Reconstruction of pseudonymized patient-trajectories in Austria’s stroke cohort using medical record-linkage of in-patient routine documentation to establish a nation-wide acute stroke cohort of 102,107 pseudonymized patients between 2015 and 2019

Abstract

Introduction:

Administrative health data are increasingly used for disease surveillance, quality assurance and research purposes. In Austria, reporting of a standardized dataset is mandatory for each patient.

Patients and methods:

Routine documentation includes administrative and medical data, including admission and discharge characteristics, disease-diagnosis using ICD-10, medical procedure codes, and coding of involved hospital departments. Since 2015, a three-step pseudonymization on these data is provided including a pseudonym using secure hash algorithm 256, a non-recalculable record-ID, and age-groups of 5 years, allowing the reconstruction of individual patient-trajectories. We included persons aged ⩾20 years with an in-patient treatment in Austrian hospitals for acute stroke or transient ischemic attack (TIA) between 01.01.2015 and 31.12.2019 using medical record-linkage.

Results:

This totals 102,107 patients (49.3% women) with 107,055 treatment episodes. An ischemic stroke (IS) occurred in 60.9% (n = 65,133), 27.1% (n = 29,019) had a TIA, 3.3% (n = 3488) a subarachnoid hemorrhage, and 8.8% (n = 9415) an intracerebral hemorrhage (ICH). The study period covers 35.2 million person-years at risk, with a hospitalization rate for acute stroke of 221.8 per 100,000 person-years (95% CI 220.2–223.3), and 185.1 per 100,000 person-years (95% CI 183.7–186.5) for IS. Unscheduled re-admissions within 1 year occurred in 29.2% (95% CI 28.8–29.7) after IS, and 41.7% (95% CI 40.0–43.3) after ICH. Recurrent stroke occurred in 5.3% (95% CI 5.1–5.5) after IS, and 5.6% (95% CI 4.9–6.4) after ICH.

Discussion:

We present herein the details of a novel methodology to establish a nation-wide unselected Austrian stroke cohort, and to reconstruct pseudonymized individual longitudinal patient-trajectories on a national level. This approach shows potential applications in epidemiological research, quality assessment and outcome measurement.

Conclusion:

This novel approach opens new research fields, facilitates international comparison, and is needed for national benchmarking to assess the achievement of goals according to the Stroke Action Plan for Europe and augment the quality of Austria’s integrated stroke care.

Keywords

Epidemiology stroke medical record-linkage pseudonymized patient-trajectory secure hash algorithm

Introduction

In-patient routine documentation is increasingly used to facilitate disease surveillance, quality assurance and research purposes using real-world data. Usage of electronic health records creates large datasets, posing enormous potential for research, quality measurement and population-based health measures. The collective information on administrative, personal, and medical data offers opportunities for demographic as well as geographical analyses. Furthermore, unique personal identifiers pose the potential to analyze patterns of temporal disease sequences (so called patient-trajectories). Finally, regionally as well as nationally implemented health care programs can be monitored and evaluated in a real-world setting.

In Austria, routine documentation is used for renumeration purposes,¹ research on health service provision as well as epidemiologic developments,^2–5 planning purposes, for example, the computing of residency training positions,⁶ and quality assurance as national outcome measurements in the form of “Austrian in-patient quality indicators” (A-IQI).⁷ The latter present a national adaption of an international standardized set of quality indicators based on administrative data, which allows a disease specific monitoring on a hospital-based, regional, national, and international level, in order to provide external and internal quality-management and -improvement. Indicators may serve as triggers for peer-reviews in annually specified subthemes as well as morbidity and mortality conferences. Stroke, as a major factor of mortality, morbidity, and disability worldwide,⁸ has been analyzed as a specified national subtheme of A-IQI in 2014, with peer-reviews urging quality improvement in structural, processual, medical, and documentational areas. One follow-up goal to improve stroke care was the implementation of a national strategy toward integrated stroke care, which has been published in form of a quality-standard.⁹ It focuses on the optimization of stroke care under the premises of patient-orientation, cross-sectoral continuity as well as evidence-based medicine, therapies, and care.

On a European level, the Stroke Action Plan for Europe (SAP-E) has defined seven domains, for which targets for 2030 were defined.¹⁰ It has been constated, that large differences within and between countries exist in quality of stroke care, with a call for publicly available standards in quality assessment and results of quality measurement, in order to facilitate national and international comparison as well as evidence-based practice monitoring and practice change.¹⁰

Multiple studies in various medical systems have shown the potential for routinely collected health-related data in stroke research.^11–13 Unselected nation-wide cohorts show the benefit of a highly outright, unbiased dataset without additional costs. In Austria, multiple national stroke-specific data sources exist in order to quantify, measure, and improve optimal individual care and therefore correspond to national and European objectives of evidence-based integrated stroke care. These include, (i) the mandatory routine documentation of in-patient care, with which pseudonymized patient-trajectories can be reconstructed as elaborated herein, (ii) the Austrian Stroke Unit Registry (ASUR) with comprehensive information on stroke unit patients, for which methodological principles have been published earlier,¹⁴ including one federal state (Tyrol) with detailed information on all stroke patients,¹⁵ and (iii) study-specific documentation for example, within the scope of the STROKE-CARD trial.^16,17 Together, these sources ought to give a comprehensive perspective on Austria’s national integrated stroke care, furthermore facilitating geographical and temporal comparison on a national and international level.

Herein we present a novel methodological approach of medical record-linkage of Austrian in-patient routine documentation to reconstruct pseudonymized patient-trajectories in Austria’s nation-wide stroke cohort.

Material and methods

In Austria, documentation of administrative data as well as medical data is mandatory for each in- and out-patient by federal law.¹⁸ In the in-patient setting, each hospital operator is obliged to collect information systematically and completely as outlined in Supplement 1, which is regarded as “routine documentation” or a “minimum basic data set” (MBDS).

Furthermore, each hospital operator is obliged to report these data sets to the data-warehouse of the responsible federal ministry of health. Hospitals financed by provincial health funds (e.g. public hospitals) must report via the provincial governments, hospitals not financed by provincial health funds (e.g. private hospitals) must send their data directly to the federal ministry. The technical and organizational framework for data transmission between hospitals and government is provided by each province. Each data transmission involving the federal ministry of health is encrypted using Secure Copy (SCP) with public key authentication. SCP is a network protocol which uses Secure Shell (SSH) for data transfer and the same mechanisms for authentication. Therefore, it ensures authenticity and confidentiality of the data.

Type of admission and type of discharge

Characteristic attributes for both admission and discharge exist. The type of admission is characterized by two attributes. For each attribute, various options exist, as outlined in Supplements 2 and 3. The type of discharge is characterized by one attribute with various options as outlined in Supplement 4.

Diagnosis of diseases

For each hospitalization, relevant pre-existing and newly diagnosed diseases ought to be documented. Diseases are coded according to the International Classification of Diseases, Version 10 (ICD-10) by the World Health Organization (WHO). A national version has been published by the responsible Austrian federal ministry of health¹⁹ with adjustments as outlined in Supplement 5.

The diagnosis, which poses the main reason for an in-patient stay must be recorded as the primary diagnosis. Certain specifications must be taken into consideration regarding the primary diagnosis as depicted in Supplement 6.

Medical procedure codes

For each hospital admission, selected medical procedure codes (diagnostic procedures, surgery, therapeutic procedures) are reliably recorded, according to the Austrian procedure-oriented hospital financing (“Leistungsorientierte Krankenanstaltenfinanzierung” – LKF), which is the national deduction-instrument of in-patient care.

For example, using this system, specialized care on a stroke unit (SU) can be reported with the selected medical procedure code “AA040 – acute stroke-care in a stroke unit,” with further stroke specific medical procedure codes summarized in Supplement 7. The unit size for this medical procedure code is one hospital stay, which defines, that this code can be reported solely once during each hospitalization.

As of 2022, the medical procedure code “ZZ723 – structured follow-up care after ischemic stroke and transient ischemic attack” is being implemented in routine care and will be a source of valuable additional information.

Documentation of care on intensive-care-units

Admission at an intensive-care-unit (ICU) prompts the documentation of the Simplified Acute Physiology Score 3 (SAPS3) on admission, as well as daily documentation according to the “Therapeutic Intervention Scoring System” (TISS) during the hospital stay.

Austrian procedure-oriented hospital financing

The procedure-oriented hospital financing according to the Austrian LKF-Model is organized in procedure-oriented diagnosis-case-groups (“Leistungsorientierte Diagnosefallgruppen – LDF groups”), including the registered medical procedure codes, the coded diagnosis as well as the involved hospital departments and the age of the patient as direct variables of its case flat-rate calculations. Indirect parameters that are included are care on ICU, multiple recordings of medical procedure codes as well as outliers concerning the duration of hospital-stays.

Patient-trajectory reconstruction

Since 2015, pseudonymization of health-related patient data is provided on Austrian routine documentation as depicted in Figure 1. Pseudonymization is ensured (i) by one-way derivation of the health-specific personal identifier (“bereichsspezifisches Personenkennzeichen Gesundheit-Gesundheitsdokumentation” – bPK GH-GD) into a non-recalculable pseudonym (hash-ID), (ii) by one-way derivation of the admission number into a non-recalculable record ID, and (iii) by replacing the date of birth by age groups of 5 years (i.e. 80 < 85 years). Both, the Austrian dataset as well as the methods and algorithms used for pseudonymization are regulated in rigorous federal laws and associated enactments.^18,20–23 This legal basis ensures that processing of this data is compliant with the European General Data Protection Regulation (GDPR) ((EU) 2016/679).²⁴

Figure 1.

Flow-chart of pseudonymization concerning patient health data using Austrian routine documentation.

According to legal regulations (i) the health personal ID must be pseudonymized inside a hardware security module into a non-recalculable pseudonym using an algorithm which is unknown to all partners and (ii) the record ID must be pseudonymized into a non-recalculable pseudonym using the cryptological hash function secure hash algorithm (SHA) 256.

Certain conditions must be met in order to be able to reconstruct patient-trajectories via pseudonyms, namely:

the registration of the individual as an insured person with a valid social security number in the central registry of the social security.

the registration of the individual in the central Austrian population registry.

the successful matching of the patient’s base data between the central registry of the social security and the central Austrian population registry.

the transmission of adequate admission or settlement data from the health operator to the social security.

Since 01.01.2015, administrative and medical information of routine documentation can therefore be analyzed on an individual’s basis with pseudonymized data while respecting patients’ privacy. Therefore, individual patient-trajectories can be reconstructed and analyzed. The analyses of the presented data were approved by the A-IQI scientific and steering committee.

Statistical analysis

Data extraction was conducted from DIAG into Microsoft Excel (Version 2109 Build 16.0.14430.20256). We identified all individuals hospitalized due to acute stroke or TIA, with multiple hospitalizations per individual equaling the number of treatment episodes over the study period. We assigned treatment episodes of each individual to a unique year by their respective start date. Incident events are reported per 100,000 person-years (py). Age-standardized hospitalization-rates were calculated using the direct method and 95% confidence intervals modeled assuming Poisson distribution. Age-adjustment was carried out using the combined Austrian census population over the investigated study period. Demographic information of the Austrian population was obtained from the federal statistical office “Statistics Austria”.²⁵ One-year acute re-admission and recurrence rates were calculated in our cohort between 2015 and 2018, ensuring a follow-up period of 365 days. Details how unscheduled re-admissions and acute recurrences were assessed are described in Supplement 8. Age-group was considered as a confounding variable. Direct age-adjustment was carried out, using the distribution of combined cases by age group as standard population for unscheduled re-admissions and recurrences. We used R (version 4.0.3) in the RStudio environment for all calculations. Graphics were created using Microsoft Excel (version 2109 Build 16.0.14430.20256), Microsoft PowerPoint (version 2204 Build 15128.20224), and yEd Graph Editor (version 3.22).

Results

Nation-wide Austrian stroke cohort

We included all individuals with a primary diagnosis of stroke or TIA according to Supplement 9, admitted to Austrian hospitals in a time period from 01.01.2015 until 31.12.2019 (n = 117,770). According to Supplement 10, the diagnosis “I64 – Unspecified stroke” was rarely used from 2013 onward and patients with this primary diagnosis were thus excluded from our analysis. Furthermore, we excluded patients with an age <20 years.

In order to exclude non-acute hospitalizations with a primary diagnosis of stroke or TIA, the admission types were specified as Type 1 “A – Admission to in-patient medical care including admission to intensive-care-unit” and Type 2 “A – Acute admission (without prior appointment).” Furthermore, LDF groups were specified, as depicted in Supplement 11. Excluded LDF groups are depicted in Supplement 12. A respective flow-chart is depicted in Figure 2.

Figure 2.

Flow-chart of included patients using Austrian in-patient routine documentation to reconstruct pseudonymized patient-trajectories after acute stroke between 01.01.2015 and 31.12.2019.

The nation-wide stroke cohort based on reconstructed patient-trajectories between 2015 and 2019 consists of 102,107 patients with in-patient admission due to acute stroke or TIA. We recorded 107,055 treatment episodes, of which a total of 60.9% (n = 65,133) had an acute ischemic stroke (IS), 27.1% (n = 29,019) a transient ischemic attack (TIA), 3.3% (n = 3488) a subarachnoid hemorrhage (SAH), and 8.8% (n = 9415) an intracerebral hemorrhage (ICH). The total cohort consists of 49.3% women (n = 50,360), and 50.7% men (n = 51,747).

Over the study period, the Austrian population aged 20 years or older increased from 6.89 to 7.14 million (3.5%). The population aged 50 years and above increased from 3.33 to 3.59 million (7.9%). Between 2015 and 2019 the data covers 35.2 million person-years (py) at risk.

Stroke hospitalization rate

Hospitalization rate for acute stroke, including IS, ICH, and SAH was 221.8 per 100,000 py (95% CI 220.2–223.3) over the entire study period. Rates for IS, TIA, ICH, and SAH are depicted in Table 1.

Table 1.

Age-adjusted cumulative hospitalization rate per 100,000 person-years (py) including 95% confidence intervals (CI), 2015–2019.

Diagnosis	Age-adjusted hospitalization rate per 100,000 py with 95% CI
Stroke	221.8 (220.2–223.3)
Stroke and TIA	304.2 (302.4–306.1)
Ischemic stroke	185.1 (183.7–186.5)
Intracerebral hemorrhage	26.8 (26.2–27.3)
Subarachnoid hemorrhage	9.8 (9.4–10.1)

The 1-year cumulative age-adjusted hospitalization rate of patients with acute stroke per 100,000 py at risk showed a significant reduction between 2015 (225.1, 95% CI 221.6–228.7) and 2019 (206.2, 95% CI 202.8–209.6) with a relative risk reduction of 0.92 (95% CI 0.90–0.93).

Analysis of re-admissions and recurrent events

Within 1 year after the initially documented acute IS in our cohort, 29.2% (95% CI 28.8–29.7) showed an unscheduled re-admission, and 5.3% (95% CI 5.1–5.5) an acute stroke. An IS recurrence occurred in 5.0% (95% CI 4.8–5.2). In ICH, 41.7% (95% CI 40.0–43.3) showed an unscheduled re-admission within 1 year. Within 365 days, 5.6% (95% CI 4.9–6.4) had a recurrent stroke. An ICH recurrence occurred in 3.4% (95% CI 2.8–4.0) within 1 year, whilst 2.0% (95% CI 1.6–2.5) had an IS.

Discussion

We present herein a novel methodology reconstructing pseudonymized individual patient-trajectories to establish a nation-wide unselected Austrian stroke cohort using medical record-linkage of Austrian in-patient routine documentation. This approach is equally suitable for epidemiological research and quality improvement in routine patient care including the monitoring and evaluation of newly implemented health care programs using real-world data. Major strengths of the presented methodology are its longitudinal recording of every in-patient admission of each pseudonymized patient in Austria, thus allowing both the retrospective as well as prospective monitoring of hospitalizations for any cause using compulsively coded routine documentation including demographic and medical information without additional costs. To the best of our knowledge, this is the first population-based, nation-wide dataset using one-way hashing for pseudonymization to reconstruct individual patient-trajectories, representing a secure technology to protect the privacy of each individual patient.

The presented methodology is compliant with both European and Austrian law as mentioned above. The European GDPR considers pseudonymized data as personal data, for which data protection principles should apply.²⁴ Pseudonymized data is considered to be processed so that it “. . . can no longer be attributed to a specific data subject without the use of additional information, provided that such additional information is kept separately and is subject to technical and organizational measures to ensure that the personal data are not attributed to an identified or identifiable natural person” (Article 4 (5)).²⁴ Processing of health data is granted according to the GDPR, if “. . . processing is necessary for reasons of public interest in the area of public health . . . ensuring high standards of quality and safety of health care . . .” ( Article 9 (2i)).²⁴ National laws further specify, that the usage of pseudonymized health-related data include the monitoring of steering-mechanisms of the Austrian health care system, quality assurance, and scientific analyses.¹⁸ Premises for health-data processing are firstly, installed adequate safeguards that is, technical and organizational measures as for example, pseudonymization (Article 89 (1)),²⁴ and secondly, the principle that analyses ought not to be aimed toward results on identifiable natural persons.²² The one-way derivation of the health-specific personal identifier, which contrary to encryption is a non-recalculable pseudonym (hash-ID) represents such an adequate safeguard, whereas further safety specifications, such as the prohibition to publish personal identifiers are defined in national laws.^23,26

In many countries, information regarding incidence and mortality of stroke is limited.²⁷ In Austria, patients with IS, ICH, and SAH irrespective of age and disability, including individuals from nursing homes and high-level home care, are transmitted to hospital care for stroke. There are no ambulatory or extramural services for TIA management in Austria, compared to other countries with TIA clinics like Denmark or France.^28,29 This policy is stable over the last decades and will be maintained. Throughout the COVID-19 crisis, no drop in stroke hospitalization was observed. Therefore, we present hospitalization rates (rather than incidence rates), and changes in the annual cumulative hospitalization, using the introduced stable and robust methodology, representing a reliable indicator of changes in disease epidemiology. These data, (complemented with information from the ASUR) pose the potential for national disease surveillance and benchmarking as well as international comparison with countries like Canada, where stroke reports are produced using Canadian Institute for Health Information administrative data,^30,31 or Denmark,^32–34 Finland,^35,36 Norway,³⁷ Sweden,³⁸ and Czech Republic³⁹ using their respective national patient and stroke registries (Table 2). Another nationwide prospective cohort is the Japan Stroke Data Bank, which recently published interesting results on 20-year changes in hospitalization-rates, stroke severity, and outcome.⁴⁰

Table 2.

Reported incidence-rates per 100,000 in European countries and regions.

	Period	Age group	Incidence rate per 100,000	Age standardization	Source
Ischemic stroke
Southern Sweden	2001	All	197 (95% CI 177–218)	Sweden census 2015	Aked et al.⁴¹
Southern Sweden	2015–2016	All	134 (95% CI 120–149)
Dijon, France	2003–2017	18–55 years	22.9 (95% CI 20.3–25.6)		Béjot et al.⁴²
Germany	2008	All	207 (95% CI 205.7–209.0)	Germany census 2008	Günster⁴³
Norway	2010	All	168.6	No	Rand et al.⁴⁴
Norway	2015	All	145.2
Czech Republic	2011	All	211	Czech Republic census 2011	Sedova et al.³⁹
Austria^a	2015–2019	⩾20 years	185.1 (95% CI 183.7–186.5)	Austria census 2015–2019
Austria^a		20–50 years	18.9 (95% CI 18.3–19.5)
Intracerebral hemorrhage
Germany	2008	All	28 (95% CI 27.5–28.7)	Germany census 2008	Günster⁴³
Norway	2010	All	28.8	No	Rand et al.⁴⁴
Norway	2015	All	28.3
Czech Republic	2011	All	29.0	Czech Republic census 2011	Sedova et al.³⁹
Austria^a	2015–2019	⩾20 years	26.8 (95% CI 26.2–27.3)	Austria census 2015–2019
Subarachnoid hemorrhage
Germany	2008	All	9 (95% CI 8.5–9.3)	Germany census 2008	Günster⁴³
Czech Republic	2011	All	8.2	Czech Republic census 2011	Sedova et al.³⁹
Austria^a	2015–2019	⩾20 years	9.8 (95% CI 9.4–10.1)	Austria census 2015–2019
Stroke
Southern Sweden	2001	All	246 (95% CI 224–270)	Sweden census 2015	Aked et al.⁴¹
Southern Sweden	2015–2016	All	165 (95% CI 149–182)
Gothenburg	2005–2006	⩾20 years; women	216 (95% CI 203–229)	European population	Harmsen et al.⁴⁵
Gothenburg	2005–2006	⩾20 years; men	286 (95% CI 269–303)	European population
Norway	2010	All	238.8 (without SAH)	No	Rand et al.⁴⁴
Norway	2015	All	194.7 (without SAH)
Czech Republic	2011	All	241	Czech Republic census 2011	Sedova et al.³⁹
Austria^a	2015–2019	⩾20 years	221.8 (95% CI 220.2–223.3)	Austria census 2015–2019

SAH: subarachnoid hemorrhage.

Hospitalization rate.

Incidence rates are higher than hospitalization rates given the failure of medical record-linkage (8%), patients with stroke symptoms who do not seek medical help, non-hospitalization of stroke patients, and non-consideration of strokes occurring during in-hospital stays. In Tyrol, incidence rates are estimated to be about 13% higher than hospitalization rates. In this federal state, in-hospital stroke is considered and accounted for 2.9% (95% CI 2.4–3.6) of all strokes in 2019–2020 (unpublished data). Hospitalization rates for IS patients were high (2010–2013, 97.4%). Reasons for non-hospitalization were death (0.2%), denial by the patient (0.9%), on-site medical care (0.7%), and other conditions (0.7%). Recent data from the STROKE-CARD trial, showed that 2.7% of acute IS and TIA patients in a large representative cohort in the federal state of Tyrol, reported typical symptoms of TIA or stroke and did not seek medical advice.⁴⁶ However, we have no information on other federal states and the divergence between incidence and hospitalization rates are much higher for TIA.

Within 1 year after the initially documented acute stroke, 29.2% of patients showed an unscheduled re-admission. These data are similar to a large observational cohort in the USA which showed a 27.2% re-admission within 1 year after stroke.⁴⁷ A total of 5.0% were readmitted within 1 year because of a recurrent IS. These data on stroke recurrence are similar to previous research which also showed that approximately one in 20 patients experienced a recurrent stroke within the first year after the index event.^48,49 Also, the recent large-scale STROKE-CARD randomized controlled trial conducted in two centers in Austria, which included 2149 stroke and TIA patients, showed a 1-year recurrence rate of 6.4%.^16,17 After ICH, 41.7% showed an unscheduled re-admission within 1 year. The NORSTROKE study found a re-admission rate after ICH of 40.6%⁵⁰ and a study using hospital record-linkage in Scotland found a re-admission rate of 43.6% within the first year after hemorrhagic stroke.⁵¹ Literature showed a recurrent stroke risk within the first year after ICH of 4.8%⁵² (Austria: 5.6%), and a 1-year recurrent ICH risk of 2.1%⁵³ (Austria: 3.4%). This risk of recurrent ICH also seems not to increase in years 2–5 after the index event, on the other hand, the risk of IS after ICH in years 2–5 increases up to sixfold depending on the location of the ICH (deep or lobar).⁵⁴ Recently, knowledge on how to best organize post-hospital care of stroke patients has emerged.⁵⁵

Additionally, the presented methodology poses the opportunity to assess ratios of patients receiving the essential treatments of stroke, namely SU care as presented in the accompanied analysis⁵⁶, systemic thrombolysis and mechanical thrombectomy. Furthermore, in-hospital case-fatality can be assessed. Recently, a nation-wide structured, multifaceted post-stroke disease management program according to the STROKE-CARD concept^16,17 has been implemented in Austria and will be remunerated as of 2022, in order to provide a standardized 3-month out-patient follow-up visit by a multidisciplinary team to all patients who experienced a TIA or IS. The aim is to ameliorate risk factor control, assess and treat post-stroke complications as well as comorbidities, to assess cardiovascular warning signs as well as rehabilitation demands and implement continued patient education, counseling and self-empowerment.¹⁷ This initiative will collect reliable information on 3-month outcome and death. The presented methodology gives us the opportunity to assess favorable effects of the STROKE-CARD concept on unscheduled 1-year hospital re-admissions in the Austrian stroke cohort.

Periodic data audits are being performed in Austria and a high data validity can therefore be assumed. In the Austrian federal state Tyrol, ICD-10 coding accuracy for stroke was 96.9% in the eight hospitals, with false-negatives (mainly misclassification as TIAs) and false-positives balancing each other.¹⁵ Validation analyses between routine documentation and information obtained from ASUR, comparing results of unscheduled re-admissions show reassuring results.⁵⁷ Multiple validity studies have confirmed the methodological reliability of using routine documentation for stroke research.¹¹ However, studies from France and Italy both reported underestimations of stroke coding in administrative data.^12,58,59 A more recent Australian study including 14,716 patients using ICD-10 reported positive predictive values (PPV) between 85.5% for TIA, 91.4% for ICH, and 95.7% for IS between registry and administrative data, whilst I64 (which was basically not used in Austria since 2013, Supplement 10) showed low PPVs of 21,9%.¹³ The validity of stroke diagnoses was shown to be 90% (95% CI 87–93) in the National Registry of Hospitalized Patients in Czech Republic, with TIA diagnosis showing a substantially lower validity of 49% (95% CI 39–58).⁶⁰ McCormick et al.¹¹ found, that the analysis of acute stroke cannot sufficiently be obtained by primary diagnosis alone. Here, we want to emphasize our specified inclusion criteria to cater this need as described above.

There are different definitions regarding primary diagnoses in different parts of the world, with the presented Austrian definition corresponding to the one used in the United States and Australian health-care system, where the primary diagnosis is the principal reason for admission.⁶¹ In an analysis by Schmidt et al.³⁴ on the quality of stroke coding quality in the Danish national patient registry, approximately one-third of stroke admissions were encoded by the unspecific diagnosis I64, of those, two-thirds were IS.³³ In Austria, documentation of stroke diagnosis has been reviewed as part of the 2014 A-IQI analysis and the unspecific diagnosis of I64 has hardly ever been coded in Austria since 2013 (Supplement 10).

It has been elaborated earlier, that second use data should be carefully interpreted in proper knowledge of the underlying methodology.^32,35 Benefits using routine documentation include an almost complete, unselected and therefore unbiased data set of the Austrian stroke population with real-world data and standardized information on demographic, administrative and medical data (e.g. diagnoses, diagnostic as well as therapeutic procedures), all without substantial additional work and therefore at minimal cost. The presented methodology offers the opportunity to study patient-trajectories with past events (e.g. risk factors or relationship with cancer), in an anterograde direction (e.g. recurrent events, post-stroke complication or unscheduled re-admission), as well as bi-directional temporal relations between various diseases (e.g. the relationship between atrial fibrillation and stroke). Additionally, exploratory studies on rare diseases can be performed. Therefore, it is a complementary data source to the European RES-Q, the Austrian Stroke Unit Registry as well as Austrian study-specific data-sets as for example, the STROKE-CARD trial, in order to inform evidence-based quality improvements, and offers the opportunity for additional as well as in-depth analyses of A-IQI, which are regularly submitted to the involved stakeholders on a hospital-based, regional, and national level.

Limitations of pseudonymized patient-trajectory reconstruction

Completeness of the pseudonymized dataset in comparison to routine documentation from hospital remuneration was 92% between 2015 to 2019, due to the above-mentioned conditions of pseudonymization, which are altogether required for successful processing. As routine documentation in the out-patient setting shows lower cohesiveness in comparison to in-patient documentation, its use for scientific purposes is currently not recommended.

As a matter of fact, remuneration data is coded for this specific purpose. Bias concerning its medical correctness can therefore not be ruled out. However, regular stakeholder-meetings, technical reviews, and strict controlling measures, including for example, an integrated warning-system for cases with outliers, audits, plausibility-checks, and case-specific follow-ups ensure both a high quality and correct quantity of coding.

Limitations concern the documentation of acute disease manifesting during in-patient stays. As already described, the primary diagnosis does not change if a stroke occurs during a hospitalization. In-hospital strokes, however, account for only about 3% of all strokes in Austria. Moreover, it is not always possible to discriminate, whether the qualifying stroke event represents the first-ever or a recurrent stroke.

Further limitations of routine documentation concern firstly, a possible bias to more severe forms of stroke, as patients with mild symptoms may either not present to medical care or not be hospitalized, and secondly, information, which is not included in the MBDS. For example, the severity of an acute stroke is typically recorded according to the National Institute of Health Stroke Scale (NIHSS),⁶² which however, is not part of the Austrian routine documentation. Finally, case-fatality analysis using routine documentation only considers in-hospital deaths.

Conclusion

We show detailed information on the methodological approach to establish a nation-wide unselected Austrian stroke cohort using medical record-linkage of routine documentation and to reconstruct pseudonymized individual patient-trajectories in the past and future. The presented methodology and data will be needed to assess Austria’s progress in reaching the challenging goals for integrated stroke care as defined in the SAP-E,¹⁰ to compare the quality of Austria’s stroke care internationally, for national benchmarking which should aid in further augmenting the quality of stroke care, and to improve our epidemiological knowledge of stroke in Austria.

Supplemental Material

sj-docx-2-eso-10.1177_23969873221107619 – Supplemental material for Reconstruction of pseudonymized patient-trajectories in Austria’s stroke cohort using medical record-linkage of in-patient routine documentation to establish a nation-wide acute stroke cohort of 102,107 pseudonymized patients between 2015 and 2019

Supplemental material, sj-docx-2-eso-10.1177_23969873221107619 for Reconstruction of pseudonymized patient-trajectories in Austria’s stroke cohort using medical record-linkage of in-patient routine documentation to establish a nation-wide acute stroke cohort of 102,107 pseudonymized patients between 2015 and 2019 by Martin Heidinger, Wilfried Lang, Christian Boehme, Michael Knoflach, Stefan Kiechl, Peter Willeit, Rainer Kleyhons and Silvia Tuerk in European Stroke Journal

Supplemental Material

sj-tif-1-eso-10.1177_23969873221107619 – Supplemental material for Reconstruction of pseudonymized patient-trajectories in Austria’s stroke cohort using medical record-linkage of in-patient routine documentation to establish a nation-wide acute stroke cohort of 102,107 pseudonymized patients between 2015 and 2019

Supplemental material, sj-tif-1-eso-10.1177_23969873221107619 for Reconstruction of pseudonymized patient-trajectories in Austria’s stroke cohort using medical record-linkage of in-patient routine documentation to establish a nation-wide acute stroke cohort of 102,107 pseudonymized patients between 2015 and 2019 by Martin Heidinger, Wilfried Lang, Christian Boehme, Michael Knoflach, Stefan Kiechl, Peter Willeit, Rainer Kleyhons and Silvia Tuerk in European Stroke Journal

Footnotes

Acknowledgements

NA.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Informed consent

Informed consent was not sought for the present study according to federal Austrian laws and regulations. Rigorous European and federal laws and enactments concerning the usage of pseudonymized data as the presented data exist.^{18,20–24,26}

Ethical approval

This study was conducted according to federal Austrian laws and regulations. Rigorous federal laws and enactments concerning the usage of these data exist.^18,20,21 This study was completed in accordance with the Helsinki Declaration as revised in 2013. The analyses of the presented data were approved by the A-IQI scientific and steering committee.

Guarantor

ST.

Author contribution

All authors contributed to the study conception and design. Material preparation, and data collection were performed by MH and RK. Data analysis was performed by MH. The first draft of the manuscript was written by MH, and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

ORCID iDs

Martin Heidinger

Christian Boehme

Supplemental material

Supplemental material for this article is available online.

References

BMG. Das österreichische LKF-System. Wien: BMG, https://www.sozialministerium.at/dam/jcr:d45d56b6-d092-4d73-b793-daa6c6eadd70/LKF-Broschüre_BMG_2010_NACHDRUCK_2011_20200124.pdf (2010, accessed 9 April 2022)

Türk

. Qualitätsberichterstattung zur Hüft- und Knieendoprothetik in Österreich – wie Routinedaten zur Analyse von Revisionen genutzt werden. In: Dormann

Klauber

Kuhlen

(eds) Qualitätsmonitor 2019. Berlin: MWV Medizinich Wissenschaftliche Verlagsgesellschaft mbH & Co. KG, 2019, pp.19–31.

Türk

. Hüft- und Knie-Endoprothetik in Österreich. Wien: BMASGK, https://www.sozialministerium.at/dam/jcr:a32545b2-d40c-43f2-ac1f-7c6ce97804a8/endoprothetik-bericht_27.07.18_final.pdf (2018, accessed 9 April 2022).

Leitner

Türk

Heidinger

, et al Trends and economic impact of hip and knee arthroplasty in central Europe: findings from the Austrian national database. Sci Rep 2018; 8: 4707.

Haug

. Identification of gatekeeper diseases on the way to cardiovascular mortality. CoRR; abs/1908.0, http://arxiv.org/abs/1908.00920 (2019).

Heidinger

. Ärzteausbildung (ÄAO 2015) - Technische Definitionen basierend auf KEF und RZ-VO 2015 – 5. Novelle sowie SpezV 2017 – 5. Novelle. Vienna: BMSGPK, https://www.sozialministerium.at/dam/jcr:569a0a15-b25c-4214-a43d-2320a9044a04/Definitionshandbuch_ÄAO_5.0.pdf (2022).

BMSGPK. Austrian Inpatient Quality Indicators (A-IQI) - Bericht 2020. Vienna, https://www.sozialministerium.at/dam/jcr:6c0bbfdb-b553-4eef-a0d1-64d1ae992990/A-IQI_Bericht_2020.pdf (2020, accessed 9 April 2022).

Feigin

Forouzanfar

Krishnamurthi

, et al Global and regional burden of stroke during 1990–2010: findings from the Global Burden of Disease Study 2010. Lancet 2014; 383: 245–255.

BMASGK. Qualitätsstandard Integrierte Versorgung Schlaganfall. Wien: BMASGK, https://www.sozialministerium.at/dam/jcr:cf87f8be-d299-4c57-ad40-872f1818fba5/qs_schlaganfall.pdf (2018, accessed 9 April 2022).

10.

Norrving

Barrick

Davalos

, et al Action plan for stroke in Europe 2018–2030. Eur Stroke J 2018; 3: 309–336.

11.

McCormick

Bhole

Lacaille

, et al Validity of diagnostic codes for acute stroke in administrative databases: a systematic review. PLoS One 2015; 10: e0135834.

12.

Baldereschi

Balzi

Di Fabrizio

, et al Administrative data underestimate acute ischemic stroke events and thrombolysis treatments: data from a multicenter validation survey in Italy. PLoS One 2018; 13: e0193776.

13.

Ryan

Riley

Cadilhac

, et al Factors associated with stroke coding quality: a comparison of registry and administrative data. J Stroke Cerebrovasc Dis. Epub ahead of print 1 February 2021. DOI: 10.1016/j.jstrokecerebrovasdis.2020.105469

14.

Hofer

Kiechl

Lang

. Das Österreichische stroke-unit-register. Wien Med Wochenschr 2008; 158: 411–417.

15.

Willeit

Geley

Schöch

, et al Thrombolysis and clinical outcome in patients with stroke after implementation of the Tyrol Stroke Pathway: a retrospective observational study. Lancet Neurol 2015; 14: 48–56.

16.

Toell

Boehme

Mayer

, et al Pragmatic trial of multifaceted intervention (STROKE-CARD care) to reduce cardiovascular risk and improve quality-of-life after ischaemic stroke and transient ischaemic attack –study protocol. BMC Neurol 2018; 18: 187.

17.

Willeit

Toell

Boehme

, et al STROKE-CARD care to prevent cardiovascular events and improve quality of life after acute ischaemic stroke or TIA: a randomised clinical trial. EClinicalMedicine 2020; 25: 100476.

18.

BGBl. Nr. 745/1996 lastly amended 2018. Bundesgesetz über die Dokumentation im Gesundheitswesen, https://www.ris.bka.gv.at/GeltendeFassung.wxe?Abfrage=Bundesnormen&Gesetzesnummer=10011011 (1996, accessed 9 April 2022).

19.

Bundesministerium für Gesundheit und Frauen. ICD-10 BMSGPK 2022 – Systematisches Verzeichnis, https://www.sozialministerium.at/dam/jcr:64beeaa0-ec63-4864-a954-0ee1beb9e5c8/ICD-10%20BMSGPK%202022+%20-%20SYSTEMATISCHES%20VERZEICHNIS.pdf (accessed 12 June 2022).

20.

BMASGK. Handbuch zur Dokumentation (Anhang 1) - Organisation & Datenverwaltung - Aufbau und Inhalt der Datenmeldung. Wien: BMASGK, https://www.sozialministerium.at/dam/jcr:9685da35-603a-4263-8025-6be5628c99b6/ANHANG1-AUFBAUUNDINHALTDERDATENMELDUNGEN2020.pdf (2018, accessed 9 April 2022).

21.

BMASGK. Handbuch zur Dokumentation - Organisation und Datenverwaltung. Wien: BMASGK, https://www.sozialministerium.at/dam/jcr:9b94b092-9be9-401e-afc5-f3c8a276e748/HANDBUCHORGANISATIONUNDDATENVERWALTUNG2020.pdf (2019, accessed 9 April 2022).

22.

BGBl. I Nr. 165/1999 lastly amended 2021. Bundesgesetz zum Schutz natürlicher Personen bei der Verarbeitung personenbezogener Daten (Datenschutzgesetz – DSG), https://www.ris.bka.gv.at/GeltendeFassung.wxe?Abfrage=bundesnormen&Gesetzesnummer=10001597 (1999, accessed 9 April 2022).

23.

BGBl. Nr. 341/1981 lastly amended 2021. Bundesgesetz über allgemeine Angelegenheiten gemäß Art. 89 DSGVO und die Forschungsorganisation (Forschungsorganisationsgesetz – FOG), https://www.ris.bka.gv.at/GeltendeFassung.wxe?Abfrage=Bundesnormen&Gesetzesnummer=10009514 (1981, accessed 9 April 2022).

24.

European Commission. European General Data Protection Regulation, https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX%3A02016R0679-20160504&qid=1532348683434 (2016, accessed 9 April 2022).

25.

Bundesanstalt Statistik Österreich. Statistics Austria, https://www.statistik.at/en/ (accessed 12 June 2022).

26.

BMASGK. Schutz sensibler Daten. Vienna, https://www.sozialministerium.at/dam/jcr:99d679b8-3676-455e-abfc-a6ec27c4d64d/Schutz_sensibler_Daten_-_Position_der_Gesundheitssektionen_VIII_und_IX_.pdf (2019, accessed 9 April 2022).

27.

Kim

Thayabaranathan

Donnan

, et al Global stroke statistics 2019. Int J Stroke 2020; 15: 819–838.

28.

Hastrup

Johnsen

Jensen

, et al Specialized outpatient clinic vs stroke unit for TIA and minor stroke: a cohort study. Neurology 2021; 96: e1096–e1109.

29.

Lavallée

Meseguer

Abboud

, et al A transient ischaemic attack clinic with round-the-clock access (SOS-TIA): feasibility and effects. Lancet Neurol 2007; 6: 953–960.

30.

Canadian Institute for Health Information. Privacy and Security Risk Management Framework. Ontario, ON, https://www.cihi.ca/sites/default/files/document/privacy-security-risk-management-framework-en.pdf (2020, accessed 9 April 2022).

31.

Heart & Stroke Canada. 2018 Stroke Report, https://heartstrokeprod.azureedge.net/-/media/pdf-files/canada/stroke-report/strokereport2018.ashx?rev=8491d9c349f7404491f36be67f649c0b (2018, accessed 17 March 2022).

32.

Lynge

Sandegaard

Rebolj

. The Danish national patient register. Scand J Public Health 2011; 39: 30–33.

33.

Krarup

Boysen

Janjua

, et al Validity of stroke diagnoses in a national register of patients. Neuroepidemiology 2007; 28: 150–154.

34.

Schmidt

Sandegaard

, et al The Danish national patient registry: a review of content, data quality, and research potential. Clin Epidemiol 2015; 7: 449–490.

35.

Sund

. Quality of the Finnish hospital discharge register: a systematic review. Scand J Public Health 2012; 40: 505–515.

36.

Tolonen

Salomaa

Torppa

, et al The validation of the Finnish hospital discharge register and causes of death register data on stroke diagnoses. Eur J Prev Cardiol 2007; 14: 380–385.

37.

Bakken

Ariansen

AMS

Knudsen

, et al The Norwegian patient registry and the Norwegian registry for primary health care: research potential of two nationwide health-care registries. Scand J Public Health 2020; 48: 49–55.

38.

Ludvigsson

Andersson

Ekbom

, et al External review and validation of the Swedish national inpatient register. BMC Public Health 2011; 11: 450–466. DOI: 10.1186/1471-2458-11-450

39.

Sedova

Brown

Zvolsky

, et al Incidence of hospitalized stroke in the Czech Republic: the National Registry of Hospitalized Patients. J Stroke Cerebrovasc Dis 2017; 26: 979–986.

40.

Toyoda

Yoshimura

Nakai

, et al Twenty-year change in severity and outcome of ischemic and hemorrhagic strokes. JAMA Neurol 2022; 79: 61–69.

41.

Aked

Delavaran

Norrving

, et al Temporal trends of stroke epidemiology in Southern Sweden: a population-based study on stroke incidence and early case-Fatality. Neuroepidemiology 2018; 50: 174–182.

42.

Béjot

Duloquin

Thomas

, et al Temporal trends in the incidence of ischemic stroke in young adults: Dijon Stroke Registry. Neuroepidemiology 2021; 55: 239–244.

43.

Günster

. Schlaganfallversorgung in Deutschland – Inzidenz, Wiederaufnahmen, Mortalität und Pflegerisiko im Spiegel von Routinedaten, Versorgungs-Report 2011, Stuttgart, Schattauer, 2011, pp.147–163.

44.

Rand

Dahl

Viana

, et al Fewer ischemic strokes, despite an ageing population: stroke models from observed incidence in Norway 2010-2015. BMC Health Serv Res 2019; 19: 705.

45.

Harmsen

Wilhelmsen

Jacobsson

. Stroke incidence and mortality rates 1987 to 2006 related to secular trends of cardiovascular risk factors in Gothenburg, Sweden. Stroke 2009; 40: 2691–2697.

46.

Boehme

Toell

Mayer-Suess

, et al. The dimension of preventable stroke in a large representative patient cohort. Neurology 2019; 93(23): e2121–e2132. DOI: 10.1212/WNL.0000000000008573

47.

Bravata

Meehan

, et al Readmission and death after hospitalization for acute ischemic stroke: 5-year follow-up in the medicare population. Stroke 2007; 38: 1899–1904.

48.

Amarenco

. One-Year risk of stroke after transient ischemic attack or minor stroke. New Engl J Med 2016; 374: 1533–1542.

49.

Strambo

Zachariadis

Lambrou

, et al A score to predict one-year risk of recurrence after acute ischemic stroke. Int J Stroke 2021; 16: 602–612.

50.

Bjerkreim

Thomassen

Waje-Andreassen

, et al Hospital readmission after Intracerebral Hemorrhage. J Stroke Cerebrovasc Dis 2016; 25: 157–162.

51.

Christensen

Munro

. Ischemic stroke and intracerebral hemorrhage: the latest evidence on mortality, readmissions and hospital costs from Scotland. Neuroepidemiology 2008; 30: 239–246.

52.

Pennlert

Eriksson

Carlberg

, et al Long-term risk and predictors of recurrent stroke beyond the acute phase. Stroke 2014; 45: 1839–1841.

53.

Flynn

MacDonald

Murray

, et al Prescribing antiplatelet medicine and subsequent events after intracerebral hemorrhage. Stroke 2010; 41: 2606–2611.

54.

Casolla

. Five-year risk of major ischemic and hemorrhagic events after intracerebral hemorrhage. Stroke 2019; 50: 1100–1107.

55.

Boehme

Toell

Lang

, et al Longer term patient management following stroke: a systematic review. Int J Stroke 2021; 16: 917–926.

56.

Heidinger

Boehme

Knoflach

, et al. Hospitalization rates, stroke unit care and in-hospital case fatality in Austria’s stroke cohort - Analysis of key quality indicators in a nation-wide acute stroke cohort of 102,107 patients between 2015-2019. Eur Stroke J 2022.

57.

BMSGPK. Status Quo der österreichischen Schlaganfall-Versorgung. Wien, https://www.sozialministerium.at/dam/jcr:1a23e7d6-e973-40be-96d1-62d113536f2c/Status Quo der österreichischen Schlaganfall-Versorgung.pdf (2020, accessed 9 April 2022).

58.

Aboa-Eboulé

Mengue

Benzenine

, et al How accurate is the reporting of stroke in hospital discharge data? A pilot validation study using a population-based stroke registry as control. J Neurol 2013; 260: 605–613.

59.

Roussot

Cottenet

Gadreau

, et al The use of national administrative data to describe the spatial distribution of in-hospital mortality following stroke in France, 2008-2011. Int J Health Geogr 2016; 15: 2.

60.

Sedova

Brown

Zvolsky

, et al Validation of stroke diagnosis in the National Registry of Hospitalized Patients in the Czech Republic. J Stroke Cerebrovasc Dis 2015; 24: 2032–2038.

61.

AYX

Holodinsky

Zerna

, et al Use and utility of administrative health data for stroke research and surveillance. Stroke 2016; 47: 1946–1952.

62.

National Institute of Neurological Disorders and Stroke rt-PA Stroke Study Group. Tissue plasminogen activator for acute ischemic stroke. New Engl J Med 1995; 333: 1581–1587.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.03 MB

0.23 MB