Comparison of length of stay and outcomes of patients with positive versus negative blood culture results

Abstract

In the United States, sepsis is the leading cause of death in critically ill patients. The fatality rate for severe sepsis is about 40%, and treatment costs over $16 billion annually. It is critical to identify and treat the source of sepsis. While there are varying guidelines determining when to draw blood for culture, at Baylor University Medical Center at Dallas, blood cultures are ordered for patients with new onset of fever, immunosuppression, or a suspicion of an underlying infectious etiology. We conducted a retrospective study of patients who had blood cultures after hospital admission or in the emergency department in December 2013. We compared length of stay and outcomes of patients with positive versus negative blood cultures. There was no significant difference for length of stay or outcomes among patients with positive and negative blood cultures. For patients admitted from the emergency department, there was a longer length of stay for patients with positive cultures; however, the overall prognosis was not worse.

Up to 46% of hospitalized septic patients are admitted through the emergency department (ED) (1). Since many ill patients initially present to the ED, it is of utmost importance to assess and identify the source of infection for proper and rapid treatment. Thirty to fifty percent of patients presenting with a clinical syndrome of severe sepsis or shock have positive blood cultures; therefore, the international gold standard of practice when there is suspected sepsis is blood culture analysis (2). However, incubation of blood cultures can take anywhere from 3 to 7 days, and in cases of suspected sepsis, antibiotics are usually started empirically without waiting for pathogen isolation (3). Due to the absence of published guidelines that clearly state when blood cultures should be drawn (4) and to establish if there is a correlation between positive blood cultures and outcome, we conducted a retrospective study of inpatients with the diagnosis code for sepsis and all ED patients with blood cultures performed over 1 month. We aimed to determine if a positive blood culture predicted increased morbidity and mortality among patients presenting to Baylor University Medical Center at Dallas (BUMC) who were suspected of having sepsis.

METHOD

We searched our electronic health system for two patient populations during the month of December 2013: all hospitalized patients who had blood cultures drawn if sepsis was suspected (hospital sepsis cohort) and all patients who had blood cultures performed in the ED. Patients who had cultures drawn during hospitalization may not have had cultures drawn in the ED prior to admission.

For all patients in this study, one or more blood cultures were obtained in a BacTAlert FA (30 mL) and/or FN (40 mL) blood culture bottle (bioMérieux, Durham, NC) in the ED or in the hospital at BUMC. The samples were sent via courier to the affiliated reference laboratory, med fusion (Lewisville, Texas). Upon arrival, the bottles were placed on the BacTAlert automated blood culture system. Any positive culture was flagged and reported to caregivers. Samples were defined as negative if an organism was not detected within 5 days.

Clinical information was derived from the patient's electronic medical records. Demographic information included age and gender. Other data included type of pathogen present in blood culture, length of stay, in-hospital mortality, intensive care unit (ICU) admission, and hospital admission or readmission for those patients who presented to the ED. P values were calculated by converting the z value (from Microsoft Excel 2007) to a P value (calculated on Graphpad).

RESULTS

In December 2013, a total of 135 patients who presented to the ED at BUMC had blood cultures drawn. In addition, 54 hospitalized patients had the diagnosis code for sepsis, for a total of 189 patients. The patients' ages ranged from 19 to 101 years (mean 64). There were 89 men and 100 women.

Hospital sepsis cohort

For the hospital sepsis cohort, the patients (n = 54) were admitted through the ED but had blood cultures drawn after hospitalization. Thirty-four patients had negative blood cultures, and 20 patients had positive cultures. For those with positive cultures, the most common organisms recovered were Escherichia coli (3, 15%), alpha streptococci (viridans group) (3, 15%), and beta hemolytic streptococci (3, 15%). The remaining organisms recovered were Staphylococcus aureus (2), coagulase-negative staphylococci (2), Bacteroides fragilis (1), Proteus mirabilis (1), Klebsiella pneumoniae (1), Streptococcus pneumoniae (1), Enterococcus faecalis (1), Propionibacterium acnes (1), and Capnocytophaga canimorsus (1) (Table 1).

Table 1.

The most common organisms in the hospital sepsis cohort

Organism	n
Escherichia coli	3
Alpha streptococci	3
Beta hemolytic streptococci	3
Staphylococcus aureus	2
Coagulase-negative staphylococci	2
Bacteroides fragilis	1
Proteus mirabilis	1
Klebsiella pneumoniae	1
Streptococcus pneumoniae	1
Enterococcus faecalis	1
Propionibacterium acnes	1
Capnocytophaga canimorsus	1

Our data did not show statistically significant differences in the outcomes between hospitalized patients with positive blood cultures and those with negative cultures. Twenty-five (73.5%) of the patients with negative cultures were admitted to the ICU versus 18 (90%) of the patients with positive cultures (P = 0.335, NS). For patients with positive cultures, the average length of stay was 5.6 days (range 0–23, SD 5.44) compared with 6.5 days (range 0–29, SD 8.28) for those with negative cultures (P = 0.63, NS) (Figure 1). Eleven (32%) patients with negative blood cultures and 6 (30%) patients with positive blood cultures died during their stay (P = 0.9998, NS). Therefore, for hospitalized patients, positive blood cultures may not be a predictor of mortality or ICU admission in patients who meet the clinical criteria for sepsis.

Figure 1.

Outcomes for the hospitalized sepsis cohort. There were no statistically significant differences in the (a) length of hospital stay or (b) number of intensive care unit (ICU) admissions among patients who had positive or negative blood culture results.

ED cohort

The second group of patients consisted of those who had blood cultures initially drawn in the ED. A total of 134 patients who presented to the ED in December 2013 had blood cultures drawn. Of this group, 124 patients (93%) had negative cultures, and 10 (7%) had positive cultures. For patients with positive cultures, the most common organisms were coagulase-negative staphylococci (3), alpha streptococci (viridans group) (2), and S. aureus (2). The remaining microorganisms present were E. coli (1), P. acnes (1), and Micrococcus sp. (1).

Of the 10 patients with positive cultures, three were directly discharged from the ED. These patients had positive cultures for coagulase-negative staphylococci, P. acnes, and Micrococcus sp., which are common contaminants (5). One patient who was positive for coagulase-negative staphylococci was transferred to an outside hospital. Of the six remaining patients with positive blood cultures who were admitted to the hospital (two with S. aureus, two with alpha streptococci viridans group, and one each with E. coli and coagulase-negative staphylococci), only one patient (17%) who was positive for alpha streptococci (viridans group) was admitted to the ICU. One admitted patient with a positive culture for coagulase-negative staphylococci died (17%) (Figure 2).

Figure 2.

Outcomes for the emergency department (ED) cohort. (a) Most patients with both positive and negative cultures were admitted to the medical/surgical floor rather than the intensive care unit (ICU). (b) A majority of patients with negative and positive cultures were discharged directly from the ED or discharged after admission. Only one patient from each group died. (Note that this group excludes transfers.)

Of the 124 patients with negative blood cultures, 112 (90%) were either discharged directly from the ED or left against medical advice; one died in the ER; three were transferred to an outside hospital (2%); and eight were admitted—two (25%) to the ICU and six (75%) to the medical/surgical floor (Figure 2).

Our data showed no significant difference in ICU admission among patients with positive blood cultures compared with patients with negative cultures. However, the average length of stay of patients with positive cultures was longer than that for patients with negative cultures. Of the 9 patients with negative cultures who were admitted to the hospital, the average length of stay was less than 1 day (range 0–16 days). The average length of stay for the patients with positive cultures was 5.1 days (range 0–12; P = 0.0001). Therefore, for patients presenting to the ED, a positive blood culture could be a predictor of an increased length of hospital stay (Figure 3).

Figure 3.

In the emergency department cohort, a positive blood culture was associated with an increased length of stay compared with a negative blood culture.

DISCUSSION

Approximately 200,000 episodes of bacteremia occur in the United States each year, with an incidence of about 10 per 1000 hospital admissions (4). In fact, bacterial bloodstream infections are associated with a mortality of 14% to 37%. Bloodstream infections include infective endocarditis, primary bacteremia, and those secondary to focal infections such as pneumonia, abscesses, urinary tract infections, and osteomyelitis (4). As a result, if sepsis or any of these aforementioned illnesses are suspected in the ED, blood cultures are often liberally ordered to help direct physicians towards proper treatment. As a result, only 4% to 7% of blood cultures drawn are positive (4). Our data also show that most of the patients who are hospitalized and who present to the ED do not have positive cultures.

The presence of systemic inflammatory response syndrome (SIRS) criteria is often used to identify patients with possible sepsis and establish when to draw a blood culture. One of the most common SIRS presentations is a new fever, and most physicians have a low threshold for ordering blood cultures for patients with a newly developed fever (6). Our data indicate that most of the patients who had blood cultures drawn had a history of fever or a newly developed fever (data not shown). However, most of the patients with a history of fever did not have positive blood cultures and were discharged. Since there does not appear to be an increased incidence of positive blood cultures and worse prognosis among patients with fever, perhaps blood cultures are not needed for patients only presenting with fever when the patient is well enough to be discharged from the ED.

Among the medical community, it is generally accepted that the presence of blood pathogens can indicate a worse prognosis. However, Previsdomini et al did not show an increased length of ICU stay and mortality among patients with positive blood cultures (6). Similar to Previsdomini et al, our data also did not show a statistically significant difference in ICU admission and mortality among those with positive and negative cultures; however, our data showed that a positive blood culture trended towards longer hospital stays (6).

Skin and environmental contaminants in blood cultures can lead to unnecessary antibiotic therapy and longer hospital stays (6). It is likely that the three patients with positive cultures in the ED cohort who were directly discharged from the ED had contaminants in the culture. In fact, as many as half of the cultures that are positive represent probable contaminants (4). Bates et al found that contaminant results independently increased subsequent laboratory charges by 20% (5). Therefore, it is important to identify common pathogens that likely represent contaminants in order to prevent increased medical costs.

Limitations of this study include a sample from a single medical facility with a modest patient number due to a short study period. There may have been patients with potential infection who did not have blood cultures performed. As Shapiro et al identified, physicians have a lower threshold for ordering blood cultures for patients who have comorbidities. Therefore, our study may have bias towards finding these comorbidities as predictors of death (8). Approximately two-thirds of patients hospitalized for sepsis in the United States are 65 and older (1). Hence, a single site study could be biased towards this age group. Comparing data with multiple sites, including a longer duration for data collection and a younger patient population, could help control for confounders (1). In addition, patients may have died due to causes unrelated to infection.

Published guidelines do not clearly indicate when blood cultures should be drawn; cultures are often obtained on patients presenting with fever, chills, focal infections, leukocytosis, or suspected endocarditis (4). At BUMC, patients with a suspected underlying infectious etiology, fever, and those who are immunosuppressed routinely have blood cultures drawn in the ED. After reviewing the data for December 2013, it appears there is a statistically significant increase in the length of stay among patients with positive blood cultures drawn in the ED. However, there does not appear to be an increase in mortality or an increase in hospital admissions and ICU stay among patients with positive cultures in the ED. Although protocols recommend blood cultures to rule out an underlying infectious etiology, our data show that blood cultures may have limited utility for patients not admitted from the ED or patients solely presenting with fever who appear clinically appropriate for discharge from the ED.