Predicting Long Term Survival in Patients With Primary Intracerebral Hemorrhage. Development and Validation of the ICHCat in Primary Care.
Predicting Long Term Survival in Patients With Primary Intracerebral Hemorrhage. Development and Validation of the ICHCat in Primary Care.
The clinical evidence shows that patients with a first episode of intracerebral hemorrhage (ICH) are increasingly old and with greater comorbidity with a recognized impact over mortality. The prediction of the outcome of ICH is not only crucial in the emergin attention to identify those patients with favorable criteria that can benefit from possible treatments; but also after hospital discharge, in primary care where the prediction should facilitate the organization and management of a wide variety of resources: familiar, health and social welfare. Even though there are different scales that predict mortality, these are not sufficiently useful in choosing a treatment or do not provide sufficient data to the family to decide.
Due to the characteristics of the population with ICH described in these works, it seems useful to propose a prognostic index (ICHCat) to identify the variables associated to its incidence and mortality and that, in addition, to make adjustments in the comparisons of the survival between different series of patients or different treatment modalities in primary care.
The authors carried out a multicentric and retrospective study of a cohort with an episode of ICH in the period between 2006-2016. The objective of this phase is to design a predictive index (ICHCat) of mortality in patients with an episode of ICH. Patient outcome was followed until death all causes or study end (30.09.2016) since date of ICH in the electronic health record. All people included were managed by the Public Health System in Catalonia. Registry information was collected from the government-run healthcare provider responsible for all inpatient care in the county.
Predictor Variables
Predictor variables were considered if they were viewed as commonly measured and available in primary care and with potential evidence of an association with ICH risk. Candidate variables included:
Sex: woman (0) man (1) Age: <80 year-old (1), ≥80 year-old (2). Number CCP criteria: <4 (0) ≥4 (1). Charlson comorbidity index. Short version. Current medications were asked about during the home visit and confirmed in medical records. Polypharmacy (defined as five or more daily medications): <5 (0), between 5-9 (1), and ≥10 (2). Oral anticoagulants (acenocumarol or warfarin) with TTR ≥60% (1), if TRT <60% (2) or New Oral Anticoagulants NOACs (0). Antidepressants and/or, sedating or other drugs affecting the neurologic system: man (1), woman (2). If there was a diagnostic of "atrial fibrillation", CHA2DS2VASC and HAS-BLED scores were included.
Recurrent falls or fall risk: no (0), yes (1). Hypertension not controlled by therapy (≥ 160/90 mmHg): no (0), yes (1). BP was measured as an average of separated follow-up measurements last six months.
Alcoholism abuse vs dependence: no (0), yes (1) Presence de cognitive impairment: a disease-specific diagnosis of cognitive impairment and/or prescription of specific treatment, without specification of sub-type or severity, was used and mesured by Pfeiffer test [2]: [0-2 errors] = Intact Intellectual Functioning (1); [≥3 errors] = Mild to severe Intellectual Impairment (2)].
Presence de disability: score in [Barthel ≥60 (1) <60 (2)] or in [Rankin <4 (1) 5(2)] were used to assess dependence in ADL Functional status: (0) Autonomus (1) Residual with caretaker (2) Dependence with caretaker (3) Long stay
Socio familiar risk: score in Gijon[18] scale 10-14 (1) ≥15 (2)]
Primary Outcome Given the limited survival, the primary outcome was death all causes occurring within the 5-year follow-up after the ICH episode.
Predictions of ICH mortality risk were based on Cox proportional-hazards regression models. All potential predictors were considered in a multivariate logistic regression, and a backward step selection procedure was carried out to pick the variables that composed the best model. Subsequently design a predictive model of multivariate Cox regression analysis to define the weight of each of the pathologies in the mortality. To assign the weight according to the value HR took into account only those with a HR ≥1.2 in the multivariate model approximating the value of HR to the nearest whole number:
The final score for each patient (ICH) was made up of the sum of their scores by age and comorbidity. To check the validity of the index we define the following levels of mortality risk:
The authors estimate the survival probability of each risk group according to the Kaplan-Meier method and compared these curves using the Log-Rank test. It will be used ROC curves i the AUC to assess the ability of the index ICH to perform properly the assignment of patients to different risk groups. The hazard ratio according to the proportional hazards model categorized in their levels of risk, in order to observe the increase in the risk of death that occurs from a risk group to the next. The statistical analysis consisted of a multivariate logistic regression model. The model selection has been caried out with an AIC-led forward step selection method. To ensure internal validity, 20% of data has been discarded in the phase of constructing the model and only used to assess model performance using ROC curves and the AUC.
In a second phase, there will be a retrospective study of validation in the cohort of patients PCC of the territory with an episode of ICH.
Inclusion Criteria:
Exclusion Criteria: