The study was approved by the ethics committee of Henan provincial people's hospital and was performed in accordance with the Declaration of Helsinki (Supplemental Fig. 1). The guardians of the included children agreed to participate in the study and signed the written consents.

The data of the study were collected from the medical records of Henan provincial people's hospital. Briefly, all patients admitted to the hospital were told that their medical records might be used anonymously for scientific research. If the patients or their families agreed, the written informed consent forms would be signed. After the patients were discharged from the hospital, their medical records would be included in the scientific research database.

A total of 1000 children with common pneumonia, and 1000 children with severe pneumonia were randomly included from the database, and all of them were admitted to the hospital between January 2010 and December 2020 due to pneumonia. Their age was limited to 18 to 60 months.

Pneumonia was diagnosed and classified using the criteria of the World Health Organization (WHO), which included: a history of cough or difficulty breathing, and if present, respiratory rate > 50 breaths/minute (age < 12 months) or > 40 breaths/minute (age ≥ 12 months).11 Severe pneumonia indicated that the disease met the above diagnostic criteria and was accompanied by at least one of the danger signs, which included (1) nasal flaring, (2) subcostal retractions, (3) convulsions, (4) Blantyre coma scale < 5, (5) “alert, voice, pain, unresponsive” scale less than “alert” and (6) inability to drink or feed. Common pneumonia was defined as meeting the above diagnostic criteria without the danger signs. Children with tuberculosis infection, congenital lung diseases, dysplasia disorders and other serious diseases were excluded. All the children with pneumonia were treated according to the protocolized guidelines for the diagnosis and treatment of pneumonia in China and the world.11

Other 1000 healthy children were randomly included from the database, and all of them came to the hospital for consultation at the same period, but no obvious disease was detected. They served as controls. Their ages ranged from 18 to 60 months.

Research data were obtained from the database using predefined questionnaires. For the children with pneumonia, age, gender, ethnic group, height, weight, household income, clinical manifestations (i.e., hyperpyrexia, dehydration, diarrhea and hypoxemia), serological data (i.e., serum white blood cell (WBC) count, creatinine level and CRP) and use of antibiotics were collected. For the controls, age, gender, ethnic group, height, weight and household income were collected.

Hyperpyrexia was a body temperature greater than 39°C. Hypoxaemia was defined as SpO2 < 90%. High creatinine referred to plasma creatinine levels greater than 35 mmol/l in infants and greater than 62 mmol/l in children.

Generally, obesity is assessed using the body mass index (BMI), which is calculated based on weight and height. However, children under 60 months of age are in a rapid growth and development phase. To increase the comparability, BMI Z-scores were calculated in the study to standardize the data, and its formula was (BMI - mean of BMI) / (standard deviation of BMI).12 The mean and standard deviation of BMI for the local population were obtained from the local Center for Disease Control and Prevention. A BMIZ score greater than 2.00 was considered overweight, a score greater than 3.00 was diagnosed as obesity, and the score between −2.00 and 2.00 was defined as normal.12

The serum levels of several lipids, lipoproteins and apolipoproteins were adopted to reflect lipid metabolism. The lipid metabolic markers included triglycerides (TG), total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), very low-density lipoprotein cholesterol (VLDL-C), high-density lipoprotein cholesterol (HDL-C), apolipoprotein A1 (ApoA1), apolipoprotein B (ApoB) and lipoprotein A (Lp-a).

These data were also collected from the database, and the blood specimen collection and testing process was briefly described below. When each child presented to the hospital, his or her fasting venous blood was routinely collected to measure a range of serological markers (such as blood lipids) in order to guide the diagnosis and treatment. The blood sample was immediately sent to the laboratory center of the hospital. After centrifugation and extraction, the serum sample was used to measure the levels of these lipid metabolic markers mentioned above. Serum levels of TG, TC, LDL-C, VLDL-C and HDL-C were detected using an ADVIA-1650 Chemistry System (Bayer Co., USA). Serum levels of ApoA1, ApoB and Lp-a were determined by several latex-particle-enhanced immunonephelometric assays on a BN ProSpec nephelometer (Dade Behring Co., Germany). Each serological marker was measured twice, and their average value was recorded in the database.

According to the Shapiro-Wilk test, all continuous variables in this study did not follow a normal distribution. Therefore, these variables were expressed by medians (25th quartiles, 75th quartiles), while the differences between groups were compared by Kruskal-Wallis H test. Categorical variables were expressed by frequencies, and the difference of the two categorical variables was measured using the Chi-square test. If there were three categorical variables to be compared at the same time, the maximum and minimum values were compared. The linear association of two continuous variables was determined using spearman correlation analysis. A P-value less than 0.05 was considered statistically significant. The Association of the lipids, lipoproteins and apolipoproteins with the risk of childhood pneumonia and hypoxemia was determined using univariate and multivariate logistic regression. The odds ratios (ORs) and 95% confidence intervals (CIs) were reported. In the multivariate analyses, demography data, clinical manifestations, serological data and use of antibiotics were adjusted. If a 95%CI did not include value one, it was considered statistically significant. All analyses were conducted using SPSS 23.0 (USA).

As shown in Table 1, compared with the common case group, the children in the severe case group had more opportunities to suffer from hyperpyrexia, dehydration, diarrhea, hypoxemia, high creatinine, high CRP (P < 0.001, P < 0.001, P < 0.001, P < 0.001, P < 0.001, P < 0.001), to receive the cephalosporins treatment (P < 0.001) and to change the antibiotics (P < 0.001). Compared with the severe case group, the children in the common case group had more possibilities to receive the macrolides treatment (P <  0.001). In addition, there was no difference in WBC count between the common case and severe case groups (P = 0.106). There was no difference in age, gender, and nationality, BMI Z-score, and household income among all the three groups (P = 0.826, P = 0.267, P = 0.214, P = 0.080, P = 0.199). Among them, household income indicated average monthly income per capita.

As shown in Table 2, with the exception of Lp-a (P = 0.433), the serum levels of the other lipid metabolic markers differed significantly in the control, common case, and severe case groups (P < 0.001, P <  0.001, P < 0.001, P < 0.001, P < 0.001, P < 0.001, P = 0.014). Briefly, compared with the common case group, the serum levels of TG, TC, LDL-C, and VLDL-C were significantly higher and the serum levels of HDL-C and ApoA1 were significantly lower in the severe case group (P < 0.05). Compared with the control group, the serum level of TC was significantly higher in the common case group (P < 0.05).

As shown in Table 3, the higher level of TC in serum was associated with the increased risk of common pneumonia (OR = 1.110, 95%CI = 1.014 ∼ 1.214). In addition, the other lipid metabolic markers were not related to the risk of the disease (Table 3).

Besides, the higher levels of TG, TC, LDL-C, VLDL-C and ApoB were associated with the elevated risk of severe pneumonia (OR = 1.407, 95%CI = 1.336 ∼ 1.480; OR = 1.947, 95%CI = 1.741 ∼ 2.175; OR = 1.153, 95%CI = 1.116 ∼ 1.189; OR = 1.310, 95%CI = 1.222 ∼ 1.404; OR = 1.075, 95%CI = 1.003 ∼ 1.151). On the contrary, the lower levels of HDL-C and ApoA1 were related to the increased risk of severe pneumonia (OR = 0.903, 95%CI = 0.873 ∼ 0.933; OR = 0.921, 95%CI = 0.891 ∼ 0.952). In addition, the serum of Lp-a was not associated with the risk of the disease (OR = 1.012, 95%CI = 0.979 ∼ 1.046).

As shown in Table 4, the higher level of TG in serum was associated with the risk of hypoxemia in this group (OR = 1.172, 95%CI = 1.028 ∼ 1.336). However, the serum levels of the other lipid metabolic markers were not related to the risk of hypoxemia in the common case group (Table 4).

Similar to the common case group, only the higher level of TG in serum was relate to the increased risk of hypoxemia in the severe case group (OR = 1.142, 95%CI = 1.072 ∼ 1.215). And the serum levels of the other lipid metabolic markers were not related to the risk of hypoxemia in this group (Table 4).

As shown in supplemental Table 1, in the common case group, the serum level of HDL-C was inversely associated with the serum level of C-reactive protein (ρ = −0.097, P = 0.002). In the severe case group, the serum level of HDL-C was also inversely associated with the serum level of C-reactive protein (ρ = −0.343, P < 0.001), and the serum level of TC was positively correlated with the serum level of the inflammatory marker (ρ = 0.064, P = 0.042).