On July 14, NCCOR’s Connect & Explore Webinar Series discussed childhood obesity declines, disparities, and opportunities to reconsider the design and impact of policies and interventions.
Access the slides (PDF)
Speakers were Shiriki Kumanyika, emeritus professor of epidemiology at the University of Pennsylvania Perelman School of Medicine; chair of the African American Collaborative Obesity Research Network (AACORN); and president of the American Public Health Association (APHA) and Tim Lobstein, director of policy at the World Obesity Federation. During the webinar, there were a number of participant questions. Drs. Kumanyika and Lobstein have responded here.
Single strategies implemented in isolation seem less likely to achieve decreases in obesity across multiple target audiences. Are you aware of multilevel approaches implementing concurrent (two or more) strategies shown to be effective for higher and lower socioeconomic status (SES), and white and minority target audiences?
Dr. Lobstein: I agree that multi-component interventions should be better, on the principle that the more you can change the determinants of behavior the more you are likely to change the behavior itself. I cannot name a specific multi-component intervention shown to be effective for obesity reduction across socioeconomic status (SES) and ethnic strata, but I would refer you to the excellent review by Professor Sir Michael Marmot for the U.K. Government “Fair Society, Healthy Lives.”
Dr. Kumanyika: This question is difficult to answer because the number of relevant studies in which effectiveness can be compared across race/ethnicity and SES subgroups is quite limited due to study design or methodology. For a discussion of the status of this evidence see: Brennan et al, American Journal of Preventive Medicine, 2014;46(1):e1 to e16; Kumanyika et al, Obesity Reviews, 2014(Supp4);177-203; and Beauchamp et al, Obesity Reviews, 2015:15,541-554. These reviews identify at least a few studies that indicate similar effectiveness in lower- and higher-risk groups. Beauchamp et al conclude that community based interventions that address structural factors in environments work better or as well in lower vs. higher SES contexts.
The cross-sectional data shown during the webinar is very compelling. Is there any cohort data demonstrating trends in childhood obesity and if so, is that data similar or different?
Dr. Kumanyika: At the time of the webinar, I hadn’t looked into data that allow examination of weight trajectories over time within the same cohort. For the population trends we rely on the National Health and Nutrition Examination Survey (NHANES) repeated cross-sectional surveys of probability samples of the U.S. population. In that sense we can view each age group as a “feeder system” for the next oldest age group in later surveys. So when we look at what’s happening in 2- to 5-year-olds now and then we look four years from now at 6- to 9-year-olds, we can think of this as a continuation of the story for the current 2- to 5-year-olds.
As follow up to the webinar, I did identify some examples of analyses of racial/ethnic differences in weight trajectories in a specific cohort of children. See McCormick et al, Academic Pediatrics, 2014;12(6):639-645 for an analysis in 2- to 12-year-olds and Harris et al, Arch Pediatr Adolesc Med, 2009 Nov;163(11):1022-8, for an analysis of trajectories from adolescence to young adulthood.
Most of the statistics shown were for children older than age 2. Research has shown that breastfeeding has a protective effect on obesity. Yet again, the statistics show breastfeeding is least prevalent among lower-educated, communities of color, single women, and younger women. Could some of the disparities be reduced if breastfeeding rates were increased?
Dr. Kumanyika: The prevalence of high weight for recumbent length for children ages 0-2 in the 2011-2012 NHANES data is not significantly different by race/ethnicity. The prevalence is lowest for non-Hispanic white children compared to the other groups, but all estimates have very wide confidence intervals (see Ogden et al, JAMA, 2014;311(8):806-814). Nevertheless, as you suggest, increasing breastfeeding rates is considered to be an important potential strategy for prevention of overweight and obesity and is especially important for the demographic groups with relatively low breastfeeding rates.
We saw data that indicated black girls have the highest rates of obesity compared to white, Mexican, and other peers. But how do we take into account differences in body type and could that possibly be a confounder in the obesity data?
Dr. Kumanyika: Although on average there are differences by race/ethnicity in the association of body mass index (BMI) with percent body fat, the obesity prevalence differences are in some cases larger than one might expect based on measurement differences. Also relevant, the time trends I showed provide for within-group comparisons that would not be affected by ethnic differences in measurement. Several trends indicate increasing prevalence in black and Mexican American children in opposition to the trends in non-Hispanic, white children. For a discussion of the ethnic differences in the interpretation of BMI, see Freedman et al, Obesity, 2008;16:1105-1111. Body type differences (e.g., body fat distribution) are not reflected in the prevalence data. Data on type 2 diabetes incidence confirm that obesity is of relatively greater concern in black and Hispanic compared to non-Hispanic white children: 26.7/100,000 in African Americans, 17.2/100,000 in Hispanics, and 4.5/100,000 in non-Hispanic whites ages 10-19 years. See the Centers for Disease Control and Prevention’s Diabetes and Youth page.
Is there a guide for examining interventions from an evidence-based criterion approach?
Dr. Lobstein: There is a ‘101’ introduction to the issue available via HealthKnowledge, and three further useful guides from the World Health Organization, European Commission, and the National Obesity Observatory. Remember that public health interventions should not be assessed in the same way that medical interventions are. See, for example, Rychetnik, Frommer, Hawe, and Shiell (2002).
Why does the socioeconomic status obesity gap in England seem to be wider than its European counterparts?
Dr. Lobstein: I’m not sure that they are. The difference in child obesity levels in families defined according to the education of parents was quite wide in some countries and not so wide in others and we didn’t even have the United Kingdom on that graph. I think one of the reasons why the U.K. data shows such a clean, smooth gradient is because it includes all children. Nearly all the other data you get are samples and often you get a bit of bias creeping in where some of the children don’t want to be measured or some of the parents don’t want their children to be measured. The beauty of the U.K. data is that it includes 97 percent of children in the age group. And there are other reasons why countries differ: For example, the overall level of inequality of income in a country has a determining factor. Countries with higher income inequality tend to have higher rates of poor health, including higher rates of obesity, diabetes, and so on. I think we can show the same for U.S. states; those with greater disparities of income also have higher rates of poor health. Norway versus Greece has hugely different levels of childhood obesity. What’s happening there? That sort of disparity is extraordinary and well worth looking at since it must indicate some fairly significant drivers. I think you could say the same for states in America comparing low rates in Colorado and California, to the high rates of obesity in parts of Mississippi. Why have we got these extraordinary differences and what does it tell us about the drivers of obesity in society?
What are other examples of universal proportionality, other than TV ads?
Dr. Lobstein: Put simply, the measures that are both universal and proportionate are those measures that apply to everybody but which are particularly beneficial for those most in need. I haven’t had a chance to do the research I suggested in the webinar to evaluate a wide range of interventions. But perhaps any intervention that promotes widely a ‘good practice’ which is currently poorly taken up among the most in need would be a start. For example, promotion of baby-friendly hospitals to help initiate breastfeeding, and promotion of maternity support in employment law to ensure breastfeeding can continue for at least six months without substantial financial penalty. If everyone has access to kindergartens, then food policies providing good nourishment for infants would be valuable, as of course are school food policies which provide low-cost good nourishment to all. Perhaps you can think of more!
How do we address communities that may not prioritize weight or obesity? What are we to do being that such communities may still affect health care, etc.?
Dr. Lobstein: I guess the answer has to be: go in and talk to them! We have to work with communities and that means addressing the problems that they themselves think that they have before trying to solve the problems you think they have.
How would you regard targeting a low-income population with a variety of interventions in a setting? For example, what about designing a program that addresses unhealthy choices in a way that gives options, support, and education (i.e., programs, community sessions and events, classes) to the full community, thereby reducing the feeling or concept of separation?
Dr. Lobstein: Sounds good in principle, but similar to the previous comment, you need to check out that the beneficiaries are part of the design of the intervention, so that it meets the needs that they themselves recognize. And you might want to look at the literature on building ‘social capital’—building the resources and networks that communities can use to support themselves and improve their own well-being.