259.00–402.00) | <0.001*** |
| CRP (mg/L) (ref: 0–6 mg/L) | 18.48 (9.59–33.30) | 10.48 (4.76–20.61) | <0.001*** |
| PCT (ng/mL) (ref: <0.05 ng/mL) | 0.15 (0.10–0.27) | 0.09 (0.05–0.14) | <0.001*** |
| ALT (U/L) (ref: 0–50 U/L) | 14.00 (11.00–18.00) | 14.00 (12.00–19.00) | 0.49 |
| AST (U/L) (ref: 17–59 U/L) | 32.00 (28.00–37.00) | 31.00 (26.75–37.00) | 0.20 |
| LDH (U/L) (ref: 120–246 U/L) | 314.00 (273.00–371.00) | 313.00 (268.83–354.03) | 0.36 |
| Treatment | |||
| Azithromycin | 149.00 (87.13) | 145.00 (83.33) | 0.32 |
| Switching to doxycycline from azithromycin | 14.00 (8.19) | 12.00 (6.90) | 0.65 |
| Doxycycline | 8.00 (4.68) | 13.00 (7.47) | 0.28 |
| Steroids | 89.00 (52.05) | 68.00 (39.08) | 0.02* |
| Viral coinfection | 70.00 (40.94) | 35.00 (20.11) | <0.001*** |
| Bacterial coinfection | 18.00 (10.53) | 21.00 (12.07) | 0.65 |
| Allergic rhinitis/sinusitis | 17.00 (9.94) | 7.00 (4.02) | 0.03* |
Data are presented as median (IQR) or n (%). *, P<0.05; ***, P<0.001. ALT, alanine aminotransferase; AST, aspartic acid transferase; CRP, C-reactive protein; FB, flexible bronchoscopy; IQR, interquartile range; LDH, lactate dehydrogenase; PCT, procalcitonin; PLT, platelet; ref, reference; WBC, white blood cell.
Independent factors for FB intervention
DAG analysis was used to derive the minimal sufficient adjustment set for each exposure. The results showed that viral coinfection required adjustment for age and sex, whereas bacterial coinfection, fever duration, and cough duration had no backdoor paths to FB. Multivariate logistic regression including these six variables (viral coinfection, bacterial coinfection, fever duration, cough duration, age, and sex) showed no multicollinearity (VIF <4). Viral coinfection was an independent risk factor for FB intervention [odds ratio (OR) =23.271, 95% confidence interval (CI): 9.299–64.900, P<0.001], whereas bacterial coinfection (OR =0.215, 95% CI: 0.088–0.507, P<0.001) and male sex (OR =0.146, 95% CI: 0.061–0.313, P<0.001) were independent protective factors. Fever duration (OR =1.076, P=0.08), cough duration (OR =0.956, P=0.09), and age (OR =0.935, P=0.16) were not significant predictors (Table 2). The Hosmer-Lemeshow test indicated good model fit [χ2 =6.7908, degree of freedom (df) =8, P=0.56].
Table 2. Multivariate logistic regression analysis of factors associated with FB.
| Variables | B | OR (95% CI) | P value | VIF |
|---|---|---|---|---|
| Viral coinfection | 3.147 | 23.271 (9.299–64.900) | <0.001*** | 3.53 |
| Bacterial coinfection | −1.536 | 0.215 (0.088–0.507) | <0.001*** | 1.64 |
| Sex (males) | −1.924 | 0.146 (0.061–0.313) | <0.001*** | 2.89 |
| Fever duration before admission | 0.073 | 1.076 (0.993–1.167) | 0.08 | 1.07 |
| Cough duration before admission | −0.045 | 0.956 (0.905–1.007) | 0.09 | 1.03 |
| Age | −0.068 | 0.935 (0.850–1.025) | 0.16 | 1.10 |
***P<0.001. B, unstandardized regression coefficient; CI, confidence interval; FB, flexible bronchoscopy; OR, odds ratio; VIF, variance inflation factor.
Nomogram construction and validation
A nomogram was constructed to visualize the predictive model, integrating the six variables from the multivariate logistic regression (Figure 2A). The predicted probability of FB intervention can be calculated using the following logistic regression equation: P=1/(1 + exp−[0.415 + 3.147 × (viral coinfection) − 1.536 × (bacterial coinfection) + 0.073 × (fever duration before admission) − 0.045 × (cough duration before admission) − 0.068 × (age) − 1.924 × (male sex)]), where viral coinfection =1 if present (0 otherwise), bacterial coinfection =1 if present (0 otherwise), male sex =1 for male (0 for female), and fever duration before admission, cough duration before admission, and age are continuous variables measured in days and years, respectively. An Excel-based calculator is provided in Table S1 to facilitate clinical use at the bedside. The model showed acceptable discrimination for FB intervention, with an AUC of 0.741 (95% CI: 0.689–0.792) (Figure 2B). The optimal cutoff value was 0.56, yielding a sensitivity of 51.5% and a specificity of 85.1% for predicting FB utilization (Youden index =0.365). Internal validation via 10-fold cross-validation yielded a cross-validated AUC of 0.705 (95% CI: 0.651–0.760), indicating minimal optimism bias and stable discriminatory performance of the model. The calibration curve showed good agreement between the predicted probabilities and observed outcomes (Figure 2C). DCA showed that the nomogram had good clinical utility. At a threshold probability of 0.1–0.3, the cross-validated net benefit was 0.3766, compared with 0.3659 for the “treat all” strategy, indicating greater clinical benefit for guiding FB intervention in children with lobar MPP (Figure 2D).
Figure 2.
Nomogram for predicting FB intervention in children with lobar pneumonia caused by MP. (A) Clinical prediction nomogram incorporating viral/bacterial coinfection, fever/cough duration before admission, age, and sex. (B) ROC curve with an AUC of 0.741 and a cutoff value of 0.56, indicating model discrimination. (C) Calibration curve demonstrating good agreement between the predicted and observed FB probabilities. (D) DCA confirming the clinical net benefit of the nomogram. AUC, area under the curve; CV, cross-validation; DCA, decision curve analysis; FB, flexible bronchoscopy; MP, Mycoplasma pneumoniae; ROC, receiver operating characteristic.
Discussion
Lobar pneumonia caused by MP in children is characterized by extensive inflammation involving entire or multiple lung lobes (5). This study aimed to develop a nomogram to help clinicians identify children who are likely to receive FB intervention. Baseline comparisons showed that the FB group had higher rates of pleural effusion, viral coinfection, allergic rhinitis/sinusitis, and steroid use, along with lower PLTs and elevated CRP and PCT. DAG analysis identified six potential causal indicators: viral coinfection, bacterial coinfection, sex, age, fever duration, and cough duration. Multivariate analysis revealed that viral coinfection was a strong risk factor for FB (OR >23), whereas bacterial coinfection and male sex were protective factors. The nomogram incorporating these six variables demonstrated moderate discrimination (AUC 0.741; cross-validated AUC 0.705) and good calibration, as supported by both the Hosmer-Lemeshow test and the calibration curve. The threshold probability range of 0.1–0.3 reflects the clinical trade-off between failing to perform FB in children with rapid progression and performing unnecessary invasive procedures. The modest net benefit over treat-all is clinically meaningful, as even small selection improvements might reduce unnecessary procedures.
In our study, viral coinfection emerged as the strongest independent risk factor for FB intervention, with an OR exceeding 23. This finding aligns with previous reports showing that viral-MP coinfection occurs in 27–56% of children with MPP and is associated with a more severe clinical course, including prolonged fever, increased inflammatory burden, and a higher likelihood of airway mucus plug formation (13,14). Such coinfection may synergistically exacerbate airway epithelial damage and mucus hypersecretion, thereby increasing the likelihood of FB intervention (15). Many viruses, such as parainfluenza virus, influenza virus, rhinovirus, adenovirus, and respiratory syncytial virus (RSV), have been detected in patients with MPP (16). MP-adenovirus coinfection is most common in children with refractory MPP and severe community-acquired pneumonia, which can aggravate the condition of children and most patients have lung consolidation and/or atelectasis, and require FB intervention (17,18). Therefore, patients with lobar pneumonia caused by MP and viral coinfection may have a higher likelihood of receiving FB intervention. However, this strong association should be interpreted with caution, as differential testing intensity between the FB and non-FB groups may have inflated the detection rate of viral coinfection in the FB group.
Bacterial coinfection, most commonly with Streptococcus pneumoniae or Haemophilus influenzae, is frequently observed in children with MPP (19). Recent studies have reported bacterial coinfection rates of approximately 20% in refractory MPP (14,20). However, its impact on disease severity remains debated. While some studies have reported that bacterial or viral-bacterial coinfections are associated with more severe disease and worse outcomes (12), others have reported that specific types of coinfections may influence treatment response rather than initial severity (21). In our study, bacterial coinfection was unexpectedly identified as an independent protective factor against FB (OR =0.215). Several hypotheses may explain this finding. First, children with bacterial coinfection may benefit from timely and effective antibiotic therapy. Moreover, azithromycin, the primary empirical antibiotic for MP, is active against common Gram-positive bacteria and may act synergistically with other antibiotics, such as cephalosporins, thereby enhancing bacterial clearance, reducing bacterial load and toxin production, and ultimately decreasing airway inflammation, mucus hypersecretion, and the likelihood of FB. In contrast, most respiratory viruses—such as adenovirus, influenza, and RSV—lack specific antiviral therapy, leading to prolonged virus-induced epithelial damage and mucus plug formation, which in turn increase the likelihood of FB intervention. Nevertheless, these hypotheses require validation in prospective cohorts with systematic treatment protocols to exclude potential treatment bias and residual confounding.
Previous studies did not report sex differences in patients with refractory MPP or severe MPP (22,23). In this study, male sex was identified as an independent protective factor, indicating that, after adjusting for confounders, female children in this cohort were more likely to receive FB. According to univariate analysis, the proportion of male patients did not differ significantly between the FB and non-FB groups. This finding was consistent with those of previous studies (24,25). However, after adjustment for confounders identified by the DAG (primarily viral coinfection and inflammatory severity), male sex emerged as an independent protective factor against FB intervention. This reversal indicates that the unadjusted association was confounded by more prevalent factors in male patients, such as viral coinfection, which strongly drives the likelihood of FB utilization. After accounting for these factors, male sex itself was associated with a lower likelihood of FB, possibly reflecting sex-specific differences in clinical presentation and treatment response. Future research should explore whether sex-specific inflammatory responses or differences in airway anatomy may affect the possibility of FB intervention in MPP patients.
Compared with previously published prediction models for bronchoscopy-related outcomes in MPP patients, our nomogram offers several practical advantages. Most prior models targeted specific complications or alternative procedures (BAL), whereas ours directly addresses the clinical decision of whether to perform FB (9-11). Moreover, our model relies solely on routine clinical and laboratory indicators, facilitating implementation even in resource-limited settings. Our model showed moderate discrimination (AUC 0.741; cross-validated 0.705), somewhat lower than that of some models that use specialized predictors (10,26). This likely reflects the inherent difficulty of predicting a procedure influenced by institutional practice, physician discretion, and parental preferences. Importantly, our model predicts FB utilization rather than objective necessity, a critical distinction given that FB decisions reflect local practice patterns. Currently, no universally accepted minimum performance standard exists for models guiding invasive procedures such as FB. Our tool is intended as a decision-support aid to complement clinical judgment, not as a definitive indication, and should be prospectively evaluated in real-world settings before widespread adoption.
This study has several limitations. First, this was a single-center retrospective study, which may introduce selection bias and limit generalizability. Second, the sample size was modest, particularly for subgroups such as those with pleural effusion or specific coinfections, which limited the statistical power to detect additional predictors. Third, although we restricted laboratory predictor measurements to values obtained within 72 hours of admission and before the FB decision, the retrospective design precluded strictly standardized sampling timing; some values may still have been obtained after clinical deterioration that influenced the FB decision. Fourth, the difference in testing intensity between groups may have introduced detection bias: children who underwent FB received BAL with more comprehensive pathogen testing, whereas non-FB patients were tested primarily via nasopharyngeal swabs, potentially overestimating the association of viral coinfection with FB utilization. Fifth, because FB utilization is influenced by non-clinical factors (physician discretion, parental consent, and resource availability), our model predicts local decision-making behavior rather than objective patient needs. Sixth, although DAG analysis is more commonly used in causal inference than in predictive modeling, we adopted it to avoid adjusting for mediators. Seventh, although we used bootstrapping for calibration, we did not apply optimism correction for the AUC. The minimal difference in the AUC (0.036) suggests that overfitting is negligible, and the model offers a pragmatic risk stratification tool pending prospective external validation.
Conclusions
We developed and internally validated a nomogram incorporating viral coinfection, bacterial coinfection, sex, age, fever duration, and cough duration to predict FB intervention in children with lobar pneumonia caused by MP. The model demonstrated acceptable discrimination and good calibration, offering a preliminary tool to optimize resource allocation and improve clinical decision-making during MP epidemics.
Supplementary
The article’s supplementary files as
Acknowledgments
The abstract of this work was presented at the 2026 Annual Meeting of the Guangdong Medical Association Pediatric Branch, held in Guangzhou, China, on July 10-12, 2026.
Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of Guangdong Women and Children Hospital (No. 20251037). Informed consent was waived, as the data were analyzed anonymously.
Footnotes
Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://tp.amegroups.com/article/view/10.21037/tp-2026-0453/rc
Funding: This work was supported by the Medical Scientific Research Foundation of Guangdong Province, China (No. B2026230).
Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tp.amegroups.com/article/view/10.21037/tp-2026-0453/coif). The authors have no conflicts of interest to declare.
Data Sharing Statement
Available at https://tp.amegroups.com/article/view/10.21037/tp-2026-0453/dss
References
- 1. Kutty PK, Jain S, Taylor TH, et al. Mycoplasma pneumoniae Among Children Hospitalized With Community-acquired Pneumonia. Clin Infect Dis 2019;68:5-12.. 10.1093/cid/ciy419 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Meyer Sauteur PM, Beeton ML, European Society of Clinical Microbiology and Infectious Diseases (ESCMID) Study Group for Mycoplasma and Chlamydia Infections (ESGMAC), and the ESGMAC Mycoplasma pneumoniae Surveillance (MAPS) study group . Mycoplasma pneumoniae: delayed re-emergence after COVID-19 pandemic restrictions. Lancet Microbe 2024;5:e100-1.. 10.1016/S2666-5247(23)00344-0 [DOI] [PubMed] [Google Scholar]
- 3. Li H, Li S, Yang H, et al. Resurgence of Mycoplasma pneumonia by macrolide-resistant epidemic clones in China. Lancet Microbe 2024;5:e515.. 10.1016/S2666-5247(23)00405-6 [DOI] [PubMed] [Google Scholar]
- 4. Yan C, Xue GH, Zhao HQ, et al. Current status of Mycoplasma pneumoniae infection in China. World J Pediatr 2024;20:1-4.. 10.1007/s12519-023-00783-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Zheng Y, Mao G, Dai H, et al. Early predictors of delayed radiographic resolution of lobar pneumonia caused by Mycoplasma pneumoniae in children: a retrospective study in China. BMC Infect Dis 2024;24:414.. 10.1186/s12879-024-09289-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Wei D, Zhao Y, Zhang T, et al. The role of LDH and ferritin levels as biomarkers for corticosteroid dosage in children with refractory Mycoplasma pneumoniae pneumonia. Respir Res 2024;25:266.. 10.1186/s12931-024-02892-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Wang YS, Zhou YL, Bai GN, et al. Expert consensus on the diagnosis and treatment of macrolide-resistant Mycoplasma pneumoniae pneumonia in children. World J Pediatr 2024;20:901-14.. 10.1007/s12519-024-00831-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Wu X, Lu W, Sang X, et al. Timing of bronchoscopy and application of scoring tools in children with severe pneumonia. Ital J Pediatr 2023;49:44.. 10.1186/s13052-023-01446-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Wu X, Lu W, Liu W, et al. Predictive value of an early comprehensive assessment model for refractory mycoplasma pneumoniae pneumonia and internal validation. BMC Infect Dis 2025;25:744.. 10.1186/s12879-025-11133-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Liang M, Zhang H, Li Y, et al. A prediction model for bronchoalveolar lavage in children with Mycoplasma pneumoniae pneumonia and consolidation. Sci Rep 2026;16:3152.. 10.1038/s41598-025-32941-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Wang P, Duan R, Wang Q, et al. Bronchoalveolar lavage cytokine profile and clinical features as risk predictors of plastic bronchitis in children with Mycoplasma pneumoniae pneumonia. Ital J Pediatr 2025;51:175.. 10.1186/s13052-025-02041-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Wang F, Wang YW, Wang JY, et al. Impact of pathogen co-detection on disease severity and clinical outcomes in children with Mycoplasma pneumoniae pneumonia. World J Pediatr 2026;22:536-45.. 10.1007/s12519-026-01053-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Wu X, Chen R, Mo D, et al. Analysis of Mycoplasma pneumoniae pneumonia and viral coinfection in southern Guangzhou after the COVID-19 pandemic. J Thorac Dis 2024;16:6789-98.. 10.21037/jtd-24-729 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Yuan L, Mingyue D, Zhou L. Analysis of the characteristics of mixed infections with Mycoplasma pneumoniae in children. Sci Rep 2025;15:9414.. 10.1038/s41598-025-94292-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Piao M, Liu N, Meng F, et al. Clinical analysis of Mycoplasma pneumoniae pneumonia combined with adenovirus infection in children with lobar pneumonia. Eur J Clin Microbiol Infect Dis 2025;44:2331-41.. 10.1007/s10096-025-05194-6 [DOI] [PubMed] [Google Scholar]
- 16. Huang L, Shi T. Co-Detection of ADV, Influenza B, and HPIV: Independent Risk Factors for SMPP with Changes in NPIs. Viruses 2025;17:1266.. 10.3390/v17091266 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Li F, Zhang Y, Shi P, et al. Mycoplasma pneumoniae and Adenovirus Coinfection Cause Pediatric Severe Community-Acquired Pneumonia. Microbiol Spectr 2022;10:e0002622.. 10.1128/spectrum.00026-22 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Zhou Y, Wang J, Chen W, et al. Impact of viral coinfection and macrolide-resistant mycoplasma infection in children with refractory Mycoplasma pneumoniae pneumonia. BMC Infect Dis 2020;20:633.. 10.1186/s12879-020-05356-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Li X, Pan L, Li Z, et al. Etiology and clinical characteristics of pathogen co-detection in pediatric Mycoplasma pneumoniae pneumonia: A multicenter retrospective study. Int J Infect Dis 2026;162:108230.. 10.1016/j.ijid.2025.108230 [DOI] [PubMed] [Google Scholar]
- 20. Li C, Rao J, Wang X, et al. Navigating the outbreak: a comprehensive analysis of pediatric Mycoplasma pneumoniae pneumonia via targeted next-generation sequencing in Wuhan, 2022-2023. Microbiol Spectr 2025;13:e0246324.. 10.1128/spectrum.02463-24 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Liang X, Wei R, Liang D, et al. Associations of Mycoplasma pneumoniae load, co-infections, and macrolide resistance with clinical-laboratory profiles in hospitalized pediatric pneumonia: a targeted next-generation sequencing study of bronchoalveolar lavage fluid. Ann Clin Microbiol Antimicrob 2026;25:30.. 10.1186/s12941-026-00870-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Chen C, Chen S, Yang C, et al. Risk factors for refractory Mycoplasma pneumoniae in Chinese children: a meta-analysis. Front Pediatr 2025;13:1512689.. 10.3389/fped.2025.1512689 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Yang S, Lu S, Guo Y, et al. A comparative study of general and severe mycoplasma pneumoniae pneumonia in children. BMC Infect Dis 2024;24:449.. 10.1186/s12879-024-09340-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Wang L, Xie Q, Xu S, et al. The role of flexible bronchoscopy in children with Mycoplasma pneumoniae pneumonia. Pediatr Res 2023;93:198-206.. 10.1038/s41390-021-01874-z [DOI] [PubMed] [Google Scholar]
- 25. Lu J, Zhang J, Wang G, et al. Effects of bronchoalveolar lavage on Mycoplasma Pneumoniae pneumonia: A propensity score matched-cohort study. Front Pediatr 2022;10:1066640.. 10.3389/fped.2022.1066640 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Jian X-q, Li M, Wang X-y. The predictive value of MUC5AC levels in the sputum of children with Mycoplasma pneumoniae pneumonia treated with fiberbronchoscopy. Microbiol Spectr 2025;13:e0010325.. 10.1128/spectrum.00103-25 [DOI] [PMC free article] [PubMed] [Google Scholar]