Study name Country Reference Follow‐up Study design Funding	Original Cohort (N total) Exclusion criteria Study population (n, sex and age at baseline(a))	Ascertainment of outcome	Exposure groups(a) n/person‐years	N deaths	Model covariates	Results
MPP Sweden Schomburg et al. (2019) 9.3 yr (median) Prospective cohort Mixed	N = 33,346 Population sampled: General population Excluded: CVD event prior to baseline n = 4,366 Sex (% F): 31 Ethnicity: Caucasian Age (yr): 70	Death ascertained through record linkage with the Swedish Hospital Discharge Register, the Swedish Cause of Death Register, the Stroke in Malmö Register, and the Swedish Coronary Angiography and Angioplasty Registry.	N per quintile of plasma Sepp1 (mg/L, median (range)) Q1 3.7 (0.4–4.3): 873 Q2: 4.7 (4.3–5.1): 873 Q3: 5.5 (5.1–5.9): 874 Q4: 6.3 (5.9–6.9): 873 Q5: 7.7 (6.9–20): 873	N, per quintile Q1(ref): 314 Q2: 214 Q3: 193 Q4: 175 Q5: 215	Age, gender, current smoking, systolic blood pressure, use of antihypertensive medication, diabetes mellitus, LDL‐cholesterol, HDL‐cholesterol, and body mass index	HR (95% CI) Q1(ref): 1 Q2: 0.73 (0.61, 0.87) Q3: 0.66 (0.55, 0.79) Q4: 0.57 (0.48, 0.69) Q5: 0.69 (0.58, 0.82)
ilSIRENTE Italy Giovannini et al. (2018) 10 yr Prospective cohort Public	N = 364 Population sampled: General population Excluded: insufficient blood sample n = 347 Sex (% F): 67 Ethnicity: Caucasian Age (yr): 85	Survival status was obtained from general practitioners and confirmed by the National Death Registry.	N above/below median serum Se (μg/L) Low Se ≤ 105.3: 175 High Se > 105.3: 172	N, per group Low Se: 135 High Se: 113	Model 1: crude Model 2: age, gender, IADL scale score, BMI, Cancer Model 3: age, gender, IADL scale score, BMI, Cancer, HDL‐cholesterol, IL‐6, CRP	HR (95% CI), high Se vs low Se Model 1: 0.66 (0.51, 0.85) Model 2: 0.69 (0.53, 0.89) Model 3: 0.71 (0.54, 0.92)
Rivalta + Reggio Emilia Italy Vinceti et al. (2016a) 27 yr Prospective cohort Mixed	N = 97,780 Population sampled: subjects continuously residing in Rivalta from 1974 to 1985, exposed to high‐Se‐contaminated tap water (n = 2,065) and unexposed municipal population as controls (n = 95,715) n = 97,780 Sex (% F): Exposed: 51 Unexposed: 53 Ethnicity: Caucasian Age (yr): 5–95+	Death ascertained using Reggio Emilia mortality register	person‐years, per group of Se in tap water (inorganic Se) (μg/L) Exposed 8–10: 46,268 Unexposed < 1: 2,067,862	N, per group Exposed: 663 Unexposed: 34,598	age and calendar year	HR (95% CI), exposed vs unexposed 0.99 (0.91, 1.05)
SWHS/SMHS China Sun et al. (2016) 13.9 yr (F)/8.37 y r (M) (mean) Prospective cohort Public	N = 136,421 Population Sampled: General population, without history of cancer, resident in Shanghai Excluded: immediately lost to follow‐up after study enrolment, missing data, extreme energy intake n = 133,957 (F 73,854/M 60,103) Ethnicity: Asian Age (yr): 40–70	Annual record linkage with the Shanghai Vital Statistics Registry. All possible matches identified through the linkage were verified by home visits	Energy‐adjusted quintile of Se intake (b) (μg/1,000 kcal per d, median) assessed through SFFQ F/M Q1 16: 14,771/12,021 Q2 21: 14,771/12,022 Q3 25: 14,769/12,022 Q4 30: 14,771/12,016 Q5 42: 14,772/12,022	N, per quintile F/M Q1(ref): 1,863/1,136 Q2: 1,285/885 Q3: 1,033/819 Q4: 892/745 Q5: 676/632	age, birth cohort, education, income, marital status, occupation, BMI, physical activity, total energy intake, dietary fat intake, supplement use, smoking status, drinking status, history of hypertension, diabetes, CHD or stroke and family history of cancer	HR (95% CI) Females Q1(ref): 1 Q2: 0.96 (0.89, 1.03) Q3: 0.92 (0.84, 1.00) Q4: 0.90 (0.82, 0.99) Q5: 0.79 (0.71, 0.88) Males Q1(ref): 1 Q2: 0.91 (0.83, 0.99) Q3: 0.86 (0.78, 0.95) Q4: 0.82 (0.73, 0.91) Q5: 0.79 (0.70, 0.89)
PREDIMED Spain Henríquez‐Sánchez et al. (2016) 4.3 yr (median) Prospective cohort Public	N = 7,447 Population sampled: Participants of the PREDIMED RCT, without CVD at enrolment and with either T2DM or ≥3 risk factors (smoking, hypertension, dyslipidemia, overweight or obesity, or family history of premature CVD) Excluded: no follow‐up; total energy intake < 800 or > 4,000 kcal/d n = 7,015 Sex (% F): 58 Ethnicity: Caucasian Age (yr): 55–80	Survival status obtained from the continuous contact with participants and their families (during trial phase), contact with family physicians, yearly comprehensive review of all medical records and yearly consultation of the National Death Index	Quintile of Se intake assessed through SFFQ NR Person‐years Q1: 6,070 Q2: 6,190 Q3: 6,210 Q4: 6,333 Q5: 6,273	N, per quintile Q1(ref): 73 Q2: 73 Q3: 62 Q4: 53 Q5: 58	Recruitment center, intervention group, age, sex, education, marital status, BMI, smoking habit, alcohol consumption, total energy intake, energy‐adjusted intake of saturated fatty acids, polyunsaturated fatty acids, monounsaturated fatty acids and glycaemic index and medical history of hypertension, diabetes, dyslipidemia and cancer	HR (95% CI) Q1(ref): 1 Q2: 0.75 (0.53, 1.05) Q3: 0.88 (0.62, 1.25) Q4: 0.72 (0.49, 1.06) Q5: 0.74 (0.49, 1.10) p trend 0.212
SES Sweden Alehagen et al. (2016) 6.8 yr Prospective cohort Mixed	N = 449 Population sampled: General population from a rural municipality n = 449 Sex (% F): 52 Ethnicity: Caucasian Age (yr): 70–80	The mortality information was obtained from the National Board of Health and Welfare in Sweden, which registers all deaths of Swedish citizens.	N, per quartile of serum Se (μg/L) Q1 < 57.2: 107 Q2 57.2–67.1: NR Q3 67.1–76.1: NR Q4 > 76.1: 111	N, per quartile Q1: 41 Q2‐4: 81	Model 1: crude Model 2: male gender, smoking, ischemic heart disease, diabetes, chronic obstructive pulmonary disease and ejection fraction < 40% (echocardiography)	HR (95% CI), Q1 vs Q2‐4 (ref) Model 1: 1.67 (1.15, 2.44) Model 2: 1.43 (1.02, 2.00)
NHANES III (1988–1994) USA Eaton et al. (2010) 13.4 yr Prospective cohort Public	N = 10,531 Population Sampled: General population, ≥ 35 yr Excluded: missing serum Se and serum creatinine; renal insufficiency n = 9,304 Sex (% F): Low Se: 64 Normal Se: 52 Ethnicity: 78% non‐Hispanic, white Age (yr): Low Se: 56.4 ± 1.3 Normal Se: 52.1 ± 0.4	Survival status obtained through the National Death Index	N above/below median serum Se (μg/L) Low Se (≤ 98 ng/mL): 418 Normal Se (> 98 ng/mL): 8,886	Age‐standardised death rate per 100,000 person‐years Low Se (≤ 98 ng/mL): 3,509 Normal Se (> 98 ng/mL): 2,305	Model 1: age‐adjusted Model 2: age, gender, race, geography, smoking, BMI, systolic BP, diabetes, total‐to‐HDL cholesterol ratio, physical activity, lifetime alcohol use, history of CHD or stroke, and family history of CHD	HR (95% CI), low vs normal Se Model 1: 1.67 (1.4, 1.91) Model 2: 1.65 (1.41, 1.93)
NHANES (1996–2006) USA Li et al. (2020b) 10.2 yr (median) Prospective cohort Public	N = 41,474 Population sampled: General population Excluded: missing Se data, aged < 18 yr n = 2,903 Sex (% F): 51 Ethnicity: 56% White, 44% Non‐White Age (yr): 61.94 ± 13.73	Mortality data extracted from the 1999–2006 NHANES public‐use linked mortality files	n, per quartile of serum Se (μg/L) Q1 ≤ 124.0: 696 Q2 125.0–135.0: 746 Q3 136.0–147.0: 721 Q4 ≥ 148.0: 740	N, per quartile Q1: 244 Q2: 208 Q3: 197 Q4: 209	Model 1: age, gender and BMI Model 2: Model 1 + systolic blood pressure, total cholesterol, C‐reactive protein, alcohol consumption, smoking, race, history of hypertension, cardiovascular disease, diabetes, estimated glomerular filtration rate, use of lipid‐lowering and antiplatelet drugs.	HR (95% CI) Model 2 Q1(ref): 1 Q2: 0.69 (0.57, 0.84) Q3: 0.63 (0.52, 0.77) Q4: 0.64 (0.53, 0.77) p for trend < 0.01 Model 3 Q1(ref): 1 Q2: 0.62 (0.47, 0.81) Q3: 0.57 (0.42, 0.75) Q4: 0.60 (0.45, 0.78) p for trend < 0.01
NDNS UK Bates et al. (2011) 14 yr Prospective cohort Public	N = 1,054 Population Sampled: General population aged ≥ 65 years; community living participants Excluded: missing plasma Se; lost to follow up n = 826 Sex (% F): 48 Ethnicity: Caucasians Age (yr): 76.6 ± 7.4	Death registries. All participants agreed to be flagged to the National Register of Births and Deaths, and whose status (i.e. as still alive or registered as having died) was known in September 2008.	Plasma Se (μmol/L) Males 950 ± 218 (375–2,376) Females 924 ± 211 (461–1,786)	NR	Model 1: Age and sex Model 2: Model 1 + other significantly predictive nutrient variables Model 3: Model 2+ inclusion of a1‐antichymotrypsin, plasma creatinine, total and HDL‐cholesterol and albumin concentrations, BMI, systolic blood pressure, current smoking index, number of prescribed drugs being taken, self‐reported health score, physical activity score and index of poverty	HR (95% CI), per SD plasma Se Model 1: 0.76 (0.69, 0.84) Model 2: 0.82 (0.73, 0.91) Model 3: 0.83 (0.73, 0.94)
InCHIANTI Italy Lauretani et al. (2008) 6 yr Prospective cohort Public	N = 1,155 Population sampled: General population, aged ≥ 65 yr Excluded: missing blood sample or incomplete data n = 1,042 Sex (% F): 57 Ethnicity: Caucasians Age (yr): 75.6 ± 7.4	Mortality data collected from the General Mortality Registry of the Tuscany Region and death certificates deposited at the municipality of residence.	Quartiles of plasma Se (μmol/L) Q1: < 0.839 Q2: 0.839–0.934 Q3: 0.935–1.037 Q4: > 1.037 n per quartile NR	N, per quartile Q1: 98 Q2: 64 Q3: 43 Q4: 32	Model 1: age and sex Model 2: age, sex, education, BMI, congestive heart failure, peripheral artery disease, stroke, Parkinson's disease and chronic obstructive pulmonary disease	HR (95% CI) Model 1 Q1: 1.82 (1.20, 2.75) Q2: 1.48 (0.96, 2.29) Q3: 1.14 (0.72, 1.81) Q4 (ref): 1 Model 2 Q1: 1.62 (1.06, 2.43) Q2: 1.42 (0.94, 2.18) Q3: 1.09 (0.68, 1.74) Q4 (ref): 1
WHAS I and II USA Ray et al. (2006) 5 yr Prospective cohort Public	N WHAS I = 1,002 N WHAS II = 436 Population sampled: Community living women with and without physical disabilities Excluded: missing serum Se, aged < 70 or > 79 yr n = 632 Sex: females Ethnicity: NR Age (yr): Died 75 Lived 73.9	Vital status determined through follow‐up interviews with proxies, obituaries, and matching with the National Death Index	Serum Se (μmol/L geometric mean (95% CI)) 1.52 Died: 1.43 (1.38–1.49) Lived: 1.54 (1.52–1.57)	Total = 89	Model 1: age and education Model 2: age, education, current smoking, alcohol use, BMI, fair‐to‐poor appetite, diabetes mellitus, cardiovascular disease, and renal disease	HR (95% CI), per 1 SD of Log e Se Model 1: 0.68 (0.55, 0.83) Model 2: 0.71 (0.56, 0.9)
Asturias Spain González et al. (2007) 6 yr (mean 4.3 yr) Prospective cohort Mixed	N = 304 Population sampled: Elderly living in institutions Excluded: previous history of cancer, mental impairment or CVD, in a wheelchair, with terminal disease, missing biochemical parameters, use of vitamin and/or mineral supplements or following a special diet n = 215 Sex (% F): 59 Ethnicity: Caucasians Age: ≥ 60 y	Vital status obtained from the institutions the participants were residing at every year	N, per quintile of serum Se (μmol/L, range) Q1 ≤ 0.92: 47 Q2 0.93–1.04: 37 Q3 1.05–1.14: 45 Q4 1.15–1.25: 43 Q5 > 1.26: 43	N, per quintile Q1: 17 Q2: 14 Q3: 10 Q4: 11 Q5: 8	Age, sex, smoking habit, BMI and cognitive score	RR (95% CI) Q1(ref): 1 Q2: 1.13 (0.54, 2.39) Q3: 0.69 (0.29, 1.60) Q4: 0.81 (0.36, 1.82) Q5: 0.56 (0.23, 1.40) p for trend = 0.158
EVA France Akbaraly et al. (2005) 9 yr Prospective cohort Mixed	N = 1,389 Population sampled: General population, residents in Nantes, aged 59–71 y Excluded: NA n = 1,389 Sex (% F): 59 Ethnicity: Caucasians Age (yr): Died 65.4 ± 3.0 Lived 65.0 ± 3.0	Vital status collected from town hall civil registries	N, per quartile of plasma Se (μmol/L, median (range)) Q1: 0.87 (0.18–0.95): 337 Q2: 1.03 (0.96–1.09): 350 Q3: 1.15 (1.10–1.21): 286 Q4: 1.32 (1.22–1.97): 361	Total = 101 N per quartile NR	Age, sex, smoking, alcohol consumption, medication, obesity, diabetes, dyslipidaemia and vascular disease	RR (95% CI) Q1: 3.34 (1.71, 6.53) Q2: 2.49 (1.25, 4.94) Q3: 1.67 (0.78, 3.56) Q4 (ref): 1
General Population Trial China Wei et al. (2004) 15 yr Prospective cohort Public	N = 29,584 Population sampled: healthy individuals aged 40–69 yr from 4 Linxian communes, participants of the General Population Trial Excluded: NR n = 1,103 (sub‐sample from the trial) Sex (% F): 45 Ethnicity: Asian Age (yr): 56.6 ± 8.0	Village doctors ascertained mortality among participants through monthly follow‐up	Quartiles of serum Se (μmol/L) Q1: ≤ 0.77 Q2: > 0.77–0.91 Q3: > 0.91–1.06 Q4: > 1.06 n per quartile of Se NR	Total = 516 N per quartile of Se NR	Sex, age, cholesterol, smoking, drinking, and BMI Assignment to treatment during trial period not included as did not affect estimates	RR (95% CI) Q1(ref): 1 Q2: 1.01 (0.79, 1.30) Q3: 0.96 (0.75, 1.23) Q4: 0.93 (0.72, 1.19) p for trend = 0.57
Uppsala Sweden Kilander et al. (2001) 25.7 yr Prospective cohort Public	N = 2,322 Population sampled: men aged 50 yr resident in Uppsala Excluded: died within 2 yr of baseline n = 2,285 Sex: M Ethnicity: Caucasians Age (yr): 48.6–51.1 y	Ascertainment of deaths from the Swedish death register	Serum Se (μg/L), by education level Low: 73.3 Medium: 77.7 High: 83.0	Total = 630	age	RR (95% CI), per 1 SD of serum Se increase 0.87 (0.80, 0.95)
Finnish Rural Finland Virtamo et al. (1985) 5 yr Prospective cohort Public	N = 1,710 Population sampled: men from rural areas of eastern and western Finland Excluded: serum Se ≥ 45 μg/L n = 328 East 84 West 244 Sex: males Ethnicity: Caucasians Age: 55–74 y	Data on deaths obtained from the National Death Certificate Register	Serum Se < 45 μg/L	NR	Model 1: age and area Model 2: Model 1 + smoking, serum cholesterol, diastolic blood pressure, previous CHD, blood haemoglobin, FEV0.7 S, alcohol consumption	RR (95% CI) Model 1 1.5 (1.1–2.0) Model 2 1.4 (1.0–2.0) Among men initially free of CHD Model 1: 1.4 (0.9, 2.4) Model 2: 1.5 (0.9, 2.4)

BMI: body mass index; BP: blood pressure; CI: confidence interval; CRP: c‐reactive protein; CVD: cardiovascular disease; CHD: coronary heart disease; EVA: Etude du Vieillissement Arte'riel study; F: females; FEV: forced expiratory volume; HDL: high density lipoprotein; HR: hazard ratio; IADL: Instrumental Activities of Daily Living; ilSIRENTE: Invecchiamento e Longevità nel Sirente; InCHIANTI: Invecchiare in Chianti; LDL: low density lipoprotein; M: males; MPP: The Malmö Preventive Project; NDNS: National Diet and Nutrition Survey; NHANES: National Health and Nutrition Examination Survey; NR: not reported; PREDIMED: Prevención con Dieta Mediterránea (Prevention with Mediterranean Diet); PY: person‐years; Qx: quartile/quintile x; RCT: randomised controlled trial; RR: relative risk; SD: standard deviation; Se: selenium; SES: South‐East Sweden cohort; SFFQ: semi‐quantitative food frequency questionnaire; SMHS: Shanghai Men's Health Study; SWHS: Shanghai Women's Health Study; UK: United Kingdom; USA: United States of America; WHAS: Women's Health and Aging Study; yr: year.

(a) Mean ± SD (range), unless specified otherwise.

(b): Dietary intakes were adjusted for energy intake using the nutrient density method and expressed as per μg/4184 kJ (1,000 kcal) per day (Willett al., 1997).