Characteristics and outcomes of culture-negative versus culture-positive severe sepsis

Severe sepsis is a major cause of morbidity and mortality in both developed and developing countries [1]. Mortality rates remain high at 30% and rise to 60% in the presence of septic shock despite significant advancement in treatment modalities [2]. Bacteria are by far the most common causative microorganisms in sepsis, and cultures are positive in about 50% of cases [3]. Failure to administer antibiotics to which the pathogens are susceptible is associated with increased mortality [4]. Thus, early broad-spectrum antibacterial agents are recommended as a means to improve survival [5].

Less is known though about the other half of the equation: sepsis for which etiologic agents are not found. It is commonly thought that cultures may lack the sensitivity to detect all infecting bacteria [6]. Beyond this, and aside from data from a few multicenter epidemiological studies, which suggest that severity of illness and mortality are not significantly affected by microbiological documentation in sepsis [7‐12], the medical literature is surprisingly devoid of information about patients with culture-negative sepsis.

The aim of our study was hence to compare the characteristics and outcomes of culture-negative versus culture-positive severe sepsis.

The study included 415 culture-negative patients (41.5%) and 586 culture-positive patients (58.5%) who were admitted to our ICU for severe sepsis. Table 1 describes their characteristics at baseline and on day one of the ICU stay. Compared to culture-positive patients, culture-negative patients were more likely to be women, have fewer comorbid conditions, less tachycardia, higher blood pressure, lower procalcitonin levels, lower APACHE II and SOFA scores, and less cardiovascular, central nervous system, and coagulation failures. Culture-negative patients were less likely to be treated with vasoactive agents on the first day of ICU stay.

As shown in Table 2, the lungs were commoner sites of infection, while liver abscesses, biliary tract, urinary tract, soft tissue and skin infections, infective endocarditis and primary bacteremia were less common in culture-negative than in culture-positive patients.

Table 3 lists the cultures performed within the two days before and the two days after ICU admission and the culture positivity rates. While more cultures were obtained from bile, liver abscesses, and soft tissue and skin in the culture-positive group than in the culture-negative group, there were no significant differences in the proportion of patients for which other cultures were performed in the two groups. Blood, urine, and endotracheal aspirate cultures were most frequently performed. A median of two blood cultures, one urine culture, and one endotracheal aspirate culture were obtained for both the culture-negative and the culture-positive patients.

Table 4 features the microbiology. Gram-positive bacteria were isolated in 257 patients (25.7%) while Gram-negative bacteria were isolated in 390 patients (39.0%). Among these patients, 196 (19.6%) had only Gram-positive infections, 329 (32.9%) only Gram-negative infections, while 61 (6.1%) had mixed Gram-positive and Gram-negative infections. Staphylococcus aureus and Klebsiella pneumonia were the commonest Gram-positive and Gram-negative microorganisms respectively.

Patient outcomes are presented in Table 5. Culture-negative patients had a shorter duration of hospital stay, and lower ICU mortality and hospital mortality (35.9% versus 44.0%, P = 0.01) than culture-positive patients.

Table 6 details the variables associated with hospital mortality. While culture positivity was associated with higher mortality on univariable analysis, it did not feature as an independent predictor of mortality after accounting for other covariates on logistic regression analysis. The same applies to the administration of inappropriate antibiotics on the day of ICU admission. Multivariable analysis revealed the following independent predictors of mortality: age, time from hospitalization to ICU admission, lung, bone and joint infections, infective endocarditis, primary bacteremia, Acute Physiology Score, coagulation and hepatic failures, and mechanical ventilation on the day of ICU admission. The logistic regression model fitted well and there was no multicollinearity. In a separate analysis including specific microorganisms, Pseudomonas aeruginosa was the only pathogen which, when isolated independently, increased mortality (odds ratio (OR) 2.02, 95% CI 1.08 to 3.79, P = 0.03).

Among the culture-positive patients, 265 were nonbacteremic while 321 were bacteremic. Hospital mortality was similar in both subgroups. The nonbacteremic but culture-positive subgroup had a higher mortality rate than the culture-negative group but this was not statistically significant; the bacteremic subgroup had a significantly higher mortality rate than the culture-negative group (Figure 1). Again among culture-positive patients, 467 received appropriate antibiotics on the first day of ICU stay while 119 did not. Hospital mortality was higher in the latter subgroup. The culture-positive subgroup that received appropriate antibiotics had a higher mortality rate than the culture-negative group but this was not statistically significant; the culture-positive subgroup that did not receive appropriate antibiotics had a significantly higher mortality rate than the culture-negative group (Figure 2).

To the best of our knowledge, no previous study has focused on the differences between culture-negative and culture-positive severe sepsis. The main findings of this study are that patients with culture-negative sepsis had fewer comorbidities and lower severity of illness than those with culture-positive sepsis. Although culture-negative patients had a shorter hospitalization and lower ICU mortality and hospital mortality than culture-positive patients, culture positivity per se was not independently associated with mortality on multivariable analysis.

Causative microorganisms were not found in 41.5% of our patients. Multicenter studies published in the last decade found that culture-negative patients accounted for 28%, 35%, 38%, and 48% of all cases of severe sepsis in North American, Spanish, French, and Canadian ICUs, respectively [10, 16‐18]. The corresponding figure was 40% in the pan-European Sepsis Occurrence in Acutely Ill Patients (SOAP) study [11]. In the United States, 49% of hospitalized patients with sepsis were culture-negative [3]. Meanwhile, the Extended Prevalence of Infection in Intensive Care (EPIC) II study, which reported prevalence - and not incidence - found that 30% of all infections in ICUs worldwide were culture-negative [19]. While we found more Gram-negative pathogens, others have found a predominance of Gram-positive microorganisms [3, 11].

Despite the high prevalence of culture-negative sepsis, studies which focus on the outcomes of such patients are surprisingly limited. In the 1990s, Rangel-Frausto and colleagues found a mortality of 25% among 577 patients with severe sepsis and septic shock in a teaching hospital and no difference in outcomes between culture-negative and culture-positive patients [9]. Mortality was also similar among 310 culture-negative and 742 culture-positive patients with severe sepsis in a multicenter study by Brun-Buisson and colleagues [8]. In the 2000s, a North American study led by Kumar and colleagues found similar mortality among 608 culture-negative and 1,546 culture-negative patients [10], while the pan-European SOAP study found similar ICU mortality rates (40% versus 39%) between 468 culture-negative and 454 culture-positive septic patients [11]. Our study, on the other hand, found a significantly lower hospital mortality rate for culture-negative than for culture-positive severe sepsis (35.9% versus 44.0%). Although this difference was greater for bacteremic patients than for culture-positive but nonbacteremic patients, there was no significant difference in mortality between these latter two subgroups (43.0% versus 44.9%). The reasons why we found a difference in mortality when the earlier studies did not are not immediately clear. These studies were not primarily designed for such a comparison and did not provide details on the severity of illness in these two groups of patients. In Rangel-Frausto et al.'s cohort, culture-negative patients were less likely to have acute kidney injury and shock, but in Brun-Buisson et al.'s cohort, culture-negative patients had more hypotension [8, 9]. In our study, culture-negative patients were clearly less sick than their culture-positive counterparts: they had fewer comorbidities, less hemodynamic instability and organ failures, and lower APACHE II scores. After adjustment for covariates including severity of illness, identification of microorganisms was not independently associated with mortality, a finding similarly reported recently by the French OUTCOMEREA database [12].

The question that then begs to be answered is: what exactly is the cause of culture-negative sepsis? Is it merely a milder form of sepsis compared to culture-positive sepsis? While our study cannot answer these questions, its findings provide some insight into four possibilities. First, it is known that cultures lack the sensitivity to identify all bacteria. Postulated reasons include prior antibiotic exposure, sampling error, insufficient volume for blood cultures, poor transport conditions, and slow-growing or fastidious bacteria [6]. Polymerase chain reaction (PCR)-based molecular techniques may improve detection rates, and many patients with clinical sepsis are indeed PCR-positive but culture-negative [20‐22]. In the PIRO system for staging sepsis, the letter 'I' refers to the nature and extent of the infection [23]. It may be hypothesized that the lower severity of culture-negative sepsis in our study was at least in part due a milder insult and lower bacterial burden [24], and correspondingly, the inability to capture the microorganisms on cultures. While it is possible that antibiotic pretreatment might have contributed to negative cultures, this is less likely for blood cultures that were usually performed before antimicrobial therapy. In addition, although we did not differentiate community-acquired from hospital-acquired infections, the short median (interquartile range (IQR)) lag time from presentation to the hospital to ICU admission of 0 days (0 to 2) suggests that most patients had the former, where the incidence of antibiotic pretreatment is likely to be lower. The letter 'P' in the PIRO system refers to predisposition [23], and given the trend toward more culture-positive sepsis among diabetic patients in our cohort, it is conceivable that diabetics are more prone to having large bacterial loads [25].

Second, it is clear from our data that certain infections are less common in culture-negative than in culture-positive patients, and vice versa. This is in part due to the nature of some infections, for example, liver abscesses are less likely to be culture-negative [26], and in part due to the diagnostic criteria for other infections, for example, the importance of blood culture positivity for primary bacteremia and in the revised Duke criteria for infective endocarditis.

Third, some of our culture-negative patients might have had nonbacterial sepsis. Fungi account for approximately 5% of cases of sepsis in ICUs and are generally more readily detected than viruses and parasites [16, 17]. Our study design mandated the exclusion of such microorganisms. While we are confident that most patients with fungal sepsis were diagnosed and thus excluded, and that parasites are extremely rare in our urban setting, it is plausible that undetected viruses contributed to a significant proportion of culture-negative sepsis. Using reverse-transcription PCR assays on bronchoalveolar lavage fluid and nasopharyngeal swab specimens, Choi and colleagues recently demonstrated that severe pneumonia was due to viruses 36% of the time [27]. It is not routine clinical practice in most ICUs, including ours, to test for viruses in pneumonia, and the lungs were a commoner site of infection in culture-negative than in culture-positive patients in our study. Meanwhile, serum procalcitonin levels, which are a marker of bacterial as opposed to viral infections, were lower in our culture-negative group, again bringing up the possibility that viruses played a role [28].

Importantly, a fourth conceivable explanation for our culture-negative group is that some of the patients did not actually have sepsis. The 1992 ACCP/SCCM Consensus Conference criteria that we and many other investigators used to define sepsis [11, 17, 18] - as well as subsequent adaptations - are based on a composite of clinical and laboratory data and will inevitably include a heterogeneous set of diagnoses, some of which are false positives and unrelated to infections [13, 14, 23]. In a study by Heffner and colleagues, 32% of culture-negative patients who were initially identified as having severe sepsis in the emergency department were subsequently found to have noninfectious mimics while 16% had illnesses of indeterminate etiology [29]. Among our culture-negative patients, 74.5% had a lung infection, compared to 64% in the European SOAP cohort [11]. This might have been due to the fact that ours is a medical ICU, but one could also postulate that some of our patients had mimickers of pneumonia such as heart failure. However, because we used all available clinical information from ICU admission till discharge or death before labeling severe sepsis, including noninvasive and invasive hemodynamic monitoring where indicated, it is likely that the proportion of patients who did not actually have sepsis in our study was smaller than in Heffner and colleagues' cohort.

What are the implications of our study's findings? Early administration of broad-spectrum antibiotics is recommended by the Surviving Sepsis Campaign guidelines in an effort to improve outcomes in sepsis, culture-positive or culture-negative [5]. Kumar and colleagues showed that every hour of delay in the administration of effective antibiotics from the onset of septic shock resulted in an increase in mortality [10]. This association occurred even among culture- negative patients, for which antimicrobial therapy was deemed appropriate if they were consistent with national guidelines modified to local flora for the clinical syndrome. In contrast, similar to the majority of studies in the literature [4], our study defined the administered antibiotics to be inappropriate only if they did not match the in vitro susceptibility of the identified pathogens, that is, only in culture-positive patients. In so doing, we found the mortality rate to be much higher among culture-positive patients who were administered inappropriate antibiotics than culture-negative patients (55.5% versus 35.9%). There was also a trend toward higher mortality among culture-positive patients who received appropriate antibiotics (41.9%) than culture-negative patients. These findings do not invalidate the Surviving Sepsis Campaign's recommendation, especially since it is impossible to accurately predict one's culture status at presentation. On the clinical front, we echo the Surviving Sepsis Campaign guidelines' advice for cautious consideration of antimicrobial therapy using clinician judgment and available clinical information when cultures are unrevealing [5]. On the research front, we propose that it is time for more in-depth studies of culture-negative sepsis. Such investigations could come in the form of multiplex PCR amplification techniques for the quantification of bacteria, fungi, and viruses to elucidate the false-negative and true-negative rates of cultures [20, 24], and interventional trials comparing algorithms to escalate, continue, narrow, or cease antibiotics coupled with a search for noninfectious etiologies when pathogens are not detected [30].

Our study has several limitations. First, it compartmentalizes sepsis into two main groups based on the identification, or lack thereof, of pathogenic microorganisms, but in reality, both groups are a mixed bag of diagnoses [31]. As discussed at length above, the culture-negative group probably included some patients with nonbacterial sepsis and patients without sepsis. Nonetheless, the incidence of culture-negative sepsis in our cohort mirrors that in multiple studies internationally [3, 10, 11, 16‐18], and this reinforces the need to better understand this real-world phenomenon. As for the culture-positive group, Pseudomonas aeruginosa was the only bacteria that independently increased mortality, a finding that has also been seen in European ICUs [11]. Besides, bacteremia is often seen as a harbinger of bad outcomes and thus may not be equivalent to the identification of pathogens in bodily fluids other than blood [31, 32]. This notwithstanding, our analyses suggest no difference in mortality between bacteremic and nonbacteremic culture-positive sepsis. Second, because we left the performance of cultures to the discretion of the managing physicians, we are unable to rule out the fact that some patients were culture-negative simply because of inappropriate sampling. Nonetheless, given that most cultures were performed in an equal proportion of patients in the culture-negative and the culture-positive groups, this is less likely to be a significant contributing factor to our findings. Third, as ours was a single-center study conducted in a medical ICU, the details of its findings may not be extrapolated to all ICU patients. Fourth, to determine the appropriateness of antimicrobial therapy, we did not record the exact timing of administration while reviewing the antibiotics given on the first day of ICU stay. This timeline is longer than the three-hour window period from the onset of sepsis suggested by the Surviving Sepsis Campaign [5]. Nonetheless, we found in a previous study that antibiotics were administered within three hours 75.0% of the time at our ICU [33]. Fifth, as our study was observational in nature, unadjusted and hidden confounders might have influenced our results and conclusions. To illustrate, although it is unlikely that significant differences exist in the acute management of the culture-negative and the culture-positive groups, especially when microbiological results would not have been available at presentation, it should still be acknowledged that data on treatments such as fluid resuscitation are lacking. Our study also has several strengths. It is the first and largest prospective epidemiological study dedicated to the difference between culture-negative and culture-positive severe sepsis. We only included microorganisms that were deemed by the managing physicians to be pathogens as opposed to colonizers or contaminants, with reference to the International Sepsis Forum Consensus Conference guidelines [14]. To optimize accuracy, data checks were performed by the investigators.

Our study identified significant differences between culture-negative and culture-positive severe sepsis, with the former group having fewer comorbidities, milder severity of illness, shorter hospitalizations, and lower ICU mortality and hospital mortality. However, after adjusting for all covariates, culture positivity did not independently predict mortality.