The details of the SSaSS trial have been reported elsewhere [8]. In brief, SSaSS was an open-label large-scale, cluster-randomised, controlled trial conducted in 600 villages across five provinces of Northern China (Hebei, Liaoning, Ningxia, Shanxi, and Shaanxi). The provinces were selected because of the high prevalence of hypertension and stroke. The villages were randomised to receive salt substitute or a continued usual diet with regular salt. The SSaSS study showed a protective effect of sodium reduction using a salt substitute on the risk of cardiovascular events and mortality [9]. The headache outcome was added during the study protocol amendment (protocol date: 15/5/2019, ethics approval date: 27/6/2019), and the analysis plan was finalized before data lock [9].

The ethics committees of Peking University Health Science Centre, China (IRB00001052-13069), and the University of Sydney, Australia (2013/888), approved the study. An independent Data Monitoring Committee oversaw the study, and informed consent was obtained from all the participants.

Adults with a high risk of developing a stroke were included in this study. Approximately 35 individuals from each of the 600 villages were recruited using the following inclusion criteria: (a) prior history of stroke (regardless of underlying aetiology) and/or (b) age ≥60 years with uncontrolled high blood pressure (BP) defined as systolic BP ≥ 140 mmHg after two measurements in people using BP-lowering drugs or systolic BP ≥ 160 mmHg after two measurements in individuals receiving no medication for high BP. In addition, the participant or a household member had to own a phone and provide another person’s number as an emergency contact.

Exclusion criteria were: (a) using potassium-sparing diuretics or potassium supplements; (b) diagnosis of severe renal impairment; (c) any concern about using salt substitute; (d) life expectancy less than six months from the trial commencement; (e) eating most meals outside of the home. Informed consent was obtained from every participant, and group consent was obtained from local county Bureaus of Health leadership.

Villages were randomly assigned in a 1:1 ratio to the intervention group or the control group. Randomisation was stratified according to county, using a central computerised process. Random assignment of the villages was performed after all the participants in the province had been recruited and all baseline survey data had been collected. Based on the nature of the study, the intervention’s delivery could not be masked.

There were two groups in the study: the Intervention group that received salt substitute and the control group that continued the usual diet.

Intervention: Participants and their households in villages randomised to the intervention used the salt substitute sufficient for all daily cooking and seasoning needs instead of the regular salt. Salt substitution is a form of salt that has 75% sodium chloride and 25% potassium chloride by mass compared to regular salt which is 100% sodium chloride. Each household received about 20 g of salt substitute per person per day (maximum of 20 kg per year per household) free of charge. Participants were advised to use the salt substitute instead of regular salt and use it less frequently than their previous salt usage habits. In addition, participants received general health advice regarding stroke prevention and healthy behaviours.

Control: Villages in the control group continued the usual diet with regular salt (100% sodium chloride) and were advised to use salt less frequently than their previous salt usage habits. This group also received general health advice for stroke prevention and healthy behaviours at the beginning of the study, like the intervention group.

After obtaining informed consent, baseline data including age, gender, body mass index (BMI), any history of smoking, history of stroke, hypertension, diabetes Mellitus, and use of anti-hypertensive drugs were collected via interview, thereafter a physical examination which included systolic BP (SBP), and diastolic BP (DBP), assessments was performed. The village doctor was responsible for distributing the salt substitute but was not involved in any outcome collection or assessments. The trained outcome assessors oversaw evaluations in a similar way in all the included villages and were instructed not to ask about the randomisation status. Outcome assessment was performed in person if the participants were available in the village on the visit day; otherwise, it was undertaken via phone call or by rescheduling to another day.

Headache frequency and severity outcomes were evaluated at final follow-up only (five years) by recording the number of days in the past month that a participant experienced a headache (headache frequency)—(a) 0 days, (b) 1–10 days, (c) 11–20 days, (d) >20 days, or (e) unknown. Participants reporting headache were asked whether the severity of headaches was mild, moderate, severe or unknown. Given this was a pragmatic, large scale trial, no details about headache were collected at baseline and no details of headache type were recorded.

Primary analyses were performed according to the intention-to-treat principle. Statistical analyses were performed using SAS software, version 9·4 (SAS Institute). Binary outcome of headache incidence (0 days vs ≥1 days) was analysed using hierarchical log-binomial regression model: results presented as rate ratios, 95% confidence intervals, and p values. Ordered categorical outcomes of headache frequency and headache severity were analysed using hierarchical ordered logistic regression model: results presented as odds ratios, 95% confidence intervals, and p values. All hierarchical models adjusted for clustering at the village level. For continuous outcomes including pulse pressure, blood pressure reduction, change in sodium excretion and potassium excretion: mean difference estimates, confidence intervals and p values were obtained from an analysis of covariance that allowed for clustering.

Subgroup analysis was performed to calculate the risk of having headache compared to no headache in the last month for groups defined by baseline age, BMI, SBP and DBP, sex, education, and history of disorders such as stroke, hypertension, and diabetes.

The study was registered with ClinicalTrials.gov in March 2014 with the identifier number NCT02092090.

The trial was funded by the National Health and Medical Research Council of Australia (NHMRC) Project Grant (APP1049417), NHMRC Program Grant (APP1052555), and NHMRC Centre for Research Excellence Grant (APP1117300).