Effect of Probiotics on the Incidence of Healthcare-Associated Infections in Mechanically Ventilated Neurocritical Care Patients

Abstract

Background:

Healthcare-associated infections (HAIs) are seen in 17% of critically ill patients. Probiotics, live nonpathogenic microorganisms, may aid in reducing the incidence of infection in critically ill patients. We hypothesized that administration of probiotics would be safe and reduce the incidence of HAIs among mechanically ventilated neurocritical care patients.

Methods:

We assembled 2 retrospective cohorts of mechanically ventilated neurocritical care patients. In the preintervention cohort from July 1, 2011, to December 31, 2011, probiotics were not used. In the postintervention group from July 1, 2012, to December 31, 2012, 1 g of a combination of Lactobacillus acidophilus and Lactobacillus helveticus was administered twice daily to all patients who were mechanically ventilated for more than 24 hours.

Results:

There were a total of 167 patients included, 80 patients in the preintervention group and 87 patients in the postintervention group. No patients in the preintervention group received probiotics. Eighty-five (98%) patients in the postintervention group received probiotics for a median of 10 days (interquartile range, 4–20 days). There were 14 (18%) HAIs in the preintervention group and 8 (9%) HAIs in the postintervention group (P = .17). Ventilator days, lengths of stay, in-hospital mortality, and discharge disposition were similar between the pre- and postintervention groups. There were no cases of Lactobacillus bacteremia or other adverse events associated with probiotics use.

Conclusion:

Probiotics are safe to administer in neurocritical care patients; however, this study failed to demonstrate a significant decrease in HAIs or secondary outcomes associated with probiotics. (Nutr Clin Pract. 2016;31:116–120)

Healthcare-associated infections (HAIs), including central line–associated bloodstream infection (CLABSI), catheter-associated urinary tract infection (CAUTI), ventilator-associated pneumonia (VAP), and surgical site infections, are seen in 17% of critically ill patients.¹ Intensive care unit (ICU) patients are at increased risk of HAIs because of decreased immune function and increased permeability of the digestive tract.² Specifically, brain-injured patients have T-helper cell inhibition that can put them at increased risk of infection.³

Both T-helper and B cells in the digestive system are important for preventing infection. Dendritic cells in the intestinal lamina propria assess molecular patterns of bacteria via Toll-like receptors and aid in the maturation of T-helper cells. B cells in intestinal lymph nodes are activated by antigen-presenting cells exposed to bacteria that have crossed the intestinal epithelium.⁴ Probiotics, live nonpathogenic microorganisms, may aid in reducing the incidence of infection in critically ill patients by improving the immune response of the intestinal mucosa, aiding in the maturation of T-helper cells, and activating B cells to secrete polymeric IgA and limit the inflammatory response.^4–6 A small study of patients with traumatic brain injury demonstrated decreased length of ICU stay in patients treated with probiotics. In addition, these patients had improved T-helper cell balance and a trend toward fewer HAIs.³ Probiotics can also benefit patients by reestablishing normal digestive flora and hindering pathogenic bacteria from colonizing.² Ease of administration, low cost, and low side effect profile increase the attractiveness of probiotic administration further.⁷

Based on encouraging preliminary studies, we aimed to use probiotics among patients in our neurocritical care unit (NCCU). We hypothesized a priori that administration of probiotics would be safe and reduce the incidence of HAIs among mechanically ventilated patients.

Methods

Study Design and Population

To measure safety and efficacy of probiotics, we assembled 2 retrospective cohorts of mechanically ventilated patients admitted to a 12-bed NCCU in a tertiary care academic medical center. In the preintervention cohort from July 1, 2011, to December 31, 2011, probiotics were not used. In the postintervention group from July 1, 2012, to December 31, 2012, one packet (100,000,000 colony-forming units [CFU]/packet) or 4 tablets (1,000,000 CFU/tablet) of a combination of Lactobacillus acidophilus (gasseri) and Lactobacillus helveticus (bulgaricus) (Becton Dickinson, Sparks, MD) were administered twice daily (8 am and 8 pm) by nursing to all patients who were mechanically ventilated for more than 24 hours (see Supplementary Material). Patients who were immunocompromised (transplant patients on immunosuppression, history of human immune-deficiency virus, current chemotherapy, or based on the neurocritical care attending’s or pharmacist’s discretion) or had a lactose allergy were not placed on the protocol. The study was designed with a phase-in period in June 2012 to increase probiotic protocol compliance. Vulnerable populations, including children younger than 18 years old, pregnant women, and prisoners, were excluded from data collection. The study protocol was approved by the institutional review board.

Measures

For purposes of this study, we defined HAIs as CLABSI, CAUTI, VAP, catheter-associated ventriculitis (CAV), and Clostridium difficile infection (CDI). Incidence of HAIs during admission to the NCCU, as documented by our Department of Epidemiology and Infection Control using the Centers for Disease Control and Prevention/National Healthcare Safety Network definitions, was recorded.^8–11 Admission diagnosis, antibiotic days, ventilator days, lengths of ICU and hospital stays, in-hospital mortality, and discharge disposition were abstracted from the medical record. All available blood culture data were reviewed for potential Lactobacillus bacteremia. The primary outcome was the incidence of HAIs in mechanically ventilated NCCU patients. Secondary outcomes measured included number of antibiotic days, number of ventilator days, lengths of ICU and hospital stay, in-hospital mortality, and discharge status.

Statistical Analysis

Categorical variables were analyzed using Fisher’s exact test. Nonnormally distributed continuous variables were analyzed using the Wilcoxon rank sum test. Significance was defined as P < .05 (2-tailed). The analysis performed was based on intention to treat. Statistical analyses were conducted using Stata SE versions 10 and 13 (StataCorp LP, College Station, TX).

As a sensitivity analysis, segmented regression (ie, interrupted time-series analysis) was used to test for changes in hospital length of stay, ICU length of stay, and ventilator days after initiation of the probiotics use policy. We analyzed outcomes by dividing time into periods of 23 days with 8 periods before the intervention and 8 periods after the intervention (180 days total in each period). Outcome variables were transformed using the natural logarithm so that they would be normally distributed for the interrupted time-series analysis. Regression models included terms for baseline level and trend as well as terms to estimate the changes in level and trend beginning with the first postintervention period (July 2012). For all models, the Durbin-Watson statistics for first-order autocorrelation were nonsignificant; thus, a term for autocorrelation was not included in the segmented regression models. Cohort characteristics did not appear to change over time; therefore, outcomes were not standardized to the sample population distribution.

Results

There were a total of 167 patients included, 80 patients preintervention in 2011 when probiotics were not used and 87 patients postintervention in 2012 when probiotics were implemented in the NCCU for mechanically ventilated patients (Table 1). Median (interquartile range [IQR]) age (59 years [46–74] vs 62 years [54–74], P = .49) and percent male (45% vs 48%, P = .76) were similar between the pre- and postintervention groups, respectively. Admission diagnoses were also similar between groups, with the majority of patients being admitted for traumatic brain injury, intracerebral hemorrhage, and subarachnoid hemorrhage (P = .17). No patients in the preintervention group received probiotics. Ninety-eight percent (n = 85) of patients in 2012 received probiotics for a median of 10 days (IQR, 4–20). The 2 patients who did not receive probiotics had their goals of care advanced to comfort measures prior to initiation of probiotics. Use of enteral nutrition (EN) and steroids was similar between groups. Antibiotic days between the pre- and postintervention groups were similar (median [IQR], 4 [0.7–14] days vs 6.6 [1.0–11.0] days, P = .27, respectively); however, more patients in the probiotic group received antibiotics (91% vs 79%, P = .03).

Table 1.

Characteristics of Patients: Preintervention (No Probiotics) vs Postintervention (Probiotics).^a

Baseline Characteristics	Preintervention (2011) (n = 80)	Postintervention (2012) (n = 87)	P Value
Age, median (IQR), y	59 (46–74)	62 (54–74)	.49
Male sex	36 (45)	42 (48)	.76
Admission diagnosis			.17
Traumatic brain injury	22 (28)	25 (29)
Intracerebral hemorrhage	15 (19)	22 (25)
Subarachnoid hemorrhage	12 (15)	13 (15)
Ischemic stroke	10 (13)	8 (9)
Status epilepticus	5 (6)	6 (7)
Brian tumor	3 (4)	7 (8)
Guillain-Barre syndrome	3 (4)	1 (1)
Myasthenia gravis	3 (4)	0 (0)
Cervical spine disease	3 (4)	0 (0)
Encephalitis	1 (1)	1 (1)
Encephalopathy	1 (1)	0 (0)
Amyotrophic lateral sclerosis	1 (1)	0 (0)
Multiple sclerosis	1 (1)	0 (0)
Spinal cord injury	0 (0)	4 (5)
Lactobacillus	0 (0)	85 (98)	<.0001
Lactobacillus days, median (IQR)	0 (0–0)	10 (4–20)	<.0001
Steroid use	21 (26)	23 (26)	.99
Antibiotic use	63 (79)	79 (91)	.03
Antibiotic days, median (IQR)	4 (0.7–14)	6.6 (1.0–11.0)	.27
Enteral nutrition use	65 (81)	74 (85)	.54

IQR, interquartile range.

^aValues are presented as number (%) unless otherwise indicated.

There were 14 (18%) HAIs in the preintervention group and 8 (9%) HAIs in the postintervention group (P = .17) (Table 2). There were no cases of CLABSI or CDI in the probiotic group. No patients had more than 1 HAI. Ventilator days, ICU days, and hospital days were similar between groups (Table 2). In addition, in-hospital mortality (38% and 33%) and discharge disposition were similar between the pre- and postintervention groups. There were no cases of Lactobacillus bacteremia or other adverse events with administration of probiotics.

Table 2.

Comparison of Outcomes in Patients: Preintervention (No Probiotics) and Postintervention (Probiotics).^a

Outcomes	Preintervention (2011) (n = 80)	Postintervention (2012) (n = 87)	P Value
Hospital-associated infections	14 (18)	8 (9)	.17
Catheter-associated urinary tract infections	4 (5)	5 (5.75)	.99
Central line-associated bloodstream infection	3 (4)	0 (0)	.11
Catheter-associated ventriculitis	3 (4)	1 (1)	.35
Clostridium difficile-associated diarrhea	2 (3)	0 (0)	.23
Ventilator-associated pneumonia	2 (3)	2 (2)	.99
Ventilator days, median (IQR)	7 (3–12)	6 (4–13)	.78
Intensive care days, median (IQR)	9 (5–21)	12 (7–20)	.23
Hospital days, median (IQR)	14 (7.5–32.5)	17 (10–31)	.23
Discharge disposition			.47
Home	12 (15)	6 (7)
Acute rehabilitation unit	23 (29)	32 (37)
Skilled nursing facility	10 (13)	13 (15)
Hospice	3 (4)	4 (5)
Transfer	1 (1)	3 (4)
Against medical advice	1 (1)	0 (0)
In-hospital mortality	30 (38)	29 (33)

IQR, interquartile range.

^aValues are presented as number (%) unless otherwise indicated.

Table 3 displays the segmented regression results of the time series pre- and postintervention for the dependent variables of mean hospital length of stay, mean ICU length of stay, and mean ventilator days. Probiotic use did not have a significant impact on any of these secondary outcomes. These results support the findings of our primary analysis.

Table 3.

Segmented Linear Regression Model Results for Change in Outcome After Probiotic Policy Change.

	Model Variable (95% Confidence Interval)
Outcome	Intercepta	Baseline Linear Trendb	Level Changec	Postintervention Linear Trendd
Log-transformed hospital length of stay, mean dayse	42.57 (10.92 to 74.21) 2.90 (2.30 to 3.50)	−2.90 (−8.49 to 2.68) −0.04 (−0.16 to 0.06)	3.78 (−21.20 to 28.77) 0.22 (−0.44 to 0.88)	0.85 (−0.51 to 2.20) 0.03 (−0.01 to 0.07)
Log-transformed intensive care length of stay, mean dayse	20.21 (3.28 to 37.14) 2.50 (1.96 to 3.04)	−0.66 (−2.86 to 1.55) −0.03 (−0.11 to 0.05)	0.96 (−7.97 to 9.88) 0.16 (−0.39 to 0.72)	0.54 (−0.68 to 1.77) 0.03 (−0.02 to 0.09)
Log-transformed ventilator use, mean dayse	15.2 (3.50 to 26.89) 2.14 (1.59 to 2.70)	−0.79 (−2.41 to 0.83) −0.04 (−0.13 to 0.04)	2.27 (−4.63 to 9.17) 0.23 (−0.36 to 0.83)	0.04 (−0.83 to 0.91) 0.00 (−0.06 to 0.07)

^aThe intercept is the model estimate for the rate in the first 23-day study period.

^bBaseline linear trend describes the trend in the outcome per 23-day study period in the preprobiotics change period (July 2011 to December 2011).

^cLevel change is the estimated instantaneous change in the outcome in the first postintervention period (July 2012) compared with the expected rate based on the baseline trend.

^dPostintervention linear trend describes the change in trend in the postintervention change period (July 2012 to December 2012). The trend in the postintervention change period is the sum of the baseline trend plus the change in trend.

^eOutcomes were transformed using a natural logarithm to make the distributions approximately normal.

Discussion

Although there was a numerical decrease in the primary outcome of HAIs in mechanically ventilated neurocritical care patients postintervention, this retrospective analysis failed to detect a significant difference. Our rate of HAIs preintervention (18%) is similar to a previously published rate of 17%, while our rate postintervention is lower at 9%.¹ A meta-analysis of 23 trials, including over 1000 mixed critical care patients, previously demonstrated that ICU patients receiving probiotics are at decreased risk of infection (relative risk [RR], 0.82; 95% confidence interval [CI], 0.69–0.99; P = .03).⁶ However a smaller study of brain-injured patients likewise did not find a significant decrease in the incidence of HAIs (35% in the probiotic group vs 58% in the control group, P = .095).³

Specifically examining VAP did not reveal a benefit for the probiotic group. All patients in this study, both pre- and postintervention, were placed on a ventilator bundle as part of our ICU protocol. This bundle consisted of elevating the head of bed to 30 degrees, cleansing the mouth with chlorhexidine, administering stress ulcer prophylaxis with a proton pump inhibitor, and spontaneous awakening and breathing trials. Three meta-analyses previously indicated that critically ill patients receiving probiotics are at decreased risk of VAP.^6,12,13 The smaller sample size in our study restricted to neurocritical care patients may explain our observed lack of association.

Administration of probiotics also failed to demonstrate a decrease in number of antibiotic days, number of ventilator days, lengths of ICU and hospital stays, in-hospital mortality, or improvement in discharge status. Meta-analyses of general critical care patients also failed to exhibit a benefit from administration of probiotics on lengths of stay or mortality.^6,12,13

Our study did show that administration of probiotics in neurocritically ill patients was safe. There have been case reports with secondary infection due to probiotics,^14–16 but we had no reports of Lactobacillus bacteremia in our patients and took care to exclude immunocompromised patients to minimize risk of adverse events. In addition to being safe, compliance with administration of probiotics was high, with 98% of patients receiving this intervention. Adherence to the protocol was ensured on daily rounds by the neurocritical care team. Because of a transition of goals of care, 2 patients were appropriately not placed on the protocol.

There were several limitations to our study. As with all observational studies, unmeasured variables may have influenced outcomes. For example, the type of EN each patient received was not recorded and may be a limitation secondary to unknown interactions between the probiotic we chose and specific nutrition formulas. The interrupted time-series sensitivity analysis relied on the assumption that no other policies or interventions were introduced at the same time as this policy. Also, it relied on the assumption that the time trends in the outcomes before the policy would have continued throughout the study period in the absence of the probiotics policy. Another limitation of our study was its retrospective nature and inability to account for biases from unmeasured changes in care over the time period. All patients during the preselected time frames were included, and all patients had complete data, minimizing selection bias. Finally, our study was underpowered. We would have required 265 patients in each treatment group (530 total) to detect an 8% difference with 80% power.

Other benefits of probiotics, such as decrease in antibiotic-associated diarrhea, might exist but were not analyzed in this study.¹⁷ Interestingly, more patients received antibiotics postintervention, but there was no CDI. Decreased incidence of CDI in critically ill patients has previously been demonstrated.⁷ It is not clear why patients postintervention received more antibiotics. This did not seem to be a direct effect of probiotic administration or due to increased incidence of HAIs.

Ideal preparation, dosing, and duration of probiotics are unknown and were a limiting factor to this study. ^6,12,13 The probiotic that we chose is the only one available on formulary at our hospital. Using a different probiotic could have yielded different results. Using a higher dose or continuing probiotics for the entire length of the patient’s hospital stay could have also shown benefit. A final limitation of our study was that it was based on a single institution, which may limit generalizability of our findings to other institutions. While we were able to demonstrate that probiotics can be safely administered in neurocritically ill patients, more studies need to be conducted with larger sample sizes to evaluate the appropriate agent, dosing, and benefit of probiotics in this population.