A clinical prediction model for rapidly differentiating pulmonary tuberculosis from community acquired pneumonia in children | Italian Journal of Pediatrics

Patients and definitions

We performed a retrospective study about children with a discharge diagnosis of PTB and community acquired pneumonia (CAP) who were initially admitted to the Respiratory Department of Tianjin Children’s Hospital between January 2018 and December 2023. All patients were randomly divided into a modeling group and a validation group at a 7:3 ratio based on established methodologies for clinical prediction model [4], as shown in Fig. 1. The distinction between the TB group and the non-TB group is based on whether they have microbiologic or pathologic confirmation. Children were defined as those under the age of 18 years old in this study. All the children completed the following detection within 24 h of admission: MP-DNA detected by MP polymerase chain reaction (PCR) tests by nasopharyngeal swab, phlegm respiratory pathogen eight (influenza A and influenza B, respiratory syncytial virus, adenovirus, metapneumovirus, parainfluenza virus type 1, 2, 3), blood respiratory pathogen IgM antibody nine (eosinophilic lung Legionella bacteria, Mycoplasma pneumoniae, Chlamydia, Rickettsia, adenovirus pneumonia, syncytial virus, influenza A virus, influenza B virus, and parainfluenza), tuberculosis and bacterial culture. This report was approved by the Ethics Committee of the Children’s Hospital of Tianjin (Approval Number: 2022-LXKY-013), in accordance with the ethical standards of the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Written informed consents were obtained from the parents for publication of this report. All patients’ data were analyzed anonymously.

The inclusion criteria for the PTB group and CAP group were as follows:

Diagnostic criteria for PTB in our study

The diagnostic criteria for PTB are as follows [3, 5]:

1. 1)

   Having symptoms and signs related to pulmonary tuberculosis.

2. 2)

   Imaging is consistent with active pulmonary tuberculosis.

3. 3)

   Tuberculin skin test (TST) and/or interferon-gamma release assays (IGRAs) positive.

4. 4)

   Phlegm, induced sputum, bronchoalveolar lavage fluid, gastric juice, pleural effusion, tissue samples, etc. are positive for acid fast staining, MTB culture, molecular biology testing for MTB nucleic acid, or pathological results are positive.

Diagnostic criteria for CAP in our study

The diagnostic criteria for CAP are as follows [6]:

1. 1)

   Newly appeared cough, sputum or worsening of pre-existing respiratory symptoms, accompanied by purulent sputum, with or without chest pain.

2. 2)

   Fever.

3. 3)

   Signs of lung consolidation and/or consolidation or hear wet sounds.

4. 4)

   White blood cell count > 10 × 10⁹/L or < 4 × 10⁹/L, with or without left shift of cell nucleus.

5. 5)

   Chest X-ray display film localized infiltrative shadows or interstitial changes, with or without pleural effusion; (1) Add clause (5) to any of clause (4), excluding tuberculosis, non-infectious pulmonary interstitial diseases, pulmonary edema and other diseases. All selected children were hospitalized, with a course of illness of ≤ 7 days at admission, and pneumonia occurring after 48 h of hospitalization was excluded.

The exclusion criteria were as follows: ① PTB group combined with other pathogenic infections. ② CAP group with positive TB microbiological testing. ③ The age was more than 18 years old.

Data collection

The clinical data of all children were collected as following: (1) basic information: name, gender, age, weight, full term, spontaneous labor, countryside, previous BCG vaccination, TB contact history; (2) clinical manifestations: fever present, fever duration, maximum body temperature, light fever after noon, basic disease, night sweat, cough, wheeze, hemoptysis, chest pain, abdominal pain, blood-stained sputum, weight loss, illness duration.

Weight loss can be defined in any of the following ways [7]:

(1) Reported noticeable weight loss (2) Very low weight or underweight.

(3) Confirmed weight loss since last visit (4) Flattening of growth curve.

History of TB contact [7]:

1. 1)

   Close contact with a person with TB at home.

2. 2)

   Contact may be with a person with TB from outside the household (e.g., carer, grandparent, relative) with whom the child has had frequent contact • In older children and adolescents, contact with a source case is often outside the household, such as at school or in the neighbourhood.

3. 3)

   A source case with bacteriologically confirmed PTB is more likely to infect contacts than cases with bacteriologically negative PTB.

4. 4)

   Treatment regimen and treatment response of the source case should be determined.

5. 5)

   If no source case is identified, any contact with a person with chronic cough should always be investigated; that person should be assessed for possible TB.

6. 6)

   Timing of contact: children usually develop TB within 12 months following exposure; most (90%) develop TB within 6 months.

Children were considered to have basic disease if they had any of the following: asthma/reactive airway disease, bronchopulmonary dysplasia, cancer, cerebral palsy, cystic fibrosis, diabetes, Down syndrome, eczema, food allergies, gastroesophageal reflux disease, genetic/metabolic disorders, heart disease, immunodeficiency, intellectual disability/developmental delay, kidney disease, liver disease, neurological disorders, other blood disorders, sickle cell disease/trait, seizure disorder, transplant recipient, and/or other conditions [8].

Statistical analysis

Statistical analyses were performed using SPSS 24.0. Data with a normal distribution were expressed as mean ± standard deviation (x ± s), and comparisons were made using the independent sample t-test. Kolmogorov-Smirnov test was used to check the normal distribution. Data that did not follow a normal distribution were described by median (P25, P75), with comparisons conducted using the Mann–Whitney U-test. The chi-square test was applied to categorical data. Logistic regression was used to identify significant risk factors, which were then utilized to construct a model and visualize it with a nomogram. The nomogram was assessed for differentiation, calibration, and clinical net benefit. The model’s performance was evaluated using the Receiver Operating Characteristic (ROC) curve and the area under the curve (AUC). Calibration was checked with a calibration plot, and clinical benefit was determined through decision curve analysis in both training and validation datasets. Statistical significance was set at P < 0.05.