Deriving a Passive Surveillance Stroke Severity Indicator From Routinely Collected Administrative Data: The PaSSV Indicator

Association Between Immigration Status and Ambulatory Secondary Stroke Preventive Care in Ontario, Canada

BACKGROUND AND OBJECTIVES

Secondary stroke preventive care includes evaluation and control of vascular risk factors to prevent stroke recurrence. Our objective was to evaluate the quality of ambulatory stroke preventive care and its variation by immigration status in adult stroke survivors in Ontario, Canada.

METHODS

We conducted a population-based administrative database-derived retrospective cohort study in Ontario, Canada. Using immigration records, we defined immigrants as those immigrating after 1985 and long-term residents as those arriving before 1985 or those born in Canada. We included community-dwelling stroke survivors 40 years and older with a first-ever stroke between 2011 and 2017. In the year following their stroke, we evaluated the following metrics of stroke prevention: testing for hyperlipidemia and diabetes; among those with the condition, control of diabetes (hemoglobin A1c ≤7%) and hyperlipidemia (low-density lipoprotein <2 mmol/L); medication use to control hypertension, diabetes, and atrial fibrillation; and visit to a family physician and a specialist (neurologist, cardiologist, or geriatrician). We determined age and sex-adjusted absolute prevalence difference (APD) between immigrants and long-term residents for each metric using generalized linear models with binomial distribution and an identity link function.

RESULTS

We included 34,947 stroke survivors (median age 70 years, 46.9% women) of whom 12.4% were immigrants. The receipt of each metric ranged from 68% to 90%. Compared with long-term residents, after adjusting for age and sex, immigrants were slightly more likely to receive screening for hyperlipidemia (APD 5.58%; 95% CI 4.18-6.96) and diabetes (5.49%; 3.76-7.23), have visits to family physicians (1.19%; 0.49-1.90), receive a prescription for antihypertensive (3.12%; 1.76-4.49) and antihyperglycemic medications (9.51%; 6.46-12.57), and achieve control of hyperlipidemia (3.82%; 1.01-6.63). By contrast, they were less likely to achieve diabetes control (-4.79%; -7.86 to -1.72) or have visits to a specialist (-1.68%; -3.12 to -0.24). There was minimal variation by region of origin or time since immigration in immigrants.

DISCUSSION

Compared with long-term residents, many metrics of secondary stroke preventive care were better in immigrants, albeit with small absolute differences. However, future work is needed to identify and mitigate the factors associated with the suboptimal quality of stroke preventive care for all stroke survivors.

Age-Specific Association of Co-Morbidity With Home-Time After Acute Stroke

OBJECTIVE

To examine the association of co-morbidity with home-time after acute stroke and whether the association is influenced by age.

METHODS

We conducted a province-wide study using linked administrative databases to identify all admissions for first acute ischemic stroke or intracerebral hemorrhage between 2007 and 2018 in Alberta, Canada. We used ischemic stroke-weighted Charlson Co-morbidity Index of 3 or more to identify those with severe co-morbidity. We used zero-inflated negative binomial models to determine the association of severe co-morbidity with 90-day and 1-year home-time, and logistic models for achieving ≥ 80 out of 90 days of home-time, assessing for effect modification by age and adjusting for sex, stroke type, comprehensive stroke center care, hypertension, atrial fibrillation, year of study, and separately adjusting for estimated stroke severity. We also evaluated individual co-morbidities.

RESULTS

Among 28,672 patients in our final cohort, severe co-morbidity was present in 27.7% and was associated with lower home-time, with a greater number of days lost at younger age (-13 days at age < 60 compared to -7 days at age 80+ years for 90-day home-time; -69 days at age < 60 compared to -51 days at age 80+ years for 1-year home-time). The reduction in probability of achieving ≥ 80 days of home-time was also greater at younger age (-22.7% at age < 60 years compared to -9.0% at age 80+ years). Results were attenuated but remained significant after adjusting for estimated stroke severity and excluding those who died. Myocardial infarction, diabetes, and cancer/metastases had a greater association with lower home-time at younger age, and those with dementia had the greatest reduction in home time.

CONCLUSION

Severe co-morbidity in acute stroke is associated with lower home-time, more strongly at younger age.

Validation of the Passive Surveillance Stroke Severity Score in Three Canadian Provinces

BACKGROUND

Stroke outcomes research requires risk-adjustment for stroke severity, but this measure is often unavailable. The Passive Surveillance Stroke SeVerity (PaSSV) score is an administrative data-based stroke severity measure that was developed in Ontario, Canada. We assessed the geographical and temporal external validity of PaSSV in British Columbia (BC), Nova Scotia (NS) and Ontario, Canada.

METHODS

We used linked administrative data in each province to identify adult patients with ischemic stroke or intracerebral hemorrhage between 2014-2019 and calculated their PaSSV score. We used Cox proportional hazards models to evaluate the association between the PaSSV score and the hazard of death over 30 days and the cause-specific hazard of admission to long-term care over 365 days. We assessed the models' discriminative values using Uno's c-statistic, comparing models with versus without PaSSV.

RESULTS

We included 86,142 patients (n = 18,387 in BC, n = 65,082 in Ontario, n = 2,673 in NS). The mean and median PaSSV were similar across provinces. A higher PaSSV score, representing lower stroke severity, was associated with a lower hazard of death (hazard ratio and 95% confidence intervals 0.70 [0.68, 0.71] in BC, 0.69 [0.68, 0.69] in Ontario, 0.72 [0.68, 0.75] in NS) and admission to long-term care (0.77 [0.76, 0.79] in BC, 0.84 [0.83, 0.85] in Ontario, 0.86 [0.79, 0.93] in NS). Including PaSSV in the multivariable models increased the c-statistics compared to models without this variable.

CONCLUSION

PaSSV has geographical and temporal validity, making it useful for risk-adjustment in stroke outcomes research, including in multi-jurisdiction analyses.

Association of Neighborhood Deprivation With Thrombolysis and Thrombectomy for Acute Stroke in a Health System With Universal Access

BACKGROUND AND OBJECTIVES

The association between socioeconomic status and acute ischemic stroke treatments remain uncertain, particularly in countries with universal health care systems. This study aimed to investigate the association between neighborhood-level material deprivation and the odds of receiving IV thrombolysis or thrombectomy for acute ischemic stroke within a single-payer, government-funded health care system.

METHODS

We conducted a population-based cohort study using linked administrative data from Ontario, Canada. This study involved all community-dwelling adult Ontario residents hospitalized with acute ischemic stroke between 2017 and 2022. Neighborhood-level material deprivation, measured in quintiles from least to most deprived, was our main exposure. We considered the receipt of thrombolysis or thrombectomy as the primary outcome. We used multivariable logistic regression models adjusted for baseline differences to estimate the association between material deprivation and outcomes. We performed a sensitivity analysis by additionally adjusting for hospital type at initial assessment. Furthermore, we tested whether hospital type modified the associations between deprivation and outcomes.

RESULTS

Among 57,704 patients, those in the most materially deprived group (quintile 5) were less likely to be treated with thrombolysis or thrombectomy compared with those in the least deprived group (quintile 1) (16.6% vs 19.6%, adjusted odds ratio [aOR] 0.76, 95% CI 0.63-0.93). The association was consistent when evaluating thrombolysis (13.0% vs 15.3%, aOR 0.78, 95% CI 0.64-0.96) and thrombectomy (6.4 vs 7.8%, aOR 0.73, 95% CI 0.59-0.90) separately. There were no statistically significant differences between the middle 3 quintiles and the least deprived group. These associations persisted after additional adjustment for hospital type, and there was no interaction between material deprivation and hospital type (p interaction >0.1).

DISCUSSION

We observed disparities in the use of thrombolysis or thrombectomy for acute ischemic stroke by socioeconomic status despite access to universal health care. Targeted health care policies, public health messaging, and resource allocation are needed to ensure equitable access to acute stroke treatments for all patients.

Change in Hospital Risk-standardized Stroke Mortality Performance With and Without the Passive Surveillance Stroke Severity Score

BACKGROUND

Adjustment for baseline stroke severity is necessary for accurate assessment of hospital performance. We evaluated whether adjusting for the Passive Surveillance Stroke SeVerity (PaSSV) score, a measure of stroke severity derived using administrative data, changed hospital-specific estimated 30-day risk-standardized mortality rate (RSMR) after stroke.

METHODS

We used linked administrative data to identify adults who were hospitalized with ischemic stroke or intracerebral hemorrhage across 157 hospitals in Ontario, Canada between 2014 and 2019. We fitted a random effects logistic regression model using Markov Chain Monte Carlo methods to estimate hospital-specific 30-day RSMR and 95% credible intervals with adjustment for age, sex, Charlson comorbidity index, and stroke type. In a separate model, we additionally adjusted for stroke severity using PaSSV. Hospitals were defined as low-performing, average-performing, or high-performing depending on whether the RSMR and 95% credible interval were above, overlapping, or below the cohort's crude mortality rate.

RESULTS

We identified 65,082 patients [48.0% were female, the median age (25th,75th percentiles) was 76 years (65,84), and 86.4% had an ischemic stroke]. The crude 30-day all-cause mortality rate was 14.1%. The inclusion of PaSSV in the model reclassified 18.5% (n=29) of the hospitals. Of the 143 hospitals initially classified as average-performing, after adjustment for PaSSV, 20 were reclassified as high-performing and 8 were reclassified as low-performing. Of the 4 hospitals initially classified as low-performing, 1 was reclassified as high-performing. All 10 hospitals initially classified as high-performing remained unchanged.

CONCLUSION

PaSSV may be useful for risk-adjusting mortality when comparing hospital performance. External validation of our findings in other jurisdictions is needed.

External validation of the Passive Surveillance Stroke Severity Indicator

BACKGROUND

The Passive Surveillance Stroke Severity (PaSSV) Indicator was derived to estimate stroke severity from variables in administrative datasets but has not been externally validated.

METHODS

We used linked administrative datasets to identify patients with first hospitalization for acute stroke between 2007-2018 in Alberta, Canada. We used the PaSSV indicator to estimate stroke severity. We used Cox proportional hazard models and evaluated the change in hazard ratios and model discrimination for 30-day and 1-year case fatality with and without PaSSV. Similar comparisons were made for 90-day home time thresholds using logistic regression. We also linked with a clinical registry to obtain National Institutes of Health Stroke Scale (NIHSS) and compared estimates from models without stroke severity, with PaSSV, and with NIHSS.

RESULTS

There were 28,672 patients with acute stroke in the full sample. In comparison to no stroke severity, addition of PaSSV to the 30-day case fatality models resulted in improvement in model discrimination (C-statistic 0.72 [95%CI 0.71-0.73] to 0.80 [0.79-0.80]). After adjustment for PaSSV, admission to a comprehensive stroke center was associated with lower 30-day case fatality (adjusted hazard ratio changed from 1.03 [0.96-1.10] to 0.72 [0.67-0.77]). In the registry sample (N = 1328), model discrimination for 30-day case fatality improved with the inclusion of stroke severity. Results were similar for 1-year case fatality and home time outcomes.

CONCLUSION

Addition of PaSSV improved model discrimination for case fatality and home time outcomes. The validity of PASSV in two Canadian provinces suggests that it is a useful tool for baseline risk adjustment in acute stroke.

The Allure of Big Data to Improve Stroke Outcomes: Review of Current Literature

PURPOSE OF REVIEW

To critically appraise literature on recent advances and methods using "big data" to evaluate stroke outcomes and associated factors.

RECENT FINDINGS

Recent big data studies provided new evidence on the incidence of stroke outcomes, and important emerging predictors of these outcomes. Main highlights included the identification of COVID-19 infection and exposure to a low-dose particulate matter as emerging predictors of mortality post-stroke. Demographic (age, sex) and geographical (rural vs. urban) disparities in outcomes were also identified. There was a surge in methodological (e.g., machine learning and validation) studies aimed at maximizing the efficiency of big data for improving the prediction of stroke outcomes. However, considerable delays remain between data generation and publication. Big data are driving rapid innovations in research of stroke outcomes, generating novel evidence for bridging practice gaps. Opportunity exists to harness big data to drive real-time improvements in stroke outcomes.

Predictive Model and Mortality Risk Score during Admission for Ischaemic Stroke with Conservative Treatment

Background: Stroke is the second cause of mortality worldwide and the first in women. The aim of this study is to develop a predictive model to estimate the risk of mortality in the admission of patients who have not received reperfusion treatment. Methods: A retrospective cohort study was conducted of a clinical–administrative database, reflecting all cases of non-reperfused ischaemic stroke admitted to Spanish hospitals during the period 2008–2012. A predictive model based on logistic regression was developed on a training cohort and later validated by the “hold-out” method. Complementary machine learning techniques were also explored. Results: The resulting model had the following nine variables, all readily obtainable during initial care. Age (OR 1.069), female sex (OR 1.202), readmission (OR 2.008), hypertension (OR 0.726), diabetes (OR 1.105), atrial fibrillation (OR 1.537), dyslipidaemia (0.638), heart failure (OR 1.518) and neurological symptoms suggestive of posterior fossa involvement (OR 2.639). The predictability was moderate (AUC 0.742, 95% CI: 0.737–0.747), with good visual calibration; Pearson’s chi-square test revealed non-significant calibration. An easily consulted risk score was prepared. Conclusions: It is possible to create a predictive model of mortality for patients with ischaemic stroke from which important advances can be made towards optimising the quality and efficiency of care. The model results are available within a few minutes of admission and would provide a valuable complementary resource for the neurologist.

Association of Neighborhood-Level Material Deprivation With Health Care Costs and Outcome After Stroke

BACKGROUND AND OBJECTIVES

To determine the association between material deprivation and direct health care costs and clinical outcomes following stroke in the context of a publicly funded universal health care system.

METHODS

In this population-based cohort study of patients with ischemic and hemorrhagic stroke admitted to the hospital between 2008 and 2017 in Ontario, Canada, we used linked administrative data to identify the cohort, predictor variables, and outcomes. The exposure was a 5-level neighborhood material deprivation index. The primary outcome was direct health care costs incurred by the public payer in the first year. Secondary outcomes were death and admission to long-term care.

RESULTS

Among 90,289 patients with stroke, the mean (SD) per-person costs increased with increasing material deprivation, from $50,602 ($55,582) in the least deprived quintile to $56,292 ($59,721) in the most deprived quintile (unadjusted relative cost ratio and 95% confidence interval 1.11 [1.08, 1.13] and adjusted relative cost ratio 1.07 [1.05, 1.10] for least compared to most deprived quintile). People in the most deprived quintile had higher mortality within 1 year compared to the least deprived quintile (adjusted hazard ratio [HR] 1.07 [1.03, 1.12]) as well as within 3 years (adjusted HR 1.09 [1.05, 1.13]). Admission to long-term care increased incrementally with material deprivation and those in the most deprived quintile had an adjusted HR of 1.33 (1.24, 1.43) compared to those in the least deprived quintile.

DISCUSSION

Material deprivation is a risk factor for increased costs and poor outcomes after stroke. Interventions targeting health inequities due to social determinants of health are needed.

CLASSIFICATION OF EVIDENCE

This study provides Class II evidence that the neighborhood-level material deprivation predicts direct health care costs.

Age-Specific and Sex-Specific Trends in Life-Sustaining Care After Acute Stroke

Background

Temporal trends in life‐sustaining care after acute stroke are not well characterized. We sought to determine contemporary trends by age and sex in the use of life‐sustaining care after acute ischemic stroke and intracerebral hemorrhage in a large, population‐based cohort.

Methods and Results

We used linked administrative data to identify all hospitalizations for acute ischemic stroke or intracerebral hemorrhage in the province of Ontario, Canada, from 2003 to 2017. We calculated yearly proportions of intensive care unit admission, mechanical ventilation, percutaneous feeding tube placement, craniotomy/craniectomy, and tracheostomy. We used logistic regression models to evaluate the association of age and sex with life‐sustaining care and determined whether trends persisted after adjustment for baseline factors and estimated stroke severity. There were 137 358 people with acute ischemic stroke or intracerebral hemorrhage hospitalized during the study period. Between 2003 and 2017, there was an increase in the proportion receiving care in the intensive care unit (12.4% to 17.7%) and mechanical ventilation (4.4% to 6.6%). There was a small increase in craniotomy/craniectomy, a decrease in percutaneous feeding tube use, and no change in tracheostomy. Trends were generally consistent across stroke types and persisted after adjustment for comorbid conditions, stroke‐center type, and estimated stroke severity. After adjustment, women and those aged ≥80 years had lower odds of all life‐sustaining care, although the disparities in intensive care unit admission narrowed over time.

Conclusions

Use of life‐sustaining care after acute stroke increased between 2003 and 2017. Women and those at older ages had lower odds of intensive care, although the differences narrowed over time. Further research is needed to determine the reasons for these findings.

Comparing regression modeling strategies for predicting hometime

Background

Hometime, the total number of days a person is living in the community (not in a healthcare institution) in a defined period of time after a hospitalization, is a patient-centred outcome metric increasingly used in healthcare research. Hometime exhibits several properties which make its statistical analysis difficult: it has a highly non-normal distribution, excess zeros, and is bounded by both a lower and upper limit. The optimal methodology for the analysis of hometime is currently unknown.

Methods

Using administrative data we identified adult patients diagnosed with stroke between April 1, 2010 and December 31, 2017 in Ontario, Canada. 90-day hometime and clinically relevant covariates were determined through administrative data linkage. Fifteen different statistical and machine learning models were fit to the data using a derivation sample. The models’ predictive accuracy and bias were assessed using an independent validation sample.

Results

Seventy-five thousand four hundred seventy-five patients were identified (divided into a derivation set of 49,402 and a test set of 26,073). In general, the machine learning models had lower root mean square error and mean absolute error than the statistical models. However, some statistical models resulted in lower (or equal) bias than the machine learning models. Most of the machine learning models constrained predicted values between the minimum and maximum observable hometime values but this was not the case for the statistical models. The machine learning models also allowed for the display of complex non-linear interactions between covariates and hometime. No model captured the non-normal bucket shaped hometime distribution.

Conclusions

Overall, no model clearly outperformed the others. However, it was evident that machine learning methods performed better than traditional statistical methods. Among the machine learning methods, generalized boosting machines using the Poisson distribution as well as random forests regression were the best performing. No model was able to capture the bucket shaped hometime distribution and future research on factors which are associated with extreme values of hometime that are not available in administrative data is warranted.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12874-021-01331-9.

Sex differences in direct healthcare costs following stroke: a population-based cohort study

BACKGROUND

The economic burden of stroke on the healthcare system has been previously described, but sex differences in healthcare costs have not been well characterized. We described the direct person-level healthcare cost in men and women as well as the various health settings in which costs were incurred following stroke.

METHODS

In this population-based cohort study of patients admitted to hospital with stroke between 2008 and 2017 in Ontario, Canada, we used linked administrative data to calculate direct person-level costs in Canadian dollars in the one-year following stroke. We used a generalized linear model with a gamma distribution and a log link function to compare costs in women and men with and without adjustment for baseline clinical differences. We also assessed for an interaction between age and sex using restricted cubic splines to model the association of age with costs.

RESULTS

We identified 101,252 patients (49% were women, median age [Q1-Q3] was 76 years [65-84]). Unadjusted costs following stroke were higher in women compared to men (mean ± standard deviation cost was $54,012 ± 54,766 for women versus $52,829 ± 59,955 for men, and median cost was $36,703 [$16,496-$72,227] for women versus $32,903 [$15,485-$66,007] for men). However, after adjustment, women had 3% lower costs compared to men (relative cost ratio and 95% confidence interval 0.97 [0.96,0.98]). The lower cost in women compared to men was most prominent among people aged over 85 years (p for interaction = 0.03). Women incurred lower costs than men in outpatient care and rehabilitation, but higher costs in complex continuing care, long-term care, and home care.

CONCLUSIONS

Patterns of resource utilization and direct medical costs were different between men and women after stroke. Our findings inform public payers of the drivers of costs following stroke and suggest the need for sex-based cost-effectiveness evaluation of stroke interventions with consideration of costs in all care settings.

Access to Mechanical Thrombectomy for Ischemic Stroke in the United States

[Figure: see text].

Using random forests to model 90-day hometime in people with stroke

Background

Ninety-day hometime, the number of days a patient is living in the community in the first 90 after stroke, exhibits a non-normal bucket-shaped distribution, with lower and upper constraints making its analysis difficult. In this proof-of-concept study we evaluated the performance of random forests regression in the analysis of hometime.

Methods

Using administrative data we identified stroke hospitalizations between 2010 and 2017 in Ontario, Canada. We used random forests regression to predict 90-day hometime using 15 covariates. Model accuracy was determined using the r-squared statistic. Variable importance in prediction and the marginal effects of each covariate were explored.

Results

We identified 75,745 eligible patients. Median 90-day hometime was 59 days (Q1: 2, Q3: 83). Random forests predicted hometime with reasonable accuracy (adjusted r-squared 0.3462); no implausible values were predicted but extreme values were predicted with low accuracy. Frailty, stroke severity, and age exhibited inverse non-linear relationships with hometime and patients arriving by ambulance had less hometime than those who did not.

Conclusions

Random forests may be a useful method for analyzing 90-day hometime and capturing the complex non-linear relationships which exist between predictors and hometime. Future work should compare random forests to other models and focus on improving the accuracy of predictions of extreme values of hometime.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12874-021-01289-8.

Potential diabetes overtreatment and risk of adverse events among older adults in Ontario: a population-based study

AIMS/HYPOTHESIS

More than 25% of older adults (age ≥75 years) have diabetes and may be at risk of adverse events related to treatment. The aim of this study was to assess the prevalence of intensive glycaemic control in this group, potential overtreatment among older adults and the impact of overtreatment on the risk of serious events.

METHODS

We conducted a retrospective, population-based cohort study of community-dwelling older adults in Ontario using administrative data. Participants were ≥75 years of age with diagnosed diabetes treated with at least one anti-hyperglycaemic agent between 2014 and 2015. Individuals were categorised as having intensive or conservative glycaemic control (HbA_1c <53 mmol/mol [<7%] or 54-69 mmol/mol [7.1-8.5%], respectively), and as undergoing treatment with high-risk (i.e. insulin, sulfonylureas) or low-risk (other) agents. We measured the composite risk of emergency department visits, hospitalisations, or death within 30 days of reaching intensive glycaemic control with high-risk agents.

RESULTS

Among 108,620 older adults with diagnosed diabetes in Ontario, the mean (± SD) age was 80.6 (±4.5) years, 49.7% were female, and mean (± SD) diabetes duration was 13.7 (±6.3) years. Overall, 61% of individuals were treated to intensive glycaemic control and 21.6% were treated to intensive control using high-risk agents. Using inverse probability treatment weighting with propensity scores, intensive control with high-risk agents was associated with nearly 50% increased risk of the composite outcome compared with conservative glycaemic control with low-risk agents (RR 1.49, 95% CI 1.08, 2.05).

CONCLUSIONS/INTERPRETATION

Our findings underscore the need to re-evaluate glycaemic targets in older adults and to reconsider the use of anti-hyperglycaemic medications that may lead to hypoglycaemia, especially in setting of intensive glycaemic control.