Outcomes measured by mortality rates, quality of life and degree of autonomy in the first year in stroke units in Spain

CONOCES was a prospective, observational, epidemiological, naturalistic, multicentre study of costs and outcomes of the disease in the Spanish healthcare setting in a sample of patients who had suffered a confirmed stroke [19]. The inclusion criteria were: being over 18 years of age; confirmed clinical diagnosis of first ischaemic or haemorrhagic stroke within 24 hours of onset; admission to a stroke unit; voluntary participation in the study; and signed informed consent by the patient and/or their primary caregiver. During the recruitment phase patients who met the inclusion criteria were enrolled consecutively in the study in all the participant centres. The study variables were collected by the medical team through personal interviews with the patient and the primary caregiver during the hospital stay in the stroke units and at the follow-up visits at 3 and 12 months. In the first data collection all the autonomy and quality of life scales were recorded at hospital discharge. Only the mRS was also recorded at admission. The protocol was approved by the Ethics Committee of Hospital Clínic i Provincial de Barcelona.

The variables were socio-demographic (age, gender and residence) and clinical (activation of stroke protocol, time to neurological care, aetiology, recurrence, presence of atrial fibrillation (AF), thrombolytic therapy and death during the follow-up period) in nature. HRQOL and degree of autonomy, as well as their determining factors, were also measured by determining neurological status on admission using the National Institutes of Health Stroke Scale (NIHSS) [20] and by assessing the patients’ disability by measuring functional status with the Barthel Index (BI) [21] and degree of functional dependence with the modified Rankin Scale (mRS) at discharge, three months and one year [22]. BI and mRS are the most commonly used measurement scales to assess the level of autonomy of stroke patients. mRS has become the most widely used clinical outcome measure for stroke in clinical trials. The scale ranks from 0 to 6, from perfect health with no symptoms to death [22]. Although the mRS classifies deaths in category 6, we decided to exclude them from the functional status analyses. BI is an ordinal scale used to measure performance in daily living activities. Each performance item is rated on this scale with a given number of points assigned to each level or ranking. It uses ten variables to describe mobility, feeding, toilet use, dressing, bathing, faecal and urinary incontinence, help needed with grooming and other activities. A higher number is associated with a greater likelihood of being able to live at home with a degree of independence following discharge from hospital. Each item can score 0, 5 or 10 points, so that BI score range from 0 to 100, corresponding to five levels of dependence: independent (100 points), low dependence (91–99 points), moderate dependence (61–90 points), severe dependence (21–60 points), total dependence (0–20) [21,23]. NIHSS has become the most widely-used scale to assess initial neurological status, in the follow-up of patients on treatment and also for routine clinical monitoring in stroke units [24,25]. It is composed of 11 items, each of which scores a specific ability between 0 and 4. For each item, a score of 0 typically indicates normal function in that specific ability, while a higher score is indicative of some level of impairment. The individual scores from each item are summed in order to calculate a patient’s total NIHSS score. The maximum possible score is 42, with the minimum score being 0 [24,25].

HRQOL was measured using the generic questionnaire EuroQuol (EQ-5D) at the same three time points [26]. The EQ-5D questionnaire was completed by the patients at the scheduled follow-up visits for the study. EQ-5D is a standardised instrument to measure self-reported health status with respect to five dimensions (mobility, self-care, usual activities, pain/discomfort, anxiety/depression) and three levels of severity (no problems, some or moderate problems, and extreme problems). It is characterised by the fact that it provides a single index value for each health state which can have a value of 1 or less, where 1 is equivalent to full health and 0 is death. These were normalised using conversion tables obtained in the general Spanish population [27]. When the patient was unable to communicate, the interviewer conducted the test to a close relative or caregiver [28].

Statistical differences in the comparison of contingency tables were analysed using the chi-square test, taking a value of α equal to or less than 0.05. Given that the variables are not normally distributed, the Wilcoxon test was used to measure the statistical significance of differences between means, as with ordinal data. All statistical analyses were carried out using SPSS software (version 21; SPSS Inc., Chicago, IL).

To analyse the magnitude of the observed change in the quality of life, the value of the final EQ-5D was compared with the original and with a sample of general population over 65 using the effect size [29]. The EQ-5D sample used to describe general population characteristics proceeded from a random sample of general population from the Canary Islands [30] which mean score was 0.73 and standard deviation 0.32. Cohen defined as non-significant an effect size of less than 0.2, an effect size between 0.2 and 0.5 as small, between 0.5 and 0.8 as moderate, and greater than 0.8 as large [29].

To analyse the association of clinical variables on admission with functional capacity and HRQOL after a year, statistical logistic regression analyses were performed in which the dependent variables were mortality, the value 1 in EQ-5D, modified Rankin Scale level 0–2, Barthel below 61 (total or severe dependence) and Barthel 100. The independent variables were age, gender, arrival time to neurological care, aetiology (ischaemic or haemorrhagic), AF, NIHSS, hospital arrival time and thrombolytic therapy (yes/no). The likelihood of receiving thrombolytic therapy was further investigated in a logistic regression model in which the explanatory variables were age, gender, NIHSS and the time of arrival.

Furthermore, we used an approach based on the analysis of ROC (Receiver Operating Characteristic) curves to measure sensitivity to change. This allowed us to assess the ability of the instruments used on admission, such as the NIHSS, the mRS and the BI, to detect patients with good prognosis, using the same methodology that is used to determine whether or not a diagnostic test identifies the condition of interest [31]. The ROC curve is obtained by plotting sensitivity for each breakpoint on the y-axis and its corresponding 1-specificity on the x-axis. If the instrument is sensitive, it has an ROC curve where sensitivity peaks have associated minimum values of 1-specificity. Conversely, if the instrument is insensitive, it has an ROC curve near to the reference line (line dividing the axis into two equal parts, which indicates the minimum level of discrimination). To characterise the instrument as discriminant, areas greater than 70% are required. The statistical method also calculated the statistical significance of the curve.