Can Digital Health Tools Replace Cardiometabolic Coaching?

A well-designed digital behavior program can match face-to-face behavioral care for cardiometabolic risk markers. It may therefore be a sensible option when appointments are expensive, scarce, or impractical. But the evidence does not show that adding an app, wearable, or digital coaching package to an existing coaching program improves results.

That distinction comes from two different kinds of evidence. A large 2025 meta-analysis found no statistically significant difference between digital and nondigital behavioral interventions across 11 risk markers. A much smaller randomized trial found no added benefit when researchers put a technology package on top of structured in-person support. Together, they support a restrained conclusion: digital delivery can substitute for behavioral care in the short to medium term; it has not proved itself a multiplier.

Digital and face-to-face programs had similar results across 11 markers

In a 2025 systematic review and meta-analysis by Berhe et al., researchers pooled 34 randomized controlled trials involving 17,389 people. Participants averaged 50.2 years old, and trial sizes ranged from 65 to 7,610 people. The investigators compared digital behavioral interventions with nondigital interventions for body weight, BMI, waist circumference, systolic and diastolic blood pressure, HbA1c, fasting glucose, LDL and HDL cholesterol, triglycerides, and total cholesterol.

Meta-regression found no statistically significant difference between delivery modes for any of those 11 outcomes. The result held in studies lasting 26 weeks or less and in longer studies, and it did not change across risk-of-bias strata. Both digital and nondigital programs improved several risk factors compared with no-intervention controls.

This is meaningful evidence that a digital program need not be a second-rate version of behavioral care. The analysis does not establish formal non-inferiority, however. A non-significant difference is not identical to a pre-specified non-inferiority result with a clinically meaningful margin. The review supports statistical similarity in its included trials, not proof that every app delivers the same result as every clinician.

The earlier 2021 Akinosun et al. meta-analysis reached a compatible conclusion in people with established cardiovascular disease: digital technology interventions could modify cardiovascular risk factors. That narrower population helps show that digital delivery can work in higher-risk patients, though it does not answer every question about long-term use outside trials.

Replacing coaching differs from stacking technology onto it

The strongest direct test of digital augmentation comes from the SMART Health randomized trial. Researchers assigned 104 schoolteachers with, or at risk of, type 2 diabetes to a wait-list control group, five face-to-face visits with a psychologist and exercise specialist, or the same five visits plus a six-month technology-based behavior-change package.

At nine months, neither active intervention improved the primary outcome—daily pedometer steps—versus the wait-list group. Researchers also found no significant group-by-time effects for blood pressure, waist circumference, BMI, fasting glucose, HbA1c, or plasma lipids. Adding the digital package did not improve any measured outcome over face-to-face support alone.

There was one suggestive difference in self-reported resistance training. At three months, both active arms exceeded control: face-to-face support increased training by 136 minutes per month (P<.01; d=0.33), while face-to-face support plus technology increased it by 145 minutes per month (P<.001; d=0.40). By nine months, only the face-to-face-only arm retained that advantage over control.

It would be tempting to call that proof that technology erodes the benefit of human support. It is not. The trial had about 35 people per arm and was underpowered for many cardiometabolic secondary outcomes. More importantly, even face-to-face support did not beat control on its primary aerobic endpoint. The study offers a directional warning against assuming a hybrid program automatically works better, but it cannot establish why the digital layer failed to add value.

Digital delivery has evidence as an alternative route to behavioral support. It does not have evidence as a performance-enhancing attachment for an already effective coaching relationship.

"Digital" is not one intervention

Program design matters. In the Dulce Digital-Me randomized comparative trial, 310 Latine adults with poorly managed type 2 diabetes received either static text-based education and support or a dynamic program with personalized feedback and goal setting. Participants averaged 52.1 years old; 69.7% were female; baseline HbA1c averaged 9.3%. The adaptive approach performed better than static digital education on diabetes outcomes.

The implication is practical. A program that sends generic reminders is not interchangeable with one that responds to goals, progress, and barriers. The large meta-analysis averages across many digital designs, so its overall delivery-mode finding should not be read as a guarantee for low-effort or poorly tailored products.

Evidence from a smaller, more specific setting points in the same direction. The iSMART-DM trial randomized 80 East Asian adults with non-insulin-treated type 2 diabetes in a 2:1 ratio to smartwatch, mobile app, and structured education (54 people) or structured education alone (26 people). Over 12 weeks, the intervention reported modest glycemic and self-care improvements. It broadens the geographic evidence base, but it does not test whether digital delivery matches face-to-face behavioral coaching.

The unanswered question is durability

Most of the evidence concerns supervised trial settings, where enrollment, follow-up, and study procedures can keep people engaged. Berhe et al. explicitly note adherence, user engagement, and long-term sustainability as problems for digital interventions. Their duration subgroup analysis—26 weeks or less versus more than 26 weeks—does not establish what happens after 12, 18, or 24 months. The review also does not report comparative dropout rates, usage frequency, or engagement trajectories that would show whether digital participation fades faster than in-person participation.

Generalizability also has limits. Twenty-three of the 34 trials in the Berhe review took place in the United States. Results may differ where broadband access, digital literacy, reimbursement, staffing, and health-care infrastructure differ. Publication bias was not quantified in the available report, and the subgroup effect sizes for diet-focused versus physical-activity-focused interventions were not available for precise comparison.

Cost remains another unproven part of the sales pitch. A 2025 randomized trial of a gamified activity program included cost-efficiency among its endpoints, but it compared digital care with usual care rather than structured face-to-face coaching, and no extractable economic result resolves the comparison. Equal outcomes do not automatically mean equal cost per unit of risk reduction.

Behavioral tools still have a clinical ceiling

Digital programs belong in the behavioral-care tier. They do not replace medication when a person has established disease or needs larger risk reduction. For example, a 2024 meta-analysis of intermittent fasting in 1,101 adults with prediabetes or type 2 diabetes found a mean weight reduction of 4.56 kg versus control diets or calorie restriction (95% CI, −6.23 to −2.83 kg; P=.001). A separate 2025 meta-analysis of GLP-1 receptor agonists and tirzepatide, covering 18 randomized trials and 97,800 participants with cardiovascular or renal disease, found a lower risk of its primary composite outcome with GLP-1 receptor agonists (RR 0.88; 95% CI reported in the study) and reported heart-failure benefit in people with obesity, not in non-obese patients.

Those interventions answer different clinical questions. An app can help deliver exercise, eating, monitoring, and self-management support. It cannot substitute for a medication decision in someone whose cardiovascular or renal risk calls for one.

What to do with this evidence

Choose the format that gives you sustained access to a credible behavioral program. If a quality digital program is affordable and you will use it, current trial evidence says it is a reasonable substitute for face-to-face behavioral care for weight, blood pressure, glycemic, and lipid risk markers. If you already have structured professional support, do not assume another dashboard will improve the outcome. It may simply give your phone one more opportunity to ask whether you have logged lunch.

The evidence still cannot tell us whether digital and face-to-face care remain equally effective over years of ordinary life, or whether either model delivers the best result for a particular patient. What it does show is narrower and more useful: delivery mode alone is not the deciding factor. Program quality and continued engagement are the parts that still have to do the work.