Cette thèse porte sur la modélisation des comportements individuels de vaccination. Lorsque la vaccination est libre, les individus peuvent décider ou non de se vacciner, et ces décisions influencent la propagation de la maladie. Nous proposons une approche de modélisation flexible qui permet de prendre en compte ces décisions individuelles dans les modèles de simulation épidémiologiques.Dans le Chapitre 1, nous étudions la décision de vaccination face à une maladie inspirée de la rougeole. Nous montrons comment inclure les décisions de vaccination dans un modèle épidémiologique en les calculant comme un point fixe fonctionnel. Nous obtenons des résultats pour un modèle avec taux de mort et de naissance, et perte d'immunité vaccinale. Nos résultats sur longue période de temps mettent en évidence des pics épidémiques récurrents. À titre de comparaison, nous produisons également des résultats pour des individus ayant un comportement adaptatif.Les objectifs du Chapitre~2 sont (i) de montrer que la boucle entre prévalence et comportements individuels ne peut pas être négligée dans les évaluations de politiques de santé publique, et (ii) de présenter un outil pour les inclure dans ces évaluations. Nous développons l'exemple de la vaccination obligatoire contre la rougeole. Notre modèle épidémiologique est le modèle SIR habituellement utilisé pour représenter la rougeole. Nos résultats suggèrent que l'anticipation de la vaccination obligatoire peut conduire à une augmentation transitoire de la prévalence avant l'éradication à long terme de la maladie. Ceci conduirait à d'importants transferts d'utilité entre générations. Ironiquement, dans notre scénario, des individus anti-vaccins sont parmi ceux qui bénéficient le plus de la vaccination obligatoire.Dans le Chapitre~3, nous partons du constat que la comparaison des coûts de vaccination avec le risque d'être infecté par la rougeole peine à expliquer la couverture vaccinale relativement élevée (bien que souvent insuffisante) dans les pays développés. Nous discutons l'hypothèse selon laquelle la vaccination est un comportement coopératif. Nous mettons en œuvre des concepts d'équilibre et de punition habituellement utilisés en théorie des jeux répétés en donnant des arguments pour leur utilisation dans le contexte de la vaccination. Nos résultats indiquent que la menace d'une punition peut expliquer la vaccination lorsqu'elle serait normalement sous-optimale. / We tackle the issue of including individual vaccination decisions in epidemiological models. We draw on the example of Measles vaccination, and we focus on strategic interactions and anticipatory behavior. We contribute to a fuller account of such behaviors by developing a modeling approach intended as a tool for practitioners and theorists.In Chapter 1, we show how the interplay between individual anticipatory vaccination decisions and the otherwise biological dynamics of a disease may lead to the emergence of recurrent patterns. We consider a Measles-like outbreak, rational and far-sighted individuals, vital dynamics, and waning vaccine efficacy. This chapter illustrates the versatility of our approach. For comparison, we provide results for individuals with adaptive behavior.In Chapter 2, we investigate the effect of anticipatory behavior in a scenario where Measles vaccination becomes mandatory. When mandatory vaccination is announced in advance, we show that individuals may alter their vaccination behavior, thus causing an increase in prevalence before Measles is ultimately eradicated. These transition effects lead to non negligible welfare differences between generations. We consider an anti-vaccinationist subpopulation with a higher vaccination cost, and exhibit scenarios where anti-vaccinationists are among those who benefit the most from mandatory vaccination.In Chapter 3, we discuss whether coalitions of vaccinating individuals can account for the relatively high vaccination coverages observed in developed countries. We explain why and how retaliation concepts usually found in repeated games can be used in the context of vaccination, even though individuals vaccinate only once. This allows us to model how vaccinating individuals might retaliate against those who refuse vaccination. We show that retaliation threats can sustain vaccination where it would otherwise be suboptimal for individuals.

L’épizootie de fièvre aphteuse de 2001 telle qu’elle s’est déroulée en Grande-Bretagne a montré les limites de certains des systèmes de surveillance, d’intervention, d’anticipation et d’aide à la décision actuellement en place. Parallèlement, l’assouplissement des modalités d’utilisation de la vaccination a enrichi la palette des outils utilisables par le gestionnaire de risque. Une réévaluation pour la France a semblé ainsi opportune. Pour cela, le projet de recherche s’est proposé tout d’abord d’étudier l’organisation des différentes filières d’animaux de rente en France. L’analyse du réseau d’élevages français par la méthode d’analyse des réseaux sociaux a permis de caractériser les relations entre établissements et d’estimer une vulnérabilité structurale constante face à la diffusion de maladies apportées essentiellement par le réseau d’échanges bovins.Dans une deuxième partie, la diffusion de la maladie dans ce réseau d’élevages a été étudiée en s’appuyant sur des simulations générées par un modèle spécifiquement construit. L’efficacité et les coûts de différentes stratégies de lutte envisageables (combinaisons d’abattage sanitaire, abattage préventif et vaccination) ont été ensuite comparés. Les résultats obtenus ont montré l’absence d’une stratégie unique optimale à la fois sur le plan épidémiologique et économique et la nécessité de s’adapter au contexte local. En effet, la nature de la stratégie optimale varie selon la région concernée et l’acteur considéré. Ces travaux ont permis de créer un outil de réflexion pour aider à la décision, mais également un outil d’entraînement et de formation collective pour se préparer à une telle gestion de crise. / The 2001 Foot-and-mouth disease epizootic in the United Kingdom has illustrated limits in some established surveillance, control, anticipation and decision-making systems. At the same time, the relaxation of vaccination procedures has enhanced the set of usable tools by risk managers. A reassessment in France seemed needed.For this purpose, the research project firstly proposed to study the organization of the different livestock industries in France. The analysis of French holding network by using the method of social network analysis has led to characterize the relationships between holdings and to determine a constant structural vulnerability to diseases spread especially due to bovine trade network.Secondly, the disease spread in the holding network has been studied with simulations generated by a specifically built model. The cost-effectiveness of different control strategies (combining infected herd depopulation, pre-emptive culling and vaccination) were then compared. Results showed that no single epidemiological and economical optimal strategy existed and local context must be taken into account. Indeed, the cost-effective strategy changed according to the concerned area and the considered players.This research work allowed creating a thinking tool for decision-making but also a training tool for crisis management preparation.

<p>Samhället är mer eller mindre konstant utsatt för risker. De flesta av dem är så små, och har kanske till och med så pass obetydande effekter för individen om de realiseras, att vi negligerar dem i våra dagliga beslut. Vid vissa tillfällen står vid dock inför beslut som är betydligt svårare, t.ex. i vilken mån vi ska handla kostsamma försäkringar för stora olyckor som högst osannolikt inträffar men skulle få katastrofala konsekvenser. Liknande situationer uppstår för centrala beslutsfattare där man fattar beslut att genomföra riskreducerande åtgärder i samband med potentiella katastrofer. Sådana åtgärder finansieras genom skattemedel och kanske kan man då kräva att myndigheter åtminstone försöker redovisa de förväntade kostnaderna och fördelarna som följs av besluten. För detta ändamål kan man använda kostnads-nyttoanalys, vilket vanligtvis rekommenderas för riskhantering inom t.ex. vägtrafik, eller i viss mån den mindre avancerade kostnads-effektanalysen. Riskanalysens utformning är bl.a. beroende av om man hanterar risk (man kan tilldela utfallen sannolikheter) eller osäkerhet (man känner till utfallen men kan inte tilldela dem sannolikheter) och begränsas vanligtvis av en mängd ovissheter, såsom värden för olika ingående parametrar (t.ex. värdet på ett statistiskt liv).</p><p>Detta arbete tar upp teori angående värdering ex ante av kostnader i samband med potentiella katastrofer och hur dessa värden kan användas under risk eller osäkerhet i en kvantitativ riskanalys. Med hjälp av det försöker jag beskriva i stora drag hur man skulle kunna värdera den ekonomiska nyttan av riskreducerande åtgärder i samband med den s.k. fågelinfluensan. Mina slutsatser är att en probabilistisk ansats och en kvantitativ riskanalys försvåras av bl.a. ett knapphändigt material för sannolikhetstolkningar, förekomsten av kollektiv risk och svårigheter i att värdera t.ex. statistiska liv. Socialstyrelsen är nationell samordnare av arbetet med förebyggande åtgärder i samband med potentiella pandemier men varken använder eller rekommenderar andra myndigheter att använda nyttobaserade beslutskriterier som en del av beslutsunderlaget i samband med den s.k. fågelinfluensan (i detta fall). Kanske borde det tillhöra deras uppgifter att se över möjligheterna att använda samhällsekonomiska utredningar vid sådana situationer. Istället använder de vad som liknar en försiktighetsprincip, vilket också kan försvaras under förutsättningen att riskerna är påtagliga. För stora och långsiktiga beslut borde man istället införa en snävare anti-katastrof-princip innefattande en prioriteringsordning för när man, baserat på om det är möjligt och utredningsekonomiskt motiverat, ska genomföra kostnads-nyttoanalyser och när man istället ska genomföra kostnads-effektanalyser. De kan sedan vara en del av ett större beslutsunderlag.</p> / <p>Our society is constantly exposed to risks. Most of them are so small, and maybe even have so little effect if they occur, that we disregard them in our daily decisions. Quite often, however, we face decisions that are harder for us to make, as whether or not to buy expensive insurance to cover for accidents that have only a small probability to occur but also catastrophic consequences. Similar situations appear for central decision makers when they decide whether or not to reduce a certain catastrophic risk for the society as a whole. Those actions are financed through taxes and therefore it might be fair to say that the authorities should at least try to account for the possible benefits and costs associated with a risk reducing programme. For this you may use Cost-Benefit Analysis (CBA), which is usually recommended for use in, for example, risk management for traffic risks, or the less advanced Cost-Effectiveness Analysis (CEA). The shape and form of the risk analysis is to a great deal dependent of whether you are dealing with risk (it’s possible to assign the predicted states probabilities) or uncertainty (it’s not possible to assign the predicted states probabilities) and is also usually limited by a number of other uncertainties, say in value parameters (e.g. the value of a statistical life, VSL).</p><p>This paper deals with theories regarding how to value (ex ante) the costs of potential catastrophes and how these values could be used in dealing with risk or uncertainty in a quantitative risk analysis. From there I try to describe how, possibly, you could value the benefits from programmes reducing risks of an avian influenza. My conclusions are that a probabilistic approach in a quantitative analysis is complicated by, among other things, the short amount of data necessary to compute objective probabilities, the existence of collective risks and for example difficulties in measuring VSL. The Swedish National Board of Health and Welfare (Socialstyrelsen) is the national coordinator for all work regarding risk reducing actions for potential pandemics, but the board neither uses nor recommends other authorities to use utility-based criteria as a part of their decision material when dealing with, in this case, the bird flue. Quite possibly it should be a part of their work to look into the possibilities of using economic evaluations in these situations. Instead they use what resembles a Precautionary Principle, which also can be used on justified grounds as long as the risks are highly present. What should be done for large and long term decision making, though, is to define a narrower Anti-Catastrophe Principle containing orders of priorities for when, based on if its possible and financially motivated, to use CBA and when to use CEA. Those analyses could then be used along with any other decision material.</p>

Lévaluation économique en santé a pour objectif dévaluer la rentabilité relative dactions de santé en comparant leurs implications en termes de coûts et leurs résultats. Cette discipline occupe une place croissante dans la prise de décisions de santé publique et est essentielle pour maintenir un système de soins de santé de qualité et accessible à tous. Son rôle devrait se renforcer dans le futur en raison des progrès constants de la médecine, de la limitation des budgets de santé publique et de la régulation accrue du secteur de la santé. Ces dernières années, lostéoporose est devenue un véritable problème de santé publique. Cette maladie chronique se caractérise par une diminution de la masse osseuse et une détérioration de la micro-architecture du tissu osseux, conduisant à une fragilité osseuse accrue et à une augmentation du risque de fracture. Les fractures liées à lostéoporose sont responsables dune morbidité élevée, dun excès de mortalité et elles imposent un poids financier considérable à la société. Les dépenses de santé dévolues à cette maladie sont consacrées principalement à la prévention, au diagnostic et au traitement de lostéoporose et de ses conséquences. Dans un souci de rationaliser les décisions de santé publique, il est indispensable dallouer efficacement les ressources disponibles. Dans cette perspective, lévaluation économique de stratégies de prise en charge de lostéoporose savère un outil de première importance pour aider les preneurs de décisions à optimiser leurs choix. Tel est lobjet de la présente thèse qui se décompose en quatre parties. La première partie se consacre au développement et à la validation dun modèle innovant de microsimulation de Markov. Cette technique de modélisation permet de représenter la complexité de lostéoporose avec un niveau élevé de précision, de manière à garantir la fiabilité des résultats. La deuxième partie a pour objectif dévaluer la rentabilité économique de stratégies de dépistage de lostéoporose et de fournir un ensemble de recommandations pour accroître leur efficience. La troisième partie sintéresse à lévaluation économique de nouveaux traitements de lostéoporose (le ranélate de strontium et le dénosumab). Il est important dévaluer si les améliorations de santé, générées par ces traitements, sont obtenues à un coût raisonnable. La quatrième partie étudie spécifiquement les implications cliniques et économiques de la faible adhérence des patients aux traitements actuels contre lostéoporose. Les informations contenues dans cette thèse ont pour objectif daider les preneurs de décisions à allouer efficacement les ressources financières consacrées à lostéoporose. Elles peuvent également savérer utiles pour sensibiliser la population générale, le corps médical et les autorités de santé publique à limportance du dépistage de lostéoporose et dun suivi adéquat ainsi quaux rôles et aux enjeux de lévaluation économique dans le processus décisionnel dans le secteur de la santé. Health economic evaluation is a method for evaluating the value for money of medical technologies by comparing alternative options in terms of their costs and consequences. This method is frequently used to inform decision makers about how to allocate scarce resources. Economic evaluation has become increasingly important nowadays, with the increasing health care expenditures, the rapid introduction of new medical technologies, and the extending role of economic evaluations in health care decision making. Osteoporosis becomes an increasingly major health problem around the world. It is a disease characterized by low bone mass with microarchitectural disruption and increased skeletal fragility, leading to increased fracture risk. Osteoporotic fractures result in significant morbidity and mortality, and impose a huge financial burden on health care systems. Financial resources are used to prevent, diagnose and treat osteoporosis. These resources must be allocated in an efficient manner. The general purpose of this thesis is therefore to contribute to the economic evaluation of osteoporosis management. More specifically, this thesis is divided into four parts. The first part is devoted to the development and the validation of a new Markov microsimulation model. This modelling approach accurately encompasses the complexity of osteoporosis, increasing the reliability of the results. The second part estimates the cost-effectiveness of osteoporosis screening strategies and provides recommendations for their implementation. The third part estimates the cost-effectiveness of new anti-osteoporotic therapies (strontium ranelate and denosumab). It is important to assess whether new drugs represent a good value for money compared to the relevant alternatives. Finally, the last part of the thesis investigates the clinical and economic implications of non-adherence to osteoporosis medications. This thesis provides new information on the cost-effectiveness of osteoporosis management, which may be useful to inform decisions makers about resources allocation for patients with osteoporosis. The results may also be used to raise awareness among public health authorities, the medical profession and the general population regarding the burden of osteoporosis, the importance of screening and adequate follow-up, as well as to the role of economic evaluation in health care decision making.

Samhället är mer eller mindre konstant utsatt för risker. De flesta av dem är så små, och har kanske till och med så pass obetydande effekter för individen om de realiseras, att vi negligerar dem i våra dagliga beslut. Vid vissa tillfällen står vid dock inför beslut som är betydligt svårare, t.ex. i vilken mån vi ska handla kostsamma försäkringar för stora olyckor som högst osannolikt inträffar men skulle få katastrofala konsekvenser. Liknande situationer uppstår för centrala beslutsfattare där man fattar beslut att genomföra riskreducerande åtgärder i samband med potentiella katastrofer. Sådana åtgärder finansieras genom skattemedel och kanske kan man då kräva att myndigheter åtminstone försöker redovisa de förväntade kostnaderna och fördelarna som följs av besluten. För detta ändamål kan man använda kostnads-nyttoanalys, vilket vanligtvis rekommenderas för riskhantering inom t.ex. vägtrafik, eller i viss mån den mindre avancerade kostnads-effektanalysen. Riskanalysens utformning är bl.a. beroende av om man hanterar risk (man kan tilldela utfallen sannolikheter) eller osäkerhet (man känner till utfallen men kan inte tilldela dem sannolikheter) och begränsas vanligtvis av en mängd ovissheter, såsom värden för olika ingående parametrar (t.ex. värdet på ett statistiskt liv). Detta arbete tar upp teori angående värdering ex ante av kostnader i samband med potentiella katastrofer och hur dessa värden kan användas under risk eller osäkerhet i en kvantitativ riskanalys. Med hjälp av det försöker jag beskriva i stora drag hur man skulle kunna värdera den ekonomiska nyttan av riskreducerande åtgärder i samband med den s.k. fågelinfluensan. Mina slutsatser är att en probabilistisk ansats och en kvantitativ riskanalys försvåras av bl.a. ett knapphändigt material för sannolikhetstolkningar, förekomsten av kollektiv risk och svårigheter i att värdera t.ex. statistiska liv. Socialstyrelsen är nationell samordnare av arbetet med förebyggande åtgärder i samband med potentiella pandemier men varken använder eller rekommenderar andra myndigheter att använda nyttobaserade beslutskriterier som en del av beslutsunderlaget i samband med den s.k. fågelinfluensan (i detta fall). Kanske borde det tillhöra deras uppgifter att se över möjligheterna att använda samhällsekonomiska utredningar vid sådana situationer. Istället använder de vad som liknar en försiktighetsprincip, vilket också kan försvaras under förutsättningen att riskerna är påtagliga. För stora och långsiktiga beslut borde man istället införa en snävare anti-katastrof-princip innefattande en prioriteringsordning för när man, baserat på om det är möjligt och utredningsekonomiskt motiverat, ska genomföra kostnads-nyttoanalyser och när man istället ska genomföra kostnads-effektanalyser. De kan sedan vara en del av ett större beslutsunderlag. / Our society is constantly exposed to risks. Most of them are so small, and maybe even have so little effect if they occur, that we disregard them in our daily decisions. Quite often, however, we face decisions that are harder for us to make, as whether or not to buy expensive insurance to cover for accidents that have only a small probability to occur but also catastrophic consequences. Similar situations appear for central decision makers when they decide whether or not to reduce a certain catastrophic risk for the society as a whole. Those actions are financed through taxes and therefore it might be fair to say that the authorities should at least try to account for the possible benefits and costs associated with a risk reducing programme. For this you may use Cost-Benefit Analysis (CBA), which is usually recommended for use in, for example, risk management for traffic risks, or the less advanced Cost-Effectiveness Analysis (CEA). The shape and form of the risk analysis is to a great deal dependent of whether you are dealing with risk (it’s possible to assign the predicted states probabilities) or uncertainty (it’s not possible to assign the predicted states probabilities) and is also usually limited by a number of other uncertainties, say in value parameters (e.g. the value of a statistical life, VSL). This paper deals with theories regarding how to value (ex ante) the costs of potential catastrophes and how these values could be used in dealing with risk or uncertainty in a quantitative risk analysis. From there I try to describe how, possibly, you could value the benefits from programmes reducing risks of an avian influenza. My conclusions are that a probabilistic approach in a quantitative analysis is complicated by, among other things, the short amount of data necessary to compute objective probabilities, the existence of collective risks and for example difficulties in measuring VSL. The Swedish National Board of Health and Welfare (Socialstyrelsen) is the national coordinator for all work regarding risk reducing actions for potential pandemics, but the board neither uses nor recommends other authorities to use utility-based criteria as a part of their decision material when dealing with, in this case, the bird flue. Quite possibly it should be a part of their work to look into the possibilities of using economic evaluations in these situations. Instead they use what resembles a Precautionary Principle, which also can be used on justified grounds as long as the risks are highly present. What should be done for large and long term decision making, though, is to define a narrower Anti-Catastrophe Principle containing orders of priorities for when, based on if its possible and financially motivated, to use CBA and when to use CEA. Those analyses could then be used along with any other decision material.

Prieš priimant sprendimą ar verta medicininę naujovę taikyti klinikinėje praktikoje tiek medicinine, tiek ekonomine prasme, atliekama gydymo metodų efektyvumo ekonominė analizė. Kaštų efektyvumo analizė – sudėtinė sveikatos ekonomikos dalis. Ji nustato ryšį tarp taikomo gydymo metodo naudojamų išteklių (kaštų) ir teikiamo naudingumo sveikatai (poveikio sveikatai). Šio darbo tikslas: išnagrinėjus ekonominės analizės metodus, pateikti kaštų efektyvumo analizės modelį bei patikrinti jį „N“ sveikatos priežiūros įstaigoje taikomiems gydymo metodams. Suformuluotam tikslui pasiekti keliami uždaviniai: 1. Remiantis mokslinės literatūros analize, pateikti kaštų efektyvumo analizės kaip sudėtinės sveikatos ekonomikos dalies charakteristiką; 2. Išnagrinėti kaštų efektyvumo analizės taikymą šiandieninėse Lietuvos sveikatos priežiūros įstaigose; 3. Pasiūlyti kaštų efektyvumo analizės modelį; 4. Atlikti „N“ sveikatos priežiūros įstaigoje taikomų gydymo metodų kaštų efektyvumo analizę. Tyrimo objektas: sveikatos priežiūros įstaigos kaštai. Išanalizavus ekonominės analizės taikymo galimybes, atlikta kaštų efektyvumo analizė „N“ sveikatos priežiūros įstaigoje. Taikyti du ekonominės analizės metodai: kaštų naudingumo ir kaštų efektyvumo analizės metodai. Analizės metu įvertinti gydymo metodų kaštai ir klinikinis poveikis sveikatai. Tyrime analizuojami trys gydymo metodai: standartinis gydymo metodas – judesio terapija salėje bei du alternatyvūs gydymo metodai – judesio terapija vandenyje ir... [toliau žr. visą tekstą] / Before making a decision to employ an innovation in medicine on both medical and economical dimensions, performed economic analysis of effectiveness of health interventions. Cost effectiveness analysis is constituent of health economic. It estimates the relation between cost ant effectiveness of health interventions. The aim of this research: after study of methods of economic evaluation, to supply the model of cost effectiveness analysis and to test it for health interventions of “N” health care institution. The tasks to reach the established aim of this study are: 1. By analysis of literature, to supply the reference of cost effectiveness analysis as constituent of health economic; 2. To study the use of cost effectiveness analysis in nowadays Lithuanian’s health care institutions; 3. To offer the model of cost effectiveness analysis; 4. To perform the cost effectiveness analysis of health interventions of “N” health care institution. The object of research is: cost of health care institution. Having studied the opportunities of economic analysis, performed the cost effectiveness analysis in “N” health care institution. There were used two methods of economic analysis: cost-effectiveness and cost-utility methods. On the analysis estimated cost and effectiveness of health interventions. In the research analyzed three health interventions: standard health intervention – physiotherapy in exercises room and two alternatives health interventions – physiotherapy in the water and... [to full text]

The partial removal of carious dentine was studied by means of a literature review and a multicenter randomized controlled clinical trial. The clinical trial compared the effectiveness of an alternative treatment for deep caries lesions and the stepwise excavation in Public Health Services in Brazil. The treatment consists of partial removal of carious dentine followed by restoration in one session. A cost-effectiveness analysis the two treatments was performed. The clinical performance of amalgam and resin restorations placed in deep caries lesions with or without decayed tissue beneath them was also evaluated. Inclusion criteria: patients with ≥ nine years old, permanent molars with deep caries lesions and absence of periapical alterations, pulp sensitivity; absence of spontaneous pain; negative percussion test. The subjects were assigned to: test-group - partial removal of carious dentine (PDR) and restoration, and control-group - stepwise excavation (SW). SW consists of partial removal of carious tissue, indirect pulp capping with calcium hydroxide cement; temporary filling; cavity re-opening after 60 days, removal of the remaining soft carious tissue and filling. Clinical and radiological exams were performed annually. The outcomes were: (1) pulp sensitivity to cold test and absence of periapical alterations, assuming those parameters as indicators of pulp vitality; and (2) success of the restoration. To determine the cost-effectiveness of the treatments, the discounted cash flow method was adopted. The data were submitted to Kaplan-Meier, Log-rank test and logistic regression analysis, P<0.05. There were performed 299 treatments, 146 SW and 153 tests. There were no differences between the groups regarding baseline characteristics - age, gender and family income. The number of teeth evaluated after one and two years were 180 and 122. After one year of treatment performance, the therapy success rates were 97.9% and 74.1% of success in test and control groups respectively (P<0.000). After 2 years of follow-up, therapy survival rates of PDR and SW were 93.7% and 73.3% respectively (P=0.000). A total of 29 therapeutic failures were observed: PDR group - pulpitis (n=3), osteitis (n=1), hyperemia (n=2); SW group - pulpitis (n=15), necrosis (n=6), extraction (n=1) and restoration fracture (n=1). None of the variables studied showed a significant causal influence on the success rate, besides the type of treatment. After two year of follow-up, 181 restorations had been evaluated, 86 from the SW group and 95 from the PDR group; 65.8% were from the resin composite group and 34.2% were amalgam restorations. The survival analysis of the treatment associated with the filling material showed no difference in the rate of success (P=0.564). Regarding the treatment, both groups presented a similar rate of success: SW=95.3% and PDR=94.7% (P=0.928). Resin composite restorations presented 96.8% of success and amalgam restorations presented 94.1% of success (P=0.446). The reason for failure was fracture of filling material. The PDR provides an economy of R$ 143.37 (67.78%) per treatment compare to SW and 2.39% in the overall economy in the annual cost of the public health center. Partial caries removal could be performed as definitive treatment and the procedure of re-opening the cavity to remove the residual infected dentine is not necessary. The maintenance of carious dentine does not interfere in the maintenance of pulp vitality. The presence of decayed tissue in deep caries lesions does not seem to interfere with the survival of the restorations. Performing the partial removal of carious dentine in one session generates benefits for the public finances (direct economy), for the public health services (increase in number of treatments performed) and for the patients (comfort and time).

The partial removal of carious dentine was studied by means of a literature review and a multicenter randomized controlled clinical trial. The clinical trial compared the effectiveness of an alternative treatment for deep caries lesions and the stepwise excavation in Public Health Services in Brazil. The treatment consists of partial removal of carious dentine followed by restoration in one session. A cost-effectiveness analysis the two treatments was performed. The clinical performance of amalgam and resin restorations placed in deep caries lesions with or without decayed tissue beneath them was also evaluated. Inclusion criteria: patients with ≥ nine years old, permanent molars with deep caries lesions and absence of periapical alterations, pulp sensitivity; absence of spontaneous pain; negative percussion test. The subjects were assigned to: test-group - partial removal of carious dentine (PDR) and restoration, and control-group - stepwise excavation (SW). SW consists of partial removal of carious tissue, indirect pulp capping with calcium hydroxide cement; temporary filling; cavity re-opening after 60 days, removal of the remaining soft carious tissue and filling. Clinical and radiological exams were performed annually. The outcomes were: (1) pulp sensitivity to cold test and absence of periapical alterations, assuming those parameters as indicators of pulp vitality; and (2) success of the restoration. To determine the cost-effectiveness of the treatments, the discounted cash flow method was adopted. The data were submitted to Kaplan-Meier, Log-rank test and logistic regression analysis, P<0.05. There were performed 299 treatments, 146 SW and 153 tests. There were no differences between the groups regarding baseline characteristics - age, gender and family income. The number of teeth evaluated after one and two years were 180 and 122. After one year of treatment performance, the therapy success rates were 97.9% and 74.1% of success in test and control groups respectively (P<0.000). After 2 years of follow-up, therapy survival rates of PDR and SW were 93.7% and 73.3% respectively (P=0.000). A total of 29 therapeutic failures were observed: PDR group - pulpitis (n=3), osteitis (n=1), hyperemia (n=2); SW group - pulpitis (n=15), necrosis (n=6), extraction (n=1) and restoration fracture (n=1). None of the variables studied showed a significant causal influence on the success rate, besides the type of treatment. After two year of follow-up, 181 restorations had been evaluated, 86 from the SW group and 95 from the PDR group; 65.8% were from the resin composite group and 34.2% were amalgam restorations. The survival analysis of the treatment associated with the filling material showed no difference in the rate of success (P=0.564). Regarding the treatment, both groups presented a similar rate of success: SW=95.3% and PDR=94.7% (P=0.928). Resin composite restorations presented 96.8% of success and amalgam restorations presented 94.1% of success (P=0.446). The reason for failure was fracture of filling material. The PDR provides an economy of R$ 143.37 (67.78%) per treatment compare to SW and 2.39% in the overall economy in the annual cost of the public health center. Partial caries removal could be performed as definitive treatment and the procedure of re-opening the cavity to remove the residual infected dentine is not necessary. The maintenance of carious dentine does not interfere in the maintenance of pulp vitality. The presence of decayed tissue in deep caries lesions does not seem to interfere with the survival of the restorations. Performing the partial removal of carious dentine in one session generates benefits for the public finances (direct economy), for the public health services (increase in number of treatments performed) and for the patients (comfort and time).

Les études observationnelles (non-randomisées) sont principalement constituées de données ayant des particularités qui sont en fait contraignantes dans un cadre statistique classique. En effet, dans ce type d'études, les données sont rarement continues, complètes et indépendantes du bras thérapeutique dans lequel les observations se situent. Cette thèse aborde l'utilisation d'un outil statistique paramétrique fondé sur la dépendance entre les données à travers plusieurs scénarios liés aux études observationnelles. En effet, grâce au théorème de Sklar (1959), les copules paramétriques sont devenues un sujet d'actualité en biostatistique. Pour commencer, nous présentons les concepts de base relatifs aux copules et aux principales mesures d'association basées sur la concordance retrouvées dans la littérature. Ensuite, nous donnons trois exemples d'application des modèles de copules paramétriques pour autant de cas de données particulières retrouvées dans des études observationnelles. Nous proposons d’abord une stratégie de modélisation de l'analyse coût-efficacité basée uniquement sur une réécriture des fonctions de distribution jointes et évitant les modèles de régression linéaire. Nous étudions ensuite, les contraintes relatives aux données discrètes, particulièrement dans un contexte de non-unicité de la fonction copule, nous réécrivons le score de propension grâce à une approche novatrice basée sur l'extension d'une sous-copule. Enfin, nous évoquons un type particulier de données manquantes : les données censurées à droite, dans un contexte de régression, grâce à l'utilisation de copules semi-paramétriques. / Observational studies (non-randomized) consist primarily of data with features that are in fact constraining within a classical statistical framework. Indeed, in this type of study, data are rarely continuous, complete, and independent of the therapeutic arm the observations are belonging to. This thesis deals with the use of a parametric statistical tool based on the dependence between the data, using several scenarios related to observational studies. Indeed, thanks to the theorem of Sklar (1959), parametric copulas have become a topic of interest in biostatistics. To begin with, we present the basic concepts of copulas, as well as the main measures of association based on the concordance founded on an analysis of the literature. Then, we give three examples of application of models of parametric copulas for as many cases of specific data found in observational studies. We first propose a strategy of modeling cost-effectiveness analysis based essentially on rewriting the joint distribution functions, while discarding the use of linear regression models. We then study the constraints relative to discrete data, particularly in a context of non-unicity of the copula function. We rewrite the propensity score, thanks to an innovative approach based on the extension of a sub-copula. Finally, we introduce a particular type of missing data: right censored data, in a regression context, through the use of semi-parametric copulas.

Introdução: O estudo MASS III não mostrou diferença significativa entre a cirurgia de revascularização miocárdica (CRM) com e sem circulação extracorpórea (CEC) em relação ao desfecho composto primário de morte por todas as causas, infarto agudo do miocárdio, acidente vascular cerebral ou revascularização adicional em pacientes com doença coronariana multiarterial. No entanto, a custo-efetividade dessas estratégias permanece desconhecida. Métodos: Pacientes com doença coronariana multiarterial estável e função ventricular esquerda preservada foram randomizados para CRM com CEC (n=153) ou sem CEC (n=155). Os dois grupos eram bem semelhantes quanto às características basais. A análise dos custos foi realizada a partir da perspectiva do sistema público de saúde brasileiro, e as utilities foram avaliadas pelo questionário SF-6D. Um modelo de Markov, com base nos dados de 5 anos de seguimento, foi utilizado para extrapolar os custos e os anos de vida ajustados pela qualidade (QALY) para doença coronariana crônica. Resultados: A qualidade de vida de ambos os grupos melhorou significativamente após a cirurgia durante o seguimento, em comparação com os dados pré-cirurgia, embora os ganhos de vida adquiridos (LYG) e QALYs tenham sido semelhantes entre os grupos durante o seguimento de 5 anos. Os custos para o período total do estudo não diferiram entre os grupos sem e com CEC (R$ 19.180,65 e R$ 19.909,18, respectivamente, p=0,409). Ao longo de um horizonte de tempo ajustado para a expectativa de vida da população do estudo, a razão de custo-efetividade incremental da CRM com versus sem CEC foi R$ 45.274 por QALY ganho, que foi robusto nas simulações de Monte Carlo e nas análises de sensibilidade. Para um limiar de custo-efetividade de R$ 34.212 por QALY ganho, a CRM sem CEC tem 65% de probabilidade de ser custo-efetiva quando comparada com CRM com CEC. Conclusão: Cirurgia de revascularização miocárdica sem CEC é clinicamente tão segura e efetiva quanto a cirurgia com CEC e parece ser uma estratégia economicamente atraente em comparação com a CRM com CEC em pacientes com doença arterial coronariana estável / Background: The MASS III trial revealed that in patients with multivessel coronary disease, no significant difference was observed between on-pump and off-pump coronary artery bypass surgery (CABG) in the primary composite outcome. However, long-term cost-effectiveness of these strategies is unknown. Methods: Patients with stable multivessel coronary artery disease and preserved left ventricular function were randomized to onpump (n=153) or off-pump CABG (n=155). The 2 groups were well matched for baseline characteristics. Costs analysis was conducted from a Brazilian public healthcare system perspective, and health state utilities were assessed using the SF-6D questionnaire. A Markov\'s model based on the 5- year in-trial data was used to extrapolate costs and quality-adjusted life-years (QALY) for chronic coronary disease. Results: Both groups\' quality of life improved significantly after surgery during follow-up compared with baseline, and life-years gained (LYG) and QALY gains were similar between on-pump and off-pump CABG over the 5-year time frame of the trial. The costs for the overall period of the trial - the mean cost in U.S. dollars per patient - did not differ significantly between the off-pump group and the on-pump group ($5674.75 and $5890.29 respectively, p=0.409). Over a lifetime horizon, the incremental cost-effectiveness ratio of on-pump vs. off-pump CABG was $12,576 per QALY gained, which was robust in Monte Carlo replications and in sensitivity analyses. Using a cost-effectiveness threshold of $10,122 per QALY gained, off-pump has 65% probability of being cost-effective versus on-pump CABG. Conclusions: Off-pump CABG was clinically as safe and effective as on-pump CABG and appears to be an economically attractive strategy compared with on-pump CABG among patients with stable coronary artery disease