Automated retinal image quality assessment on the UK Biobank dataset for epidemiological studies

Morphological changes in the retinal vascular network are associated with future risk of many systemic and vascular diseases. However, uncertainty over the presence and nature of some of these associations exists. Analysis of data from large population based studies will help to resolve these uncertainties. The QUARTZ (QUantitative Analysis of Retinal vessel Topology and siZe) retinal image analysis system allows automated processing of large numbers of retinal images. However, an image quality assessment module is needed to achieve full automation. In this paper, we propose such an algorithm, which uses the segmented vessel map to determine the suitability of retinal images for use in the creation of vessel morphometric data suitable for epidemiological studies. This includes an effective 3-dimensional feature set and support vector machine classification. A random subset of 800 retinal images from UK Biobank (a large prospective study of 500,000 middle aged adults; where 68,151 underwent retinal imaging) was used to examine the performance of the image quality algorithm. The algorithm achieved a sensitivity of 95.33% and a specificity of 91.13% for the detection of inadequate images. The strong performance of this image quality algorithm will make rapid automated analysis of vascular morphometry feasible on the entire UK Biobank dataset (and other large retinal datasets), with minimal operator involvement, and at low cost.

Affections dermatologiques et troubles dépressifs : leur association chez les personnes âgées du Québec

Résumé : Les liens entre les maladies dermatologiques et les troubles de santé mentale, dont la dépression, sont reconnus depuis longtemps. Cependant, peu d’études de population ont examiné cette problématique et aucune n’a été faite auprès des aînés. Le but de ce mémoire est d’explorer l’association entre les affections dermatologiques et les troubles dépressifs chez les personnes âgées vivant à domicile. Pour ce faire, deux études ont été réalisées. La première visait à décrire les caractéristiques dermatologiques de la population à l’étude et la deuxième avait pour but de tester l’hypothèse d’une association bidirectionnelle entre les problèmes mentionnés. Il s’agit d’une analyse secondaire des données de l’Enquête sur la Santé des Aînés (ESA) qui a été menée auprès d’un échantillon représentatif de la population âgée (≥ 65 ans) vivant à domicile au Québec. Des mesures répétées à un an d’intervalle (T1 et T2) ont été obtenues auprès de 2 cohortes successives fixes. Les données de l’enquête ESA ont été appariées à celles des registres de la Régie de l’assurance maladie du Québec (RAMQ). Les troubles dépressifs ont été définis en se basant sur les critères du DSM-IV et les affections dermatologiques ont été mesurées à partir de deux sources de données (enquête et registres administratifs de la RAMQ). Des modèles autorégressifs bivariés ont été utilisés pour tester l’hypothèse d’association bidirectionnelle entre les affections dermatologiques et les troubles dépressifs. Nos résultats ont montré que près de 13% et 21% des répondants ont rapporté des affections dermatologiques auto-rapportées ou avaient été diagnostiqués selon les registres de la RAMQ. En plus, près de 6% des participants rapportaient un trouble dépressif probable au T1 et au T2. Nos résultats suggèrent la présence d’une association synchronique (transversale) entre les affections dermatologiques et les troubles dépressifs. Ce projet souligne l’importance d’évaluer et d’explorer la cooccurrence de ces deux pathologies afin d’améliorer la prise en charge des individus atteints simultanément par ces deux problèmes. Nous suggérons l'inclusion des affections dermatologiques dans les futures études épidémiologiques visant à explorer les liens entre les troubles de santé mentale et de santé physique chez les personnes âgées.//Abstract : The relationship between skin conditions and mental health disorders, which includes depression, has long been recognized. However, few population - based studies have examined this issue and none were carried out in older - adults. The aim of this project was to explore the associations between skin conditions and depressive disorders affecting the elderly living at home. To do this, two studies were conducted; the first aimed to describe the dermatological features of the study population. The second was designed to test the hypothesis of a bidirectional association between the conditions mentioned above. We carried out secondary data analyses from data collected in the Study on the Health of Seniors (ESA ) survey, which consisted of a representative sample of the elderly population (≥ 65 years) living at home in Quebec. Two repeated measurements one year apart (T1 and T2) were obtained from two fixed successive cohorts. Participants in both phases of the investigation and with available health service information from Quebec’s health insurance plan agency (Régie de l'assurance maladie du Québec - RAMQ) were selected for this project. Depressive disorders were defined based on DSM - IV criteria and dermatological conditions were measured from two data sources (survey and administrative records). Cross - lagged panel models were used to test the hypothesis of association between the two mentioned conditions. Our results showed that nearly 13% and 21% of respondents have self - reported and diagnosed skin conditions, respectively. In addition, about 6% of participants reported symptoms that were consistent with a probable depressive disorder on T1 and T2. Our results suggested the presence of synchronous (cross - sectional) associations between skin conditions and depressive disorders in the elderly. This research highlights the importance of assessing and exploring the co - occurrence of these two conditions to improve the management of individuals who are affected. We suggest the inclusion of dermatological conditions in future/further studies exploring the comorbidity between mental and physical health in the older adults.

Risk prediction of prevalent diabetes in a Swiss population using a weighted genetic score--the CoLaus Study.

AIMS/HYPOTHESIS: Several susceptibility genes for type 2 diabetes have been discovered recently. Individually, these genes increase the disease risk only minimally. The goals of the present study were to determine, at the population level, the risk of diabetes in individuals who carry risk alleles within several susceptibility genes for the disease and the added value of this genetic information over the clinical predictors. METHODS: We constructed an additive genetic score using the most replicated single-nucleotide polymorphisms (SNPs) within 15 type 2 diabetes-susceptibility genes, weighting each SNP with its reported effect. We tested this score in the extensively phenotyped population-based cross-sectional CoLaus Study in Lausanne, Switzerland (n = 5,360), involving 356 diabetic individuals. RESULTS: The clinical predictors of prevalent diabetes were age, BMI, family history of diabetes, WHR, and triacylglycerol/HDL-cholesterol ratio. After adjustment for these variables, the risk of diabetes was 2.7 (95% CI 1.8-4.0, p = 0.000006) for individuals with a genetic score within the top quintile, compared with the bottom quintile. Adding the genetic score to the clinical covariates improved the area under the receiver operating characteristic curve slightly (from 0.86 to 0.87), yet significantly (p = 0.002). BMI was similar in these two extreme quintiles. CONCLUSIONS/INTERPRETATION: In this population, a simple weighted 15 SNP-based genetic score provides additional information over clinical predictors of prevalent diabetes. At this stage, however, the clinical benefit of this genetic information is limited.

Utilisation de la coronarographie en Suisse: une étude de population. [The use of coronary vessel radiography in Switzerland: a population study].

All patients having undergone a coronarography during 1984 have been surveyed in Switzerland. This retrospective study has used existing data in the 13 centers practicing this diagnostic procedure. 4921 coronarographies were carried out in 1984, amongst 4359 patients. In terms of population-based rates, the national figures are 77 procedures/100,000 residents, and 68 patients/100,000 residents. Female rates are one fourth of the male rates (27/100,000 versus 112/100,000). For both sexes, the highest utilization rates are for the age groups 60-64. Swiss figures are relatively low when compared with other developed countries. However, patterns of utilization are very different within the country: according to the Canton of residence of the patient, the utilization rates (standardized for age and sex) vary from 8/100,000 to 160/100,000. There is a distinct gradient from south-west to north-east, which closely corresponds to the distribution of centers practicing the procedure. More intriguing is the fact that cardiovascular mortality shows an inverse geographical gradient, with the highest mortality in Cantons having the lowest rate of coronarography. Various reasons for the observed variations are discussed, in relation with differences in supply of diagnostic and therapeutic equipments, but also in relation with various patterns of demand related to differential morbidity rates and/or differential patterns of clinical decision.

Linking Global Youth Tobacco Survey (GYTS) data to the WHO Framework Convention on Tobacco Control (FCTC): the case for the Seychelles

Tobacco control has been recognized as a main public health concern in Seychelles for the past two decades. Tobacco advertising, sponsoring and promotion has been banned for years, tobacco products are submitted to high taxes, high-profile awareness programs are organized regularly, and several other control measures have been implemented. The Republic of Seychelles was the first country to ratify the WHO Framework Convention on Tobacco Control (FCTC) in the African region. Three population-based surveys have been conducted in adults in Seychelles and results showed a substantial decrease in the prevalence of smoking among adults between 1989 and 2004. A first survey in adolescents was conducted in Seychelles in 2002 (the Global Youth Tobacco Survey, GYTS) in a representative sample of 1321 girls and boys aged 13-15 years. The results show that approximately half of students had tried smoking and a quarter of both boys and girls had smoked at least one cigarette during the past 30 days. Although "current smoking" is defined differently in adolescents (>or=1 cigarette during the past 30 days) and in adults (>or=1 cigarette per day), which precludes direct comparison, the high smoking prevalence in youth in Seychelles likely predicts an increasing prevalence of tobacco use in the next adult generation, particularly in women. GYTS 2002 also provides important data on a wide range of specific individual and societal factors influencing tobacco use. Hence, GYTS can be a powerful tool for monitoring the situation of tobacco use in adolescents, for highlighting the need for new policy and programs, and for evaluating the impact of current and future programs.

High expectation in non-evidence-based smoking cessation interventions among smokers--the CoLaus study.

To assess the preferred methods to quit smoking among current smokers. Cross-sectional, population-based study conducted in Lausanne between 2003 and 2006 including 988 current smokers. Preference was assessed by questionnaire. Evidence-based (EB) methods were nicotine replacement, bupropion, physician or group consultations; non-EB-based methods were acupuncture, hypnosis and autogenic training. EB methods were frequently (physician consultation: 48%, 95% confidence interval (45-51); nicotine replacement therapy: 35% (32-38)) or rarely (bupropion and group consultations: 13% (11-15)) preferred by the participants. Non-EB methods were preferred by a third (acupuncture: 33% (30-36)), a quarter (hypnosis: 26% (23-29)) or a seventh (autogenic training: 13% (11-15)) of responders. On multivariate analysis, women preferred both EB and non-EB methods more frequently than men (odds ratio and 95% confidence interval: 1.46 (1.10-1.93) and 2.26 (1.72-2.96) for any EB and non-EB method, respectively). Preference for non-EB methods was higher among highly educated participants, while no such relationship was found for EB methods. Many smokers are unaware of the full variety of methods to quit smoking. Better information regarding these methods is necessary.

Excessive alcohol consumption in young men: is there an association with their earlier family situation? A baseline-analysis of the C-SURF-study (Cohort Study on Substance Use Risk Factors).

AIMS: To determine whether parental factors earlier in life (parenting, single parent family, parental substance use problem) are associated with patterns of alcohol consumption among young men in Switzerland. METHODS: This analysis of a population based sample from the Cohort Study on Substance Use Risk Factors (C-SURF) included 5,990 young men (mean age 19.51 years), all attending a mandatory recruitment process for the army. These conscripts reported on parental monitoring and rule-setting, parental behaviour and family structure. The alcohol use pattern was assessed through abstention, risky single occasion drinking (RSOD), volume drinking and dependence. Furthermore, the impact of age, family socio-economic status, educational level of the parents, language region and civil status was analysed. RESULTS: A parental substance use problem was positively associated with volume drinking and alcohol dependence in young Swiss men. Active parenting corresponded negatively with RSOD, volume drinking and alcohol dependence. Single parent family was not associated with a different alcohol consumption pattern compared to standard family. CONCLUSION: Parental influences earlier in life such as active parenting (monitoring, rule-setting and knowing the whereabouts) and perceived parental substance use problem are associated with alcohol drinking behaviour in young male adults. Therefore, health professionals should stress the importance of active parenting and parental substance use prevention in alcohol prevention strategies.

Prevalence of overweight and obesity among migrants in Switzerland: association with country of origin.

Migrants tend to present higher overweight and obesity levels, but whether this relationship applies to all nationalities has seldom been studied. The present study aimed to assess the prevalence of overweight and obesity according to nationality in adults. Cross-sectional population-based samples. Five-year nationwide interview surveys (Swiss Health Surveys - SHS) from 1992 to 2007 (n 63 766) and a local examination survey (CoLaus Study in Lausanne 2004-2006, n 6743). Participants were separated into Swiss, French, German, Italian, Portuguese, Spanish nationals, those from the former Republic of Yugoslavia and from other European and other countries. Compared with Swiss nationals, German and French nationals presented a lower prevalence of overweight and obesity, whereas nationals from Italy, Spain, Portugal and the former Republic of Yugoslavia presented higher levels. Adjusting the SHS data for age, gender, education, smoking, leisure-time physical activity and survey year, a lower risk for overweight and obesity was found for German (OR = 0·80, 95 % CI 0·70, 0·92) and French (OR = 0·74, 95 % CI 0·61, 0·89) nationals, whereas higher risks were found for participants from Italy (OR = 1·45, 95 % CI 1·33, 1·58), Spain (OR = 1·36, 95 % CI 1·15, 1·61), Portugal (OR = 1·25, 95 % CI 1·06, 1·47) and the former Republic of Yugoslavia (OR = 1·98, 95 % CI 1·69, 2·32). Similar findings were observed in the CoLaus Study for Italian (OR = 1·63, 95 % CI 1·29, 2·06), Spanish (OR = 1·54, 95 % CI 1·17, 2·04) and Portuguese (OR = 1·49, 95 % CI 1·16, 1·91) participants and for those from the former Republic of Yugoslavia (OR = 5·34, 95 % CI 3·00, 9·50). Overweight and obesity are unevenly distributed among migrants in Switzerland. Migrants from Southern Europe and from the former Republic of Yugoslavia present higher prevalence rates. This suggests that preventive messages should be tailored to these specific populations.

Mirror extreme BMI phenotypes associated with gene dosage at the chromosome 16p11.2 locus.

Both obesity and being underweight have been associated with increased mortality. Underweight, defined as a body mass index (BMI) ≤ 18.5 kg per m(2) in adults and ≤ -2 standard deviations from the mean in children, is the main sign of a series of heterogeneous clinical conditions including failure to thrive, feeding and eating disorder and/or anorexia nervosa. In contrast to obesity, few genetic variants underlying these clinical conditions have been reported. We previously showed that hemizygosity of a ∼600-kilobase (kb) region on the short arm of chromosome 16 causes a highly penetrant form of obesity that is often associated with hyperphagia and intellectual disabilities. Here we show that the corresponding reciprocal duplication is associated with being underweight. We identified 138 duplication carriers (including 132 novel cases and 108 unrelated carriers) from individuals clinically referred for developmental or intellectual disabilities (DD/ID) or psychiatric disorders, or recruited from population-based cohorts. These carriers show significantly reduced postnatal weight and BMI. Half of the boys younger than five years are underweight with a probable diagnosis of failure to thrive, whereas adult duplication carriers have an 8.3-fold increased risk of being clinically underweight. We observe a trend towards increased severity in males, as well as a depletion of male carriers among non-medically ascertained cases. These features are associated with an unusually high frequency of selective and restrictive eating behaviours and a significant reduction in head circumference. Each of the observed phenotypes is the converse of one reported in carriers of deletions at this locus. The phenotypes correlate with changes in transcript levels for genes mapping within the duplication but not in flanking regions. The reciprocal impact of these 16p11.2 copy-number variants indicates that severe obesity and being underweight could have mirror aetiologies, possibly through contrasting effects on energy balance.

Association of major and minor ECG abnormalities with coronary heart disease events.

CONTEXT: In populations of older adults, prediction of coronary heart disease (CHD) events through traditional risk factors is less accurate than in middle-aged adults. Electrocardiographic (ECG) abnormalities are common in older adults and might be of value for CHD prediction. OBJECTIVE: To determine whether baseline ECG abnormalities or development of new and persistent ECG abnormalities are associated with increased CHD events. DESIGN, SETTING, AND PARTICIPANTS: A population-based study of 2192 white and black older adults aged 70 to 79 years from the Health, Aging, and Body Composition Study (Health ABC Study) without known cardiovascular disease. Adjudicated CHD events were collected over 8 years between 1997-1998 and 2006-2007. Baseline and 4-year ECG abnormalities were classified according to the Minnesota Code as major and minor. Using Cox proportional hazards regression models, the addition of ECG abnormalities to traditional risk factors were examined to predict CHD events. MAIN OUTCOME MEASURE: Adjudicated CHD events (acute myocardial infarction [MI], CHD death, and hospitalization for angina or coronary revascularization). RESULTS: At baseline, 276 participants (13%) had minor and 506 (23%) had major ECG abnormalities. During follow-up, 351 participants had CHD events (96 CHD deaths, 101 acute MIs, and 154 hospitalizations for angina or coronary revascularizations). Both baseline minor and major ECG abnormalities were associated with an increased risk of CHD after adjustment for traditional risk factors (17.2 per 1000 person-years among those with no abnormalities; 29.3 per 1000 person-years; hazard ratio [HR], 1.35; 95% CI, 1.02-1.81; for minor abnormalities; and 31.6 per 1000 person-years; HR, 1.51; 95% CI, 1.20-1.90; for major abnormalities). When ECG abnormalities were added to a model containing traditional risk factors alone, 13.6% of intermediate-risk participants with both major and minor ECG abnormalities were correctly reclassified (overall net reclassification improvement [NRI], 7.4%; 95% CI, 3.1%-19.0%; integrated discrimination improvement, 0.99%; 95% CI, 0.32%-2.15%). After 4 years, 208 participants had new and 416 had persistent abnormalities. Both new and persistent ECG abnormalities were associated with an increased risk of subsequent CHD events (HR, 2.01; 95% CI, 1.33-3.02; and HR, 1.66; 95% CI, 1.18-2.34; respectively). When added to the Framingham Risk Score, the NRI was not significant (5.7%; 95% CI, -0.4% to 11.8%). CONCLUSIONS: Major and minor ECG abnormalities among older adults were associated with an increased risk of CHD events. Depending on the model, adding ECG abnormalities was associated with improved risk prediction beyond traditional risk factors.

Serum 25-hydroxyvitamin D level and kidney function decline in a Swiss general adult population.

Molecular evidence suggests that levels of vitamin D are associated with kidney function loss. Still, population-based studies are limited and few have considered the potential confounding effect of baseline kidney function. This study evaluated the association of serum 25-hydroxyvitamin D with change in eGFR, rapid eGFR decline, and incidence of CKD and albuminuria. Baseline (2003-2006) and 5.5-year follow-up data from a Swiss adult general population were used to evaluate the association of serum 25-hydroxyvitamin D with change in eGFR, rapid eGFR decline (annual loss >3 ml/min per 1.73 m(2)), and incidence of CKD and albuminuria. Serum 25-hydroxyvitamin D was measured at baseline using liquid chromatography-tandem mass spectrometry. eGFR and albuminuria were collected at baseline and follow-up. Multivariate linear and logistic regression models were used considering potential confounding factors. Among the 4280 people included in the analysis, the mean±SD annual eGFR change was -0.57±1.78 ml/min per 1.73 m(2), and 287 (6.7%) participants presented rapid eGFR decline. Before adjustment for baseline eGFR, baseline 25-hydroxyvitamin D level was associated with both mean annual eGFR change and risk of rapid eGFR decline, independently of baseline albuminuria. Once adjusted for baseline eGFR, associations were no longer significant. For every 10 ng/ml higher baseline 25-hydroxyvitamin D, the adjusted mean annual eGFR change was -0.005 ml/min per 1.73 m(2) (95% confidence interval, -0.063 to 0.053; P=0.87) and the risk of rapid eGFR decline was null (odds ratio, 0.93; 95% confidence interval, 0.79 to 1.08; P=0.33). Baseline 25-hydroxyvitamin D level was not associated with incidence of CKD or albuminuria. The association of 25-hydroxyvitamin D with eGFR decline is confounded by baseline eGFR. Sufficient 25-hydroxyvitamin D levels do not seem to protect from eGFR decline independently from baseline eGFR.

Ideal Body Size as a Mediator for the Gender-Specific Association Between Socioeconomic Status and Body Mass Index: Evidence From an Upper-Middle-Income Country in the African Region.

While obesity continues to rise globally, the associations between body size, gender, and socioeconomic status (SES) seem to vary in different populations, and little is known on the contribution of perceived ideal body size in the social disparity of obesity in African countries. We examined the gender and socioeconomic patterns of body mass index (BMI) and perceived ideal body size in the Seychelles, a middle-income small island state in the African region. We also assessed the potential role of perceived ideal body size as a mediator for the gender-specific association between SES and BMI. A population-based survey of 1,240 adults aged 25 to 64 years conducted in December 2013. Participants' BMI was calculated based on measured weight and height; ideal body size was assessed using a nine-silhouette instrument. Three SES indicators were considered: income, education, and occupation. BMI and perceived ideal body size were both higher among men of higher versus lower SES (p< .001) but lower among women of higher versus lower SES (p< .001), irrespective of the SES indicator used. Multivariate analysis showed a strong and direct association between perceived ideal body size and BMI in both men and women (p< .001) and was consistent with a potential mediating role of perceived ideal body size in the gender-specific associations between SES and BMI. Our study emphasizes the importance of gender and socioeconomic differences in BMI and ideal body size and suggests that public health interventions that promote perception of healthy weight could help mitigate SES-related disparities in BMI.

Feature Selection and Classification Using Age Layered Population Structure Genetic Programming

The curse of dimensionality is a major problem in the fields of machine learning, data mining and knowledge discovery. Exhaustive search for the most optimal subset of relevant features from a high dimensional dataset is NP hard. Sub–optimal population based stochastic algorithms such as GP and GA are good choices for searching through large search spaces, and are usually more feasible than exhaustive and deterministic search algorithms. On the other hand, population based stochastic algorithms often suffer from premature convergence on mediocre sub–optimal solutions. The Age Layered Population Structure (ALPS) is a novel metaheuristic for overcoming the problem of premature convergence in evolutionary algorithms, and for improving search in the fitness landscape. The ALPS paradigm uses an age–measure to control breeding and competition between individuals in the population. This thesis uses a modification of the ALPS GP strategy called Feature Selection ALPS (FSALPS) for feature subset selection and classification of varied supervised learning tasks. FSALPS uses a novel frequency count system to rank features in the GP population based on evolved feature frequencies. The ranked features are translated into probabilities, which are used to control evolutionary processes such as terminal–symbol selection for the construction of GP trees/sub-trees. The FSALPS metaheuristic continuously refines the feature subset selection process whiles simultaneously evolving efficient classifiers through a non–converging evolutionary process that favors selection of features with high discrimination of class labels. We investigated and compared the performance of canonical GP, ALPS and FSALPS on high–dimensional benchmark classification datasets, including a hyperspectral image. Using Tukey’s HSD ANOVA test at a 95% confidence interval, ALPS and FSALPS dominated canonical GP in evolving smaller but efficient trees with less bloat expressions. FSALPS significantly outperformed canonical GP and ALPS and some reported feature selection strategies in related literature on dimensionality reduction.

Feature Selection and Classification Using Age Layered Population Structure Genetic Programming

The curse of dimensionality is a major problem in the fields of machine learning, data mining and knowledge discovery. Exhaustive search for the most optimal subset of relevant features from a high dimensional dataset is NP hard. Sub–optimal population based stochastic algorithms such as GP and GA are good choices for searching through large search spaces, and are usually more feasible than exhaustive and determinis- tic search algorithms. On the other hand, population based stochastic algorithms often suffer from premature convergence on mediocre sub–optimal solutions. The Age Layered Population Structure (ALPS) is a novel meta–heuristic for overcoming the problem of premature convergence in evolutionary algorithms, and for improving search in the fitness landscape. The ALPS paradigm uses an age–measure to control breeding and competition between individuals in the population. This thesis uses a modification of the ALPS GP strategy called Feature Selection ALPS (FSALPS) for feature subset selection and classification of varied supervised learning tasks. FSALPS uses a novel frequency count system to rank features in the GP population based on evolved feature frequencies. The ranked features are translated into probabilities, which are used to control evolutionary processes such as terminal–symbol selection for the construction of GP trees/sub-trees. The FSALPS meta–heuristic continuously refines the feature subset selection process whiles simultaneously evolving efficient classifiers through a non–converging evolutionary process that favors selection of features with high discrimination of class labels. We investigated and compared the performance of canonical GP, ALPS and FSALPS on high–dimensional benchmark classification datasets, including a hyperspectral image. Using Tukey’s HSD ANOVA test at a 95% confidence interval, ALPS and FSALPS dominated canonical GP in evolving smaller but efficient trees with less bloat expressions. FSALPS significantly outperformed canonical GP and ALPS and some reported feature selection strategies in related literature on dimensionality reduction.

Risque de cancer du poumon associé aux expositions environnementales de fumées de soudage : 2 études cas-témoins basées sur la population montréalaise.

Objectif : Examiner la relation entre une exposition professionnelle aux fumées de soudage au gaz et à l’arc et le risque de développer un cancer du poumon, chez des travailleurs provenant de différents milieux de travail et exposés à de nombreuses substances de différentes concentrations. Méthodes : Nous avons mené deux études cas-témoins basés sur la population montréalaise (1979-86 et 1996-2001), incluant respectivement 857 et 736 cas, ainsi que des témoins appariés selon la fréquence des cas, le sexe et le lieu de résidence. Un historique détaillé des emplois occupés fut obtenu par entrevue et évalué par une équipe d’experts afin d’estimer l’intensité, la durée ainsi que l’exposition cumulative à un grand nombre de substances. Les fumées de soudage au gaz et à l’arc sont parmi les substances évaluées et nous avons pu calculer un rapport de cote (RC) et son intervalle de confiance de 95% (IC 95%) pour le risque de cancer du poumon en utilisant la régression logistique et en ajustant pour l’historique de tabagisme et plusieurs autres covariables pertinentes. Des analyses subséquentes ont permis d’explorer d’autres voies, comme la modification des effets observés par le tabagisme. Résultats : Les résultats obtenus à partir des deux études étant assez similaires, nous avons donc pu les combiner pour former une seule étude. Aucune association statistiquement significative n’a été trouvée entre le cancer du poumon et les fumées de soudage au gaz (RC=1,13; IC 95%=0,90-1,42) et les fumées de soudage à l’arc (RC=1,01; IC 95%=0,80-1,26). Par contre, dans le sous-groupe des non-fumeurs et fumeurs très légers, nous avons trouvé un risque accru de cancer du poumon en relation avec les fumées de soudage au gaz (RC=2,78; IC 95%=1,66-4,65) et à l’arc (RC=2,20; IC 95%=1,32-3,70). En se concentrant sur ce sous-groupe, nous avons trouvé un risque encore plus élevé de développer un cancer du poumon pour ceux exposé à des doses plus substantielles, pour le soudage au gaz (RC=4,63; IC 95%=2,14-10,03) et à l’arc (RC=3,45; IC 95%=1,59-7,48). Discussion : Globalement, aucun excès de risque causé par une exposition aux fumées de soudage n’a été détecté, mais un excès a été trouvé chez les non-fumeurs et fumeurs légers uniquement et ce, pour les deux types de fumées de soudage. L’effet des fumées de soudage peut être voilé chez les fumeurs, ou bien les non-fumeurs pourraient être plus vulnérables.