An overview of positive cultures and clinical outcomes in septic patients: a sub-analysis of the Prehospital Antibiotics Against Sepsis (PHANTASi) trial

Sepsis remains one of the most important causes of morbidity and mortality worldwide [1]. Every year, over 30 million persons develop sepsis of which more than five million die [2]. Although mortality rates have been decreasing, the incidence of sepsis continues to increase [2]. This is due to several factors, among which are as follows: an aging population, antibiotic resistance, improved recognition of sepsis, and use of immunotherapy [2, 3].

One of the cornerstones of the diagnosis and treatment of sepsis is the collection of (blood) cultures. This is of importance since detection of the organism that caused sepsis provides possibilities for targeted antimicrobial therapy [1]. However, culture-negative sepsis is common, and for example, Kumar et al. [4] found 29% cases of culture-negative septic shock in their cohort. Bernard et al. [5] found a negative blood culture in approximately 68% of cases with severe sepsis.

Previous studies have focused on the association between (blood) cultures and patient outcomes [6‐10]. However, these studies did not produce consistent results and were only conducted in subpopulations, such as those with septic shock who were admitted to the intensive care unit (ICU) [7]. No prior studies on this subject have been conducted in septic patients who were transported by Emergency Medical Service (EMS). It is known that these septic patients who were transported by EMS are more seriously ill than those transported otherwise. Thus, sepsis patients that were transported by EMS personnel are a representative sample of the general sepsis population [11]. As these patients often require urgent care, it is quintessential to determine the relationship between (blood) culture outcomes and mortality in this group of patients in order to optimize their treatment.

Several theories have been described in literature to formulate an explanation for the low yield of micro-organisms in sepsis patients, such as prior antibiotic treatment, insufficient sampling, transport problems, and insufficient techniques. Finally, it is speculated that not all patients actually have sepsis but an alternative diagnosis [6‐8]. Although previous studies have suggested that prior oral antibiotics administration might lead to decreased (blood) culture yield, limited research has been conducted to assess this association [6‐8]. In an effort to concur the above sketched clinical dilemmas, the primary aims of this study were to investigate the association between culture result (positive or negative) and 28-day mortality, 90-day mortality, and the number of organ systems affected in sepsis patients. Another primary aim was to determine what the effect is of prior administration of antibiotics on culture outcome. The secondary aims of this study were to describe the association between the administration of pre-hospital antibiotics (2 g ceftriaxone in the ambulance) and mortality in culture-positive sepsis patients as well as describing the different specimen sources identified in the various types of cultures.

Patients, who had a culture-positive sepsis, had a significantly higher mortality rate than those who had a culture-negative sepsis. We found culture-negative sepsis in 56.4%, and this incidence in our cohort was higher than that seen in earlier studies that describe incidences of 30.6–47.1% [6‐8]. Several factors might have led to this difference in the outcome. We excluded cultures that were most likely positive due to contamination. Apart from this, we included all sepsis patients, and not only those with severe sepsis or sepsis shock. This in turn might have led to our patients having a lower bacterial burden and thereby making it more difficult to detect a pathogen with a conventional culture. Another possible explanation is that the administration of pre-hospital antibiotics in the intervention group contributed to the higher rate of culture-negative sepsis. Culture-negative sepsis was seen more often in the intervention group compared to the control group (59.2% vs 53.9%).

When comparing culture-positive patients with culture-negative ones, we saw several notable differences between these groups. Culture-positive patients had higher qSOFA scores, a higher severity of illness, a longer hospital LOS, and a higher level of the CRP. These clinical parameters suggest that culture-positive patients might have been sicker than culture-negative patients. Furthermore, higher mortality rates were found in culture-positive patients compared to culture-negative patients, and the culture-positive patients more often had dysfunction of ≥ 3 organ systems. All these findings suggest a higher burden of disease in the culture-positive group. Although we cannot completely rule out the possibility of patients having a positive culture status due to colonization of specimens in non-sterile culture sites, our findings were confirmed by a sensitivity analysis after exclusion of positive rectum cultures.

Several earlier studies were conducted on the association between culture-positive status in sepsis patients and mortality. Like in our cohort, Kethireddy et al. [7] found less impressive clinical parameters in culture-negative septic shock patients compared to culture-positive counterparts. However, survival was similar in culture-negative patients and culture-positive patients. Interestingly, a delayed administration of antibiotics led to a similar increase in mortality in both groups, supporting the hypothesis that culture-negative sepsis patients do not have an alternative non-infectious diagnosis. Gupta et al. [6] also looked at the association between culture status and mortality and found a higher mortality rate in culture-negative severe sepsis patients. A possible explanation given is the lack of guidance for specific antibiotic treatment in culture-negative patients. However, as selection of culture-positive patients was based on codes in patient records, undercoding of specific pathogens might have led to incorrect classification of patients into the culture-negative group [6]. Like in our study, Phua et al. [8] found higher severity of illness, a longer LOS, and more organ dysfunction in culture-positive severe sepsis compared to culture-negative severe sepsis. However, no independent association between a positive culture and mortality was found in a multivariate model. As the authors describe this model included all covariates available at baseline, inclusion of covariates that are to be part of the causal pathway of sepsis might have led to overadjustment [8, 19]. Therefore, we chose not to include sepsis severity in the multivariate model, although previous work has shown that the probability of finding bacteremia is dependent on the clinical context [20].

Although we think it is unlikely that patients in our cohort suffered from alternative non-infectious diagnoses because of strict inclusion criteria and retrospective chart analysis by an expert panel, study inclusion was carried out before formulation of the SEPSIS-3 criteria. Therefore, we included patients that do not fulfill clinical SEPSIS-3 criteria and do not meet the current definition of sepsis.

A possible explanation for the low mortality rates seen our cohort, next to the fact that other studies have been conducted in cohorts of patients with a higher severity of disease only (severe sepsis or septic shock), is that all EMS personnel participating in the PHANTASi trial was trained to recognize sepsis in a timely manner. Apart from this, awareness of sepsis was improved during the study period through the use of several media. This led to better recognition and improved treatment.

In the subgroup of patients that met SEPSIS-3 criteria, we only found a significant association between a positive blood culture and 90-day mortality. We can however not see these specific findings in the light of other literature, as no research has been performed in the field of culture positivity in the whole population of sepsis patients after formulation of SEPSIS-3 criteria. In our cohort, we found that 55.2% of patients that did not meet clinical SEPSIS-3 criteria (qSOFA < 2) had organ dysfunction of one or more organ systems (Table 4). Meta-analysis has shown a low sensitivity of the qSOFA score in screening for mortality outside of the ICU [21]. In a recent article, the same authors pray for utilization of SIRS criteria outside of the ICU [22]. Therefore, we do not exclude the possibility of review of the use of qSOFA score as a clinical criterion for sepsis, as fast recognition of sepsis is found to reduce mortality [23].

Insufficient techniques may have contributed to the high amount culture-negative cases of sepsis seen in general. Bloos et al. found a positive PCR (polymerase chain reaction) in approximately 20% of culture-negative sepsis patients [24]. Earlier studies have showed that multiplex real-time PCR leads to a faster diagnosis and to a reduction of the amount of days of inadequate antibiotic treatment [25‐27].

Quick adequate antimicrobial therapy is of great importance as long-term treatment with broad-spectrum antibiotics leads to resistant micro-organisms and is associated with superinfections with C. difficile and fungi [28].

We found that culture-negative sepsis was most common in respiratory tract infection and that urinary tract infection was most common in culture-positive sepsis. These findings were similar to those by Phua et al. [8]. A possible explanation could be that sepsis due to pneumonia is more often caused by viruses and/or fungi, which is not routinely tested for in most hospitals. There was a variability in culture positivity across the different hospital sites. This is probably due to the differences in pre-analytical (the time from collection to incubation) between hospital locations. Some hospitals are located in rural areas and others in urban areas, where driving time to the closest hospital is usually shorter. Earlier research has shown that longer pre-analytical time leads to lower blood culture positivity [29]. Moreover, guidelines for antibiotic treatment differ between different hospital location, which might also influence culture results.

Of note, in culture-positive sepsis patients, higher mortality rates were observed in the intervention group compared to the control group. A possible explanation for this could be that EMS personnel deliberately assigned the intervention group protocol to sicker patients.

An association between prior administration of antibiotics at home and a negative pre-hospital culture was not seen. However, as the amount of patients with a positive pre-hospital blood culture was low (13.8% of patients in the invention group), we cannot exclude the possibility that due to obtainment of a culture early in the disease process, a lower disease burden at that time may have influenced the results.

Our study contains several strengths. Firstly, to the best of our knowledge, this is the first study which investigates culture outcomes in all sepsis patients, instead of a subset of patients [6‐8]. In addition, we excluded cultures that were likely false-positive due to contamination which was not done in two of the previously mentioned studies [6, 7].

Furthermore, we used data of a large multi-center trial that consisted of patients from both urban and rural areas which allow our results to be extrapolated easier to other countries.

Our study also holds limitations. First, we did not have data on the levels of the pro-calcitonin. This would have yielded extra information, as this biomarkers is found superior to the CRP for the diagnosis of sepsis in most studies [30].

Secondly, choosing a cut-off of 3 in analyzing the amount of organ dysfunction might be considered somewhat arbitrary. However, earlier research has shown that involvement of ≥ 3 organ systems is associated with an in-hospital mortality rate of more than 50% [31].

Our results show that culture-positive sepsis is associated with a higher mortality rate and that culture-positive sepsis patients more often have ≥ 3 organ systems affected during the sepsis episode. Future studies should focus on the etiological mechanisms that lead to the differences in culture outcomes in sepsis patients in order to further reduce mortality rates of sepsis.