Trials that put a coach on top of a weight-loss app, against the same kind of app alone, couldn't show the coach adds weight loss. They couldn't rule it out either.
To count as health coaching in a 2023 review of weight-loss apps paired with coaches, a programme had to be at least partly built around the person, and an app discussing and deciding things with the patient met that test.
People who added a coach to the app lost 1.80 kg (about 4 lb) more on average than people with the app alone, within a likely range from 4.54 kg more lost to 0.93 kg less. That is the answer from the five trials a 2023 review counted as testing exactly this. The range is too wide to tell a coach worth several kilos from one worth nothing.
The review's authors put their answer in one line: they "found no statistical evidence of additional benefits" from the coaching. Not finding a benefit is not the same as finding there isn't one, and a range like that leaves room for both.
The review is by Han Shi Jocelyn Chew and four colleagues, published in 2023 in the Journal of Medical Internet Research. It gathered 14 randomised trials, where people are put into groups by chance. All of them tested weight-loss programmes that paired a tracking app with coaching, in 2,478 adults.
Everyone in them was an adult with overweight or obesity. Every result comes from each trial's first check after its programme ended.
Most of the coaches were dietitians or psychologists. In some trials the comparison group had a tracking app of its own, and in the others it had no app at all.
The five trials with an app on both sides are the ones that match the choice you're weighing. They are not where most of the review's headline number comes from. That number is bigger.
Most of the people in the trials that pitted a coach against the app alone were in two that, by their own reports, never tested a person against no person. Your exact question rests on far fewer people than it first seems.
- The review's headline number leans mostly on trials whose comparison groups had no app at all, so it can't show how much of that came from the coach.
- Programmes that gave more coaching sessions didn't show more weight lost, a pattern across trials that can't say what extra sessions would do for any one person.
- The review compared results only at the first weigh-in after each programme ended, so it can't say whether any difference lasted.
- All 14 trials landed at unclear or high risk on the reviewers' check for anything in their set-up that could skew the results, with people dropping out the commonest worry.
Where the 2.15 kg Comes From
The review's own summary leads with 2.15 kg: the extra weight lost with an app and a coach, measured against every comparison group put together.
Put together, they were two different kinds of group. Of the 11 trials behind the 2.15 kg, five gave the other group the same kind of tracking app. Six gave it no app at all: a waitlist, ordinary care, group classes with printed handouts, or texts and a website.
Split the two kinds of group apart and the review gives two answers. Against no app at all, app plus coach came out 2.27 kg ahead in six trials; against the same kind of app alone, 1.80 kg on average in five, too uncertain to count.
The 2.15 kg leans on the first answer: in the review's averaging some trials count for more than others, and the six no-app trials carried 70.4% of the say.
The summary paragraph behind the 2.15 kg ends with its own small print. The combined programmes were "only superior to usual care for weight loss", usual care being the review's name for groups with no app and no coaching.
A comparison with no app can't separate what the app did from what the coach did. So the review's headline number mostly measured an app and a coach against no app.
Which leaves the smaller number, the one with an app on both sides, and a fair question about it: does too uncertain to count mean the coach does nothing?
Too Uncertain to Count Is Not Zero
Not necessarily. The average sits on the coach's side, and its range reaches as far as 4.54 kg (about 10 lb) more lost with a coach.
The authors themselves flag the risk of reading a result like theirs as a zero: split like this, their comparison may have been too small to show a difference that is really there.
They give two readings of their own result. Either a coach adds little to a tracking app, or the coaching needs to be more thorough than what was tested, with more intensive feedback sessions, social support and problem-solving.
In these trials, then, the coach's extra was not shown, and not ruled out. That makes the next question matter more: what exactly did this review count as a coach?
Reviewing 14 trials that added coaching to weight-loss apps, the researchers concluded that app-plus-coach programmes could improve weight-related results, but that what a coach adds to an app still needs more research.
What This Review Counted as Health Coaching
The review set a definition, and on paper it asks a lot. Coaching had to be built around the person, at least in part, with goals they set themselves, learning about themselves and their health, being taught, and being held to account. If that last part is why a coach appeals to you, it was part of the brief.
Most programmes met it with people, from dietitians and psychologists to health coaches, most of them one-to-one and three in groups. Two are a different story. In the review's own table, two programmes are marked automated. One is described in its own trial's title as "a fully automated behavioral intervention by email, web, and mobile phone".
That programme came from a trial in people with prediabetes, meaning blood sugar above normal but not yet diabetes. It had more say than any other trial in the average behind the 2.15 kg: 17.4% on its own, and 21.0% together with the second automated programme.
Both automated programmes were compared with no app at all. So neither is inside the five-trial answer about the app alone. What they change is the headline number: part of what it credits to coaching came from programmes the review's own table marks automated.
The definition left room for that. The part about being built around the person could be met by "active discussion and shared decision-making between the app and patients".
That leaves the five trials that answer the question you came with. By the review's table, every one of them had a person doing the coaching.
The Head-to-Head Trials, by Their Own Reports
For two of those five, the trial's own report and the review's table disagree on the thing you'd want settled: whether one group had a person and the other didn't.
The largest, with 502 people, appears in the table as coaching by a dietitian over 20 sessions. In its own report, everyone had one 90-minute session with a dietitian at the start. After that, one group also got messages that a computer program picked from their logged food, activity and weight, up to three a day.[1]
The authors describe that support as coming "without ongoing human counseling".[1] The other group logged the same way, without the messages.
The second was a 12-week pilot of 64 people, set up to see whether people could use and accept a new app, not to test its effect on weight. Both groups got the same counsellor-led online programme, each group led by a registered dietitian or a clinical psychologist, but not by the same person.[2]
Otherwise only the tracking app differed, with the programme's tracking lessons changed to match: an app the researchers built, called Slip Buddy, against MyFitnessPal.[2]
The review's table lists that one as group coaching by a dietitian and a psychologist, against an app alone. By the trial's own report, there was a counsellor on both sides.
Together, those two trials hold 566 of the 700 people in the review's coach-versus-app comparison, and they are the two where the app-alone group lost slightly more.
That leaves 134 people in three small trials that, by the review's own table, set a person against no person. None of this changes the review's own figure for the five trials: 1.80 kg on average, too uncertain to count, left exactly as printed. What it changes is how many of those 700 people were in trials asking your question.
And the researchers who ran that largest trial came away with a view on people that isn't what you might expect.
Researchers who tested computer-picked daily messages in a 502-person weight-loss trial, with no ongoing human counselling, found engagement fell without personal interaction and suggested some human contact may be needed.
The Case for a Person
Those researchers didn't conclude that people can be replaced. In their own words, "without personal interaction, engagement declined to the point that it interfered with the intervention delivery" (getting the programme to people).[1]
They also wrote that their expanded messages hadn't made up for the in-person group sessions of their earlier trial, "suggesting that some form of interpersonal interaction may be needed".[1]
A larger, newer analysis leans the same way. Across 68 trials of weight-loss programmes with digital parts, and more than 14,500 people, contact with a specialist was linked with 1.08 kg more weight lost at six months.[3] The specialist was a trained expert such as a dietitian reached mostly by email, chat or video.[3] Its authors list that contact among three "best bets" for weight loss.[3]
They still call no single part of a digital programme essential.[3] Even their specialist contact included automated emails sent in a dietitian's name.[3] Their analysis also shares trials with this review, so the two can't simply be added together.
So people may well matter, and the head-to-head evidence that a coach adds weight loss on top of an app is thin. Whether any of it fits you depends on who these trials enrolled, and when they weighed them.
Who These Trials Were About
The apps will look familiar: MyFitnessPal, Lose It!, a Fitbit app and the Noom Coach app all appear in the review's table. The review didn't rate the apps. It looked at what adding a coach to one did.
None of the coaches is listed as a personal trainer. By the review's count, coaching ran from 2 sessions to 24, and across the trials, more sessions did not line up with more weight lost. That compares programmes, not people, and several programmes didn't report their numbers.
The people were adults with overweight or obesity, 39 years old on average, which the authors themselves call young, and mostly in the US. Trials in people with mental health conditions such as major depression, or chronic illnesses such as heart failure, were left out.
Each trial's first check after its programme came anywhere from 2 to 24 months in, and the review reports nothing later.
The reviewers also rated each trial for how likely its set-up was to skew its results. None came out low. Eight were rated unclear and six high. The most common problem was people dropping out, with too little said about how their missing data were handled.
One result is reported as separating the coach from the app: waist size, where the review found a difference against the app alone that was statistically significant, meaning unlikely to be chance. It comes with two problems.
The paper prints its two waist results, one against an app alone and one against no app, with their labels swapped between the text and the figure. And the authors' explanation, more muscle and a more toned body, went untested: muscle wasn't one of the review's outcomes.
So what does all of that leave for someone deciding about the coach?
The Weight-Loss App With Coaching Study, Read to the End
Chew and colleagues (2023) reviewed 14 randomised trials of weight-loss programmes that paired a tracking app with a coach, in adults with overweight or obesity. Against an app alone, in five trials, the coached groups lost 1.80 kg more on average, a difference the authors did not find statistically significant; against no app at all, they lost 2.27 kg more.
Read to the end, the small print adds three things. The headline 2.15 kg mostly compared an app and a coach with no app. Some of what counted as coaching was marked automated. And the head-to-head test of a person rests on three small trials that, by the review's own table, set a person against no person.
What a coach is worth to you is a question these trials hand back. They show what adding one did to weight, on average, in trials like these, and how uncertain that still is.
Every programme in the review shared one part: a tracking app and the logging that came with it. How much of the weight loss that part did on its own is a separate question. So is whether a coach met face to face would change the answer.
The authors offer one more reason a coach might add so little, and it's about the app rather than the coach. Most tracking apps, they point out, already come with several habit-building tools: the log itself, setting goals, making plans, and reminders that nudge you.
If that's right, the person on offer isn't being added to an empty phone. Part of a coach's job may already be done by your app, and whatever a coach adds has to come from what's left over.
That's their reasoning, not something the trials tested: the review never checked which tools went with more weight lost. Across other kinds of paid help, extras stacked on help people already had mostly came out close too.
Before you change anything
Trial by trial, average ages ran from 20 to 55. The volunteers in one trial could be a generation apart from those in another. Average BMI (weight for height) ran from 26.9 to 35.9.
The review never says how many were men and how many were women. Children were ruled out from the start.
Ten of the trials came from the US, with one each from Australia, Belgium, Korea and Japan.
And the review doesn't report whether the coaching happened face to face or at a distance, only whether it was one-to-one or in a group.
Coaching without an app was never part of the comparison. The authors set that pairing aside as outside their focus, so the review can't say what a coach does on their own.
The coaching itself differed widely between trials, in how long it ran and the setting it came in. The authors name that as one possible reason they found no clear extra from the coach.
Money was never measured in the review. It looked at weight, body measures such as waist size and blood sugar, eating and activity. So it can speak to what a coach did for weight, not to whether one is worth it.
The trials didn't tell one story. Their results differed from each other by more than chance would explain. The review turns that spread into a range for where a new trial like these might land.
That range runs from 4.72 kg more lost to 0.43 kg less. Even the combined result, then, doesn't promise that the next programme like these would come out ahead.
The reviewers also looked for signs that trials with unimpressive results had gone unpublished, and found none. With about a dozen trials to check, a test like that can only see so much.
Here's a suspicion worth following. Maybe the logging itself, the part you'd keep with or without a coach, is behind much of the weight lost, in a way these trials couldn't pull apart. It's a fair thing to wonder, and it has its own evidence. FitChef's page on a review of 12 calorie-tracking trials looks at what keeping a food log does to weight when the log is the thing being tested.
What This Study Found
All findings from this paper, in plain language.
- Programmes combining a tracking app with a coach led to more weight lost than all the comparison groups put together.
- Against people using the same kind of app without a coach, the coached groups' extra weight loss was too uncertain to tell apart from chance.
- Against people who had neither an app nor a coach, app-plus-coach programmes led to clearly more weight lost, though that can't separate the app from the coach.
- The results against an app alone and against no app were not clearly different from each other.
- App-plus-coach programmes shrank waists more than all the comparison groups put together.
- Against an app alone, coaching went with a bigger drop in waist size, though the authors warn this comes from a smaller slice of the data.
- Body mass index, a measure of weight for height, did not clearly change compared with the comparison groups.
- People in app-plus-coach programmes took in fewer calories a day than people in the comparison groups.
- A blood test of average blood sugar over recent months fell slightly more with app-plus-coach programmes.
- A drop in one kind of blood fat was reported, but that result rests on a single trial.
- Body fat, physical activity, blood pressure and cholesterol showed no clear difference between the groups.
- Programmes with more coaching sessions did not show more weight or waist loss than programmes with fewer.
- The authors concluded they found no statistical evidence that a coach adds to what an app alone does, and that more research is needed.
- The reviewers rated none of the 14 trials at low risk of a set-up that could skew the results.
- A check for unpublished trials with disappointing results found no sign of any for weight or waist size.