An impact of dietary intervention on blood pressures among diabetic and/or hypertensive patients with high cardiovascular disorders risk in northern Thailand by cluster randomized trial

Abstract Background Global sodium intake remains above the recommended levels to control blood pressure (BP). We aimed to evaluate the efficacy of a dietary intervention on BP through salt reduction among community‐dwelling participants with high risk of cardiovascular disorders (CVD). Methods This cluster randomized trial (February 2012 to January 2013) included cooking instruction using the pocket salt meter among patients with diabetes and/or hypertension who were treated at health center in Thailand. Based on health centers, 8 clusters of eligible participants were randomly allocated to the 4 intervention and 4 control groups. Dietary intervention was performed at baseline, 1 month, and 3 months in intervention group. In both groups, systolic and diastolic BPs, and estimated 24 hours salt intake based on overnight urine samples were measured at baseline, 6 months, and 12 months. Results A total of 753 participants were enrolled (374 in the intervention group and 379 in the control group). In the mixed‐effects model, there were significant difference in SBP and estimated salt intake after adjusting covariates at 6 months (adjusted differences between groups [95% CI]; −7.55 [−5.61 to −9.49] mm Hg P < .01; −0.66 [−0.40 to −0.92] g/day P = .03). However, these differences were not observed at 12 months (adjusted differences between groups [95% CI]; −1.83 [0.34 to −4.00] mm Hg P = .48; −0.42 [−0.17 to −0.67] g/day P = .16). There were no differences in DBP in both follow‐ups. Conclusions These results may suggest the effectiveness of a visually based dietary intervention targeting salt intake reduction in short term, but the effectiveness discontinued in long term. Clinical trial number: The International Standard Randomized Controlled Trial Number Register (ISRCTN39416277) on January 3, 2012.


| BACKG ROU N D
Hypertension is a major cause of cardiovascular events, both in developed and in developing countries. An estimated 1.39 (1.34-1.44) billion people had hypertension in 2010, 349 (337-361) million in high-income countries and 1.04 (0.99-1.09) billion in low-and middle-income countries. 1 Elevated blood pressure (BP) is estimated to cause 7.5 million deaths, about 12.8% of all deaths. Hypertension accounts for 57 million disability-adjusted life years (DALYS) or 3.7% of total DALYS. 2 High salt intake has been associated with high BP, thereby contributing to cardiovascular disease (CVD). [3][4][5][6] Salt intake reduction has been established as an important public health strategy for prevention of hypertension. 7-10 Hypertension (raised blood pressure ≥140/90 mm Hg) is common among Thais as well as worldwide and is on the rise. One out of 4 adult Thais has hypertension.
Hypertension causes more than 50 000 deaths annually. 11 Several intervention trials have shown that a low-sodium diet has the potential to reduce BP. 12,13 In spite of these evidences, global sodium intake still does not reach the levels recommended by the World Health Organization (WHO) which is below 5 g/day, and Thailand is relatively higher sodium intake levels (5.31 g/day) in the world. [14][15][16] Although the importance of salt reduction is emphasized in Thailand as well as worldwide, the evidences regarding salt reduction intervention are still limited and there are few trials to use visualization tools to inform participants of the salt content. Under this background, we conducted a cluster randomized trial (CRT) in the northernmost province of Thailand (Risk Patients by Advanced Health Education Intervention [RESIP-CVD Study]). 17 In the trial, the effect of an intensive health education intervention that used visualization tools to help increase awareness of daily salt intake was examined and baseline data were already reported. 18 We aimed to evaluate the effectiveness of the intervention on blood pressure through salt intake reduction among those with a high CVD risk in northern Thailand using longitudinal data analysis.

| Study site
The study was conducted in Muang district, Chiang Rai, the northernmost province of Thailand. The patients with diabetes and/ or hypertension were diagnosed and treated by the doctors at the hospital. When they became stable, the patients were referred to the clinics of health centers according to their residential places. All health centers in Thailand are taken care by nurse and public health officers, instead of doctors. There were about 31 health centers in survey site where patients with diabetes and/or hypertension could receive the medications, and we screened all health centers for eligibility criteria. Every cluster had a chance to be allocated to control and intervention arm through the procedure. Figure 1 shows flow diagram of participants. We screened patients who had visited diabetes and/or hypertension clinics of health centers, and enrolled those who met the eligibility criteria which screened participants, the Framingham general CVD risk score (>15%). 17 Details of the eligibility criteria and exclusion criteria were described in our previous report which analyzed baseline data and enrolled those who met the following eligibility criteria: (a) diabetic and/or hypertensive patients with high CVD risk according to the Framingham general CVD risk score (>15% of the 10 years cardiovascular risk, which were calculated using the information on gender, SBP, total cholesterol [mg/ dL], high-density lipoprotein cholesterol [mg/dL], hypertension medication, cigarette smoker, and diabetes treatment) and (b) patients who were willing to participate in the study. Patients fulfilling the following criteria were excluded: (a) pregnant or trying to become pregnant; (b) age less than 35 years; (c) documented type I diabetes mellitus; (d) undergoing long-term steroid therapy (more than two weeks); (e) undergoing long-term nonsteroidal anti-inflammatory drug (NSAID) therapy (ie, every day for at least one year); (f) cancer; (g) known secondary hypertension such as primary aldosteronism, Cushing's syndrome, or pheochromocytoma; (h) severe chronic pulmonary diseases requiring home oxygen therapy; (i) chronic renal disease (creatinine ≥2.0 mg/ dL); (j) congestive heart failure; and (k) a known diagnosis of CVD. 17,18 Informed consent was obtained from all participants.

| Clusters
The units of randomization were health centers. A statistician who did not know about study and intervention was asked to produce random sequenced numbers using Stata software, version 11. 17 Then, eligible clusters were randomly allocated to the intervention and control groups by simple randomization. 17

K E Y W O R D S
awareness, dietary, hypertension, intervention, nutrition, salt intake

| Selecting clusters and randomization
Eligibility of cluster was set in order to prevent an empty cluster as that any health center with a total of fewer than 200 patients attending the hypertension and diabetes clinic was not included. Within each cluster, a total of 100 eligible participants were recruited. 17 We selected 8 of 31 health centers, applying eligibility criteria for a cluster. The unit of randomization was a cluster, which was a health center. Cluster size was 100, and cluster number was eight. Eligible eight clusters were randomized to four clusters in intervention and four clusters in control arm applying simple randomization. On the launching day of the study, we invited cluster representatives and randomization procedure was conducted. Before randomization, researchers and cluster representatives could not know in which arm each cluster will be located.
After the randomization, clusters were informed about the details of the study procedure, separately for intervention clusters and control clusters, in order to minimize the contamination through second-hand education message between participants of two arms.
Furthermore, we recognized that we got the control arms and intervention arms, coincidently situating at the opposite side of the motor expressway in the study after random allocation, 17 and people residing in such opposite places did not cross the road to access health care at primary care units. 17

| Recruitment within a cluster
Among all eligible cases in each cluster of primary care unit, eligible 100 participants were able to be randomly recruited. The cluster representatives knew their allocated arm after random allocation, and the study procedure of only their allocated arm was explained to them.
Individual participants were given information pertaining to the study group in their allocated cluster only after enrollment.

| Baseline characteristics
Baseline data have been reported previously. 18 We interviewed participants to collect data on their baseline characteristics including age, gender, medical history (hypertension, dyslipidemia, diabetes mellitus, CVD, cerebrovascular disease, and kidney disease), and family medical history (hypertension, diabetes mellitus, F I G U R E 1 Flow diagram of participants 31 clusters were assessed for eligibility 23 clusters were excluded due to ineligibility 8 clusters were randomized (n = 795) 4 clusters were assigned to control group (n = 402) 4 clusters were assigned to intervention group (n = 393) 374 were included in the analysis 379 were included in the analysis 19 were excluded due to discontinuation 23 were excluded due to discontinuation CVD, and cerebrovascular disease). We also inquired about lifestyle characteristics listed in Breslow's 7 health practices and defined the components of a healthy lifestyle, as indicated in parentheses as follows: alcohol consumption (<3 days/week), smoking behavior (current nonsmoker), exercise frequency (≥2 times/ week), body mass index (BMI; 18.5-24.9), sleep hours (6-9), breakfast consumption habits (every morning), and snacking between meals (no). 19,20 A healthy lifestyle was defined as having at least 6 healthy practices.
Awareness and motivation related to salt intake were assessed by a self-administered questionnaire, and detail was already reported as baseline data analysis. 18 We collected BMI, systolic blood pressure (SBP), diastolic blood pressure (DBP), total cholesterol Daily salt intake, which was the primary outcome, was estimated by A KME-03 salinity checker (Kono ME Institute, Kanagawa, Japan) using overnight urine samples. 18,22 Detail of the method of data collection was already reported as baseline data analysis. 17 We instructed participants to store overnight urine appropriately. The participants then brought the collected urine to the health centers, and staff members measured the estimated 24 hours urinary salt excretion. This procedure was performed for three successive nights, and the average measurement was used as the estimated daily salt intake.

| Sample size calculation
Detail of sample size calculation was already described in our previous report. 17 For a conventional randomized control trial, a total of 240 participants, randomizing 120 into each arm, will have enough power (90%) to detect the minimal difference of systolic blood pressure of 130 (±20) and 120 (±20) mm Hg, diastolic pressure of 90 (±20) and 80 (±20) mm Hg, difference in 24 hours estimated salt intake of 10 (±5) and 8 (±4) g/day, and CVD event rate of 0.15% and 0.08% between the control and intervention groups with a 95% confidence interval. 23,24 The cluster randomized trial design used in RESIP-CVD might result in loss of power and reduced efficiency. In this cluster randomized trial, the number of participants in each cluster was the same, and other variable factors such as geography, food tradition, race, and dietary habits were assumed not to be different for delineating the intracluster coefficient. To compensate the design effect, the inflation factor has been calculated by the formula (Deff = 1 + (m−1)ρ) where Deff is inflation factor for design effect, m was the size of each cluster, which was 30, and ρ is the intracluster coefficient which is assumed to be 0.07. The sample required for a cluster randomized trial was inflated to 720. With a loss to follow-up of approximately 10%, we expect that the sample size of 800 patients would yield sufficient power to detect the desired minimal difference in primary and secondary outcomes between the intervention and control groups. All eight clusters of the RESIP-CVD study will enroll 100 eligible participants.

| Intervention
In the intervention group, participants received a health education

| Outcomes
The primary outcome was estimated daily salt intake, and secondary outcomes were SBP and DBP. Follow-up was performed at 6 and 12 months. The secondary outcomes of this study were SBP and DBP at 6 and 12 months. SBP and DBP were measured using the automated sphygmomanometer HEM-907 (Omron Co., Ltd) by welltrained research nurses and calculated by calculating the mean of two upper arm blood pressure measurements taken for participants who had been seated for at least five minutes.
At each follow-up, the baseline survey used to assess baseline characteristics was conducted again for each participant.
Anthropometric measurements, SBP and DBP measurements, instruction on overnight urine collection, and estimation of daily salt intake from urine were all performed in the same manner as baseline.

| RE SULTS
In total, 795 participants were enrolled in the study and 753 completed both follow-ups (control group: 379; intervention group: 374).
Baseline characteristics of participants are shown in Table 1

| D ISCUSS I ON
To the best of our knowledge, this RCT was the first trial to assess the effectiveness of a dietary intervention involving cooking instruction and visualization of salt content of food by a digital handheld pocket salt meter in Thailand.
Our results showed that SBP was significantly improved in the intervention group compared to the control group at 6 months, while the significance was not observed at 12 months, suggesting that the intervention may have had a short-term effect. Several previous trials have indicated the effectiveness of salt intake reduction interventions in lowering BP. An experimental intervention study using a specific diet (emphasizing fruits, vegetables, and low-fat dairy foods; including whole grains, poultry, fish, and nuts; and low in fats, red meat, sweets, and sugar-containing beverages) coupled with reduced sodium intake (50, 100, and 150 mmol/d at 2100 kcal) reported that the diet and reduced sodium intake were each associated with significant decreases in BP and that these two factors combined produced the greatest reductions. 13 The PREMIER trial, which was a multicenter, randomized controlled trial of 810 participants with high-normal hypertension, revealed that mean SBP diet group (P < .001). 26 The results of these studies, as well as ours, suggest that a dietary intervention may significantly improve BP in both the general and high-risk populations.

Intervention group (N = 374) P-value
In addition to BP, estimated salt intake was significantly reduced  Note: Changes in variables between groups and associated test of effect estimated regression adjusting clustering and baseline covariates: body mass index, alcohol consumption (less than 3 days per week), smoking behavior (current nonsmoker), exercise frequency (2 times or more per week), sleep hours (6-9), breakfast (every morning), snack between meals (no), antihypertensive drug use (yes), antidyslipidemic drug use (yes), antidiabetic drug use (yes), total cholesterol (mg/dL), hemoglobin A1c (%), awareness of seriousness of CVD (yes), and motivation to reduce salt intake (yes).
we registered only participants with higher cardiovascular risks in the present study, and it is required to assess participants with low cardiovascular risks in future analysis. Third, the automated device used to estimate 24 hours salt intake based on overnight urine was validated in Japanese individuals, but not in other ethnicities.
In addition, validity and reliability of a digital handheld pocket salt meter were not fully estimated. To assess the validity and reliability of a digital handheld pocket salt meter in other ethnicities, additional study is needed. Fourth, we could not collect the detail of medication including diuretics and could not consider the effect of diuretics on urinary sodium excretion, and validated assessment considering diuretics is required. Fifth, we did not collect the detail of other dietary and nutrient information such as potassium and protein intake, and it is possible that our results will be mod-

| CON CLUS ION
Our trial data revealed the effectiveness of a dietary intervention combining cooking instruction with visualization of SBP and estimated salt intake in short term, while the effectiveness discontinued in long term. This type of dietary intervention targeting salt intake reduction may be required for BP management, and a developed intervention is necessary for long-term management.