FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2368708758> ?p ?o ?g. }

• W2368708758 endingPage "1952" @default.
• W2368708758 startingPage "1946" @default.
• W2368708758 abstract "HomeStrokeVol. 47, No. 7Use and Utility of Administrative Health Data for Stroke Research and Surveillance Free AccessResearch ArticlePDF/EPUBAboutView PDFView EPUBSections ToolsAdd to favoritesDownload citationsTrack citationsPermissionsDownload Articles + Supplements ShareShare onFacebookTwitterLinked InMendeleyReddit Jump toSupplemental MaterialFree AccessResearch ArticlePDF/EPUBUse and Utility of Administrative Health Data for Stroke Research and Surveillance Amy Y.X. Yu, MD, Jessalyn K. Holodinsky, MSc, Charlotte Zerna, MD, Lawrence W. Svenson, PhD, Nathalie Jetté, MD, Hude Quan, PhD and Michael D. Hill, MD Amy Y.X. YuAmy Y.X. Yu From the Department of Clinical Neurosciences (A.Y.X.Y., C.Z., N.J., M.D.H.), Department of Community Health Sciences (A.Y.X.Y., J.K.H., L.W.S., N.J., H.Q., M.D.H.), O’Brien Institute for Public Health (N.J., H.Q., M.D.H.), and Hotchkiss Brain Institute (N.J., M.D.H.), University of Calgary, Calgary, Alberta, Canada; Surveillance and Assessment Branch, Alberta Health, Edmonton, Alberta, Canada (L.W.S.); and School of Public Health, University of Alberta, Edmonton, Alberta, Canada (L.W.S.). Search for more papers by this author , Jessalyn K. HolodinskyJessalyn K. Holodinsky From the Department of Clinical Neurosciences (A.Y.X.Y., C.Z., N.J., M.D.H.), Department of Community Health Sciences (A.Y.X.Y., J.K.H., L.W.S., N.J., H.Q., M.D.H.), O’Brien Institute for Public Health (N.J., H.Q., M.D.H.), and Hotchkiss Brain Institute (N.J., M.D.H.), University of Calgary, Calgary, Alberta, Canada; Surveillance and Assessment Branch, Alberta Health, Edmonton, Alberta, Canada (L.W.S.); and School of Public Health, University of Alberta, Edmonton, Alberta, Canada (L.W.S.). Search for more papers by this author , Charlotte ZernaCharlotte Zerna From the Department of Clinical Neurosciences (A.Y.X.Y., C.Z., N.J., M.D.H.), Department of Community Health Sciences (A.Y.X.Y., J.K.H., L.W.S., N.J., H.Q., M.D.H.), O’Brien Institute for Public Health (N.J., H.Q., M.D.H.), and Hotchkiss Brain Institute (N.J., M.D.H.), University of Calgary, Calgary, Alberta, Canada; Surveillance and Assessment Branch, Alberta Health, Edmonton, Alberta, Canada (L.W.S.); and School of Public Health, University of Alberta, Edmonton, Alberta, Canada (L.W.S.). Search for more papers by this author , Lawrence W. SvensonLawrence W. Svenson From the Department of Clinical Neurosciences (A.Y.X.Y., C.Z., N.J., M.D.H.), Department of Community Health Sciences (A.Y.X.Y., J.K.H., L.W.S., N.J., H.Q., M.D.H.), O’Brien Institute for Public Health (N.J., H.Q., M.D.H.), and Hotchkiss Brain Institute (N.J., M.D.H.), University of Calgary, Calgary, Alberta, Canada; Surveillance and Assessment Branch, Alberta Health, Edmonton, Alberta, Canada (L.W.S.); and School of Public Health, University of Alberta, Edmonton, Alberta, Canada (L.W.S.). Search for more papers by this author , Nathalie JettéNathalie Jetté From the Department of Clinical Neurosciences (A.Y.X.Y., C.Z., N.J., M.D.H.), Department of Community Health Sciences (A.Y.X.Y., J.K.H., L.W.S., N.J., H.Q., M.D.H.), O’Brien Institute for Public Health (N.J., H.Q., M.D.H.), and Hotchkiss Brain Institute (N.J., M.D.H.), University of Calgary, Calgary, Alberta, Canada; Surveillance and Assessment Branch, Alberta Health, Edmonton, Alberta, Canada (L.W.S.); and School of Public Health, University of Alberta, Edmonton, Alberta, Canada (L.W.S.). Search for more papers by this author , Hude QuanHude Quan From the Department of Clinical Neurosciences (A.Y.X.Y., C.Z., N.J., M.D.H.), Department of Community Health Sciences (A.Y.X.Y., J.K.H., L.W.S., N.J., H.Q., M.D.H.), O’Brien Institute for Public Health (N.J., H.Q., M.D.H.), and Hotchkiss Brain Institute (N.J., M.D.H.), University of Calgary, Calgary, Alberta, Canada; Surveillance and Assessment Branch, Alberta Health, Edmonton, Alberta, Canada (L.W.S.); and School of Public Health, University of Alberta, Edmonton, Alberta, Canada (L.W.S.). Search for more papers by this author and Michael D. HillMichael D. Hill From the Department of Clinical Neurosciences (A.Y.X.Y., C.Z., N.J., M.D.H.), Department of Community Health Sciences (A.Y.X.Y., J.K.H., L.W.S., N.J., H.Q., M.D.H.), O’Brien Institute for Public Health (N.J., H.Q., M.D.H.), and Hotchkiss Brain Institute (N.J., M.D.H.), University of Calgary, Calgary, Alberta, Canada; Surveillance and Assessment Branch, Alberta Health, Edmonton, Alberta, Canada (L.W.S.); and School of Public Health, University of Alberta, Edmonton, Alberta, Canada (L.W.S.). Search for more papers by this author Originally published12 May 2016https://doi.org/10.1161/STROKEAHA.116.012390Stroke. 2016;47:1946–1952Other version(s) of this articleYou are viewing the most recent version of this article. Previous versions: January 1, 2016: Previous Version 1 IntroductionDespite declining age-standardized stroke incidence in high-income countries, stroke incidence is rising in low- and middle-income countries.1 Globally, the absolute burden of stroke was high in 2010 with 16.9 million first-ever strokes, 5.9 million stroke-related deaths, and 102 million disability-adjusted life-years lost.1 These numbers are projected to increase. Surveillance provides an understanding of stroke frequency, burden, distribution, and determinants. These data are essential for monitoring trends over time, guiding judicious resource allocation, and for the design, implementation, and evaluation of interventions aimed at stroke prevention, treatment, and rehabilitation.2 Collecting data specifically for research purposes can be costly and time-consuming, limiting the sample size, period of follow-up, and geographical distribution of subjects. Surveillance requires continuous data collection in large geographic areas over years; therefore, attention has been paid to secondary use data.Health services utilization data, or administrative health data, provide a wealth of information for health services researchers and for stroke surveillance. However, the information collected and ascertainment methods are heterogeneous between countries and even between jurisdictions within a country, making the data vulnerable to selection and measurement bias. Comparing international data is also challenging.3 In this review, we discuss the strengths and weaknesses of administrative health data for stroke surveillance.What Are Administrative Health Data?Administrative health data are routinely generated through interactions with healthcare systems. They are collected for payment, monitoring, planning, priority setting, and evaluation of health services provision. Sources include, but are not limited to, hospitalizations, emergency department and ambulatory care visits, and physician billings. Unlike prospective clinical research data collection, administrative health data are accumulated in a distributed manner over a prolonged period of time. As a result, these data capture a large number of individuals and a wide range of demographic information, including race/ethnicities, geographical areas (eg, rural versus urban), and institutions (eg, community versus long-term care). Furthermore, they reflect real-world practice at a hospital, regional, or population level.As these data are not generated for research purposes, they are commonly termed secondary use research data. Clinicians and researchers exhibit skepticism on the validity of the data. However, when these data are applied to an appropriate question with validated case definitions, high-quality and reliable conclusions can be inferred. The growing use of administrative health data for research purposes has led to the publication of international reporting standards and standards specific to neurological diseases.4–6Content and Coding of Administrative Health DataIn this article, we focus on data collected from acute-care hospitals, which typically contain patient identifiers, demographics, basic timeline information, including admission and discharge dates, procedures, and 1 primary and multiple secondary diagnoses. Diagnoses are coded using the International Classification of Diseases (ICD).7 The history of classification of disease begins in 1893 when Jacques Bertillon presented the first International List of Causes of Death. This classification was revised every 10 years until 1948, when the World Health Organization established the sixth version of the ICD. In the current 10th iteration (ICD-10), disease groupings are coded using an alphanumeric system and separated into chapters.7 Strokes are coded under chapter IX: Diseases of the Circulatory System (I60.X–I64.X) and transient ischemic attacks are listed separately under chapter VI: Diseases of the Nervous System (G45.X, except G45.4; Table I in the online-only Data Supplement). Other stroke syndromes, such as neonatal cerebral ischemia (P91.0) or transient retinal artery occlusion (H34.0x) are classified separately. Until 1975, the ICD was revised every decade by the World Health Organization Steering Committee. Because of the lengthy process of consulting international experts, more time has been allocated to the most recent revisions. There are substantial delays between the evolution of disease knowledge and clinical practice and their reflection in ICD revisions. For instance, ICD-11 has been under development since 2014 and is anticipated to be finalized in 2018. The ICD-11 will better reflect the modern understanding of stroke, bringing together stroke diagnoses in a single chapter. Furthermore, high costs associated with updating software and training personnel lead to delays in the implementation of new coding systems.3 Some Canadian provinces implemented ICD-10-CA in 2012, the United States switched to ICD-10 in 2015, and other countries are still using the ICD-9.8 Outpatient billing is largely based on ICD-9 in North America.Although ICD-10 codes form the backbone structure of disease classification worldwide, many countries have their own adaptations of the codes, beyond the 3-digit level, creating barriers in direct comparisons between jurisdictions, or merging of the data.3 For example, the ICD-10-CA/Canadian Classification of Health Interventions is the Canadian adaptation for coding diagnoses and procedures, whereas the ICD-9-CM and ICD-10-CM contain the American clinical modifications, which includes procedural codes, adapted by the US National Center for Health Statistics. The ICD-11 is expected to address some of the challenges of coding heterogeneity, as it will be implemented internationally as a single version, translated into multiple languages, digitalized for ongoing modifications, and will contain definitions to allow for standardized application.Validity, Completeness, and Limitations of Administrative Health DataTo understand validity and completeness, the coding process must first be reviewed. The Figure illustrates the flow of events between the onset of a patient’s stroke symptoms and the point where information is collected into a research data repository. Sources of error and bias exist at each level.Download figureDownload PowerPointFigure. Flowchart of information pathway from clinical symptoms to a research administrative health database. *Clinical encounter include clinical evaluation, diagnostic tests, treatments in the emergency department (ED), or outpatient clinic (generalist or specialist office, including stroke prevention clinic). †Documentations include paper chart, electronic medical records, diagnostic imaging and laboratory test reports, consultation notes, and discharge summaries. ‡Administrative databases may be at regional or national-level government source or private sources, such as insurance companies. §Dependent on jurisdiction-specific physician remuneration process.PopulationFirst, because data are generated by routine healthcare utilization, patients with stroke who do not present to medical attention or who are misdiagnosed will not be captured. This causes a selection bias toward stroke patients having more severe and nonfatal symptoms and those who have access to medical coverage. Second, the coverage of administrative health data depends on the setting of the healthcare encounter. Hospital inpatient admissions usually have consistent and mandatory reporting requirements, but there is more variability in the availability of ambulatory care data. For example, the Canadian Institute of Health Information (CIHI) Discharge Abstract Database contains discharge diagnosis codes from all Canadian acute-care hospitals. In contrast, the National Ambulatory Care Reporting System receives data for emergency department visits, selected hospital-based ambulatory clinics, and day surgeries on a voluntary basis. Most of the National Ambulatory Care Reporting System data come from 2 of the 10 Canadian provinces (Ontario and Alberta).9 Third, the population covered by administrative health databases depends on the healthcare models of a given country or jurisdiction within a country. The population covered may be complete and relatively stable or fragmented and variable over time. The Canadian healthcare system is organized, administered, and paid by each province, and the government mandates high-level national standards, including reporting of all Canadian hospitalizations to CIHI.10 This single repository allows for accurate denominators, ie, all hospitalizations in Canada. The United Kingdom has a similar healthcare system model provided through the National Health Service.11 In contrast, countries under a multipayer healthcare system, such as the United States, have several databases for hospital admissions. The Medicare Claims Database is often used in US administrative health data studies. It contains all acute-care hospitalizations for eligible individuals (those aged ≥65, individuals aged <65 with select disabilities, and individuals with end-stage renal disease regardless of age). Utilizing this database would give a complete sample of hospitalizations for elderly individuals, but would not be generalizable to those aged <65 years. Other sources of US administrative health data include the Medicaid Claims database (all acute-care hospitalizations for low-income families, children, pregnant women, and individuals with disabilities), the Veterans Health Administration database (all acute-care hospitalizations for active military veterans), various Health Management Organization databases (hospitalization data for members), and the Healthcare Cost and Utilization Project databases from which the National Inpatient Sample is drawn. The Healthcare Cost and Utilization Project databases contain information about all-payer hospitalizations in 47 of the 50 states; some states also contribute data on emergency department and surgical visits. From the Healthcare Cost and Utilization Project databases, the National Inpatient Sample is generated annually using a sample of 8 million hospital discharges from 1000 hospitals to approximate a 20% stratified sample of hospital discharges.12 Another example of a multipayer system is Australia, which has both a public and a private healthcare system.13 As the population captured can differ so widely, it is important for investigators to (1) know what population the database captures, (2) be specific about this population when reporting study results, and (3) be cautious with generalization of the results.DocumentationCoding is the process of translating the medical record information into codes. Therefore, data quality depends on 2 major factors: (1) documentation: the quality of data provided in the medical record and (2) coding: the quality of training/expertise of the coder and the applicability of the codes to the medical condition(s) reported in the medical record. The ICD terminology is not specific to health conditions and individual coding guidelines. The coders, who can be coding specialists, administrative clerks, or in certain settings physicians, extract information from article and electronic records, diagnostic imaging and laboratory reports, consultation notes, and discharge summaries to obtain the primary and secondary diagnoses, including comorbidities and complications. Completeness and legibility of the documentation are critical for fidelity of translation of the medical record to the appropriate treatment and diagnostic codes. The Table illustrates how physician documentation influences coding of the most responsible diagnosis. We emphasize that currently, there is no uniform definition of the most responsible diagnosis. It can vary between the principal reason for admission (eg, United States and Australia) or the condition with the highest resource use (eg, Canada).3,8 For example, a patient is admitted for myocardial infarction and has a stroke post admission, leading to a prolonged hospitalization. In Canada, stroke becomes the main source of resource use and in the United States, myocardial infarction remains the principal reason for admission. Furthermore, the Table illustrates how timing of diagnosis is coded. Although most countries do not mandate this level of coding precision, timing of diagnosis is required in the United States (called present on admission), Canada (diagnosis type), and Australia (diagnosis onset type).8,14 In Canada, a type 1 diagnosis is a condition present on admission, type 2 is a condition that arose after admission, and type 3 are secondary diagnoses. In the scenarios presented, physician A documents that atrial fibrillation is the cause of stroke and therefore, presumed to be present on admission (type 1), whereas physician B fails to document the relationship between atrial fibrillation and stroke and it is coded as a complication (type 2). This distinction is relevant for comorbidity adjustment. Type 2 diagnoses are often excluded from comorbidity scores, such as the Charlson–Deyo and Elixhauser indices, as they are interpreted as complications, rather than as comorbidities.15 Furthermore, in the context of stroke, baseline stroke severity is a stronger overall predictor of outcome and often remains an important residual confounder as stroke severity is not coded. In the United States, initiatives are underway to include the National Institutes of Health Stroke Scale in the ICD-10-CM (Schwamm LH, personal communications).Table. Example of Coding From Physician DocumentationDocumentationAdministrative Codes*Interpretation of CodesPhysician A85 male with diabetes mellitus, ischemic cardioembolic stroke caused by atrial fibrillation with hemorrhagic transformationI63.4 (MRDx)I48.90 (DxType1)E11.52 (DxType3)Cerebral infarction caused by embolism of cerebral arteriesAtrial fibrillation, unspecifiedType 2 diabetes mellitus with certain circulatory complicationsPhysician B85 male, type 2 diabetes mellitus, stroke and hemorrhage on CT, atrial fibrillationI61.9 (MRDx)I48.90 (DxType2)E11.52 (DxType3)Intracerebral hemorrhage, unspecifiedAtrial fibrillation, unspecifiedType 2 diabetes mellitus with certain circulatory complicationsScenario: 85-year-old man with diabetes mellitus type 2 admitted with an ischemic stroke with hemorrhagic transformation and is found to have atrial fibrillation on the second day of admission. CT indicates computed tomography; DxType, diagnosis type; and MRDx, most responsible diagnosis.*These represent real codes generated by a coding specialist in a tertiary-care hospital based on hypothetical patient information.Coding is limited by the accuracy of the physician’s diagnosis. If the treating physician did not recognize that the intracerebral hemorrhage was because of a hemorrhagic transformation of an ischemic stroke, the coding will not represent the truth because the chart did not. Differences in stroke coding have been shown between urban and rural hospitals. More common use of unspecified stroke type codes in rural than in urban facilities has been reported, suggesting challenges in distinction of ischemic versus hemorrhagic strokes in rural areas.16 When a patient returns to a rural center after treatment at an urban one, the more specific code follows them back, suggesting that differences in access to advanced neurovascular imaging and clinical expertise contribute to diagnostic ascertainment and documentation, and thus coding quality. In ambulatory settings, charting is sparser and clinical contact briefer, so that the diagnostic and coding accuracy may differ based on outpatient, emergency department, or inpatient encounters. A diagnostic code of transient ischemic attack may not have the same sensitivity and specificity for patients discharged form the emergency department compared with patients discharged after a hospital stay. In addition, changes in clinical practice may influence documentation and coding. Stroke and transient ischemic attack hospitalization rates using discharge abstract database have been shown to be decreasing.17–19 Although this finding is consistent with the advances in stroke knowledge and care, it is important to recognize that the decreased admissions for transient ischemic attacks may be largely explained by practice variations: more patients with mild or transient symptoms are being evaluated and discharged from the emergency departments without admission. Physicians must recognize that they are active participants in administrative health data generation, and coding is limited by documentation quality.CodingThe question of accuracy of administrative health data is best addressed in 2 ways: how well does the ICD code reflect (1) the clinical encounter and (2) the underlying medical condition? There is a jurisdiction-specific recipe book for coding and the quality of training and expertise of the coder does matter. For example, CIHI in Canada, and the Centers for Medicare and Medicaid Services and the National Center for Health Statistics in the United States, publish standards for coding and reporting.20,21 CIHI regularly publishes reports on data quality. In a recent study, 19 coders from CIHI reabstracted charts from 85 hospitals across Canada and found 96% accuracy for interventions, 89% for significant diagnoses, and 76% for most responsible diagnosis.22 Another study evaluated the intercoder agreement for strokes using the ICD-9-CM and found generally high agreement with an overall κ value of 0.77 for primary diagnosis of stroke.23 This study was conducted in Italy where coding is done by physicians. Most studies validating ICD codes select patients who have records with the diagnostic code of interest.24–26 The charts are then reviewed for accuracy. Fundamentally, this approach is a comparison of the agreement between coders and reabstractors of health records and has shown moderate to high sensitivity and specificity. In the Brain Attack Surveillance in Corpus Christi project, the authors compared active and passive surveillance of cerebrovascular events in Nueces County, Texas, during a full calendar year.27 The active surveillance arm consisted of screening, in real-time, admission, and emergency department visit logs in the 6 hospitals of the region, using 8 cerebrovascular screening diagnostic terms as well as routine visits to medical offices, nursing homes, radiology centers, and the office of the medical examiner. Passive surveillance used ICD-9 discharge codes. Although passive surveillance missed ≈1 in 4 cases, mostly outpatient strokes or transient ischemic attacks, active surveillance also missed 7.1% of the cases, mostly because of missing screening terms in the hospital admission. Active surveillance had a positive predictive value of 12.2% compared with the 72.8% for passive surveillance. Interestingly, the active surveillance arm only identified 17 (2.3%) out-of-hospital cerebrovascular cases, indicating that only a relatively low proportion of patients with stroke do not present to an acute-care hospital.Finally, because administrative health data are collected for accounting purposes, financial incentives may influence coding. Code creep, a term that has been coined since the 1980s refers to the situations when codes leading to higher reimbursement will be preferred.28 An American study showed that 61% of the coding discrepancies financially favored the hospitals.29 This led the government to require physicians to attest to the accuracy of diagnostic codes, a decision later eliminated in 1995. A more recent American study estimated an annual 2.2% increase on physician visit payments attributable to code creep.30 A similar phenomenon has been described in Canada.31Data LinkageData linkage between databases may be used to create a research database. However, linkage algorithms may result in incomplete or erroneous matching. In Canada, the Discharge Abstract Database has the advantage of identifying >99% of Canadian residents using a unique provincial or territorial personal healthcare number. This allows for data from multiple databases within a province or territory to be linked reliably.32 Most countries, including the United States, Australia, and many Asian countries do not have access to unique indentifiers and data must be linked using probabilistic methods.32 Finally, delays inherent in the compilation of large data sets may limit the real-time use of administrative health data.Administrative Health Data for Stroke SurveillanceDespite the limitations described, administrative health data are a valuable tool for observational epidemiology, where identification of patterns can generate hypothesis that need to be tested by further studies. The American Heart Association Stroke Report estimates the national burden of stroke by compiling results from (1) the National Center for Health Statistics, (2) national and regional cross-sectional surveys, such as the Behavioral Risk Factor Surveillance System, and (3) longitudinal studies, such as the Greater Cincinnati/Northern Kentucky Stroke Study and the Framingham Heart Study.33–35 In Canada, surveillance systems like the Canadian Chronic Diseases Surveillance System and Heart and Stroke Foundation of Canada produce stroke reports using CIHI administrative health data and cross-sectional survey results from the Canadian Community Health Survey.36,37 On a global scale, administrative health data are the only sustainable tool for monitoring disease burden.Administrative health data allow for the estimation of distributions and prevalence of determinants of stroke. A Canadian study identified that South Asian patients have a higher prevalence of diabetes mellitus, dyslipidemia and hypertension, whereas white patients have a higher prevalence of atrial fibrillation, alcohol, and tobacco use.38 Administrative health data may reveal inequities in health or healthcare resources and demonstrate systematic differences of health across socially and economically defined subgroups. Patients of lower socioeconomic status are reported to be less likely to present in a timely manner to the hospital, receive thrombolysis, or specialist care by a neurologist when having acute ischemic stroke.39Because administrative health data contain large numbers of patients over long periods of time, they are useful to study disease associations with rare risk factors. Although congenital heart disease is relatively uncommon, assessing province-wide administrative health data for over a decade showed that patients with adult congenital heart disease were ≈9 to 12 times more likely to have an ischemic stroke compared with the general population and 5 to 6 times more likely to have hemorrhagic strokes.40 Administrative health data are adjuncts to other study designs. They allow for assessment of the impact of research knowledge translation on a large population scale. For example, the implementation of integrated systems of stroke care was associated with a population-wide reduction in mortality after stroke in Canada.41 Another Australian observational cohort study used administrative health data to evaluate the effect of a stroke unit implementation. They found reduced inpatient mortality, increased discharge rates to home, and no increased length of stay.42 However, like other forms of observational epidemiological data, administrative health data do not allow conclusions on causality.Future Initiatives and ConclusionsWith the upcoming ICD-11, we can look forward to advances in the standardization of coding at an international level and improvements in the accuracy of capturing diagnosis timing and severity of disease to allow for better identification of comorbidities.14 For the first time in history, the ICD-11 Steering Committee welcomes all experts and stakeholder to participate in the ICD-11 revision through their online platform (http://apps.who.int/classifications/icd11/browse/l-m/en). In summary, administrative health data are a valuable source of information for the surveillance of stroke and stroke risk factors, when used appropriately. Recognizing their limitations and strengths is important, and authors should be specific when designing and reporting studies using administrative health data.AcknowledgmentsWe acknowledge the special contribution of Mrs Chris Makar, Coding Coordinator, Data Collection, Foothills Medical Centre.Soures of FundingDr Jetté is the holder of a Canada Research Chair Tier 2 in Neurological Health Services Research. Dr Yu holds a Clinician Fellowship Award from Alberta Innovates Health Solutions.DisclosuresDr Hill reports grants from Covidien (Medtronic), grants from Alberta Innovates Health Solutions, grants from Heart & Stroke Foundation, grants from Hotchkiss Brain Institute, grants from the Canadian Stroke Prevention Network (Institute of Circulatory and Respiratory Health–Canadian Institutes for Health Research), grants from Calgary Stroke Program, Department of Clinical Neurosciences, University of Calgary, nonfinancial support from Alberta" @default.
• W2368708758 created "2016-06-24" @default.
• W2368708758 creator A5025499620 @default.
• W2368708758 creator A5031383231 @default.
• W2368708758 creator A5056711512 @default.
• W2368708758 creator A5068320077 @default.
• W2368708758 creator A5080252318 @default.
• W2368708758 creator A5084100416 @default.
• W2368708758 creator A5090161362 @default.
• W2368708758 date "2016-07-01" @default.
• W2368708758 modified "2023-10-03" @default.
• W2368708758 title "Use and Utility of Administrative Health Data for Stroke Research and Surveillance" @default.
• W2368708758 cites W1600386408 @default.
• W2368708758 cites W1604828483 @default.
• W2368708758 cites W1870272006 @default.
• W2368708758 cites W1974295393 @default.
• W2368708758 cites W1976494066 @default.
• W2368708758 cites W1991731220 @default.
• W2368708758 cites W2001372145 @default.
• W2368708758 cites W2004198484 @default.
• W2368708758 cites W2027679199 @default.
• W2368708758 cites W2039288075 @default.
• W2368708758 cites W2049706465 @default.
• W2368708758 cites W2057367182 @default.
• W2368708758 cites W2061326496 @default.
• W2368708758 cites W2118338777 @default.
• W2368708758 cites W2129642207 @default.
• W2368708758 cites W2134891979 @default.
• W2368708758 cites W2155291422 @default.
• W2368708758 cites W2155477484 @default.
• W2368708758 cites W2163278718 @default.
• W2368708758 cites W2171017045 @default.
• W2368708758 cites W2174373014 @default.
• W2368708758 cites W2186150902 @default.
• W2368708758 cites W2267205120 @default.
• W2368708758 cites W2318106356 @default.
• W2368708758 cites W2915142973 @default.
• W2368708758 doi "https://doi.org/10.1161/strokeaha.116.012390" @default.
• W2368708758 hasPubMedId "https://pubmed.ncbi.nlm.nih.gov/27174527" @default.
• W2368708758 hasPublicationYear "2016" @default.
• W2368708758 type Work @default.
• W2368708758 sameAs 2368708758 @default.
• W2368708758 citedByCount "42" @default.
• W2368708758 countsByYear W23687087582016 @default.
• W2368708758 countsByYear W23687087582017 @default.
• W2368708758 countsByYear W23687087582018 @default.
• W2368708758 countsByYear W23687087582019 @default.
• W2368708758 countsByYear W23687087582020 @default.
• W2368708758 countsByYear W23687087582021 @default.
• W2368708758 countsByYear W23687087582022 @default.
• W2368708758 countsByYear W23687087582023 @default.
• W2368708758 crossrefType "journal-article" @default.
• W2368708758 hasAuthorship W2368708758A5025499620 @default.
• W2368708758 hasAuthorship W2368708758A5031383231 @default.
• W2368708758 hasAuthorship W2368708758A5056711512 @default.
• W2368708758 hasAuthorship W2368708758A5068320077 @default.
• W2368708758 hasAuthorship W2368708758A5080252318 @default.
• W2368708758 hasAuthorship W2368708758A5084100416 @default.
• W2368708758 hasAuthorship W2368708758A5090161362 @default.
• W2368708758 hasBestOaLocation W23687087581 @default.
• W2368708758 hasConcept C127413603 @default.
• W2368708758 hasConcept C177713679 @default.
• W2368708758 hasConcept C2780645631 @default.
• W2368708758 hasConcept C545542383 @default.
• W2368708758 hasConcept C71924100 @default.
• W2368708758 hasConcept C78519656 @default.
• W2368708758 hasConcept C99508421 @default.
• W2368708758 hasConceptScore W2368708758C127413603 @default.
• W2368708758 hasConceptScore W2368708758C177713679 @default.
• W2368708758 hasConceptScore W2368708758C2780645631 @default.
• W2368708758 hasConceptScore W2368708758C545542383 @default.
• W2368708758 hasConceptScore W2368708758C71924100 @default.
• W2368708758 hasConceptScore W2368708758C78519656 @default.
• W2368708758 hasConceptScore W2368708758C99508421 @default.
• W2368708758 hasIssue "7" @default.
• W2368708758 hasLocation W23687087581 @default.
• W2368708758 hasLocation W23687087582 @default.
• W2368708758 hasOpenAccess W2368708758 @default.
• W2368708758 hasPrimaryLocation W23687087581 @default.
• W2368708758 hasRelatedWork W2029715266 @default.
• W2368708758 hasRelatedWork W2046478069 @default.
• W2368708758 hasRelatedWork W2063939897 @default.
• W2368708758 hasRelatedWork W2072633245 @default.
• W2368708758 hasRelatedWork W2119292156 @default.
• W2368708758 hasRelatedWork W2144793026 @default.
• W2368708758 hasRelatedWork W2919783417 @default.
• W2368708758 hasRelatedWork W3182778946 @default.
• W2368708758 hasRelatedWork W3212470821 @default.
• W2368708758 hasRelatedWork W381629422 @default.
• W2368708758 hasVolume "47" @default.
• W2368708758 isParatext "false" @default.
• W2368708758 isRetracted "false" @default.
• W2368708758 magId "2368708758" @default.
• W2368708758 workType "article" @default.