Diet Tracking Does Not Need To Be Perfect

Diet tracking or journaling can be very helpful for a weight loss journey, but keeping an accurate tally of everything you put into your mouth 24/7 can be harder than you think, especially when you take into account the occasional between-meal snack, midnight snacks, or samples at the grocery store. Most health experts agree that diet tracking/journalling can be an important part of successful weight loss, but this study suggests that it does not have to be 100% accurate/perfect to achieve significant weight loss. 

For this study, 153 people who were enrolled in a weight loss program that included journaling their food intake were tracked for 6 months to ascertain what the optimal accuracy thresholds were for diet tracking to predict 3%, 5%, and 10% weight loss during the study period. 

“We partnered with WeightWatchers, who was planning on releasing a new Personal Points program, and they wanted to get empirical data via our clinical trial,” says co-author and Department of Allied Health Sciences Professor Sherry Pagoto. “Dietary tracking is a cornerstone of all weight loss interventions, and it tends to be the biggest predictor of outcomes. This program lowers the burden of that task by allowing zero-point foods, which do not need to be tracked,” she continues.

The researchers wanted to find ways to help make the process of tracking food to be less burdensome, as many programs seem to have people counting calories on an endless loop that could potentially last for the remainder of their lives. “That’s just not sustainable. Do users need to track everything every single day or not necessarily,” said Professor Pagoto. 

After 6 months the data was analyzed using a method called receiver operating characteristics curve analysis to see if any patterns emerged that could be associated with successful weight loss, and using the ROC method the researchers discovered the number of days that people need to track their food/drink intake in order to achieve clinically significant weight loss. The researchers note that the results are promising when you take into account the goal of a 6-month program is typically 10%, which is within the range where one would expect to begin to see health benefits that have been demonstrated in clinical trials. 

“It turns out, you don’t need to track 100% each day to be successful,” says Assistant Professor in the Department of Allied Health Sciences Ran Xu. “Specifically in this trial, we find that people only need to track around 30% of the days to lose more than 3% weight and 40% of the days to lose more than 5% weight, or almost 70% of days to lose more than 10% weight. The key point here is that you don’t need to track every day to lose a clinically significant amount of weight.”

“A lot of times people feel like they need to lose 50 pounds to get healthier, but actually we start to see changes in things like blood pressure, lipids, cardiovascular disease risk, and diabetes risk when people lose about 5-to-10% of their weight,” Prof. Pagoto says. “That can be accomplished if participants lose about one to two pounds a week, which is considered a healthy pace of weight loss.”

When the diet trajectories were tracked over the 6-month period, it led to identifying 3 distinct trajectories: 

• High trackers or Super Users are those who tracked food on most days of the week and lost on average 10% of their weight;
• A group that started out tracking regularly but their tracking declined over time to be only 1 day a week by the fourth month and lost an average of 5% of their body weight;
• And the third group called low trackers began by tracking/journalling 3 days a week then dropped to not tracking at all by the third month and lost an average of  2% of their body weight. 

“One thing that is interesting about this data is, oftentimes in the literature, researchers just look at whether there is a correlation between tracking and overall weight loss outcomes. Ran took a data science approach to the data and found there is more to the story,” Pagoto says. “Now we’re seeing different patterns of tracking. This will help us identify when to provide extra assistance and who will need it the most.”

“For me, what’s exciting about these digital programs is that we have a digital footprint of participant behavior,” says Xu. “We can drill down to the nitty-gritty of what people do during these programs. The data can inform precision medicine approaches, where we can take this data science perspective, identify patterns of behavior, and design a targeted approach.”

“Before, it felt like we were flying in the dark or just going by anecdotes or self-reported measures, but it’s different now that we have so much user data. We need data science to make sense of all these data. This is where team science is so important because clinical and data scientists think about the problem from very different perspectives, but together, we can produce insights that neither of us could do on our own. This must be the future of this work,” says Pagoto.

Xu agrees: “From a data science perspective, machine learning is exciting but if we just have machine learning, we only know what people do, but we don’t know why or what to do with this information. That’s where we need clinical scientists like Sherry to make sense of these results. That’s why team science is so important.”