This study provides further insight to the aetiological diagnosis and distributions of in-hospital CAP in a low antibiotic resistance prevalence setting. Aetiology was confirmed by routine microbiological testing in 29.1 % of included cases. In addition, strategic diagnostic stewardship measures demonstrated that efforts to target microbiological sampling frequencies and techniques turned out successful, in terms of enhanced microbiological diagnostic yield.
The proportions of patients that routinely underwent procedures to collect samples from the respiratory tract were comparable to previous studies [3, 9, 12-14]. A diagnostic yield of 4.9 % in blood cultures and 39.5 % in cultured respiratory tract samples is also within the range found in comparable studies. It is noteworthy that a particularly high overall diagnostic yield of 51.7 % was observed in expectorated or induced sputum, and this outperformed other anatomical sampling sites.
Expectorated or induced sputum is the standardized procedure for tuberculosis management. However, it’s role in common lower respiratory tract infections has faded over time, although approximately 75 % of patients can produce an adequate sputum sample at admission [15], and that sensitivity of sputum examination is >75 % for detecting bacterial pathogens [16]. In our study, expectorated or induced sputum provided considerable diagnostic yield of 51.7 %, although this diagnostic strategy was only applied to 22.3 % of patients diagnosed with CAP. Furthermore, by relatively modest interventional efforts, the diagnostic yield of expectorated or induced sputum increased to 62.0 %. We are not aware of similar results from interventions aiming at enhancing diagnostic yield from expectorated or induced sputum at a ward-level in CAP. It seems that such samples nonetheless provide valuable microbiological confirmations in CAP and should be the preferred method for respiratory tract sampling. However, sampling from the respiratory tract is inevitable hampered by infection control measures in viral pandemic situations. Our study was underpowered to detect whether diagnostic yield from lower respiratory tract samples were benefitted from increased numbers or quality of expectorated or induced sputum.
NTHi as the most prevalent CAP pathogen in our study, is in line with recent studies from Denmark [3] and Germany [17]. Traditionally, community-acquired lower respiratory tract infections in patients with structural pulmonary diseases, especially chronic obstructive pulmonary disease (COPD), are more likely to be caused by NTHi [4, 18]. Of notice, we found COPD in only 27 % of patients with NTHi infection. This may indicate that clinical practice guidelines in Nordic countries underestimate the prevalence of NTHi infections in CAP, and thereby offer inadequate therapy recommendations. The potential emerging relative prevalence of NTHi in CAP, may be related to pneumococcal vaccination, although an absolute increase is also possible [19].
No clinical signs or symptoms in acute respiratory tracts infections are pathogen specific. International guidelines on diagnostic strategies and antimicrobial therapy in in-hospital CAP often favour thorough microbiological evaluation and testing, in particular in severe infections [20]. Even so, exposure to special transmission settings, underlying comorbid conditions, and disease severity all represent considerable pitfalls to microbiological testing and empiric antimicrobial therapy outcomes. In our study, the lack of a consistent diagnostic testing strategy was evident at the study sites.
A recent review claims that representative respiratory tract secretions applied to highly sensitive nucleic acid amplification tests (NAAT) today have the capacity to detect common viral and bacterial pathogens as well as selected drug-resistant determinants [21]. Turnaround time for NAAT tests targeting multiple viral and bacterial pathogens are increasingly rapid and may decline to minutes. In terms of antimicrobial stewardship, a negative test may withhold empirical coverage, and a positive test may permit individualized pathogen-directed therapy. Further, efforts to establish a reliable microbiological diagnosis in pneumonia have proved beneficial in terms of clinical outcomes and resource utilization. Both mortality [22], overall antimicrobial therapy consumption [23], broad spectrum antibiotic consumption [24], infection-control practices [25], and length of stay [26], are significantly reduced by such strategy. Our study was conducted with the use of traditional cultures of respiratory tract secretions. NAAT provided aetiological confirmation in only 16 % of test in our study.
The diagnostic yield of any strategy to detect the infecting bacteria in CAP is likely to be influenced by the timing of specimen collection in view of antimicrobial therapy. In our study, 20.6 % of included cases received antimicrobial therapy before microbiological sampling. A rigorous study of CAP among immune-competent adults, demonstrated that the infecting agent was significantly more frequently detected in blood cultures prior to empirical antimicrobial therapy [12]. The same finding did not apply for respiratory tract specimens. International guidelines have previously stated that pre-treatment Gram stain and culture of expectorated sputum should be performed only if good-quality specimens can be obtained and quality performances measures for collection, transport, and processing of samples can be met [27]. In a recent published systematic review, Gram staining of sputum samples still seem to provide valuable diagnostic information, in particular for S. pneumonia and H. Influenzae detections, in an antibiotic stewardship perspective [28].
Severity assessment in pneumonia is not routinely conducted and documented in clinical practice, especially outside of intensive care settings. The CRB65-score is uniformly recommended to aid empirical antimicrobial therapy in all settings, and to assess microbiological diagnostic strategies [7]. With few exceptions, the study group calculated the CRB65-score retrospectively in our study. This may indicate that other undocumented approaches, if any, to assess disease severity, exist. In our cohort, the distributions of CRB65-score of 1 or 2 was 69-77 %, and CRB65-score of 3-4 was 4-10 % among all study sites. These findings indicate that included cases were largely non-severe CAP, and that severity did not differ significantly between study sites.
Antimicrobial stewardship measures are considered crucial to prevent harmful outcomes from antimicrobial resistance [29]. In countries with low AMR prevalence, microbiological confirmed cases of CAP allow for pathogen-directed, narrow-spectrum therapy. Of importance, only 29 % of included hospitalizations for CAP cases underwent microbiological diagnostic approach in our study. This should encourage clinicians to reinforce sampling techniques and to scale up sampling numbers, preferably lower respiratory tract secretions.
Testing for respiratory viruses in a broad panel scale is encouraged by antibiotic stewardship guidelines to reduce inappropriate antimicrobial usage [30]. This recommendation relies on studies that have classical pre- and post-intervention models, to calculate the reduction of antibiotic consumption. Other strategies, combining NAAT testing with serum biomarkers or host immune-response analyses, shows promising results [31]. We did not undertake antibiotic usage calculations in the present study. Moreover, we wanted to describe the aetiology of in-hospital CAP in a region with low prevalence of antimicrobial resistance, and to highlight that diagnostic yield from lower respiratory tract specimens may increase with the use of simple efforts to sustain adequate sampling.
The study has some limitations. Firstly, all data from included cases were extracted retrospectively. Secondly, inclusion criteria relied on the attending doctor’s ability to correctly catalogue patient data. Thirdly, we may have missed designated respiratory tract specimens collected in primary health care settings prior to hospitalization. Fourthly, details on the individual patient’s ability to comply with testing strategy recommendations were not available. Fifthly, respiratory tract samples were stored overnight, and for three hospitals transported to the laboratory before handled. Finally, microbiology results may be affiliated by the non-identical in-house laboratory protocols and procedures among the laboratories.
In conclusion, this study shows that modest efforts to scale up sampling frequencies and enhance sampling techniques, provided significantly more microbiological confirmations in in-hospital CAP. Also, expectorated or induced sputum outperformed other respiratory secretions. We advise others to conduct similar interventions in order to establish rigorous cost-benefit analyses for the role of such interventions, and to calculate the potential reduction of antimicrobial consumption.