Lindsey Smith is a doctoral student in nutrition epidemiology at UNC, Chapel Hill.
Public health and nutrition researchers need to know what people eat in order to evaluate how diet affects health and life, and to make recommendations about health policies. That should be easy, right? We in a world where 3-D printers print pizza and our phones tell us which taco truck has the best carnitas.
But, as anyone who studies nutrition epidemiology can tell you, collecting data on diet is a tricky business. Each assessment method is prone to a set of biases that make it difficult to get an accurate read on the type and amount of food people eat. For example, the 24-hour recall, the method that is used by major national nutrition surveys, asks people to remember and report everything they ate in the last 24 hours. The problem? People forget. It can be difficult to remember every bite of food that crossed your lips, and even more difficult to figure out exactly how much you had.
And even when people do remember, they often lie. Social desirability bias is the tendency of participants to give researchers answers that are socially acceptable, and leads to underreporting of unhealthful foods or stigmatized behaviors like smoking or drinking. What’s worse is that this type of bias is often stronger amongst those with the very conditions we’re hoping to study—like type 2 diabetes or obesity—making it even more difficult to determine the truth from the data.
In China, the picture gets even more complicated. The China Health and Nutrition Survey is an ongoing cohort study beginning in 1989 that was designed to how understand the rapid social and economic changes in China have impacted the demography, diet, and health of the population. Although the survey employs gold-standard techniques of collecting diet data, including multiple consecutive 24-hour recalls and household inventories of the food supply, capturing all foods and beverages people consume is exorbitantly difficult.
For one thing, in China, people tend to eat meals family-style, where a number of dishes are shared between diners who eat from the common dishes using chopsticks, making it difficult to figure out just how much any one person ate. Secondly, nutrition information on foods eaten away from home is lacking, which poses an increasing problem as a growing urban population eats out more and more at the street vendors, small shops, fast food restaurants, and convenience stores dotting every corner.
Because of these issues, it’s unsurprising that researchers recently noticed an odd quirk in the data. Despite the rapid proliferation of packaged beverages like soda and sports drinks, the diet data recorded only very low levels of consumption of these new drinks. Because China’s diet patterns are Westernizing so rapidly, capturing accurate information on processed packaged foods like sugar-sweetened beverages is essential to understanding the link between these new diet patterns and health.
In response to this dilemma, we created a new diet assessment methodology to see if we could improve our ability to accurately record the new types of beverages people are drinking and how much of it. In collaboration with Edmund Seto and Jenna Hua, environmental health researchers at the University of California, Berkeley, and the staff at the Shanghai Center for Disease Control, we developed a smartphone-assisted 24-hour recall. In this methodology, participants use a Samsung smartphone to take video recordings of everything they eat and drink for three days. At the end of each day, the participant reviews the video with an interviewer to help boost their memory of what he or she consumed. In addition, a special app on the phone prompts participants to respond to a short survey every few hours, asking if they had anything to drink since the last survey. The phones also collect information about the participant’s location and physical activity, which allows us to assess their exposure to the food environment: what restaurants they encounter during their everyday life.
In March of 2013, I traveled to Shanghai with Jenna to work with our CDC collaborators to test the new method in a small group of young adults living in and around the city. We knew that, like all field-based researchers, we should expect the unexpected; as we encountered challenges like the outbreak of the bird flu in our study site, or the difficulty in getting internet-based apps to run despite the proverbial “Great Firewall”, we were almost forced to shut down the study. Luckily, we were able to collect data on over 100 participants.
We’re only now analyzing the data, but our hopes are high that this new technology may lead to a more accurate reflection of what people eat and drink. It might not be as fancy that pizza-slinging 3-D printer, but at least it’s a step in the right direction.