This retrospective analysis of laboratory tests in hospitalized CAP patients did not show significant differences in the identified micro-organisms between patients with and without goat and poultry farms in the vicinity of their home address.

Based on previous research, in which a shift in the respiratory microbiome with higher abundance of S. pneumoniae in hospitalized CAP patients close to poultry farms was observed [3], we hypothesized to find a higher percentage of positive tests for S. pneumoniae in patients exposed to livestock. However, in our study a positive urinary antigen test for S. pneumoniae was not significantly associated with poultry or goat farm presence around the patients’ home address. An explanation might be that we used results from urinary antigen tests, a relative insensitive test compared to the 16S-rRNA based sequencing to determine S. pneumoniae presence used in the study by Smit et al. [3] Moreover, Smit et al. used a one kilometer distance as cut-off point for poultry farm presence instead of the two kilometers distance we defined. The number of people living within one kilometer of goat or poultry farms in our study was too low for proper statistical analysis. When analyzing the urinary antigen tests for CAP episodes from JBH only, we did find a statistically higher percentage of positive urinary antigen tests for S. pneumoniae in patients living close to poultry farms (15.6%) compared to patients living more than two kilometers from poultry farms (10.1%). Compared to Gelre hospital, urinary antigen tests were more often performed at JBH and a higher percentage of patients from JBH live within two kilometers of a livestock farm.

Prior studies regarding CAP and proximity to livestock farming used electronic medical records from GPs, reflecting mostly outpatient CAP. Our present study was among hospitalized CAP patients, i.e. a more severe patient population. The identified micro-organisms in these hospitalized patients might not entirely reflect the etiology of CAP patients visiting their GP. No patterns with goat or poultry farm exposure were detected for the micro-organisms included in a respiratory PCR panel (e.g. human metapneumovirus, adenovirus, rhinovirus, coronavirus, C. pneumoniae, C. psittaci and M. pneumoniae). Because of differences in diagnostic testing strategies, results from this PCR panel were only available for CAP episodes from Gelre Hospital. Unfortunately, the low prevalence of goat farms around the home addresses of these patients (3.6%) hampered the analyses of these micro-organisms due to limited statistical power.

In around 75% of the CAP episodes included in our study, no pathogens were detected. The failure to detect a pathogen has often been described in scientific literature, with estimates ranging from 20 to 45% of patients. [13,14,15,16,17] The high percentage of diagnostic failure reported in this study, might be related to the fact that our data was limited to routine diagnostics performed in these patients. A comprehensive and broad diagnostic package can increase diagnostic yield. However, we did not include test results from serology and bacterial culture, because these results often require an extra interpretation step.

During the large Q-fever epidemic in the Netherlands, from 2007 to 2010, Coxiella burnetii, was clearly an important zoonotic cause of CAP among people living close to Q-fever-affected goat farms. An observational study on CAP etiology in the Netherlands, that coincided with this Q-fever outbreak, showed that C. burnetii was the second most common pathogen in that study population. [16] However, in more recent years, after the epidemic, it was unlikely that Q-fever still played a role. C. burnetii seropositivity among CAP patients was independent from goat farm exposure and the increased CAP incidence was also found around Q-fever negative farms. [2, 4] In our study, very little to no Q-fever diagnostics had been requested in the CAP episodes included from the JBH, the hospital which is located within the former Q-fever epidemic area.

Our study was limited by multiple factors. First, the analysis was hampered by the small number of identified micro-organisms and the low prevalence of livestock exposure around some of the home addresses of patients. Therefore we only analyzed the laboratory results in relation to livestock farm exposure by crude univariate analysis. The small numbers and low prevalence also did not allow for further classification of the livestock exposure variable. We chose to handle a 2 km distance cut-off based on the results from previous research, but further classification using buffers or regression frameworks, as used in previous studies, was not appropriate. Second, as we did not have additional patient characteristics such as comorbidities, lifestyle factors and medication use, we could not address potential confounding by certain risk factors in our analyses. The age of the CAP patients was similarly distributed between the groups with and without livestock farm proximity. Finally, potential selection bias is also of concern because not every patient was tested for each pathogen. We retrospectively used laboratory data initially performed for diagnostic purposes, meaning that the data is dependent on the prevailing testing strategies in both hospitals. It might be that physicians only test for certain pathogens in the more severe patient group. Still the overall pathogen detection rate of around 25%, as presented in Table 1, was comparable between the CAP patients with and without goat and poultry farm exposure.

This explorative study did not provide information about possible causes of the increased CAP risk around goat and poultry farms. The previous associations are based on diagnostic data from GPs. In the Netherlands, around 80% of all CAP patients are managed by the GP. The causative pathogen, however, remains unknown in the majority of CAP patients as microbiological testing is not included in the pneumonia guidelines of the Dutch College of General Practitioners. [18, 19] To gather insight in CAP etiology related to livestock farming in a primary care setting, we started a prospective study in CAP patients visiting their GP including extensive microbiological diagnostics by multiplex PCR and respiratory microbiome analyses.