Deep divefrom Episode 6Fat loss11 October 202621 min

The Dietitian, the App and the Fitbit

A deep dive into one paper: what was known before, what the researchers did, what the numbers say and what they don't.

The paper: Impact of Dietetic Counseling Combined With Digital Tools on Obesity-Related Indices in Greek Adults: The GATEKEEPER Study (Obesity)

MP3

Chapters

Transcript

The script the two AI voices read from. The times are worked out from the script, not heard, so a line can start a few seconds early or late.

Read the whole episode

Cold open

Host 1You heard the trailer — this is the whole story. Welcome to the Hastrup Review of Science — where artificial intelligence reads real science, so you don't have to pretend you understood the paper. One paper today: nearly a thousand Greek adults, forty dietitians, a web platform and a box of fitness trackers.

Host 2The paper is called Impact of Dietetic Counseling Combined With Digital Tools on Obesity-Related Indices in Greek Adults: The GATEKEEPER Study. It is by Rafaela Makri, Odysseas Androutsos, Yannis Manios and colleagues at Harokopio University in Athens and the University of Thessaly, published this September in the journal Obesity.

Host 1And the claim, in one sentence?

Host 2Over three months, monthly face-to-face sessions with a dietitian were followed by smaller waists, lower weight and better blood sugar, and adding a web platform or a fitness tracker on top did not improve any of that. The tracker group did report a little more physical activity.

Host 1So the gadgets lost to the human, at least on the waistline. Let's see how firmly. First, why is anyone measuring waists at all?

Why waists matter

Host 2As established background, doctors noticed through the twentieth century that a handful of problems tend to travel together: belly fat, high blood pressure, high blood sugar and an unhealthy pattern of blood fats. In the late nineteen-eighties the endocrinologist Gerald Reaven called the cluster syndrome X, and it later became known as metabolic syndrome.

Host 1How is it defined?

Host 2The paper uses the usual working definition: you have metabolic syndrome if you have at least three of five things. A large waist, high fasting glucose, high triglycerides, low HDL cholesterol, or high blood pressure.

Host 1Why does the waist get so much attention among those five?

Host 2Because fat stored around the organs, visceral fat, behaves differently from fat under the skin. The authors describe it as the key driver, pushing the body toward insulin resistance and inflammation, and waist circumference is the cheapest everyday stand-in for how much of it you carry.

Host 1How common is this?

Host 2The paper cites roughly one in four adults worldwide. For Greece, older population studies put it around twenty-four percent, rising with age, and one primary care study in Crete found about seventy-four percent among older, higher-risk patients.

Host 1Greece being the home of the Mediterranean diet.

Host 2Which is part of the authors' point. They note Greece has high obesity rates and has been drifting away from the traditional Mediterranean pattern, so a country famous for a healthy diet now has a sizeable midlife metabolic problem.

What was known

Host 1What did we know before this paper?

Host 2As established background, the strongest evidence for lifestyle change comes from large prevention trials around the turn of the millennium. The Finnish Diabetes Prevention Study and the American Diabetes Prevention Program both found that structured diet and exercise counselling cut the rate of new type two diabetes by about fifty-eight percent in people at high risk.

Host 1So talking to people about food works.

Host 2Structured, repeated, personal counselling works, on average and with effort. The catch has always been cost and reach, because a trained professional's time is expensive, and that is exactly where digital tools came in.

Host 1What was the promise of the digital side?

Host 2Self-monitoring. Logging what you eat, counting steps, weighing yourself regularly, getting feedback and reminders between appointments. The authors cite meta-analyses showing app-based programmes give modest but real improvements in weight, mostly when combined with professional support.

Host 1Modest but real. Has anyone found the opposite?

Host 2Yes, and this is useful background for today. In twenty sixteen the IDEA trial in the United States, led by John Jakicic and published in JAMA, gave young adults in a weight-loss programme a wearable activity monitor, and after two years they had lost less weight than the group without one, about three and a half kilograms against nearly six.

Host 1So adding a gadget has already backfired once.

Host 2Once, in one population, over a much longer period. It showed that more data about yourself does not automatically mean more behaviour change.

Host 1Where does the GATEKEEPER name come from?

Host 2GATEKEEPER is a European Union Horizon twenty twenty project on technology for healthy ageing, with pilot sites in several countries. This paper reports what the authors call Greek Use Case one, and an earlier paper from the same trial already reported effects on blood pressure and blood vessels.

The open question

Host 1So the open question here is?

Host 2If you already give people at risk a monthly session with a dietitian, does adding a digital platform, or a platform plus a tracker and a smart scale, shrink their waists further? And does that hold in adults over fifty-five, in a country where the authors note digital literacy among older people is below the European average?

Host 1What did the authors expect?

Host 2They hypothesised that the digital tools would help. Their stated expectation was that better self-monitoring and personalised feedback would mean better adherence, and therefore bigger reductions in waist size and better blood sugar than standard care.

Who took part

Host 1Who took part?

Host 2Adults aged fifty-five or older, living independently, with at least one risk factor for metabolic syndrome. That could be a large waist, more than ninety-four centimetres for men or eighty for women, or raised triglycerides, low HDL cholesterol, raised fasting glucose, or blood pressure of one hundred thirty over eighty-five or higher.

Host 1Only one risk factor needed?

Host 2Only one, so not everyone had metabolic syndrome itself. In practice they were a heavy group: average body mass index about thirty-three, which is in the obese range, and an average waist of around one hundred eight to one hundred ten centimetres.

Host 1How many people?

Host 2Twelve hundred were screened and nine hundred eighty-four enrolled. Seventy-four percent were women and twenty-six percent men, recruited through private dietetics practices, community centres for older people, municipal health centres and hospitals.

Host 1Where in Greece?

Host 2About seventy percent in the Attica region around Athens and thirty percent in Central Greece, across ten cities including Larissa, Volos, Ioannina and Trikala. That split was chosen to mirror the twenty twenty-one census.

Host 1Who was left out?

Host 2People with dementia or cognitive impairment, severe hearing or vision problems, conditions that would stop them taking part, anyone living in an institution, and anyone already in another trial.

The three arms

Host 1How were they split up?

Host 2Randomly, one to one to one, using a computer-generated sequence made by independent staff. Three hundred forty-four went to standard care and three hundred twenty to each of the two digital arms.

Host 1What did standard care mean?

Host 2Monthly face-to-face sessions with a dietitian for three months: a dietary assessment, nutrition education, lifestyle advice and an individual diet plan, following Greek national guidelines for people with these risk factors.

Host 1And the second arm?

Host 2The same counselling plus access to a web platform built by the Centre for Research and Technology Hellas. On it you could log your meals, set goals, plan meals, book appointments, get personalised recommendations and message your dietitian, who could send reminders and encouragement back.

Host 1And the third arm gets the gadgets.

Host 2The counselling, the platform, plus a Fitbit Inspire two wristband for activity and sleep and a Fitbit Aria Air digital scale. The device data synced automatically to the platform, so the dietitian could see it.

Host 1Who delivered all this?

Host 2Forty dietitians, thirty in Attica and ten in Central Greece. The whole intervention lasted three months, give or take a week.

Host 1So everyone got the human. The question is purely what the technology adds.

Host 2Exactly, and that is worth holding on to. There is no arm without counselling, so the trial can compare the three arms with each other but cannot say how much of anyone's improvement came from the counselling itself.

The measurements

Host 1How did they measure the outcomes?

Host 2The main outcome, set in advance, was change in waist circumference. It was measured standing, at the natural waist between the lowest rib and the top of the hip bone, with a non-stretch tape at the end of a normal breath out, taking the average of three readings.

Host 1And the rest?

Host 2Weight and height for body mass index, and bioelectrical impedance for body fat percentage, muscle mass and a visceral fat score. For blood sugar, fasting glucose and H b A one c, which reflects average blood sugar over the previous two to three months.

Host 1What is bioelectrical impedance, briefly?

Host 2As background, it passes a tiny electrical current through the body and estimates fat and lean tissue from how easily it flows, since water-rich muscle conducts better than fat. It is cheap and quick, but it is an estimate, sensitive to hydration, and it varies between machines.

Host 1And physical activity?

Host 2A questionnaire, the short International Physical Activity Questionnaire, where people report how many days and minutes they walked, did moderate and vigorous activity in the past week. Each kind is weighted by intensity and summed into what are called MET minutes per week.

Host 1So activity is self-reported.

Host 2Entirely self-reported, which will matter later.

The analysis

Host 1How did they analyse it?

Host 2With linear mixed-effects models, which track each person's measurements over time and compare how the groups changed, adjusted for age and sex. The key number is the difference between groups in how much they changed, with a ninety-five percent confidence interval.

Host 1What about people who dropped out?

Host 2Thirty people were excluded because their baseline data were incomplete, leaving nine hundred fifty-four in the analysis. People who missed the follow-up stayed in, under the assumption that their missing values were missing at random, meaning not related to how they were doing.

Host 1Is that a safe assumption?

Host 2It is a standard one, but untestable. If people who were struggling were the ones who stopped showing up, it would flatter every group, and the text we have does not give follow-up numbers per arm.

Host 1Were the groups well matched at the start?

Host 2Not perfectly, which the authors report openly. The standard care group was older, about sixty-four against sixty-one or sixty-two, and less educated: about forty-nine percent had more than twelve years of education, against sixty-two and sixty-three percent in the digital arms.

Host 1Anything else?

Host 2The digital arms also reported more physical activity at the start, and the standard care group had a slightly higher body fat percentage. The authors ran a sensitivity analysis also adjusting for education and baseline activity and say the results held, but those numbers are not shown.

Results: the waist

Host 1Now the results. The waist.

Host 2Everyone's waist got smaller. The standard care group lost six point three centimetres on average, the platform group four point nine, and the platform plus devices group four point seven.

Host 1So the human-only group did best.

Host 2Yes. The between-group difference was about one point four centimetres in favour of standard care over the platform group, with a confidence interval from about zero point four to two point four, and about one point six centimetres over the devices group, interval roughly zero point six to two point six.

Host 1Is one and a half centimetres a lot?

Host 2It is statistically clear but small: less than the width of a finger, on a waist of about one hundred ten centimetres. The honest summary is that the digital arms did not do better and did slightly worse on this measure.

Results: weight and body fat

Host 1Weight?

Host 2Body mass index fell by between one point two and one point seven units across the groups. By my own rough arithmetic, at the average height of about one metre sixty-three, the standard care drop works out to roughly four and a half kilograms.

Host 1And between groups?

Host 2Standard care again lost a bit more than the platform group, about zero point four units, while the devices group was not clearly different from standard care. Body fat percentage fell by a little over two points in every group, with no meaningful differences between them.

Host 1The visceral fat score?

Host 2It fell in all three arms. Standard care beat the platform group by about a third of a point, a small difference, and was essentially level with the devices group.

Host 1Did people lose muscle along with fat?

Host 2Very little. Changes ranged from minus zero point two to minus zero point four kilograms, and the authors describe muscle as stable. The devices group's small drop was itself statistically distinguishable from zero, but no group differed from another.

Results: blood sugar

Host 1Blood sugar.

Host 2Fasting glucose fell in all groups: about eight milligrams per decilitre in standard care, seven in the platform group and five in the devices group. None of the between-group differences were statistically clear.

Host 1And H b A one c?

Host 2Down by zero point two nine percentage points in standard care, zero point two two with the platform and zero point one eight with devices, from a starting point of about six percent. Again, no clear differences between groups.

Results: physical activity

Host 1So the one place the technology won?

Host 2Self-reported physical activity. All groups increased, but the platform plus devices group went up by about ninety-five MET minutes a week, against thirty-nine in standard care. The difference, about fifty-five MET minutes, had a confidence interval from fifteen to ninety-six.

Host 1Translate fifty-five MET minutes for me.

Host 2Using the questionnaire's own weighting, where brisk walking counts about three point three, fifty-five MET minutes is roughly seventeen extra minutes of walking per week. That is my conversion, not the paper's, but it gives the scale: a couple of minutes a day.

Host 1And the platform without the wristband?

Host 2No clear difference from standard care on activity. Sitting time fell by between about forty minutes and an hour a day in all groups, with no differences between them.

A word from our sponsor

Host 1Before the big picture, a quick word from our sponsor.

Host 2Does your beard feel dry, wiry or hard to handle by the end of the day? Then you are in luck. The Hastrup Review of Science is brought to you by beard oil.

Host 2A blend of light plant oils that softens the hair and soothes the skin underneath. Just a few drops, worked in from the root to the tip.

Host 1I use it every morning. It keeps everything smooth and well groomed, even after a long, hard day.

Host 2Available in unscented and woodsy varieties, widely in stock and very reasonably priced.

Host 1And a little goes a very long way. My partner noticed the difference right away.

Host 2Beard oil. Our referral link is in the show notes.

Why it matters

Host 1So, what is the principle here?

Host 2That a tool for self-monitoring is not the same thing as a behaviour change. In this trial, giving older adults a platform and a wristband on top of regular dietitian visits added work and information, but not extra centimetres off the waist.

Host 1Why might the human-only group have done slightly better?

Host 2The authors offer three reasons: digital literacy among older Greek adults may have limited engagement, self-monitoring tends to fade without reinforcement, and three months may be too short. A further possibility, which is my speculation and not tested here, is that time spent on the platform partly replaced attention in the room rather than adding to it.

Host 1What does this change for whom?

Host 2For a health service deciding how to spend money on adults over fifty-five at metabolic risk, it suggests the first priority is regular contact with a dietitian. Digital add-ons should earn their place with evidence that people actually use them.

Host 1And for someone listening who just bought a fitness tracker?

Host 2It may nudge you to move a little more, which is what the activity numbers hint at. On its own short-term evidence, this trial gives no reason to think it will change your waistline beyond what good dietary advice already does.

Host 1What does it not mean?

Host 2It does not mean digital tools are useless, or that they make people worse. The digital arms improved substantially too, the differences were small, and the trial says nothing about longer periods, younger people, or settings where there is no dietitian at all.

Host 1And it does not prove that the counselling did the work.

Host 2Correct. Every arm got counselling, so the within-group improvements could also include people simply being in a study, getting measured, and natural drift back toward average among those recruited for having high readings.

Caveats

Host 1Now the caveats.

Host 2First, three months. Short-term weight loss is common in lifestyle programmes, and as established background, keeping it off over years is the harder problem.

Host 1Second?

Host 2Nobody measured whether people actually used the digital tools. The authors state plainly that there was no process evaluation: engagement, usability and adherence to the platform and devices were not recorded, so we cannot tell a tool that did not work from a tool that was not opened.

Host 1That seems like the central gap.

Host 2It is, and the authors acknowledge that it limits how the digital results can be interpreted.

Host 1Third?

Host 2The activity result rests on a questionnaire. People wearing a step counter for three months are more aware of their activity, and possibly more inclined to report it generously, so a self-reported gain in that arm is the weakest kind of evidence for real extra movement.

Host 1What about the measurements themselves?

Host 2Body composition came from impedance devices, and the authors note that different machines were used across centres, which adds noise and possible bias. They also say researcher bias in measuring waists and height cannot be ruled out, and the paper does not describe whether the people holding the tape knew who was in which group.

Host 1The baseline imbalances?

Host 2They matter a little. The standard care group was older and less educated, and randomisation was not stratified, so chance produced those gaps. The adjusted and sensitivity analyses reduce the worry, but the sensitivity results are only described, not shown.

Host 1Anything odd in the design?

Host 2One small thing: the power calculation was set to detect a difference of one standard deviation between groups, which would be a very large effect. With three hundred people per arm the trial could in fact detect much smaller differences, so it was not underpowered, but the stated rationale is unusual.

Host 1Who is in the sample?

Host 2About three quarters women, all over fifty-five, all living independently in Greece, recruited partly through dietetics practices and health services. People who sign up through a dietitian's office may already be more motivated than the average person at risk.

Host 1And the data?

Host 2Available on request from the authors, not publicly, because of data protection rules. The trial is registered on ClinicalTrials dot gov, and the funding was the European Union's Horizon twenty twenty programme, with no conflicts of interest declared.

What to watch

Host 1What should we watch for next?

Host 2Usage data. If the GATEKEEPER teams can report how often people logged in, synced devices and messaged dietitians, we could see whether the people who engaged did better.

Host 1Anything else?

Host 2Longer follow-up, to see whether the counselling gains last once the monthly visits stop, and whether the activity nudge from the wristband persists. And objective activity measurement in all arms, not only the one that was given a tracker.

Host 1And the bigger field?

Host 2The authors raise the drug question themselves: G L P one medicines now produce large weight losses, and they argue behaviour change stays essential for keeping weight off and preserving muscle. Trials that test where dietitians and digital tools fit alongside those drugs would be the natural next step.

Host 1Bottom line?

Host 2A large, registered, real-world trial with a modest and honest result: monthly dietitian sessions were followed by meaningful improvements, and a platform or tracker on top added nothing to waist size or blood sugar in three months. The tracker may have bought a little extra walking, but without data on who used what, the digital story stays unfinished.

Sign-off

Host 1That is the deep dive: Impact of Dietetic Counseling Combined With Digital Tools on Obesity-Related Indices in Greek Adults, by Makri, Androutsos, Manios and colleagues, in the journal Obesity. The link is in the show notes.

Host 2Every Friday this same feed carries the roundup episode: five new papers as trailers, and you pick which become deep dives. This has been the Hastrup Review of Science.