Nonpharmacologic Interventions for Reducing Blood Pressure in Adults With Prehypertension to Established Hypertension.

Background Nonpharmacologic interventions that modify lifestyle can lower blood pressure (BP) and have been assessed in numerous randomized controlled trials and pairwise meta-analyses. It is still unclear which intervention would be most efficacious. Methods and Results Bayesian network meta-analyses were performed to estimate the comparative effectiveness of different interventions for lowering BP. From 60 166 potentially relevant articles, 120 eligible articles (14 923 participants) with a median follow-up of 12 weeks, assessing 22 nonpharmacologic interventions, were included. According to the surface under the cumulative ranking probabilities and Grading of Recommendations Assessment, Development and Evaluation (GRADE) quality of evidence, for adults with prehypertension to established hypertension, high-quality evidence indicated that the Dietary Approach to Stop Hypertension (DASH) was superior to usual care and all other nonpharmacologic interventions in lowering systolic BP (weighted mean difference, 6.97 mm Hg; 95% credible interval, 4.50-9.47) and diastolic BP (weighted mean difference, 3.54 mm Hg; 95% credible interval, 1.80-5.28). Compared with usual care, moderate- to high-quality evidence indicated that aerobic exercise, isometric training, low-sodium and high-potassium salt, comprehensive lifestyle modification, breathing-control, and meditation could lower systolic BP and diastolic BP. For patients with hypertension, moderate- to high-quality evidence suggested that the interventions listed (except comprehensive lifestyle modification) were associated with greater systolic BP and diastolic BP reduction than usual care; salt restriction was also effective in lowering both systolic BP and diastolic BP. Among overweight and obese participants, low-calorie diet and low-calorie diet plus exercise could lower more BP than exercise. Conclusions DASH might be the most effective intervention in lowering BP for adults with prehypertension to established hypertension. Aerobic exercise, isometric training, low-sodium and high-potassium salt, comprehensive lifestyle modification, salt restriction, breathing-control, meditation and low-calorie diet also have obvious effects on BP reduction.

credible interval

Dietary Approach to Stop Hypertension

diastolic blood pressure

Grading of Recommendations Assessment, Development and Evaluation

systolic blood pressure

surface under the cumulative ranking

weighted mean difference

Clinical Perspective

What Is New?

Single interventions including Dietary Approach to Stop Hypertension (DASH; ranked first), aerobic exercise, isometric training, low‐sodium and high‐potassium salt, salt restriction, breathing‐control, and meditation are associated with effective reduction of systolic blood pressure and diastolic blood pressure.

Comprehensive lifestyle modification, as a combined intervention, can lower both systolic and diastolic blood pressure effectively.

For patients who are overweight and obese, low‐calorie diet or low‐calorie diet plus exercise could lower blood pressure levels more than exercise alone.

What Are the Clinical Implications?

High‐quality evidence suggests that DASH might be the most effective intervention to lower blood pressure for adults with prehypertension to established hypertension.

Aerobic exercise, isometric training, low‐sodium and high‐potassium salt, comprehensive lifestyle modification, salt restriction, breathing‐control, and meditation should also be recommended to lower blood pressure for patients with hypertension who are receiving pharmacotherapy.

Weight loss from low‐calorie diet or low‐calorie diet plus exercise could lower blood pressure level more than exercise alone among people who are overweight and obese.

Hypertension is an important worldwide public health problem. As populations age, adopt unhealthy lifestyles, and increase their body weight, the number of people with hypertension will continue to increase, reaching close to 1.5 billion by 2025. Studies have proven that hypertension is a strong risk factor for severe cardiovascular events, including myocardial infarction and stroke, if uncontrolled. , , Compared with people who are normotensive, patients with prehypertension have a higher risk of developing sustained hypertension and cardiovascular disease. , Pharmacotherapy with first‐line antihypertensive agents has significant effects in lowering blood pressure (BP) but also has side effects, treatment resistance, and financial burden. Effective, widely available, low‐cost, and sustainable strategies are needed to prevent and manage hypertension.

Numerous randomized controlled trials (RCT), systematic reviews, and meta‐analyses have assessed the BP‐lowering effects of nonpharmacologic interventions. , , US, Canadian, and European guidelines for hypertension recommend different nonpharmacologic interventions to prevent and manage hypertension. , , However, their suggestions were based on traditional meta‐analysis, which can only compare the relative efficacy of pairs of interventions. A study that can compare the BP‐lowering effects of different nonpharmacologic interventions comprehensively is urgently needed to provide concrete evidence of the practice of nonpharmacologic interventions. Network meta‐analyses can synthesize direct and indirect evidence in a network of studies that compare multiple interventions. This approach has the potential to rank the competing treatments according to the studied outcome and determine the best available option for intervention. , ,

The aim of our study was to assess the comparative effectiveness of different nonpharmacologic interventions for reducing BP in adults with prehypertension to established hypertension and to determine the most efficacious intervention.

Methods

This network meta‐analysis is reported according to the Preferred Reporting Items for Systematic Reviews and Meta‐analyses (PRISMA) extension statement for reporting of systematic reviews incorporating network meta‐analyses of healthcare interventions. All supporting data are available within the article and its online supplementary files. The study protocol can be found online (https://www.crd.york.ac.uk/PROSP​ERO/displ​ay_record.php?Recor​dID=67522). An extended description of the methods is reported in Data S1.

Search Strategies, Eligibility Criteria, and Information Sources

We searched PubMed, Embase, Cochrane Central Register of Controlled Trials, Clini​calTr​ials.gov, and the EU Clinical Trials Register up to July 1, 2019, to identify eligible studies. We limited searches to English‐language publications and supplemented them by perusing reference lists of reviews and retrieved literature. The search strategies are presented in Data S1.

We included RCTs of at least 4 weeks' duration that compared the BP‐lowering effects of nonpharmacologic interventions for adult patients (aged ≥18years) with primary hypertension or prehypertension. Patients with hypertension were defined as those with office systolic BP (SBP) ≥140 mm Hg and/or diastolic BP (DBP) ≥90 mm Hg without taking antihypertensive medication or those with established hypertension using antihypertensive medication, even if BP was <140/90 mm Hg. Prehypertension was defined as an office SBP of 120 to 139 mm Hg and/or DBP of 80 to 89 mm Hg, according to the Joint National Committee in the Seventh Report. Eligible interventions were different nonpharmacologic therapies. Comparators were other nonpharmacologic therapies or usual care as a control. We excluded studies that enrolled participants who had a history of heart failure, renal disease, stroke, serious mental or physical illness, malignancy, diabetes mellitus, or metabolic syndrome. We also excluded studies that focused on postmenopausal or pregnant women or single‐sex populations. However, studies that focused on participants who used alcohol habitually or who were overweight or obese were included because those features could be modified.

Study Selection and Data Collection Process

Two reviewers (Y.L. and D.L.) independently screened the titles and abstracts of all potentially eligible studies. Three reviewers (J.F., L. Zhang, and L. Zhou) performed full‐text review to identify studies that met all criteria for inclusion in the quantitative synthesis. Disagreements were resolved by discussion.

Pairs of independent reviewers extracted relevant data from each eligible study in duplicate, and discrepancies were resolved by discussion among reviewers. We extracted data on characteristics and demographics of study participants, mean baseline and follow‐up SBP and DBP, dropout, and other information.

End Points and Handling of Missing Data

Reductions of SBP and DBP after intervention were separately evaluated as co–primary end points, and the summary estimates were calculated by using the mean difference and SE. If the SEs of the mean differences were not available from included articles, we either estimated SEs based on the sample size, median, and range or based on the mean difference, sample size, and P value. We also imputed these data by estimsting correlation coefficient values between baseline and follow‐up.

Statistical Analysis

Traditional meta‐analyses were conducted using a random‐effects model for every direct comparison. Next, Bayesian random‐effects network meta‐analyses were performed using the GeMTC package (R 3.4.3) based on the Markov‐chain Monte Carlo method. , Comparative effect estimates are presented as the weighted mean difference (WMD) and 95% credible interval (CrI) because all end points were continuous variables. Trace plots and the Brooks–Gelman–Rubin statistic were assessed to ensure convergence. Network consistency between direct and indirect evidence was analyzed by the node‐splitting method, and its bayesian P value was reported. Statistical heterogeneity of studies and the global heterogeneity of network meta‐analysis were also examined using the I statistic.

The relative rankings of different nonpharmacologic interventions were calculated using surface under the cumulative ranking (SUCRA) probabilities and were presented graphically (WinBUGS 1.4.3 [BUGS Project] and Stata 14.0 [StataCorp]). Sensitivity analyses were conducted by omitting data from specific studies, including studies with high risk of bias, studies started before 1999 (international diagnostic criteria for hypertension were issued by the World Health Organization in 1999), studies with end points of home BP or 24‐hour ambulatory BP, or studies targeted to special population (participants who used alcohol habitually or who were overweight or obese). Metaregression analyses were also performed by adding covariates (mean or median age, mean body mass index [BMI], proportion of participants taking antihypertensive medicine, and proportion of female patients). In addition, subgroup network meta‐analyses were conducted in different subgroups defined by study duration or region of origin of study participants. Because all analyses were based on bayesian framework, no multiplicity was adjusted. Publication bias was assessed using the comparison‐adjusted funnel plot and the netfunnel command (Stata 14.0).

Risk‐of‐Bias Assessment and Certainty of Evidence

Two reviewers (J.F. and L. Zhang) assessed the risk of bias separately for each included study using the Cochrane risk‐of‐bias tool (RevMan 5.3). They also assessed the quality of evidence contributing to each direct, indirect, and network estimate independently using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) method of network meta‐analysis. , Discrepancies were resolved by discussion with another reviewer (Y.L.).

Results

A total of 60 166 articles were identified in the initial systematic search, and 888 potentially eligible articles were retrieved as full text. Overall, 120 articles (corresponding to 126 RCTs) with 14 923 participants met the inclusion criteria and were included in the network meta‐analysis (Figure 1).

Figure 1

PRISMA flow chart of the study selection for the network meta‐analysis.

*In case of multiple publications from the same population, only the study with the largest sample size was included. For studies published more than once, only the study with the most informative and complete data was included. Any additional publications were excluded to avoid double counting data from the same trial. PRISMA indicates Preferred Reporting Items for Systematic Reviews and Meta‐analyses.

Study Characteristics

This network meta‐analysis covered 22 nonpharmacologic interventions including dietary approaches, physical exercise, approaches to reduce stress or lose weight, restriction of alcohol intake, combined interventions, and comprehensive lifestyle modification. All 22 interventions have been practiced in clinical or community trials, and brief descriptions and median intensity of all interventions and usual care are presented in Table 1. , , , , , , , , , , , , , , , , , , , , , , , , , , , , , Baseline and basic characteristics of included studies are presented in Table S1. Overall, 8530 participants were randomly assigned to intervention groups and 6393 to usual care; the mean age of all participants was 51.2 years; the median proportion of female patients was 0.49 (range, 0.05–0.88); median study duration was 12 weeks (range, 4–144); studies recruited participants from Europe (49.20%), the United States (24.60%), Asia (13.49%), and Africa (3.18%), and included Black Americans (9.53%); of 126 included studies, 91 (72.22%) recruited only patients with hypertension, 27 (21.43%) recruited patients with hypertension and prehypertension, and 8 (6.35%) recruited only patients with prehypertension (Table 2). The mean SBP and DBP levels of adults with prehypertension to established hypertension were 136.74 and 86.27 mm Hg, respectively. The mean SBP and DBP levels of patients with hypertension were 143.80 and 87.51 mm Hg, respectively.

Table 1

Coding Guide for Components of Nonpharmacologic Interventions

Intervention/Abbreviation	Brief Descriptions	Median Intensity
Dietary approach
DASH 34 , 35	Participants' diet strictly follows the DASH eating pattern, which recommends a diet rich in fruits, vegetables, whole grains, and low‐fat dairy with reduced sodium and saturated and total fat content	Eating on the DASH pattern every day
Low‐sodium and high‐potassium salt 36 , 37	Participants receive either a salt substitute (25%−30% potassium chloride, 50%−65% sodium chloride, and 5%–10% calcium and magnesium sulfate) to cover all cooking, or test food cooked using salt substitution	5 g of low‐sodium and high‐potassium salt every day
Salt restriction 38 , 39	The goal is to restrict daily sodium intake <100 mmol (5.85 g salt). Professional instructors give participants detailed advice about how to reduce their salt intake and to avoid foods that contain large amount of salt and also offer metric salt‐spoon or placebo to participants	Restrict sodium intake <100 mmol (5.85 g salt) every day
Physical exercise
Aerobic exercise 40 , 41	Participants are supervised by project staff to perform exercise (eg, treadmill or brisk walking, jogging, bicycle training, swimming, ball games), at least 30 min/time; almost all were moderate or high intensity (60%–90% of the maximum heart rate or maximum oxygen consumption)	3 d/wk, 50 min/time
Isometric training 42 , 43	Participants perform isometric training, which involves sustained contraction against an immovable load or resistance with no or minimal change in length of the involved muscle group. Training consisted of four 2‐min isometric contractions at 30% MVC using alternate hands with a programmed handgrip dynamometer, with a 1‐min rest period between each contraction for 3 d per week	3 d/wk, bilateral contractions at 30% of MVC
Resistance training 44 , 45	Participants perform active movement progress through muscle to overcome external resistance, such as leg press, leg curl, knee extension, chest press, seated row, overhead press, triceps dip, and biceps curl, 50–60 min/d, 2–3 d/wk	3 d/wk
Tai chi 46	Tai chi js a set of Chinese systematic calisthenic exercises with slow circular movements and requires the muscles to remain relaxed while making sustained movement. Participants are taught by instructors with expertise to finish each session, which includes warm‐up exercises, tai chi practice, and cool‐down exercise	3 d/wk, 50 min with 50% to 60% Vo 2max
Qigong 47	Qigong, a traditional Chinese health and fitness exercise, includes qi gong ba duan jin, shu xin ping xue gong and dao yin shu qigong. Qigong experts help participants to reconstruct this instrument using a warming‐up exercise, qigong, and cool‐down exercise	Qigong classes 2 d/wk, home practice 2 d/wk
Interventions to reduce stress
Breathing control 48 , 49	Use of a device guides participants toward slow and regular breathing in the evening (the goal is <10 breaths/min with accumulating ≥40 min of therapeutic breathing per week)	Every day, 15 min/time
Meditation 50	Transcendental meditation is considered the principal approach for stress reduction. Participants are instructed by a professional meditation instructor and then practice 20 min twice a day while sitting comfortably with eyes closed	Practice meditation 20 min twice a day
MBSR 51 , 52	MBSR is a multicomponent group intervention that provides systematic training in mindfulness meditation as a self‐regulation approach to stress reduction and emotion management. It can be explored through activities including but not limited to gentle stretching and mindful yoga, a meditative body scan, mindful breathing, and mindful walking	Practice MBSR techniques 45 min every day
PMR 53 , 54	PMR involves directing the participants' attention to tense and relax various muscle groups throughout the body systematically to achieve deep relaxation	Practice PMR techniques 15–20 min twice a day
Yoga 55	Participants are instructed by a professional yoga instructor through yoga home training or a yoga class and practice yoga at least 30 min/d, 3 d/wk	Practice yoga 3 d/wk, 45 min
Interventions to lose weight
Low‐calorie diet 56 , *	Participants who are overweight or obese using the low‐calorie diet induce weight loss are provided with detailed guidelines on the daily number of servings from each food group and on fat intake to achieve weight loss of ≤10% of each participant's baseline body weight. To enhance compliance with the low‐calorie diet, participants are provided with food diaries that assisted them in recording intake	Low‐caloric diet every day for weight loss
Exercise 56 , *	Participants who are overweight or obese in the exercise training group are provided with an individualized exercise prescription consisting of 30–40 min exercise (eg, aerobic exercise or others), at least 3 d/wk, keeping 60%–80% of the maximum heart rate. To enhance compliance, details of each exercise session are recorded in a training diary and reviewed by the study's counselor	Exercise 3 d/wk, reach 60%–80% peak heart rate
Low‐calorie diet plus exercise 56 , *	Participants who are overweight or obese using the exercise training plus low‐calorie diet for weight loss are provided with detailed guidelines on a low‐calorie diet to achieve weight loss and decrease BMI. In addition, they perform systematic exercise training, 30–45 min/d, at least 3 d/wk, keeping 60%–80% of the maximum heart rate	Low‐caloric diet for losing weight, with exercise 3 d/wk, reaching 60%–80% peak heart rate
Restrict alcohol
Alcohol restriction 57 , 58 , †	Participants reduce their alcohol consumption to <14 drinks weekly or 50% cut or total abstinence, with education for alcohol restriction provided by investigators	Reduce alcohol intake by half or abstain
Combined intervention
Aerobic exercise+DASH 59	Participants follow the DASH eating pattern and perform aerobic exercise	At least 5 d/wk, 30–60 min aerobic exercise plus DASH
Aerobic exercise+resistance training 60	Participants attend an aerobic exercise session and a resistance training session at the center at least twice a week	At least 2 d/wk, endurance training and resistance training
Salt restriction+DASH 35	Participants follow the DASH eating pattern with salt restriction (sodium intake <100 mmol/d)	Follow diet every day
Salt restriction+low‐calorie diet plus exercise 61 , *	Participants who are overweight or obese follow a low‐sodium (80 mmol/d) diet with low‐calorie intake to achieve weight loss of 4.5 kg	Low‐sodium and low‐calorie diet every day; 3 d/wk, reach 60%–80% peak heart rate
Comprehensive lifestyle modification
Comprehensive lifestyle modification 62 , 63	Participants are recommended to comprehensively modify their lifestyle, such as lose weight, restrict sodium intake, reduce alcohol consumption, increase physical exercise to a moderate degree, give up cigarette smoking, and learn to manage stress	Use lifestyle modification every day
Control group
Usual care	Participants keep usual lifestyle and do not change during the period of intervention

Brief descriptions of 22 interventions plus usual care (as control) are summarized, with 17 nonpharmacologic interventions targeted to the general population with hypertension or prehypertension. BMI indicates body mass index; DASH, Dietary Approaches to Stop Hypertension; MBSR, mindfulness‐based stress reduction; MVC, maximum voluntary contraction; and PMR, progressive muscle relaxation.

Nonpharmacologic intervention targeted only people who were overweight and obese who had hypertension or prehypertension.

Nonpharmacologic intervention targeted only people who used alcohol habitually who had hypertension or prehypertension.

Table 2

Details of Included Studies (N=126)

Study Details	n (%)
Region of origin of study participants
Europe	62 (49.20)
America (all)	31 (24.60)
Asia	17 (13.49)
America (Black Americans)*	12 (9.53)
Africa	4 (3.18)
Year thestudy started
1973–1998	54 (42.86)
1999–2019	72 (57.14)
Study design
Parallel	108 (85.71)
Crossover	18 (14.29)
Study duration, wk
<12	55 (43.65)
12–24	54 (42.86)
>24	17 (13.49)
Usage of antihypertensive medications
Yes	43 (34.13)
No	65 (51.58)
Not reported	18 (14.29)
Health status of recruited participants
Hypertension and prehypertension (mixed)	27 (21.43)
Hypertension only	91 (72.22)
Prehypertension only	8 (6.35)

America (Black Americans) studies are those from America that were done in Black participants.

Risk of Bias Within Studies

Of 126 included RCTs, 41 (32.54%) were judged to have low risk of bias, and only 9 (7.14%) were judged to have high risk of bias; all other studies (60.32%) were judged to have moderate risk of bias (Table S2 and Figure S1). All comparison‐adjusted funnel plots of network meta‐analysis for outcomes did not show distinct asymmetry, which suggested no evidence of publication bias in this study (Figure S2).

Network Meta‐Analysis in Adults With Prehypertension to Established Hypertension (BP ≥120/80 mm Hg)

The results of traditional meta‐analyses showed that, compared with usual care, 10 interventions were more effective for lowering both SBP and DBP (Table S3).

Network meta‐analysis included 126 RCTs (14 923 participants) with 22 interventions and usual care. All 22 nonpharmacologic interventions had direct comparison with usual care, and 14 interventions compared directly with at least one other intervention (Figure 2A). Comparative effect estimates of 22 nonpharmacologic interventions in lowering BP are presented in Figure S3. Because indirect comparisons provided observational evidence in network meta‐analysis, we focused on the effective BP‐lowering estimates of interventions that were supported by the combination evidence of direct and indirect comparisons (Figure S3).

Figure 2

Network geometry used to assess the comparative effects of 22 nonpharmacologic interventions.

A, Adult with prehypertension to established hypertension. B, Patients with hypertension. The nodes represent 22 nonpharmacologic interventions and usual care. The size of every node is proportional to the number of randomly assigned participants (sample size). Each line represents a direct comparison, and the width of the lines is proportional to the number of studies comparing every pair of interventions. The coding guide, which provides a description of each intervention component, can be found in Table 1. DASH indicates Dietary Approaches to Stop Hypertension; MBSR, mindfulness‐based stress reduction; and PMR, progressive muscle relaxation.

In terms of lowering SBP, 15 interventions were shown to be more effective than usual care (Figure 3A). Based on SUCRA, the following interventions ranked ahead: tai chi (WMD, 13.47 mm Hg [95% CrI, 9.30–17.64]), Dietary Approaches to Stop Hypertension (DASH; WMD, 6.97 mm Hg [95% CrI, 4.50–9.47]), aerobic exercise plus DASH (WMD, 11.20 mm Hg [95% CrI, 2.81–19.61]), low‐calorie diet (WMD, 6.50 mm Hg [95% CrI, 2.78–10.17]), aerobic exercise (WMD, 6.60 mm Hg [95% CrI, 4.98–8.23]), isometric training (WMD, 5.77 mm Hg [95% CrI, 1.41–10.16]), low‐sodium and high‐potassium salt (WMD, 8.21 mm Hg [95% CrI, 4.99–11.43]), comprehensive lifestyle modification (WMD, 4.63 mm Hg [95% CrI, 1.32–7.94]), and the others include salt restriction, salt restriction plus low‐calorie diet and exercise, breathing‐control, low‐calorie diet plus exercise, meditation, yoga, and alcohol restriction (Figure 3A and Figure S4A). In addition, low‐calorie diet lowered SBP level more than exercise (WMD, 5.36 mm Hg [95% CrI, 0.45–10.25]) for participants who were overweight and obese (Figure S3).

Figure 3

Forest plots for mean changes of blood pressure in adults with prehypertension to established hypertension.

A, Systolic blood pressure. B, Diastolic blood pressure. Mean changes of blood pressure are reported in WMD and 95% CrI for intervention vs usual care. Rectangle represents the point estimate for the effect of each intervention. Horizontal lines indicate 95% CrI. Tables on the left of the forest plot show, for each intervention, the number of direct comparison studies, number of participants, rankings of SUCRA probabilities and quality of evidence. Interventions are ranked according to the rankings of SUCRA. The quality of evidence was classified as high, moderate, low, or very low. CrI indicates credible interval; DASH, Dietary Approaches to Stop Hypertension; MBSR, mindfulness‐based stress reduction; NA, not available; PMR, progressive muscle relaxation; SUCRA, surface under the cumulative ranking; and WMD, weighted mean difference.

In terms of lowering DBP, based on SUCRA, low‐calorie diet (WMD, 4.56 mm Hg [2.22–6.89]), tai chi (WMD, 7.76 mm Hg [95% CrI, 4.72–10.81]), DASH (WMD, 3.54 mm Hg [95% CrI, 1.80–5.28]), qigong (WMD, 6.74 mm Hg [95% CrI, 0.74–12.72]), aerobic exercise (WMD, 4.44 mm Hg [95% CrI, 3.31–5.57]), comprehensive lifestyle modification (WMD, 3.38 mm Hg [95% CrI, 1.01–5.76]), isometric training (WMD, 4.01 mm Hg [95% CrI, 1.07–6.93]), and low‐calorie diet plus exercise (WMD, 3.35 mm Hg [95% CrI, 1.41–5.32]), followed by low‐sodium and high‐potassium salt, salt restriction plus low‐calorie diet and exercise, breathing‐control, yoga, and salt restriction, were more effective than usual care (Figure 3B and Figure S4B). In addition, aerobic exercise was slightly better than salt restriction in lowering DBP (WMD, 1.82 mm Hg [95% CrI, 0.33–3.31]) (Figure S3).

The quality of evidence for interventions in comparisons with usual care are summarized in Table S4. We focused on high‐ or moderate‐quality evidence given the large number of results from the GRADE framework. In terms of lowering SBP and DBP, the quality of evidence for DASH and meditation were rated as high, and as moderate for low‐calorie diet, isometric training, aerobic exercise, comprehensive lifestyle modification, resistance training, alcohol restriction, breathing‐control and progressive muscle relaxation. There was also moderate confidence supporting the use of low‐sodium and high‐potassium salt and yoga in lowering SBP and the use of salt restriction in lowering DBP (Figure 3 and Table S4).

Network Meta‐Analysis in Patients With Hypertension (BP ≥140/90 mm Hg)

In traditional meta‐analyses of patients with hypertension, 10 interventions were more effective in lowering BP compared with usual care (Table S3).

Network meta‐analysis in patients with hypertension included 91 studies (7291 participants) with 19 interventions and usual care (Figure 2B). Comparative effect estimates were presented in Figure S5, and we focused on the effective BP‐lowering estimates of interventions that were supported by the combination evidence of direct and indirect comparisons. Ten interventions were shown to be more effective than usual care in lowering SBP and DBP (Figure 4). Tai chi (WMD, 12.75 mm Hg [95% CrI, 6.54–18.98]), DASH (WMD, 8.69 mm Hg [95% CrI, 5.23–12.19]), and low‐calorie diet (WMD, 7.78 mm Hg [95% CrI, 3.53–11.91]) were ranked first in lowering SBP; low‐calorie diet (WMD, 4.98 mm Hg [95% CrI, 2.03–7.89]) was ranked higher in lowering DBP than DASH (WMD, 4.54 mm Hg [95% CrI, 1.91–7.18]); regardless of lowering either SBP or DBP, aerobic exercise, isometric training, low‐sodium and high‐potassium salt, yoga, meditation, salt restriction, and breathing‐control followed low‐calorie diet and DASH (Figure 4 and Figure S6).

Figure 4

Forest plots for mean changes of blood pressure in patients with hypertension.

A, Systolic blood pressure. B, Diastolic blood pressure. Mean changes of blood pressure are reported in WMD and 95% CrI for intervention vs usual care. Rectangle represent the point estimate for the effect of each intervention. Horizontal lines indicate 95% CrI. For each intervention, tables on the left of the forest plot show the number of direct comparison studies, number of participants, rankings of SUCRA probabilities, and quality of evidence. Interventions are ranked according to the rankings of SUCRA. The quality of evidence was classified as high, moderate, low, and very low. CrI indicates credible interval; DASH, Dietary Approaches to Stop Hypertension; MBSR, mindfulness‐based stress reduction; NA, not available; PMR, progressive muscle relaxation; SUCRA, surface under the cumulative ranking; and WMD, weighted mean difference.

The quality of evidence for interventions in comparisons with usual care in patients with hypertension was similar to that for adults with prehypertension to established hypertension (Table S5). In addition, the quality of evidence for salt restriction was rated as high regardless of lowering either SBP or DBP (Figure 4 and Table S5).

Assessment of Heterogeneity and Inconsistency

The global I values were 74.26% and 76.70% for mean SBP change and mean DBP change, respectively, in network meta‐analysis of adults with prehypertension to established hypertension, and the global I values were 63.25% and 74.60% in the analysis of patients with hypertension. The test of inconsistency showed significant differences in only 2 comparisons (tai chi versus usual care, tai chi versus aerobic exercise) between direct and indirect results. Details of heterogeneity and consistency are given in Table S6.

Network Sensitivity, Metaregression, and Subgroup Analyses

We conducted sensitivity, metaregression, and subgroup analyses based on all existing data. Sensitivity analyses omitting studies with high risk of bias, studies started before 1999, or studies with end points of home BP or 24‐hour ambulatory BP did not significantly alter the results of overall network meta‐analysis (Table S7). Sensitivity analysis was carried out by omitting special populations along with specified interventions, and the SUCRA rankings of the other interventions did not change except that the rankings of the deleted interventions were missing (Table S7). The results of metaregression analyses showed that covariates (mean age, mean body mass index, proportion of participants taking antihypertensive medicine, and proportion of female patients) did not affect the results of this study (Table S8). The BP‐lowering effects of interventions among different subgroups defined by study duration and region of origin of study participants were not statistically different (Table S9).

Discussion

To the best of our knowledge, this study is the first to estimate the aggregate BP effects of 22 nonpharmacologic interventions through network meta‐analysis including patients with hypertension and prehypertension. In the results, which combined the SUCRA rankings and GRADE quality of evidence and overcame the lack of head‐to‐head trials, DASH ranked as the most effective intervention for lowering BP, followed by aerobic exercise, isometric training, low‐sodium and high‐potassium salt, and comprehensive lifestyle modification. Meditation and breathing‐control were considered to be relatively better among stress‐reduction interventions but were less effective than the above‐mentioned interventions. Salt restriction was also supported for lowering BP, especially in patients with hypertension.

Nonpharmacologic interventions, including dietary approaches, are a cornerstone for the prevention and treatment of hypertension. The DASH diet promotes consumption of whole grains, vegetables and fruits, lean meat, and fat‐free dairy products and the inclusion of micronutrients in the diet. These foods are also naturally low in sodium and contain nutrients, which may help lower BP. This diet can also decrease concentrations of total cholesterol and LDL (low‐density lipoprotein), which may predict a reduction of ≈13% in the 10‐year Framingham risk score for cardiovascular disease. Our report demonstrates that eating a DASH diet every day has a significant effect on lowering BP compared to usual care, which is in keeping with previous meta‐analysis. In addition, our network meta‐analysis suggests DASH to be the most effective intervention, based on its top SUCRA ranking and high‐quality evidence supporting.

The World Health Organization has proposed that a 30% reduction in salt or sodium intake may reduce the risk of hypertension. In our study, salt restriction (sodium intake <100 mmol, equivalent to 5.85 g salt) can significantly lower SBP, which is consistent with the result of a previously published meta‐analysis. Because the quality of evidence was rated down by risk of bias, inconsistency, and publication bias, there is low confidence supporting the use of salt restriction for lowering SBP in adults with prehypertension to established hypertension. However, the risk of bias was due to a study that was published in 1973 with insufficient information, and high heterogeneity (I=77.1%) was mainly produced by combining the results of participants with hypertension and prehypertension. In the analysis for only patients with hypertension, the quality of evidence for salt restriction was considered high.

For people with a long‐established habit of high salt intake, it is difficult to attain and maintain long‐term voluntary salt control, and alternative approaches with equivalent effects are needed. A salt substitute with low‐sodium and high‐potassium content and an acceptable salty flavor would be an ideal population‐wide preventative strategy. In our network meta‐analysis, moderate‐quality evidence supports the BP‐lowering effect of low‐sodium and high‐potassium salt (≈5–8 g) in adults with prehypertension to established hypertension.

An extensive body of research suggested that physical activity may have beneficial effects on BP in people with hypertension. Published meta‐analyses have also confirmed the efficacy of physical activity in lowering BP. , , In our study, moderate‐ to high‐intensity aerobic exercise (at least 3 days weekly, 30 minutes per time, achieving 60% to 90% of the maximum heart rate) and isometric training (3 days weekly, bilateral contractions at 30% of maximum voluntary contraction), followed behind DASH in lowering BP significantly.

Tai chi fared relatively better based on SUCRA rankings. However, highly ranked interventions would result in misleading inferences when most evidence is of low or very low quality. Based on the very low quality of evidence (severe inconsistency, imprecision, and publication bias), tai chi disappeared from our recommendations. However, tai chi was still among the interventions with a potentially better effectiveness profile, although the BP‐lowering effect was potentially spurious according to the analysis of existing data, and more RCTs should be conducted to evaluate this result further.

In addition to single nonpharmacologic interventions, comprehensive lifestyle modification is also effective in lowering BP and has been evaluated in several RCTs. , , , In our network meta‐analysis, it is not surprising that moderate‐quality evidence supported comprehensive lifestyle modification for lowering both SBP and DBP. However, based on SUCRA rankings, this intervention did not seem to be the most effective, possibly because different studies have different approaches to modify multiple unhealthy lifestyles, which may bring heterogeneity and affect the results. Benefits of other combined interventions (aerobic exercise plus DASH, salt restriction plus low‐calorie diet and exercise, salt restriction plus DASH, aerobic exercise plus resistance training) could not be judged because of insufficient studies and low quality of evidence until now.

This study also extends findings from previous traditional meta‐analyses that aerobic exercise seems more effective than salt restriction with respect to lowering BP, whereas the differences in comparative effect among other effective nonpharmacologic interventions mentioned above were modest (Figures S3 and S5), signaling potential equivalence of these interventions for lowering BP. Low‐calorie diet and low‐calorie diet plus exercise could lower BP level more than exercise among participants who were overweight and obese because of participants' weight loss with these 2 interventions.

Considering that the BP‐lowering effects of nonpharmacologic interventions may be affected by study duration, we performed subgroup analysis. The results showed that the BP‐lowering effects of interventions among different subgroups were not statistically different. Despite the differences that were not statistically significant, we observed that low‐sodium and high‐potassium salt and aerobic exercise lowered BP more over 12 to 24 weeks, and the BP‐lowering effects of salt restriction, low‐calorie diet, and comprehensive lifestyle modification were decreased with the extension of duration. These results may be caused by different persistence over time. For breathing‐control, DASH, and isometric training, the duration of most studies was <12 weeks. Consequently, more RCTs should be conducted to assess the long‐term effects of nonpharmacologic interventions.

Our network meta‐analysis strictly excluded studies involving adults with resistant hypertension, who are particularly salt‐sensitive and reacted differently from patients with primary hypertension in terms of salt‐related interventions. Studies involving patients with diabetes mellitus or metabolic syndrome were also excluded because these conditions might influence the effects of nonpharmacologic interventions. , The participants who used alcohol habitually or who were overweight or obese in our study were also free of diabetes mellitus and metabolic syndrome; therefore, these special participants were treated equally as patients with hypertension and prehypertension. After omitting these specified interventions along with their corresponding participants, the SUCRA rankings of the other interventions did not change except that the rankings of the deleted interventions were missing.

This analysis has several limitations. First, for the 22 interventions included in the network meta‐analysis, 8 were only directly compared with usual care. The effects of these interventions were estimated with direct evidence; however, this did not affect the evaluation and rankings of these 8 interventions because direct evidence has a higher rating than indirect evidence. In addition, many indirect comparisons were assessed as being of low or very low quality in the GRADE framework, which largely restricts the interpretation of these results. Inconsistency existed in the comparison of tai chi versus aerobic exercise; however, this did not affect the estimates of other interventions seriously. Second, our study reported only the effectiveness of nonpharmacologic interventions in lowering BP, lacking secondary end points such as rate of BP control, incidence of hypertension, and mortality due to complications of hypertension, as most RCTs included in this study provided data of mean BP or changes in BP. Third, smoking cessation as a nonpharmacologic intervention was not included in our study because existing RCTs on smoking cessation in patients with hypertension or prehypertension were not truly intervened. Music therapy was also not included because of a wide variety of music was used, and there was no comparable control group. Fourth, most RCTs included in this study had short‐ or moderate‐term follow‐up. Fifth, we only reviewed publications in English.

Conclusions

This network meta‐analysis showed that, among 22 nonpharmacologic interventions, DASH was the most effective intervention in lowering BP for adults with prehypertension to established hypertension. Aerobic exercise, isometric training, low‐sodium and high‐potassium salt, comprehensive lifestyle modification, breathing control, meditation, and low‐calorie diet also have obvious effects in lowering BP. Moreover, our findings suggest that salt restriction be used for lowering BP, especially in patients with hypertension.

Sources of Funding

Yang, Sun, and Y. Zhao received funding from the Chinese Academy of Engineering advisory project: “International comparative study on strategies for the prevention and control of a major disease (hypertension) in China's cold regions” (2015‐XZ‐32).

Disclosures

None.

Data S1

Tables S1–S9

Figures S1–S6

References ,

Click here for additional data file.