Secondary Stroke Prevention in People With Schizophrenia

Authors: Moira K. Kapral, Joan Porter, Paul Kurdyak, Amy Y. X. Yu, Emilie Matheson, Jiming Fang, Leanne K. Casaubon, Eshita Kapoor, Kathleen A. Sheehan

Abstract

Background

People with schizophrenia are less likely than those without to be treated for cardiovascular disease. We aimed to evaluate the association between schizophrenia and secondary preventive care after ischemic stroke.

Methods and Results

In this retrospective cohort study, we used linked population‐based administrative data to identify adults who survived 1 year after ischemic stroke hospitalization in Ontario, Canada between 2004 and 2017. Outcomes were screening, treatment, and control of risk factors, and receipt of outpatient physician services. We used modified Poisson regression to model the relative risk of each outcome among people with and without schizophrenia, adjusting for age and other factors. Among 81 163 people with ischemic stroke, 844 (1.04%) had schizophrenia. Schizophrenia was associated with lower rates of screening for hyperlipidemia (60.5% versus 66.0%, adjusted relative risk [aRR] 0.88 [95% CI, 0.84–0.93]) and diabetes (69.4% versus 73.9%, aRR 0.93 [95% CI, 0.89–0.97]), prescription of antihypertensive medications (91.2% versus 94.7%, aRR 0.96 [95% CI, 0.93–0.99]), achievement of target lipid levels (low‐density lipoprotein <2 mmol/L) (30.6% versus 34.6%, aRR 0.86 [95% CI, 0.78–0.96]), and outpatient specialist visits (55.3% versus 67.8%, aRR 0.78 [95% CI, 0.74–0.83]) or primary care physician visits (94.5% versus 98.5%; aRR 0.96 [95% CI, 0.95–0.98]) within 1 year. There were no differences in prescription of antilipemic, antiglycemic, or anticoagulant medications, or in achievement of target hemoglobin A1c ≤7%.

Conclusions

People with stroke and schizophrenia are less likely than those without to receive secondary preventive care. This may inform interventions to improve poststroke care and outcomes in those with schizophrenia.

Clinical PerspectiveWhat Is New? This study found that in people with recent stroke, those with schizophrenia were less likely to receive screening for diabetes or hyperlipidemia and to achieve target lipid levels and had fewer visits to primary care providers or specialists with expertise in stroke and cardiovascular disease. What Are the Clinical Implications? Future work should focus on understanding the causes and consequences of these gaps in care. In the interim, given the elevated risk of recurrent stroke and death in this population, tailored interventions may be needed to improve the cardiovascular care of people with stroke and schizophrenia.

Schizophrenia is a major mental illness that is associated with increased stroke incidence, recurrence, and case fatality. ^1^ , ^2^ , ^3^ , ^4^ , ^5^ , ^6^ Cardiovascular risk factor modification can reduce the risk of recurrent stroke, and best practice guidelines include recommendations for blood pressure control, diabetes treatment, lipid management, anticoagulation for atrial fibrillation, carotid revascularization, and modification of lifestyle behaviors encompassing diet, physical activity, and smoking cessation. ^7^ , ^8^ , ^9^

Risk factor modification is of particular importance in those with schizophrenia, in whom there is a higher prevalence of hypertension, hyperlipidemia, diabetes, and smoking compared with the general population, and who may experience metabolic side effects such as obesity and insulin resistance from antipsychotic pharmacotherapy. ^10^ , ^11^ , ^12^ Despite this, a meta‐analysis of the quality of primary and secondary prevention screening and treatment for those with cardiovascular disease found that people with schizophrenia were less likely to receive any preventive services compared with the general population, and the disparity was largest for those with schizophrenia compared with other mental illnesses. ^13^ Few prior studies, however, have focused on stroke or on comprehensive stroke secondary prevention measures. Stroke, which usually results in hospitalization and contact with the health care system, represents an important opportunity for the identification and management of vascular risk factors.

In this study, our objective was to compare rates of screening for diabetes and hyperlipidemia; prescriptions for antiglycemic, lipid‐lowering, antihypertensive, and anticoagulant medications; and receipt of outpatient physician services in adults with recent stroke with and without schizophrenia. We hypothesized that schizophrenia would be associated with lower rates of screening and treatment.

Methods

Setting, Data Sources, and Study Sample

This was a population‐based retrospective cohort study set in Ontario, Canada, a province of >14 million people. Residents are insured for all medically necessary hospital, physician, and diagnostic services under a publicly funded health care system, and prescriptions for medications included in the provincial drug formulary are covered for those aged 65 years and older. We used inpatient hospitalization data from the Canadian Institute for Health Information to extract the first record of adults 18 years and older with a primary diagnosis of acute ischemic stroke and discharged alive between April 1, 2004 and March 31, 2018, using validated codes (International Classification of Diseases, Tenth Revision, Canada [ICD‐10‐CA] codes I63.x, I64.x, and H34.1). ^14^ By law in Ontario, all hospitalizations are included in the Canadian Institute for Health Information database. We excluded people with an invalid health card number, non‐Ontario residents, strokes that occurred after admission to the hospital, and those who were admitted to long‐term care or received any palliative care in the 1 year following discharge. Because testing and treatment decisions may vary with first and subsequent stroke, we focused on first stroke and applied a lookback to 1991 to exclude individuals with a prior hospitalization for hemorrhagic or ischemic stroke (ICD‐9 codes 430.x, 431.x, 434.x, 436.x; ICD‐10‐CA codes I60.x, I61.x, I63.x, I64.x, H341). We also excluded those who died within 1 year of discharge in order to have a full year of data for analyses of medications and laboratory tests. In secondary analyses, we included those who did not survive to 1 year. Please see Figure S1 for the cohort flow chart.

We linked this incident ischemic stroke cohort to other population‐based administrative databases including the Ontario Laboratory Information System for laboratory results, the physician claims database for physician visits, the Ontario Drug Benefits Database for prescription claims (only available for the population 65 years and older), the Registered Persons Database for demographics and death date, and the Postal Code Conversion File and the Canada Census for income quintile allocation and identification of rural place of residence.

Exposure

We identified schizophrenia using a validated algorithm ^15^ that consisted of meeting any 1 of 3 conditions indexed to the admission date of the stroke and with lookback to 1988 as

(1) any hospitalization in a general hospital bed with a primary diagnosis of schizophrenia or schizoaffective disorder (ICD‐9295.x or ICD‐10‐CA F20.x or F25.x); or (2) any hospitalization in a psychiatric hospital bed with a primary diagnosis of schizophrenia (DSM‐IV 295.x); or (3) 3 outpatient physician billings for schizophrenia (ICD‐9295.x) within a 3‐year period.

This diagnostic algorithm has a sensitivity of 97% and a specificity of 65% for the diagnosis of schizophrenia. ^15^

We identified the comorbidities of diabetes, hypertension, atrial fibrillation, and hyperlipidemia based on validated algorithms and published definitions ^16^ , ^17^ , ^18^ , ^19^ (see Table S1 for details) and calculated Charlson comorbidity scores and grouped scores as <2 (minimal comorbidity) or ≥2 (moderate to high comorbidity). ^20^ , ^21^ We assigned neighborhood‐level income quintile by linking each patient's postal code to the dissemination area of their principal residence and determining the median income of each neighborhood based on the Canada Census cycle that corresponded to the index date. Rural residence was determined based on postal code, and was defined as a community with a population size under 10 000, outside the commuting zones of census metropolitan and census agglomeration areas. ^22^

Secondary Prevention Indicator Definitions

We did not have a predefined primary end point, but rather described a spectrum of quality metrics. We evaluated the following indicators for stroke secondary prevention, based on guidelines ^7^ , ^8^ : (1) screening for diabetes; (2) screening for hyperlipidemia; (3) in the subgroup with hypertension, filling of a prescription for antihypertensive medications; (4) in the subgroup with diabetes, filling of a prescription for antiglycemic medications; (5) in the subgroup with hyperlipidemia, filling of a prescription for lipid‐lowering medications; (6) in the subgroup with atrial fibrillation, filling of a prescription for anticoagulant medications; (7) in the subgroup with diabetes, achievement of glycosylated hemoglobin ≤7%; (8) in the subgroup with hyperlipidemia, achievement of low‐density lipoprotein ≤2 mmol/L (70 mg/dL); (9) any outpatient visit to a general practitioner, medical specialist likely to have expertise in cardiovascular disease (neurologist, internist, cardiologist, or geriatrician), or psychiatrist (see Table S2 for details). For all indicators, we reported whether the service was provided within 1 year of discharge. We did not include testing done during the index admission, because our hospitalization databases do not include laboratory information. Analyses of medications were limited to the subgroup aged 65 years and older. Our data sources did not allow us to analyze other recommended processes of care, such as screening for atrial fibrillation, achievement of target blood pressure control, use of antiplatelet agents (because acetylsalicylic acid is available over the counter and may not be captured in the prescription claims database), or management of smoking, physical inactivity, and other lifestyle factors. In addition, although current guidelines ^12^ recommend a target low‐density lipoprotein of ≤1.8 mmol/L, we selected a target of ≤2.0 mmol/L to align with guidelines from the early part of our study period.

Statistical Analysis

We compared baseline characteristics of those with and without schizophrenia using standardized differences of the mean, where differences of ≥0.1 are generally considered indicative of clinical significance. ^23^ For context, we also compared baseline characteristics of our study cohort with patients who were excluded for either prior stroke or for death within 1 year of discharge. For outcomes, we used χ^2^ tests to compare the percentage of those with and without schizophrenia who received the secondary prevention service adjusting the denominator as appropriate to include only those who were eligible for the service. We calculated 95% CIs based on Wald confidence limits for the binomial proportions. We did not adjust P values or CIs for multiple hypothesis testing.

We used Poisson regression modified by using a robust error variance procedure (sandwich estimation) ^24^ to model the relative risk of the indicator outcome among people with and without schizophrenia surviving to 1 year from discharge, adjusting for age, sex, neighborhood income quintile, Charlson score (grouped as 0–1 and ≥2), and rural residence. We did not adjust for individual comorbid conditions. We modeled age as a continuous variable using restricted cubic splines with 5 knots located at the 5th, 27.5th, 50th, 72.5th, and 95th percentiles of age. ^25^ In secondary analyses, we included those who did not survive to 1 year.

Standard Protocol Approvals, Registrations, and Patient Consent

Data sets used in this project were linked using unique encoded identifiers and analyzed at ICES (formerly known as the Institute for Clinical Evaluative Sciences) using SAS version 9.4 (SAS Institute Inc., Cary, NC). ICES is an independent, nonprofit research institute whose legal status under Ontario's health information privacy law allows it to collect and analyze health care and demographic data, without consent, for health system evaluation and improvement. The use of data is authorized under section 45 of Ontario's Personal Health Information Protection Act, which does not require review by a Research Ethics Board.

Data Availability

The data set from this study is held securely in coded form at ICES. While legal data‐sharing agreements prohibit ICES from making the data set publicly available, access may be granted to those who meet prespecified criteria for confidential access, available at www.ices.on.ca/DAS. The full data set creation plan and underlying analytic code are available from the authors upon request, understanding that the computer programs may rely upon coding templates or macros that are unique to ICES and are therefore either inaccessible or may require modification.

Results

Over the period of this study, 81 163 people were hospitalized for an incident ischemic stroke and survived to 1 year, 844 (1.04%) of whom also had a diagnosis of schizophrenia. Those with schizophrenia, compared with those without, were younger (median age 64 years versus 71), more likely to be female (50.0% versus 44.5%), to live in the lowest income quintile neighborhoods (41.5% versus 22.8%), to have diabetes (42.4% versus 31.9%), to have an overall higher level of comorbidity as measured by a Charlson score ≥2 (52.1% versus 45.5%), and were less likely to have a history of atrial fibrillation (10.0% versus 18.8%) (Table 1). There was no difference in the proportion of those with hyperlipidemia or seen at a comprehensive stroke center. Compared with those excluded for prior stroke and those who did not survive to 1 year, patients in our study cohort were younger, less likely to be female, and had lower comorbidity (Table S3).

People with schizophrenia, compared with those without, were less likely to be tested for diabetes (69.4% versus 73.9%, adjusted relative risk [aRR] 0.93 [95% CI, 0.89–0.97]) or hyperlipidemia (60.5% versus 66.0%, aRR 0.88 [95% CI, 0.84–0.93]), and among those with these conditions, people with schizophrenia were less likely to fill a prescription for antihypertensive medication (91.2% versus 94.7%, aRR 0.96 [95% CI, 0.93–0.99]) or achieve control of hyperlipidemia (30.6% versus 34.6%, aRR 0.86 [95% CI, 0.78–0.96]) (Table 2 and Figure). There were no differences in prescription of antilipemic (90.1% versus 93.7%, aRR 0.96 [95% CI, 0.91–1.01]), antiglycemic (72.2% versus 72.7%, aRR 0.94 [95% CI, 0.86–1.02]), or anticoagulant (82.0% versus 85.2%, aRR 0.95 [95% CI, 0.84, 1.07]) agents, or in achievement of target hemoglobin A1c ≤7% (38.5% versus 45.1%, aRR 0.90 [95% CI, 0.79–1.02]). Those with schizophrenia were less likely than those without to have any outpatient visit to a primary care physician or medical specialist within 1 year of discharge (primary care 94.5% versus 98.5%, aRR 0.96 [95% CI, 0.95–0.98]; medical specialist 55.3% versus 67.8%, aRR 0.78 [95% CI, 0.74–0.83]) and were more likely to have at least 1 visit to a psychiatrist within 1 year (42.8% versus 4.9%, aRR 7.14 [95% CI, 6.55–7.78]) (Table 2 and Figure; data for psychiatrist visits not shown). Unadjusted results are shown in Table S4. Results were similar when people who did not survive to 1 year were included (Tables S5 and S6).

Discussion

In this large, population‐based study conducted in a province where all residents receive hospital, physician, and laboratory services at no cost, we found that people with recent ischemic stroke and schizophrenia were less likely than those without to receive screening for diabetes or hyperlipidemia and to achieve target lipid levels, and had fewer visits to primary care providers or specialists with expertise in stroke and cardiovascular disease. These findings are concerning given the elevated risk of recurrent stroke and death in this population, as well as the younger age at stroke onset in those with schizophrenia, conferring more years at risk for future events.

Disparities in the screening and management of cardiovascular diseases in people with mental health conditions have been previously documented in a variety of health care settings, with a meta‐analysis finding that undertreatment is particularly marked in those with schizophrenia compared with other conditions. ^5^ , ^13^ , ^26^ This study adds to the existing literature by focusing specifically on a cohort with stroke and identifies gaps in screening and management even in this selected high‐risk population with clear indications for therapy, a recent hospitalization creating opportunities for close contact with the health care system, and (with the exception of a minimal co‐pay for prescriptions) no out‐of‐pocket expenses for the interventions under study. Although the absolute differences in care were modest, these persisted after adjustment for age, sex, and comorbid conditions, were seen across the continuum of stroke secondary preventive care, from screening to treatment to control of risk factors, and occurred despite a similar proportion of people with and without schizophrenia receiving care at comprehensive stroke centers.

Potential explanations for gaps in outpatient care include socioeconomic barriers to health care access (difficulty navigating the health care system, payment for transportation to attend appointments, lack of mobile phones for communication), provider factors (lack of awareness or comfort in prescribing cardiovascular medications by mental health professionals, lack of attention to medical comorbidity in people with psychiatric disorders), patient factors (mental illness–related cognitive and social factors that create barriers to engagement with health services or limit participation in behavioral interventions such as exercise and dietary change ^27^ , ^28^ , ^29^ ), or bias on the part of health care providers. ^3^ The differences in physician services after stroke were particularly striking, with lower use of both primary care and medical specialist visits in those with schizophrenia, and suggest missed opportunities for physician assessments with a specific focus on the secondary prevention of cerebrovascular disease. Given the high proportion of people with schizophrenia who receive care by psychiatrists after stroke, better integration of medical care within the mental health care system (for example, through psychiatric and stroke specialist or primary care provider colocation and comanagement) ^10^ , ^12^ , ^30^ could be a useful tool for improving cardiovascular care in this population. This study did not assess care provided during the index hospitalization, but prior work has shown that people with schizophrenia are less likely to be prescribed medications for secondary stroke prevention at hospital discharge, ^2^ suggesting that interventions to optimize care in the inpatient setting may also be needed. Other potential strategies include a focus on lifestyle modification such as nutritional counseling, smoking cessation, and physical activity interventions to reduce risk factors, ^31^ chronic illness self‐management programs, ^27^ improving health literacy, and use of technology‐based tools such as automated telephone reminders to improve medication adherence. ^32^ Finally, interventions need to extend beyond individual patients, providers, and the health care system, and include policies that address the broader systemic factors that affect health in people with schizophrenia, such as education, housing, income, environment, and community and mental health supports. ^33^

It is worth noting that the observed differences in poststroke care in those with and without schizophrenia are smaller than those previously seen in the context of other medical conditions such as acute myocardial infarction, diabetes, and cancer, where schizophrenia has been associated with at least a 50% lower likelihood of receiving recommended interventions. ^27^ , ^34^ , ^35^ , ^36^ Ontario has a well‐established coordinated stroke system that aims to optimize acute stroke care and secondary prevention for all. ^37^ Further study of the relative differences in care and outcomes for physical health conditions in those with schizophrenia may help to identify optimal care delivery models and determine whether any components of the stroke system can be applied to the management of other physical diagnoses.

Study limitations include lack of medication data for those aged <65 years. In addition, we are likely to have overestimated medication use because we considered a medication as being prescribed if a single prescription was filled in the year following discharge and did not assess discontinuation or nonadherence. The algorithm used to identify people with schizophrenia had a specificity of only 65%, so we anticipate that some people were misclassified as having schizophrenia and this would have biased our findings to the null. We also did not have information on some variables that could affect processes of care, such as the severity and management of psychiatric illness, severity of stroke, the use of specific antipsychotic medications, the presence of conditions such as smoking and obesity, or lifestyle factors such as diet and physical activity. Due to limitations in our available data, we were unable to evaluate all aspects of stroke secondary preventive care and could not determine whether patients were seen by neurologists with specific stroke expertise or in dedicated stroke prevention clinics. We did not have information on laboratory testing performed during the index stroke hospitalization, which would have led to underestimates of receipt of screening. We did not have information on the reasons for differences in care and did not evaluate their clinical consequences or potential interventions. Our study was conducted in a province with publicly funded health care and a well‐established stroke system, and the findings may not be generalizable to other jurisdictions. Finally, we restricted our analyses to people who survived for 1 year after stroke, and this may have biased our findings in either direction. Despite these limitations, this study is strengthened by the inclusion of all people hospitalized with stroke during the study time frame across an entire province, the use of a validated algorithm to identify those with schizophrenia, linkages to population‐based laboratory and medication databases to measure processes of care, and conduct within a health care system with universal coverage for the services under consideration.

People with schizophrenia are less likely than those without to receive secondary preventive care after stroke. While our study does not provide information on the causes or consequences of these gaps in care, given the known association between schizophrenia and stroke incidence, mortality, and case fatality (the latter in part due to deaths from cardiovascular disease), addressing deficiencies in cardiovascular risk factor modification in those with schizophrenia should be a focus of future tailored interventions.

Sources of Funding

This study was funded by a grant‐in‐aid from the Heart and Stroke Foundation of Canada (HSFC G‐17‐0018192).

Disclosures

None.

Supporting information