In this sub-analysis of a prospective observational cohort of septic patients, we found that septic patients with positive culture was associated with demographic characteristics, comorbidities, patient source, infection type and clinical features of the patients. PSM revealed that positive culture was associated with a significant decrease in mortality.
Patient characteristics including older age and male gender seemed to associated with positive culture, as was also indicated in previous studies[6, 7]. The reason for this tendency was unclear, we thought this could partially contributed to that patients with older age usually had more severe sepsis, while septic patients more severe was likely to have positive cultures, as found in the present study. However, the association between male gender and culture positive remained to be identified.
Patients from emergency medicine or internal medicine both had a tendency of negative cultures, while patients transferred from other hospitals were seemed to associate with positive cultures, as similar findings were also revealed in the previous studies[6, 7, 12]. This was consistent with the spectrum of the isolated pathogens, which was predominated by Gram-negative bacteria, usually isolated from hospital-acquired infections.
Comorbidities could affect the results of cultures, while the influence remained inconsistent. In the present study, we found that coronary heart failure and solid tumor were associated with culture positive. Previous study had found that diabetes mellitus may possibly related to a positive culture[6]. We extrapolated that some chronic comorbidities were more susceptible to have a micro-environment to have a positive culture, which could be related to a compromised immune system.
We found a higher proportion of patients with a positive culture who had priori exposure to anti-biotics. We extrapolated this could mostly due to the investigated populations, with more than half of the patients were hospital-acquired infections, and was very likely to have priori anti-biotics exposures. In a previous study, recent antibiotic exposure was found to be associated with increased hospital mortality in Gram-negative bacteremia infection with severe sepsis or septic shock, which was partly consistent with our findings[13].
The septic patients with positive culture seemed to be more severe compared with negative ones, as APACHE II and SOFA score was significantly higher in culture-positive patients. This was probably could be explained by that the patients with positive results was related with more bacteria load and was more likely to be isolated.
Previous studies had revealed that culture-positive was associated with organ dysfunctions, so in this study, we also included laboratory findings at admittance, aimed to find some laboratory parameters that could be potentially associated with positive culture. Unfortunately, we found no connections between positive culture and laboratory parameters that reflected the organ function. We found from the multivariable regression, the only factor we’ve found as significantly related to culture positive was procalcitonin, as consistent with previous studies, that septic patients had a significantly elevated serum level of procalcitonin, indicating that procalcitonin might be a predictor[14].
In a previous study, the authors introduced machine learning methodology in prediction of the culture positivity in suspected bloodstream infections[15]. However, in the in their study, the authors found the variables that has irrelevant to the clinical meanings[16], and in the present study, we found some risk factor that could associated with culture positive in septic patients.
The mortality between culture positive and negative patients remained controversial in previous studies[2, 6–8, 12, 17, 18]. in the present study, we found a higher mortality in patients with positive culture compared with negative cultures, from the crude results, however with no statistical significance (37.5% vs. 34.5%, P = 0.165). We subsequently performed a PSM, and interestingly, we found a decrease in mortality after we matching the demographic information, source, infection type and severity of the septic patients (38.2% vs. 48.2%, P = 0.012). We noted a higher proportion of patients with complete resolve (15.9% vs. 10.4%). In the previous study, the authors found that patients with positive culture more likely to receive appropriate antibiotics treatment[12, 19], and this might partially explain the higher percentage of patients with complete resolve and decrease in mortality in positive culture. As in our study, we also noticed a higher proportion of patients with targeted antibiotic treatment, both before and after PSM. We further compared the mortality between septic patients between targeted and empirical antibiotics treatment, and interestingly, we found a significant decrease in mortality after PSM (37.0% vs. 46.2%, P = 0.031), which was not observed before PSM (36.6% vs. 35.8%, P = 0.759). We extrapolated the appropriate use of antibiotics might be one of the reasons that contributed to the decreased in mortality in culture positive patients, while in the crude results, the effect of appropriate antibiotic use was masked by the cofounding factors.
We thought the factors associated with culture positive in septic patients were neither very strong predictors, nor could be manipulated in the clinical practice. On the other side, technical improvement[20, 21], multiple sending of the samples definite affect culture positive, as culture positive was very likely to be associated with decrease in mortality, we thought it would be much encouraged for a given septic patients to obtain positive culture whenever possible.
Our study had some limitations, we did not include the difference in antibiotics administration between patients with positive and negative cultures. We assumed that the discrepancies in mortality in two groups of patients was contributed to the targeted antibiotics management which was guided by the results of the positive culture, which needed further investigations. We found no factors that could intervened for to improve the culture positive.