Repeat Blood Culture Positive for B. pseudomallei Indicates an Increased Risk of Death from Melioidosis

Melioidosis, a bacterial infection caused by Burkholderia pseudomallei, is notoriously difficult to cure despite appropriate antimicrobial therapy and has a mortality rate of up to 40%. We demonstrate that a blood culture positive for B. pseudomallei taken at the end of the first and/or second week after hospitalization for melioidosis is a strong prognostic factor for death (adjusted odds ratio = 4.2, 95% confidence interval = 2.1–8.7, P < 0.001 and adjusted odds ratio = 2.6, 95% confidence interval = 1.1–6.0, P = 0.03, respectively). However, repeat cultures of respiratory secretions, urine, throat swabs, or pus/surface swabs provide no prognostic information. This finding highlights the need for follow-up blood cultures in patients with melioidosis.

Melioidosis is a community-acquired infectious disease caused by the Gram-negative bacillus Burkholderia pseudomallei . 1 This organism is present in the environment in areas where melioidosis is endemic, and infection is acquired through skin inoculation, inhalation or ingestion. Melioidosis is the most common cause of community-acquired bacteremic pneumonia in northern Australia 2 and the third most frequent cause of death from infection diseases in northeastern Thailand (after human immunodeficiency virus/acquired immunodeficiency syndrome and tuberculosis). 1 Common manifestations include pneumonia, hepatosplenic abscesses, septic arthritis, parotid and prostatic abscesses, and urinary tract infection. Persons with suspected melioidosis require hospital admission cultures of blood, respiratory secretions, urine, throat swab, pus if abscesses are identified and accessible, and swabs from any skin lesions. Burkholderia pseudomallei is not part of the normal human flora, and isolation of this organism from any specimen is diagnostic for melioidosis.
Death occurs in 40% of patients with melioidosis in our setting. 1 Several factors have been identified that are associated with a higher risk of death, including older age, pneumonia, markers of organ dysfunction, positive blood culture, and high quantitative count of B. pseudomallei in blood or urine. 3 , 4 Follow-up cultures are easy to perform and relatively inexpensive, but there is limited information on the significance of positive B. pseudomallei cultures over time and whether these provide additional prognostic information. We evaluated the hypothesis that follow-up qualitative and/or quantitative culture can be used to predict survival outcome in patients with melioidosis.
The patients described here were consecutive individuals presenting to Sappasithiprasong Hospital between January 1, 1997 and December 31, 2006 with culture-confirmed melioidosis who were identified as part of on-going surveillance associated with clinical trials. The hospital diagnostic laboratory was contacted daily to identify patients with one or more cultures positive for B. pseudomallei . Rounds of the medical and intensive care wards were conducted daily to identify additional cases. Any patient not already known to us who we suspected of having melioidosis on the basis of admission clinical features had samples obtained for culture, including blood, respiratory secretions (sputum, or tracheal aspirate if intubated), urine, throat swab, pus, and surface swabs from skin lesions.
Microbiologic specimens were processed as described previously. 5 All patients with microbiologically confirmed melioidosis were subsequently visited daily to obtain information on history, examination findings, and outcome (defined as survival to discharge or in-hospital death). Repeat blood culture was performed at the end of the first and second weeks after diagnosis in patients who were failing to improve despite parenteral antimicrobial therapy. In addition, other specimen types that were positive for B. pseudomallei at admission were obtained every week until negative results were obtained, and patients in whom new signs and symptoms developed and were suspected of new organ involvement had a repeat screening of relevant clinical specimens.
For purposes of analysis, clinical specimens obtained within the first three days were grouped as admission samples, those obtained during days 4-10 as first follow-up samples, and those obtained after day 10 as second follow-up samples. A few relatives took moribund patients home to die and these patients were assumed to have died.
Ethical approval for all clinical trials was obtained from the Ethical and Scientific Review Subcommittee of Thai Ministry of Public Health. [6][7][8][9] Quantitative culture of blood, urine, sputum, and pus was performed in adult patients (> 18 years of age) during 2004-2006 as described. 10 Simple and multivariable logistic regression analysis was used to examine the association between culture result and outcome. Sex, age, history of diabetes, pneumonia, hypotension, and duration of in-hospital intravenous antimicrobial treatment received were included in the multivariable analysis. Drugs used were amoxicillin/clavulanate, ceftazidime, and carbapenems. Fisher's exact test and the Mann-Whitney test were used to compare categorical variables and continuous variable between groups, respectively. There were 2,243 patients admitted to Sappasithiprasong Hospital during 1997-2006 with their first episode of culture-confirmed melioidosis. Of these patients, 1,314 (59%) were male and 929 (41%) were female. Median age was 49 years (interquartile range [IQR] = 37-60 years), and 212 patients (9%) were less than 15 years of age. Death occurred in 956 patients (43%).
On admission, blood, respiratory secretions, urine, throat swab, and pus/surface swab were obtained for culture from 1,997, 841, 1,152, 874, and 862 patients, respectively ( Table 1 ). Of these patients, blood, respiratory secretions, urine, throat swab, and pus/surface swab culture were positive for B. pseudomallei in 1,160 (58%), 550 (65%), 255 (22%), 324 (37%) and 761 (85%) patients, respectively. Having a blood, respiratory secretion, urine, or throat swab culture obtained on admission that was positive for B. pseudomallei was strongly associated with death ( P ≤ 0.01, for all specimens) ( Table 1 ). Having a pus/surface swab culture obtained on admission that was positive for B. pseudomallei was associated with a survival outcome, and most of these patients (58%, 443 of 761 patients) had only one localized lesion. Associations between death and a B. pseudomallei -positive respiratory secretion, urine, or throat swab culture on admission was independent of bacteremia ( Table 2 ). Table 1 Prognostic value of a positive culture of blood, respiratory secretion, urine, throat swab, and pus/surface swab culture on admission, first follow-up sampling period (4-10 days after admission), and second follow-up sampling period (after day 10) in patients with melioidosis, Thailand *  Multivariable logistic regression demonstrated that follow-up blood cultures positive for B. pseudomallei were strongly associated with death (adjusted odds ratio = 4.2, 95% confidence interval = 2.1-8.7, P < 0.001 and adjusted odds ratio = 2.6, 95% confidence interval = 1.1-6.0, P = 0.03, respectively). However, this association was not observed in follow-up culture of other specimen types ( Table 1 ).
The prognostic value of follow-up blood culture was evaluated further in a subset of patients who had quantitative blood culture. Thirteen patients had quantitative blood culture on admission and one or more follow-up cultures ( Figure 1 ). All 13 patients received ceftazidime or a carbapenem drug within the first two days after admission, and a decrease in bacterial count in blood was observed in all patients. Most patients with a positive quantitative follow-up blood culture (80%, 4 of 5 patients) had an admission bacterial count in blood greater than 1 colony-forming unit (CFU)/mL, and most patients with a negative quantitative follow-up blood culture (88%, 7 of 8 patients) had an admission bacterial count in blood of less than 1 CFU/mL ( P = 0.03). Of these 13 patients, two patients who had a bacterial count in blood ≤ 100 CFU/mL at admission (patients A and B), and one patient who had persistent bacteremia (0.2 CFU/mL on day 2, 0.1 CFU/mL on day 17, and 0.1 CFU/mL on day 26; patient D) died ( Figure 1 ). Patient D was a 66-year old man with diabetes mellitus and non-Hodgkin lymphoma, and was post-chemotherapy. The patient who had a follow-up blood culture of 4.7 CFU/mL on day 9 (patient C) improved and was transferred to continue parenteral therapy at a community hospital.
Our study shows that follow-up blood culture is an important predictor of late in-patient death from melioidosis. This finding agrees with observations for other infectious diseases. 11 , 12 Response to antimicrobial therapy for melioidosis is often slow (mean duration of fever of 9 days), 13 and markers for patients who survive the first week of in-patient care but remain at risk of death are not well established. Our study indicates that patients with melioidosis should receive weekly follow-up blood culture until clearance of the organism from blood Our results from a small subset of patients who had followup quantitative blood culture indicated that positive followup blood culture was associated with several factors, including high bacterial load on admission and an immunocompromise condition associated with a presumed failure to eradicate the infecting bacterium. We found that most of the patients who had follow-up positive blood culture had a bacterial count in blood of more than 1 CFU/mL on admission. However, one patient who had a bacterial count in blood of 0.2 CFU/mL failed to clear the organism from the blood (patient D). This finding could have been caused by his markedly impaired immune function and defective phagocytic cell function associated with chemotherapy. This finding emphasizes that in the clinical setting, intensive and detailed management is required of patients with positive follow-up blood culture, even without data on quantitative count.
Although a positive culture of respiratory secretions, urine, or sputum on admission was associated with death, this association was not observed for follow-up cultures of these specimen types. Positive cultures of respiratory secretions and urine on admission represent organ involvement, including pneumonia and urinary tract infection, which have been shown to be independent prognostic factors for death. 3 This study extends previous findings by describing the value of a throat swab on admission as a prognostic factor, an observation that was independent of blood and urine culture results (as demonstrated by multivariable logistic regression). Furthermore, a throat swab is better than a sputum culture because it can be performed in every patient and not only patients with respiratory symptoms and sputum production.  Lack of an association between death and persistently positive respiratory secretions, urine, throat swab, and pus/surface swab is new evidence supporting previous observations. 14 This finding adds to our experience that a small number of patients with persistent culture of these specimens can nonetheless be cured with oral eradicative treatment. 14 Localized lesions are difficult to eradicate and slow to resolve, but routine follow-up culture of the related sample types does not appear to have any clinical utility.
A limitation of this study is that not all clinical specimens were obtained from every patient. For example, there were only 779 patients with respiratory symptoms who had respiratory secretions and blood obtained for culture at admission ( Table 2 ). In addition, follow-up cultures were not performed in every patient, and the timing of repeat cultures was variable. Nonetheless, these limitations do not affect the main conclusion of this study that blood culture should be followed up at least weekly in melioidosis patients and until a negative result is observed.