The retrospective analysis of the present data shows that approximately 40% (n = 132) of all 327 patients included had a PCT of <0.1 ng/ml and more than two thirds (n = 225) had <0.25 ng/ml. We hypothesised that a PCT <0.1 ng/ml indicated absence of bacterial infection or—even more important—absence of bacterial infection requiring antibiotic treatment. In patients with PCT values of <0.25 ng/ml, bacterial infection was assumed to be unlikely. To find out whether these hypotheses were robust, we considered ARI in patients not treated with antibiotics to definitely have been of viral aetiology or self-limiting bacterial infection. We estimated it appropriate to exclude five patients from our analysis who were treated with antibiotics for other reasons (e.g. urinary tract infection), 17 patients who had already started with antibiotics prior to admission and of four children with chronic illnesses since these patients would also be excluded from a prospective trial. We then had a closer look into the remaining 12 patients with serum PCT concentrations of <0.1 ng/ml who started with antibiotic treatment on admission (postulating that in these patients bacterial respiratory tract infection had been suspected on admission by the paediatrician in charge). Specifically, were there children with low PCT who nevertheless proved to have bacterial infection? As far as could be judged from retrospective analysis of clinical, laboratory, microbiological or radiological findings in these 12 children, there was no substantial criterion for underlying bacterial infection. Antibiotic treatment could probably have been withheld safely. Therefore, we suggest a PCT serum concentration of <0.1 ng/ml as an adequate cutoff value to prospectively guide treatment decision in otherwise healthy children with ARI.
As shown in Table 1 (column 5), 40% of patients with PCT values ≥0.5 ng/ml had not received antibiotics (neither initially nor in the clinical course) without adverse effects as far as could be judged from retrospective analysis. This reflects the very restrictive antibiotic treatment policy in patients with ARI in our institution. Thus, unlike Christ-Crain et al. in adult patients with ARI, we would not strongly encourage initial antibiotic treatment in children presenting with ARI and PCT values ≥0.25 ng/ml (especially ≥0.5 ng/ml) in a prospective intervention trial. The most beneficial cutoff in infants and children still has to be determined.
Remarkably, virus testing was positive in about two thirds of patients with PCT values <0.5 ng/ml and only less in those with a PCT ≥0.5 ng/ml (54%). Comparing the percentages of RSV, adenovirus or rhinovirus detections, there was no relevant difference between high or low PCT values. It is known that viral infections, e.g. RSV, rhinovirus or hMPV, may facilitate bacterial adherence to respiratory epithelial cells [11, 15, 16] and thus may lead to bacterial superinfection. Therefore, viral–bacterial co-infections might account for a number of cases with PCT values of ≥0.25 ng/ml. On the other hand, previous studies have come to the conclusion that PCT cannot reliably distinguish between viral or bacterial aetiology but reflects more the invasiveness and severity of microbial invasion [2, 9, 13]. Thus, elevated PCT values of ≥0.25 ng/ml might rather indicate the presence of more invasive viral rather than true bacterial infection in some of our patients, too.
These retrospective considerations illustrate again the dilemma in the differentiation between viral and bacterial respiratory tract infections, mainly due to the limitations of viral and especially bacterial detection methods.
A further interesting aspect of our analysis was that all six patients with positive PCR testing for M. pneumoniae (n = 5) or C. pneumoniae (n = 1) had a PCT of <0.1 ng/ml. In another five patients where atypical pneumonia had been diagnosed with radiograph only, PCT concentrations were below 0.25 ng/ml, too. This is in agreement with previous data in adult patients showing that PCT can be helpful in differentiating typical from atypical pneumonia [12]. On the other hand, PCT seems to increase and even correlate with the prognosis of adult patients with community-acquired pneumonia due to Legionella pneumophila as shown in a recent study [10]. It is common clinical practice to treat patients with verified or suspected atypical lower respiratory tract infections with macrolides or doxycycline. In the literature, however, the indication for antimicrobial therapy in atypical pneumonia and its effectiveness are discussed rather controversially [8, 22, 23]. Within a possible prospective PCT-guided intervention trial, children with atypical (predominantly M. pneumoniae and C. pneumoniae, only rare cases of L. pneumophila in childhood) respiratory tract infections and PCT levels <0.1 ng/ml would obviously not be treated with antibiotics. Given the controversial discussion on this issue, however, there seems to be no relevant loss of safety withholding antibiotic therapy in these children. If there are strong clinical or radiological hints for atypical pneumonia, respective antibiotic therapy should be discussed.
Concerning the ten cases with positive blood cultures, we retrospectively judged five blood cultures positive for coagulase-negative staphylococci and Corynebacterium spp. as bacterial contamination as none of these children presented with clinical symptoms of bacteraemia or sepsis nor did any of the laboratory findings, inclusive of PCT, match with severe bacterial infection. Additionally, we judged the culture positive for S. aureus as contamination since the patient did not develop clinical signs of infection (confirmed by personal follow-up call after discharge). Only one of these six patients (with detection of coagulase-negative staphylococci) was treated with antibiotics for pneumonia. We think the PCT values remained low in these patients in absence of veritable bloodstream infection and should not be considered as falsely negative.
Yet, in three clinically very ill patients with blood cultures twice positive for Streptococcus pneumoniae and once for Salmonella enteritidis, significantly elevated PCT values (38.9, 10.66 and 13.22 ng/ml) were found (in addition to highly elevated CRP and leucocyte counts). This confirms previous studies that report on a significant induction of PCT in septic patients [19, 24].
Korppi and Kroger [14] postulated that a CRP of >40 mg/l made sole viral aetiology of respiratory tract infection in children rather unlikely. In more recent studies, CRP concentrations of even 50 to 60 mg/l have been reported to discriminate more precisely bacterial infection from other types of inflammation [7, 18, 20, 26]. In our study, CRP was ≤40 mg/l in 11 cases and 43 mg/l in one of the 12 patients with a PCT <0.1 ng/ml. None of these patients had clear evidence for significant bacterial infection. Thus, inclusion of the above-mentioned CRP criteria might further support our hypothesis that a PCT serum concentration of <0.1 ng/ml is an adequate parameter to withhold antibiotic treatment in otherwise healthy children with ARI.
In conclusion, in spite of methodical shortcomings in a retrospective study such as this, we assume that antibiotic therapy could probably have been withheld without adverse effects in these children who suffered from ARI but were otherwise healthy and had PCT concentrations of <0.1 ng/ml. The results of this study give hope that PCT might develop into a reliable marker to identify children with ARI not requiring antimicrobial treatment and consequently help in reducing the tremendous overuse of antibiotics in children with ARI. This hypothesis, however, remains to be proven in a prospective randomised intervention trial.