Healthspan, explained through peer-reviewed evidence. Every recommendation on evidage carries an explicit evidence level (Level 1–4) and links to primary sources (PubMed / DOI).
How this ranking is built
evidage scores every candidate habit on four weighted axes:
| Axis | Weight | What it measures |
|---|---|---|
| Effect size | 35% | Hazard ratios, relative risks, mortality reduction — the measurable size of the benefit |
| Evidence certainty | 30% | Number of RCTs / meta-analyses, consistency across studies, sample size |
| Ease of adoption | 20% | Can an ordinary person start today, and keep going? |
| Cost | 15% | Money, time, opportunity cost |
The ranking is not fixed. When major new evidence lands, positions move the following month.
📐 August 2026: we recalculated every score. Here is why.
This month we did not swap individual items in and out. We re-scored all ten entries against our own four-axis method, because the published scores had drifted out of line with it.
Three problems had accumulated:
- #5 (UPF restriction, 7.825) scored lower than #10 (light exposure, 7.925) — the scores and the ordering contradicted each other
- Duplicate position labels (“1-3”, “6-8”) coexisted, and there was no #9 at all
- Microplastics reduction carried a score of 8.625 that did not reflect the fact that its human evidence rests on a single observational study of 257 people — the “evidence certainty” axis is worth 30% of the total
evidage’s core promise is that every recommendation carries an explicit evidence level. A ranking table that does not follow its own stated method breaks that promise. So this month we re-scored all ten items and published the full axis-by-axis breakdown for the first time.
As a result some items moved by as much as four places. This reflects corrected scoring, not new research. The breakdown table near the end lets you check the arithmetic yourself.
🥇 #1: Smoking cessation 9.20
[Level 1] [Lifestyle] [Strongly recommended] [July: top-3 group →]
- Smokers vs never-smokers: ~10 years difference in life expectancy (Jha P, et al. N Engl J Med 2013;368:341-350)
- Quitting by age 35 eliminates almost the entire gap
- All-cause mortality HR 0.36 (quitting by 50, women)
In one line: the single largest effect size of any lifestyle intervention. Every other healthy habit is partly cancelled out while you keep smoking.
🥈 #2: 150–300 minutes of moderate aerobic exercise per week 9.10
[Level 1] [Exercise] [Strongly recommended] [July: top-3 group →]
- All-cause mortality HR 0.69 (meeting WHO guidelines)
- Cardiovascular disease HR 0.64; dementia HR 0.72
- Clear dose–response, with maximum benefit around 300 min/week
In one line: roughly 8,000 steps a day, with 20 of those minutes at a brisk pace, gets most people to the WHO threshold.
🥉 #3: Resistance training 2–3× per week 8.45
[Level 1] [Exercise] [Strongly recommended] [July: top-3 group →]
- 30–60 minutes per week of muscle-strengthening activity is associated with a 10–17% lower risk of all-cause mortality, cardiovascular disease, total cancer and diabetes
(Momma H, et al. Br J Sports Med 2022;56(13):755-763. DOI: 10.1136/bjsports-2021-105061 — systematic review and meta-analysis of 16 studies) - Grip strength +7 kg associated with −12% all-cause mortality; highest group −33% (LaMonte MJ, et al. JAMA Netw Open 2026, OPACH cohort, 5,472 women)
📝 Correction (16 Aug 2026): earlier versions cited “Momma H, et al. (2024) Br J Sports Med 58(4):224-229″. That reference does not exist. The correct citation is the 2022 paper above. The effect size has also been corrected from “HR 0.79” to “10–17% lower risk”.
In one line: aerobic work alone will not prevent sarcopenia — and the benefit plateaus at 30–60 min/week, so you do not need long sessions.
4️⃣ #4: Keep blood pressure below 120/80 mmHg 8.25 ⬆️ +4
[Level 1] [Medical & Testing] [Strongly recommended] [July: #8 ⬆️]
- SPRINT trial: targeting systolic <120 mmHg cut cardiovascular events 25% (N Engl J Med 2015;373:2103-2116)
- All-cause mortality reduced 27%
- Cognitive decline reduced 15% (SPRINT-MIND)
🔺 Why it climbed four places
Not because of new research. Blood pressure control is the one medical intervention that scores near-maximum on both effect size and evidence certainty — and those two axes alone account for 65% of the total weight. Placing it at #8 in July simply failed to apply our own method. Even after penalising it on cost and ease (medication, ongoing clinic visits), it lands at #4.
First step: measure at home, in the morning, after waking. If readings stay above 130, talk to a doctor. Salt reduction, the DASH pattern and 150 min/week of exercise are first-line.
5️⃣ #5: 7–9 hours of sleep, at consistent times — regularity matters more than duration 8.10 ⬆️ +1
[Level 2] [Sleep] [Strongly recommended] [July: #6 ⬆️]
- Sleep regularity predicts mortality more strongly than sleep duration
(Windred DP, et al. Sleep 2024;47(1):zsad253. PMID: 37738616)- UK Biobank, 60,977 participants, >10 million hours of accelerometer data, mean follow-up 7.1 years, 3,010 deaths
- Sleep Regularity Index at the 5th percentile (SRI = 41): all-cause mortality HR 1.53 (95% CI 1.41–1.66, vs the median)
- At the 95th percentile (SRI = 75): HR 0.90 (95% CI 0.81–1.00)
- The SRI carries mortality information beyond the variability of sleep duration — but not the reverse
- Under 6 hours: HR 1.12; over 9 hours: HR 1.23 (inverted U-curve)
📝 Correction (16 Aug 2026): earlier versions cited “Zhao et al., 2024” and an HR of 1.30. No such study exists. The correct source is Windred DP, et al. Sleep 2024 (PMID: 37738616), and the correct figure is HR 1.53 at the 5th percentile.
First step: fix your wake time first, within ±30 minutes, weekends included. Bedtime tends to follow on its own.
6️⃣ #6: Mediterranean diet 7.98 ⬇️ −2
[Level 1] [Diet] [Strongly recommended] [July: #4 ⬇️]
- Cardiovascular events reduced 30% (PREDIMED, N Engl J Med 2018;378:e34)
- All-cause mortality RR 0.92 (Dinu M, et al., meta-analysis)
- Dementia hazard −28% (MIND-diet based); type 2 diabetes −31% (PREDIMED-Plus)
Why it slipped: the evidence has not weakened at all (still Level 1). The drop reflects an honest cost penalty — olive oil, nuts and oily fish are genuinely expensive in some markets, Japan included.
7️⃣ #7: Restricting ultra-processed foods (UPF) 7.80 ⬇️ −2
[Level 1] [Diet] [Strongly recommended] [July: #5 ⬇️]
- Umbrella review across 32 health outcomes found consistent associations between UPF exposure and adverse outcomes (Lane MM, et al. BMJ 2024;384:e077310)
- Cardiovascular-related mortality +50%, type 2 diabetes +12%, depression +22%
First step: read the ingredients list. If it contains substances you would never find in a home kitchen — emulsifiers, thickeners, artificial flavourings — it is NOVA Group 4. Decide at the shelf, not at the table.
8️⃣ #8: Maintaining social connection 7.75 ⬇️ −1
[Level 1] [Mental Wellbeing] [Strongly recommended] [July: #7 ⬇️]
- Social isolation: all-cause mortality 1.26× (Holt-Lunstad J, et al. PLoS Med 2010, meta-analysis of 148 studies)
- Dementia risk 1.57× with loneliness; cardiovascular disease 1.29×
- The WHO published its Commission on Social Connection report in 2025, framing loneliness as a public-health issue
First step: deliberately maintain the relationships you already have. A standing weekend activity with the same group covers exercise and connection at once.
9️⃣ #9: Light hygiene (circadian management) 7.68 ⬆️ +1
[Level 2] [Sleep] [Recommended] [July: #10 ⬆️]
- Bright morning light (>10,000 lux, 30 min) stabilises melatonin rhythm
- Evening blue light suppresses melatonin secretion by up to 50%
- Morning light exposure: 24% improvement on the Hamilton depression scale in RCT settings
In one line: 10–30 minutes of morning sunlight, and block blue light at night. Zero cost, highest ease-of-adoption score in the list.
🔟 #10: Reducing microplastics exposure 7.08 ⬇️
[Level 3] [Lifestyle] [Conditionally recommended] [July: labelled “6-8” ⬇️]
- Micro- and nanoplastics were detected in carotid plaque from 58% of 257 patients
- Those patients had a composite endpoint of myocardial infarction, stroke or death at HR 4.53 (95% CI 2.00–10.27) over a mean 33.7 months
(Marfella R, et al. N Engl J Med 2024;390:900-910. DOI: 10.1056/NEJMoa2309822) - Animal and cell work shows activation of pro-atherogenic genes in endothelial cells, oxidative stress and mitochondrial dysfunction
🔻 Why the score moved from 8.625 to 7.08
July’s 8.625 essentially passed the raw effect size straight through to the total. But evidence certainty carries 30% of the weight, and the human evidence here is one observational study — single-centre, n=257, 34 months. We left the effect size untouched at 8.5 and corrected only the certainty axis to 5.0, which yields 7.08 and tenth place.
The signal is large enough to keep it in the list, but we have moved it to Level 3 (conditional): worth doing, not worth prioritising over the nine items above it.
First step: stop microwaving food in plastic containers; move from bottled water to tap water in glass or steel; replace single-use plastic boards and cutlery. All of these are close to free and carry no downside — low certainty is no reason not to do them.
🚨 Warning section: Alcohol 2.50 (outside the ranking — unchanged)
[Level 1] [Lifestyle] [Avoid] [July: warning section →]
We continue the reclassification made in the July edition, moving alcohol from “recommended habit” to “habit to avoid”.
- The apparent “J-curve” is largely explained by confounding — former smokers and people who quit drinking due to illness were pooled into the “non-drinker” reference group
- The WHO states that no level of alcohol consumption is safe for health
- The 2025–2030 US Dietary Guidelines and ACC positions point the same way
Alcohol is a Group 1 carcinogen. “Less is better” is where the evidence currently stands. There is no health reason to start drinking.
Movement this month
| Aug | Item | Score | July position | Change |
|---|---|---|---|---|
| 1 | Smoking cessation | 9.20 | top-3 group | → |
| 2 | Aerobic exercise 150–300 min | 9.10 | top-3 group | → |
| 3 | Resistance training 2–3× | 8.45 | top-3 group | → |
| 4 | Blood pressure <120/80 | 8.25 | #8 | ⬆️ +4 |
| 5 | Sleep 7–9 h + regularity | 8.10 | #6 | ⬆️ +1 |
| 6 | Mediterranean diet | 7.98 | #4 | ⬇️ −2 |
| 7 | UPF restriction | 7.80 | #5 | ⬇️ −2 |
| 8 | Social connection | 7.75 | #7 | ⬇️ −1 |
| 9 | Light hygiene | 7.68 | #10 | ⬆️ +1 |
| 10 | Microplastics reduction | 7.08 | “6-8” label | ⬇️ |
| — | Alcohol (warning) | 2.50 | warning section | → |
📐 Full four-axis breakdown
Each axis is scored 0–10. Total = effect×0.35 + certainty×0.30 + ease×0.20 + cost×0.15
| # | Item | Effect (35%) |
Certainty (30%) |
Ease (20%) |
Cost (15%) |
Total |
|---|---|---|---|---|---|---|
| 1 | Smoking cessation | 10.0 | 10.0 | 6.0 | 10.0 | 9.20 |
| 2 | Aerobic exercise | 9.0 | 10.0 | 8.0 | 9.0 | 9.10 |
| 3 | Resistance training | 8.0 | 9.0 | 8.0 | 9.0 | 8.45 |
| 4 | Blood pressure <120/80 | 9.0 | 10.0 | 6.0 | 6.0 | 8.25 |
| 5 | Sleep + regularity | 8.0 | 8.0 | 7.0 | 10.0 | 8.10 |
| 6 | Mediterranean diet | 8.5 | 9.0 | 7.0 | 6.0 | 7.98 |
| 7 | UPF restriction | 8.0 | 8.0 | 7.0 | 8.0 | 7.80 |
| 8 | Social connection | 8.0 | 8.0 | 6.0 | 9.0 | 7.75 |
| 9 | Light hygiene | 6.5 | 7.0 | 9.0 | 10.0 | 7.68 |
| 10 | Microplastics reduction | 8.5 | 5.0 | 7.0 | 8.0 | 7.08 |
Why alcohol is not in this table: these four axes measure how much value there is in adopting something. Effect size, ease of adoption and cost all presuppose that you do the thing. They are meaningless when applied to a habit we recommend avoiding — scoring alcohol highly on “ease of adoption” would perversely raise its total. The 2.50 figure is the recommendation score carried over from July’s reclassification and is not directly comparable with the ten items above.
How to read this table: microplastics reduction at #10 scores higher on effect size (8.5) than several items above it — it is the certainty score of 5.0 that pulls it down. Light hygiene at #9 has a modest effect size of 6.5 but earns its place on ease (9.0) and cost (10.0). Seeing which axis earned the points tells you whether a recommendation fits your situation.
Important, but not in the top 10
- Vitamin D supplementation — no significant all-cause mortality effect in VITAL (n=25,871, 5.3 years), though cancer mortality fell 17% in sub-analysis. The 2025 VITAL telomere analysis (Am J Clin Nutr, PMID: 40409468) reported that 2,000 IU/day over four years preserved roughly 140 base pairs of leukocyte telomere length. A candidate for entry in coming months.
- Shingles vaccination (RZV) — Rayens E, et al. Nat Commun 2026;17:2056 reported a 51% lower dementia risk in twice-vaccinated adults aged 65+, significant after accounting for healthy vaccinee bias and larger in women. Observational, so Level 2; confirmatory RCTs are ongoing.
- Omega-3 (EPA) — clearly effective in high cardiovascular risk (REDUCE-IT), but 2026 reports on cognitive outcomes have split the field.
- Time-restricted eating — RCTs on meal timing are accumulating; evidence still developing.
- Screening — colonoscopy from age 45, every 10 years, cuts colorectal cancer mortality ~50%.
🔬 Editors’ picks this month
1. Shingles vaccine and dementia — the largest analysis yet
Rayens E, Sy LS, Qian L, et al. (2026). Recombinant zoster vaccine is associated with a reduced risk of dementia. Nature Communications, 17, 2056. DOI: 10.1038/s41467-026-69289-0
2. Sleep regularity predicts mortality
Windred DP, Burns AC, Rutter MK, et al. (2024). Sleep regularity is a stronger predictor of mortality risk than sleep duration: A prospective cohort study. Sleep, 47(1), zsad253. PMID: 37738616
3. Ultra-processed food — umbrella review of 32 outcomes
Lane MM, et al. (2024). Ultra-processed food exposure and adverse health outcomes: umbrella review of epidemiological meta-analyses. BMJ, 384, e077310.
⚠️ Disclaimer
- This page is written from peer-reviewed literature, but is not a substitute for medical care.
- It does not guarantee the prevention or treatment of any condition. Consult a physician about your own symptoms and medication.
- Evidence levels reflect the literature at the time of writing and may change as new large studies appear.
- Where earlier editions carried incorrect bibliographic information, this edition marks the fix explicitly as 📝 Correction.
evidage editorial team / Hydrowing Lab Inc. / 16 August 2026
