1 What are the different ways to account for SES in an analytic model when investigating racial/ethnic health disparities? (Hint: you should have three options). Discuss the interpretations/implications of each approach as it relates to the interest in understand health disparities by race/ethnicity. Draw a DAG for each option and reference it in your response (you do not have to post this!).
- Regression analysis and adjusting for confounding. The authors in the Hasden et al paper used generalized estimating equations, which is a form of regression accounting for clustering. They then analysed for crude associations and adjusted associations controlling for relevant covariates, and interaction models to determine whether racial/ethnic differences in GWG varied by pre-pregnancy weight class – this method assumes that the confounding covariates are all independent. After reading the Merlo et al paper- I thought that the multi-level regression analysis would have been better suited for this paper particularly because they found that disparities in inadequate GWG appears to vary by pre-pregnancy weight class – showing that there was some effect modification going on. Given the fact that these factors were not independent- it is likely that in just adjusting for clustering, assessing for interaction – there was still residual confounding
- Mediation analysis as described in the article by Lorch et al. The authors conducted a retrospective study and sought to determine the importance of socioeconomic factors, maternal comorbid conditions, antepartum and intrapartum complications of pregnancy, and fetal factors in mediating racial disparities in fetal deaths. They found that the factors that mediate racial/ethnic disparities in fetal death differ depending on the racial/ethnic group.
- Multi-level regression analysis or Multiple-group path analysis as described in the paper by Merlo et al. This method demonstrated it was possible to provide relevant epidemiological information just by observing how differences in systolic blood pressure were partitioned between the individual level and the neighbourhood level. The authors were able to demonstrated using both statistical clustering and the social epidemiological concept of contextual phenomenon demonstrate that people from the same neighbourhood are more similar to each other than to people from different neighbourhoods with respect to the health outcome variable. (In my little epi knowledge- I know that one of the disadvantages for ecological studies was the ecological fallacy and then also the - - and I am unable to make the connection as to whether the authors imply that this methodology can be used to reduce this)
- Structural equation modeling
2 Think about multilevel influences on a health outcome of interest to you. Discuss how you would study this, including measurement and analytic approaches you would use to account for exposures across multiple levels.
I am interested in adolescent psychosocial functioning and sexual reproductive health outcomes in areas with high HIV prevalence and incidence. Food insecurity, poverty and parental HIV/AIDS have been identified as important drivers of HIV risk and vulnerability among AGYW.1-3 Food insecure and impoverished girls have been shown to engage in behaviours that put them at risk for HIV infection.1,2,4-7 Studies have shown that food insecurity is associated with inconsistent condom use, transactional sex, intergenerational sex, multiple concurrent sexual partners, forced sex, sexually transmitted infections (STIs) and HIV infection.6,8-13 Adolescents residing with an adult parent/guardian who is chronically ill due to HIV/AIDS are at particularly high risk of living in a food insecure and impoverished household, which may further enhance vulnerability to sexual risk behaviours with negative outcomes. Several studies have found higher rates of emotional and behavioural problems, and higher sexual risk among uninfected adolescents with HIV-infected parents as compared with adolescents with uninfected parents.14-20
My project is looking at the impact of a household level income generating agricultural intervention delivered on sexual behavior, psychological well being, and use of Sexual Reproductive Health (SRH) and HIV services among adolescent girls and young women aged 15 to 21 years. I believe that the intervention will work through a number of pathways mainly 1) improving food security, 2) increasing household wealth, 3) increasing school attendance, 4) gender empowerment and 5) improving intra-household functioning
My study is nested within a larger cluster randomised trial. I plan to survey the AGYW (in both control and intervention arm) at baseline and endline using a set of standardised scales, qualitative methods and participant observation. For my quantitative analysis, I plan to use the following methods (which I am still learning) include
- Assessing the direct and indirect intervention effects using structural equation modeling to examine pathways from the intervention through baseline-to-endline changes in mediating outcomes to changes in primary health outcomes.
- I believe that while food security and household wealth are important proximal factors – that the key mediators are intra household functioning and parental stress and mental health as well as school attendance – for these I will do mediation analysis using a causal inference approach that is able to yield estimates of indirect effects in the presence of non-continuous outcomes, interactions, and clustered data – given that we have 16 clusters.
- I am a little confused but Randomisation is supposed to distribute confounding covariates equally between the intervention and control. However, because the primary intervention was developed to target adult parents and children less than 5 years of age, I am convinced that there may be some residual confounding and as such there is definitely a role for multivariable regression analysis adjusting for confounders. Undeniably in looking at the relationship between food insecurity and household wealth and adolescent SRH and psychosocial outcomes- there are multiple and varied confounders such as age, education
- I am also intrigued by the multilevel regression analysis suggested by Merlo et al. The authors note that the assumption made in usual regression analyses is the independence of individual measures. If this assumption is violated, the results of the regression analysis are biased. It is very likely that my variables are not independent. There is also the obvious fact that my study has 16 clusters that are clustered around geographic regions. As such – it looks as if this multi-level regression analysis rather than a simple multivariate regression analysis adjusting for clustering. My main question would be how this multi-level regression analysis differs from generalized estimating equations or mixed effects modeling – I am not sure I get the difference. Are they one and the same thing?
1. Weiser SD, Leiter K, Bangsberg DR, et al. Food insufficiency is associated with high-risk sexual behavior among women in Botswana and Swaziland. PLoS Med. Oct 2007;4(10):1589-1597; discussion 1598.
2. Miller CL, Bangsberg DR, Tuller DM, et al. Food insecurity and sexual risk in an HIV endemic community in Uganda. AIDS and behavior. Oct 2011;15(7):1512-1519.
3. Pascoe SJS, Langhaug LF, Mavhu W, et al. Poverty, Food Insufficiency and HIV Infection and Sexual Behaviour among Young Rural Zimbabwean Women. PloS one. 2015;10(1):e0115290.
4. Krishnan S, Dunbar MS, Minnis AM, Medlin CA, Gerdts CE, Padian NS. Poverty, gender inequities, and women's risk of human immunodeficiency virus/AIDS. Annals of the New York Academy of Sciences. 2008;1136:101-110.
5. Rollins N. Food insecurity--a risk factor for HIV infection. PLoS Med. Oct 2007;4(10):1576-1577.
6. Weiser SD, Young SL, Cohen CR, et al. Conceptual framework for understanding the bidirectional links between food insecurity and HIV/AIDS. Am J Clin Nutr. Dec 2011;94(6):1729S-1739S.
7. Tsai AC, Weiser SD. Population-based study of food insecurity and HIV transmission risk behaviors and symptoms of sexually transmitted infections among linked couples in Nepal. AIDS and behavior. Nov 2014;18(11):2187-2197.
8. Palar K, Kushel M, Frongillo EA, et al. Food Insecurity is Longitudinally Associated with Depressive Symptoms Among Homeless and Marginally-Housed Individuals Living with HIV. AIDS and behavior. Aug 2015;19(8):1527-1534.
9. Palar K, Laraia B, Tsai AC, Johnson MO, Weiser SD. Food insecurity is associated with HIV, sexually transmitted infections and drug use among men in the United States. AIDS (London, England). Jun 1 2016;30(9):1457-1465.
10. Vogenthaler NS, Kushel MB, Hadley C, et al. Food insecurity and risky sexual behaviors among homeless and marginally housed HIV-infected individuals in San Francisco. AIDS and behavior. Jun 2013;17(5):1688-1693.
11. Tsai AC, Bangsberg DR, Frongillo EA, et al. Food insecurity, depression and the modifying role of social support among people living with HIV/AIDS in rural Uganda. Social science & medicine (1982). Jun 2012;74(12):2012-2019.
12. Tsai AC, Hung KJ, Weiser SD. Is food insecurity associated with HIV risk? Cross-sectional evidence from sexually active women in Brazil. PLoS Med. 2012;9(4):e1001203.
13. Tsai AC, Leiter K, Heisler M, et al. Prevalence and correlates of forced sex perpetration and victimization in Botswana and Swaziland. American journal of public health. Jun 2011;101(6):1068-1074.
14. Breuer E, Myer L, Struthers H, Joska JA. HIV/AIDS and mental health research in sub-Saharan Africa: a systematic review. African journal of AIDS research : AJAR. Jun 2011;10(2):101-122.
15. Sherr L, Mueller J. Where is the evidence base? Mental health issues surrounding bereavement and HIV in children. Journal of Public Mental Health. 2009;7(4):31-39.
16. Sherr L, Varrall R, Mueller J, et al. A systematic review on the meaning of the concept 'AIDS Orphan': confusion over definitions and implications for care. AIDS Care. May 2008;20(5):527-536.
17. Guo Y, Li X, Sherr L. The impact of HIV/AIDS on children's educational outcome: a critical review of global literature. AIDS Care. 2012;24(8):993-1012.
18. Operario D, Underhill K, Chuong C, Cluver L. HIV infection and sexual risk behaviour among youth who have experienced orphanhood: systematic review and meta-analysis. Journal of the International AIDS Society. 2011;14:25.
19. Brackis-Cott E, Mellins CA, Block M. Current Life Concerns of Early Adolescents and Their Mothers:: Influence of Maternal Hiv. The Journal of Early Adolescence. February 1, 2003 2003;23(1):51-77.
20. Olayo MMR. Household HIV/AIDS status and sexual debut among adolescents in Kenya. African Population Studies. December 2011 2011;25(2):457-470.