Every research figure on this page comes from a published study. Each one shows its source and its limits.
Leaders want results they can read and a program people actually use. Here is what the research says about the features EPX3 is built on.
return-on-investment estimates in a review of 56 workplace blood-testing programs came from programs with physician- or coordinator-led follow-up. None of the 14 screening-only programs reported a return. Reported returns clustered between about 1.5 to 1 and 4 to 1.
Swadźba et al., BMC Health Services Research, Sept. 29, 2026. Observational; the authors warn it may reflect reporting bias.
saved for every $1 spent, in Johnson & Johnson’s own in-house worksite health program. Medical spending grew 3.7 points a year slower than at similar large companies, 2002 to 2008.
Henke et al., Health Affairs, 2011. A J&J result, not an EPX3 result. Observational.
In 1,228 employees who used employer-sponsored coaching, well-being did not change after one visit. It rose about 2 points after 2 to 3 visits and about 5 points after 4 or more. Absence and presenteeism fell more with more visits.
Sagui-Henson et al., Telemedicine and e-Health, 2022. No comparison group; overall effects were small.
points of vitality over six months in a randomized trial of a 2.5-day energy-management course (240 employees). Purpose in life rose 5.22 vs 0.27. Weight, blood pressure and blood sugar did not differ.
Das et al., American Journal of Health Promotion, 2020. Run by independent Tufts researchers; funded by Johnson & Johnson.
In a review of 43 studies, AI that reduces workload (assistive) was generally linked to lower burnout. AI used to monitor or control workers was linked to more strain. Most of the strongest evidence is from healthcare.
Fang et al., Frontiers in Psychology, 2026.
These are published findings about other programs. They are not EPX3 results and do not predict yours. EPX3 has not yet run a cohort. Results differ across studies, and some of the best-run trials found no short-term return. We show both.
Your results are reviewed by a physician and explained in plain language. Then you get coaching on movement, sleep and food.
Program design.
In a randomized trial of an energy-management course, vitality and purpose in life rose more than in the control group. Sleep problems, fatigue and depressive symptoms also improved.
Das et al., 2020. Funded by Johnson & Johnson; run by independent researchers.
In 11,484 people, a five-minute walk every 30, 60 or 120 minutes left people less tired and in a better mood for two weeks. Work performance did not suffer. Hourly walks were judged the best balance.
British Journal of Sports Medicine, 2026 (the Body Electric Challenge). Self-reported surveys; two weeks; no control group.
In 40 young workers, those with long periods of uninterrupted muscle activity had about three times the rate of neck and shoulder pain.
Hanvold et al., 2013. Small study.
Your employer sees only anonymous group numbers, for groups of ten or more. Your own results stay with you and the clinic.
Program design.
| A generic app or standard wellness program | The EPX3 design | |
|---|---|---|
| What the best trials found | The largest randomized trial of a generic program (32,974 employees) found no significant effect on clinical markers, spending or absence. A review of 25 return-on-investment studies found only 2 rigorous ones, with no positive short-term return. | EPX3 has not yet run a cohort. Its design follows what the research favors, listed in the rows below. The pilot measures the result. |
| Who follows up on results | Often no one. In a 2026 review of 56 programs, none of the 14 screening-only programs reported a return. | A physician reviews each person’s results and leads a weekly check-in. |
| Coaching | Often one-time or self-serve. | Weekly check-ins, so everyone gets more than one visit. In one study of 1,228 employees, gains appeared after two or more visits (no comparison group). |
| Privacy | Varies by vendor. | Employers see only anonymous group numbers, for groups of ten or more. |
| AI | May be used to monitor. | AI is taught as a tool the employee directs. In a 2026 review, assistive AI was generally linked to lower burnout and monitoring AI to more strain. |
| What gets measured | Sign-ups and logins. | Before-and-after group measures: absence, turnover, engagement, plus health and energy measures. |
in the largest randomized trial of a generic workplace wellness program (32,974 employees, 160 worksites, 18 months): no significant effect on 10 clinical markers, 38 spending and use measures, or absence. A review of 25 return-on-investment studies found only 2 rigorous ones, with no positive short-term return.
Song and Baicker, JAMA, 2019. Baid et al., American Journal of Preventive Medicine, 2021.
No study here tests EPX3. We do not promise savings, a return on your money, or a health result. We promise to measure, and to show you what changed.
EPX3 has not yet run a cohort.
Some reviews suggest it takes years to see financial effects in claims data. Short pilots measure engagement, health and energy first.
Our reading of the research limits above.
Three employer numbers, before and after: absence, turnover and engagement. Plus group measures of health and energy. We report what changed, whatever it is. Groups of ten or more only.
Every study we cite, with who it studied, what it found and where it falls short. We attribute each figure to its own source and never present another program’s result as ours.
| Source | Who and what | What it found | Limits |
|---|---|---|---|
| Henke et al., Health Affairs, 2011 | Johnson & Johnson’s in-house worksite health program, 2002 to 2008, compared with similar large companies | $565 saved per employee per year; $1.88 to $3.92 saved per $1 spent; medical spending grew 3.7 points a year slower | One company’s own program. Observational. Not a test of EPX3. |
| Swadźba et al., BMC Health Services Research, Sept. 29, 2026 | Systematic review of 56 workplace programs with blood-based screening | All 20 return-on-investment estimates came from programs with physician- or coordinator-led follow-up; none of 14 screening-only programs reported a return; returns clustered between about 1.5 to 1 and 4 to 1 | Mostly US observational studies; six randomized trials, none at low risk of bias. Authors warn of reporting and selection bias. |
| Song and Baicker, JAMA, 2019 | Randomized trial, 32,974 employees, 160 worksites, 18 months, generic workplace wellness program | Exercise and weight management improved. No significant effect on 10 clinical markers, 38 spending and use measures, or absence and job outcomes | A different kind of program from EPX3. Shown here so the picture is balanced. |
| Baid et al., American Journal of Preventive Medicine, 2021 | Systematic review of 25 return-on-investment evaluations | Only 2 were rigorous and low in bias; no positive short-term return | Short-term focus. Shown for balance. |
| Das et al., American Journal of Health Promotion, 2020 | Randomized trial, 240 employees, 12 worksites, 2.5-day energy-management course from Johnson & Johnson’s Human Performance Institute, six-month follow-up | Vitality +12.65 vs +4.98; purpose in life +5.22 vs +0.27; better sleep, less fatigue and fewer depressive symptoms than controls | No difference in weight, blood pressure, blood sugar or activity. Funded by Johnson & Johnson; run by independent Tufts researchers. Dollar return not measured. |
| Sagui-Henson et al., Telemedicine and e-Health, 2022 | 1,228 employees using employer-sponsored video coaching (Modern Health), average 2.4 visits | Well-being unchanged after one visit; about +2 points after 2 to 3 visits; about +5 after 4 or more. Absence and presenteeism fell more with more visits | No comparison group. Overall effects small. One vendor’s platform. |
| Zhou et al., PLOS Medicine, 2026 | 91,292 UK Biobank adults wearing accelerometers | Each extra hour of prolonged sitting: hazard ratio 1.09 for cancer death. Each extra hour of interrupted sitting: 0.82 | Observational. A link, not proof of cause. |
| Healy et al., Diabetes Care, 2008 | Australian accelerometer study of adults | Breaks in sitting were linked to smaller waist, lower triglycerides and lower two-hour glucose, independent of total sitting time | Observational. |
| Veiersted et al., 1993; Hanvold et al., 2013 | EMG studies of 30 new workers and 40 young workers | A low rate of brief muscle pauses predicted chronic neck and shoulder pain; long uninterrupted muscle activity went with about three times the pain rate | Small studies. |
| Fang et al., Frontiers in Psychology, 2026 | Systematic review of 43 studies on AI at work and burnout | Assistive AI generally linked to lower burnout; monitoring and algorithmic control linked to more strain | Strongest evidence is from healthcare settings. |
Testimonials on this site are labeled as coming from Dr. Majette’s earlier program. Any projection is labeled as a projection. This page is general information, not legal or medical advice.
A free 30-minute call. No long-term commitment.