Effect of imbalanced sampling and missing data on associations between gender norms and risk of adolescent HIV

BACKGROUND

Despite strides towards gender equality, inequalities persist or remain unstudied, due potentially to data gaps. Although mapped, the effects of key data gaps remain unknown. This study provides a framework to measure effects of gender- and age-imbalanced and missing covariate data on gender-health research. The framework is demonstrated using a previously studied pathway for effects of pre-marital sex norms among adults on adolescent HIV risk.

METHODS

After identifying gender-age-imbalanced Demographic and Health Survey (DHS) datasets, we resampled responses and restricted covariate data from a relatively complete, balanced dataset derived from the 2007 Zambian DHS to replicate imbalanced gender-age sampling and covariate missingness. Differences in model outcomes due to sampling were measured using tests for interaction. Missing covariate effects were measured by comparing fully-adjusted and reduced model fitness.

FINDINGS

We simulated data from 25 DHS surveys across 20 countries from 2005-2014 on four sex-stratified models for pathways of adult attitude-behaviour discordance regarding pre-marital sex and adolescent risk of HIV. On average, across gender-age-imbalanced surveys, males comprised 29.6% of responses compared to 45.3% in the gender-balanced dataset. Gender-age-imbalanced sampling significantly affected regression coefficients in 40% of model-scenarios (N = 40 of 100) and biased relative-risk estimates away from gender-age-balanced sampling outcomes in 46% (N = 46) of model-scenarios. Model fitness was robust to covariate removal with minor effects on male HIV models. No consistent trends were observed between sampling distribution and risk of biased outcomes.

INTERPRETATION

Gender-health model outcomes may be affected by sampling gender-age-imbalanced data and less-so by missing covariates. Although occasionally attenuated, the effect magnitude of gender-age-imbalanced sampling is variable and may mask true associations, thus misinforming policy dialogue. We recommend future surveys improve balanced gender-age sampling to promote research reliability.

FUNDING

Bill & Melinda Gates Foundation grant OPP1140262 to Stanford University.

Quantification of Gender Bias and Sentiment Toward Political Leaders Over 20 Years of Kenyan News Using Natural Language Processing

Background: Despite a 2010 Kenyan constitutional amendment limiting members of elected public bodies to < two-thirds of the same gender, only 22 percent of the 12th Parliament members inaugurated in 2017 were women. Investigating gender bias in the media is a useful tool for understanding socio-cultural barriers to implementing legislation for gender equality. Natural language processing (NLP) methods, such as word embedding and sentiment analysis, can efficiently quantify media biases at a scope previously unavailable in the social sciences.

Methods: We trained GloVe and word2vec word embeddings on text from 1998 to 2019 from Kenya’s Daily Nation newspaper. We measured gender bias in these embeddings and used sentiment analysis to predict quantitative sentiment scores for sentences surrounding female leader names compared to male leader names.

Results: Bias in leadership words for men and women measured from Daily Nation word embeddings corresponded to temporal trends in men and women’s participation in political leadership (i.e., parliamentary seats) using GloVe (correlation 0.8936, p = 0.0067, r² = 0.799) and word2vec (correlation 0.844, p = 0.0169, r² = 0.712) algorithms. Women continue to be associated with domestic terms while men continue to be associated with influence terms, for both regular gender words and female and male political leaders’ names. Male words (e.g., he, him, man) were mentioned 1.84 million more times than female words from 1998 to 2019. Sentiment analysis showed an increase in relative negative sentiment associated with female leaders (p = 0.0152) and an increase in positive sentiment associated with male leaders over time (p = 0.0216).

Conclusion: Natural language processing is a powerful method for gaining insights into and quantifying trends in gender biases and sentiment in news media. We found evidence of improvement in gender equality but also a backlash from increased female representation in high-level governmental leadership.

Gender-related data missingness, imbalance and bias in global health surveys

Global surveys have built-in gender-related biases associated with data missingness across the gender dimensions of people's lives, imbalanced or incomplete representation of population groups, and biased ways in which gender information is elicited and used. While increasing focus is being placed on the integration of sex-disaggregated statistics into national programmes and on understanding effects of gender-based disparities on the health of all people, the data necessary for elucidating underlying causes of gender disparities and designing effective intervention programmes continue to be lacking. Approaches exist, however, that can reasonably address some shortcomings, such as separating questions of gender identification from biological sex. Qualitative research can elucidate ways to rephrase questions and translate gendered terms to avoid perpetuating historical gender biases and prompting biased responses. Non-health disciplines may offer lessons in collecting gender-related data. Ultimately, multidisciplinary global collaborations are needed to advance this evolving field and to set standards for how we measure gender in all its forms.

Association of collective attitudes and contraceptive practice in nine sub-Saharan African countries

BACKGROUND

There is ample evidence that gender norms affect contraceptive practice; however, data are mostly qualitative with limited geographical scope. We investigated that association quantitatively using collective community-level attitudes towards premarital sex and wife-beating as proxies for gender norms.

METHODS

Data came from nationally representative Demographic and Health Surveys (2005-2009) for women of reproductive age (15-49 years) in nine sub-Saharan African countries. Using multilevel logistic models, controlling for individual covariates and community-level indicators of women's empowerment, we assessed the community-level association of gender norms regarding premarital sex and wife-beating with individual contraception uptake and demand satisfied among fecund sexually active women. Norms were approximated as 'collective attitudinal norms' from female/male residents (aged 15-49 years) from the same community. We assessed the magnitude and significance of the community-level effects and attributed variance across communities. The same analysis was replicated for each country.

RESULTS

In a fully-adjusted model with a pooled sample of 24 404 adolescent women, the odds of contraception use increased with a 1 standard deviation (SD) increase in the variation of collective permissive attitudes towards premarital sex of female (odds ratio (OR) = 1.08, 95% confidence interval (CI) = 1.02-1.15) and male (OR = 1.11, 95% CI = 1.05-1.17) peers (15-24 years), while odds of contraceptive use declined by 10% (OR = 0.90, 95% CI = 0.85-0.96) with collective accepting attitudes towards wife-beating of women aged 15-49 years. Similar results were found in separate models that controlled for adults' permissive attitudes towards premarital sex. The community-level attributed variance (V2 = 1.62, 95% CI = 1.45-1.80) represented 33% (intra-class correlation (ICC) = 33.0, 95% CI = 30.0-35.4) of the total variation of contraception use, and attitudes towards premarital sex and violence jointly explained nearly 26% of that V2 variance. The community-level shared of attributed variation of contraceptive use varied significantly across countries, from 3.5% in Swaziland (ICC = 3.5, 95% CI = 0.8-13.7) to 60.2% in Nigeria (OR = 60.2, 95% CI = 56.0-64.2).

CONCLUSIONS

Overall, significant positive associations of collective permissive attitudes of both adolescent and adult women towards premarital sex were found for use of, and demand for, contraception, whereas collective accepting attitudes towards wife-beating were negatively associated with the use and demand for contraception. Ours is the first study to define quantitatively the influence of proxies for gender norms at the community level on women's family planning decisions. These findings offer new insights for understanding the role of sex-related attitudes and norms as important factors in shaping contraceptive practices and improving the effectiveness of family planning policies by targeting individuals as well as their groups of influence.

Gender Norms and Weight Control Behaviors in U.S. Adolescents: A Prospective Cohort Study (1994-2002)

PURPOSE

The aim of this article was to determine the relationship between gender norms and weight control behaviors in U.S. adolescents.

METHODS

We analyzed prospective cohort data from the National Longitudinal Study of Adolescent to Adult Health (N = 9,861), at baseline in 1994-1995 (ages 11-18 years, Wave I), 1-year follow-up (ages 12-19 years, Wave II), and 7-year follow-up (ages 18-26 years, Wave III). The primary exposure variable was a measure of one's gender normativity based on the degree to which males and females behave in ways that are similar to the behaviors of their same-gender peers. The outcome variable was an individual's weight control attempts (trying to lose or gain weight) and behaviors (dieting, fasting/skipping meals, vomiting, or weight-loss pills/laxatives/diuretics to lose weight or ate different/more foods than usual or taking supplements to gain weight).

RESULTS

In logistic regression analyses controlling for potential confounders, a higher baseline individual gender normativity score (higher femininity in females and higher masculinity in males) was associated with weight loss attempts (β = .10; p = .01) and weight loss behaviors (β = .18; p < .001) in girls but was associated with weight gain attempts (β = .18; p < .001) and behaviors (β = .16; p < .001) in boys at 1-year follow-up. Higher individual gender normativity score was protective of weight loss attempts (β = -.15; p < .001) and weight loss behaviors (β = -.17; p < .001) in males but not females at 7-year follow-up. Loess plots provided visualizations of significant relationships.

CONCLUSIONS

Gender norms may reinforce a thinner body ideal for girls but a larger ideal for boys.

Parent-Related Normative Perceptions of Adolescents and Later Weight Control Behavior: Longitudinal Analysis of Cohort Data From Brazil

PURPOSE

Body image-related norms can be imposed by parents and can shape adolescents' body satisfaction in consequential ways, yet evidence on long-term effects is scarce. Longitudinal data from a country with strong body image focus provided a unique opportunity to investigate long-term influences of normative parent-related perceptions.

METHODS

Multinomial logistic regression was used on data from a 1993 birth cohort in Brazil to investigate the association of normal-body mass index (BMI) adolescents' perception of their parent's opinion of their weight at age 11 years with their weight control attempts at 18 years, testing a mediating role for body dissatisfaction at age 15 years. All models controlled for body dissatisfaction at age 11 years and BMI change between ages 11 and 15 years.

RESULTS

A total of 1150 boys and 1336 girls were included. Girls were more likely than boys to diet without nutritionist advice to lose weight (51.5% vs. 34.3% among boys) and use medication to gain weight (12.7% vs. 4.2%). Normal-BMI adolescents who reported at age 11 years that their parents thought they were thin had higher odds of feeling thinner than ideal at age 15 years (odds ratio 2.8, 95% confidence interval 1.8-3.2; and odds ratio 2.0, 95% confidence interval 1.5-2.7) among boys and girls, respectively). Feeling thinner than ideal at age 15 years was associated among girls with higher odds of weight gain attempts at age 18 years. Similar patterns appeared among girls reporting that their parents thought they were fat at age 11 years, feeling fatter than ideal at age 15 years and having higher odds of weight loss attempts at age 18 years. Body dissatisfaction was a statistically significant mediator among girls but not boys.

CONCLUSIONS

A long-term influence of parent-related perceptions via a likely trajectory of body dissatisfaction is evident among girls.

Gender norms and health: insights from global survey data

Despite global commitments to achieving gender equality and improving health and wellbeing for all, quantitative data and methods to precisely estimate the effect of gender norms on health inequities are underdeveloped. Nonetheless, existing global, national, and subnational data provide some key opportunities for testing associations between gender norms and health. Using innovative approaches to analysing proxies for gender norms, we generated evidence that gender norms impact the health of women and men across life stages, health sectors, and world regions. Six case studies showed that: (1) gender norms are complex and can intersect with other social factors to impact health over the life course; (2) early gender-normative influences by parents and peers can have multiple and differing health consequences for girls and boys; (3) non-conformity with, and transgression of, gender norms can be harmful to health, particularly when they trigger negative sanctions; and (4) the impact of gender norms on health can be context-specific, demanding care when designing effective gender-transformative health policies and programmes. Limitations of survey-based data are described that resulted in missed opportunities for investigating certain populations and domains. Recommendations for optimising and advancing research on the health impacts of gender norms are made.

Neighborhood contributions to psychological distress among Latino New York City adults

OBJECTIVE

Neighborhood-level factors such as ethnic densities and social cohesion have been negatively associated with psychological distress among Latino Americans. Yet, existing evidence is based on either specific neighborhood factors or particular Latino subgroups. The objective of the study was to assess difference in psychological distress between each of four Latino subgroups (Puerto Ricans, Dominicans, Mexicans, other Latinos) and non-Latino white adults in New York City, and quantify total neighborhood-level influence on these differences.

DESIGN

We used the combined Community Health Survey data from 2009, 2010, and 2012 surveys. We estimated the odds ratios (ORs) for self-reported non-specific psychological distress (Kessler-6 questions scale ≥ 13) by race/ethnicity using logistic regression models. We further adjusted these estimates for both observed and unobserved neighborhood-level confounding using the conditional pseudolikelihood method for complex survey data.

RESULTS

Puerto Ricans were more likely to be psychologically distressed (OR = 1.58, 95% CI = 1.18, 2.12) compared with non-Latino whites, whereas the opposite was seen in other Latino subgroups. Accounting for full neighborhood-level confounding increased the disparity for Puerto Ricans (OR = 1.79, 95% CI = 1.26-2.54). For the other subgroups, lower odds of psychological distress were no longer observed or attenuated after accounting for neighborhood-level confounding. Overall neighborhood-level factors were associated with lower psychological distress at greater extent among all Latinos subgroups versus non-Latino whites in New York City.

CONCLUSION

Despite substantial variations of psychological distress across Latino subgroups, the study shows evidence that neighborhood-level factors might play a protective role in all Latino subgroups.

Balloon-assisted coil embolization of intracranial aneurysms is not associated with increased periprocedural complications

BACKGROUND

The balloon-assisted coil embolization (BACE) technique represents an effective tool for the treatment of complex wide-necked intracranial aneurysms; however, its safety is a matter of debate. This study presents the authors' institutional experience regarding the safety of the BACE technique.

METHODS

428 consecutive patients with 491 intracranial aneurysms (274 acutely ruptured and 217 unruptured) treated with conventional coil embolization (CCE) or with BACE were retrospectively reviewed. All procedure-related adverse events were reported, regardless of clinical outcome. Thromboembolic events, intraprocedural aneurysm ruptures, device-related complications, morbidity and mortality were compared between the CCE and BACE groups.

RESULTS

The total rate of procedural and periprocedural adverse events was 9.6% (47/491 embolizations). Thromboembolic events, intraprocedural aneurysmal rupture and device-related complications occurred in 2.4%, 3.9% and 3.3% of procedures, respectively. The risk of thromboembolic events and device-related problems was similar between the CCE and BACE groups. A trend towards a higher risk of intraprocedural aneurysm rupture was observed in the BACE group (not statistically significant). The total cumulative morbidity and mortality for both groups was 2.6% (11/428 patients) and there was no statistically significant difference in the morbidity, mortality and cumulative morbidity and mortality rates between the two groups.

CONCLUSION

In this series of patients with acutely ruptured and unruptured aneurysms, the BACE technique allowed treatment of aneurysms with unfavorable anatomic characteristics without increasing the incidence of procedural complications.