Introduction

Stenotrophomonas maltophilia (S maltophilia) is a non-fermenting, gram-negative bacillus which in recent years, has emerged as an opportunistic pathogen in human host. It increasingly causes nosocomial human infections particularly among patients who are immunosuppressed and have an underlying debilitating medical condition [7, 10, 12, 13, 15–17]. It causes a variety of clinical syndromes like bacteremia, respiratory tract infection, urinary tract infection, ocular infections, skin, soft tissue and wound infections, bone and joint infections, meningitis, and endocarditis [6, 25, 29]. Although it was initially thought to be of lower virulence, several studies [7, 10, 14, 17] indicate that the mortality rates from S. maltophilia infection have been rising especially in patients with malignancy, organ transplant, and among those admitted in intensive care units (ICU).

Currently, among children, there are only a few reports describing S. maltophilia infection [6, 14, 16, 21, 32]. The majority of these studies have focused on S. maltophilia in cystic fibrosis patients [1, 8, 9, 19], in patients with hematological malignancies/bone marrow transplant [2, 23, 25, 28], and in the neonatal ICU population [6, 16, 21, 32]. However, there are no studies which describe the clinical characteristics of S. maltophilia infections among children with heart disease cared for in the cardiac ICU.

Therefore, our objective was to investigate the incidence, clinical characteristics, and outcome of S. maltophilia infection among children 0–6 months of age with underlying cardiac disease. We hypothesized that unique risk factors are associated with poor outcome with S. maltophilia infection in this patient population.

Materials and Methods

Study Design and Setting

This single center, retrospective, observational study was conducted in the cardiovascular ICU (CVICU) at Arkansas Children’s Hospital, a 330 bedded urban tertiary care children’s hospital. The study was approved by the Institutional Review Board (IRB) and waiver was obtained for informed consent from the IRB.

Study Population

The study included all children between 0 and 6 months of age hospitalized in CVICU for at least 48 hours and had positive cultures for S. maltophilia at any biologic site between the period of January 1st, 2006 and December 31st, 2010. All study patients were identified by querying the institutional medical records and microbiology records. Subjects were excluded if they were admitted to non-ICU units or transferred from an outside facility with S. maltophilia infection.

Study Definitions

Infection was defined as recovery of S. maltophilia from blood, cerebral spinal fluid (CSF), urine, eye, wound, or bronchio-alveolar lavage (BAL). All hospital acquired- health-care associated infections were defined using the standard Centers for Disease Control/National Healthcare Safety definition criteria [15]. For the purposes of the study, (a) ventilator-associated pneumonia (VAP) was defined as per the CDC guidelines for VAP criteria, i.e., patients with new or progressive and persistent infiltrate on ≥2 Chest x-rays along with desaturations or increased oxygen requirement or increased ventilator settings and with temperature instability (fever or hypothermia) OR leukocytosis and left shift or leukopenia (<4,000/>15,000) OR bradycardia or tachycardia OR apnea, tachypnea or retractions were defined as having infection and patients who did not qualify this criteria were defined as colonized. We defined bloodstream infection as isolation of Stenotrophomonas from blood and urinary tract infection as >50,000 colony count of Stenotrophomonas from catheterized urine sample and wound infection as isolation of Stenotrophomonas without any mixed growth from wound along with clinical features of wound infection, (b) Prolonged ICU and hospital length of stay (LOS) was defined as the LOS above the 75th percentile cutoff for LOS for the entire study cohort, (c) A day for the study was defined with a cutoff of 12 midnight while defining outcomes, (d) Ventilator-free days were defined as twenty-eight days minus number of days the subjects were not on mechanical ventilator via tracheostomy or oral endotracheal tube, (e) Days of infection was defined as number of days from first positive culture for S. maltophilia to first day of negative culture, (f) Broad spectrum antibiotics duration was defined as days of usage of extended spectrum beta-lactams, carbapenems, or intravenous quinolones about 4 weeks prior to first positive S. maltophilia culture, (g) Negative culture during hospitalization was defined as negative culture present before death or discharge, (h) Treatment duration days were defined as days treated with appropriate antibiotics till the presence of first negative culture in subjects with one infection and till last negative culture during hospitalization in subjects with multiple infections, and (i) Lymphopenia was defined as absolute lymphocyte count (ALC) less than 1500 cells/microliter of blood.

Data Variables

Data abstracted include: demographic variables, ICU outcomes (mortality, ICU length of stay (LOS), hospital LOS), underlying medical/surgical condition(s), and laboratory parameters. Additional S. maltophilia infection- specific data collected were site, days to first positive culture, broad spectrum antibiotic use (type and duration) 30-days prior to infection, presence, type and number of invasive indwelling catheters (central line, surgical drains, urinary catheter), resistance pattern of infection, and other bacterial/viral infections.

Statistical Analysis

All the data were analyzed using R v2.15.0 (R Development Core Team, Vienna, Austria). Descriptive statistics were expressed as median (interquartile range) for continuous variables and frequency and percentage for categorical variables. Mann–Whitney U tests were carried out to compare the distributions of continuous variables between the survivors and non-survivors to hospital discharge, while the proportions of categorical data were compared between the two groups using χ 2 tests. A histogram was plotted to represent the distribution of the ALCs (Fig. 1). Kaplan–Meier curves were plotted for the time from day of hospital admission to S. maltophilia infection and time from start of mechanical ventilation to S. maltophilia infection among survivors and non-survivors, respectively (Figs. 2 and 3). Log-rank tests were performed to evaluate the difference in the survival curves between survivors and non-survivors. p-values less than 0.05 were considered to indicate statistical significance.

Fig. 1
figure 1

Distributional plot of ALC prior to infection

Full size image
Fig. 2
figure 2

Kaplan Meier plot for time from hospital admission to Stenotrophomonas infection stratified by death

Full size image
Fig. 3
figure 3

Kaplan Meier plot for time from mechanical ventilation to Stenotrophomonas infection stratified by death

Full size image

Results

Patient Characteristics

We identified a total of 38 patients between the ages of 0–6 months hospitalized in CVICU for ≥48 hours with positive culture for S. maltophilia. Six of these patients were excluded from the study as they did not fulfill the CDC criteria for VAP and were considered to be colonized with S. maltophilia. The overall incidence for the 5-year study period was 0.8 % (n = 32/3656) with an incidence of 0.7 % in 2006 (n = 5/713), 1.05 % in 2007 (n = 7/662), 1.4 % in 2008 (n = 10/729), 1.06 % in 2009 (n = 8/755), and 0.25 % in 2010 (n = 2/794). Among 32 eligible patients, there were 47 S. maltophilia infections, including 44 episodes of VAP, one blood stream infection, one sternal wound infection, and one urinary tract infection.

Demographics of study patients are shown in Table 1. The median age of the overall study population was 60 days. Among those who acquired S. maltophilia infection after admission in CVICU, 21.8 % were admitted in neonatal ICU and one (3.1 %) was admitted in pediatric ICU prior to transfer to the CVICU. The overall in-hospital all-cause mortality for the study group was 37.5 % (n = 12/32).

Table 1 Comparison of demographics between survivors and non-survivors
Full size table

Comparison of Survivors and Non-survivors with S. maltophilia

Survivors had 56 ventilator-free days when compared to 15.5 days among non-survivors (p value = 0.006). Number of days with positive cultures for S. maltophilia was 30 days in non-survivors compared to 11 days in survivors (p value = 0.008).

The ALC immediately prior to S. maltophilia infection was significantly higher in survivors versus non-survivors (1,665 vs 960/mm3; p value = 0.01) (Fig. 1). All patients (n = 32) had leukocytosis prior to infection with S. maltophilia but there was no statistical differences between survivors and non-survivors. Among the comorbid conditions, the need for renal dialysis (67 %; n = 8/12 among non-survivors vs 30 %; n = 6/20; p value = 0.04) and presence of stroke (17 %; n = 7/12 among non-survivors compared to 0 % in survivors; p value = 0.05) were found to be significantly different among survivors and non-survivors.

A majority of positive cultures for S. maltophilia were poly-microbial in 75 % (n = 9/12) of non-survivors compared to 55 % (n = 11/20) among the survivors. The most common bacteria which were involved in poly-microbial infection were pseudomonas (n = 10/32), enterobacter (n = 8/32), and klebsiella (n = 7/32). 95 % of survivors (n = 19/20) had negative culture during hospitalization in contrast to 50 % (n = 6/12) among non-survivors (p value = 0.003).

The median time between ICU admission and ICU acquired S. maltophilia and the median time from mechanical ventilation to ICU acquired infection in survivors and non-survivors are depicted as Kaplan–Meier plots in Figs. 2 and 3.

Antibiotic Therapy

The median duration of antibiotic therapy among survivors was 10 days versus 24 days among non-survivors (p value = 0.011). 97 % (n = 31/32) of S. maltophilia strains were sensitive to TMP-SMX therapy and 91 % (n = 29/32) to levofloxacin and ticarcillin/clavulanate. Resistance pattern in our patient cohort was low and it did not change over time even in patients who had multiple episodes of infection with S. maltophilia.

Broad spectrum antibiotic use prior to infection with S. maltophilia was noted in 75 % of non-survivors (n = 9/12) versus 60 % (n = 12/20) of survivors. The median duration of broad spectrum antibiotics before S. maltophilia infection was noted to be 7.5 days in non-survivors and 6 days in survivors. 50 % of all patients (n = 16/32) and 66.6 % of non-survivors (n = 8/12) had prior treatment with carbapenem about 2–4 weeks before S. maltophilia infection. Clinical characteristics of patients are demonstrated in Table 2 and outcome of study patients is demonstrated in Table 3.

Table 2 Infection and antimicrobial-related characteristics
Full size table
Table 3 Comparison of outcomes between survivors and non-survivors
Full size table

There were two patients who were older than 6 months of age who qualified for diagnosis with S. maltophilia VAP based on criteria. One of them was 5 years of age, had ALC of 2,370 and 3,600 at admission and prior to infection, respectively, was treated with trimethoprim/sulfamethoxazole for 16 days, was on broad spectrum antibiotics for about a month prior to infection, had multiple positive cultures, the isolate was resistant to ticarcillin/clavulanate and did not survive. The second patient was 18 months of age, had ALC of 553 and 1,440 at admission and prior to infection, respectively, was treated with ticarcillin/clavulanate for 10 days, was on broad spectrum antibiotics for 21 days prior to infection, cleared the positive culture, the isolate was susceptible to trimethoprim/sulfamethoxazole, ticarcillin/clavulanate, and levofloxacin and survived.

Discussion

This is the first study to characterize S. maltophilia nosocomial infection exclusively among infants (0–6 months) with heart disease hospitalized in a dedicated CVICU. The resistance to TMP-SMX was low and the resistance pattern did not change over 5-year study period, even in patients with multiple episodes of documented infections.

The overall incidence rate of S. maltophilia nosocomial infection was 0.8 % over the study period of 5 years and there were no significant increase in the incidence at our institution despite recent trends of increase in the overall incidence of S. maltophilia infection in ICUs. Earlier reports from the late 1990 s indicated an incidence of 0.6–0.9 %. However, in recent years, several studies have indicated a much higher incidence of up to 8 %. However, unlike these studies which included adults as well as children our study focused specifically on infants with complex heart disease.

Lymphopenia prior to S. maltophilia infection was noted to be significantly more frequent among non-survivors compared to survivors. It is possible that lymphopenia conferred a degree of immunosuppression among non-survivors and may have contributed for them to have longer duration of positive cultures and to be treated for longer duration with antibiotics. In contrast to our report, however, Mazzoni et al. have noted that neutropenia was an associated risk factor for developing S. maltophilia bacteremia among adult patients with malignancy. Even though we found lymphopenia more frequently among our non-surviving subjects, this study is limited in defining it as a risk factor for acquiring S. maltophilia infection.

The use of broad spectrum antibiotics prior to S. maltophilia infection in two-thirds of the study cohort is consistent with other studies which mention antibiotic exposure as a risk factor for infection with S. maltophilia. Furthermore, carbapenem use in only 50 % of all infected patients in our study is similar to some those reported by other studies [35]. It appears that rather than exposure to one particular antibiotic, exposure to broad spectrum antibiotics is important in S. maltophilia infection. However, in contrast, prior use of TMP-SMX which was identified to be associated with an increased risk for bacteremia [4] was only used in 15.6 % of our patient population. We found 97 % of the isolates were sensitive to TMP-SMX and 91 % of the isolates to levofloxacin and ticarcillin/clavulanate. The other antibiotics were not tested for susceptibility at our institution. The resistance rate is much less when compared to one article [4] which mentioned 10 % resistance rate to TMP/SMX and 54 % to ticarcillin/clavulanate but similar rates to TMP/SMX as another article [27]. The resistance pattern did not change significantly even in isolates from the same patient with recurrent episodes of infection. In more than half of the patients in our study (62.5 %), bacterial co-pathogens were isolated from the respiratory tract when S. maltophilia infection was diagnosed. This frequency of polymicrobial infections is slightly higher than previously reported data in patients with S. maltophilia pneumonia and S. maltophilia bacteremia [ [27], [31], [23] and [9] ].

S. maltophilia infection in our cohort was associated with significant morbidity (prolonged need for mechanical ventilation, prolonged ICU-LOS, and hospital LOS). It was also associated with poor outcomes as non-survivors had longer duration of positive cultures and antibiotics therapy and it took longer to clear the infection. The findings are consistent with several other studies [ [10], [32], [33], [5], [23], and [9] ] which showed a similar increase in morbidity in patients with S. maltophilia infection who have severely debilitating underlying illness and/or immunosuppression. These data suggest that acquisition of S. maltophilia may be a marker of severity of underlying illness than of it being a pathogen with high virulence. Most studies which suggest high attributable mortality rates from S. maltophilia infection [1, 5, 17, 18, 20, 27] had large incidence of S. maltophilia bacteremia than localized respiratory tract involvement [8, 23, 26, 30]. In our study, the majority of S. maltophilia infections were localized as pneumonia with only one episode of bacteremia noted.

The crude all-cause mortality was 37.5 % which may be related to their underlying medical/surgical condition rather than to S. maltophilia infection. This study is limited to calculate infection-attributed mortality rate as we did not include matched controls without S. maltophilia infection. Prior studies indicate infection-attributed mortality rate in the range of 14.7–37.5 % [1, 5, 11, 13, 17, 18, 20, 22, 27] and infection-attributed mortality rate for bacteremia to be as high as 51 % [8, 9, 11, 22–24, 26, 30, 34]. Risk factors for increased mortality described in the literature include bacteremia, shock, thrombocytopenia, neutropenia, multi organ dysfunction, ICU admission, underlying pulmonary disease, pulmonary source of isolate, polymicrobial cultures, delayed effective treatment, prior use of carbapenem, renal failure, and liver failure in the ICU [1, 3, 5, 9, 11, 18, 24, 27]. Our study population is unique as it only included infants with congenital heart disease but shared some of the common risk factors noted in other studies like ICU admission and pulmonary source of isolate. However, in our cohort we did not find bacteremia, shock, polymicrobial cultures, neutropenia or thrombocytopenia to be higher among non-survivors compared to survivors.

Our study has several potential limitations related to its retrospective observational study design, and a small sample size from a single study center. Since molecular typing of S. maltophilia isolates was not performed in clinical routine, we cannot provide data on potential transmission of S. maltophilia isolates in our study. The small sample size as well as the absence of a comparison group precluded us from being able to conduct a multivariate analysis for risk factors.

Conclusion

To our knowledge, this is the first study in the literature looking at incidence, clinical characteristics and outcomes related to ICU-acquired S. maltophilia infection in pediatric population. S. maltophilia infection is found to be rare in this particular patient cohort at our institution during the study period over 5 years. Lymphopenia was noted prior to infection in non-survivors and these patients took a longer time to clear the infection and needed prolonged duration of antibiotic therapy. The all-cause in-hospital mortality was found to be 37.5 % and S. maltophilia infection was associated with increased morbidity with increased length of stay in ICU, increased length of hospitalization, increased duration of mechanical ventilation, and longer duration of treatment with antimicrobials similar to another article [4]. Prospective, multi-center, case–control studies among larger pediatric ICU population will be important to further define risk factors for acquiring S. maltophilia infection among children with cardiac disease.