Yes / Secondary use of data via integrated health information technology is fundamental to many healthcare policies and processes worldwide. However, repurposing data can be problematic and little research has been undertaken into the everyday practicalities of inter-system data sharing that helps explain why this is so, especially within (as opposed to between) organisations. In response, this article reports one of the most detailed empirical examinations undertaken to date of the work involved in repurposing healthcare data for National Clinical Audits. Methods: Fifty-four semi-structured, qualitative interviews were carried out with staff in five English National Health Service hospitals about their audit work, including 20 staff involved substantively with audit data collection. In addition, ethnographic observations took place on wards, in ‘back offices’ and meetings (102 hours). Findings were analysed thematically and synthesised in narratives. Results: Although data were available within hospital applications for secondary use in some audit fields, which could, in theory, have been auto-populated, in practice staff regularly negotiated multiple, unintegrated systems to generate audit records. This work was complex and skilful, and involved cross-checking and double data entry, often using paper forms, to assure data quality and inform quality improvements. Conclusions: If technology is to facilitate the secondary use of healthcare data, the skilled but largely hidden labour of those who collect and recontextualise those data must be recognised. Their detailed understandings of what it takes to produce high quality data in specific contexts should inform the further development of integrated systems within organisations.

This diploma thesis deals with Information system of basic registries and its implementation. The information system consists of four main registries (Registry of Inhabitants, Registry of Persons, Registry of Territorial Identification, Addresses and Real Estates and Registry of Rights and Obligations) and system of ORG. The thesis defines crucial documents which deal with this project and the basic terms of Basic Registries Act No. 111/2009 Coll. and other related acts are explained. This thesis also compares the public administration before and after the activation of basic registries and the most important part is public survey which discovers knowledge about basic registries and responds to determined hypotheses.

Introdução: As estatísticas de mortalidade são usadas pelo mundo inteiro e por isso precisa ter dados confiáveis e comparáveis. Uma das estratégias para melhorar sua qualidade é a automação de etapas do fluxo dos dados. O Iris é um sistema que codifica as causas de morte da declaração de óbito (DO) e seleciona a causa básica de morte. Objetivos: adaptar o software para uso no Brasil, testá-lo e comparar seu uso com a codificação manual e com a seleção do sistema usado na rotina brasileira, o Sistema de Seleção da Causa Básica de Morte (SCB). Métodos: foram utilizadas uma amostra de declarações de óbito de óbitos ocorridos no período de janeiro a junho de 2014 em nove cidades representando as cinco regiões do Brasil. Foram selecionados óbitos pelas causas mais comuns, com menção de doença transmissível, assim como óbitos infantis, maternos e óbitos com menção de causa externa ou cirurgia. A fase 1 visou a conclusão do dicionário e a fase 2 testar e comparar o uso do Iris. Resultados: 1848 DO, com uma média de 3,1 linhas preenchidas por DO. Foram realizadas 618 ajustes ou adições nas tabelas do dicionário ou nas tabelas de padronização. Em 45,9 por cento das DO o Iris codificou todas as causas de morte presente na DO assim como selecionou a causa básica de morte. Das DO que o Iris não conseguiu concluir o processo, a maioria (97,8 por cento) dessas rejeições foi por dificuldade em encontrar ou processar um código da CID. A concordância do Iris com a codificação manual nos níveis de 4 caracteres, 3 caracteres e 1 caractere dos códigos da CID-10 foi de 73,3 por cento, 78,2 por cento e 83,9 por cento respectivamente. Isso representou uma discordância em 49,1 por cento das DO. A concordância da causa básica de morte com o SCB foi de 74,2por cento, 84,3 por cento e 91,8 por cento nos níveis de 4 caracteres, 3 caracteres e 1 caractere dos códigos da CID-10 respectivamente. A principal causa de discordância (71 por cento) foi por codificações diferentes. Após as pequenas correções, o Iris finaliza 66,6 por cento das DO. Conclusão: As discordâncias tiveram como causa os hábitos e rotinas dos codificadores que variam entre eles, enquanto no Iris a codificação de uma causa é sempre a mesma. As tabelas de decisão do SCB precisam ser revistas. Os dados mostram que é possível usar esta ferramenta e que ela vai diminuir o trabalho dos codificadores. O Iris teve uma boa taxa de finalização, semelhante a países que o utilizam na sua rotina. / Introduction: Mortality statistics is used all over the world and therefore needs reliable and comparable data. One of the strategies to improve quality is an automated data collection. Iris is a system that codes the causes of death of the death certificate (DC) and selects the underlying cause of death. Objective: To adapt Iris software to Brazil finalising the Portuguese dictionary, to test Iris and to compare it with manual coding and with the selection of the Basic Death Cause Selection System (Portuguese acronym is SCB). Methods: The sample was death certificates occurred from January to June of 2014 in nine cities representing the five regions of Brazil. Were selected to compose the sample: routine deaths, DC with mention of communicable disease, infant death, maternal death and DC with mention of external cause or surgery. Phase 1 aimed to complete the dictionary and phase 2 aimed to test and compare Iris. Results: The sample was1848 DC, with an average of 3.1 lines filled by DC. There were 618 adjustments or additions to dictionary tables or standardization tables. In 45.9 per cent of DC Iris coded all causes of death and selected the underlying cause of death. Of the DC that Iris was unable to complete the process, the majority (97.8 per cent) of these rejections were due to difficulty in finding or processing an ICD code. Iris agreement with manual coding at the 4-character, 3- character, and 1-character levels of ICD-10 codes was 73.3 per cent, 78.2 per cent and 83.9 per cent, respectively. This represented a disagreement in 49.1 per cent of DC. The concordance of the underlying cause of death with SCB was 74.2 per cent, 84.3 per cent and 91.8 per cent at the 4- character, 3-character and 1-character levels respectively. The main cause of discordance (71 per cent) was by different codes for the same cause of death. After the small corrections, Iris finalised 66.6 per cent of the DC. Conclusion: The disagreements were caused by coders routines that can vary between them, whereas in Iris the codification of a cause is always the same. The SCB decision tables need to be reviewed. The data show that it is possible to use this tool and that it will decrease the work of the coders. Iris had a good finalisation rate, similar to countries that use it in their routine.

Les patients coronariens stables ou stabilisés sont à haut risque d’évènements cardiovasculaires. Ils représentent une population hétérogène avec une présentation clinique, un terrain et un pronostic pouvant être extrêmement variables d’un patient à l’autre. Pourtant, d’après les recommandations internationales, tous doivent bénéficier d’une prise en charge relativement comparable basée sur des essais cliniques réalisés dans des sous-populations restreintes de patients stables et instables, pour la plupart anciens, et ne correspondant plus à la prise en charge actuelle des patients. Préciser les déterminants du pronostic de cette population, et notamment les stratégies thérapeutiques, est un enjeu majeur.Les antagonistes du système rénine-angiotensine (IEC/ARA2) font partie de l’arsenal médicamenteux de tout patient coronarien. Pourtant leur intérêt, en association aux antiagrégants plaquettaires et statines, est incertain chez les patients sans dysfonction ventriculaire gauche qui constituent un sous-groupe important parmi les patients stables.Le registre international REACH a évalué l’impact des IEC/ARA2 dans cette population avec 4 ans de suivi. La méthodologie statistique utilisée a été une analyse observationnelle avec ajustement ou avec appariement selon le score de propension. Il n’a pas été mis en évidence de bénéfice des IEC/ARA2 sur le critère de jugement principal composite associant décès cardiovasculaire – IDM – AVC, de même que sur le critère de jugement secondaire associant décès cardiovasculaire – IDM – AVC – Hospitalisation pour évènement athéro-thrombotique ou sur les critères tertiaires comprenant individuellement chacun des critères de jugement secondaire ainsi que sur la mortalité toute cause. Enfin il n’est pas ressorti non plus de bénéfice franc dans les sous-groupes d’analyse. Les résultats ont été concordants lorsque les analyses ont été réalisées pour les IEC seuls ou pour les ARA2 seuls, et ont été confortés par diverses analyses de sensibilité.Ces données méritent confirmation dans une cohorte indépendante. C’est l’un des objectifs du registre CLARIFY, registre de 32703 patients coronariens stables ou stabilisés, dont le suivi à 5 ans est terminé. Dans ce registre contemporain international, le taux global à 5 ans de mortalité toute cause a été de 7,9%, de mortalité non cardiovasculaire de 5% et de mortalité cardiovasculaire de 2,9%. Un évènement cardiovasculaire comprenant infarctus du myocarde (fatal ou non), angor instable, revascularisation coronaire par angioplastie ou pontage est survenu chez 15,9% des patients.Tout comme les IEC/ARA2, l’impact des bétabloquants dans la prise en charge du coronarien stable ou stabilisé, sans dysfonction ventriculaire est également controversé. Cette classe médicamenteuse est en cours d’évaluation dans CLARIFY. L’analyse tient compte du type de bétabloquant, de la dose prescrite, des éventuelles intolérances amenant à modifier leur utilisation, de la présence et de l’ancienneté d’un infarctus du myocarde et la fraction d’éjection ventriculaire gauche.CLARIFY a également pour objectif d’approfondir les déterminants du pronostic de la maladie coronarienne stable, avec une analyse spécifiquement focalisée sur la présence de symptômes angineux, d’ischémie myocardique et sur leur combinaison, en fonction de l’utilisation des méthodes de revascularisation myocardiques, pour mieux comprendre les mécanismes responsables des évènements cardiovasculaires et évoluer vers une prise en charge plus personnalisée. / Stable or stabilized coronary artery disease patients are at high risk for cardiovascular events. They represent a heterogeneous population. The clinical presentation, the context and the prognosis can be extremely variable from one patient to another. However, according to the international guidelines, those patients should be given a relatively comparable treatment based on clinical trials realized in restricted subpopulations of stable and unstable patients. Most of these trials are old, and no longer correspond to the current management. Specifying the determinants of the prognosis of this population, and in particular the therapeutic strategies, is a major challenge.The antagonist receptors of renin-angiotensin system (ACEI/ARB) are a part of the treatment of any coronary artery disease patient. Yet their interest in the prognosis of this population without left ventricular dysfunction in association with antiplatelet agents and statins is uncertain.The contemporary REACH registry has assessed the impact of ACEI/ARB in this population with a 4-year of follow-up. The statistical methodology used was based on the propensity score. After adjustment or matching with the propensity score, there was no benefit of ACEI/ARB on the primary endpoint of cardiovascular death - MI - stroke. No benefit was found on the secondary endpoint of cardiovascular death - MI - stroke - hospitalization for atherothrombotic events. No benefit was found on the tertiary criteria including individually each of the secondary endpoints and on any cause mortality. Finally,there was no clear benefit in the analyzes subgroups. These results were consistent when the analyzes were performed for ACEI alone or for ARB alone. They were also supported by sensitivity analyzes.These data should be confirmed or reversed in an independent cohort. This will be one of the many objectives of the CLARIFY registry, that enrolled 32,703 stable or stabilized coronary artery disease patients. The 5-year follow-up is complete. In this international contemporary registry, the overall 5-year rate of total mortality was 7.9%, non-cardiovascular mortality was 5% and cardiovascular mortality was 2.9%. A cardiovascular event including myocardial infarction (fatal or not), unstable angina, coronary revascularization by angioplasty or bypass surgery occured in 15.9% of patients.Like ACEI/ARB, the impact of betablockers on the management of stable or stabilized coronary artery disease without left ventricular dysfunction is also controversial. This drug class is being evaluated in CLARIFY. The analyzis takes into account the type of beta-blocker, the prescribed dose, any intolerance leading to changes in their use, the history of a myocardial infarction, and the left ventricular ejection fraction.CLARIFY will help to more define the determinants of the prognosis of stable coronary artery disease, with a more particular focus on symptomatic or not, ischemic or not, and revascularized or not, in order to better understand the mechanisms responsible for cardiovascular events, and evolve towards a more personalized and cost-effective care.

In the development process of novel cancer drugs, one important aspect is to identify patient populations with a high risk of early death so that resources can be focused on patients with the highest medical unmet need. Many cancer types are heterogeneous and there is a need to identify patients with aggressive diseases, meaning a high risk of early death, compared to patients with indolent diseases, meaning a low risk of early death. Predictive modeling can be a useful tool for risk stratification in clinical practice, enabling healthcare providers to treat high-risk patients early and progressively, while applying a less aggressive watch-and-wait strategy for patients with a lower risk of death. This is important from a clinical perspective, but also a health economic perspective since society has limited resources, and costly drugs should be given to patients that can benefit the most from a specific treatment. Thus, the goal of predictive modeling is to ensure that the right patient will have access to the right drug at the right time. In the era of personalized medicine, Artificial Intelligence (AI) applied to high-quality data will most likely play an important role and many techniques have been developed. In particular, Graph Neural Network (GNN) is a promising tool since it captures the complexity of high dimensional data modeled as a graph. In this work, we have applied Network Representation Learning (NRL) techniques to predict survival, using pseudonymized patient-level data from national health registries in Sweden. Over the last decade, more health data of increased complexity has become available for research, and therefore precision medicine could take advantage of this trend by bringing better healthcare to the patients. However, it is important to develop reliable prediction models that not only show high performances but take into consideration privacy, avoiding any leakage of personal information. The present study contributes novel insights related to GNN performance in different survival prediction tasks, using population-based unique nationwide data. Furthermore, we also explored how privacy methods impact the performance of the models when applied to the same dataset. We conducted a set of experiments across 6 dataset using 8 models measuring both AUC, Precision and Recall. Our evaluation results show that Graph Neural Networks were able to reach accuracy performance close to the models used in clinical practice and constantly outperformed, by at least 4.5%, the traditional machine learning methods. Furthermore, the study demonstrated how graph modeling, when applied based on knowledge from clinical experts, performed well and showed high resiliency to the noise introduced for privacy preservation. / I utvecklingsprocessen för nya cancerläkemedel är en viktig aspekt att identifiera patientgrupper med hög risk för tidig död, så att resurser kan fokuseras på patientgrupper med störst medicinskt behov. Många cancertyper är heterogena och det finns ett behov av att identifiera patienter med aggressiv sjukdom, vilket innebär en hög risk för tidig död, jämfört med patienter med indolenta sjukdom, vilket innebär lägre risk för tidig död. Prediktiv modellering kan vara ett användbart verktyg för riskstratifiering i klinisk praxis, vilket gör det möjligt för vårdgivare att behandla patienter olika utifrån individuella behov. Detta är viktigt ur ett kliniskt perspektiv, men också ur ett hälsoekonomiskt perspektiv eftersom samhället har begränsade resurser och kostsamma läkemedel bör ges till de patienter som har störst nytta av en viss behandling. Målet med prediktiv modellering är således att möjliggöra att rätt patient får tillgång till rätt läkemedel vid rätt tidpunkt. Framför allt är Graph Neural Network (GNN) ett lovande verktyg eftersom det fångar komplexiteten hos högdimensionella data som modelleras som ett diagram. I detta arbete har vi tillämpat tekniker för inlärning av grafrepresentationer för att prediktera överlevnad med hjälp av pseudonymiserade data från nationella hälsoregister i Sverige. Under det senaste decennierna har mer hälsodata av ökad komplexitet blivit tillgänglig för forskning. Även om denna ökning kan bidra till utvecklingen av precisionsmedicinen är det viktigt att utveckla tillförlitliga prediktionsmodeller som tar hänsyn till patienters integritet och datasäkerhet. Den här studien kommer att bidra med nya insikter om GNNs prestanda i prediktiva överlevnadsmodeller, med hjälp av populations -baserade data. Dessutom har vi också undersökt hur integritetsmetoder påverkar modellernas prestanda när de tillämpas på samma dataset. Sammanfattningsvis, Graph Neural Network kan uppnå noggrannhets -prestanda som ligger nära de modeller som tidigare använts i klinisk praxis och i denna studie preserade de alltid bättre än traditionella maskininlärnings -metoder. Studien visisade vidare hur grafmodellering som utförs i samarbete med kliniska experter kan vara effektiva mot det brus som införs av olika integritetsskyddstekniker.

Diss. (sammanfattning) Linköping : Univ., 2002. / Härtill 4 uppsatser.

Diss. (sammanfattning) Linköping : Univ., 2003. / Härtill 5 uppsatser.

Introdução: As estatísticas de mortalidade são usadas pelo mundo inteiro e por isso precisa ter dados confiáveis e comparáveis. Uma das estratégias para melhorar sua qualidade é a automação de etapas do fluxo dos dados. O Iris é um sistema que codifica as causas de morte da declaração de óbito (DO) e seleciona a causa básica de morte. Objetivos: adaptar o software para uso no Brasil, testá-lo e comparar seu uso com a codificação manual e com a seleção do sistema usado na rotina brasileira, o Sistema de Seleção da Causa Básica de Morte (SCB). Métodos: foram utilizadas uma amostra de declarações de óbito de óbitos ocorridos no período de janeiro a junho de 2014 em nove cidades representando as cinco regiões do Brasil. Foram selecionados óbitos pelas causas mais comuns, com menção de doença transmissível, assim como óbitos infantis, maternos e óbitos com menção de causa externa ou cirurgia. A fase 1 visou a conclusão do dicionário e a fase 2 testar e comparar o uso do Iris. Resultados: 1848 DO, com uma média de 3,1 linhas preenchidas por DO. Foram realizadas 618 ajustes ou adições nas tabelas do dicionário ou nas tabelas de padronização. Em 45,9 por cento das DO o Iris codificou todas as causas de morte presente na DO assim como selecionou a causa básica de morte. Das DO que o Iris não conseguiu concluir o processo, a maioria (97,8 por cento) dessas rejeições foi por dificuldade em encontrar ou processar um código da CID. A concordância do Iris com a codificação manual nos níveis de 4 caracteres, 3 caracteres e 1 caractere dos códigos da CID-10 foi de 73,3 por cento, 78,2 por cento e 83,9 por cento respectivamente. Isso representou uma discordância em 49,1 por cento das DO. A concordância da causa básica de morte com o SCB foi de 74,2por cento, 84,3 por cento e 91,8 por cento nos níveis de 4 caracteres, 3 caracteres e 1 caractere dos códigos da CID-10 respectivamente. A principal causa de discordância (71 por cento) foi por codificações diferentes. Após as pequenas correções, o Iris finaliza 66,6 por cento das DO. Conclusão: As discordâncias tiveram como causa os hábitos e rotinas dos codificadores que variam entre eles, enquanto no Iris a codificação de uma causa é sempre a mesma. As tabelas de decisão do SCB precisam ser revistas. Os dados mostram que é possível usar esta ferramenta e que ela vai diminuir o trabalho dos codificadores. O Iris teve uma boa taxa de finalização, semelhante a países que o utilizam na sua rotina. / Introduction: Mortality statistics is used all over the world and therefore needs reliable and comparable data. One of the strategies to improve quality is an automated data collection. Iris is a system that codes the causes of death of the death certificate (DC) and selects the underlying cause of death. Objective: To adapt Iris software to Brazil finalising the Portuguese dictionary, to test Iris and to compare it with manual coding and with the selection of the Basic Death Cause Selection System (Portuguese acronym is SCB). Methods: The sample was death certificates occurred from January to June of 2014 in nine cities representing the five regions of Brazil. Were selected to compose the sample: routine deaths, DC with mention of communicable disease, infant death, maternal death and DC with mention of external cause or surgery. Phase 1 aimed to complete the dictionary and phase 2 aimed to test and compare Iris. Results: The sample was1848 DC, with an average of 3.1 lines filled by DC. There were 618 adjustments or additions to dictionary tables or standardization tables. In 45.9 per cent of DC Iris coded all causes of death and selected the underlying cause of death. Of the DC that Iris was unable to complete the process, the majority (97.8 per cent) of these rejections were due to difficulty in finding or processing an ICD code. Iris agreement with manual coding at the 4-character, 3- character, and 1-character levels of ICD-10 codes was 73.3 per cent, 78.2 per cent and 83.9 per cent, respectively. This represented a disagreement in 49.1 per cent of DC. The concordance of the underlying cause of death with SCB was 74.2 per cent, 84.3 per cent and 91.8 per cent at the 4- character, 3-character and 1-character levels respectively. The main cause of discordance (71 per cent) was by different codes for the same cause of death. After the small corrections, Iris finalised 66.6 per cent of the DC. Conclusion: The disagreements were caused by coders routines that can vary between them, whereas in Iris the codification of a cause is always the same. The SCB decision tables need to be reviewed. The data show that it is possible to use this tool and that it will decrease the work of the coders. Iris had a good finalisation rate, similar to countries that use it in their routine.

Les objectifs de cette thèse étaient d’étudier les tendances du risque de second cancer primitif (SPC) selon l’année de diagnostic d’un premier cancer des voies aéro-digestives supérieures (VADS) dans le Bas-Rhin, de produire les premières estimations de l’incidence des SPC à l’échelle nationale en France et d’estimer la survie des patients atteints d’un cancer des VADS selon la présence d’antécédents de cancer. Ce travail a montré que : 1) L’excès de risque de SPC des VADS et de l’œsophage a diminué de 53% entre 1975 et 2006 dans le Bas-Rhin, mais que le risque de SPC du poumon est resté stable ; 2) Le risque de SPC en France est augmenté de 36% chez les patients atteints de cancer par rapport à la population générale ; 3) La survie des hommes atteints d’un cancer des VADS était fortement associée à la présence d’antécédents de cancer. Des perspectives se dégagent de ce travail en termes de recherche épidémiologique, de recherche clinique et de politiques de santé publique. / The objectives of this PhD thesis were: to study the trends of the risk of second primary cancer (SPC) among patients with a head and neck (HNSCC) cancer in Bas-Rhin, to provide first nationwide estimates of the risk of SPC in France and to assess the survival of patients with a HNSCC depending on their history of cancer. This work showed that : 1) The excess risk of SPC of head and neck and esophagus sites decreased by 53% over three decades among patients with a HNSCC, and that the excess risk of SPC of the lung did not change significantly. 2) The risk of SPC among cancer survivors in France was increased by 36% compared to the general population. 3) History of cancer was strongly associated with survival among HNSCC patients. Several epidemiological and clinical research perspectives can be established based on this work. These results also present an interest in a public health perspective in the framework of the third cancer plan.

Stroke thrombolysis is a method for restoring cerebral blood flow after ischemic stroke, with high priority in the Swedish national guidelines. implementation of stroke thrombolysis in Swedish routine stroke care has shown marked differences between demographic groups, hospital types, and regions. The general aim of this thesis were to examine the implementation of ischemic stroke thrombolysis in Swedish routine stroke care with an equity perspective; to gain more insight into the factors that influence implementation, how the treatment has reached patient groups, and differences in long-term outcomes between women and men. Analysis of data from research interviews with clinicians working within stroke care displayed that the facilitators of and barriers to the implementation of stroke thrombolysis could broadly be categorized into those related to individuals, to social interactions and context, and to organizational and resource issues. Key facilitating factors expressed in interviews were work pride and motivation, good leadership, involvement of all staff members in the implementation process, and quality assurance. Major barriers concerned lack of competence and experience, outdated attitudes regarding stroke management, counterproductive power structures, lack of continuity, and insufficient human resources. National quality register data displayed that stroke thrombolysis treatment expanded to reach more patients with mild deficits. Groups with higher education were more likely to receive treatment, compared to groups with lower educational level. These education group differences have, however, decreased over time in relative terms, but not in absolute terms. Further, there were considerable between-hospitals differences in treatment rates for patients with milder deficits, associated with hospital’s overall stroke thrombolysis rates. Moreover, larger non-university hospitals displayed treatment rate differences between educational groups that were not attributable to patient characteristics. Among thrombolysis-treated women and men, that was independent in ADL before their stroke and survived the first year post-stroke, women experienced higher probability to be dependent in ADL at both 3 and 12 months post-stroke, compared to men. This difference remained significant despite comprehensive adjustments for individual characteristics, symptom severity, and acute effects from stroke thrombolysis. This thesis displays that clinicians face barriers and facilitators at several levels, suggesting implementation interventions could be targeted towards both the individual-, the social interactions and context-, and also the organisation and available resources level. Assurance of clinicians’ individual competence, peer support, and clinical leadership seem to be important areas to intervene. Stroke thrombolysis rates have expanded over time, and an increase in stroke thrombolysis delivery to patients with mild stroke symptoms has contributed to this increase. However, it seems considerable differences between hospitals inhibit equity of care delivery. Further, socioeconomically disadvantaged groups receive less often stroke thrombolysis. Type of hospital seems to play a role, yet the reasons for this difference are not fully understood. This thesis also display that stroke thrombolysis-treated women that survive 1 year after stroke, appears to face higher risk for dependency in ADL, compared to men. / <p>Incorrect ISBN in print version 978-91-760-711-1. Correct ISBN should be 978-91-7601-711-1.</p>