Patient Cohort and Procedures
This multicenter prospective cohort study was performed in 10 Dutch mixed medical-surgical intensive care units (ICUs) of 1 university-affiliated and 9 non–university-affiliated hospitals. The cohort included adult patients, admitted to the ICU, who remained in a coma after CPR, and who were treated with mild hypothermia. Exclusion criteria were preexisting diseases with a life expectancy of less than 6 months, severe disability before CPR, CPR due to hypovolemic shock, and absence of informed consent. The cooperating hospitals were not obligated to keep records of patients admitted after CPR who were not eligible for the study. Included patients were treated with hypothermia (32°C–34°C) in combination with sedative drugs in accordance with the local hospital's protocol. Sedative drugs used were midazolam, propofol, and opiates. Patients who died during hypothermia treatment were not excluded from analyses.
The research protocol and consent procedures were approved by the ethics committees of all collaborating hospitals. Informed consent was obtained from a legal representative shortly after the patient's hospital admission. When the patient regained consciousness and was able to judge the situation properly, informed consent was also obtained from the patient.
Neurologic examination, consisting of the Glasgow Coma Scale (GCS) score and brainstem reflexes, was performed 48 and 72 hours after CPR by the attending ICU physician or a neurologist.
Serum samples for determination of NSE levels were drawn on ICU admission, 12 hours after reaching the target temperature of 32°C to 34°C, and 36 and 48 hours after collapse. Samples were stored at −20°C for later analysis in a central laboratory (Radboud University Nijmegen Medical Center), using monoclonal 2-sided single-incubation immunoluminometric assay with a Liaison automated analyzer (DiaSorin, Saluggia, Italy). Results of serum NSE levels were not available to the treating physicians.
Cortical N20 responses of median nerve SEPs were recorded with standard procedures during hypothermia.14 A second SEP was performed in patients who remained in a coma after regaining normal body temperature and who were thought to have cleared all sedative drugs and metabolites. Recordings were assessed by the local clinical neurophysiologists and documented as “absent” (bilaterally absent cortical N20 responses after left and right median nerve stimulation, in the presence of a cervical potential), “present” (cortical N20 response present on at least 1 side), or “undeterminable” (technically insufficient recording). Undeterminable SEP results were considered as present. The results of the SEPs recorded during hypothermia were not disclosed to the treating physicians to avoid any influence of the test results on treatment decisions. If a second SEP was performed after the patient had regained a normal body temperature, the results were disclosed to the treating physicians and treatment decisions were recorded.
Neurologic outcome was assessed with the Glasgow Outcome Scale (GOS), 1 week, 1 month, and 6 months after admission via a telephone interview with the patient or a legal representative.15 The primary outcome of the study was poor outcome, defined as death, vegetative state, or severe disability after 6 months (GOS 1–3). Secondary outcomes were death or vegetative state (GOS 1–2) 1 month after CPR and mortality after 1 week, 1 month, and 6 months. The study protocol contained no guidelines for withholding or withdrawing treatment. These decisions were at the discretion of the treating physician. In the Netherlands the results of SEP in normothermic patients are used in most hospitals to predict prognosis and absent cortical responses are a reason to withdraw treatment.8
Additional data collected were gender, age, in- or out-of-hospital cardiac arrest, presenting rhythm (ventricular fibrillation [VF] or ventricular tachycardia [VT] vs asystole or pulseless electrical activity [PEA]), time from collapse to bystander basic life support, time from collapse to return of spontaneous circulation, previous medical history as described in the patient's chart, duration of hypothermia, time of discontinuation of sedative drugs, use of sedative drugs or neuromuscular blocking agents during SEP, use of sedative drugs during neurologic examination, and reasons for limiting or withdrawing care. Scored reasons for treatment restrictions were divided into neurologic (options: neurologic examination, SEP result, or other neurologic reasons) and non-neurologic (options: cardiac/hemodynamic complications or other reasons).
Patient characteristics were described according to their distribution. Variables were expressed as mean and standard deviation (continuous and normally distributed) or as medians and interquartile ranges (not normally distributed). Categorical variables were expressed as percentages (n). Differences between categorical variables were calculated by the χ2 test. The GOS was used as the reference variable for a patient's outcome. The percentages of correct identification of patients with a poor outcome were expressed as sensitivity and specificity. The false-positive rate (FPR) was calculated as 1 minus specificity.16, 17 The FPR is considered a single proportion. Due to the low observed proportions, 95% confidence intervals (CIs) were calculated using the recommended method of the Confidence Interval Analysis software.18, 19 Sensitivity, specificity, and FPR were calculated for all predictors of neurologic outcome (72 hours: motor score, pupillary light responses, corneal reflexes; NSE at admission, during hypothermia, and 36 and 48 hours after collapse; SEP during hypothermia and normothermia). Area under the receiver operating characteristic (ROC) curve was calculated for the NSE in relation to the GOS after 6 months. Statistical significance was considered to be at p < 0.05. When appropriate, statistical uncertainty was expressed by their 95% CIs. All analyses were performed with SPSS 18.1.