Comparison of antibiotic and acyclovir usage before and after the implementation of an on-site FilmArray meningitis/encephalitis panel in an academic tertiary pediatric hospital: a retrospective observational study

Acute meningoencephalitis can be caused by a variety of pathogens. In case of bacterial or herpes simplex virus (HSV) infection, early initiation of antibiotics or acyclovir is essential and associated with better outcomes [1‐4]. However, in febrile infants undergoing evaluation for meningoencephalitis, the most common infectious agents found are viruses other than HSV, which usually cause self-limiting diseases, do not require anti-infective therapy and are not affected by the treatment with antibiotics and/or acyclovir [5].

It is often difficult to clearly differentiate between the disease-causing organisms using clinical or laboratory information (such as cerebrospinal fluid (CSF) cell counts, inflammatory markers etc.). Particularly in young infants, symptoms are often unspecific and may overlap, [5, 6] while no reliable biomarkers for diagnosing bacterial infections are available [7]. Therefore, early initiation of anti-infective therapy is common practice, [8, 9] resulting in unnecessary usage of antimicrobials. However, parenteral antibiotic and acyclovir administration can lead to serious adverse effects, such as catheter-associated complications, [10] and side effects, such as allergic reactions, [11] diarrhea [12] and nephrotoxicity [13]. In addition, antibiotic-associated changes in the child’s microbiome have been shown to have sometimes long term consequences on the patient’s health [14].

New molecular methods, such as multiplex PCR (mPCR) tests, have been increasingly introduced into clinical routine procedures to allow for simultaneous and more rapid testing for a variety of pathogens. Several authors have already suggested a positive effect of a FilmArray Meningitis/Encephalitis Panel (FA ME Panel) (i.e. mPCR) on empiric treatment. Quick verification or exclusion of the presence of organisms may enable clinicians to early optimize antimicrobial therapy and hence possibly reduce therapy-associated complications and healthcare costs [15‐18]. In a previous study conducted at our institution we retrospectively analyzed all patients receiving mPCR testing over the period of one year [19]. To further investigate the impact of the FA ME Panel on empiric antibiotic and acyclovir usage in children with suspected meningoencephalitis, we decided to perform a retrospective observational study using a historical control group of patients prior to the implementation of the FA ME Panel.

In this study, we compared empiric anti-infective usage before and after the implementation of a FA ME Panel in a pediatric hospital. Our data indicate that the introduction of an on-site mPCR into clinical routine procedures is associated with reduced empiric therapy in children with suspected meningoencephalitis. Overall, LoT and DoT of antibiotics and acyclovir were significantly lower. When stratifying for age, a significant reduction for LoT and DoT of antibiotics was only observed for infants, while acyclovir treatment was significantly shorter for both, infants and older children.

Our results suggest that the implementation of rapid molecular testing for meningoencephalitis in pediatric hospitals can lead to earlier optimization of empiric therapy, as seen in several recent studies [26‐29]. However, in contrast to our study, these reports refer to single PCR assays. By using a mPCR, this effect might be even greater because of the simultaneous testing for a variety of pathogens. Recent studies by Rogers et al. and Subramony et al. have shown the benefits of combining several etiological organisms in one mPCR for acute respiratory tract diseases [30, 31]. Another study recently published, evaluated the impact of the FA ME Panel on antibiotic therapy in children with confirmed CNS infection by comparing antibiotic usage before and after its introduction. However, only patients with a discharge diagnosis of meningitis or encephalitis were included [32]. In contrast to that study, our study includes all patients with suspected CNS infection and also analyzes the usage of antiviral agents.

During the mPCR period, the majority of pathogens detected in CSF were enteroviruses and HHV-6, that usually cause self-limiting diseases and only need supportive care [33]. This is in line with the published literature where enteroviruses and HHV6 are the most common detected pathogens [25, 34]. There were two cases of bacterial meningitis and no case of HSV encephalitis, confirming that the incidence of these infectious organisms in children is low [35‐37]. However, as they are associated with high morbidity and mortality, especially when anti-infective therapy is delayed, prompt initiation of empiric treatment is common practice [1‐4]. This often leads to unnecessary usage of antibiotics and acyclovir, associated costs and side effects. Moreover, Gaensbauer et al. noticed an increase in acyclovir usage, while no increase in HSV diseases was observed, [37] hence, further highlighting the necessity for rapid diagnostic testing. It has been shown that PCR results may be negative very early in HSV encephalitis [38]. Thus, in patients with high suspicion of HSV infection due to clinical findings and anamnesis, careful interpretation of a negative HSV PCR result is required before discontinuing empiric acyclovir therapy. In these cases a second lumbar puncture and repeated testing might be necessary.

The biggest difference between both study groups regarding anti-infective usage was observed in infants. In this age group most viral pathogens were found and LoT and DoT of antibiotics and acyclovir, were significantly lower. In each study group, most infants were younger than three months old. These patients often present with unspecific symptoms and usually receive empiric therapy while undergoing several diagnostic procedures, including lumbar puncture [5]. In older patients, symptoms are more specific and thus these can be managed without anti-infective treatment more often [39]. Moreover, the highest incidence of bacterial meningitis was found to be in infants below the age of six months [40]. Therefore, additional rapid molecular testing may be of greater benefit for young infants.

The implementation of our in-house mPCR has facilitated more rapid pathogen detection, while covering a broad range of infectious organisms and raising awareness for other viral agents. Prior to the implementation of the FA ME Panel, tests for viruses other than HSV were rarely ordered. This lack of awareness for other viral agents was also seen in other institutions [16, 41‐43]. Hence, many viral CNS infections are likely to have remained undetected, resulting in unnecessary continuation of anti-infective therapy. Since the introduction of the FA ME Panel, all patients with suspected meningoencephalitis are tested for 14 different pathogens, including 7 viruses, 6 bacteria and a yeast. Moreover, in our institution testing for viral pathogens used to be sent out to a reference laboratory, usually taking between two and five days to receive results. By implementing an on-site PCR, results are available sooner than before, hence enabling clinicians to earlier adapt therapy procedures. Thus, it is likely that the increase in the number of detected viral pathogens and the reduction in anti-infective therapy seen in our study are a result of both, frequent testing for a wider variety of pathogens and more rapid pathogen detection. However, during the mPCR period only two of seven possible viral pathogens included in the mPCR were detected. This might raise the question if using in-house PCR assays for HSV, enterovirus and HHV-6 instead of using the mPCR, may be more cost effective in some institutions. Some authors even implied that the sensitivity of a singleplex assay to detect viral agents is greater than that of a mPCR [15, 17, 44]. However, this approach involves the risk of missing a viral pathogen and requires a certain level of medical experience. To further investigate this questioning, additional prospective (multicenter) studies with longer time periods are needed to possibly include more detected infectious agents.

In our study, we found a significant difference in the number of patients with confirmed viral CNS infection (5/46 (10.9%) in the control vs. 14/46 (30.4%) in the mPCR group, p = 0.038). The number of patients being admitted during enteroviral season was similar in both study groups (21/46 (45.7%) in the control and 26/46 (56.5%) in the mPCR group, p = 0.404). Thus, comparisons regarding enterovirus results and testing orders are not biased by seasonality. Comparing the relative frequency of patients with detected enteroviruses in relation to the number of patients receiving enteroviral PCR testing between both study groups, a higher percentage can be seen in the control group (27.8% in the control vs. 19.6% in the mPCR group). However, the overall higher detection rate of enteroviruses after the implementation of the mPCR (10.9% in the control vs. 19.6% in the mPCR group) underlines that a targeted approach using singleplex assays involves the risk of missing viral agents. In previous studies, rapid detection of enteroviruses in pediatric patients was already shown to be associated with reduced antibiotic usage [20, 29, 39, 45]. Hence, the higher detection rate of enteroviruses after the implementation of the mPCR, is likely to have contributed to the reduction in anti-infective therapy seen in our study.

In contrast to the mPCR group, no HHV-6 was found in the control group. We suggest this being mainly due to the limited number of tests being ordered for HHV-6 by clinicians (n = 2). However, the role of this viral agent regarding CNS infections remains unclear. HHV-6-positivity may represent a primary infection, a latent state of infection, a reactivation or chromosomal integration [25, 46]. Hence terminating empiric anti-infective therapy based on a positive HHV-6 result in CSF alone is not appropriate. The significance of HHV-6 positivity should be interpreted in the context of the patient, including clinical symptoms, immune status, laboratory results and cranial imaging [25, 46].

Some authors have suggested that the positive impact of rapid testing on antibiotic and acyclovir reduction may be even increased with faster turn-around-time [26, 29, 39, 47]. In our study rapid verification of a viral pathogen by the mPCR enabled clinicians to withhold anti-infective therapy for two infants, as results had been available prior to administration (1 enterovirus and 1 HHV-6). In another case acyclovir therapy had not been initiated due to a mPCR result being negative for HSV shortly after admission. These findings demonstrate that by faster turn-around-time, antibiotic and acyclovir administration can be completely avoided, as previously seen in a study by Van et al. [26] Our mPCR is not run outside the microbiology laboratory working hours. If it was run 24 h a day, 7 days a week, its potential influence on empiric therapy might be even greater [26]. However, it is only feasible to withhold anti-infective treatment if the patient is clinically stable and the CSF cell count is either normal or moderately increased [48]. Due to the often rapid course of meningoencephalitis we hospitalize these children and monitor them until CSF culture results are negative after 48 h of incubation. Furthermore, despite faster PCR turnaround times, several difficulties might be encountered in the clinical setting that make it difficult for pediatricians to completely withhold anti-infective therapy. Pediatric patients, especially infants, often present with unspecific symptoms that make it difficult to distinguish between viral and bacterial infection and there is always the risk for false positive or negative results [5, 6, 22]. In addition, the detection of a viral infection does not rule out a concomitant bacterial infection [16].

We did not observe a significant reduction in length of stay after the introduction of the FA ME Panel into our clinical routine setting. Despite the significant higher amount of proven viral CNS infections for infants in the mPCR group (12/29 (41.4%) in the mPCR vs. 3/29 (10.3%), p = 0.015 in the control group), median hospital LOS for infants was 6.0 days in both study groups. As previously described by Archimbaud et al. [39], enterovirus positive infants were often not immediately discharged after pathogen detection. These infants are kept for observation until having recovered and appearing clinically well.

Our study has several limitations. First, this is a single center study, which means that our results may not be representative of other hospitals. Second, the sample size was small. Third, by its retrospective nature, there is always the risk for information bias and missing data. Moreover, by excluding patients with bacterial meningitis, we cannot make any conclusions on the impact of the FA ME Panel for these children. During the mPCR period, only two bacterial CNS infections were detected. In both cases the mPCR was positive for bacterial pathogens the same day as lumbar puncture and confirmed by conventional culture. In both cases, the positive result has not changed the empiric therapy, and culture confirmation is necessary to assess antimicrobial susceptibility to optimize treatment. Furthermore, we focused on patients with high suspicion of CNS infection, hence all were receiving anti-infective therapy. Patients, for whom bacterial or HSV meningoencephalitis was excluded based on presenting symptoms or CSF analysis were not included in our study. In these cases, the mPCR cannot influence empiric therapy, and is therefore of lesser importance in these situations. For 32/46 (69.6%) children the FA ME Panel showed negative results. Confirmatory testing by singleplex PCR was only performed in 8/32 (25.0%) patients. All of these were tested for HSV and 7 of these also for enterovirus. All were confirmed negative. However, for the other patients with FilmArray negative samples, no confirmatory testing was performed.

To summarize, our study provides data regarding the potential influence of a FA ME Panel on anti-infective therapy after its implementation in a pediatric setting. Our results suggest that in children with suspected meningoencephalitis that are treated with empiric antibiotics and/or acyclovir, rapid results provided by a mPCR can possibly reduce anti-infective therapy or prevent treatment in the first place.