Background

Streptococcus pneumoniae (S. pneumoniae) is a leading cause of serious infectious diseases associated with significant morbidity and mortality worldwide [1]. S. pneumoniae typically colonises the pharynx and the upper respiratory tract of healthy individuals. However, when the host immune function is weak, the pathogen can cause invasive pneumococcal disease(IPD) which can be transmitted to various sterile sites, such as the blood, cerebrospinal fluid (CSF), pleural space, and peritoneal fluid [2]. The mortality rate of children with IPD is approximately 5.3-27.5% [3,4,5], with more than 1 million deaths recorded annually worldwide; moreover, approximately 25-50% of survivors develop serious neurological sequelae [6, 7]. A surveillance report showed that China had the largest number of patients with pneumococcal-associated diseases compared with 12% among ten African and Asian countries; furthermore, approximately 30,000 children younger than five years of age died from IPD in 2000 [7]. With the widespread application of pneumococcal conjugate vaccines (PCVs), the incidence of IPD significantly declined by 51% between 2000 and 2015 [8]. However, PCV vaccination is not routinely considered by governments of many countries, including China, and the immunisation rate is generally low. Consequently, the incidence and mortality rates of the disease are high in these countries.

The clinical signs and symptoms of IPD vary among different cases and are related to host immunity, bacterial virulence, and the infection site [9]. Moreover, the incidence of antibiotic resistance in S. pneumoniae has increased in recent years owing to the prevalence of multidrug-resistant organisms [10]. Moreover, vaccination and inappropriate antibiotic administration cause challenges in identifying patients with pneumococcal infections at an early stage. Previous studies have evaluated the clinical characteristics of paediatric patients with IPD and revealed several predictive variables associated with mortality, such as younger age, meningitis, and underlying diseases [11,12,13,14]. However, data on antimicrobial resistance and the risk factors for poor outcomes in children with IPD in China are limited.

In this study, we aimed to analyse the characteristics and antimicrobial resistance of pneumococcal strains and the risk factors for poor outcomes in paediatric patients with IPD in Hangzhou to identify a better strategy for reducing the incidence and mortality of IPD in China.

Methods

Study cohort

We performed a retrospective analysis using clinical data from the Pediatric Intensive Care (PIC) database [15], a large paediatric specific, single-center, bilingual database containing comprehensive clinical records between 2010 and 2018 from the Children’s Hospital of Zhejiang University School of Medicine, China [16]. A total of 178 paediatric patients with IPD and positive cultures of S. pneumoniae from the blood, CSF, pleural effusion, and other normally sterile body fluids were included in the study. Patients’ demographic details, vital signs, clinical examination results, and outcomes were documented. We analysed the first positive cultures obtained during the intensive care unit (ICU) stay of the enrolled patients. The Institutional Review Board of the Children’s Hospital of Zhejiang University School of Medicine (Hangzhou, China) reviewed and approved the study protocol. The requirement for individual patient consent was waived because of the retrospective observational nature of the study and the anonymisation of the medical records.

Data collection

Clinical data, including demographic characteristics such as age and sex and vital signs such as temperature, heart rate (HR), respiratory rate (RR), systolic blood pressure (SBP), diastolic blood pressure (DBP), laboratory values, and outcomes, were collected. Infection and microorganism-associated variables included the sample type with a positive culture and antimicrobial resistance testing. Other laboratory values were obtained from the first blood sample collected after ICU admission. The primary diagnosis was based on the International Statistical Classification of Diseases and Related Health Problems, 10th edition (ICD-10) system. Sepsis was defined according to the Sepsis 3.0 as previously described [17].

Antimicrobial resistance testing

Bacterial identification was performed using an automatic microorganism identification system (Vitek-2 Compact; bioMérieux, France). Antimicrobial resistance testing of S. pneumoniae was performed using the Vitek-2 AST-GP68 Test Kit according to the manufacturer’s instructions, as well as the E-test for penicillin resistance testing as supplementary experiments. The antibiotic susceptibility breakpoints for S. pneumoniae were determined at the time of detection in accordance with the latest edition of the guidelines of the Clinical and Laboratory Standards Institute at the time of detection.

Statistical analysis

All the statistical analyses were performed using R version 3.4.3 [18, 19]. Continuous variables with a normal distribution were tested using the unpaired Student’s t-test and are presented as the mean ± standard deviation (SD), whereas non-normally distributed continuous variables were tested using the Mann–Whitney U test and are presented as the median (interquartile range, IQR). Categorical variables were tested using chi-square analysis or Fisher’s exact test and are presented as the frequencies (proportions). Univariate and multivariate logistic regression models, shown as odds ratios (ORs) and 95% confidence intervals (CIs), were used to identify factors associated with poor outcomes. Statistical significance was defined as a two-sided P-value of < 0.05.

Results

1. Epidemiology and clinical characteristics

Over the 8 years from 2010 to 2018, 13,449 patients were admitted to the ICUs of the Children’s Hospital of Zhejiang University School of Medicine, and 178 patients (1.32%) had an IPD. Characteristics of the patients with and without IPDs are shown in Supplementary Table 1. Among the patients with IPD, 100 (56.18%) were male and 78 (43.82%) were female, with an average age of 22.63 (IQR 10.41–54.83) months. Among them, 2 (1.12%) patients were newborns (≤ 28 days), 51 (28.65%) patients were aged 29 days to 1 year, 118 (66.29%) patients were aged 2 to 10 years, and the remaining 7 (3.93%) patients were older than 11 years. A seasonal trend was observed, with 61.8% of all cases diagnosed during winter and spring, between December and May. The surgical intensive care unit (SICU) had the greatest number of patients with IPD (35.96%) while the cardiac intensive care unit (CICU) contained younger patients with IPD aged ≤ 1 year. The most frequent cause of ICU admission was cardiovascular disorder (30.34%). The respiratory tract (pleural effusion, 87.64%) was the most common site for positive S. pneumoniae culture, whereas newborns seemed more susceptible to bloodstream infections. Almost half of the patients with IPD (43.26%) used vasopressors during their ICU stay. The average length of ICU stay was 1.96 (IQR 0.91–6.14) days, and the average length of hospital stay was 13.85 (IQR 8.65–22.85) days. In total, 61 (34.27%) patients had sepsis, and seven patients died, with an in-hospital mortality rate of 3.93% (Table 1).

Table 1 Clinical and demographic characteristics of included patients in different age groups
Full size table

2. Antibiotic resistance

As shown in Table 2, antibiotic sensitivity tests were conducted with isolates from 163 patients (91.57%). The resistance (resistant + intermediate, R + I) of the S. pneumoniae isolates were highest for erythromycin (98.62%, 143 of 145), followed by tetracycline (90.06%, 145 of 161), and SMZ-Co (84.57%, 137 of 162). Penicillin resistance occurred in eight (72.73%) isolates from patients with meningitis, whereas penicillin resistance occurred in only six (4.05%) isolates from patients without meningitis. In addition, the resistance rates to ceftriaxone and cefotaxime were 45.45% and 54.55%, respectively, among isolates from patients with meningitis and 22.30% and 25.00%, among isolates from patients without meningitis. All isolates were sensitive to vancomycin, linezolid, moxifloxacin, telithromycin, ofloxacin, and levofloxacin. The number of multidrug-resistant (MDR) isolates (those resistant to ≥ three antibiotics at the same time) was 138 (84.66%).

Table 2 Antibiotic resistance results of pneumococcal isolates from children with invasive pneumococcal disease
Full size table

3. Comparison of clinical characteristics between survivors and nonsurvivors

Among the seven patients with IPD who died, six (85.71%) were male, with an average age of 17.97 (IQR 11.23–36.18) months (Table 3). The PICU and general ICU had the most nonsurvivors, whereas the SICU had the most survivors. Neurological diseases were observed in most patients (n = 3; 42.68%). A comparison of vital signs revealed that the differences in temperature, HR, RR, SBP, and DBP between the surviving and nonsurviving patients were not significant (all P > 0.05). A comparison of laboratory test results revealed that the PaO2 level (P = 0.030) was significantly greater in patients who survived than in those who died, whereas the D-dimer (P = 0.038), fibrinogen (Fib; P < 0.001), and C-reactive protein (CRP) levels (P = 0.002) were higher in patients who died than in those who survived. Among the patients with IPD who died, 28.57% used vasopressors, which was not significantly different from the percentage of surviving patients. In addition, the length of hospital stay was 4.41 (IQR 1.08–9.95) days in patients who died was significantly shorter than that in surviving patients (P = 0.045).

Table 3 Comparison of clinical characteristics between survivors and nonsurvivors of IPD children
Full size table

4. Risk factors for sepsis among patients with IPD

Among the 61 patients with IPD who developed sepsis, 35 (57.38%) were male, with an average age of 18.40 (IQR 6.87–55.97) months (Table 4). Among the patients with IPD, the PICU had the most patients with sepsis (45.90%), whereas the SICU had the most patients without sepsis (42.74%). In patients with sepsis, the most frequent causes of ICU admission were neurological and respiratory disorders (both 24.59%). Higher temperatures, HR, and RR were found in patients with sepsis than in those without (all P < 0.05). A comparison of laboratory test results revealed that the international normalised ratio (INR), prothrombin time (PT), D-dimer levels, CRP levels, and lactate levels were significantly higher in patients with sepsis than in those without sepsis. In paediatric patients with IPD who developed sepsis, the platelet, haemoglobin, PaO2, and albumin levels were significantly lower than those in patients without sepsis. Hospital data showed that the duration of hospital and ICU stays in the sepsis group were significantly longer than those in the nonsepsis group (both P < 0.001). The difference in the in-hospital mortality rate between the sepsis group and nonsepsis groups was not statistically significant (Table 4). Temperature (OR 3.80, 95% CI 1.62–8.87; P = 0.0021), PaO2 (OR 0.99, 95% CI 0.98-1.00; P = 0.0266), and albumin (OR 0.89, 95% CI 0.80–0.99; P = 0.0329) were found to be independent risk factors for sepsis in children with IPD (Table 5).

Table 4 Clinical characteristics of invasive pneumococcal disease in children with or without sepsis
Full size table
Table 5 Logistic regression analysis for risk factors of sepsis in IPD children
Full size table

Discussion

IPD is a significant health concern in children worldwide. We conducted a retrospective single-center study involving patients with IPD in the ICUs of a specialised hospital in Hangzhou, China, from 2010 to 2018. This study focused on the epidemiology, bacterial antimicrobial resistance, and prognostic prediction of patients with IPD with the aim of enhancing our understanding and early treatment of this disease in paediatric patients admitted to the ICU.

The epidemiology of IPD in children varies according to region and age. In our cohort, 66.29% of children with IPD were aged 2–10 years, which differs from data from other studies [3, 20]. Among patients with IPD, the overall mortality rate was 3.93%, which was lower than that reported in other studies, where the mortality rate was 6.25-7.8% [21]. More than half (61.8%) of patients developed IPD during the cold season. Understanding the epidemiology of IPD will aid in the development of effective preventive measures. Various underlying diseases, including cancer, asthma [22], renal disease, congenital heart disease (CHD), and immune deficiency [23] are often observed in patients with IPD and are considered risk factors for IPD. In our study, cardiovascular disorders, especially CHD, were the most common underlying diseases in children with IPD, indicating strong susceptibility to S. pneumoniae infection in these patients.

Antimicrobial resistance is a growing challenge in the treatment of IPD in children worldwide, particularly in Asia. Data from the Asian Network for Surveillance of Resistant Pathogens showed that China has the highest rate of multidrug and macrolide resistance among S. pneumoniae isolates [24]. In the present study, the rate of erythromycin resistance was the highest in S. pneumoniae, which is similar to findings in other regions of China, including Suzhou [20], Shanghai [25], and Beijing [26]. However, the proportion of erythromycin-resistant isolates in India is approximately 37% [27], which is lower than that in China. Consistent with the previous findings [3], our study revealed high rates of resistance to tetracycline and SMZ-Co. Notably, the resistance rates to penicillin, ceftriaxone, and cefotaxime were significantly higher in the meningeal isolates than in the non-meningeal isolates. Therefore, different infectious conditions should be considered during individual antimicrobial selection in patients with and without meningitis. Moreover, all isolates were sensitive to vancomycin, linezolid, moxifloxacin, telithromycin, ofloxacin, and levofloxacin. Another study of 123 hospitalised children with IPD in Shanghai, China, also showed that vancomycin, linezolid, and levofloxacin were sensitive antibiotics for all detected S. pneumoniae isolates [3]. In our study, the proportion of observed MDR bacteria (84.66%) was similar to that reported in another study in China (81.2%) [28] and other Asian countries (83.3%) [24] but was substantially greater than that in India [27], reflecting the high geographical and genetic diversity of S. pneumoniae. Appropriate surveillance and antibiotic stewardship are crucial for the management of resistant strains.

Patients with IPD may experience poor outcomes including death and sepsis. In this study, the in-hospital mortality rate of children with IPD was 3.93%, and most nonsurvivors were males aged < 5 years. In addition, 61 (34.27%) patients with IPD developed sepsis and had longer hospital and ICU stays, indicating severe infection complicated by organ dysfunction. Predicting risk factors for poor prognosis in children with IPD aids in the early identification of severe cases and guides treatment decisions. Previous studies have shown that age, meningitis, underlying diseases, penicillin resistance, and inappropriate initial antibiotic therapy are risk factors for IPD-related mortality [11, 29, 30]. Our results showed that PaO2 levels decreased significantly in nonsurvivors and were independent predictors of sepsis in children with IPD. Similarly, a retrospective single-centre study in Beijing indicated that respiratory failure was an independent risk factor for mortality in patients with IPD [28]. Therefore, preventing hypoxemia and ensuring a sufficient oxygen supply are effective measures for reducing the mortality of patients with IPD.

Dysfunction of coagulation plays a crucial role in the pathogenesis of sepsis-related host dysregulation and organ failure. Increasing evidence suggests that the D-dimer is a valuable biomarker for predicting the prognosis of patients with different infectious conditions [31,32,33,34]. In the present study, D-dimer levels were significantly higher in patients who died and in those with sepsis. A D-dimer cutoff value of 3,000 ng/mL helps in identifying patients with a two-fold increase in the occurrence of multiple organ dysfunction syndrome in S. pneumoniae-invasive infections [35]. In addition, Fib levels were higher in patients who died than in survivors, and the INR and PT were significantly higher in patients with sepsis. These results highlight that medical staff should especially note patients with IPD and coagulopathy and ensure prompt management to reduce the risk of poor outcomes.

As an indicator of inflammation, CRP levels were significantly higher in nonsurvivors and patients with sepsis in our study. A previous study that compared IPD and pneumococcal pneumonia showed that a CRP level ≥ 17.0 mg/dL was a useful factor for suspicion of IPD [36]. Similarly, high CRP levels, leukopenia and thrombocytopenia are associated with mortality in paediatric patients with IPD, and these factors warrant special attention upon admission [37]. Moreover, we found that temperature (OR 3.80, 95% CI 1.62–8.87) was an independent predictor of sepsis in children with IPD, indicating that severe inflammation participated in disease progression. Intense inflammation is a hallmark of IPD and appropriate anti-inflammatory treatment is necessary.

Serum albumin (ALB) levels are also an independent risk factor for sepsis in children with IPD. Hypoalbuminaemia is common in critically ill patients because of decreased hepatocyte synthesis and altered distribution from vessels to tissues [38]. Recently, increasing evidence has indicated that decreased serum ALB concentration is a marker of inflammation rather than that of nutritional status [39, 40]. A nationwide multicentre study of 3967 patients with sepsis admitted to the ICU revealed that a decrease in serum ALB concentration one week after admission was a strong predictor of mortality in younger adults [41]. Additionally, low serum ALB levels are associated with an increased risk of IPD [42]. ALB-containing solutions are recommended for the resuscitation of patients with sepsis because of the lower mortality rate of patients treated with these solutions than that of patients treated with other fluid resuscitation strategies [43]. However, another multicentre study showed that compared with crystalloids alone, ALB replacement in combination with crystalloids did not improve the survival rate at 28 or 90 days in patients with severe sepsis [44]. Therefore, additional high-quality studies are required to explore the effects of ALB replacement therapy.

This study has several limitations. First, owing to its retrospective, single-centre nature, the sample size was relatively small. However, as one of the national regional medical centres for children, a substantial proportion of inpatients are from all over the country; thus, they could, at least in part, represent the clinical characteristics of IPD. Second, the serotypes of S. pneumoniae isolates were not determined. Third, our study did not include paediatric patients with non-IPDs in the ICUs, which may create a bias. In addition, because of the lack of data concerning vaccination of patients in the PIC database, we did not show relevant data on immunisation with PCVs. Therefore, large-scale multicentre studies, including the serotyping of pneumococcal isolates and immunisation with PCVs, are required in the future.

Conclusions

This study demonstrated that paediatric IPD deserves attention in China. Most children with PID in the ICUs were 2–10 years old. Antimicrobial resistance tests of S. pneumoniae isolates revealed high resistance to erythromycin, tetracycline, and SMZ-Co. All isolates were sensitive to vancomycin, linezolid, moxifloxacin, telithromycin, ofloxacin, and levofloxacin. Patients with IPD may experience poor outcomes including death and sepsis. The in-hospital mortality rate was 3.93%, and 34.27% of patients had sepsis. Temperature, PaO2 levels, and serum ALB concentration were independent risk factors for sepsis in children with IPD. Considering that this was a retrospective single-centre study, further prospective, multicentre studies are required to validate our findings.