Association of Socioeconomic Status With Ischemic Stroke Survival

Rosa Maria Vivanco-Hidalgo, MD, PhD, MPH; Aida Ribera, PhD; Sònia Abilleira, MD, PhD


Stroke. 2019;50(12):3400-3407. 

In This Article


Data Availability Statement

The data that support the findings of this study are available from the corresponding author on reasonable request.

Data Sources

The Catalan Central Registry of Insured Persons collects sociodemographic information of all residents of Catalonia (7.5 million people) and identifies them with a unique healthcare ID code. This healthcare ID code allows for keeping track of Catalan residents across several health administration databases including the Catalan Central Registry of Insured Persons dataset, which aside of sociodemographic data includes date (but not cause) of decease, the acute hospitals discharge dataset (CMBD-HA, Conjunt Mínim Bàsic de Dades d'Hospitalitzats d'Aguts), the Pharmacy claims databases, and the database for primary care (CMBD-AP, Conjunt Mínim de Dades d'Atenció Primària) that records information on comorbidity and date of diagnosis. Both CMBD databases register the information using International Classification of Diseases (ICD) codes (ICD, Ninth Revision, for CMDB-HA and ICD, Tenth Revision [ICD-10], for CMBD-AP), and Pharmacy Claims Databases use WHOCC–ATC/DDD Index codes (World Health Organization Collaborating Centre–Anatomical Therapeutic Chemical/Defined Daily Dose). The registry has an automated data validation system to check consistency of data and identify potential errors, and external audits are performed periodically to ensure quality and reliability of data.

Detailed descriptions of the ascertainment methods for incident stroke cases, cardiovascular risk factors, comorbidities, and deaths are described elsewhere.[8]

Study Population and Design

Using the CMBD-HA dataset, we identified first ischemic stroke patients (ICD, Ninth Revision, codes 433.x1, 434.xx, and 436.5) hospitalized from January 1, 2015, to December 31, 2016. Patients were followed up from admission until death or until October 1, 2018 (end of follow-up).

Individual Socioeconomic Status

In Catalonia, all residents are granted universal healthcare by law and the use of the health system services is free at the point of use, with the only exception of drug dispensation.[9] Drug dispensation follows a system of copayment calculated according to individuals' income or to the social security benefits received. Individual socioeconomic status was derived from the information used to calculate the levels of copayment and these appeared in categories predefined in the Catalan Central Registry of Insured Persons as follows: exempts (nonworking population or people receiving noncontributory pension), <€18 000 ($US 20 468) income per year; €18 000 to €100 000 ($US 113 710) income per year; >€100 000 income per year.

PCSA Index

To strengthen territorial equity in the allocation of primary care resources, in 2015, the Catalan Health Department developed a socioeconomic deprivation indicator representative of the primary care service areas and linked to adverse health outcomes. The 13 variables used in this index of deprivation are related to the presence of social inequalities on which the primary care has a mitigating effect. The indicator score ranges from 0 (less deprived) to 100 (more deprived).[10]

Cardiovascular Risk Factors, Comorbidity Index, and Other Covariables

Information regarding cardiovascular risk factors was obtained by combining patients' active diagnoses from the CMBD-AP (hypertension ICD-10 codes I10 and I15.9; dyslipidemia ICD-10 codes E78–78.9; diabetes mellitus ICD-10 codes E10–14.9; and atrial fibrillation ICD-10 code I48) and patients' active medication from Pharmacy Claims Databases (C02, C02K, C02L, C02N, and C02LX codes for antihypertensive agents; A10A, A10B, and A10X codes for insulin and other blood glucose-lowering drugs; C10A and C10B codes for lipid modifying agents).

The comorbidity index (Adjusted Morbidity Groups [AMGs]) score was also collected from the CMBD-AP. The AMG index is a morbidity measurement recently developed and adapted to the Spanish healthcare system, which enables classification of the population into 6 morbidity groups. AMG is comparable to other comorbidity measures, and its explanatory value has been checked by comparing it with morbidity measures such as the Charlson index or the number of chronic diseases.[11]

Information regarding reperfusion therapies (ICD, Ninth Revision, codes 99.10 and 39.74 for thrombolytic and endovascular treatment, respectively) was collected from CMBD-HA.

Standard Protocols, Registrations, and Patients' Consent

No patients were involved in setting the research question or the outcome measures, nor were they involved in the design or implementation of the study. No patients were asked to advise on interpretation or writing up of results. This observational study used retrospective deidentified data from different health-administrative databases. Therefore, neither informed consent nor ethical committee approval was required according to the current legislation in Catalonia.

Statistical Analysis

Descriptive statistics were estimated by individual socioeconomic categories and PCSA index quartiles. Because of the low number of patients in the individual socioeconomic category of >€100 000 income per year (50 cases), we decided to merge the categories €18 000 to €100 000 income per year and >€100 000 income per year. Quartile 1 of PCSA index represents the least deprived area, whereas quartile 4 represents the most deprived.

Survival Analysis. We estimated survival using crude fatality rates (at 30 days, 1 year, 2 years, and 3 years) and Kaplan-Meier curves, to examine survival rates adjusted by sex, AMG, and age between individual socioeconomic categories and according to primary care service area quartiles.

Because we were interested in evaluating the impact of individual socioeconomic status on short- (30 days) and long-term survival after stroke, and socioeconomic status varies between PCSA, we fitted mixed-effects logistic and survival models, which take into account random effects because of clustering of patients in PCSA.

We fitted 4 a priori models including potential confounding variables in successive steps to assess the changes in the magnitude of association between the individual socioeconomic status and survival: model 1, individual socioeconomic status categories plus sex, age, and reperfusion therapy; model 2, model 1 plus AMG index; model 3, model 2 plus cardiovascular risk factors; model 4, model 3 plus PCSA index.

All 2-way interactions with individual socioeconomic status were tested. We checked the proportional hazards assumption for all covariates using graphical methods (inspection of log minus log plot of survival).

We used Stata statistical software, version 15, for all analyses.