BSI represent a critical complication associated with hospitalization of very low birth weight (VLBW) infants, contributing to longer stay, and different long-term adverse outcomes. This phenomenon is well-understood and described [3, 7, 18–21], contrary to the subject under discussion, in which, unfortunately, data concerning antibiotic consumption are sparse and incomplete.
The proportion of microbiologically confirmed cases of BSI observed in the present study indicates similarity, but not identicalness, with other national programs. In the analysis of differences, attention should be paid to the applied definition of infections and different significance of microbiological findings for various types of surveillance. In American NHSN, in which to confirm LC-BSI, it was necessary to obtain at least 2 identical blood cultures, clinical sepsis was observed in 6.7–12.7% of infections [22], i.e. twice less frequently than in this study—significantly less restrictive in evaluating microbiological results. It unfortunately, indicates too rare application of the capabilities of contemporary microbiology in everyday clinical practice of the studied NICUs. This is confirmed by the results of a Cypriot study, in which LC-BSI constituted 96% of all BSIs [23]. Another matter is the debated problem of legitimacy of repeated blood drawing for cultures from VLBW newborns. Currently, it is more and more frequently assumed that, in this patient population, it is more justifiable to draw a single full-volume sample than to take two or more—even in the event of a result revealing typical skin contaminants, i.e. coagulase-negative staphylococci [24]. This is confirmed by the definitions adopted in the Netherlands [25], in NICHD Vermont Oxford Network [3] and the ones employed in the German national program called Neo-KISS [17].
The obtained data show that the use of diagnostics, even based on the standard, basic level, meaning culture, is a proceeding effectively affecting the consumption of antibiotics, and therefore, the costs of therapy, i.e. they reduce the consumption of antibiotics expressed by DDD.
The indicator most commonly used to assess the consumption of drugs, including antibiotics, is defined daily dose. This recognized international standard applicable in measuring the consumption of antibiotics is based on the average dose for the treatment of adults. For this reason, it is an indicator which, in relation to children, should be used with caution and one should take into account its limitations [26, 27].
Howether, both measures: DDD and the length of therapy were used by different authors with the application of various denominators—the number of admissions, number of person days or with respect to the treatment of a single patient [13, 15].
To evaluate the consumption of antibiotics in pediatric wards, particularly neonatal units, Gerber et al. employed DOT [15]. Depending on the type of neonatal unit, they found the length of treatment to be in the range from 5.7 in medical NICUs to 34.3 DOT in surgical NICUs. Median DOT in this study was about forty percent higher than in the medical NICUs in the study by Gerber; however, due to distinct populations and different degrees of detail, it is difficult to explicitly compare these values. Studies concerning antibiotic consumption in the neonatal population are not numerous, but what is more important: the published papers present a differentiated approach to the subject and are carried out with the use of diverse methodology and for various needs [5, 11, 13, 26, 28].
However, the main objective of this study was the analysis of the evaluation of antibiotic consumption in the treatment of one form of infection in a narrow and specific patient population, giving special consideration to the possibility of its use in the evaluation of the effectiveness and accuracy of microbiological diagnostics as an element of surveillance of infections in NICU.
And so, in NICUs covered by the study, significantly lower consumption of antibiotics expressed by DDD was observed in the case of LC-BSI treatment, compared to those in which the etiological agent was not isolated. On this basis, it can be concluded that the microbiological diagnosis of BSI in newborns treated in NICU was conducted properly and the results of microbiological tests were used in targeted therapy, what made it possible to obtain a reduction in the consumption of antimicrobial drugs and, consequently, the cost of treatment. In Polish NICUs, costs of medication account for nearly one-fifth of the total cost of treatment, the amount of which is inversely proportional to a child’s birth weight and directly proportional to the length of hospitalization [20].
A different situation was observed in the population of the PNSN neonates, who developed necrotising enterocolitis, wherein the length of therapy and consumption rates were not affected by the isolation of the potential etiological agent [29]. But with NEC, difficulties present themselves as regards obtaining material for microbiological examination, which would enable the isolation of the etiological agent. The quoted results concerning antibiotic consumption in NEC cases and the ones demonstrated in the present study regarding BSI illustrate the possibilities of how analyses in this respect could be utilized in infection control and, in particular, in evaluation of adequacy and effectiveness of microbiological diagnostics.
On the other hand, no significant differences in DOT values in cases of LC-BSI and microbiologically unconfirmed BSI were observed. This is contradictory to the current approach to modern antimicrobial stewardship: in newborn population, with suspected BSI, it is recommended to terminate antibiotic treatment after 48 h since the identification of symptoms, if the infection was not confirmed microbiologically. Even traditional diagnostics based on the culture method ensures obtaining a positive result (information on microbial etiology of infection) within 48 h [30, 31].
For antimicrobial stewardship to efficiently and effectively influence the reduction of antibiotic consumption, but not to decrease patient safety, in neonatal units, a principle of daily detailed review of the situation of neonates treated with antibiotics should be introduced, so as to minimize the intake of antibiotics in children, whose blood cultures and other clinical specimens tested negative and symptoms of infection are no longer observed or infectious origin of the disease was excluded. The lack of significant differences in DOT values of laboratory confirmed vs. not-confirmed BSI cases would point to the fact that these recommendations are not applied in the PNSN wards.
It should also be noted that rapid diagnostic molecular methods (which enable rapid assessment of the need to implement or discontinue therapy with vancomycin) could be implemented to decrease glycopeptides consumption, because glycopeptides are not easy to use in neonates [32, 33]. For it has been observed that the use of glycopeptides in the case of BSI caused by Enterobacteriaceae was lower by only approx. 20% in comparison with BSI caused by Gram-positive cocci. Generally, in our study, in treatment of BSI as a whole glycopeptides were used the longest. Similar situation was reported in study of Sameer et al. [34].
As for aminoglycosides, only the use of the DDD indicator permitted the demonstration of their significantly increased consumption in BSI caused by Gram-positive cocci (109 DDD), in comparison with BSI that were caused by Gram-negative bacilli (56.1 DDD).
Also noteworthy is the fact that there is a large share of antifungal drugs in the treatment of the analyzed cases of BSI, 25% of the entire consumption expressed by DDD (13.3% DOT), with simultaneous, lower than anticipated, participation of yeast-like fungi isolated in microbiological testing [35, 36]. This coincides with the trends observed in other studies. According to Fridkin, the application of fluconazole in infection prophylaxis contributed to this fact, which is also confirmed by other authors [6]. Thus, the high level of antimycotic medication consumption in PNSN wards fulfilled the task of reducing the incidence of fungal infections.
Antibiotic consumption assessment using at least two different measures, as presented in our study, can be a useful tool in antibiotic stewardship [27]. In presenting case, the results of the analysis indicate the need of implementing more sensitive and faster methods of microbiological diagnostics (PCR and/or MALDI-TOF) as a first step of reducing antibiotics consumption due to faster identification of etiological factor of infection. PCR increases the sensitivity of diagnostics test and shortens the time of identification of microorganisms without culture. MALDI-TOF improves the specificity and shortens the time of identification after receiving microorganism grow in culture method [37, 38]. These diagnostics techniques are still very rarely used in Polish hospital. They are considered as expensive procedures by hospitals’ management, because complex cost-effectiveness analysis in the field of infection control, taking into account at least the cost of prolonged hospital stay, are not performed. In the study patient population occurrence of BSI significantly increases length of stay in NICU, by approximately 20 days [21].