Increase in number of eating occasions, more than increases in food portion sizes or energy density per meal responsible for increased energy intake at the population level?

We know by now that an increase in calorie consumption since the 1970s by about 500 Calories per day for Americans is primarily fueling obesity.  Physical activity decline at the population level is a lesser contributor (and it is contentious if the ‘decline’ is significant).  For references, see those from Yoni’s presentation in this post.  What is lesser known at this point is why people are eating more- what factors have the most influence on how much we eat per day?  There are many theories, and an important new paper in PLoS Medicine by Kiyah Duffey and Barry Popkin (whose credentials are impressive) gives us more insight into the relative contributions of some of them (and they are a bit counter-intuitive to conventional wisdom).

They identified 3 main theories that obesity research centralizes on (quoted from the paper):

  • increases in the frequency of eating/drinking occasions, especially snacking (EO)
  • increases in the typical portion sizes of foods and beverages (PS)
  • changes in the energy density of the foods consumed (termed “volumetrics” by Rolls and colleagues) (ED)

Each of these have small supportive bodies of research, but they have not yet been studied together.  So this study looked at relative contributions of changes in EO frequency, and PS and ED of the EOs to changes in total energy (TE) intake in adults between 1977 and 2006 in the U.S.

To cover this timeline, they used 4 US food surveys: the Nationwide Food Consumption Survey (NFCS; 1977-78), the Continuing Survey of Food Intakes of Individuals (CSFII; 1989-91 & 1994-96), and 2 NHANES (2003-2006) surveys.  Between these surveys, the sample size was 46,085 people.

Here is what they found after analysis (foods/beverages were grouped using the UNC-CH food-grouping system developed by Popkin, Haines, and Siega-Riz, and they used mathematical decomposition- see the open access paper for details):

  • Total energy intake increased by 570 Calories per day between 1977-78 and 2003-06, importantly corroborating other research with independent data sets.  Notably, an increase in 229 of these Calories per day came in only a decade- 1994-98 to 2003-06 data)
  • Portion sizes increased the greatest between 1977-78 and 1989-91 (+49 grams per EO), lesser between 1989-91 and 1994-98 (+18 g), then actually declined in 2003-06 (-2 g).
  • Energy density per EO interestingly did not change between 1977-78 and 1989-91, then declined between 1989-91 and 1994-98 (-0.02 Cal/g/EO).
  • Total number of EO increased from 3.8 EO/d in 1977-78 to 4.9 in 2003-06.

They zeroed in on food versus beverages;

  • PS of beverages increased more than foods, though they point out that food still contributed more to the increase in total energy intake (367 Cal/day for food versus 203 Cal/day from beverages) from 1977-78 to 2003-06.

Having trouble visualizing all this?  Luckily they provided a nice chart (light gray is EO, dark gray PS, white ED):

This clearly shows that the largest contributor to the annual increase in Calorie consumption was an increase in eating occasions (22 Cal/day/year) with portion size contributing less (10 Cal/day/year).  Surprisingly, there seems to be a decrease in the contribution of energy density per EO to total energy intake- it slightly offset the increase.

There are of course limitations to using data sets from different surveys (with slightly different methodologies) and those inherent to 24-hour recalls themselves, but a very interesting paper.


Duffey KJ, & Popkin BM (2011). Energy Density, Portion Size, and Eating Occasions: Contributions to Increased Energy Intake in the United States PLoS Med : 10.1371/journal.pmed.1001050

  • Stephan Guyenet

    That increase in calorie intake is just striking.  I’m not sure I believe the magnitude of it.  570 extra calories per day means people should be gaining more than a pound of extra fat per week, 52 pounds per year, etc.  That’s assuming energy expenditure doesn’t increase in response, which it does.  But I can’t imagine it could increase enough to offset most of that 570 kcal per day. 

    Either the self-reported dietary data that paper is based on are way off, or you can’t calculate weight changes simply by knowing how much food a person is eating (because the body mounts a homeostatic response that increases energy expenditure to oppose weight gain).  My opinion is that both are true to some extent.

  • Robin Judice

    Interesting in light of recent nutritionists’ recommendations to “graze” with small meals throughout the day. Wasn’t the theory that you would keep your metabolism up with “grazing” and maintain insulin levels?  

    Sounds like my Dad (who at 67 is the same weight he was a 30~slim) may actually  be doing the right thing for him with his ONE big meal a day (dinner only).

    I love your blog.

  • Pingback: Weekend Link Love - Edition 149 | Mark's Daily Apple

  • Pegheinemancihocki

    None of this explains why people are eating bigger portions more often, which to me is more to the point.  Gary Taubes does in “Why We Get Fat”: a diet too high in carbohydrates and too low in fat, thanks to the anti fat campaign that started with Ancel Keys and his bogus 7 countries study.  Eat more fat and you will need fewer meals, fewer fattening carbohydrates,  and smaller portions. 

    • Colby

      There are 10s of factors that have been identified by research that increase food consumption, and future research will give us a better idea which are contributing most.  Obesity is clearly multi-factorial and reducing obesity to 1 variable ignores a lot of research.  The insulinocentric theory of obesity is mostly confined to a few people and the internet. 

  • Pingback: Country hip hop! «

  • Pingback: CrossFit 312 » Blog Archive » 07.17.11