Neal 2021.

Study characteristics
Methods	Study design: Cluster‐randomised controlled trial Country: China Setting: Intervention conducted in villages from five provinces Hebei, Liaoning, Ningxia, Shanxi and Shaanxi in northern region. NR where outcomes were measured Comments: SSaSS: Salt Substitute and Stroke Study Aim of study: The primary objective of the SSaSS is to determine the effects of sodium reduction achieved through the use of a reduced‐sodium, added potassium salt substitute on the risk of fatal or non‐fatal stroke. The secondary objectives are to determine the effects of the intervention on (1) major vascular events, a composite of non‐fatal stroke, non‐fatal acute coronary syndrome, and vascular death, and (2) total mortality. Unit of allocation: Villages Start date: April 2014 End date: March 2021 Relevant study limitations as reported by study authors: Neal 2021: No serial assessments of blood potassium, therefore, hyperkalaemia may have been missed in some participants. Dose‐response analysis could not be conducted since only one type of LSSS was used. No concealment of intervention delivery; however, objective primary and secondary outcomes were reported. In addition, systematic searches of routinely collected health data and verification by an independent end‐point adjudication committee, the members of which were unaware of the trial group assignments, was conducted, but authors stated that "Information that could be used for adjudication of outcome events was limited, and definitive assignment of causation was difficult in many cases.” No evidence was found to suggest certainty of adjudications had any substantive implications for the primary conclusions of the trial. Huang 2020: Hyperkalaemia risk in the study did not include vulnerable populations (e.g. CKD). Increased uncertainty of the effect estimates for intermediate outcomes due to the cluster‐randomised trial design (compared to an individual‐randomised trial). Incomplete participation of invited participants and incomplete collection of 24‐hour urine collections in interim surveys (however: those that participated were broadly representative of overall study population); and absence of 24‐hour urine for two provinces at baseline Sample size calculation: Estimated sample size of 21,000 participants in 600 clusters (n = 300 in the LSSS intervention group; n = 300 in control group, with 35 participants in each cluster) to detect a >= 13% relative risk reduction for stroke in the intervention villages compared with control villages, with 90% statistical power and 95% confidence level (two‐sided). The estimate assumed a primary outcome event rate of 3.5% per year, a systolic blood pressure difference of 3.0 mmHg between randomised groups, and an ICC of 0.04.
Participants	Baseline Characteristics LSSS intervention Age: in years, mean (SD): 65.2 (8.5) Gender: Female, % (n/N): 49.7 (5221/10 505) Ethnicity/race: Chinese Smoking: Current smoking, % (n/N): 18.7 (1964/10505); past smoking, % (n/N): 32.8 (3446/10505) Body Mass Index (BMI): in kg per m2, mean (SD): 24.8 (3.6) Blood pressure status: Uncontrolled hypertension, % (n/N): 59.4 (6240/10505) Antihypertensive medication used: Any antihypertensive medication, % (n/N): 79.9 (8394/10 505); diuretic, % (n/N): 11.5 (1208/10 505); ACE inhibitor or ARB, % (n/N): 23.1 (2427/10 505); alpha‐blocker, % (n/N): 0.7 (95/10 505); beta‐blocker, % (n/N): 5 (525/10 505); calcium antagonist, % (n/N): 43.3 (4549/10 505) Cardiovascular disease or stroke: History of stroke, % (n/N): 73.3 (7700/10 505); transient ischaemic attack, % (n/N): 13.2. (1387/10 505); ischaemic heart disease, % (n/N): 16.4 (1723/10 505) Diabetes mellitus: Type 2 DM, % (n/N): 10.6 (1114/10505) Renal impairment: Known serious kidney disease, % (n/N): 0 (0/10505) Dietary potassium intake: NR Dietary sodium intake: NR Urinary potassium excretion: in (g/d), mean (SD): 1.4 (0.6) Urinary sodium excretion: in (g/d), mean (SD): 4.4 (1.8) Control Age: in years, mean (SD): 65.5 (8.5) Gender: Female, % (n/N): 49.2 (5162/10491) Ethnicity/race: Chinese Smoking: Current smoking, % (n/N): 18.9 (1983/10491); past smoking, % (n/N): 34.2 (3588/10491) Body Mass Index (BMI): in kg per m2, mean (SD): 24.9 (3.7) Blood pressure status: Uncontrolled hypertension, % (n/N): 59.2 (6211/10,491) Antihypertensive medication used: Any antihypertensive medication, % (n/N): 78.7 (8256/10 491); diuretic, % (n/N): 11.2 (1175/10 491); ACE inhibitor or ARB, % (n/N):23 (2413/10 491); alpha‐blocker, % (n/N): 1.1 (115/10 491); beta‐blocker, % (n/N): 6.3 (661/10 491); calcium antagonist, % (n/N): 40.6 (4259/10 491) Cardiovascular disease or stroke: History of stroke, % (n/N): 72.0 (7554/10 491); transient ischaemic attack, % (n/N): 14.9 (1563/10 491); ischaemic heart disease, % (n/N): 15.7 (1647/10 491) Diabetes mellitus: Type 2 DM, % (n/N): 10.5 (1102/10491) Renal impairment: Known serious kidney disease, % (n/N): 0 (0/10491) Dietary potassium intake: NR Dietary sodium intake: NR Urinary potassium excretion: in (g/d), mean (SD): 1.4 (0.6) Urinary sodium excretion: in (g/d), mean (SD): 4.2 (1.8) Overall Age: NR Gender: NR Ethnicity/race: NR Smoking: NR Body Mass Index (BMI): NR Blood pressure status: NR Antihypertensive medication used: NR Cardiovascular disease or stroke: NR Diabetes mellitus: NR Renal impairment: NR Dietary potassium intake: NR Dietary sodium intake: NR Urinary potassium excretion: NR Urinary sodium excretion: NR Inclusion criteria: Individuals in the selected villages with an elevated risk of stroke, defined as a history of stroke (regardless of type or knowledge of aetiology) and/or age 60 years or greater with uncontrolled high blood pressure (systolic blood pressure >= 140 mm Hg at visit if on blood pressure–lowering medication or systolic blood pressure >= 160 mm Hg if not on blood pressure–lowering medication). Participants were also required to be contactable by telephone and be able to nominate a friend or relative through which they could also be contacted, if necessary. Exclusion criteria: Participant or family members were excluded if they: (1) or someone living in their household had a potential contraindication to the salt substitute used in the trial; contraindications included use of a potassium‐sparing diuretic, use of a potassium supplement, or known serious kidney disease; (2) had serious renal impairment; (3) were taking potassium‐sparing diuretic; (4) were taking potassium supplement;(5) had concerns about using salt substitute; (6) ate most meals outside home; (7) were expected to live less than 6 months from date of baseline assessment by village doctors Pretreatment: No group differences Method of recruitment of participants: Two counties per province were selected by the research team based upon their willingness to participate, their proximity to the local research team and being representative of the socioeconomic level of each province. Sixty villages in each of the counties were recruited through a consent process involving the leadership of the local county Bureaus of Health (identified villages that were willing and accessible). About 35 individuals at elevated risk of stroke were recruited in each village. Informed consent obtained: Yes (written) Clusters: n = 600; cluster size: n = 35 individuals Subgroups planned/measured: NR Subgroups reported: Age > 65 years vs. age <= 65 years; female vs. male; less educated vs. more educated; cerebrovascular disease vs. no cerebrovascular disease; diabetes vs. no diabetes; hypertensive vs. non‐hypertensive; using antihypertensive medication vs. not using antihypertensive medication; SBP > 153 mmHg vs. SBP <= 153 mmHg; DBP > 89 mmHg vs. <= 89 mmHg and BMI > 24.6 kg/m2 vs. BMI <= 24.6 kg/m2   Participant flow Assessed for eligibility: NR Excluded (number with reasons): n = 1 individual (death) Randomised: n = 600 villages (n = 20,995 individuals) Allocated to LSSS intervention(s): n = 300 villages (n = 10,504 individuals) Allocated to control: n = 300 villages (n = 10,491 individuals) Received allocated LSSS intervention(s): NR Did not receive allocated LSSS intervention(s): NR Lost to follow‐up (LSSS intervention group): n = 1969 individuals (death) Discontinued intervention (LSSS intervention group): NR Analysed (LSSS intervention group): n = 10,504 (cardiovascular outcomes ‐ fatal or non‐fatal stroke; major cardiovascular outcomes; all‐cause mortality) n = 786 at 12 months (DBP, SBP); n = 1412 at 24 months (DBP, SBP); n = 584 at 36 months (DBP, SBP); n = 587 at 48 months (DBP, SBP); n = 7436 (60 months) (DBP, SBP) Excluded from analysis (LSSS intervention group): NR Received allocated control: NR Did not receive allocated control: NR Lost to follow‐up (control group): 2203 individuals (death) Discontinued intervention (control group): NR Analysed (control group): n = 10,491 (cardiovascular outcomes ‐ fatal or non‐fatal stroke; major cardiovascular outcomes; all‐cause mortality) n = 658 at 12 months (DBP, SBP); n = 1374 at 24 months (DBP, SBP); n = 584 at 36 months (DBP, SBP); n = 559 at 48 months (DBP, SBP); n = 7081 (60 months) (DBP, SBP) Excluded from analysis (control group): NR
Interventions	Intervention Characteristics LSSS intervention Theoretical basis: Lowering sodium intake and increasing potassium intake with a reduced‐sodium, added‐potassium salt substitute may reduce the risk of stroke by lowering BP. Description: Reduced sodium, added‐potassium salt substitute (30 ± 10% KCl; 70 ± 10% NaCl at average intake of 20 g per person per day, to a maximum of 20 kg per year for a household with 3 members LSSS category: ≥ 30% KCl Contains fortificant: NR Delivery: Discretionary use Duration of run‐in period: None Duration of active intervention: mean duration of 4.74 years Duration of follow‐up (as reported): none Total duration of study (as reported): mean duration of 4.74 years Timing: Participants provided with sufficient quantity to cover all cooking and food preservation activities Implementation: Village doctors advise participants to use the salt substitute more sparingly, and not more frequently, than they previously used salt. Distributed quarterly. Providers: Local provider in each county. Village doctors provided regular supply of LSSS to each household. Co‐interventions: NR Resource requirements: LSSS purchased from local providers; provided free of charge to trial households Integrity of delivery: Oral, written and other reminders (e.g. cooking apron) to reinforce use of LSSS. Adherence assessed with questionnaire in random subsets of participants at 12, 24, 36, 48 and 60 months Control Theoretical basis: To enable comparison of LSSS intervention with standard practice Description: Usual salt used by household; advice about reducing salt intake given at study commencement Contains fortificant: NR Delivery: Discretionary use Duration of run‐in period: none Duration of active intervention: mean duration of 4.74 years Duration of follow‐up (as reported): none Total duration of study (as reported): mean duration of 4.74 years Timing: Participants provided with sufficient quantity to cover all cooking and food preservation activities Implementation: Advice about reducing salt intake given at study commencement Providers: Own household supply Co‐interventions: NR Resource requirements: NR Integrity of delivery: "Participants in the control villages continue with their use of usual salt with advice about reducing salt intake given only at the commencement of the study."
Outcomes	Primary outcomes:  Diastolic Blood Pressure (DBP): Outcome measurement: automated devices (Omron Electronic Blood Pressure Monitor) with participants in a sitting position. Two measurements were taken at least 5 minutes apart with the average of the two measurements used for the analysis; time points: 12, 24, 36, 48 months (random samples of at least 20 participants drawn from a stratified random sample of at least 60 villages) and 60 months (all available and surviving participants) Systolic Blood Pressure (SBP): Outcome measurement: automated devices (Omron Electronic Blood Pressure Monitor) with participants in a sitting position. Two measurements were taken at least 5 minutes apart with the average of the two measurements used for the analysis; time points: 12, 24, 36, 48 months (random samples of at least 20 participants drawn from a stratified random sample of at least 60 villages) and 60 months (all available and surviving participants) Hypertension (as reported, or SBP > 140 mmHg or DBP > 85 mmHg): NR Blood pressure control (achieving blood pressure threshold or 'control' as prespecified): NR Cardiovascular events‐stroke: Outcome measurement: fatal and non‐fatal stroke defined as an acute disturbance of focal neurological function resulting in death or symptoms lasting more than 24 hours; imaging, clinical and laboratory data will be collected wherever possible. Time points: every 6 months for the duration of the study period (5 years) Cardiovascular events‐myocardial infarction: NR Cardiovascular events‐other: Outcome measurement: non‐fatal stroke (defined as an acute disturbance of focal neurological function resulting in symptoms lasting more than 24 hours); acute coronary syndrome (myocardial infarction or unstable angina event that leads to a hospitalisation, defined by 2 of the following: a clinical history, electrocardiogram changes, and enzyme levels). Time points: every 6 months for the duration of the study (5 years) Cardiovascular mortality: Outcome measurement: Vascular death from an acute coronary syndrome event, stroke, heart failure, suspected vascular causes, or sudden cardiac death. Time points: every 6 months for the duration of the study (5 years) Blood potassium: NR Hyperkalaemia: Outcome measurement: defined as either a serum potassium concentration greater than 6.0 mmol/L with or without other positive findings or a serum potassium concentration greater than 5.5 mmol/L with evidence of typical electrocardiogram changes; time points: duration of study period (5 years) Hypokalaemia: NR Secondary outcomes:  All‐cause mortality: Outcome measurement: A verbal autopsy completed with the help of a caregiver that witnessed the circumstances leading to death; cause of death grouped according to the chapter headings of the Revision of International Statistical Classification of Diseases and Related Health Problems (ICD‐10) version 2016, with additional granularity of reporting done for selected common outcomes of the circulatory system; time points: every 6 months for the duration of the study (5 years) Adverse events: NR Antihypertensive medication use: NR Body Mass index (BMI): NR Serum creatinine: NR Albuminuria: NR Urinary albumin‐to‐creatinine ratio (uACR): NR Fasting blood glucose: NR Blood triglycerides: NR Total blood cholesterol: NR 24‐h urinary sodium excretion: 24‐hour urine sample. Time points: 60 months 24‐h urinary potassium excretion: 24‐hour urine sample. Time points: 60 months
Notes	Funding source: Supported by research grants (APP1164206 and APP1049417) and an investigator grant (APP1197709, to Dr. Neal) from the National Health and Medical Research Council of Australia. Dr J HY Wu is supported by a UNSW Scientia Fellowship. Salt substitute donated by Jiangsu Sinokone Technology for years 3 to 4 of the study Authors name: Bruce Neal Institution: Sydney School of Public Health, University of Sydney, Sydney, Australia. Email: bneal@georgeinstitute.org.au Possible conflicts of interest (for study authors): Paul Elliott declared various grants from non‐profit organisations and is an unpaid member of 'Action on Salt and health; World Action on Salt, Sugar and health'; Dr. Yangfeng Wu received a research grant from Health Source (Chongqing) Cardiovascular Health Technology Co. (a pilot study on the effect of very low‐sodium salt substitute among patients with hypertension); other study authors declared no conflict of interest. Sources used for data extraction: Journal article(s) with results of the trial; trial protocol (published; trial registry); conference abstract Trial registration details: NCT02092090
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	The authors stated that county‐stratified random assignment occurred through a central computerised process, done by an independent biostatistician.
Allocation concealment (selection bias)	Low risk	Randomisation was conducted by an independent biostatistician through a central computerised process.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Although participants and personnel involved with study implementation, e.g. village doctors, were not blinded, the risk of bias was minimised by the measurement of objective outcomes such as clinical events or measurements such as blood pressure.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	The trial registry indicated that investigators and outcomes assessors were blinded to allocation, measurement of process indicators were intended to be measured separately to avoid unblinding, and it appeared as though an analysis plan was finalised before data unblinding. However, Huang and colleagues 2020 reported that ‘the trial has an unavoidable open design and biases consequent upon the unblinded design cannot be entirely excluded’. Endpoint adjudication committees were blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Mortality: Both groups had complete follow‐up outcome data. Cardiovascular events: Non‐fatal events were assessed among the participants for 99,473 of 99,522 person‐years (> 99.9%) of scheduled follow‐up. SBP and DBP: Death was the main reason for the attrition in both groups at 60‐months follow‐up. Reported attrition, other than death, was 10.5% (1099/10504) in the intervention vs. 11.5% (1207/10491) in the control group. Data for mortality were complete, and data for events could also be found for all surviving participants (albeit from sources with different quality and risk for misclassification). Attrition for blood pressure measurements in participants surviving at 60 months was 1099/8535 (13%) in the intervention and 1207/8288 (15%) in the control arm, however, there appeared to be no difference in the reasons for most of the missingness (followed up through nominated contact, followed up through record linkage, followed up at previous visit). Blood pressure measurements for participants followed up in person remained unaccounted for in 286/7722 (4%) participants in the intervention and 327/7408 (4%) participants in the control arm.
Selective reporting (reporting bias)	Unclear risk	Some prespecified outcomes e.g. total major vascular events (as a composite outcome) were not reported. The study authors reported subgroup analyses, which was not prespecified.
Other bias	High risk	Misclassification bias: The study authors reported that information used for adjudication of outcome events was limited, and definitive assignment of causation was difficult in many cases. However, they reported no evidence of a potential effect of misclassification of clinical end‐points on effect estimates. Further, the authors stated that the lack of serial monitoring of blood potassium in the study may have resulted in missing participants with hyperkalaemia. Definite and probable hyperkalaemia events were defined on the basis of medical reports documenting a serum potassium > 5.5 mmol/L or ECG or enzyme changes. A post hoc decision was taken to classify participants with possible hyperkalaemia (based on self‐report; not requiring any supporting documentation).
Recruitment bias (cluster‐RCTs)	Low risk	All participants in the selected villages were recruited before villages were allocated to the intervention or control group.
Comparability with individually randomised trials (cluster‐RCTs)	Unclear risk	There were no RCTs that reported on long‐term cardiovascular clinical outcomes, following LSSS interventions. To date, some RCTs have reported short term benefits in cardiovascular risk factors, such as reductions in blood pressure (CSSS Collaborative Group 2007; Zhao 2014), while others have not (Gilleran 1996; Mu 2003; Sarkkinen 2011; Suppa 1988).
Loss of clusters (cluster‐RCTs)	Low risk	The complete data on vital status for all participants indicated that it was very unlikely that entire clusters were lost from the study.
Baseline imbalance (cluster‐RCTs)	Low risk	No baseline imbalances reported between the intervention and control groups in terms of demographic characteristics or comorbidity risk
Incorrect analysis (cluster‐RCTs)	Low risk	Adjustment for clustering was made with the use of a hierarchical Poisson regression model.
Overall risk of bias	Low risk	Low risk of bias for incomplete outcome data, baseline imbalance, recruitment bias and loss of clusters