Introduction

Childhood community-acquired pneumonia (CAP) is a common illness, annually affecting ca. 1.5 million children in the European Union (of 64 million children <15 years of age) [3, 13, 17, 26]. Several direct and indirect aetiological studies suggest that Streptococcus pneumoniae, Mycoplasma pneumoniae, Chlamydia pneumoniae and Haemophilus influenzae are the most common bacteria causing CAP [14, 21, 33]. Increasing lack of susceptibility of S. pneumoniae to penicillin and macrolides, lack of susceptibility of M. pneumoniae and C. pneumoniae to betalactams and increasing betalactamase production of H. influenzae suggest that one single antibiotic may not be effective in all cases of pneumonia. Furthermore, in up to 50% of CAP cases in children, there is evidence of viral infection [14, 27].

Surprisingly, few comprehensive studies of the antibiotic treatment of CAP in children have been reported from developed countries. No prospective placebo-controlled study has been reported. This explains the wide variation in the empirical treatment recommendations [3, 17, 26]. Limited study data are available on the expected response to treatment in children with CAP. We report the clinical responses and outcomes to antibiotic treatment in 153 children hospitalised for uncomplicated CAP with carefully defined possible aetiology.

Subjects and methods

Subjects

During the 3 years between January 1st 1993 and December 31st 1995, 254 patients with radiologically confirmed childhood CAP were prospectively enrolled in the study at the Department of Paediatrics, Turku University Hospital. The main causes of hospitalisation were high fever, dyspnoea, malaise/lethargy, vomiting or poor appetite. The diagnosis was based on a simultaneous finding of infiltrates compatible with pneumonia on the chest X-ray film as well as signs and symptoms of acute infection and fever (>37.5°C). The role of 17 microbes (10 viruses and 7 bacteria) was studied. The details of the methods and microbiological findings have been reported earlier [14]. In brief, antibody tests with acute and convalescent serum samples were used for bacterial diagnosis. Viral diagnosis was based on virus culture, sensitive immunoassay for the antigens of seven viruses and a polymerase-chain-reaction assay for rhinoviruses from nasopharyngeal aspirates and serological tests with acute and convalescent sera. A blood sample for blood culture was taken from 125 patients and only one patient was found positive for S. pneumoniae. Of the 254 patients, 61 had received antibiotic treatment before hospitalisation, 23 patients had no fever on the ward, and at the time of hospitalisation and onset of antibiotic treatment, four patients had complicated pneumonia (two patients had empyema, one patient bronchiectasis and one a cystic malformation of lung). In addition, 13 patients had severe underlying chronic conditions (six had genetic syndromes, six neurocognitive disorders and one astrocytoma). All these 101 patients were excluded from the analysis and thus 153 patients were included in the study. The antibiotic treatment was chosen by the attendant physician. The median age of the 153 patients was 2.3 years (range 0.1–16.7 years). Twenty-one (14%) patients were 0 to 11 months of age, 46 (30%) 12 to 23 months of age, 49 (32%) 2 to 4 years of age and 37 (24%) were 5 years of age or older. Of the patients, 53% were boys. Of the 153 patients, 18 had stable bronchial asthma, 5 atopy, 3 had been premature, 2 had congenital heart disease, 2 had epilepsy in addition to bronchial asthma, 1 had diabetes mellitus, 1 had recovered from acute lymphoblastic leukaemia, and 123 patients (80%) were healthy.

The duration of fever (>37.5°C) and the duration of hospitalisation were studied as objective measures of recovery. These data were recorded from patient charts as hours. Body temperature was measured rectally in patients <3 years of age and in the axilla in children ≥3 years of age. Temperature was measured in the ward every 6–8 h; when a temperature was found normal, the duration of fever was marked as 3 h of the 6–8 h period. Of the 153 patients, 120 febrile patients (78%) were treated with paracetamol or naproxen.

The follow-up examination was carried out about 3–4 weeks after discharge, when, in addition to a careful history of the time after discharge and to a clinical examination, blood samples for viral and bacterial serology and a chest radiograph were taken.

Statistical analyses

Pearson’s standard chi-squared test was used to compare proportions between the groups (Fisher’s exact test when the expected count was less than 5) and the Mann-Whitney test to determine the statistical differences in continuous variables. Differences in the duration of fever between the separate infection groups (Fig. 1) were analysed using the survival analysis approach with the logit model for discrete response variables. These analyses were done using the Probit Procedure of the SAS System for Windows, version 8.2 [1].

Fig. 1
figure 1

Response to treatment: percentages of febrile (>37.5°C) patients after onset of antibiotics. A All patients. B Patients with either sole viral, sole bacterial, or mixed viral and bacterial infection. P=0.89 when comparison is made between groups of patients with sole viral and sole bacterial infection, P=0.08 between sole viral and mixed infection and P=0.04 between sole bacterial and mixed infection. C Patients with either sole RSV, sole pneumococcal, or mixed RSV and pneumococcal infection. P=0.30 when comparison is made between groups of patients with sole RSV and sole pneumococcal infection, P=0.03 between sole RSV and mixed infection and P=0.04 between sole pneumococcal and mixed infection

Results

A potential aetiology was found in 127 (83%) of 153 patients: 44 (29%) patients had evidence of sole viral and 39 (26%) of sole bacterial infection and 44 (29%) of mixed viral-bacterial infection. The probable aetiological agents were: S. pneumoniae in 60 (39%) patients, respiratory syncytial virus (RSV) in 35 (23%), rhinovirus in 33 (22%), parainfluenza viruses 1, 2, 3 in 20 (13%), non-typeable H. influenzae in 14 (9%), adenovirus in 10 (7%), M. pneumoniae in 9 (6%), influenza viruses A or B in 6 (4%), M. catarrhalis in 6 (4%), coronavirus in 4 (3%), C. pneumoniae in 4 (3%), human herpes virus 6 (HHV 6) in 2 (1%), Streptococcus pyogenes in 1 (1%) and Bordetella pertussis in 1 (1%) patient. The most common viral-bacterial infections were rhinovirus + S. pneumoniae (n=15) and RSV + S. pneumoniae (n =8) infections. On admission, the median WBC was 17.5×109/l and the median level of serum CRP 85 mg/l. Of the patients, 72% had alveolar infiltrates on the chest X-ray film. Sole bacterial and viral pneumonias differed significantly in the number of alveolar pneumonia, median CRP and median WBC (Table 1).

Table 1 Characteristics of 153 children with CAP with defined aetiology treated with antibiotics in hospital

Before hospitalisation, the median duration of fever in the 153 children was 2.0 days (range 0–11 days). On the ward, the median duration of fever after onset of antibiotic treatment was 14 h (range 2–127 h) (Table 1). Interestingly, the durations of fever in patients with sole bacterial and sole viral infection were similar, but patients with mixed viral and bacterial infections had longer durations of fever after onset of antibiotic treatment (Fig. 1). The same finding, statistically significant, was obtained when patients with sole (all other concomitant infections excluded) RSV, S. pneumoniae, and mixed RSV and S. pneumoniae pneumonia were compared (Fig. 1).

Of the 153 patients, 43 (28%) had fever lasting ≥24 h after onset of antibiotic treatment. Fever lasting for ≥48 h was considered treatment failure and was recorded in 13 (9%) of the 153 children (Table 1 and Fig. 1). Half of the patients (7/13) with treatment failure had evidence of mixed viral-bacterial infection, three (23%) had sole viral infection and two (15%) sole bacterial infection; in one patient no possible aetiology was found. The pathogens usually found in patients with treatment failure were RSV (46%) and H. influenzae (39%). Seven (5%) patients had fever lasting for ≥72 h; only one of them had sole bacterial infection (H. influenzae). Two patients had fever for ≥120 h; both had viral (one influenza A virus and one with RSV and coronavirus) infections. The median duration of hospital stay was 48 h (range 7–240 h).

Of the patients, 106 were treated with intravenous penicillin G, 13 with cefuroxime, and the rest 34 with other antibiotics (Table 2). The penicillin treatment group showed a trend for more treatment failures in children with either B. pertussis, C. pneumoniae, M. catarrhalis, or M. pneumoniae pneumonia (penicillin-non-susceptible bacteria) compared to children with sole S. pneumoniae pneumonia (2/16 versus 0/21 patients respectively; P=0.18). The initial antibiotic therapy was changed in six patients: three of them had sole viral, two mixed viral-bacterial infection, and in one patient the aetiology was not established. After discharge, all 153 children received oral antibiotics so that the total duration of antibiotic treatment was 7–10 days. Seven patients had fever at home immediately after discharge from the hospital; four of them had mixed viral-bacterial infection, two sole bacterial infection, and one sole viral infection.

Table 2 Antimicrobial therapy in 153 children with CAP

A follow-up examination was carried out in all 153 children 17–62 days (median 29 days) after admission to the hospital. No recurrent pneumonias were recorded and 94% of the patients were reported by the parents to be free of CAP-related symptoms. However, 32 (21%) patients had received a second antibiotic treatment during the follow-up, mainly for acute otitis media.

Discussion

We found that the study children rapidly recovered from CAP after onset of antibiotic therapy and no one developed complications during the antibiotic treatment. Of the patients, 91% became afebrile within 48 h and the median duration of the hospital stay was 48 h. Our detailed findings are consistent with earlier studies. In developed countries, independent of aetiology and the type of antibiotics used, 70%–94% of children with uncomplicated CAP have been found to be afebrile and clinically improving within 48 h after the start of antibiotics [7, 27, 30]. The clinical responses are often reportedly good or leading to cure in 84%–98% of CAP patients, without any more detailed daily data on their recovery [2, 10, 28]. Even children with bacteraemic pneumococcal pneumonia have become afebrile within an average of 22 h after onset of antimicrobial therapy. A rapid clinical recovery has been associated with rapid normalisation of WBC and CRP [30]. This is in sharp contrast to data on pneumonia in less-developed countries, where severe pneumonia is common and 5% of the patients die in spite of parenteral antibiotic treatment and treatment failures are not rare in patients with non-severe pneumonia (18% of patients in one study still had CAP-related symptoms after 5–7 days) [4, 9].

Only 9% of the children with CAP had fever for ≥48 h after onset of antibiotic treatment. Of these patients, 75% had evidence of viral infection, usually with concomitant bacterial infection. Many findings support the view that mixed viral-bacterial infections, which may be more common than earlier recognised, may induce a more severe inflammation and clinical illness than sole bacterial or sole viral infections. Concomitant influenza virus and Staphylococcus aureus infection may cause severe and fatal pneumonia in children and adults [6]. In one study, children with acute otitis media showing both bacteria and virus in middle ear fluid failed to respond to therapy more often than a group with bacteria alone [5]. In a common cold study, patients having both virus and S. pneumoniae, H. influenzae or M. catarrhalis in the nasopharynx had more symptoms than those with virus alone [23]. Furthermore, in another study these patients with common cold seemed to benefit from antibiotic therapy [16].

A surprising finding was that children with sole viral pneumonia showed a response to antibiotic treatment similar to that of those with sole bacterial pneumonia. The same finding was obtained when only sole RSV and sole pneumococcal pneumonia were studied. It can be argued that all our patients with CAP may have had a bacterial pneumonia and the tests used to detect a bacterial infection were too insensitive. This is supported by the finding that pneumolysin PCR analysis of blood samples may detect cases overlooked by serology [18, 31]. Furthermore, serological tests do not detect all blood culture positive cases [12]. On the other hand, our patients with sole bacterial pneumonia differed significantly from those with sole viral pneumonia both in laboratory and chest radiography findings, showing that the serological tests detected a bacterial type infection [15, 32]. Another explanation could be the high spontaneous cure rate of CAP in children, as studies with acute otitis media and acute sinusitis have also shown [24, 25]. Penicillin G is still considered a drug of choice in hospitalised children with CAP in many European countries with low penicillin resistance of pneumococci [11, 19, 26]. Of our patients, 66% were treated with penicillin G and showed a rapid and uneventful recovery. Penicillin G is no longer recommended in the United States as a first-choice drug because of limited supply and the increasing resistance of pneumococci to penicillin [20], whereas in Finland, 95% of pneumococcal strains remain sensitive to penicillin [22]. Our patients received only 1–2 days of parenteral treatment, after which they were successfully switched to oral treatment. Earlier studies have shown that the majority of children with CAP can be treated as outpatients after one injection of cephalosporin [8].

The limitations of our study must be admitted. We only studied otherwise healthy children in a highly developed country. We did not include children who already had complicated pneumonia, i.e. empyema, (two children) on admission before the onset of antibiotic treatment. The relative frequency of empyema is increasing, often requiring prolonged hospitalisation [29]. Only indirect evidence of the aetiology of CAP was obtained, and causality should be considered with some caution.

Our results show that CAP in children in a developed country is not a serious illness. In clinical practice, nearly all otherwise healthy children with pneumonia become afebrile within 1–3 days and make an uneventful recovery. Etiological investigations are not needed for the planning of therapy in these children. Only a very short parenteral treatment period is needed and the patients can then be treated with oral antibiotics as outpatients.