Every change to every link.

Outcome match, Outcome measured: The estimate measures: acute coronary events (heart attack or coronary death). measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: smoking_chances (weight 1.0), bprf_smoking (weight 0.58). reviews_counted

Outcome match, Outcome measured: The estimate measures: acute coronary events (heart attack or coronary death). measured_outcome

Grade override: Curator override to Moderate: Accepted by the verifier: an individual-participant analysis of cohorts (CHANCES) with a gradient by time since quitting. T9b proposes Suggestive because z is 3.0, on the line. New disagreement for the owner (card G3 A): kept at the verified grade until the owner decides. grade_overrideT9b

Reviews counted: Counted from the cited syntheses under the review-counting rule: smoking_chances (weight 1.0). reviews_counted

Outcome match, Outcome measured: The estimate measures: ischaemic heart disease. measured_outcome

Grade basis, Grade override: Curator override to Moderate: Not settled by the curation pass of card G12, so the verifier's best estimate (Moderate) applies. Two bodies rate the pair differently (card G9): Burden of Proof gives two stars, and the US Surgeon General (2006) judged the evidence sufficient to infer a causal relationship. The text of the 2006 report was not read. G9 applies: two bodies rate the pair differently. Burden of Proof: two stars (evidence weak; conservative lower bound about 8%). The US Surgeon General (2006) judged the evidence sufficient to infer a causal relationship. The curator grades it and names both bodies. The pooled IHD estimate is 1.27 (1.10-1.48). Moderate. The 2006 report was not read, so its wording is not quoted here. grade_override and grade_basis

Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: shs_fischer (weight 1.0), bprf_shs (weight 0.64). reviews_counted

Outcome match, Outcome measured: The estimate measures: fatal myocardial infarction only. measured_outcome

Grade basis: Suggestive: the estimate covers fatal heart attack only, for ever use of Swedish and US products (review note). Kept by the verifier. grade_basis

Sources: 21 Oct 2026 grading reform v2.0: outcome_match narrower, measured outcome recorded (cards O1 and O5). estimate_source smokeless_boffetta, estimate_design cohort synthesis, measure RR, by rule E1 (cards G8, K5, P2). adjudication recorded for each cited body that rules on causality, with pair_match (rules T3, T4, card G5). grade_basis records why the grade stands (card G3) 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): gbd2021rf removed from the sources (blocker 6): the body does not rate this pair (adjudication pair_match false), so it cannot count as a replication synthesis either; its adjudication entry goes with it. corrections.json _whyT3T4

Outcome match, Outcome measured: The estimate measures: ischaemic heart disease (pooled PSC and APCSC cohorts). measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: metab_rf_singh (weight 1.0), psc_bp (weight 0.0), peters2013_sbp_stroke_ihd (weight 0.0), bop_sbp_ihd (weight 0.64). reviews_counted

Sources: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5): Sources added (card K1): peters2013_sbp_stroke_ihd; the displayed estimate is unchanged. estimate_source metab_rf_singh, estimate_design pooled IPD, measure RR, by rule E1 (cards G8, K5, P2). triangulation recorded as objects with their sources, same exposure and same outcome (card G11). adjudication recorded for each cited body that rules on causality, with pair_match (rules T3, T4, card G5). external_grades transcribed from the caveat, the review note or the quote (rule T11). 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): triangulation.mr supportive with its sources recorded (lieb2013_bp_cad_mr) (card G11, blocker 4): Blood-pressure-raising variants raise coronary artery disease (Lieb 2013). hand value(s) ess removed (minor e): the build computes them from the counting rule (card K1). corrections.json _whyT3T4T11

Outcome match, Outcome measured: The estimate measures: CHD (non-fatal MI or fatal CHD). measured_outcome

Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: erfc_lipids (weight 1.0), gbd2021rf (weight 0.75). reviews_counted

Sources: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5). estimate_source erfc_lipids, estimate_design pooled IPD, measure HR, by rule E1 (cards G8, K5, P2). triangulation recorded as objects with their sources, same exposure and same outcome (card G11). adjudication recorded for each cited body that rules on causality, with pair_match (rules T3, T4, card G5) 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): triangulation.mr supportive with its sources recorded (ference2012_ldl_chd_mr) (card G11, blocker 4): Lifelong lower LDL cholesterol through nine variants lowers coronary heart disease (Ference 2012). corrections.json _whyT3T4

Outcome match, Outcome measured: The estimate measures: CHD. measured_outcome

Sources: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5). estimate_source erfc_lpa, estimate_design pooled IPD, measure RR, by rule E1 (cards G8, K5, P2). triangulation recorded as objects with their sources, same exposure and same outcome (card G11) 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): triangulation.mr supportive with its sources recorded (clarke2009_lpa_chd, kamstrup2009_lpa_mi) (card G11, blocker 4): LPA variants that raise lipoprotein(a) raise coronary disease (Clarke 2009; Kamstrup 2009). corrections.json _why

Outcome match, Outcome measured: The estimate measures: CHD. measured_outcome

Sources: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5). estimate_source erfc_lipids, estimate_design pooled IPD, measure HR, by rule E1 (cards G8, K5, P2). triangulation recorded as objects with their sources, same exposure and same outcome (card G11). conflict records both sides and their sources from the caveat and review note (rule C2, card G1) 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): triangulation.mr supportive with its sources recorded (do2013_tg_cad_mr) (card G11, blocker 4): Triglyceride-raising variants raise coronary artery disease after accounting for LDL and HDL (Do 2013). 23 Oct 2026 fix round 2 (verifier of the data step, second check, issues 1 to 7): conflict.side_a rewritten to state the cited genetic analysis (Do 2013) and do2013_tg_cad_mr added to conflict.sources (issue 3, rule C2). corrections.json _whyC2

Outcome match, Outcome measured: The estimate measures: CHD. measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: huxley2006_diabetes_chd (weight 1.0), erfc_diabetes (weight 0.44), peters2014_diabetes_chd (weight 0.44). reviews_counted

Sources: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5): Sources added (card K1): peters2014_diabetes_chd, huxley2006_diabetes_chd; the displayed estimate is unchanged. estimate_source erfc_diabetes, estimate_design pooled IPD, measure HR, by rule E1 (cards G8, K5, P2). triangulation recorded as objects with their sources, same exposure and same outcome (card G11). corrections.json _why

Outcome match, Outcome measured: The estimate measures: CHD. measured_outcome

Grade basis: Suggestive: a modest 1.17 from one consortium, and fasting glucose adds little to risk prediction (review note). Kept by the verifier. grade_basis

Outcome match, Outcome measured: The estimate measures: CHD. measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: psc_bmi (weight 1.0), erfc_adiposity (weight 0.29), gbd2021rf (weight 0.72). reviews_counted

Sources: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5). estimate_source erfc_adiposity, estimate_design pooled IPD, measure HR, by rule E1 (read by hand) (cards G8, K5, P2). triangulation recorded as objects with their sources, same exposure and same outcome (card G11). adjudication recorded for each cited body that rules on causality, with pair_match (rules T3, T4, card G5) 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): triangulation.mr supportive with its sources recorded (larsson2020_bmi_cvd_mr, dale2017_adiposity_mr) (card G11, blocker 4): Genetically predicted BMI raises coronary disease (Dale 2017, OR 1.36 per SD; Larsson 2020). hand value(s) ess removed (minor e): the build computes them from the counting rule (card K1). corrections.json _whyT3T4

Outcome match, Outcome measured, Related outcome: The estimate measures: cardiovascular disease. The link sits in the related-outcome layer, outside the counts and the score. measured_outcome

Reviews counted: Counted from the cited syntheses under the review-counting rule: erfc_adiposity (weight 1.0). reviews_counted

Sources: 21 Oct 2026 grading reform v2.0: outcome_match broader, measured outcome recorded (cards O1 and O5). estimate_source erfc_adiposity, estimate_design pooled IPD, measure HR, by rule E1 (cards G8, K5, P2). triangulation recorded as objects with their sources, same exposure and same outcome (card G11) 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): triangulation.mr supportive with its sources recorded (emdin2017_whr_mr, dale2017_adiposity_mr) (card G11, blocker 4): Genetically predicted waist-to-hip ratio raises coronary heart disease (Dale 2017, OR 1.48 per SD; Emdin 2017). corrections.json _why

Outcome match, Outcome measured: The estimate measures: myocardial infarction. measured_outcome

Sources: 21 Oct 2026 grading reform v2.0: outcome_match narrower, measured outcome recorded (cards O1 and O5). estimate_source metsyn_mottillo, estimate_design cohort synthesis, measure RR, by rule E1 (cards G8, K5, P2). adjudication recorded for each cited body that rules on causality, with pair_match (rules T3, T4, card G5) 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): gbd2021rf removed from the sources (blocker 6): the body does not rate this pair (adjudication pair_match false), so it cannot count as a replication synthesis either; its adjudication entry goes with it. corrections.json _whyT3T4

Outcome match, Outcome measured, Related outcome: The estimate measures: total cardiovascular disease. The link sits in the related-outcome layer, outside the counts and the score. measured_outcome

Reviews counted: Counted from the cited syntheses under the review-counting rule: pa_cvd_meta (weight 1.0). reviews_counted

Outcome match, Outcome measured, Related outcome: The estimate measures: cardiovascular disease. The link sits in the related-outcome layer, outside the counts and the score. measured_outcome

Sources: 21 Oct 2026 grading reform v2.0: outcome_match broader, measured outcome recorded (cards O1 and O5). estimate_source sedentary_pandey, estimate_design cohort synthesis, measure HR, by rule E1 (cards G8, K5, P2). adjudication recorded for each cited body that rules on causality, with pair_match (rules T3, T4, card G5) 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): gbd2021rf removed from the sources (blocker 6): the body does not rate this pair (adjudication pair_match false), so it cannot count as a replication synthesis either; its adjudication entry goes with it. corrections.json _whyT3T4

Outcome match, Outcome measured, Related outcome: The estimate measures: combined CHD and CVD events. The link sits in the related-outcome layer, outside the counts and the score. measured_outcome

Grade: Suggestive: Fitness partly marks hidden disease; CHD and CVD pooled. Rule T8a (ceiling Suggestive). Verified on 21 October 2026 (card G3). Curator record: 21 Oct 2026 grading reform v2.0: outcome_match broader, measured outcome recorded (cards O1 and O5). estimate_source crf_kodama, estimate_design cohort synthesis, measure RR, by rule E1 (cards G8, K5, P2): Grade Moderate to Suggestive (card G3, PROPOSAL section 3.1): Fitness partly marks hidden disease; CHD and CVD pooled. grade_basisT8a

Grade basis: Suggestive: Fitness partly marks hidden disease; CHD and CVD pooled. Rule T8a (ceiling Suggestive). Verified on 21 October 2026 (card G3). grade_basisT8a

Outcome match, Outcome measured, Related outcome: The estimate measures: cardiovascular events. The link sits in the related-outcome layer, outside the counts and the score. measured_outcome

Grade override: Curator override to Moderate: T8a reads the caveat's mention of reverse causation, which refers to J-shaped curves in spot-urine cohorts; the displayed estimate uses repeated 24-hour urine collections. T3 caps at Moderate. New disagreement for the owner (card G3 A): kept at the verified grade until the owner decides. grade_overrideT8aT3

Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: sodium_dr (weight 1.0), sodium_urine_pooled (weight 0.38). reviews_counted

Trial and genetic evidence: 21 Oct 2026 grading reform v2.0: outcome_match broader, measured outcome recorded (cards O1 and O5). estimate_source sodium_urine_pooled, estimate_design pooled IPD, measure HR, by rule E1 (cards G8, K5, P2). triangulation.rct supportive, same_exposure and same_outcome false (cards G6, G11): SSaSS tested a salt substitute that lowers sodium and adds potassium; the Cochrane review of salt reduction measured blood pressure. grade_override keeps Moderate: T8a reads the caveat's mention of reverse causation, which refers to J-shaped curves in spot-urine cohorts; the displayed estimate uses repeated 24-hour urine collections: T3 caps at Moderate. New disagreement for the owner (card G3 A): kept at the verified grade until the owner decides. corrections.json _whyT8aT3

Outcome match, Outcome measured: The estimate measures: total CHD (fatal and non-fatal). measured_outcome

Grade override: Curator override to Moderate: Kept by the verifier: GRADE 'low' is the starting level for observational nutrition evidence, and lipid trials and GBD support the pair. T11 reads 'very low to low certainty' as very low. grade_overrideT11

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: transfat_desouza (weight 1.0), gbd2021rf (weight 0.64). reviews_counted

Outcome match, Outcome measured: The estimate measures: incident CHD. measured_outcome

Grade: Suggestive: Five cohorts, z 2.4. Verified on 21 October 2026 (card G3). Curator record: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5). estimate_source meat_micha, estimate_design cohort synthesis, measure RR, by rule E1 (cards G8, K5, P2): Grade Moderate to Suggestive (card G3, PROPOSAL section 3.1): Five cohorts, z 2.4. grade_basis

Grade basis: Suggestive: Five cohorts, z 2.4. Verified on 21 October 2026 (card G3). grade_basis

Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: meat_micha (weight 1.0), foodgroups_schwingshackl (weight 0.58). reviews_counted

Outcome match, Outcome measured: The estimate measures: incident CHD. measured_outcome

Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: meat_micha (weight 1.0), foodgroups_schwingshackl (weight 0.58), bprf_meat (weight 0.5). reviews_counted

Outcome match, Outcome measured: The estimate measures: CHD. measured_outcome

Grade basis: The link already cites a later dose-response meta-analysis of 64 prospective studies (Sun 2023): HR 1.11 (1.05-1.17) per 250 mL a day for CHD, linear. With two syntheses (2014, 2023) T9 does not apply; T4 met; T7 met by the linear gradient; T3 fails, so the ceiling is Moderate. SSB is not on the reverse-causation watch list. grade_basisT9T4T7T3

Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: ssb_chd_xi (weight 1.0), ssb_fructose_dr (weight 0.62). reviews_counted

The estimate measures: cardiovascular events. The link sits in the related-outcome layer, outside the counts and the score. measured_outcome

Outcome match, Outcome measured: The estimate measures: CHD. measured_outcome

Grade override: Curator override to Moderate: T8c (healthy-user check, amended) proposes Suggestive: a beneficial diet judged on observational data alone. New disagreement for the owner (card G3 A): kept at the verified grade until the owner decides. grade_overrideT8c

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: wholegrain_aune (weight 1.0), fibre_reynolds (weight 0.38), gbd2021rf (weight 0.5). reviews_counted

Outcome match, Outcome measured: The estimate measures: CHD. measured_outcome

Grade override: Curator override to Moderate: T8c (healthy-user check, amended) proposes Suggestive: a beneficial diet judged on observational data alone. New disagreement for the owner (card G3 A): kept at the verified grade until the owner decides. grade_overrideT8c

Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: fruitveg_aune (weight 1.0), gbd2021rf (weight 0.58). reviews_counted

Outcome match, Outcome measured: The estimate measures: ischaemic heart disease. measured_outcome

Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: fruitveg_aune (weight 1.0), bprf_veg (weight 0.5). reviews_counted

Outcome match, Outcome measured: The estimate measures: fatal ischaemic heart disease only. measured_outcome

Grade override: Curator override to Moderate: T9a proposes Weak because the quote names four studies for non-fatal heart disease; the displayed fatal estimate pools six studies (I-squared 28%). New disagreement for the owner (card G3 A): kept at the verified grade until the owner decides. grade_overrideT9a

Reviews counted: Counted from the cited syntheses under the review-counting rule: nutslegumes_afshin (weight 1.0). reviews_counted

Trial and genetic evidence: 21 Oct 2026 grading reform v2.0: outcome_match narrower, measured outcome recorded (cards O1 and O5). estimate_source nutslegumes_afshin, estimate_design cohort synthesis, measure RR, by rule E1 (cards G8, K5, P2). triangulation.rct supportive, same_exposure false (cards G6, G11): PREDIMED tested a Mediterranean diet supplemented with nuts against a control diet. grade_override keeps Moderate: T9a proposes Weak because the quote names four studies for non-fatal heart disease; the displayed fatal estimate pools six studies (I-squared 28%): New disagreement for the owner (card G3 A): kept at the verified grade until the owner decides. corrections.json _whyT9a

Outcome match, Outcome measured: The estimate measures: total ischaemic heart disease. measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: nutslegumes_afshin (weight 1.0), gbd2021rf (weight 0.67). reviews_counted

Outcome match, Outcome measured: The estimate measures: CHD incidence. measured_outcome

Grade override: Curator override to Moderate: T8c (healthy-user check, amended) proposes Suggestive: a beneficial diet judged on observational data alone. New disagreement for the owner (card G3 A): kept at the verified grade until the owner decides. grade_overrideT8c

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: fibre_reynolds (weight 1.0). reviews_counted

Outcome match, Outcome measured: The estimate measures: CHD incidence. measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: omega3_chd_alexander (weight 1.0), fish_chd_jiang (weight 0.38). reviews_counted

Trial and genetic evidence: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5). estimate_source fish_chd_jiang, estimate_design cohort synthesis, measure RR, by rule E1 (cards G8, K5, P2). triangulation.rct null, same_exposure false (cards G6, G11): The trials tested omega-3 supplements; the link measures fish consumption corrections.json _why

Outcome match, Outcome measured: The estimate measures: CHD incidence. measured_outcome

Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: medd_grosso (weight 1.0), medd_becerra (weight 0.29). reviews_counted

Outcome match, Outcome measured: The estimate measures: IHD incidence (fatal and non-fatal). measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: alc_heavy_ihd_roerecke (weight 1.0). reviews_counted

Outcome match, Outcome measured: The estimate measures: Death from ischaemic heart disease only. measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: escape_coronary (weight 1.0), pm25_ihd_alexeeff (weight 0.55), who_air (weight 0.17), gbd2021rf (weight 0.38). reviews_counted

Outcome match, Outcome measured: The estimate measures: ischaemic heart disease. measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: hap_health_lee2020 (weight 1.0), solidfuel_cve_huang (weight 0.17). reviews_counted

Outcome match, Outcome measured: The estimate measures: new cases of ischaemic heart disease. measured_outcome

Grade basis: Moderate: the WHO review rates the evidence high quality; only a low-confidence rule (z 2.3) points lower. Accepted by the verifier. grade_basis

Outcome match, Outcome measured: The estimate measures: CHD. measured_outcome

Grade basis: Moderate: a dose-response and a large effect; only a low-confidence rule on interval width points lower. Kept by the verifier. grade_basis

Outcome match, Outcome measured: The estimate measures: incident CHD (IPD-Work). measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: ipd_work (weight 1.0). reviews_counted

Outcome match, Outcome measured: The estimate measures: new cases of coronary heart disease. measured_outcome

Sources: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5). estimate_source isolation_valtorta, estimate_design cohort synthesis, measure RR, by rule E1 (cards G8, K5, P2) 23 Oct 2026 fix round 2 (verifier of the data step, second check, issues 1 to 7): Source holt_lunstad removed from the link's sources (issue 2, card K7 A topic check, PROPOSAL section 5.1): it is a meta-analysis of social relationships and death from any cause, with no estimate for coronary heart disease, so it does not replicate this pair and must not count in the replication term. review_note: a dated sentence records that Holt-Lunstad 2010 no longer appears among the sources (issue 2). corrections.json _why

Outcome match, Outcome measured: The estimate measures: CHD. measured_outcome

Grade: Suggestive: Funnel asymmetry; treatment trials did not cut events. Rules T2, T8a, T10 (ceiling Suggestive). Verified on 21 October 2026 (card G3). Curator record: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5). estimate_source depression_chd, estimate_design cohort synthesis, measure RR, by rule E1 (cards G8, K5, P2). triangulation.rct null, same_outcome false (cards G6, G11): The trials treated depression after a heart attack, so they measure further cardiac events in people who already have heart disease: No trial is cited. Grade Moderate to Suggestive (card G3, PROPOSAL section 3.1): Funnel asymmetry; treatment trials did not cut events. grade_basisT2T8aT10

Grade basis: Suggestive: Funnel asymmetry; treatment trials did not cut events. Rules T2, T8a, T10 (ceiling Suggestive). Verified on 21 October 2026 (card G3). grade_basisT2T8aT10

Trial and genetic evidence, Sources: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5). estimate_source depression_chd, estimate_design cohort synthesis, measure RR, by rule E1 (cards G8, K5, P2). triangulation.rct null, same_outcome false (cards G6, G11): The trials treated depression after a heart attack, so they measure further cardiac events in people who already have heart disease: No trial is cited. Grade Moderate to Suggestive (card G3, PROPOSAL section 3.1): Funnel asymmetry; treatment trials did not cut events. 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): triangulation.rct null with its sources recorded (enrichd2003_trial), same_outcome false (card G11, blocker 4): ENRICHD treated depression after a heart attack and did not change death or recurrent infarction (2,481 patients); it measures further cardiac events in people who already have heart disease. corrections.json _why

Outcome match, Outcome measured: The estimate measures: CHD. measured_outcome

Grade basis: Moderate: two reviews agree; only the reverse-causation watch list points lower. Kept by the verifier. grade_basis

Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: sleep_cappuccio (weight 1.0), shortsleep_itani (weight 0.5), sleep_yin (weight 0.17). reviews_counted

Outcome match, Outcome measured: The estimate measures: CHD. measured_outcome

Sources: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5). estimate_source insomnia_umbrella, estimate_design cohort synthesis, measure RR, by rule E1 (cards G8, K5, P2). triangulation recorded as objects with their sources, same exposure and same outcome (card G11) 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): triangulation.mr supportive with its sources recorded (yuan2021_insomnia_cvd_mr) (card G11, blocker 4): Genetic liability to insomnia raises coronary artery disease (OR 1.19, Yuan 2021). corrections.json _why

Outcome match, Outcome measured: The estimate measures: CHD. measured_outcome

Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: osa_dong (weight 1.0), osa_wang (weight 0.0), osa_xie (weight 0.44). reviews_counted

Outcome match, Outcome measured: The estimate measures: coronary events (MI, coronary death and related events). measured_outcome

Grade: Suggestive: Any shift work; I-squared 85%. Rule T10 (ceiling Suggestive). Verified on 21 October 2026 (card G3). Curator record: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5). estimate_source shiftwork_vyas, estimate_design cohort synthesis, measure RR, by rule E1 (cards G8, K5, P2): Grade Moderate to Suggestive (card G3, PROPOSAL section 3.1): Any shift work; I-squared 85%. grade_basisT10

Grade basis: Suggestive: Any shift work; I-squared 85%. Rule T10 (ceiling Suggestive). Verified on 21 October 2026 (card G3). grade_basisT10

Outcome match, Outcome measured: The estimate measures: coronary artery disease. measured_outcome

Sources: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5). estimate_source sep_khaing, estimate_design cohort synthesis, measure RR, by rule E1 (cards G8, K5, P2). triangulation recorded as objects with their sources, same exposure and same outcome (card G11) 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): triangulation.mr supportive with its sources recorded (tillmann2017_education_chd_mr) (card G11, blocker 4): Genetic predisposition to more education lowers coronary heart disease (Tillmann 2017). corrections.json _why

The estimate measures: coronary artery disease. measured_outcome

Outcome match, Outcome measured, Related outcome: The estimate measures: combined high blood pressure, CHD, heart attack and stroke. The link sits in the related-outcome layer, outside the counts and the score. measured_outcome

Grade basis: Kept at Weak by the verified dry run: MR found no effect on coronary disease (mr=null), adversity recalled in adulthood, composite outcome including hypertension. The note gives these reasons. grade_basis

Outcome match, Outcome measured: The estimate measures: CHD. measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: gbd_ckd (weight 1.0), ckdpc_jama (weight 0.38). reviews_counted

Outcome match, Outcome measured: The estimate measures: ischaemic heart disease. measured_outcome

Grade: Suggestive: Shared cardiovascular predisposition. Rule T8a (ceiling Suggestive). Verified on 21 October 2026 (card G3). Curator record: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5). estimate_source preeclampsia_bellamy, estimate_design cohort synthesis, measure RR, by rule E1 (cards G8, K5, P2): Grade Moderate to Suggestive (card G3, PROPOSAL section 3.1): Shared cardiovascular predisposition. grade_basisT8a

Grade basis: Suggestive: Shared cardiovascular predisposition. Rule T8a (ceiling Suggestive). Verified on 21 October 2026 (card G3). grade_basisT8a

Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: preeclampsia_bellamy (weight 1.0), preeclampsia_wu (weight 0.67). reviews_counted

The estimate measures: cardiovascular events. The link sits in the related-outcome layer, outside the counts and the score. measured_outcome

Outcome match, Outcome measured: The estimate measures: CHD (WHI trial, intervention phase). measured_outcome

Reviews counted, Rests on primary studies: No synthesis is cited, so N_eff is set to 1 and the link is flagged as resting on primary studies. reviews_counted

Outcome match, Outcome measured: The estimate measures: major coronary events (non-fatal heart attack or coronary death). measured_outcome

Reviews counted: Counted from the cited syntheses under the review-counting rule: ctt (weight 1.0). reviews_counted

Outcome match, Outcome measured: The estimate measures: CHD. measured_outcome

Reviews counted: Counted from the cited syntheses under the review-counting rule: bp_ettehad (weight 1.0). reviews_counted

The estimate measures: cardiovascular disease. The link sits in the related-outcome layer, outside the counts and the score. measured_outcome

Outcome match, Outcome measured: The estimate measures: new cases of coronary heart disease. measured_outcome

Grade basis: Suggestive: no trial of periodontal treatment has measured coronary events, and shared causes such as smoking are hard to exclude (T8a). Kept by the verifier. grade_basisT8a

Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: perio_humphrey (weight 1.0), perio_gao (weight 0.72). reviews_counted

Trial and genetic evidence: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5). estimate_source perio_gao, estimate_design cohort synthesis, measure RR, by rule E1 (cards G8, K5, P2). triangulation recorded as objects with their sources, same exposure and same outcome (card G11) 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): triangulation.rct none (status was null) with its sources recorded (none) (card G11, blocker 4): No trial of periodontal treatment with coronary events as the outcome is cited; treatment trials measure inflammatory markers and carotid thickness. Grade stays Suggestive (card G3, blocker 4); grade_basis rewritten: Suggestive: no trial of periodontal treatment has measured coronary events, and shared causes such as smoking are hard to exclude (T8a). Kept by the verifier. corrections.json _whyT8a

Outcome match, Outcome measured: The estimate measures: CHD. measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: vitd_cvd_wang2012 (weight 1.0). reviews_counted

The estimate measures: CHD. measured_outcome

Outcome match, Outcome measured: The estimate measures: ischaemic stroke. measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: lawes2004_bp_stroke (weight 1.0), metab_rf_singh (weight 0.64), psc_bp (weight 0.0), peters2013_sbp_stroke_ihd (weight 0.0). reviews_counted

Sources: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5): Sources added (card K1): peters2013_sbp_stroke_ihd, lawes2004_bp_stroke; the displayed estimate is unchanged. estimate_source metab_rf_singh, estimate_design pooled IPD, measure RR, by rule E1 (cards G8, K5, P2). triangulation recorded as objects with their sources, same exposure and same outcome (card G11). adjudication recorded for each cited body that rules on causality, with pair_match (rules T3, T4, card G5). corrections.json _whyT3T4

Outcome match, Outcome measured: The estimate measures: haemorrhagic stroke. measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: lawes2004_bp_stroke (weight 1.0), ariesen2003_ich_risk (weight 0.0), metab_rf_singh (weight 0.64), psc_bp (weight 0.0). reviews_counted

Sources: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5): Sources added (card K1): lawes2004_bp_stroke, ariesen2003_ich_risk; the displayed estimate is unchanged. estimate_source metab_rf_singh, estimate_design pooled IPD, measure RR, by rule E1 (cards G8, K5, P2). triangulation recorded as objects with their sources, same exposure and same outcome (card G11). 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): triangulation.rct supportive with its sources recorded (chapman2004_progress_subtype) (card G11, blocker 4): In PROGRESS, blood pressure lowering cut intracerebral haemorrhage by 50% (26 to 67) among 6,105 people with a previous stroke or transient ischaemic attack. hand value(s) ess removed (minor e): the build computes them from the counting rule (card K1). corrections.json _why

Outcome match, Outcome measured: The estimate measures: new ischaemic stroke. measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: smoking_chances (weight 1.0), bprf_smoking (weight 0.58), smoking_stroke_luo (weight 0.0). reviews_counted

Outcome match, Outcome measured: The estimate measures: new haemorrhagic stroke. measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: ariesen2003_ich_risk (weight 1.0), feigin2005_sah_review (weight 0.29), bprf_smoking (weight 0.77), smoking_stroke_luo (weight 0.0). reviews_counted

Sources: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5): Sources added (card K1): ariesen2003_ich_risk, feigin2005_sah_review; the displayed estimate is unchanged. estimate_source smoking_stroke_luo, estimate_design cohort synthesis, measure OR, by rule E1 (cards G8, K5, P2). adjudication recorded for each cited body that rules on causality, with pair_match (rules T3, T4, card G5). corrections.json _whyT3T4

Outcome match, Outcome measured: The estimate measures: ischaemic stroke. measured_outcome

Grade basis: Strong: 104 cohorts (Odutayo 2016) with a 2.3-fold risk; genetic liability to atrial fibrillation raises cardio-embolic ischaemic stroke (T3, T4); anticoagulation trials remove most of the excess risk (another exposure contrast, noted). Verified at Strong on 21 October 2026. grade_basisT3T4

Trial and genetic evidence, Sources: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5). estimate_source afib_odutayo, estimate_design cohort synthesis, measure RR, by rule E1 (cards G8, K5, P2). triangulation.rct supportive, same_exposure false (cards G6, G11): Anticoagulation trials remove most of the excess stroke risk, but they test a drug and leave atrial fibrillation itself unchanged: No trial is cited. grade_override keeps Strong: Anticoagulation trials remove most of the excess stroke risk in atrial fibrillation, but they leave the exposure itself unchanged, so they are recorded with same_exposure false and rule T3 (Moderate) fails as written. No Mendelian randomisation study or adjudicating body is cited yet. New disagreement for the owner (card G3 A): kept at the verified grade until the owner decides. 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): triangulation.mr supportive (status was none) with its sources recorded (wang2022_af_phewas_mr, he2023_af_stroke_mr) (card G11, G2, blocker 4): Genetic liability to atrial fibrillation raises the risk of cardio-embolic ischaemic stroke (Wang 2022) and of stroke (He 2023); strokes caused by atrial fibrillation are almost all ischaemic. triangulation.rct supportive with its sources recorded (hart2007_af_antithrombotic_ma), same_exposure false (card G11, blocker 4): Anticoagulation trials remove most of the excess stroke risk (warfarin cut stroke by 64% in 29 trials), but they test a drug and leave atrial fibrillation itself unchanged. grade_override removed (blocker 2): the recorded genetic evidence now meets rule T3, and T4 has two lines (a cohort synthesis and MR), so Strong follows from the table. Grade stays Strong (blocker 2); grade_basis rewritten: Strong: 104 cohorts (Odutayo 2016) with a 2.3-fold risk; genetic liability to atrial fibrillation raises cardio-embolic ischaemic stroke (T3, T4); anticoagulation trials remove most of the excess risk (another exposure contrast, noted). Verified at Strong on 21 October 2026. corrections.json _whyT3T4

Outcome match, Outcome measured: The estimate measures: ischaemic stroke. measured_outcome

Trial and genetic evidence, Sources: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5): Sources added (card G11): larsson2017_t2d_stroke_mr; the displayed estimate is unchanged. estimate_source erfc_diabetes, estimate_design pooled IPD, measure HR, by rule E1 (cards G8, K5, P2). triangulation.mr supportive recorded with its sources (card G11, research pass): Genetically predicted type 2 diabetes raises large-artery stroke (OR 1.28) and small-vessel stroke (OR 1.21); the cardioembolic subtype is null. The abstract gives no estimate for all ischaemic stroke. corrections.json _why

Outcome match, Outcome measured, Related outcome: The estimate measures: cardiovascular disease. The link sits in the related-outcome layer, outside the counts and the score. measured_outcome

Grade: Suggestive: three Mendelian randomisation studies found no effect of BMI on ischaemic stroke (rule T2, ceiling Suggestive; C3 gives Suggestive by default, card G6). The displayed estimate is the ERFC figure for cardiovascular disease, so the link sits in the related-outcome layer. Lowered from Moderate by the binding rule once the genetic evidence was recorded. Curator record: 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): triangulation.mr null (status was supportive) with its sources recorded (larsson2020_bmi_cvd_mr, dale2017_adiposity_mr, marini2020_obesity_stroke_mr) (card G11, blocker 4): Three Mendelian randomisation studies found no effect of genetically predicted BMI on ischaemic stroke (Dale 2017; Larsson 2020; Marini 2020); waist-to-hip ratio, another exposure, did raise it. Grade Moderate to Suggestive (blocker 4, binding rule T2): Suggestive: three Mendelian randomisation studies found no effect of BMI on ischaemic stroke (rule T2, ceiling Suggestive; C3 gives Suggestive by default, card G6). The displayed estimate is the ERFC figure for cardiovascular disease, so the link sits in the related-outcome layer. Lowered from Moderate by the binding rule once the genetic evidence was recorded. grade_basisT2C3

Grade basis: Suggestive: three Mendelian randomisation studies found no effect of BMI on ischaemic stroke (rule T2, ceiling Suggestive; C3 gives Suggestive by default, card G6). The displayed estimate is the ERFC figure for cardiovascular disease, so the link sits in the related-outcome layer. Lowered from Moderate by the binding rule once the genetic evidence was recorded. grade_basisT2C3

Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: psc_bmi (weight 1.0), erfc_adiposity (weight 0.29). reviews_counted

Trial and genetic evidence, Sources: 21 Oct 2026 grading reform v2.0: outcome_match broader, measured outcome recorded (cards O1 and O5). estimate_source erfc_adiposity, estimate_design pooled IPD, measure HR, by rule E1 (cards G8, K5, P2). triangulation recorded as objects with their sources, same exposure and same outcome (card G11) 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): triangulation.mr null (status was supportive) with its sources recorded (larsson2020_bmi_cvd_mr, dale2017_adiposity_mr, marini2020_obesity_stroke_mr) (card G11, blocker 4): Three Mendelian randomisation studies found no effect of genetically predicted BMI on ischaemic stroke (Dale 2017; Larsson 2020; Marini 2020); waist-to-hip ratio, another exposure, did raise it. Grade Moderate to Suggestive (blocker 4, binding rule T2): Suggestive: three Mendelian randomisation studies found no effect of BMI on ischaemic stroke (rule T2, ceiling Suggestive; C3 gives Suggestive by default, card G6). The displayed estimate is the ERFC figure for cardiovascular disease, so the link sits in the related-outcome layer. Lowered from Moderate by the binding rule once the genetic evidence was recorded. corrections.json _whyT2C3

Outcome match, Outcome measured: The estimate measures: ischaemic stroke. measured_outcome

Sources: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5). estimate_source erfc_lipids, estimate_design pooled IPD, measure HR, by rule E1 (cards G8, K5, P2). triangulation recorded as objects with their sources, same exposure and same outcome (card G11). adjudication recorded for each cited body that rules on causality, with pair_match (rules T3, T4, card G5) 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): triangulation.mr supportive with its sources recorded (hindy2018_lipids_stroke_mr) (card G11, blocker 4): Genetically predicted LDL cholesterol raises ischaemic stroke, mainly the large-artery subtype (Hindy 2018). corrections.json _whyT3T4

Outcome match, Outcome measured, Related outcome: The estimate measures: cardiovascular events. The link sits in the related-outcome layer, outside the counts and the score. measured_outcome

Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: sodium_dr (weight 1.0), sodium_urine_pooled (weight 0.38). reviews_counted

Trial and genetic evidence: 21 Oct 2026 grading reform v2.0: outcome_match broader, measured outcome recorded (cards O1 and O5). estimate_source sodium_urine_pooled, estimate_design pooled IPD, measure HR, by rule E1 (cards G8, K5, P2). triangulation.rct supportive, same_exposure false (cards G6, G11): SSaSS tested a salt substitute that lowers sodium and adds potassium corrections.json _why

Estimate, Interval, lower, Interval, upper: 21 Oct 2026 grading reform v2.0: outcome_match broader, measured outcome recorded (cards O1 and O5). estimate_source interstroke, estimate_design case-control, measure OR, by rule E1 (read by hand) (cards G8, K5, P2). grade_override keeps Moderate: Kept by the verifier: the displayed INTERSTROKE all-stroke case-control odds ratio is the wrong estimate, and the two cited cohort meta-analyses support Moderate for the pair: The estimate should be replaced (card O3). 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): Displayed estimate replaced (card G3 A, kept link 64, card O3; blocker 8): the INTERSTROKE all-stroke case-control odds ratio gives way to the cited cohort synthesis (Garcia 2023, 94 cohorts), 1.25 (1.18 to 1.33) for no activity against the recommended amount, the inverse of 0.80 (0.75 to 0.85); the outcome is still stroke of any type. Grade stays Moderate (blocker 8); grade_basis rewritten: Moderate: two cohort syntheses (Li 2012; Garcia 2023) agree on lower stroke risk with activity, with a gradient (T7). No trial or genetic study is cited, so T3 caps at Moderate. The INTERSTROKE case-control estimate was replaced by the cohort synthesis (cards O3 and G3). Kept at Moderate by the verifier. grade_override keeps Moderate (blocker 8): Kept at Moderate by the verifier (card G3): two cohort syntheses with a gradient support the pair. Rule T8a reads the review note's remark on reverse causation and recall bias, which referred to the INTERSTROKE case-control estimate that has now been replaced by the cohort synthesis of 94 cohorts (Garcia 2023). corrections.json _whyT7T3T8a

Comparison, Comparison type, Caveat, Quoted sentence: INTERSTROKE reports regular physical activity at 0.60 (99% CI 0.52-0.70) for all stroke, whose reciprocal is 1.67 (1.43-1.92) for inactivity - considerably larger than the 1.26 on file. This is a case-control odds ratio with 99% intervals and is vulnerable to reverse causation and recall bias, so the true cohort value probably lies between the two; the dose-response cohort literature finds most of the benefit accrues by 8.75 mMET-hours/week, equivalent to the recommended 150 minutes. Updated 23 Oct 2026: the INTERSTROKE case-control odds ratio was replaced by the cohort dose-response synthesis (Garcia 2023), as the verified dry run and card O3 asked. review_note

Outcome match, Outcome measured, Related outcome: The estimate measures: stroke of any type (incidence, cohort synthesis). The link sits in the related-outcome layer, outside the counts and the score. measured_outcome

Grade basis, Grade override: Curator override to Moderate: Kept at Moderate by the verifier (card G3): two cohort syntheses with a gradient support the pair. Rule T8a reads the review note's remark on reverse causation and recall bias, which referred to the INTERSTROKE case-control estimate that has now been replaced by the cohort synthesis of 94 cohorts (Garcia 2023). Moderate: two cohort syntheses (Li 2012; Garcia 2023) agree on lower stroke risk with activity, with a gradient (T7). No trial or genetic study is cited, so T3 caps at Moderate. The INTERSTROKE case-control estimate was replaced by the cohort synthesis (cards O3 and G3). Kept at Moderate by the verifier. grade_override and grade_basisT7T3T8a

Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: pa_cvd_meta (weight 1.0), pa_bjsm (weight 0.69). reviews_counted

Estimate, Interval, lower, Interval, upper: 21 Oct 2026 grading reform v2.0: outcome_match narrower, measured outcome recorded (cards O1 and O5): Sources added (card K1): larsson2016_alcohol_stroke; the displayed estimate is unchanged. estimate_source ckb_alcohol, estimate_design MR, measure RR, by rule E1 (read by hand) (cards G8, K5, P2). triangulation recorded as objects with their sources, same exposure and same outcome (card G11). 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): Displayed estimate replaced (card G8 A; blocker 8): the genetic estimate per 280 g a week gives way to the cited prospective synthesis, 1.67 (1.25 to 2.23) for more than 4 drinks a day (Larsson 2016, 11 studies); the MR stays in triangulation. 23 Oct 2026 fix round 2 (verifier of the data step, second check, issues 1 to 7): Source zhang2014_alcohol_stroke added (issue 7, card K1): its abstract (PubMed 24820756) reports no significant effect of heavy drinking on haemorrhagic stroke, which the caveat now says. caveat: Zhang 2014 mentioned (issue 7). caveat_zh: the same sentence. review_note: the sentences that described the replaced genetic estimate are trimmed to a dated history (issue 7). grade stays Strong; grade_basis written so the Zhang 2014 result is on record for the owner (issue 7). corrections.json _why

Comparison, Comparison type, Studies, Quoted sentence: Review of 18 Sep 2026: the genetic estimate then on display was per 280 g of alcohol a week of genotype-predicted intake, with no heavy-versus-light contrast. Alcohol acts about twice as strongly on intracerebral haemorrhage as on ischaemic stroke, and mainly through blood pressure. Updated 23 Oct 2026: the displayed estimate is now the prospective synthesis (Larsson 2016); the genetic estimate stays as corroboration (card G8 A). review_note

Outcome match, Outcome measured: The estimate measures: intracerebral haemorrhage (prospective synthesis). measured_outcome

Grade basis: Strong: the prospective synthesis (Larsson 2016, 11 studies, 1.67, 1.25 to 2.23 for more than 4 drinks a day) and the China Kadoorie Mendelian randomisation agree (T3, T4). A second synthesis (Zhang 2014, 27 prospective studies) found no significant effect of heavy drinking on haemorrhagic stroke; it is cited and named in the caveat. Verified at Strong on 21 October 2026. grade_basisT3T4

Caveat: 23 Oct 2026 fix round 2 (verifier of the data step, second check, issues 1 to 7): Source zhang2014_alcohol_stroke added (issue 7, card K1): its abstract (PubMed 24820756) reports no significant effect of heavy drinking on haemorrhagic stroke, which the caveat now says. caveat: Zhang 2014 mentioned (issue 7). caveat_zh: the same sentence. review_note: the sentences that described the replaced genetic estimate are trimmed to a dated history (issue 7). grade stays Strong; grade_basis written so the Zhang 2014 result is on record for the owner (issue 7). 23 Oct 2026 (owner): the 2014 null synthesis (Zhang) is named in the caveat only and no longer cited, so it does not enter the replication term. corrections.json _why

Sources: 21 Oct 2026 grading reform v2.0: outcome_match narrower, measured outcome recorded (cards O1 and O5): Sources added (card K1): larsson2016_alcohol_stroke; the displayed estimate is unchanged. estimate_source ckb_alcohol, estimate_design MR, measure RR, by rule E1 (read by hand) (cards G8, K5, P2). triangulation recorded as objects with their sources, same exposure and same outcome (card G11). 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): Displayed estimate replaced (card G8 A; blocker 8): the genetic estimate per 280 g a week gives way to the cited prospective synthesis, 1.67 (1.25 to 2.23) for more than 4 drinks a day (Larsson 2016, 11 studies); the MR stays in triangulation. 23 Oct 2026 fix round 2 (verifier of the data step, second check, issues 1 to 7): Source zhang2014_alcohol_stroke added (issue 7, card K1): its abstract (PubMed 24820756) reports no significant effect of heavy drinking on haemorrhagic stroke, which the caveat now says. caveat: Zhang 2014 mentioned (issue 7). caveat_zh: the same sentence. review_note: the sentences that described the replaced genetic estimate are trimmed to a dated history (issue 7). grade stays Strong; grade_basis written so the Zhang 2014 result is on record for the owner (issue 7). 23 Oct 2026 (owner): the 2014 null synthesis (Zhang) is named in the caveat only and no longer cited, so it does not enter the replication term. corrections.json _why

Estimate, Interval, lower, Interval, upper: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5): Sources added (card K1): larsson2016_alcohol_stroke, zhang2014_alcohol_stroke; the displayed estimate is unchanged. estimate_source ckb_alcohol, estimate_design MR, measure RR, by rule E1 (read by hand) (cards G8, K5, P2). triangulation recorded as objects with their sources, same exposure and same outcome (card G11). Grade Strong to Moderate (card G12, settled by the research pass). 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): Displayed estimate replaced (card G8 A; blocker 8): the genetic estimate per 280 g a week gives way to the cited prospective synthesis, 1.14 (1.02 to 1.28) for more than 4 drinks a day (Larsson 2016, 25 studies); the MR stays in triangulation. Grade stays Moderate (blocker 8); grade_basis rewritten: Moderate: the displayed estimate is the prospective synthesis (Larsson 2016, 25 studies, RR 1.14, 1.02 to 1.28 above four drinks a day) and the China Kadoorie Mendelian randomisation is supportive (T3, T4). Strong is not supported: Zhang 2014 (27 prospective studies) found no significant effect of heavy drinking on ischaemic stroke, and the prospective effect is small. Settled at Moderate by the curation pass (card G12). 23 Oct 2026 fix round 2 (verifier of the data step, second check, issues 1 to 7): review_note: the sentences that described the replaced genetic estimate are trimmed to a dated history (issue 7). corrections.json _whyT3T4

Comparison, Comparison type, Studies, Caveat, Quoted sentence: Review of 18 Sep 2026: the genetic estimate then on display was per 280 g of alcohol a week, with no heavy-versus-light contrast, and that association is log-linear with no threshold. INTERSTROKE separately gives OR 2.09 (1.64-2.67) for high or heavy episodic intake versus never or former drinkers, for all stroke. Updated 23 Oct 2026: the displayed estimate is now the prospective synthesis (Larsson 2016); the genetic estimate stays as corroboration (card G8 A). review_note

Outcome match, Outcome measured: The estimate measures: ischaemic stroke (prospective synthesis). measured_outcome

Grade: Moderate: the displayed estimate is the prospective synthesis (Larsson 2016, 25 studies, RR 1.14, 1.02 to 1.28 above four drinks a day) and the China Kadoorie Mendelian randomisation is supportive (T3, T4). Strong is not supported: Zhang 2014 (27 prospective studies) found no significant effect of heavy drinking on ischaemic stroke, and the prospective effect is small. Settled at Moderate by the curation pass (card G12). Curator record: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5): Sources added (card K1): larsson2016_alcohol_stroke, zhang2014_alcohol_stroke; the displayed estimate is unchanged. estimate_source ckb_alcohol, estimate_design MR, measure RR, by rule E1 (read by hand) (cards G8, K5, P2). triangulation recorded as objects with their sources, same exposure and same outcome (card G11). Grade Strong to Moderate (card G12, settled by the research pass). Curator record: 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): Displayed estimate replaced (card G8 A; blocker 8): the genetic estimate per 280 g a week gives way to the cited prospective synthesis, 1.14 (1.02 to 1.28) for more than 4 drinks a day (Larsson 2016, 25 studies); the MR stays in triangulation. Grade stays Moderate (blocker 8); grade_basis rewritten: Moderate: the displayed estimate is the prospective synthesis (Larsson 2016, 25 studies, RR 1.14, 1.02 to 1.28 above four drinks a day) and the China Kadoorie Mendelian randomisation is supportive (T3, T4). Strong is not supported: Zhang 2014 (27 prospective studies) found no significant effect of heavy drinking on ischaemic stroke, and the prospective effect is small. Settled at Moderate by the curation pass (card G12). grade_basisT3T4

Grade basis: Moderate: the displayed estimate is the prospective synthesis (Larsson 2016, 25 studies, RR 1.14, 1.02 to 1.28 above four drinks a day) and the China Kadoorie Mendelian randomisation is supportive (T3, T4). Strong is not supported: Zhang 2014 (27 prospective studies) found no significant effect of heavy drinking on ischaemic stroke, and the prospective effect is small. Settled at Moderate by the curation pass (card G12). grade_basisT3T4

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: zhang2014_alcohol_stroke (weight 1.0), larsson2016_alcohol_stroke (weight 0.29). reviews_counted

Sources: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5): Sources added (card K1): larsson2016_alcohol_stroke, zhang2014_alcohol_stroke; the displayed estimate is unchanged. estimate_source ckb_alcohol, estimate_design MR, measure RR, by rule E1 (read by hand) (cards G8, K5, P2). triangulation recorded as objects with their sources, same exposure and same outcome (card G11). Grade Strong to Moderate (card G12, settled by the research pass). 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): Displayed estimate replaced (card G8 A; blocker 8): the genetic estimate per 280 g a week gives way to the cited prospective synthesis, 1.14 (1.02 to 1.28) for more than 4 drinks a day (Larsson 2016, 25 studies); the MR stays in triangulation. Grade stays Moderate (blocker 8); grade_basis rewritten: Moderate: the displayed estimate is the prospective synthesis (Larsson 2016, 25 studies, RR 1.14, 1.02 to 1.28 above four drinks a day) and the China Kadoorie Mendelian randomisation is supportive (T3, T4). Strong is not supported: Zhang 2014 (27 prospective studies) found no significant effect of heavy drinking on ischaemic stroke, and the prospective effect is small. Settled at Moderate by the curation pass (card G12). corrections.json _whyT3T4

Outcome match, Outcome measured: The estimate measures: new ischaemic stroke. measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: escape_stroke (weight 1.0), pm25_ihd_alexeeff (weight 0.55), who_air (weight 0.17), gbd2021rf (weight 0.38). reviews_counted

Outcome match, Outcome measured, Related outcome: The estimate measures: all incident stroke. The link sits in the related-outcome layer, outside the counts and the score. measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: egfr_stroke_lee (weight 1.0). reviews_counted

Outcome match, Outcome measured, Related outcome: The estimate measures: stroke (all types). The link sits in the related-outcome layer, outside the counts and the score. measured_outcome

Grade basis: Suggestive: a linear dose-response for stroke of any type, 1.17 (1.02 to 1.34), with the serving size unconfirmed (review note). Accepted by the verifier. grade_basis

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: meat_micha (weight 1.0), foodgroups_schwingshackl (weight 0.58). reviews_counted

Outcome match, Outcome measured, Related outcome: The estimate measures: stroke (all types). The link sits in the related-outcome layer, outside the counts and the score. measured_outcome

Grade override: Curator override to Moderate: T8c (healthy-user check, amended) proposes Suggestive: a beneficial diet judged on observational data alone. The link sits in the related-outcome layer. New disagreement for the owner (card G3 A): kept at the verified grade until the owner decides. grade_overrideT8c

Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: fruitveg_aune (weight 1.0), gbd2021rf (weight 0.58). reviews_counted

Outcome match, Outcome measured: The estimate measures: ischaemic stroke. measured_outcome

Grade override: Curator override to Moderate: T8c (healthy-user check, amended) proposes Suggestive; Burden of Proof rates the pair at three stars, which allows Moderate. New disagreement for the owner (card G3 A): kept at the verified grade until the owner decides. grade_overrideT8c

Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: fruitveg_aune (weight 1.0), bprf_veg (weight 0.5). reviews_counted

Outcome match, Outcome measured, Related outcome: The estimate measures: stroke (all types). The link sits in the related-outcome layer, outside the counts and the score. measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: wholegrain_aune (weight 1.0), gbd2021rf (weight 0.62). reviews_counted

Outcome match, Outcome measured, Related outcome: The estimate measures: stroke incidence (all types). The link sits in the related-outcome layer, outside the counts and the score. measured_outcome

Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: medd_grosso (weight 1.0), medd_becerra (weight 0.29). reviews_counted

Outcome match, Outcome measured: The estimate measures: ischaemic stroke. measured_outcome

Grade: C3/T2: a Mendelian randomisation study of major depression and all ischaemic stroke is null (OR 1.03, 0.92-1.15); only small-vessel stroke is positive. Same exposure, same outcome: record mr as null. Three syntheses agree on the association (Pan 2011: ischaemic HR 1.25, 6 studies; Dong 2012: all stroke RR 1.34, publication bias possible; Ashraf 2024: ischaemic HR 1.30, 44 studies). No treatment trial lowers stroke. Suggestive, consistent with depression and CHD. Curator record: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5): Sources added (card G11): dong2012_depression_stroke, ashraf2024_depression_stroke, cai2019_mdd_stroke_mr; the displayed estimate is unchanged. estimate_source depression_stroke_pan, estimate_design cohort synthesis, measure HR, by rule E1 (cards G8, K5, P2). triangulation.mr null recorded with its sources (card G11, research pass): Seventy-two genetic variants for major depression: null for all ischaemic stroke (OR 1.03, 0.92 to 1.15); positive only for small-vessel stroke (OR 1.33, 1.08 to 1.65). n_studies from the displayed review (card G12, research pass; PubMed 21934057). Grade Moderate to Suggestive (card G12, settled by the research pass). grade_basisC3T2

Grade basis: C3/T2: a Mendelian randomisation study of major depression and all ischaemic stroke is null (OR 1.03, 0.92-1.15); only small-vessel stroke is positive. Same exposure, same outcome: record mr as null. Three syntheses agree on the association (Pan 2011: ischaemic HR 1.25, 6 studies; Dong 2012: all stroke RR 1.34, publication bias possible; Ashraf 2024: ischaemic HR 1.30, 44 studies). No treatment trial lowers stroke. Suggestive, consistent with depression and CHD. grade_basisC3T2

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: depression_stroke_pan (weight 1.0), dong2012_depression_stroke (weight 0.17), ashraf2024_depression_stroke (weight 0.71). reviews_counted

Studies, Trial and genetic evidence, Sources: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5): Sources added (card G11): dong2012_depression_stroke, ashraf2024_depression_stroke, cai2019_mdd_stroke_mr; the displayed estimate is unchanged. estimate_source depression_stroke_pan, estimate_design cohort synthesis, measure HR, by rule E1 (cards G8, K5, P2). triangulation.mr null recorded with its sources (card G11, research pass): Seventy-two genetic variants for major depression: null for all ischaemic stroke (OR 1.03, 0.92 to 1.15); positive only for small-vessel stroke (OR 1.33, 1.08 to 1.65). n_studies from the displayed review (card G12, research pass; PubMed 21934057). Grade Moderate to Suggestive (card G12, settled by the research pass). corrections.json _why

Outcome match, Outcome measured, Related outcome: The estimate measures: total stroke. The link sits in the related-outcome layer, outside the counts and the score. measured_outcome

Grade basis: Suggestive: the source gives P = 0.047 for the stroke estimate, so the lower bound sits just above 1 before rounding (about 1.002) and rule T1 does not apply. One estimate for stroke of any type (related-outcome layer); short sleep is prone to reverse causation (T8c). The approved change to Weak (card G3) was conditional on the unrounded bound and is withdrawn. grade_basisT1T8c

Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: sleep_cappuccio (weight 1.0), sleep_yin (weight 0.55). reviews_counted

ci_low_exact: 21 Oct 2026 grading reform v2.0: outcome_match broader, measured outcome recorded (cards O1 and O5). estimate_source sleep_cappuccio, estimate_design cohort synthesis, measure RR, by rule E1 (cards G8, K5, P2): Grade Suggestive to Weak (card G3, PROPOSAL section 3.1): Lower bound 1.00; total stroke (confirm the unrounded bound). 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): Grade Weak to Suggestive (blocker 7): Suggestive: the source gives P = 0.047 for the stroke estimate, so the lower bound sits just above 1 before rounding (about 1.002) and rule T1 does not apply. One estimate for stroke of any type (related-outcome layer); short sleep is prone to reverse causation (T8c). The approved change to Weak (card G3) was conditional on the unrounded bound and is withdrawn. ci_low_exact 1.002 recorded, derived from the published two-sided P value of 0.047 (blocker 7). corrections.json _whyT1T8c

Outcome match, Outcome measured, Related outcome: The estimate measures: stroke (all types). The link sits in the related-outcome layer, outside the counts and the score. measured_outcome

Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: osa_dong (weight 1.0), osa_wang (weight 0.0), osa_xie (weight 0.44). reviews_counted

Outcome match, Outcome measured, Related outcome: The estimate measures: all stroke. The link sits in the related-outcome layer, outside the counts and the score. measured_outcome

Grade basis: Suggestive: publication bias is flagged and the prospective studies alone give 1.15 (review note). Kept by the verifier. grade_basis

Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: shs_fischer (weight 1.0), shs_lee_stroke (weight 0.17), bprf_shs (weight 0.62). reviews_counted

Outcome match, Outcome measured, Related outcome: The estimate measures: all strokes. The link sits in the related-outcome layer, outside the counts and the score. measured_outcome

Grade basis: Suggestive: the outcome is stroke of any type and classic risk factors explain part of the excess; genetic studies of education support an effect. Kept by the verifier. grade_basis

Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: sep_stroke_kerr (weight 1.0), sep_khaing (weight 0.55). reviews_counted

Trial and genetic evidence: 21 Oct 2026 grading reform v2.0: outcome_match broader, measured outcome recorded (cards O1 and O5). estimate_source sep_stroke_kerr, estimate_design cohort synthesis, measure HR, by rule E1 (cards G8, K5, P2). triangulation.mr supportive, same_exposure false (cards G6, G11): The genetic studies are of education, one part of socioeconomic position. grade_basis records why the grade stands (card G3) corrections.json _why

Outcome match, Outcome measured, Related outcome: The estimate measures: stroke (all types). The link sits in the related-outcome layer, outside the counts and the score. measured_outcome

Grade basis: Kept at Strong by the verified dry run: Trial evidence on total stroke is decisive and most strokes are ischaemic; this is an outcome-label problem. Set outcome_match broader and relabel instead of lowering the grade. grade_basis

Reviews counted: Counted from the cited syntheses under the review-counting rule: bp_ettehad (weight 1.0). reviews_counted

The estimate measures: ischaemic stroke. measured_outcome

Outcome match, Outcome measured, Related outcome: The estimate measures: total stroke. The link sits in the related-outcome layer, outside the counts and the score. measured_outcome

Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: preeclampsia_bellamy (weight 1.0), preeclampsia_wu (weight 0.67). reviews_counted

Outcome match, Outcome measured: The estimate measures: ischaemic stroke. measured_outcome

Grade basis: Strong: two syntheses of randomised trials (Bath and Gray 2005, 16 trials; the Cochrane review of 2015) and the WHI trial agree on more ischaemic stroke with hormone therapy (T3, T4). Verified at Strong on 21 October 2026. grade_basisT3T4

Outcome match, Outcome measured: The estimate measures: ischaemic stroke. measured_outcome

Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: ocp_urrutia (weight 1.0), coc_cochrane (weight 0.29). reviews_counted

Outcome match, Outcome measured: The estimate measures: heart failure incidence. measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: bp_hf_aune (weight 1.0). reviews_counted

Outcome match, Outcome measured: The estimate measures: heart failure. measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: dm_hf_aune2018 (weight 1.0), ohkuma_dm_hf (weight 0.17), dm_hf_cvd (weight 0.44). reviews_counted

Outcome match, Outcome measured: The estimate measures: heart failure incidence. measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: bmi_hf_aune (weight 1.0). reviews_counted

Sources: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5). estimate_source bmi_hf_aune, estimate_design cohort synthesis, measure RR, by rule E1 (cards G8, K5, P2). triangulation recorded as objects with their sources, same exposure and same outcome (card G11). adjudication recorded for each cited body that rules on causality, with pair_match (rules T3, T4, card G5) 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): triangulation.mr supportive with its sources recorded (larsson2020_bmi_cvd_mr) (card G11, blocker 4): Genetically predicted BMI raises heart failure in UK Biobank (Larsson 2020). gbd2021rf removed from the sources (blocker 6): the body does not rate this pair (adjudication pair_match false), so it cannot count as a replication synthesis either; its adjudication entry goes with it. corrections.json _whyT3T4

Outcome match, Outcome measured: The estimate measures: heart failure. measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: smk_hf_aune (weight 1.0). reviews_counted

Outcome match, Outcome measured: The estimate measures: heart failure. measured_outcome

Grade basis: Suggestive: the meta-analyses disagree, and abstainers carry raised risk (review note). Kept by the verifier. grade_basis

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: alc_hf_larsson (weight 1.0), alc_hf_suita (weight 0.55). reviews_counted

Outcome match, Outcome measured: The estimate measures: heart failure. measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: ckdpc_jama (weight 1.0). reviews_counted

Outcome match, Outcome measured: The estimate measures: heart failure. measured_outcome

Grade basis: Moderate: two reviews and a dose-response; only the reverse-causation watch list points lower. Kept by the verifier. grade_basis

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: pa_hf_pandey (weight 1.0), pa_hf_aune (weight 0.5). reviews_counted

Outcome match, Outcome measured: The estimate measures: incident heart failure (men). measured_outcome

Reviews counted, Rests on primary studies: No synthesis is cited, so N_eff is set to 1 and the link is flagged as resting on primary studies. reviews_counted

Outcome match, Outcome measured: The estimate measures: heart failure (hospital admission, incidence or death). measured_outcome

Grade basis: Weak: the long-term pooled estimate rests on 14 studies with I-squared 99.95% (review note). Accepted by the verifier. grade_basis

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: pm_hf_shah (weight 1.0), airpoll_hf_jia (weight 0.67). reviews_counted

Outcome match, Outcome measured: The estimate measures: atrial fibrillation. measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: bmi_af_aune (weight 1.0). reviews_counted

Sources: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5). estimate_source bmi_af_aune, estimate_design cohort synthesis, measure RR, by rule E1 (cards G8, K5, P2). triangulation recorded as objects with their sources, same exposure and same outcome (card G11) 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): triangulation.mr supportive with its sources recorded (larsson2020_bmi_cvd_mr) (card G11, blocker 4): Genetically predicted BMI raises atrial fibrillation in UK Biobank (Larsson 2020). corrections.json _why

Outcome match, Outcome measured: The estimate measures: atrial fibrillation. measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: af_umbrella (weight 1.0), bp_af_aune (weight 0.38). reviews_counted

Sources: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5). estimate_source bp_af_aune, estimate_design cohort synthesis, measure RR, by rule E1 (cards G8, K5, P2). triangulation recorded as objects with their sources, same exposure and same outcome (card G11). adjudication recorded for each cited body that rules on causality, with pair_match (rules T3, T4, card G5) 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): triangulation.mr supportive with its sources recorded (georgiopoulos2021_bp_af_mr, nazarzadeh2021_bp_af_mr) (card G11, blocker 4): Genetically predicted systolic blood pressure raises atrial fibrillation (OR 1.19 per 10 mmHg, Nazarzadeh 2021; Georgiopoulos 2021). corrections.json _whyT3T4

Outcome match, Outcome measured: The estimate measures: incident atrial fibrillation. measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: alc_af_larsson (weight 1.0), alc_af_gallagher (weight 0.38), alc_af_yang (weight 0.5). reviews_counted

Trial and genetic evidence, Sources: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5). estimate_source alc_af_gallagher, estimate_design cohort synthesis, measure HR, by rule E1 (cards G8, K5, P2). triangulation.rct supportive, same_outcome false (cards G6, G11): The abstinence trial measured recurrence of atrial fibrillation in people who already had it: The trial is not cited on the link. 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): triangulation.mr supportive with its sources recorded (larsson2020_alcohol_cvd_mr, yang2022_alcohol_af_mr, jiang2020_alcohol_af_mr) (card G11, blocker 4): Genetically predicted alcohol intake and atrial fibrillation: OR 1.17 (1.00 to 1.37) per SD of log drinks a week (Larsson 2020), at the edge of significance; an ALDH2 analysis in Korea found higher risk in men (Yang 2022); one two-sample analysis found no association (Jiang 2020). The dissent is noted. triangulation.rct supportive with its sources recorded (voskoboinik2020_af_abstinence), same_outcome false (card G11, blocker 4): The abstinence trial measured recurrence of atrial fibrillation in people who already had it (hazard ratio 0.55). corrections.json _why

Outcome match, Outcome measured: The estimate measures: atrial fibrillation. measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: alc_af_larsson (weight 1.0), alc_af_gallagher (weight 0.38). reviews_counted

Sources: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5). estimate_source alc_af_larsson, estimate_design cohort synthesis, measure RR, by rule E1 (cards G8, K5, P2). triangulation recorded as objects with their sources, same exposure and same outcome (card G11) 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): triangulation.mr supportive with its sources recorded (larsson2020_alcohol_cvd_mr, yang2022_alcohol_af_mr, jiang2020_alcohol_af_mr) (card G11, blocker 4): Genetically predicted alcohol intake and atrial fibrillation: OR 1.17 (1.00 to 1.37) per SD of log drinks a week (Larsson 2020), at the edge of significance; an ALDH2 analysis in Korea found higher risk in men (Yang 2022); one two-sample analysis found no association (Jiang 2020). The dissent is noted. corrections.json _why

Outcome match, Outcome measured: The estimate measures: atrial fibrillation. measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: osa_af_li (weight 1.0). reviews_counted

Outcome match, Outcome measured: The estimate measures: atrial fibrillation. measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: smk_af_aune (weight 1.0). reviews_counted

Outcome match, Outcome measured: The estimate measures: incident atrial fibrillation. measured_outcome

Reviews counted: Counted from the cited syntheses under the review-counting rule: crf_af (weight 1.0). reviews_counted

Outcome match, Outcome measured: The estimate measures: atrial fibrillation. measured_outcome

Grade: Suggestive: Bias-adjusted estimates fall to 1.24 to 1.34. Rules T9b, T10 (ceiling Suggestive). Verified on 21 October 2026 (card G3). Curator record: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5). estimate_source dm_af_huxley, estimate_design cohort synthesis, measure RR, by rule E1 (cards G8, K5, P2): Grade Moderate to Suggestive (card G3, PROPOSAL section 3.1): Bias-adjusted estimates fall to 1.24 to 1.34. grade_basisT9bT10

Grade basis: Suggestive: Bias-adjusted estimates fall to 1.24 to 1.34. Rules T9b, T10 (ceiling Suggestive). Verified on 21 October 2026 (card G3). grade_basisT9bT10

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: af_umbrella (weight 1.0), dm_af_huxley (weight 0.0). reviews_counted

Outcome match, Outcome measured: The estimate measures: peripheral artery disease. measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: pad_fowkes (weight 1.0), lu2014_smoking_pad (weight 0.17), bprf_smoking (weight 0.62), xu2024_smoking_pad_sex (weight 0.29). reviews_counted

Sources: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5): Sources added (card K1): xu2024_smoking_pad_sex, lu2014_smoking_pad; the displayed estimate is unchanged. estimate_source bprf_smoking, estimate_design cohort synthesis, measure RR, by rule E1 (cards G8, K5, P2). triangulation recorded as objects with their sources, same exposure and same outcome (card G11). adjudication recorded for each cited body that rules on causality, with pair_match (rules T3, T4, card G5). external_grades transcribed from the caveat, the review note or the quote (rule T11). corrections.json _whyT3T4T11

Outcome match, Outcome measured: The estimate measures: prevalent PAD defined by low ankle-brachial index in surveys. measured_outcome

Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: pad_fowkes (weight 1.0), pad_global2019 (weight 0.55), dm_pad_peters (weight 0.17). reviews_counted

Outcome match, Outcome measured: The estimate measures: peripheral artery disease (genetic estimate). measured_outcome

Grade override: Curator override to Moderate: Kept by the verifier (card G6): the trials compared intensive with standard lowering in people at high risk, which is another exposure contrast from lifelong higher pressure. Mendelian randomisation (1.20 per 10 mmHg) and a cohort of 4.2 million adults agree on harm. C3 proposes Suggestive. grade_overrideC3

Reviews counted, Rests on primary studies: No synthesis is cited, so N_eff is set to 1 and the link is flagged as resting on primary studies. reviews_counted

Trial and genetic evidence: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5). estimate_source bp_pad_mr_levin, estimate_design MR, measure OR, by rule E1 (read by hand) (cards G8, K5, P2). triangulation.rct contradicts, same_exposure false (cards G6, G11): The trials compared intensive with standard blood pressure lowering in people at high cardiometabolic risk, which is another exposure contrast from lifelong higher pressure. grade_override keeps Moderate: Kept by the verifier (card G6): the trials compared intensive with standard lowering in people at high risk, which is another exposure contrast from lifelong higher pressure: Mendelian randomisation (1.20 per 10 mmHg) and a cohort of 4.2 million adults agree on harm. C3 proposes Suggestive. corrections.json _whyC3

Outcome match, Outcome measured: The estimate measures: peripheral artery disease (genetic estimate). measured_outcome

Grade basis: Strong: Mendelian randomisation per 1 mmol/L LDL (the displayed estimate) and the Heart Protection Study, where statin treatment cut peripheral vascular events, agree (T3, T4). grade_basisT3T4

Reviews counted, Rests on primary studies: No synthesis is cited, so N_eff is set to 1 and the link is flagged as resting on primary studies. reviews_counted

Sources: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5). estimate_source ldl_pad_mr, estimate_design MR, measure RR, by rule E1 (read by hand) (cards G8, K5, P2). triangulation recorded as objects with their sources, same exposure and same outcome (card G11). adjudication recorded for each cited body that rules on causality, with pair_match (rules T3, T4, card G5) 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): triangulation.rct supportive with its sources recorded (hps2007_pad) (card G11, blocker 3, blocker 4): In the Heart Protection Study, simvastatin (LDL 1.0 mmol/L lower) cut first peripheral vascular events by 16% (5 to 25) among 20,536 high-risk people, a third of whom had peripheral artery disease at entry. The CTT analyses report no peripheral artery disease outcome. ctt removed from the sources (blocker 3): the CTT analysis reports no peripheral artery disease, so it is neither an adjudicator nor a trial line for this pair; its adjudication entry goes with it. Grade stays Strong (blocker 3); grade_basis rewritten: Strong: Mendelian randomisation per 1 mmol/L LDL (the displayed estimate) and the Heart Protection Study, where statin treatment cut peripheral vascular events, agree (T3, T4). corrections.json _whyT3T4

Outcome match, Outcome measured: The estimate measures: incident peripheral artery disease. measured_outcome

Reviews counted: Counted from the cited syntheses under the review-counting rule: ckdpc_pad (weight 1.0). reviews_counted

Outcome match, Outcome measured: The estimate measures: abdominal aortic aneurysm. measured_outcome

Grade basis: Moderate, the v1.4 grade. With the Burden of Proof rating attached (five stars for aortic aneurysm, card K1) the decision table allows Strong; card G10 rules out an automatic upgrade, so the later curation pass decides. grade_basis

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: smk_aaa (weight 1.0), bprf_smoking (weight 0.44), zhang2025_aaa_risk (weight 0.38). reviews_counted

Sources: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5): Sources added (card K1): bprf_smoking, zhang2025_aaa_risk; the displayed estimate is unchanged. estimate_source smk_aaa, estimate_design cohort synthesis, measure RR, by rule E1 (cards G8, K5, P2). adjudication recorded for each cited body that rules on causality, with pair_match (rules T3, T4, card G5). external_grades transcribed from the caveat, the review note or the quote (rule T11). Grade recorded as Moderate, the v1.4 grade (card G10: no automatic upgrade). corrections.json _whyT3T4T11

Outcome match, Outcome measured: The estimate measures: abdominal aortic aneurysm. measured_outcome

Reviews counted: Counted from the cited syntheses under the review-counting rule: bp_aaa (weight 1.0). reviews_counted

Outcome match, Outcome measured: The estimate measures: incident abdominal aortic aneurysm. measured_outcome

Grade override: Curator override to Suggestive: Kept by the verifier: smoking confounding and a modest effect per unit. T9a proposes Weak because the quote names three studies for thoracic aortic aneurysm; the abdominal estimate pools 12 studies. grade_overrideT9a

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: adip_aaa (weight 1.0). reviews_counted

Outcome match, Outcome measured: The estimate measures: incident type 2 diabetes. measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: bmi_t2d_abdullah (weight 1.0), adip_t2d_bmj (weight 0.71). reviews_counted

Sources: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5). estimate_source bmi_t2d_abdullah, estimate_design cohort synthesis, measure RR, by rule E1 (cards G8, K5, P2). triangulation recorded as objects with their sources, same exposure and same outcome (card G11) 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): triangulation.mr supportive with its sources recorded (yuan2020_t2d_atlas_mr, dale2017_adiposity_mr) (card G11, blocker 4): Genetically predicted adult BMI raises type 2 diabetes (Dale 2017, OR 1.98 per SD; Yuan 2020). corrections.json _why

Outcome match, Outcome measured: The estimate measures: incident type 2 diabetes. measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: adip_t2d_bmj (weight 1.0). reviews_counted

Sources: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5). estimate_source adip_t2d_bmj, estimate_design cohort synthesis, measure RR, by rule E1 (cards G8, K5, P2). triangulation recorded as objects with their sources, same exposure and same outcome (card G11) 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): triangulation.mr supportive with its sources recorded (emdin2017_whr_mr, dale2017_adiposity_mr) (card G11, blocker 4): Genetically predicted waist-to-hip ratio raises type 2 diabetes (Emdin 2017; Dale 2017, OR 1.82 per SD). corrections.json _why

Outcome match, Outcome measured: The estimate measures: incident type 2 diabetes. measured_outcome

Grade basis: Strong: a dose-response synthesis of 81 studies (Aune 2015) and Mendelian randomisation of physical activity (Yuan 2023) agree (T3, T4); the DPP trial tested a combined lifestyle programme (another exposure contrast, noted). Verified at Strong on 21 October 2026. grade_basisT3T4

Trial and genetic evidence, Sources: 21 Oct 2026 grading reform v2.0: outcome_match exact, measured outcome recorded (cards O1 and O5). estimate_source pa_t2d_aune, estimate_design cohort synthesis, measure RR, by rule E1 (cards G8, K5, P2). triangulation.rct supportive, same_exposure false (cards G6, G11): DPP tested a lifestyle programme that combined diet, exercise and weight loss (review note). grade_override keeps Strong: The cited DPP trial combined diet, exercise and weight loss, so it is recorded with same_exposure false and rule T3 (Moderate) fails as written: no trial, Mendelian randomisation study or adjudicating body corroborates physical activity alone: New disagreement for the owner (card G3 A): kept at the verified grade until the owner decides. 23 Oct 2026 fix round (verifier of the data step, blockers 1 to 10): triangulation.mr supportive (status was none) with its sources recorded (yuan2023_pa_t2d_mr, meisinger2020_pa_t2d_mr) (card G11, G2, blocker 4): Genetically predicted moderate-to-vigorous activity lowers type 2 diabetes (OR 0.70 per unit log-odds, Yuan 2023, 600,000 people for the instruments); an earlier analysis with the few accelerometer-based variants found no association (Meisinger 2020). grade_override removed (blocker 2): the recorded genetic evidence meets rule T3 and gives the second line for T4, so Strong follows from the table. Grade stays Strong (blocker 2); grade_basis rewritten: Strong: a dose-response synthesis of 81 studies (Aune 2015) and Mendelian randomisation of physical activity (Yuan 2023) agree (T3, T4); the DPP trial tested a combined lifestyle programme (another exposure contrast, noted). Verified at Strong on 21 October 2026. corrections.json _whyT3T4

Outcome match, Outcome measured: The estimate measures: incident type 2 diabetes. measured_outcome

Grade basis: Moderate: television time per hour, adjusted for physical activity; genetic studies support an effect of television viewing (review note). Kept by the verifier. grade_basis

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: sed_t2d_patterson (weight 1.0), sed_t2d_bailey (weight 0.17). reviews_counted

Outcome match, Outcome measured: The estimate measures: incident type 2 diabetes. measured_outcome

Reviews counted, Effective reviews (N_eff): Counted from the cited syntheses under the review-counting rule: diet_t2d_umbrella (weight 1.0), asb_t2d_imamura (weight 0.0). reviews_counted