Quantifying Impact of Nutrition Policy on Population Health

After finishing my PhD I knew I wanted to leave the halls of academia and apply my newly honed expertise to real-world problems. As a research scientist at the New York City Department of Health and Mental Hygiene I felt I could do exactly that. This data story is about my most memorable project: the evaluation of a citywide nutrition policy that eliminated the use of trans fats in city restaurants.

Trans fat is a type of oil or fat which became immensely popular among food manufacturers in the 1950s. It was cheaper than butter, but worked just as well to optimize the taste and shelf-life of food products. Trans fats are naturally present in animal foods (e.g. dairy) in tiny amounts. They can also be manufactured industrially and became a ubiquitous presence in the grocery aisle for pastries, baked goods and fried foods (overview here).

But trans fats are bad for the heart. The evidence for the unhealthful connection between trans fat intake and heart disease has been supported by evidence drawn from all observational and experimental studies, making the link between trans fats and heart disease one of the most robust associations in the nutrition world.

Despite the evidence, removing this ingredient from supermarket shelves has taken decades. Since New Yorkers consume restaurant foods so frequently, restaurants were considered a major point of contact for reducing trans fat consumption. So in 2005, the NYC health department rolled out a public health campaign to engage restaurants in the bid to eliminate trans fats from their prepared foods. The education and technical assistance moved trans fat presence by zilch, nada, well closer to 1% point in the wrong direction-- 50% of restaurants were using trans fats prior to the campaign and 51% were using it in a year after. The encouragement to voluntarily change behavior was not working.

This led to a formal nutrition policy in 2007 which mandated the elimination of trans fats from all restaurant-prepared foods.

Absolutely, in all the ways you might expect. By 2008 it was evident that 99% of restaurants had complied with the policy. We could tell that restaurants stopped using trans fats but until our evaluation, we had no evidence that the ban was associated with decreased presence of trans fat in the body.

As lead analyst on this project, I was tasked with quantifying any change in trans fat levels biomarkers in the blood. I had access to trans fat biomarker data from a representative population health dataset from adult New Yorkers. Similar data had been gathered before (2004) and after (2014) the policy had been implemented. In both datasets, survey respondents had reported their frequency of dining out each week. Armed with this data, I could conduct a difference-in-difference analysis to see if the trans fat policy was effective. If the policy were impactful, 2014 trans fat intake should be lower than 2004\. Importantly, there also should be a differential association of trans fat consumption with restaurant frequency prior to the ban and that association should no longer be apparent after restaurants no ceased using trans fats. By measuring the difference in declines, pre- and post- the policy, I could examine whether the policy contributed to a change in trans fat biomarkers in NYC adults\!

The peer reviewed analysis was published in the American Journal of Public Health. Average trans fat intake dropped by 57% for all New Yorkers in 2014, but those who frequented restaurants the most, had a much larger decline (62%) than those who hardly ate out (51%).

The chart below, reproduced from our publication, shows that prior to the ban, frequent restaurant goers tended to have more trans fats in their blood, relative to those who hardly dined out (dotted line). By 2014 however, there was no apparent association between restaurant frequency and blood trans fat levels. Importantly, note the biggest decline was apparent in those who reported frequenting restaurants the most.

Our study provided a critical link in the causal chain of events between policy implementation, subsequent decline in restaurant foods, the decrease in circulating trans fats in the blood, and ultimate decline in heart disease-related morbidity and mortality. Other researchers compared hospitalizations and mortality rates in jurisdictions with similar nutrition policies relative to cities which did not have such mandates in place. Their work indicates that these policies contributed to fewer cardiovascular disease-related deaths and hospitalizations.

It was exciting for me to be involved in a project that tapped into my technical and substantive interests in nutrition, chronic disease, public policy, epidemiology and data science. Public reporting on this work is available in the New York Times and other nutrition and medical publications. I’m also thrilled that after decades of evidence, the rest of the US now has a healthier default supermarket as trans fats were finally banned at the federal level last year.