Steroids and Post-resuscitation Infectious (Septic) Complications

BACKGROUND AND RATIONALE Patients successfully resuscitated after cardiac arrest experience a "sepsis-like" syndrome characterized by cytokine storm, endotoxemia, coagulopathy, and various degrees of adrenal insufficiency. These pathophysiological mechanisms contribute to the development of circulatory failure, i.e. post-resuscitation shock. Post-resuscitation shock patients resuscitated from vasopressor-requiring cardiac arrest are frequently poorly responsive to high-rate vasopressor infusions (e.g. norepinephrine ≥0.5 μg/kg/min) and intravenous fluids.

The postresuscitation systemic inflammatory response syndrome (SIRS) may be partly caused and subsequently amplified by ischemia/reperfusion (I/R)-associated disruption of the intestinal mucosal barrier. Steroids may suppress key events of I/R injury propagation. Furthermore, in shock states, stress-dose steroids improve vascular responsiveness to vasopressors and preserve monocyte and neutrophil phagocytosis, and dendritic cell function. Low-dose steroids may reduce the mortality of severely ill patients with septic shock.

Nosocomial infections constitute an important cause of postresuscitation mortality. We hypothesized that exposure to stress-dose steroids during and/or after CPR may be associated with reduced risk of death due to postresuscitation infectious complications. To test this hypothesis, we combined data from two prior, prospective studies of in-hospital cardiac arrest. These studies compared the combination of vasopressin, steroids, and epinephrine (VSE) to epinephrine alone, with respect to survival to hospital discharge and good functional outcome. Patients with post-resuscitation shock of the VSE groups received stress dose hydrocortisone (300 mg/day for 7 days maximum, followed by gradual taper at a rate of 100 mg / day and discontinuation on day 10). Patients with post-resuscitation shock of the control groups received saline placebo. Follow-up rates were high in both studies and the reported incidence of post-resuscitation infectious complications was similar in the VSE and control groups.

METHODS Study Design Retrospective analysis of prospectively collected data from two randomized, clinical studies. Study participants were hospitalized in intensive or coronary care units (ICUs or CCUs) of three tertiary care centers: Evaggelismos General Hospital and 401 Greek Army Hospital (both in Athens, Greece), and University Hospital of Larissa, Larissa, Greece.

Ethics and Approval The present analysis is not associated with any clinical intervention, and therefore, the investigators have applied for a waiver of informed consent from either the patient or his/her next of kin. Moreover, the investigators have requested the permission to confirm previously recorded microbiological data through the hospitals' electronic databases The institutional review boards (IRBs) of the aforementioned hospitals have granted their approval for the current study. Evaggelismos Hospital approval No. 14/9/1/2015; 401 Greek Army Hospital approval No. 3/2015/5/2/2015; Larissa University Hospital approval No. 58905/2014/14/1/2015. Modifications of statistical terminology [from individual patient data (IPD) meta-analysis to IPD re-analysis] and of infections' classification in the definition of the primary outcome were also approved by the Evaggelismos IRB (respective Approval Nos. 30/25/2/2016 and 29/25/2/2016) and communicated to the other 2, aforementioned IRBs. These approvals were ratified by the IRBs of the other 2 participating centers (401 Greek Army Hospital, IRB Decision No.: 4-2016/6/4/2016; Larissa University Hospital, IRB Decision No.: 5/19-5-2016/Θ.18). Additional, significant amendments of the analysis protocol aimed at primarily evaluating the effect of stress-dose steroids on lethal septic shock were approved by the Evaggelismos IRB (Approval No. 9/26/1/2017) and appropriately communicated to the other 2, aforementioned IRBs.

Analysis Endpoints are presented in the dedicated subsection. Patients The reference study population consists of 368 patients (Evaggelismos Hospital, n=288/368=78.2%) with in-hospital cardiac arrest, who required epinephrine during cardiopulmonary resuscitation (CPR) according to the 2005 Guidelines for Resuscitation. During CPR, VSE group patients (n=178) also received vasopressin and methylprednisolone and controls (n=190) the respective saline placebos. At 4 hours after CPR, there were 211 surviving patients (VSE group, n=115), who were evaluated for postresuscitation shock. One hundred three VSE group patients were then assigned to stress-dose hydrocortisone and 88 controls to saline placebo. One hundred two VSE group patients and 15 controls were actually treated with stress-dose hydrocortisone (n=117). One VSE group patient did not receive stress dose hydrocortisone due to pharmacist error; however, the patient did receive 40 mg of methylprednisolone during CPR.

Definitions Essential prerequisites for the applicability of the below-provided definitions of infections to determine their (prior) occurence in the participants of the included studies will include the presence of confirmatory investigator recordings (for the prior diagnoses of these infections) and also the retrieval of confirmatory microbiological data from hospital records.

Ventilator-associated pneumonia (VAP) will be defined as new (or worsening if already present) chest x-ray infiltrate and at least 2 of the following 3 findings: worsening of respiratory secretions (to purulent or muco-purulent), leucocytosis or leucopenia (>11000/μL or <4000/μL respectively), and fever or hypothermia (>38 degrees Celcius or <36 degrees Celcius respectively). VAP diagnosis will require at least 48 hours of postresuscitation endotracheal intubation and mechanical ventilation. VAP will be confirmed by a positive tracheobronchial aspirate sample with ≥1000000 Colony Forming Units (CFUs)/μL] and / or a concurrent blood culture positive for the same pathogen.

Ventilator-associated tracheobronchitis (VAT) will be defined as fever (>38 degrees Celcius) not attributable to another infectious complication and positive tracheobronchial aspirate culture (with ≥1000000 CFUs/μL) yielding a new bacteria (not present at intubation), and no radiographic signs of new pneumonia.

Bacteremia (or Fungemia) will be defined as isolation of a pathogen from at least one blood culture set (in cases of coagulase negative Staphylococcus species, or Corynebacterium species, 2 positive blood culture sets will be required). Septic shock will be defined as bacteremia / fungemia plus circulatory failure with at least 30% increase in norepinephrine infusion requirements after day 3 postarrest.

Urinary tract infection will be defined as fever or hypothermia and presence of at least one pathogen in a urine sample culture at a concentration of ≥100000 CFUs/μL.

Post-resuscitation shock, organ / system failures, acute respiratory distress syndrome, hyperglycemia, and survival to hospital discharge with good functional outcome will be defined as in the 2 included VSE studies. Organ failure free days and ventilator free days will be determined as in the 2 included VSE studies.

Statistical Analysis Plan All analyses will be performed with the Statistical Package for Social Sciences version 22.0 (IBM, Armonk, NY, USA). As this is not a randomized study, patient baseline characteristics will be compared between the Steroids and No Steroids groups. Distribution normality will be tested by Kolmogorov-Smirnov test. Dichotomous and categorical variables will be compared by a two-sided chi-square or Fisher's exact test. Continuous variables will be compared by a two-tailed, independent samples t test or Mann-Whitney exact U test.

We will conduct a cumulative incidence competing risks analysis (CICR) to determine cause-specific hazard ratios (CSHRs) and their 95% confidence intervals (CIs) for death due to a postresuscitation infection, i.e., lethal septic shock, death due to a noninfectious cause, and poor inhospital outcome (as defined in the outcome measures subsection).

Prespecified risk factors for physical death or poor inhospital outcome will be group (Intervention vs. Control); cardiac arrest cause (cardiac vs. non-cardiac); cardiac arrest area (monitored vs. non-monitored), initial cardiac arrest rhythm (shockable vs. non-shockable); cardiac arrest time (i.e. weekday vs. holiday and nighttime vs. morning-to-late evening), advanced life support (ALS)-related bicarbonate dose; time from resuscitation team call to ALS initiation plus ALS duration; and therapeutic hypothermia (yes vs. no).

Additional analyses will include addition of early postresuscitation mean arterial pressure as covariate to the multivariable CICR Cox models