Identified risk factors for co-infection in hospitalised children infected with adenovirus in Hangzhou

Abstract

This study aimed to describe the clinical manifestations of adenovirus infections and identify potential risk factors for co-infection with chlamydia, viruses and bacteria in hospitalised children from Hangzhou, China. From January to December 2019, the characteristics of hospitalised children infected with adenovirus at Hangzhou Children's Hospital and Zhejiang Xiaoshan Hospital were collected. The clinical factors related to co-infection with chlamydia, viruses and bacteria were assessed using multivariate logistic regression analyses. A total of 5989 children were infected with adenovirus, of which 573 were hospitalised for adenovirus infection. The severity of adenovirus respiratory infection was categorised as follows: mild (bronchiolitis, 73.6%), moderate (bronchopneumonia, 17.6%) or severe (pneumonia, 8.8%). Of the 573 children who were hospitalised, 280 presented with co-infection of chlamydia, viruses or bacteria, while the remaining 293 had only adenovirus infection. Multivariate stepwise logistic regression analyses indicated that elevated ferritin was associated with an increased risk of chlamydia co-infection (odds ratio (OR) 6.50; 95% confidence interval (CI) 1.56–27.11; P = 0.010). However, increased white blood cell (WBC) count was associated with a reduced risk of viral co-infection (OR 0.84; 95% CI 0.75–0.95; P = 0.006). The study indicated that co-infection with chlamydia could be affected by elevated ferritin levels. WBC levels could affect viral co-infection in hospitalised children infected with adenovirus.

Introduction

Human adenoviruses (HAdVs) are double-stranded non-enveloped DNA viruses belonging to the genus Mastadenovirus, family Adenoviridae. These viruses cause various clinical manifestations, including acute respiratory infections, gastroenteritis, conjunctivitis, cystitis and meningoencephalitis [1]. HAdV infections account for 5–7% of respiratory illnesses in paediatric patients and 1–7% of those in adults [2, 3]. Populations susceptible to HAdV infection include those aged <5 years, those who live in densely populated areas and immunocompromised populations [4]. HAdVs are classified into seven species (A–G) and include more than 60 serotypes [5]. The clinical manifestations and severity of the disease are significantly correlated with the serotype [6].

Outbreaks of respiratory tract HAdV have been recently reported in Jiangsu and Taiwan provinces of China, Korea, Singapore and Malaysia [7–9]. Although most cases presented with mild to moderate disease, life-threatening diseases were also detected. This was especially the case in immunocompromised populations [10]. Children who have co-infection with other pathogens are especially vulnerable to developing severe disease [11]. A previous study attempted to identify the risk factors for disease severity and found that the duration of hospitalisation, lymphocyte count and lactate dehydrogenase (LDH) levels could all affect disease severity. Independent risk factors for co-infection were not identified [12], the present study aimed to explore the clinical manifestations of HAdV infection in children and the potential risk factors for co-infection with chlamydia, viruses and bacteria in hospitalised children infected with HAdV.

Methods

Patient inclusion and exclusion criteria

Data for 5989 children with HAdV infections who presented at Hangzhou Children's Hospital and Zhejiang Xiaoshan Hospital from January to December 2019 were collected. Among these, 573 children who were hospitalised for HAdV infection were recruited to explore the risk factors for co-infection with chlamydia, viruses and bacteria. All recruited patients were <14.0 years. These patients displayed positive results in HAdV tests with a reverse transcription polymerase chain reaction assay. The exclusion criteria were as follows: (1) congenital heart disease, (2) congenital pulmonary disease, (3) malignancy, (4) severe organ dysfunction, (5) infection in other organs and (6) use of corticosteroids within 1 week. The Institutional Review Board of Hangzhou Children's Hospital and Zhejiang Xiaoshan Hospital approved this study (No. 201803). All enrolled patients’ parents or guardians provided written informed consent for research purposes.

Variable measurements

The participants’ data were collected from electronic medical records. Clinical information, laboratory results and radiological findings of 573 patients were extracted. These data included the season of onset, fever duration, hospital stay, serum amyloid A levels, procalcitonin levels, C-reactive protein (CRP) levels, white blood cell (WBC) count, neutrophil count, lymphocyte count and the severity of HAdV respiratory infection. Levels of alanine aminotransferase (ALT), aspartate aminotransferase (AST), creatine kinase (CK), creatine kinase isoenzyme (CK-MB), LDH and ferritin were also extracted. Disease severity was based on the American Thoracic Society's guidelines for the management of community-acquired pneumonia, which categorises respiratory disease as mild (bronchiolitis), moderate (bronchopneumonia) or severe (pneumonia) [13]. A Digital Diagnostic TH system with a Philips DR machine was used to obtain the chest radiographs. The automatic exposure control was adopted to set mAs and Kv values according to the patients’ age as follows: 58–65 Kv/1–2 mAs for patients aged 1–6 years or 65–80 Kv/2–3 mAs for those aged over 6 years. The GE OPTIMA 540 system was used to obtain CT images with the following parameters: 16 rows; tube voltage, 80–100 kVp for children aged 1–5 years and 120 kVp for children aged >5.0 years; tube current, 10–300 mA; pitch, 1.375:1; collimation, 40 mm; layer thickness and spacing, 5 mm.

Definition of co-infection

The co-infections were grouped into those caused by chlamydia, viruses or bacteria. No co-infection was considered as a control. Co-infection caused by chlamydia was determined by enzyme-conjugate immunoassay, while virus co-infection was tested using swabbed with a nasopharyngeal sample collection kit. Moreover, fungal and bacterial infections were assessed through blood cultures before using antibiotics. Children who showed positive results for HAdV and co-infection of chlamydia, a virus or a bacterium were regarded as displaying co-infection and were divided into chlamydia, viral or bacterial infected groups. The mixed co-infection was defined as infected two types of chlamydia, viruses or bacteria.

Statistical analysis

The collected characteristics of children were presented as median (quartile) and number (percentage) for continuous and categorical variables, respectively. The differences in characteristics among the chlamydia, viral or bacterial co-infection and control groups were assessed by Wilcoxon and χ ² tests. Multivariate logistic regression analysis was performed. These were conducted to estimate multivariate adjusted odds ratios (ORs) and 95% confidence intervals (CIs) for the risk factors related to subsequent co-infection with chlamydia, viruses or bacteria. Multivariate analyses were performed using both the full model and stepwise regression approach. All reported P values were two-sided, and P values <0.05 were considered statistically significant. Statistical analysis was performed using SPSS version 22 software (SPSS, Chicago, Illinois, USA).

Results

Epidemiologic characteristics of infected children

The prevalence of HAdV infection according to months of disease onset is shown in Figure 1. The peak months of disease onset were December, January, May and June. A dip in infections was observed in August, September, October and November. Of 5989 children with HAdV infection, the mean age was 4.17 years. Most patients were aged 6.0 months or older. A total of 1746 children were aged 6.0 months to 2.0 years, 1930 were aged 2.0–4.0 years and 2278 were aged 4.0 years or older. Only 35 patients were aged 6.0 months or younger.

Fig. 1. The epidemic curve for children with HAdV infections.

Characteristics of infected children and disease severity

A total of 573 children were hospitalised for HAdV respiratory infection, of which 73.6% showed bronchiolitis, 17.6% displayed bronchopneumonia and the remaining 8.8% exhibited severe pneumonia. Of these patients, 280 showed co-infection, while the remaining 293 were infected only with HAdV. One child presented with brachial plexus nerve injury. There were significant differences among the groups in terms of hospital stay (P = 0.002), WBC count (P < 0.001), neutrophil count (P = 0.002) and ferritin levels (P = 0.023). However, there were no significant differences in the month of onset (P = 0.250), fever duration (P = 0.174), serum amyloid A level (P = 0.954), procalcitonin level (P = 0.569), CRP level (P = 0.516), lymphocyte count (P = 0.487), ALT (P = 0.431), AST (P = 0.675), CK (P = 0.932), CK-MB (P = 0.660) and LDH (P = 0.846) (Table 1).

Table 1. Baseline characteristics of recruited children infected with adenovirus

Risk factors for co-infection with chlamydia

The results of multivariate logistic regression based on the full model revealed that the onset months of February (OR 0.12; P < 0.001), March (OR 0.11; P < 0.001), June (OR 0.32; P = 0.010), July (OR 0.41; P = 0.043) and December (OR 0.37; P = 0.028) were associated with a reduced risk of chlamydia co-infection as compared with disease onset in January. Multivariate stepwise logistic regression revealed that February (OR 0.10; P < 0.001), March (OR 0.10; P < 0.001), May (OR 0.35; P = 0.011), June (OR 0.28; P = 0.003), July (OR 0.35; P = 0.012), September (OR 0.22; P = 0.031) and December (OR 0.35; P = 0.017) were associated with a reduced risk of chlamydia co-infection as compared with disease onset in January. However, both full model (OR 6.50; P = 0.010) and stepwise (OR 6.48; P = 0.013) logistic regression found a ferritin level >1500 was associated with an increased risk of chlamydia co-infection (Table 2).

Table 2. Multivariable adjusted analyses for the risk of co-infection of chlamydia, virus and bacterial

^a Stepwise, α _inclusion = 0.05, β _exclusion = 0.10

Risk factors for co-infection with viruses

The multivariate stepwise logistic regression found that elevated WBC levels were associated with a reduced risk of viral co-infection (OR 0.84; P = 0.006). No other significant indicators for co-infection with HAdV were observed (Table 2).

Risk factors for co-infection with bacteria

The results of multivariate logistic regression based on the full model found that the onset months of February (OR 0.03; P < 0.001), June (OR 0.33; P = 0.033), July (OR 0.22; P = 0.010), August (OR 0.15; P = 0.025), November (OR 0.17; P = 0.036), December (OR 0.19; P = 0.009) and an elevated neutrophil count (OR 0.71; P = 0.008) were associated with a reduced risk of bacterial co-infection. Moreover, elevated WBC levels were associated with an increased risk of bacterial co-infection (OR 1.34; P = 0.009). The multivariate stepwise logistic regression suggested that the onset months of February (OR 0.03; P = 0.001), May (OR 0.26; P = 0.008), June (OR 0.34; P = 0.028), July (OR 0.24; P = 0.009), August (OR 0.17; P = 0.030), November (OR 0.19; P = 0.041) and December (OR 0.24; P = 0.014) were associated with a reduced risk of bacterial co-infection (Table 2).

Discussion

The onset of HAdV in children remains controversial. The results of this study based on 5989 cases found that the peak periods of disease onset were in December, January, May and June. However, a dip in infections was observed in August, September, October and November. This result suggests that the onset of HAdV is significantly correlated with climatic factors, and the potential reason for this could be seasonal variability of the co-infecting pathogens. Moreover, the most frequent age of children infected with HAdV was 6.0 months or older. Furthermore, the risk of chlamydia co-infection could be affected by the month of onset and ferritin levels >1500. In addition, elevated WBC levels were significantly associated with a lower viral co-infection risk. Finally, the onset month, neutrophil count and WBC count could affect the risk of bacterial co-infection.

In this study, we noted that 73.6% of the children with respiratory disease presented with bronchiolitis. Their clinical manifestations were non-specific, including symptoms such as cough, runny nose and fever. Of the remaining patients, 17.6% presented with bronchopneumonia and 8.8% presented with severe pneumonia. The clinical manifestations of these cases included persistent high fever, dry cough, shortness of breath and hypoxemia. These symptoms were associated with a high incidence of mortality. These results are consistent with the findings of a previous study conducted in Korea [9]. Furthermore, the fever duration ranged from 0 to 13 days. The mean temperature fluctuated from 37.4 to 39.4 °C. This was not consistent with the study conducted by Xie et al. [14]. They noted that the temperature fluctuated from 40.0 to 40.9 °C in most children, the disease duration ranged from 3.0 to 14.0 days, and most cases entered the recovery period after 2.0 weeks.

Most cases presented with mild symptoms in the early stage and 73.6% of the children displayed bronchiolitis. The imaging characteristics in these cases included increased and thickened main veins in the lungs and a higher frequency of interstitial reticular shadows in the lower lobes of both lungs. Moreover, the characteristics of the moderate to severe cases included multiple clusters of consolidated shadows in both lungs, ‘centripetal’ lesions that were primarily found in the middle and inner lungs, enhanced fusion foci, additional large foci, added emphysema, extra lung texture, fewer round foci, less pulmonary bullae and few pleural effusions. These were associated with rapid progression, and multiple clusters of consolidated shadows in both lungs. Although the cases with moderate to severe symptoms presented with ‘centripetal’ lesions, they showed no obvious irregularity in the distribution and morphology of the bronchial tree. Finally, we did not observe positive cases of encephalitis. This indicated that HAdV infections did not involve the central nervous system (CNS). These characteristics could differentiate HAdV infections from A/H1n1 infections. This is because A/H1n1 tends to be distributed along the central bronchial tree with an intact bronchial tree shape. The H1N1 virus can easily invade the CNS, yielding positive images of the CNS. It may even cause necrotising encephalitis.

This study found that elevated ferritin levels were associated with an increased risk of chlamydia co-infection in hospitalised children infected with HAdV. The potential reason for this could be that ferritin not only represents iron reserves, but also serves as an inflammatory marker [15]. Moreover, we noted that elevated WBC levels were associated with a lower risk of viral co-infection and a higher risk of bacterial co-infection. The reason for this could be that WBC also serves as an inflammatory marker, which could affect bacterial co-infection. Thus, viral co-infection could not be reflected by WBC levels. Interestingly, we noted that an elevated neutrophil count was associated with a lower risk of bacterial co-infection. This was inconsistent with previous studies [16–18] and could be explained by the distribution of neutrophil counts in hospitalised children infected with HAdV. Therefore, elevated ferritin levels in hospitalised children infected with HAdV was considered as high-risk of chlamydia co-infection, while elevated WBC level was regarded as high-risk of bacterial co-infection.

Several limitations of this study should be acknowledged: (1) because this study was a hospital-based study and designed as retrospective, which can suffer from selection bias that is often associated with hospital-based retrospective studies; (2) the data available in the current study were based on electronic medical records, while the background therapies were not addressed; (3) stratified analyses based on the characteristics of patients were not conducted owing to the small number of patients in each group; and (4) the serotypes of the adenovirus, which could affect the clinical manifestations and severity of the disease, were not addressed in this study.

Conclusions

This study describes the epidemiologic and clinical manifestations of HAdV infection in children. The risk of co-infection with chlamydia, viruses and bacteria could be affected by the month of onset, neutrophil count, WBC count and ferritin levels in hospitalised children infected with HAdV. Further prospective studies should be conducted to construct a predictive model for the prognosis of hospitalised children infected with HAdV.