Dismantling Underlying Racist Narratives in Science: Healthfulness of the Southern Diet and a Deep Dive into Dietary Pattern Analysis

By Allison Lacko

A recent study published in the Journal of the American Medical Association stirred up controversy, finding that “Southern” style dietary patterns may be the most important lifestyle factor explaining the disparity in high blood pressure between Black and White populations. News headlines like “Southern diet blamed for high rates of hypertension among blacks” or “New research shows soul food is a major contributor to high blood pressure” forced the Black community into a familiar corner: defending their food heritage.

“Southern food” at once encapsulates the rich multicultural heritage of southern foodways as well as the legacies of the forced removal of Native Americans and slavery. In comparison, the diet the authors refer to was an empirically derived dietary pattern, which is quite different from a cultural understanding of southern food. Therefore, it’s worth taking a step back to understand how an empirical diet pattern is created. We’ll circle back to why their choice of names for this dietary pattern is problematic at the end of the post.

First, some background: What’s dietary pattern analysis?

Diets encompass the kinds of food we usually eat and are influenced by culture. Dietary patterns specifically describe the ways groups of foods are combined together, accounting for the quantity, proportion and frequency of consumption. In nutrition research, this allows us to look at whether an entire pattern of eating is associated with health outcomes, rather than looking at the effect of only one type of food or nutrient. For example, the Mediterranean dietary pattern is associated with living longer and disease prevention. In some studies, a template for a healthy dietary pattern is provided by the researcher, and people are given a score based on what they say they eat on a regular basis. The Healthy Eating Index is an example. In other cases, researchers do not come with a predefined idea of what pattern participant diets should be compared to. Instead, they look for common food groupings among their study participants and define new patterns specific to these individuals.

The same research team that wrote the hypertension paper published a study in 2015 where they identified dietary patterns for the same group of participants. We’ll refer to this 2015 study as the “REGARDS Dietary Pattern Derivation” study. The researchers asked participants how often they ate certain foods in the past year. Then they took the answers from all participants to see if there were any patterns, or foods that tended to be eaten together across people. For example, people who eat a lot of cereal tend to consume a lot of milk and people who eat a lot of rice also tend to eat a lot of beans/legumes, while people who drink a lot of soda tend to avoid vegetables. This process of identifying underlying patterns is called factor analysis.

In factor analysis, a statistical program identifies underlying food patterns. The researcher then decides how many patterns explain the most variation in all of the foods eaten by the study population. Within each pattern, the software gives foods different weights depending on how important the food was in characterizing the pattern (as an example, we will provide results from the “REGARDS Dietary Pattern Derivation” study later in this post). Each individual then receives a score that reflects how closely their consumption follows a given dietary pattern by multiplying their consumption of a food times its factor weight and summing these weighted scores across all foods. Not all foods are given a factor weight in each pattern – only some foods meaningfully contribute to defining each pattern.

Who was in the study?

It’s important to remember that these food patterns best explain the variation in foods consumed within a specific group of people – namely, the participants in the study who told researchers what they ate and how often they ate it. In this case, the participants were from racially and geographically diverse backgrounds. They were part of the Reasons for Geographic and Racial Differences in Stroke (REGARDS) study. Black and White individuals over age 45 were sampled from across the country, although the study was designed to oversample Black participants and people residing in the “stroke belt” (North Carolina, South Carolina, Georgia, Alabama, Mississippi, Tennessee, Arkansas, and Louisiana).  These participants filled out a Food Frequency Questionnaire, where they reported how often they consumed 107 different food items in the past year. In total, diet data was used from 21,636 participants. The study population is an important strength of the “REGARDS Dietary Pattern Derivation” paper, because few studies investigate dietary patterns in diverse populations. At the same time, this also helps us understand why the researchers chose the name “Southern” to label one of their dietary patterns.

What patterns did they find?

Now that we have some background on factor analysis and know that the researchers had a diverse study sample, which dietary patterns explained the most variation in what the study population ate? The researchers chose five distinct patterns of food that tended to be consumed (or avoided) together. According to the authors, these patterns were then named based on the foods that were most important to forming each group.

The table shows the five dietary patterns and the relative importance of the foods that defined them.  The “Southern” dietary pattern is highlighted. You can see that it is characterized by fried food, organ meat, processed meats, eggs, and the avoidance of low-fat milk.

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4429641/
All results are from the “REGARDS Dietary Pattern Derivation” study. Foods are listed from the most important weight to the least important weight, only those with a weight greater than 0.20 are included. Note that some of the weights are negative – this indicates that these foods tend to be avoided by people who consume the positively weighted foods.

So is a “Southern” diet really related to poor health?

To circle back to the controversy generated by this study, I would like to leave you with 3 important takeaways to help answer this question.

First (and longest), this study shows how our choice of words carries weight and can perpetuate racial stereotypes. While the statistical process that derived these five patterns was objective, the naming of dietary patterns was quite subjective. In naming the “Southern” dietary pattern, the authors stray from the naming convention used for the other four patterns. No other pattern uses a label that also has carries cultural weight. The authors did this on purpose, stating they chose the name “Southern” because “this diet is similar to the culinary pattern observed in the Southeastern US,” specifically citing the high weight of organ meats and the moderate weights for shellfish. However, organ meats are commonly eaten by many non-White cultures, and shellfish are similarly weighted in the “Convenience” and “Alcohol/Salads” patterns. Would the authors have come to the same conclusion if they had been blinded to the study participants, unaware of the racial and geographic composition of the sample?

While the authors may have made a well-intentioned effort to account for the context of their study, their use of “Southern” indicates an ignorance of the social history of US food (if you’d like to learn more, How America Eats by Jennifer Wallach is a great resource). As mentioned earlier, southern food has roots in Native American and Black cultures, and many of these foods are healthful (leafy greens, yams and sweet potatoes, corn, beans, rice, herbs and spices, fish, eggs and poultry). The increase in processed, high fat and high sugar foods is pervasive in many diets in the US.

To label an unhealthy dietary pattern a “Southern” diet perpetuates a racist narrative because southern food is commonly associated with Black culture. Although White people eat southern food, it was introduced to their diets in the Southeast through Black cooks, and their relationship to southern food cannot be separated from the exploitation of Black cooks and farmers (well into the 20th century). The idea that southern food is unhealthy perpetuates a legacy of cultural appropriation. In this case, the healthy foods from Black and Native American culinary traditions that White people approve of are excluded from the researchers’ definition of “Southern” food. In fact, green leafy vegetables tend to be avoided in this diet pattern (weight = -0.22). This limits the culinary contribution of Black and Native communities to those foods that are unhealthy.

This kind of racism is often not intentional, but it is important to call it out for what it is. This is especially important in science, which aims to describe an objective reality. If we are to overcome racist narratives, especially when studying disparities, we must engage stakeholders in affected communities as partners in all stages of research. Had they discussed their findings with stakeholders in the Black community, perhaps the research team would have realized that “Southern” diet was a poor name choice for an empirically derived dietary pattern. In alignment with the other four dietary patterns, I propose renaming the “Southern” dietary pattern to the “High fat and red meat” dietary pattern.

Second, the controversy around the healthfulness of “Southern” food detracts from the more important finding of this study: that a dietary pattern high in fried foods and processed meats is unhealthy. Understanding dietary patterns in this study population is especially important since the stroke belt has high rates of diet-related chronic diseases. In addition to hypertension, this “High fat and red meat” dietary pattern has also been associated with acute coronary heart disease, reduced cognitive function, all-cause mortality and stroke in this study population.

Third, the hypertension study illustrates that it’s important to understand how diet fits in to the broader conversation about health inequity. Diet is an important mediator between race and disease (a mediator is a factor along the A -> B pathway, in this case Race -> Hypertension). We could spend an additional post dissecting the mediation analysis that lead authors to attribute most of the disparity in hypertension to the “High fat and red meat” diet. It is important to note that differences in diet between racial groups may be due to differences food availability and affordability. In addition, while diet certainly impacts our health, it is important to examine which mediators are missing from their analysis, like systemic racial discrimination.

Wrap up

In conclusion, it is important to both investigate how what we eat impacts disease outcomes while also understanding the cultural context and traditions that affect food and health. As researchers, we have a responsibility to ensure that the questions we ask, methods we use and language we choose do not perpetuate structural racism by reinforcing problematic cultural representations and norms.

In our current political context, labeling an empirically derived, unhealthy dietary pattern “Southern” may seem like small fish to fry (yes, pun intended). To be very clear, I do not believe the authors harbored any malicious intent. Yet this paper is indicative of public health’s larger struggle to dismantle racial hierarchies in a field whose goal is to promote social justice.

If we are to reduce health inequities in a way that is evidence-based and empowers communities, we must have open dialogue about breaking down racial narratives when we see them and support each other in improving public health research.

Peer-reviewed by Adante Hart, Blaide Woodburn, Charlotte Lane, Tania Aburto Soto, Natalia Rebolledo Fuentealba, Melissa Jensen, and Yiqing Wang

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

Create a website or blog at WordPress.com

Up ↑

%d bloggers like this: