This prospective study was performed from March 2013 to July 2016 in our tertiary university hospital in the Netherlands.
Population
All patients ≥18 years old were eligible for inclusion if they suffered from TBI or stroke (ischemic or hemorrhagic) and required admission to the intensive care unit. Exclusion criteria comprised expected death within 1 hour of admission, burn or inhalation injury or emergency thoracotomy or cardiopulmonary resuscitation with chest compressions before ICU arrival. Patients with known pregnancy, receiving immunosuppressive medication prior to admission, with a known do-not-resuscitate order and patients already enrolled in a concurrent ongoing interventional randomized clinical trial were also excluded.
Heart rate variability recordings
Patients were hemodynamically monitored during their ICU or medium care stay as standard care. For HRV analysis, the heart rate of patients was recorded from the patient bedside monitor with specialized software from our hospital’s technical department (HartRateMonitor3, Academic Medical Center, Amsterdam, The Netherlands). This software determines the position of the R peaks in the patient’s electrocardiography at a sampling rate of 1000Hz. Ten minute epochs of the heart rate were recorded directly after admission and subsequently four times daily during the admission period at the ICU. Daily recorded 10 minute epochs were pooled together to calculate a daily mean.
Blood samples
Blood samples were collected on admission day and 24 and 72 hours thereafter (days 0, 1 and 3). The tubes were centrifuged for 10 minutes at 1750 relative centrifugal force (RCF) and 18˚C. The upper two-thirds of the plasma was centrifuged again to obtain platelet-free plasma and stored as 250 µL aliquots at -80˚C.
Definition of infection
Infection criteria were adopted from the MARS project (Molecular Diagnosis and Risk Stratification of Sepsis)[12], which has provided criteria for infection classification that were modified from the Centers for Disease Control (CDC) and International Sepsis Forum (ISF) criteria. All available clinical, microbiological, and radiological evidence on infections was collected. Infections were scored for infection source and certainty of diagnosis resulting in definite, probable and possible infections. The MARS criteria for infections can be found in the online additional file.
Antibiotic management
Selective digestive tract decontamination (SDD), consisting of parenteral and enteral antimicrobials, was administered to patients with an expected ICU stay of >3 days. This includes a 4 day regime of intravenous cefotaxime. Also, an enteral nonabsorbable suspension containing polymyxin, tobramycin and amphotericin B was administered through a nasogastric tube 4 times a day. Orabase paste was applied 4 times a day to the oropharyngeal mucosa. Furthermore antibiotics chosen based on local protocol and expert opinion were started empirically when infection was suspected.
Whole blood stimulation
Whole blood stimulations were performed as described before[13]. Briefly, heparin-anticoagulated blood from a healthy volunteer was collected. 500 µL whole blood samples were diluted in a 1:1 ratio with RPMI (Roswell Park Memorial Institute, Buffalo, USA), supplemented with glutamine 0.3 g/L and lipopolysaccharide (LPS E.coli; O111:B4 Ultrapure SIGMA, 1 ng/ml). Subsequently, 100 µL of plasma from either infected or non-infected patients was added. Whole blood stimulated with LPS without the addition of patient plasma served as a positive control. Whole blood buffered with RPMI without the addition of patient plasma or LPS served as a negative control. The samples were then incubated for 24 hours at 5% CO2 at 37°C. After incubation, the samples were centrifuged (1200 RCF at 18°C for 10 minutes) and the upper two-thirds of the plasma was collected and stored at -20°C.
ELISA
TNF-α and IL-10 levels were measured in the supernatant collected after the whole blood stimulations by enzyme-linked immunosorbent assays (ELISA) according to the manufacturer’s instructions (R&D Systems, Abingdon, United Kingdom).
Outcome
Primary outcome was nosocomial infection and the extent of immunosuppression, defined as the outcome of ex vivo whole blood stimulations. Secondary outcomes included heart rate variability, duration of ventilation, length of hospital and ICU stay and 28-day mortality.
Statistical analyses
HRV recordings were analyzed with HRV algorithms in MATLAB software (MathWorks, Natick, USA) adhering to international standards[14]. The rationale behind the algorithms is described in more detail in the online additional file. Normalized units of high-frequency power (HFnu) as a reflection of parasympathetic activity and normalized units of low-frequency power (LFnu) as a reflection of sympathetic activation were used to assess autonomic activity[14]. Also low-frequency:high-frequency (LF:HF) ratio was analyzed, as this ratio is considered to reflect either the sympathovagal balance or sympathetic nervous system activity[10].
All patients with definite or probable infections were selected for ex vivo whole blood stimulation tests. These patients were compared to non-infected patients, matched for age and APACHE II scores.
Distribution of variables was examined for normality using the Kolmogorov-Smirnov test. Continuous variables are expressed as mean and standard deviation (SD) or median and interquartile range (IQR) and were analyzed using the Student’s t-test or the Mann-Whitney U test, depending on the distribution. Categorical factors are expressed as proportions and were analyzed by chi-square testing. Testing between infected patients, non-infected patients and controls in ex vivo whole blood stimulation tests was done with the one-way ANOVA test. P-values less than 0.05 were considered statistically significant.