Automated classification of fat-infiltrated axillary lymph nodes on screening mammograms.

OBJECTIVE: Fat-infiltrated axillary lymph nodes (LNs) are unique sites for ectopic fat deposition. Early studies showed a strong correlation between fatty LNs and obesity-related diseases. Confirming this correlation requires large-scale studies, hindered by scarce labeled data. With the long-term goal of developing a rapid and generalizable tool to aid data labeling, we developed an automated deep learning (DL)-based pipeline to classify the status of fatty LNs on screening mammograms. METHODS: Our internal data set included 886 mammograms from a tertiary academic medical institution, with a binary status of the fat-infiltrated LNs based on the size and morphology of the largest visible axillary LN. A two-stage DL model training and fine-tuning pipeline was developed to classify the fat-infiltrated LN status using the internal training and development data set. The model was evaluated on a held-out internal test set and a subset of the Digital Database for Screening Mammography. RESULTS: Our model achieved 0.97 (95% CI: 0.94-0.99) accuracy and 1.00 (95% CI: 1.00-1.00) area under the receiver operator characteristic curve on 264 internal testing mammograms, and 0.82 (95% CI: 0.77-0.86) accuracy and 0.87 (95% CI: 0.82-0.91) area under the receiver operator characteristic curve on 70 external testing mammograms. CONCLUSION: This study confirmed the feasibility of using a DL model for fat-infiltrated LN classification. The model provides a practical tool to identify fatty LNs on mammograms and to allow for future large-scale studies to evaluate the role of fatty LNs as an imaging biomarker of obesity-associated pathologies. ADVANCES IN KNOWLEDGE: Our study is the first to classify fatty LNs using an automated DL approach.

Linking specific biological signatures to different childhood adversities: findings from the HERO project.

BACKGROUND: Although investigations have begun to differentiate biological and neurobiological responses to a variety of adversities, studies considering both endocrine and immune function in the same datasets are limited. METHODS: Associations between proximal (family functioning, caregiver depression, and anxiety) and distal (SES-D; socioeconomic disadvantage) early-life adversities with salivary inflammatory biomarkers (IL-1ß, IL-6, IL-8, and TNF-α) and hair HPA markers (cortisol, cortisone, and dehydroepiandrosterone) were examined in two samples of young U.S. children (N = 142; N = 145). RESULTS: Children exposed to higher SES-D had higher levels of TNF-α (B = 0.13, p = 0.011), IL-1ß (B = 0.10, p = 0.033), and DHEA (B = 0.16, p = 0.011). Higher family dysfunction was associated with higher cortisol (B = 0.08, p = 0.033) and cortisone (B = 0.05, p = 0.003). An interaction between SES-D and family dysfunction was observed for cortisol levels (p = 0.020) whereby children exposed to lower/average levels of SES-D exhibited a positive association between family dysfunction and cortisol levels, whereas children exposed to high levels of SES-D did not. These findings were partially replicated in the second sample. CONCLUSIONS: Our results indicate that these biological response systems may react differently to different forms of early-life adversity. IMPACT: Different forms of early-life adversity have varied stress signatures, and investigations of early-life adversities with inflammation and HPA markers are lacking. Children with higher socioeconomic disadvantage had higher TNF-α, IL-1ß, and DHEA. Higher family dysfunction was associated with higher hair cortisol and cortisone levels, and the association between family dysfunction and cortisol was moderated by socioeconomic disadvantage. Biological response systems (immune and endocrine) were differentially associated with distinct forms of early-life adversities.

The role of behavioural factors in explaining socio-economic differences in adolescent health: a multilevel study in 33 countries.

Attempts to describe and explain socio-economic differences in health have mainly focused on adults. Little is known about the mechanisms of the relationship between socio-economic status (SES) and health in adolescence including inconsistent findings between SES and health among young people. Data were derived from representative samples of 13 and 15-year-old students in 33 European and North American countries (n=97,721) as part of the Health Behaviour in School-aged Children (HBSC) study 2001/2002. Multilevel logistic regression models were used to investigate socio-economic differences in self-rated health among adolescents and the contribution of health-related behaviours to the explanation of such differences. Odds ratios of self-rated health by family affluence were calculated before and after adjustment for behavioural factors (tobacco smoking, physical activity, television use, breakfast intake, consumption of fruits and vegetables). On average, adolescents from low affluent families had an odds ratio for low self-rated health of 1.84 for boys and 1.80 for girls, compared to those from high affluent families. The majority of behavioural factors were significantly associated with family affluence in all countries and explained part of the relationship between self-rated health and family affluence. Smoking, physical activity and breakfast consumption showed the largest independent effect on health. The present study suggests that behavioural factors in early adolescence partly account for the association between self-rated health and socio-economic status. Prevention programmes should target unhealthy behaviours of adolescents from lower socio-economic groups to help prevent future life-course disadvantages in terms of health and social inequalities.

Parental occupation, family affluence and adolescent health behaviour in 28 countries.

OBJECTIVES: We explored the association between socioeconomic position and four different aspects of adolescent health behavior in a wide range of European countries, the US, Canada and Israel. METHODS: Data were collected from the Health Behaviour in School-aged Children (HBSC) study in 2002. Representative samples of 13 and 15 year olds completed a standardised questionnaire during school hours in each country. Logistic regression analyses were used to investigate the independent effect of parental occupation and family affluence on tobacco and alcohol use, vegetable consumption and TV viewing. RESULTS: Family affluence showed no significant association with regular smoking in most countries, whereas an increase in smoking with decreasing occupational status was found in half of the countries. For alcohol consumption a positive association was found with family affluence in half of the countries, while no relationship with parental occupation was observed. Both measures of socioeconomic position were strong independent predictors for vegetable consumption and television viewing in almost all countries. CONCLUSIONS: The findings suggest that health behaviours that begin to develop in adolescence are less strongly influenced by parental socioeconomic position. Preventive intervention strategies should take the different socioeconomic patterns of health behaviour into account.

Patterns of adolescent physical activity, screen-based media use, and positive and negative health indicators in the U.S. and Canada.

PURPOSE: To examine: (1) how adolescent physical activity (PA) and screen-based media use (SBM) relate to physical and social health indicators, and (2) crossnational differences in these relationships. METHODS: Essentially identical questions and methodologies were used in the Health Behavior in School-Aged Children cross-sectional surveys of nationally representative samples of American (N = 14,818) and Canadian (N = 7266) students in grades 6 to 10. Items included questions about frequency of PA, SBM, positive health indicators (health status, self-image, quality of life, and quality of family and peer relationships), and negative health indicators (health complaints, physical aggression, smoking, drinking, and marijuana use). RESULTS: In regression analyses controlling for age and gender, positive health indicators were uniformly positively related to PA while two negative health indicators were negatively related to PA. However, PA was positively related to physical aggression. The pattern for SBM was generally the opposite; SBM was negatively related to most positive health indices and positively related to several of the negative health indicators. The notable exception was that SBM was positively related to the quality of peer relationships. Although there were crossnational differences in the strength of some relationships, these patterns were essentially replicated in both countries. CONCLUSIONS: Surveys of nationally representative samples of youth in two countries provide evidence of positive physical and social concomitants of PA and negative concomitants of SBM. These findings suggest potential positive consequences of increasing PA and decreasing SBM in adolescents and provide further justification for such efforts.

Associations between adolescent risk behaviors and injury: the modifying role of disability.

BACKGROUND: Adolescents with disabilities are at risk for poor health outcomes including injury. The objective of this study was to examine if disability status modifies the association between risk behavior and injury among adolescents. METHODS: The cross-sectional Health Behavior in School-Aged Children Survey was administered to a representative sample of 7235 Canadian students (grades 6-10) in 2002. Students who reported at least 1 functional difficulty due to a health condition were classified as having a disability. Engagement in up to 6 individual risk behaviors and a summative multiple risk behavior score were considered the primary exposures. Primary outcomes included medically attended injury experienced during a 12-month period. RESULTS: Sixteen percent of students reported a disability. Almost all risk behaviors and all injury outcomes were more common among students with disabilities than in those without disabilities (eg, older age group's smoking: 17.5% vs 8.9%, p = <.01; medically attended injury: 67.4% vs 51.4%, p = <.01). Clear risk gradients were observed between engagement in multiple risk behavior and all injury outcomes. The association between multiple risk behavior and injury was accentuated by disability status among older students, particularly for students with disabilities who engaged in frequent multiple risk behavior (adjusted risk ratio 1.8, 95% CI: 1.6-1.9). CONCLUSIONS: Canadian students with disabilities who engage in risk behaviors experience higher risks for medically attended injury than their nondisabled peers who engage in those same risk behaviors. Injury prevention programs that focus on risk-taking behavior should integrate the needs of this high-risk group of adolescents in order to prevent additional disability.

Researching health inequalities in adolescents: the development of the Health Behaviour in School-Aged Children (HBSC) family affluence scale.

Socioeconomic inequalities in adolescent health have been little studied until recently, partly due to the lack of appropriate and agreed upon measures for this age group. The difficulties of measuring adolescent socioeconomic status (SES) are both conceptual and methodological. Conceptually, it is unclear whether parental SES should be used as a proxy, and if so, which aspect of SES is most relevant. Methodologically, parental SES information is difficult to obtain from adolescents resulting in high levels of missing data. These issues led to the development of a new measure, the Family Affluence Scale (FAS), in the context of an international study on adolescent health, the Health Behaviour in School-Aged Children (HBSC) Study. The paper reviews the evolution of the measure over the past 10 years and its utility in examining and explaining health related inequalities at national and cross-national levels in over 30 countries in Europe and North America. We present an overview of HBSC papers published to date that examine FAS-related socioeconomic inequalities in health and health behaviour, using data from the HBSC study. Findings suggest consistent inequalities in self-reported health, psychosomatic symptoms, physical activity and aspects of eating habits at both the individual and country level. FAS has recently been adopted, and in some cases adapted, by other research and policy related studies and this work is also reviewed. Finally, ongoing FAS validation work is described together with ideas for future development of the measure.

Risk taking and recurrent health symptoms in Canadian adolescents.

OBJECTIVE: Psychosomatic health symptoms are commonly experienced by young people and are associated with chronic health problems. This study examines reports of psychosomatic health symptoms during adolescence, and associations between adolescent risk-taking and the occurrence of these symptoms. METHOD: A cross-sectional national health survey was performed in 2001/02 and involved 1746 young Canadians. Principal components analyses were used to derive a multiple risk-taking score. Multi-level regression models with individual participants (1st level) nested in schools (2nd level) were used to examine associations between individual then multiple risk-taking behaviors and psychosomatic symptoms. RESULTS: 37% of males and 48% of female students reported psychological health symptoms > once per week, while 30% of males and 41% of females reported somatic health symptoms. Strong associations were observed between individual risk-taking behaviors (cannabis use, other illicit drug use, drunkenness, smoking, unprotected sex) and psychosomatic health symptoms. A risk gradient was observed between multiple risk behaviors and these health symptoms. Youth who engaged in high levels of risk-taking behavior were two to four times more likely to report health symptoms. CONCLUSION: Adolescent lifestyles that include risk-taking behavior are strongly associated with the occurrence of psychosomatic health symptoms. Collectively, these measures are useful clinical indicators of adolescent health status.

Overweight and obesity in Canadian adolescents and their associations with dietary habits and physical activity patterns.

PURPOSE: To present recent overweight and obesity prevalence rates for 11-16-year-old Canadian youth and to examine associations between overweight and obesity with dietary habits and leisure-time physical activities. METHODS: Nationally representative sample of 11-16-year-old adolescents (n = 5890) from the Canadian component of the 2001/02 World Health Organization Health Behaviour in School-Aged Children Survey were used. Height, weight, dietary habits, and leisure-time activities were determined from self-report. Age- and gender-specific prevalence rates of overweight and obesity were calculated based on international body mass index cut-points. Logistic regression was employed to examine the association among measures of overweight, obesity, and lifestyle habits. RESULTS: Fifteen percent of 11-16-year-old Canadian youth were overweight (preobese) and 4.6% were obese in 2002. These prevalence rates were greater in boys than girls (p < .001), but did not vary according to age. There were no clear associations observed between dietary habits and measures of overweight and obesity. However, physical activity levels were lower (p < or = .05) and television viewing times were higher (p < .01) in overweight and obese boys and girls than normal-weight youth. CONCLUSIONS: The prevalence rates of overweight and obesity in Canadian youth are high. The results suggest that physical inactivity and sedentary behaviors are strongly related to obesity in Canadian adolescents.

Multiple risk behavior and injury: an international analysis of young people.

BACKGROUND: Multiple risk behavior plays an important role in the social etiology of youth injury, yet the consistency of this observation has not been examined multinationally. OBJECTIVE: To examine reports from young people in 12 countries, by country, age group, sex, and injury type, to quantify the strength and consistency of this association. SETTING: World Health Organization collaborative cross-national survey of health behavior in school-aged children. PARTICIPANTS: A multinational representative sample of 49 461 students aged 11, 13, and 15 years. MAIN EXPOSURE MEASURES: Additive score consisting of counts of self-reported health risk behaviors: smoking, drinking, nonuse of seat belts, bullying, excess time with friends, alienation at school and from parents, truancy, and an unusually poor diet. MAIN OUTCOME MEASURE: Self-report of a medically treated injury. RESULTS: Strong gradients in risk for injury were observed according to the numbers of risk behaviors reported. Overall, youth reporting the largest number (> or =5 health risk behaviors) experienced injury rates that were 2.46 times higher (95% confidence interval, 2.27-2.67) than those reporting no risk behaviors (adjusted odds ratios for 0 to > or =5 reported behaviors: 1.00, 1.22, 1.48, 1.73, 1.98, and 2.46, respectively; P<.001 for trend). Similar gradients in risk for injury were observed among youth in all 12 countries and within all demographic subgroups. Risk gradients were especially pronounced for nonsports, fighting-related, and severe injuries. CONCLUSIONS: Gradients in risk for youth injury increased in association with numbers of risk behaviors reported in every country examined. This cross-cultural finding indicates that the issue of multiple risk behavior, as assessed via an additive score, merits attention as an etiological construct. This concept may be useful in future injury control research and prevention efforts conducted among populations of young people.