Figure 1

The atlas

Click a mark to open the variable: its edges, the sign of each arrow, and the papers. Drag a mark to move it. Click a curve to open the edge. Hover for a short card. The "Strata" layout groups nodes by layer. The URL is the view.

68 edges in this cut — click a mark, a curve, or a row

1. MECHANISM0 NULLAerobic → HOMA-IRIn 48 Glasgow adults, a 12-week pedometer program that added about 3,000 steps a day on five days a week did not change fasting insulin or glucose against a no-change walking control.
2. MENDELIAN+ INCREASESAlcohol → CV eventsHolmes 2014 found that a gene variant that cuts weekly drinking by 17 percent also cut the odds of coronary heart disease by 10 percent in 261,991 European adults (odds ratio 0.90, 0.84 to 0.96). Millwood 2019 found genetically predicted drinking raised stroke in Chinese men and left heart attack unmoved.
3. MENDELIAN+ INCREASESAlcohol → MortalityKassaw 2024 found a linear Mendelian randomization rise in all-cause death with genetically predicted alcohol in 278,093 white-British UK Biobank adults: odds ratio 1.27 (1.16 to 1.39) per extra 8 grams a day.
4. MECHANISM~ NON-LINEARAlcohol → THeavy chronic drinking suppresses male testosterone. A 0.5 g/kg drink raised plasma T in thirteen men. Three weeks of 40 grams a day lowered it 6.8 percent in ten men on a controlled diet. A glass of wine is not a T protocol in reverse.
5. MECHANISM− DECREASESAshwagandha → CortisolChoudhary 2017 found that 300 mg of ashwagandha root extract twice a day for eight weeks lowered morning serum cortisol versus placebo in 52 overweight adults under chronic stress (13.41 versus 15.44 μg per deciliter at week 8, P=0.0132). Cortisol was a secondary endpoint in a weight-and-craving trial.
6. MECHANISM+ INCREASESAshwagandha → TSmith 2023 found that 200 mg of Witholytin ashwagandha root extract twice a day for twelve weeks raised free testosterone and luteinizing hormone versus placebo in the men of a trial of overweight adults aged 40 to 75 (P=0.048 and 0.002). The primary stress score did not beat placebo.
7. MECHANISM− DECREASESBerberine → HOMA-IRPerez-Rubio 2014 found that 500 mg of berberine three times a day for three months raised the Matsuda insulin-sensitivity index from 2.1 to 3.1 versus placebo in 24 people with metabolic syndrome. Zhang 2008 lowered glucose versus placebo in 116 people with type 2 diabetes; the clamp contrast missed.
8. SMOKE~ NON-LINEARCaffeine → MortalityLi 2020 pooled 21 cohort articles (10,103,115 people, 240,303 deaths) and found a nonlinear inverse between coffee and all-cause death: three cups a day against none or rare sat at relative risk 0.87 (0.84 to 0.89). Tsujimoto 2017 saw the same U in caffeine milligrams. Nobody assigned a dose.
9. MECHANISM− DECREASESChromium → HOMA-IRJamilian 2016 found that 200 micrograms of chromium picolinate a day for eight weeks moved HOMA-IR by -0.8 versus +0.9 on placebo in 64 women with polycystic ovary syndrome (P<0.001). Kim 2012 saw a similar direction in 25 overweight children. Jain 2013's picolinate arm did not.
10. MECHANISM? CONDITIONALCortisol → LeadershipHigh cortisol is the condition under which 'high T = leader' fails. The internet quote drops this term.
11. MECHANISM? CONDITIONALCreatine → StrengthAssigned creatine raised predicted one-rep strength more than lifting alone in 45 older adults (Amiri 2023) and raised fat-free mass in a five-day load (Mihic 2000), but a two-year postmenopausal trial found no extra one-rep max (Chilibeck 2023).
12. MECHANISM0 NULLDHEA-S → CognitionA year of 50 mg DHEA did not move six cognitive tests against placebo in 225 healthy adults aged 55 to 85. A Cochrane review of earlier RCTs and a 13-week crossover in 46 older men also found no cognitive benefit.
13. SMOKE0 NULLDHEA-S → StatusDHEA-S on an optimization panel is not a status intervention.
14. SMOKE? CONDITIONALE2 → DominanceCycle-phase 'preference for dominance' is a contested meta-finding, not a hormone dial.
15. SMOKE− DECREASESAerobic → MortalityMartinez-Gomez 2024 pooled 2,011,186 adults in four megacohorts and found that meeting recommended leisure-time activity tracked a 22 percent lower all-cause death rate (hazard ratio 0.78, 0.77 to 0.79). Nobody randomized walking against death.
16. SMOKE+ INCREASESPrenatal famine → HOMA-IRPeople exposed to the Dutch Hunger Winter in utero had higher 120-minute glucose at age 58, and a smaller IVGTT subsample showed worse glucose tolerance after mid-gestation famine. Nobody randomized a siege.
17. MECHANISM? CONDITIONALFenugreek → TRao 2017 found that 600 mg of Testofen fenugreek seed extract a day for twelve weeks raised total and free testosterone versus placebo in 120 healthy men aged 43 to 70. Lee-Odegard 2024 saw a saliva testosterone rise versus placebo in 95 men and missed on plasma. Rao 2020 left testosterone unchanged in men with prostate symptoms.
18. MECHANISM+ INCREASESFerritin → CognitionTreating iron deficiency in young women can move attention and memory on named tasks. Murray-Kolb and Beard (2007) showed that in a blinded 16-week trial. Two later trials in the same age band again find the gain where ferritin rose. Berglund's infant trial found no WISC-IV difference at age 7.
19. MECHANISM− DECREASESGLP-1 → CV eventsIn 17,604 adults with overweight or obesity, established heart disease, and no diabetes, weekly semaglutide 2.4 mg cut a first heart attack, stroke, or cardiovascular death by 20 percent versus placebo (Lincoff 2023). Liraglutide did the same in type 2 diabetes (Marso 2016). Oral semaglutide was not worse for that same count (Husain 2019).
20. MECHANISM+ INCREASESGLP-1 → FertilityIn 28 infertile obese women with PCOS, twelve weeks of low-dose liraglutide added to metformin, then stopped for four weeks, was followed by more IVF pregnancies per embryo transfer than metformin alone: 85.7 percent versus 28.6 percent (Salamun 2018). The file is small and stacked.
21. MECHANISM− DECREASESGLP-1 → HOMA-IRIn 1,879 people with type 2 diabetes, weekly tirzepatide cut HOMA2 insulin resistance by about 16 to 24 percent, more than weekly semaglutide 1 mg (Frias 2024). A clamp study found the same direction versus placebo (Heise 2022).
22. MECHANISM− DECREASESGLP-1 → KidneyIn 3,533 adults with type 2 diabetes and chronic kidney disease, weekly semaglutide 1.0 mg cut a first major kidney event (kidney failure, a 50 percent drop in eGFR, or death from kidney or heart causes) by 24 percent versus placebo (hazard ratio 0.76; Perkovic 2024). Daily liraglutide cut a similar renal composite (Mann 2017).
23. MECHANISM− DECREASESGLP-1 → MortalityIn SELECT, weekly semaglutide 2.4 mg cut death from any cause versus placebo in 17,604 adults with overweight or obesity and established heart disease (hazard ratio 0.81; Lincoff 2023). Daily liraglutide did the same in type 2 diabetes (Marso 2016).
24. MECHANISM− DECREASESGLP-1 → OSAIn two 52-week trials of adults with obesity and moderate-to-severe sleep apnea, weekly tirzepatide cut the apnea-hypopnea index by about 20 to 24 events per hour versus placebo (Malhotra, Grunstein, Fietze 2024).
25. MECHANISM− DECREASESGLP-1 → StrengthIn 160 adults with obesity who sat for a DXA scan in SURMOUNT-1, weekly tirzepatide cut lean mass by 10.9 percent versus 2.6 percent on placebo (Look 2025). About a quarter of the lost weight was lean on both arms. Fat mass fell more.
26. MECHANISM0 NULLGreen tea → HOMA-IRBrown 2009 found that 400 mg of EGCG twice a day for eight weeks did not change insulin sensitivity, insulin secretion, or glucose tolerance versus lactose in 88 overweight or obese men. Hsu 2011 and Fukino 2006 also found no between-group HOMA difference after a green-tea extract or powder.
27. MECHANISM? CONDITIONALHearing → CognitionIn 977 older adults with untreated hearing loss, three years of hearing aids did not slow a global cognition score versus health education; a prespecified ARIC slice of 238 people at higher risk did.
28. MECHANISM− DECREASESInositol → HOMA-IRGiordano 2011 found that 2 g of myo-inositol twice a day plus diet for six months lowered the HOMA insulin-resistance index versus diet plus placebo in 80 postmenopausal women with metabolic syndrome. D'Anna 2015 saw a smaller HOMA drop and less gestational diabetes in 220 obese pregnancies; that file was open-label.
29. SMOKE? CONDITIONALJob strain → CortisolHair cortisol ran higher with effort-reward imbalance among 81 high-workload workers, and morning saliva ran 21.7 percent higher in 48 high-strain teachers, but Karasek job strain tracked a lower diurnal cortisol area in 254 matched MESA adults. Nobody assigned the job.
30. SMOKE− DECREASESLead → CognitionHigher childhood blood lead tracked lower IQ in Rochester, the Lanphear pool, and Dunedin at age 38 after maternal IQ was in the model. Nobody randomized lead.
31. SMOKE+ INCREASESLoneliness → MortalityWang 2023 pooled 90 prospective cohorts in 2,205,199 adults and found loneliness associated with a 14 percent higher all-cause death risk (effect size 1.14, 1.08 to 1.20). Nobody randomized loneliness.
32. MECHANISM? CONDITIONALMagnesium → HOMA-IRDrenthen 2024 found that 15 mmol a day of oral magnesium for six weeks did not change clamp insulin sensitivity versus placebo in 14 adults with insulin-treated type 2 diabetes (M value 4.6 versus 4.4). An intensive-care loading dose then moved HOMA in stress hyperglycemia (Heidary 2021).
33. MECHANISM− DECREASESMed diet → HOMA-IRIn 418 older Spaniards free of diabetes, PREDIMED-Reus assigned a Mediterranean diet with free extra-virgin olive oil or mixed nuts against low-fat advice. After a median four years the pooled Mediterranean arms had about half the new type 2 diabetes (hazard ratio 0.48). The measure is diabetes incidence, not HOMA-IR.
34. SMOKE+ INCREASESTraffic noise → CV eventsIn a 2023 dose-response meta-analysis of 23 cohort papers and 18.5 million adults, mostly in Europe, each extra 10 dB of modelled road-traffic noise (as Lden) sat with a 2 percent higher risk of incident cardiovascular disease (RR 1.020, 1.006 to 1.035). Nobody randomized the street.
35. MECHANISM? CONDITIONALOmega-3 → CV eventsAssigned 4 g a day of icosapent ethyl cut a five-part ischemic composite in 8179 statin-treated adults with high triglycerides (Bhatt 2019, hazard ratio 0.75); the same dose of mixed EPA plus DHA carboxylic acids did not in 13,078 similar adults (Nicholls 2020, 0.99), and about 400 mg of EPA-DHA in margarine did not after a heart attack (Kromhout 2010).
36. MECHANISM0 NULLOSA → CognitionXu 2025 assigned 148 middle-aged adults with moderate-to-severe obstructive sleep apnea and normal cognition to CPAP plus usual care or usual care alone. At six months the Montreal Cognitive Assessment did not differ (difference -0.04, 95% CI -0.72 to 0.65).
37. MECHANISM0 NULLOSA → CV eventsMcEvoy 2016 found that CPAP plus usual care did not change the risk of a composite cardiovascular event versus usual care alone among 2717 adults with moderate-to-severe obstructive sleep apnea and established coronary or cerebrovascular disease (hazard ratio 1.10, 95% CI 0.91 to 1.32).
38. MECHANISM+ INCREASESOSA → HOMA-IRIn 80 adults with severe OSA and morbid obesity, 12 weeks of CPAP improved two-hour glucose versus conservative care. HOMA-IR did not change in that trial. Pooled CPAP trials later found a modest HOMA-IR drop versus sham or no machine.
39. MECHANISM+ INCREASESOutdoors → 25(OH)DSkin makes 25(OH)D from UV-B; Samefors 2021 then showed that encouraging 40 Swedish nursing-home residents outdoors for two summer months raised median serum 25(OH)D from 42.5 to 53.5 nmol/L.
40. SMOKE0 NULLOT → TrustIntranasal oxytocin does not raise trusting behavior: a 2026 registered report found no effect in 211 people, and a pooled equivalence test with the 2020 registered replication (532 people) put the leftover effect inside a minimal range.
41. SMOKE+ INCREASESAir pollution → MortalityHao 2023's Medicare cohort of about 73.4 million adults 65 and older found a 4.0 percent higher all-cause death rate per 3.68 µg/m³ of annual PM2.5 (HR 1.040, 1.039 to 1.041). Nobody randomized the air.
42. SMOKE0 NULLPower pose → TTwo minutes in a stance does not identify an androgen change. The larger RCTs are nulls. The TED talk is not a protocol.
43. MECHANISM− DECREASESCoQ10 → CV eventsMortensen 2015 found that 100 mg of coenzyme Q10 three times a day for two years halved a first major heart-failure event versus placebo in 420 people (15 percent versus 26 percent; hazard ratio 0.50). Morisco 1994 had already cut hospitalizations for worsening failure in 641 people on 2 mg per kg a day.
44. SMOKE0 NULLT/C → StatusT/C is a derived index. Stamp it smoke unless the paper pre-registered the ratio and showed the interaction.
45. MECHANISM+ INCREASESShift work → HOMA-IRFour assigned night shifts with a nine-hour sleep opportunity reduced clamp insulin sensitivity about 25 percent in healthy adults. Matched day-shift controls did not move.
46. MECHANISM+ INCREASESSitting → HOMA-IRIn 24 adults with type 2 diabetes, an 8-hour day of unbroken sitting produced a higher postprandial insulin curve than the same day with 3-minute walking or simple resistance breaks every half hour.
47. MECHANISM− DECREASESSleep → CortisolSix 4-hour nights raised evening cortisol in eleven young men. Four nights of 5-to-9-hour beds in eighty-eight adolescents did not move evening or morning saliva cortisol.
48. MECHANISM− DECREASESSleep → HOMA-IRExperimental sleep restriction lowers insulin sensitivity. Spiegel, Leproult, and Van Cauter put eleven young men into sleep debt and watched glucose tolerance fall into a pre-diabetic range. Later randomized weeks, including clamp studies and a six-week trial in thirty-eight women, found the same direction.
49. SMOKE~ NON-LINEARSleep → MortalityYin 2018's dose-response meta of prospective cohorts found a U-shaped link between hours of sleep and all-cause death: each hour under 7 tracked a 6 percent higher risk, and each hour over 7 tracked 13 percent. Nobody randomized time in bed.
50. MECHANISM+ INCREASESSleep → TIn men, experimental sleep loss lowers testosterone. A week of 5-hour nights cut daytime T about 10 to 15 percent. One night of total wake cut 24-hour T. All-night deprivation cut total T; REM-only did not. In 37 healthy adults, T also fell during long wake; caffeine lessened that drop.
51. SMOKE+ INCREASESSmoking → HOMA-IRIn Cho 2022, 4,043 Korean adults without diabetes who met both urine NNAL and cotinine smoker cutoffs had higher triglyceride-glucose odds than people who met neither (men 1.74, women 2.01 after age, BMI, drink, walk, income, and pack-years); Eliasson 1997's 17 quitters, against 23 who kept smoking, improved clamp insulin sensitivity in eight weeks despite modest weight gain.
52. MECHANISM0 NULLT → CV eventsLincoff 2023 found that daily testosterone gel was noninferior to placebo for a first major cardiac event among 5198 men aged 45 to 80 with hypogonadism and preexisting or high-risk cardiovascular disease (hazard ratio 0.96, 95% CI 0.78 to 1.17).
53. MECHANISMUNSIGNEDT → RiskLarge randomized trials do not show that a testosterone dose moves lottery risk. A preregistered study of 1,000 men found no effect of a single nasal dose. Four weeks of testosterone in 200 postmenopausal women also left risk attitudes unchanged.
54. MECHANISM? CONDITIONALT → StatusIn observational samples, testosterone tracks status-seeking mainly when cortisol is low; the interaction is small, heterogeneous, and still contested.
55. LOOP+ INCREASEST → Win/lossThe other half of the loop: higher T states make another challenge more likely. Evidence is thinner than the win→T half.
56. MECHANISM? CONDITIONALT → UG rejectExogenous T changes bargaining — but not in the folk direction, and not in every lab.
57. MECHANISM− DECREASESLifting → HOMA-IRWaters 2022 randomized 160 older adults with obesity and found that resistance training during matched diet-induced weight loss of about 9 to 10 percent raised the OGTT insulin sensitivity index 39 percent; combination training raised it 86 percent.
58. MECHANISM~ NON-LINEARLifting → TAhtiainen 2016: twelve months of resistance training left testosterone production and clearance unchanged in five men near 28 and eight men near 70; a 5×10RM session could raise serum T by slowing clearance.
59. MECHANISM? CONDITIONALEating window → HOMA-IRIn eight men with prediabetes, a supervised 6-hour early eating window improved insulin sensitivity without weight loss; later RCTs show the effect depends on clock time and calories.
60. MECHANISM+ INCREASESUPF → HOMA-IRIn Hall's ward trial and Hamano's Tokyo crossover, ultra-processed menus matched for presented calories raised ad-lib intake and weight; Chen's US cohorts associate the same food class with later type 2 diabetes, not with measured HOMA-IR.
61. MECHANISM0 NULL25(OH)D → CognitionKang 2022 found that 2000 IU a day of vitamin D3 versus placebo did not slow cognitive decline among 4218 adults aged 60 and older in VITAL (pooled annual global-score difference 0.01, 95% CI -0.01 to 0.02). Corbett 2025 and Pham 2023 then also found no cognitive gain.
62. MECHANISM0 NULL25(OH)D → DepressionOkereke 2020 found that 2000 IU a day of vitamin D3 versus placebo did not change the risk of depression or clinically relevant depressive symptoms among 18353 adults aged 50 and older in VITAL-DEP (hazard ratio 0.97, 95% CI 0.87 to 1.09). Rahman 2023 then found the same null on PHQ-9 in D-Health.
63. MENDELIAN0 NULL25(OH)D → EmploymentGenetic instruments for 25(OH)D do not buy you a job. The observational D–success link was a marker.
64. SMOKE0 NULL25(OH)D → IncomePeople with money and outdoor lives have higher 25(OH)D. That is not D causing the money.
65. MENDELIAN~ NON-LINEAR25(OH)D → MortalitySutherland 2022 found an L-shaped Mendelian randomization link between genetically predicted 25(OH)D and all-cause death in 307,601 white-European UK Biobank adults, with risk falling until about 50 nmol/L, while Trummer 2013 and Ong 2019 found no linear genetic effect.
66. SMOKE+ INCREASESWidowhood → MortalityIn 373,189 elderly US couples followed from 1993 to 2002, a wife's death sat with 18 percent higher all-cause death hazard in the husband (HR 1.18, 1.16 to 1.19) and a husband's death sat with 16 percent in the wife (HR 1.16, 1.14 to 1.17). Nobody assigned widowhood.
67. LOOP+ INCREASESWin/loss → TWinning a meaningful contest raises T relative to losing. That T change then colors the next contest. A loop, not an arrow.
68. MECHANISM+ INCREASESZinc → TMahajan 1982 found that 50 mg of elemental zinc as acetate a day for six months raised serum testosterone from 2.8 to 5.2 ng per milliliter versus no change on placebo in 20 men on hemodialysis. Prasad 1997 then saw testosterone fall under zinc restriction in four young men and rise after zinc gluconate in nine older men.