Study for the Improvement of Long-Term Outcome Prediction in Patients in Coma After Cardiac Arrest

HOPE conducts a 2.5-year multicenter prospective cohort study in Germany on outcome prediction in the early stages of coma caused by cerebral hypoxia-ischemia.

The main features of the project are:

• Control for a self-fulfilling prophecy
• Long-term follow-up (12 months) covering acute and neurorehabilitation phases
• Use of sensitive measures of level of consciousness
• Further in-depth and systematic examination of the predictive value of somato-sensory evoked potentials (SSEP)

Using as a reference existing guidelines and previous research, we consider the following factors to be predictors of a negative outcome with a high specificity of > 90% with small 95% confidence intervals: loss of cortical N20 SEP potentials, loss of more than one brain stem reflex, non-reactive EEG, or a NSE above 33 mcg/L. We refer to the presence of any of these specific unfavourable test results during the ICU treatment as a "negative predictor". Therefore, our null hypothesis is that the probability of a favourable outcome, despite the presence of a negative predictor, is < 10%. We will use Simon's two-stage design. The null hypothesis will be tested against a one-sided alternative. In the first stage, we will accrue 72 patients. If there are 8 or fewer good outcomes in these 72 patients, the study will be discontinued because the null hypothesis could not be rejected. Otherwise, we will recruit 100 additional patients to gather a total of 172. The null hypothesis will be rejected if 28 or more favourable outcomes are observed in these 172 patients. This design yields a type I error rate of 0.05 and power of 80% if the true response rate is 25%, i.e. if indeed 25% or more patients have a favourable outcome despite a negative predictor. This calculation is based on the assumption that 25% of the enrolled patients have at least one negative predictor.

HOPE has seven data entry points (i.e. study visits).The primary endpoints are completion of the follow-up period or patient death. HOPE collects clinical and demographic data in accordance to Utstein-style recommendations. The most crucial part of the study is determining the current level of consciousness by means of the revised version of the Coma Recovery Scale (CRS-R). The CRS-R allows a precise categorization of the current level of consciousness based on a structured and systematic clinical examination of auditory, visual, motor, oromotor, communication and arousal functions. Other assessment instruments employed yield information about functional outcome/ADL, neuropsychological performance, and health-related quality of life.

Study centers will collect data prospectively following a data acquisition timeline (e.g. t0 to t6). Visits t0 to t2 will take place at acute-setting ICUs; visits t3 to t5 at neurorehabilitation centers; and visit t6 at the patients' location of residence 12 months after day 0. A manual and data dictionary will be provided to all centres along with the case report forms (CRF). All study centers will send pseudonymized data weekly by certified mail to the Central Study Centre (CSC) located at the Department of Neurology of the University of Munich. All study centres, including the CSC, will check the data for completeness, plausibility, accuracy, consistency and outliers. Legal representatives of coma patients, who decline to participate, will be asked to provide data on socio-demographic variables and reason for declining. The CSC will store the data at a database and will be responsible for safeguarding and analysing it. Patients/Legal representatives will be asked to remain in the study for a further follow-up. After completion of the follow-up period, datasets will be irreversibly anonymised.

We will analyze the positive predictive value of a dichotomous prognostic marker (favorable or unfavourable SSEP). For this purpose, we will use two definitions of favorable outcome, i.e. a functional and a behavioral one. If outcome is defined with respect to independence in daily life, we will use a modified Rankin Scale score of 0-3 (no symptoms - moderate disability, but able to walk independently) to categorize a favorable functional outcome. In the context of a catastrophic event such as cardiac arrest with high likelihood of permanent VS, it is our experience that some patients and families consider regaining functional communications skills also a favourable outcome. In a separate analysis, we will therefore use the recovery of full consciousness as measured by the CRS-R to define a favorable behavioral outcome. To analyse the effect of covariates on functional and behavioral outcomes, we will use mixed effects regression models, including both fixed and random effects. This method of analyzing longitudinal data is highly effective to examine change trajectories with several unequally spaced waves of data. Covariates will be sociodemographic variables, prognostic variables and treatment factors (e.g. use of therapeutic hypothermia), adjusted for age and sex.