No description available.

No description available.

In public health, biomedical, epidemiological, and other applications, data collected are often measured with error. When mismeasured data is used in a regression analysis, not accounting for the measurement error can lead to incorrect inference about the relationships between the covariates and the response. We investigate measurement error in the covariates of two types of regression models.  For each we propose a fully Bayesian approach that treats the variable measured with error as a latent variable to be integrated over, and a semi-Bayesian approach which uses a first order Laplace approximation to marginalize the variable measured with error out of the likelihood. The first model is the matched case-control study for analyzing clustered binary outcomes. We develop low-rank thin plate splines for the case where a variable measured with error has an unknown, nonlinear relationship with the response. In addition to the semi- and fully Bayesian approaches, we propose another using expectation-maximization to detect both parametric and nonparametric relationships between the covariates and the binary outcome. We assess the performance of each method via simulation terms of mean squared error and mean bias. We illustrate each method on a perturbed example of 1--4 matched case-control study. The second regression model is the generalized linear model (GLM) with unknown link function. Usually, the link function is chosen by the user based on the distribution of the response variable, often to be the canonical link. However, when covariates are measured with error, incorrect inference as a result of the error can be compounded by incorrect choice of link function. We assess performance via simulation of the semi- and fully Bayesian methods in terms of mean squared error. We illustrate each method on the Framingham Heart Study dataset. The simulation results for both regression models support that the fully Bayesian approach is at least as good as the semi-Bayesian approach for adjusting for measurement error, particularly when the distribution of the variable of measure with error and the distribution of the measurement error are misspecified. / Ph. D.

An important prerequisite for running a successful insurance business is to predict risk. By forecasting the future in as much detail as possible, competitive advantages are created in terms of price differentiation. This work aims at using spatial point processes to provide a proposal for how the geographical position of the customer can be used in developing risk differentiation tools. For spatial variation in claim frequency an approach is presented which is common in spatial epidemiology by considering a group of policyholders, with and without claims, as a realisation of a multivariate Poisson point process in two dimensions. Claim costs are then included by considering the claims as a realisation of a point process with continuous marks. To describe the spatial variation in relative risk, demographic and socio-economic information from Swedish agencies have been used. The insurance data that have been used come from the insurance company If Skadeförsäkring AB, where also the work has been carried out. The result demonstrates problems with parametric modelling of the intensity of policyholders, which makes it difficult to validate the spatial varying intensity of claim frequency. Therefore different proposals of non-parametric estimation are discussed. Further, there are no tendencies that the selected information is able to explain the variation in claim costs. / En viktig förutsättning för att kunna bedriva en framgångsrik försäkringsverksamhet är att prediktera risk. Genom att på en så detaljerad nivå som möjligt kunna förutse framtiden skapas konkurrensfördelar i form av prisdifferentiering. Målet med detta arbete är att med hjälp av spatiala punktprocesser ge ett förslag på hur kunders geografiska position kan utvecklas som riskdifferentieringsverktyg. För spatial variation i skadefrekvens presenteras ett tillvägagångssätt som är vanligt inom spatial epidemiologi genom att betrakta en grupp försäkringstagare, med och utan skador, som en realisering av en multivariat Poissonprocess i två dimensioner. Skadekostnaderna inkluderas sedan genom att betrakta skadorna som en punktprocess med kontinuerliga märken. För att beskriva spatial variation i relativ risk används demografisk och socioekonomisk information från svenska myndigheter. De försäkringsdata som använts kommer från If Skadeförsäkring AB, där också arbetet har utförts. Resultatet påvisar problem med att parametriskt modellera intensiteten för försäkringstagare, vilket medför svårigheter att validera den skattade spatiala variationen i skadefrekvens, varför olika ickeparametriska förslag diskuteras. Vidare upptäcktes inga tendenser till att variationen i skadekostnad kan förklaras med den utvalda informationen.

Yes / Adults with amblyopia ('lazy eye'), long-standing strabismus (ocular misalignment) or both typically do not experience visual symptoms because the signal from weaker eye is given less weight than the signal from its fellow. Here we examine the contribution of the weaker eye of individuals with strabismus and amblyopia with both eyes open and with the deviating eye in its anomalous motor position. The task consisted of a blue-on-yellow detection task along a horizontal line across the central 50 degrees of the visual field. We compare the results obtained in ten individuals with strabismic amblyopia with ten visual normals. At each field location in each participant, we examined how the sensitivity exhibited under binocular conditions compared with sensitivity from four predictions, (i) a model of binocular summation, (ii) the average of the monocular sensitivities, (iii) dominant-eye sensitivity or (iv) non-dominant-eye sensitivity. The proportion of field locations for which the binocular summation model provided the best description of binocular sensitivity was similar in normals (50.6%) and amblyopes (48.2%). Average monocular sensitivity matched binocular sensitivity in 14.1% of amblyopes' field locations compared to 8.8% of normals'. Dominant-eye sensitivity explained sensitivity at 27.1% of field locations in amblyopes but 21.2% in normals. Non-dominant-eye sensitivity explained sensitivity at 10.6% of field locations in amblyopes but 19.4% in normals. Binocular summation provided the best description of the sensitivity profile in 6/10 amblyopes compared to 7/10 of normals. In three amblyopes, dominant-eye sensitivity most closely reflected binocular sensitivity (compared to two normals) and in the remaining amblyope, binocular sensitivity approximated to an average of the monocular sensitivities. Our results suggest a strong positive contribution in habitual viewing from the non-dominant eye in strabismic amblyopes. This is consistent with evidence from other sources that binocular mechanisms are frequently intact in strabismic and amblyopic individuals.

Background Self-harm is a major and growing public health issue among young people worldwide. Self-harm is an important risk factor for suicide, which is one of the leading causes of death for young people. Although suicide rates are declining overall, this trend is not seen in young people. Young people with mental distress and/or suicidal thoughts are reluctant to seek help, and often drop out of treatment initiated after a self-harm episode. Many young people who self-harm have had contact with healthcare before their first self-harm episode, but often for reasons other than suicidal thoughts or psychiatric problems. In this context, physical illness is associated with increased risk for self-harm and suicide among young people. The present thesis investigated how young people perceived the help and support they received before and after an episode of self-harm. A further aim was to map the inpatient somatic healthcare contacts young patients had before an episode of self-harm, and determine any relationship to risk for self-harm and suicide. Method Four studies were conducted using qualitative and quantitative methods. Participants were people aged 16-24 years. The definition of self-harm was based on the intentional self-harm criteria in the International Classification of Diseases, tenth revision, which includes all forms of self-harm without ascribing suicidal intent. In the first two studies, 10 respective 9 participants with a first healthcare contact for self-harm were interviewed during 2009-2011. The interviews covered participants’ knowledge and experience of professional care before their healthcare contact for self-harm. Participants were interviewed a second time 6 months later about their experiences with professional care during the period since their initial interview. Qualitative content analysis was used for all interviews. For the next two studies, we selected 16,235 participants with a first hospitalisation for self-harm during 1999-2009 from the Swedish National Inpatient Register. These cases were compared with matched controls to determine the odds of having been admitted with a non-psychiatric diagnosis during the year preceding the self-harm admission. To assess risk for suicide, data were retrieved from the Swedish Cause of Death Register for all deceased participants until 2013, and group differences were determined using survival analysis. Results In the first interview, participants described how they wanted more information on where they could turn for professional help. They also wanted different help-seeking pathways and emphasised the importance of the quality of professional contact. After 6 months, participants stressed the importance of being able to rely on professionals and treatment. Their life circumstances significantly affected their treatment, and practical help was appreciated. The register studies showed that young people admitted for self-harm were more likely to have been hospitalised with symptomatic diagnoses such as abdominal pain and syncope/collapse, and somatic illnesses such as epilepsy and diabetes mellitus type 1. A higher proportion of cases (4.5%; women 2.6%, men 8.8%) died during the study period than controls (0.3%; women 0.2%, men 0.6%) (p<0.001). For both cases and controls, a higher proportion of those with a previous somatic admission died from suicide during the study period than those without a somatic admission (cases: 4.2% vs. 2.8%, p<0.05). For cases with a somatic admission, the hazard ratio was 1.43 (95% confidence interval 1.04-1.98) compared with those without somatic admissions (controlled for age, sex and psychiatric admission). Survival of cases with a previous somatic admission compared with those without was 98.4% versus 99.2% after the first year, 97.8% versus 98.9% after the second year, and 95.5% versus 96.9% after the tenth year. Conclusion These findings suggest that healthcare providers need to find new ways to reach young people at risk for suicidal behaviour. Access to professional help should be easy and direct. Treatment for young people after self-harm should be flexible, and be receptive to input from the patient. The importance of and need for basic practical help should not be overlooked. Somatic healthcare contact provides an opportunity for intervention, particularly as psychiatric problems can manifest as physical symptoms, and physical illness is a risk factor for self-harm and suicide.

Le cancer du col représente un problème de santé publique au Marco. L’objectif principal de cette thèse est d’améliorer les connaissances sur le cancer du col de l’utérus au Maroc en terme d’épidémiologie descriptive et analytique notamment par l’étude des principaux facteurs associés à ce cancer, son retard de diagnostic et sa prise en charge.Méthodologie Pour répondre à notre objectif, trois études ont été menées. La première étude, rétrospective sur dossier d’hospitalisation, sur les stades de diagnostic et les résultats des traitements du cancer du col. La deuxième étude, transversale, sur les facteurs liés au retard de diagnostic du cancer du col et la troisième étude, cas-témoins multicentrique avec appariement individuel sur l’âge, sur les facteurs de risque du cancer du col de l’utérus. Résultats La première étude a montré la vulnérabilité socioéconomique de la population des femmes atteintes de cancer du col et a mis le point sur les problèmes du retard de diagnostic et de l’accessibilité aux établissements de soins auxquels elles sont confrontées. L’étude a mis l’accent sur la durée du suivi insuffisante et le nombre élevé des perdues de vue. La deuxième étude, a montré que les femmes âgées de moins de 50 ans, analphabètes, non mariées, les femmes au foyer, les femmes résidant en milieu rural, les femmes résidant à plus de 100Km du lieu du diagnostic du cancer, les femmes sans antécédents de cancer du col et les femmes n’ayant pas eu comme premier symptôme des saignements gynécologiques étaient plus à risque du retard de diagnostic du cancer du col. L’étude cas-témoins a confirmé que l’infection à HPV est le facteur de risque majeur du cancer du col. Les autres facteurs de risque du cancer invasif du col utérin étaient : la parité élevée, le faible niveau d'éducation, l’existence de multiples partenaires sexuelles chez le mari, les rapports sexuels pendant les menstruations et les antécédents d’IST. La limite majeure de notre étude cas-témoins est liée au manque de puissance statistique pour l’étude de l’association avec d’autres facteurs tels que les facteurs nutritionnels. Conclusion Les études réalisées dans le cadre de notre thèse, en dépit de leurs limites, constituent une source d’information pour la recherche scientifique sur le cancer du col au Maroc et dans les pays similaires. Elles contribuent à l’amélioration des connaissances sur le cancer du col au Maroc et par là des attitudes et pratiques des différents corps sanitaires vis-à-vis de ce cancer. Elles ont permis de définir les populations à risque de cancer du col et à risque de retard de diagnostic et, ainsi de mieux les cibler dans les programmes de dépistage d’information et d’éducation. Les résultats de nos travaux de recherche pourraient représenter une « situation de référence » utile pour tout programme d’évaluation concernant le cancer du col au Maroc. Ceci devrait permettre dans le futur de mieux évaluer à long terme l’efficacité des programmes de dépistage, de vaccination, des traitements et de l’éducation pour la santé. / The main objective of this work was to improve knowledge about cervical cancer in Morocco in terms of descriptive and analytical epidemiology including the study of the main factors associated with cancer, the delay in diagnosis and its management. Methods : To meet our objectives, we conducted three studies. The first (Study of diagnosis stages and treatment outcomes for cervical cancer), is a retrospective study using data from hospital records of patients with cervical cancer in the major cancer centers in Morocco. The second study (Study of factors related to delay in diagnosis of cervical cancer), transversal, with recruitment of cervical cancer patients. The third study (Study of risk factors for cervical cancer) was a multicenter case-control study with individual matching on age. Results : The first study showed the socio-economic vulnerability of the population of women with cervical cancer and the problems of delayed diagnosis and access to health facilities. In addition, this study focused on the inadequate length of the follow-up and the high number of lost to following-up. The second study showed that women aged less than 50 years, illiterate, unmarried, women at home, women living in rural areas, women living more than 100km from the place of cancer diagnosis, women without history of cervical cancer and women who have not had, as the first symptom, a gynecological bleeding were more at risk of delayed diagnosis of cervical cancer. The case-control study confirmed that HPV infection is the major risk factor for cervical cancer. The other risk factors for invasive cervical cancer identified in our study were: the high parity, the low educational level, husband's multiple sexual partners, sex during menstruation and the history of venereal diseases. Conclusion : The studies conducted in the context of our thesis, inspite of their limitations, are considered as a rich source of information for scientific research on cervical cancer in Morocco and in similar countries. They have improved the knowledge on cervical cancer in Morocco and thus should improve the knowledge, attitudes and practice of different health practitioner’s for cervical cancer. They helped to define population at risk of cervical cancer and of delayed diagnosis. Therefore, it allows them to be better targeted by the screening, information and education programs. The results of our research also represent a "reference situation" for any evaluation program for cervical cancer in Morocco. This should help, in the future, to better assess the long-term effectiveness of screening programs, vaccination, treatment and health education.

Le rôle de l'environnement dans l'étiologie des leucémies aigües de l’enfant (LA) fait aujourd'hui l'objet de recherches intenses. Dans ce contexte, le présent travail a pour objectif d’étudier la relation entre l’incidence de LA et la proximité des centrales nucléaires de production d’électricité (CNPE) et des lignes à haute tension (LHT). Avant cette analyse fine, un premier travail a consisté à étudier les variations départementales de l'incidence de LA.Les cas inclus dans ces études sont toutes les LA du Registre National des Hémopathies malignes de l’Enfant sur la période étudiée : 1990-2004 pour l’étude de l’incidence départementale et 2002-2007 pour les études de l’association avec les facteurs d’exposition environnementale. Dans l’approche cas-témoins principalement utilisée pour ces dernières, les 30 000 sujets témoins constitue un échantillon représentatif de la population pédiatrique française sur la période d’’intérêt. D’autre part, la géolocalisation des adresses des sujets et des sources d'exposition permet de définir des critères de proximité en relation avec la probabilité et/ou l'intensité d'exposition aux facteurs d'intérêt. • L’étude des LA par département n’a pas mis en évidence de tendance ni de structure spatiale dans l’incidence à ce niveau géographique : que ce soit globalement, par classe d’âge, par sexe ou par sous-type de leucémie. • Sur la période 2002-2007 contrairement aux périodes précédentes, un quasi-doublement de l’incidence des LA à moins de 5 km des CNPE a été mis en évidence, avec une approche cas-témoin comme avec l’étude d’incidence. Ce résultat n’était pas spécifique d’une CNPE ou d’un type de CNPE et non lié à la cartographie des émissions aériennes de radioactivité par les CNPE. • L’association trouvée entre l’incidence de LA et la proximité aux LHT de plus de 225 kV (<50 m) semble restreinte aux enfants de moins de 5 ans ou n’habitant en milieu urbain ; aucune association n’a été trouvée avec la proximité aux LHT de moins de 150 kV. / The role of the environment in the etiology of childhood acute leukemia (AL) is currently investigated. In this context, the aim of the present work is to study the association between the incidence of AL and the proximity no nuclear power plants (NPP) and to high voltage overhead power lines (HV OLs). At first, the geographical variations of AL have been studied at the Département level.The cases included in the studies are all cases of AL of the French National Registry of Childhood Haemopatopoietic Malignancies on the studied periods: 1990-2004 for the study of incidence on Départements and 2002-2007 for the studies of association between incidence of AL and environmental exposure factors. Concerning those latter studies, a case-control approach has been used. The control sample, representative of the French pediatric population, contains 30,000 subjects and has been drawn by the INSEE. The precise localization of addresses of subjects and of exposure sources in relation with the type of sources is essential to build indicators of exposure reflecting the probability and intensity of exposure. • The study of AL by Département has highlighted neither trend nor spatial structure in the incidence at this geographical level globally as well as by age, gender and subtype of leukemia.• On 2002-2007, on the contrary of on previous periods, the incidence of AL at less than 5 km from a NPP was nearly twice higher than expected, with the case-control study as well as with the incidence approach. This result was not specific to any age group, NPP, a type of NPP and was not associated with the geographic zoning of gaseous discharges of NPPs. • The study of the proximity to HV OLs highlighted an association between the incidence of AL and the close proximity (< 50 m) of lines of more than 225 kV, association which was restricted to children of less than 5 y.o. or living in non-urban areas; but not with the proximity to lines of less than 150 kV.

A dengue é um importante problema de saúde pública, responsável por cerca de 25.000 mortes anuais em áreas subtropicais do mundo. Desde 2001, há uma tendência de aumento da incidência de formas fatais de febre hemorrágica da dengue (FHD) no Brasil, com aumento dramático de casos graves em menores de 15 anos de idade a partir de 2007, especialmente na região nordeste do país. O objetivo deste estudo caso-controle foi avaliar fatores de risco associados a óbito em crianças com dengue grave. Avaliamos a condição clínica de pacientes internados que morreram de dengue (n=18) e comparamos com pacientes internados com dengue grave que sobreviveram (controles, n=77). Os pacientes incluídos no estudo foram menores de 13 anos de idade internados em hospitais de São Luís, nordeste do Brasil, com diagnóstico laboratorial confirmado de dengue. O diagnóstico de infecção aguda de dengue foi confirmado pela detecção de anticorpos IgM específicos de dengue através do MAC-ELISA (IgM Antibody Capture Enzyme-Linked Immunosorbent Assay) ou pela detecção do DENV em soro, sangue ou víscera pela técnica de Transcrição Reversa - Reação em Cadeia de Polimerase (RT-PCR). Sinais de choque descompensado (extremidades frias, cianose e letargia) e hemoptise foram fortemente associados a óbito, o que está de acordo com a mais recente classificação da Organização Mundial de Saúde (OMS) para dengue grave. Epistaxe e vômitos persistentes também foram fortemente associados a óbito. Embora incluídos na mais recente classificação de dengue da OMS como sinais de alarme, epistaxe e vômitos incoercíveis não estão incluídos na definição da OMS para dengue grave. Estes achados necessitam ser explorados em estudos posteriores. Como unidades de terapia intensiva são frequentemente limitadas em cenários com poucos recursos, qualquer informação que possa distinguir, dentre os pacientes com dengue grave, aqueles com maior risco de evolução a óbito, pode ser crucial. / Dengue is a major public health problem, responsible for about 25,000 deaths in subtropical areas every year. In Brazil, the incidence of fatal forms of dengue hemorrhagic fever has increased since 2001. In particular, there has been a dramatic increase in severe cases in patients younger than 15 years of age since 2007, especially in the Northeastern region of the country. The purpose of this case-control study was to evaluate risk factors associated with death in children with severe dengue. The clinical condition of hospitalized patients with severe dengue who died (cases, n=18) was compared with that of hospitalized patients with severe dengue who survived (controls, n=77). Inclusion criteria for this study were: age under 13 years; hospital admission in São Luis, Northeastern Brazil; and laboratory-confirmed diagnosis of dengue. The diagnosis of acute dengue infection was confirmed by detection of dengue-specific IgM antibodies using an IgM Antibody Capture Enzyme-Linked Immunosorbent Assay (MAC-ELISA) or by DENV detection in serum, blood, or viscera by a Reverse Transcription - Polymerase Chain Reaction (RT-PCR). Death was strongly associated with signals of hypotensive shock (cold extremities, cyanosis and lethargy) and hemoptysis. These associations are in accordance with the most recent World Health Organization (WHO) case classification for severe dengue. We also found that epistaxis and persistent vomiting were strongly associated with death, both are included as warning signs in the WHO classification of dengue, but they are not included in the most recent WHO definition of severe dengue. These findings should be explored in further studies. Because intensive care units are often limited in resource-poor settings, any information that can help to distinguish patients with severe dengue with higher risk to progress to death may be crucial.

Objetivos: Investigar os fatores associados à hospitalização por pneumonia, em crianças menores de cinco anos de idade, no município de Ribeirão Preto - SP. Métodos: Estudo epidemiológico com delineamento do tipo caso-controle de base hospitalar, com alocação de 345 casos e 345 controles. Fatores socioeconômicos, reprodutivos, ambientais, perinatais, nutricionais, relativos ao cuidado à criança e à morbidade prévia foram considerados variáveis explanatórias. Os dados foram coletados por meio da aplicação de um questionário pré-codificado que contemplou o elenco de variáveis do estudo, incluindo-se o Instrumento de Avaliação da APS - PCATool. Odds ratios (OR) brutos e ajustados, com respectivos intervalos de confiança (95%) foram calculados, aplicando-se a regressão logística multivariada e seguindo-se os pressupostos da abordagem hierarquizada, buscando-se um modelo explicativo que contemplasse as relações hierárquicas existentes entre as exposições e o desfecho, sendo as análises desenvolvidas no software STATA, versão 12.0. Resultados: Renda familiar superior a R$700,00 foi responsável por uma redução de 32% na chance de hospitalização das crianças por pneumonia (OR=0,68; IC95%=0,47-0,98). Paridade >=2 representou um expressivo aumento na chance de hospitalização (categoria 2 partos: OR=4,60, IC95%=2,18-9,72; categoria >=3 partos: OR=3,25, IC95%=1,55-6,81), enquanto o intervalo interpartal >=48 meses e o ganho de peso na gestação de 10 Kg ou mais apresentaram efeito protetor para o desfecho (OR=0,28, IC95%=0,14-0,56 e OR=0,68, IC95%=0,47-0,97, respectivamente). Frequência à creche foi responsável por um aumento de 67% na chance de hospitalização por pneumonia (OR=1,67, IC95%=1,16-2,41). As crianças desnutridas apresentam uma chance duas vezes maior de serem hospitalizadas pela doença (OR=2,53; IC=1,06-6,05) enquanto aquelas com excesso de peso apresentam uma redução de 63% nessa chance (OR=0,37; IC=0,14-0,99); no entanto, questiona-se a plausibilidade biológica desse efeito protetor. A situação vacinal não atualizada foi responsável por um aumento de quase 3 vezes na chance de hospitalização por pneumonia (OR=2,81; IC=1,76-4,49). As crianças que fizeram uso pregresso de medicamentos apresentaram uma chance 67% maior de serem hospitalizadas por pneumonia (OR=1,67; IC=1,00-2,78; p=0,049). Crianças com sibilância prévia apresentaram o dobro de chance de serem hospitalizadas pela doença (OR categoria 1 episódio = 2,13; IC95%=1,31-3,47; OR categoria >=3 episódios = 2,37; IC95%=1,35-4,15). A exclusão de pneumonias aspirativas dentre os casos pode ter contribuído para uma maior proporção de crianças com refluxo referido entre os controles, levando a uma associação inversa à esperada (efeito de proteção) entre refluxo gastroesofágico e hospitalização por pneumonia (OR=0,55; IC=0,31-0,99). Escores Essenciais da APS acima de 3,17 foram responsáveis por um efeito protetor em relação à hospitalização por pneumonia, reduzindo as chances de hospitalização em 43% (OR para a categoria >3,41 = 0,57; IC=0,32-0,99) a 50% (OR para a categoria >3,17 e <=3,41 = 0,50; IC=0,28-0,88). Conclusões: O modelo explicativo obtido pelo presente estudo é composto, em grande parte, por variáveis relacionadas ao cuidado à criança ou às características da mãe e da família. Considerando-se os procedimentos referentes ao planejamento do estudo, à execução da coleta de dados e às análises estatísticas empregadas, reitera-se a consecução de validade interna para o estudo, sendo possível afirmar que o modelo obtido é explicativo do fenômeno da hospitalização por pneumonia, na população estudada. / Objectives: To investigate the factors associated with hospitalization due to pneumonia in children under five years of age in the city of Ribeirão Preto - SP, Brazil. Methods: Epidemiological study with a hospital-based case-control design, including 345 cases and 345 controls. Socioeconomic, reproductive, environmental, perinatal, nutritional, childcare and previous morbidity factors were considered as explanatory variables. The data were collected through the application of a pre-coded questionnaire that addressed the study variables and included the Primary Care Assessment Tool - PCATool. Gross and adjusted odds ratios (OR) were calculated with their respective confidence intervals (95%), applying multivariate logistic regression in accordance with the premises of the hierarchized approach, looking for an explanatory model that considered the existing hierarchical relations between the exposures and the outcome. The analyses were developed in STATA software, version 12.0. Results: A family income superior to R$700 was responsible for a 32% reduction in children\'s chance of hospitalization due to pneumonia (OR=0.68; 95%CI=0.47-0.98). Parity>=2 represented a considerable increase in the chance of hospitalization (category 2 births: OR=4.60, 95%CI=2.18-9.72; category >=3 births: OR=3.25, 95%CI=1.55-6.81), while the inter-birth interval >=48 months and the weight gain of 10 Kg or more during pregnancy represented a protective effect against the outcome (OR=0.28, 95%CI=0.14-0.56 and OR=0.68, 95%CI=0.47-0.97, respectively). Attending kindergarten was responsible for a 67% increase in the chance of hospitalization due to pneumonia (OR=1.67, 95%CI=1.16-2.41). Malnourished children present twice as many chances of being hospitalized due to the disease (OR=2.53; CI=1.06-6.05), while children with overweight present a 63% reduction in that chance (OR=0.37; CI=0.14-0.99); the biological plausibility of this protective effect is questioned though. An outdated vaccine situation was responsible for almost three times as many chances of hospitalization due to pneumonia (OR=2.81; CI=1.76-4.49). Children with earlier medication use revealed a 67% higher chance of being hospitalized due to pneumonia (OR=1.67; CI=1.00-2.78; p=0.049). Children with earlier wheezing presented twice as many chances of being hospitalized due to the disease (OR category 1 episode = 2.13; 95%CI=1.31-3.47; OR category >=3 episodes = 2.37; 95%CI=1.35-4.15). The exclusion of aspiration pneumonias from the cases may have contributed to a greater proportion of children with reflux among the control, leading to an inverse association (protective effect) between gastroesophageal reflux and hospitalization due to pneumonia (OR=0.55; CI=0.31-0.99). Essencial Scores of PHC superior to 3.17 were responsible for a protective effect with regard to hospitalization due to pneumonia, reducing the chances of hospitalization by 43% (OR for the category >3.41 = 0.57; CI=0.32-0.99) to 50% (OR for the category >3.17 and <=3.41 = 0.50; CI=0.28-0.88). Conclusions: The explanatory model obtained in this study largely includes variables related to childcare or the mother\'s and family\'s characteristics. In view of the study planning and data collection procedures and the statistical analyses applied, the internal validity of the study is highlighted, based on which it can be affirmed that the obtained model explains the phenomenon of hospitalization due to pneumonia in the study population.