Study population
Our institution is a teaching non-for-profit tertiary private hospital of Santiago, Chile. Since 1997 we include all consecutive 18 year or older patients with an acute stroke admitted to Clínica Alemana de Santiago in our prospective stroke registry (Registro de Enfermedades Cerebrovasculares Clínica Alemana: RECCA).
In this study we selected patients with AIS, complete data on the 14-point Mediterranean Diet Adherence Screener (MEDAS) score and informed consent, admitted from February 2017 to July 2019. We excluded patients with transient ischemic attacks.
All variables were prospectively collected during patient hospitalization. Ischemic stroke was defined according to current practice.(11) Ischemic stroke etiology was classified using the Trial of Org 10172 in Acute Stroke Treatment (TOAST) classification.(12) Hypertension, Diabetes Mellitus, Insulin resistance, hyperlipidemia and mood disorder were defined as present in patients with the previous clinical diagnosis or in those under treatment for each specific risk factor. Vessel occlusion was defined present when found in intracranial o cervical extracranial arteries on acute vascular neuroimaging as per institutional protocol. Previous dependency was defined according to the modified Rankin Scale (mRS). Patients with suspected stroke were immediately evaluated by the neurologist. Acute neurological assessment includes: the National Institutes of Health Stroke Scale (NIHSS) score (13) and a neuroimaging stroke protocol consisting of: non-contrast Computed Tomography (CT) scan, Diffusion Weighted Magnetic Resonance Imaging (DWI-MRI) and cervico-craneal CT angiography (CTA) or magnetic resonance angiography (MRA). Additionally, many patients are studied with Transcranial Doppler and some Digital Subtraction Angiography (DSA). Etiologic evaluation includes electrocardiogram monitoring, Echocardiography and Carotid Ultrasonography if appropriate as well as coagulation, hematological and biochemical analyses.
Adherence to a Mediterranean diet was measured by the Spanish version of the MEDAS score, which we adapted to the local Chilean terms for an ongoing population-based stroke study.(14) The MEDAS is a 14-point questionnaire validated and used in the PREDIMED study to assess the adherence to a Mediterranean diet pattern (supplement); higher scores are associated with higher adherence, being 14 points the maximum score.(15) We used this short screener as it has been shown to be a valid tool for rapid assessment of adherence to Mediterranean diet pattern and proposed as useful in time-limited clinical and research settings. (16) Since January 2017, the stroke fellow or vascular neurology staff obtains the MEDAS score prospectively in all patients admitted with an AIS diagnosis, using a paper form previous to discharge in one session. Exposure before the index stroke was assessed by asking the patients (85%) or next of kin (15%) to refer to their last month usual diet pattern. Adherence to Mediterranean diet pattern was defined as low (score of 0-6 points) or high (score of 7-14 points) based on the median MEDAS scores in our cohort. The primary outcome was severity defined by the admission NIHSS score.
The local Ethics Committee and the Institutional Review Board of the hospital approved the study registry protocol and written informed consent was obtained in every patient as local regulatory law requests.
Statistical analyses
Sociodemographic characteristics, cardiovascular risk factors, medication use, clinical presentation, etiology, treatments and NIHSS scores were compared by adherence to the MEDAS score using Chi2 or Fisher´s test for frequencies. T-test was used for normally distributed continuous variables and Wilcoxon log-rank test for non-normally distributed continuous variables in the univariate analyses.
We investigated the association of adherence to Mediterranean diet and NIHSS scores computed as an ordinal variable using a Generalized Linear Model (GLM) with gamma distribution and link function identity adjusting for the following explanatory variables: Diabetes Mellitus, sedentary lifestyle, alcohol consumption, Body Mass Index (BMI), total and Low-Density Lipoprotein (LDL) cholesterol. These were chosen because their frequency distribution was significantly different in both groups in univariate comparison. In this model we further adjusted for prognostic variables usually associated with severity in prior studies, using Hosmer–Lemeshow criteria with a cut-off point of p < 0.25.(17) These prognostic variables were age, previous dependency (mRS 3-5), atrial fibrillation (AF), sedentary lifestyle, time from symptom onset to emergency consultation, vessel occlusion, admission glycemia, and cardioembolic etiology. In order to decrease over-adjustments, we only included cardioembolic stroke and not atrial fibrillation in the models. (18)(19)(20) In this model we added 1 to all NIHSS scores, because 62 (16.8%) patients had a cerebral infarction but an NIHSS of 0. In order to account for the effect of premorbid disability on severity we performed a sensitivity analysis stratifying by previous disability, excluding those with mRS 3-5 from the GLM model.
As a secondary analysis we performed an ordinal logistic regression (OLR) to determine the independent association of low or high adherence to Mediterranean diet to quartiles of NIHSS scores. NIHSS quartiles are: 0-1= q1; 2-3=q2; 4-9=q3; 10-maximum=q4. The prognostic and confounding variables in the ORL model were the same as in the previous model: Age, Diabetes mellitus, alcohol consumption, sedentary lifestyle, previous dependency, time from symptom onset to hospital admission, vessel occlusion, body mass index, admission glucose, total cholesterol, LDL cholesterol and cardioembolic stroke etiology.
All statistical analyses were performed with Stata 14.0. An alfa error < 0.05 was considered significant. The paper is reported according to the STROBE guidelines.(21)