aDivisiUniversitySuzhou, CSuzhou, Creceived an

Drs. ZoSee pa*Corre

65125097.**Corr

65880929.E-mail

sxwang200

0002-9149http://dx.do

Meta-Analysis of Effects of Voluntary Slow BreathingExercises for Control of Heart Rate and Blood Pressure in

Patients With Cardiovascular Diseases

Yan Zou, MNa,b, Xin Zhao, PhDa, Yun-Ying Hou, MNb, Ting Liu, MNb, Qing Wu, MNb,Yu-Hui Huang, PhDc,**, and Xiao-Hua Wang, PhDa,*

Rising heart rate (HR) and elevated blood pressure (BP) cause a greater frequency of

on of C, Suzhohina; anhina. Md acceu andge 153spondin

espondi

addr1@163

/17/$ -i.org/1

cardiovascular events. Many patients cannot maintain target HR and BP using pharma-cological therapies. To evaluate the effectiveness of voluntary slow breathing exercises inreducing resting HR and BP, we searched Embase (1974 to April 2016), PubMed (1966 toApril 2016), the Cochrane Central Register of Controlled Trials (issue 4, April 2016), andPEDro (www.pedro.org.au; 1999 to April 2016). The primary outcome was the mean changein HR at rest. Secondary outcomes included changes in systolic blood pressure (SBP) anddiastolic blood pressure (DBP) as well as compliance with the breathing training. Finally,we included 6 studies consisting of 269 subjects. Practice of the breathing exercises resultedin statistically significant HR reduction (mean difference: L1.72 beats/min, 95% CI L2.70to L0.75). Reductions were seen in SBP (mean difference: L6.36 mm Hg, 95% CI L10.32to L2.39) and DBP (mean difference: L6.39 mm Hg, 95% CI L7.30 to L5.49) comparedwith the controls. Trial durations ranged from 2 weeks to 6 months. In conclusion,the existing evidence from randomized controlled trails demonstrates that short-termvoluntary slow breathing exercises can reduce resting HR, SBP, and DBP for patientswith cardiovascular diseases. � 2017 Elsevier Inc. All rights reserved. (Am J Cardiol2017;120:148e153)

Voluntary slow breathing exercises (VSBEs) are definedbased on a self-controlled breathing rate to achievedecreased respiratory rate and increased respiratory ampli-tude (tidal volume).1 It is an easy-operated/practical methodfor patients with cardiovascular disease. However, its effectson reductions in heart rate (HR) and blood pressure (BP) arestill controversial. Silva et al.2 found that there was nosignificant reduction in HR or systolic blood pressure (SBP)after deep breathing exercises for patients with coronaryartery disease, hypertension, and diabetes mellitus. Mean-while, a study by Dixhoorn et al3 revealed that slowbreathing was related to beneficial effects on resting HR formyocardial infarction patients. Another recent study alsodemonstrated that slow breathing training produced avaluable reduction in resting HR and SBP in hypertensivepatients.4 Thus, we performed a meta-analysis to analyze theimpact of VSBE on HR and BP for patients with cardio-vascular disease.

ardiology, The First Affiliated Hospital of Soochowu, China; bSchool of Nursing, Soochow University,d cCyrus Tang Hematology Center, Soochow University,anuscript received February 1, 2017; revised manuscriptpted March 28, 2017.Zhao are considered as first authors of this work.for disclosure information.g author: Tel: (86) 512-65221447; fax: (86) 512-

ng author: Tel: (86) 512-65880877-3507; fax: (86) 512-

ess: hyhui20126@163.com (Y.-H. Huang) or.com (X.-H. Wang).

see front matter � 2017 Elsevier Inc. All rights reserved.0.1016/j.amjcard.2017.03.247

Methods

Studies were eligible if they met the following standards:(1) studied cardiovascular disease mainly including coro-nary heart disease, hypertension, and heart failure; (2) pa-tients were aged over 18 years; (3) HR and (or) BP was anoutcome; (4) the language of the studies was English; and(5) were randomized controlled trails (RCTs) with designsinvolving VSBE and natural breathing as the interventionand control arms, respectively. Exclusion criteria were: (1)abstracts, review studies, case reports, or editorials; (2)repeated reports or low-quality studies; (3) studies notproviding enough needed data or data were unobtainablefrom original investigators; (4) subjects with other seriousdiseases or complications; (5) device-guided breathing ex-ercises; and (6) only 1 intervention.

Embase (1974 to April 2016), PubMed (1966 to April2016), the Cochrane Central Register of Controlled Trials(issue 4, April 2016), and PEDro (www.pedro.org.au; 1999to April 2016) were searched to obtain studies meeting theeligibility criteria. Terms used included “breathing exercise/slow breathing” and “heart rate/blood pressure.” We alsosearched the citations of full-text studies retrieved. Tworeviewers screened studies through the titles and abstractsindependently to confirm whether the study met the inclu-sion criteria. Data extracted independently by 2 investigatorsincluded study characteristics and intervention information.Disagreements were resolved in consultation with an arbi-trator. We contacted the investigators of original studies formissing information when necessary.

Primary outcomes included the mean change in restingHR. Secondary outcomes included changes in SBP and

www.ajconline.org

Table 1Methodological quality of included studies

1 2 3 4 5 6 7 8 9 10 11 Score Level

Dixhoorn 1998 þ þ - - þ - - - þ þ þ 5 2Silva 2014 þ þ - þ - - - þ - þ þ 5 2Jones 2015 þ þ þ þ - - - þ þ þ þ 7 1Modesti 2010 þ þ - þ þ þ - þ þ þ þ 8 1Mourya 2009 þ þ þ þ - - þ þ þ þ þ 8 1Sundaram 2012 þ þ þ - - - þ þ þ þ þ 7 1

1 ¼ eligibility criteria specified; 2 ¼ subjects randomly allocated to groups; 3 ¼ concealed allocation; 4 ¼ group similarity at baseline; 5 ¼ blinding ofsubjects; 6 ¼ blinding of therapists; 7 ¼ blinding of assessors; 8 ¼ outcome measures obtained from more than 85% of subjects; 9 ¼ intention-to-treat analysis;10 ¼ reporting of between-group statistical comparison results; 11 ¼ point measures and measures of variability reported; þ ¼ present; - ¼ no informationavailable or not done.

Figure 1. Flow chart of search results.

Miscellaneous/Slow Breathing Exercise and Meta-Analysis 149

diastolic blood pressure (DBP) as well as compliance withthe intervention. When data were shown only in graphicalform, we extracted information from the figures using theGetdata Graph Digitizer, version 2.25 (http://getdata-graph-digitizer.com/). When the trial provided only the meanand SD before and after the intervention in eachgroup and those change between baseline and finalmeasurement were missing, the SD of the change could betransformed by the formula in the Cochrane Handbook 4.2.2

(SDðCÞ ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiSDðBÞ2 þ SDðFÞ2�2� R� SDðBÞ � SDðFÞ

q;

C, change value; B, baseline value; F, final value).5 TheR values for HR, SBP, and DBP were 0.75, 0.76, and 0.76in the experimental group and 0.93, 0.8, and 0.54 in the

control group, respectively.6 Because the R value wasestimated using data from other studies, we chose sensitivityanalysis to test the stability of the result.

We used the Physiotherapy Evidence Database scale7 toevaluate the methodological quality of each included study.For assessing bias, the tool involves 11 items7 (Table 1).The total scale score is 10. The second item to the eleventhitem each count as a point. Item 1, used to evaluate theexternal reality, is not accounted for in the total score.Scores of 9 to 10 represent the best quality, scores of 6 to 8represent good quality, scores of 4 to 5 represent generalquality, and scores < 4 represent poor quality.8

The meta-analysis was performed using Review Manager5.3 software. We used it to calculate the mean difference

Table 2Clinical characteristics of the patient populations from included studies at baseline

Study Population Number Age (M�SD) years Sex (men/women)

C I C I C I

Dixhoorn 1998 Post-myocardial infarction, Netherlands 39(51.3%) 37(48.7%) _ _ 74/2Silva 2014 Myocardial arterial disease (14 diabetes mellitus),India 20(50%) 20(50%) 50-59(47.5%) 30/10Jones 2015 Hypertension, Thailand 10(50%) 10(50%) 50.4�5.4 53.4�4.3 _ _Modesti 2010 Hypertension, Italy 24(45.3%) 29(54.7%) 58 (53-61)* 58(54-62)* 16/10 16/13Mourya 2009 Stage 1 essential hypertension, India 20(50%) 20(50%) 20-60 12/8 10/10Sundaram 2012 Essential hypertension, India 20(50%) 20(50%) 52.1�4.58 53.0�5.4 12/8 14/6

C¼ control group; I¼ intervention group; e ¼ no information available.* The data are represented by the mean (95% CI).

Table 3Interventions of included studies

Study Duration Intervention Comparison Outcome

Dixhoorn 1998 3 months Exercise rehabilitation þ breathing relaxation Exercise rehabilitation Change in HR, respiration rate, and respiratorysinus arrhythmia

Silva 2014 0.5 month Perform deep breathing exercises twice a day for10 min.

Breathing naturally Change in HR, SBP, DBP, anxiety anddepression

Jones 2015 2 months Breathing with a duty cycle (IT:TRT) of 0.4 withTRT of 10 s; resting for 5 s after every 6breaths; performing at home for 30 min, 2times/day

Normal daily living Change in HR, SBP, DBP and heart ratevariability

Modesti 2010 6 months 4-6 beats/min, performing ‘abdominal’ breathingwith a 1:2 inspiration: expiration ratio for 20min every day.

Reading books Change in office, mean-24 h, daytime and night-time SBP, DBP and HR. Change in QoL

Mourya 2009 3 months Right and left nasal breathing alternately for15 min, approx. 5-6 beats/min twice daily;each nasal breath for 6 s.

Breathing naturally Change in SBP and DBP, S/L ratio, 30:15 ratio,Valsalva ratio, E/I ratio, handgrip test, coldpressor response to SBP and DBP.

Sundaram 2012 1 month Two times/week slow breathing exercises Breathing naturally Change in HR, SBP, DBP and respiratory rate,Change in 6 Minute Walk Distance

30:15 ratio ¼ immediate heart rate response to standing; E/I ratio ¼ heart rate variation with respiration; IT ¼ inspiration time; QoL ¼ quality of life;S/L ratio ¼ standing-to-lying ratio; TRT ¼ total respiratory time.

150 The American Journal of Cardiology (www.ajconline.org)

and 95% CI. Heterogeneity was evaluated by testing theclinical characteristics of the enrolled studies as well as byformal statistical testing using chi-square and I2 tests. Whereno heterogeneity was present, we performed a fixed-effectmeta-analysis. If substantial heterogeneity (I2>50%) wasdetected, we sought the direction of effect, and whereapplicable, used a random-effects analysis.

Results

Our initial search revealed a total of 2,549 records.Screening progress was shown in Figure 1. Six RCTs wereultimately included; characteristics of patient populationsfrom included studies are presented in Tables 2 and 3. Studywith a low risk of bias was defined as a study fulfilling 6 ormore of the 11 criteria, whereas a study meeting < 6 of thecriteria had a high risk of bias. The scores of the 6 RCTs forrisk of bias ranged from 5 to 8 (Table 1), indicating a lowrisk of bias.

HR results of 229 patients were reported on 5 studiesincluded.2e4,9,10 Overall, VSBE resulted in an HR decreaseof 1.72 beats/min (95% CI �2.7 to �0.75, p ¼ 0.0005) withlow heterogeneity (I2 ¼ 13%; Figure 2, 1.1.1). According tothe Nfs0.05 ¼ (SZ/1.64)2�S (Z representing the Z value of

each single study; S representing the number of all enrolledstudies) to calculate a fail-safe number (Nfs0.05),

11 the Nfs0.05

was 182. That is, that another 182 negative studies would beneeded to reverse this result, thus indicating that the result isstable.

Five studies including data from 193 patients showed theeffect of VSBE on BP.2,4,9,10,12 SBP was reduced by6.36 mm Hg (95% CI �10.32 to �2.39; p ¼ 0.002) withsignificant heterogeneity (I2 ¼ 80%; Figure 3, 2.1.1). Byreviewing the studies, we conducted a subgroup meta-analysis after excluding a study by Mourya et al. Theresult revealed that the heterogeneity (I2 ¼ 29%; Figure 3,2.1.2) and SBP (mean difference: �4.63 mm Hg, 95%CI �7.47 to �1.79) were significantly reduced. The Nfs0.05

for SBP was 23, which indicated that the result was stable.DBP was reduced by 6.39 mm Hg (95% CI �7.30 to �5.49;p <0.00001), and the heterogeneity was moderate (42%;Figure 4, 3.1.1). The Nfs0.05 regarding DBP was 183, whichindicated that the result was stable.

Data on compliance with VSBEwere reported on 3 studies.In the study by Dixhoorn3 and Mourya,12 79% and 90%compliance rates were reported, respectively. In the study byModesti et al,9 patients performed exercises 5.1 times/week (ofthe 7 requested), for 22 min/day (of 30 minutes) on average.9

Figure 2. VSBE versus control: effect on HR.

Miscellaneous/Slow Breathing Exercise and Meta-Analysis 151

There were 3 studies3,4,12 that did not include the originaldata we needed. After contacting the investigators, aninvestigator4 provided the relevant results. Therefore, weused the formula to calculate the SD of mean change. Weperformed sensitivity analysis, excluding 2 studies3,4 fromthe HR comparison and another 2 studies4,12 from the BPcomparison. The results indicated there were still significanteffects on improving resting HR, SBP, and DBP (Figure 2,1.1.2; Figure 3, 2.1.3; and Figure 4, 3.1.2).

Discussion

Based on the inclusion and exclusion standards, weselected all RCTs published in English to explore the effectof VSBE on resting HR and BP. To avoid missing thestudies of the effects of VSBE on HR and BP, we did notinclude “diseases of participants” in the search strategy.Ultimately, the diseases of the participants in the includedstudies were hypertension and coronary artery disease.

Many experimental and clinical observations have shownthat ischemic heart disease and heart failure can reducebaroreflex sensitivity, which leads to sympathetic over-activity and suppression of parasympathetic activity.13 Theorigin of hypertension is characterized by such a charac-teristic of autonomic imbalance.14 A low breathing ratethrough activating the Hering-Breuer reflex could improvebaroreflex sensitivity,14 improve cardiac vagal tone, andmodulate sympathetic overactivity, thereby decreasingresting HR and BP.15 Changes of autonomic imbalance andbaroreflex sensitivity promptly vanish after the restoration ofa normal breathing rate.16 However, recently a randomizedstudy17 has demonstrated that VSBE can induce chronicautonomic changes in the modulation of baroreflex sensi-tivity, ambulatory BP, renal resistive index, and HRvariability.

Resting HR reduction was closely related to thedecreased risk of cardiovascular events and all-cause

death for patients with coronary heart disease and hy-pertension.18,19 The risk of cardiovascular death or hos-pital admission increased by 3% with every beat increasefrom baseline HR and 16% with every 5 beats/min in-crease in patients with heart failure.20 When resting HRincreased by 5 beats/min in patients with coronary heartdisease and left ventricular dysfunction, risks increasedby 8% for cardiovascular death, 7% for myocardialinfarction, and 16% for admission for heart failure.21

In addition, data from 4,065 patients suggested that foreach beat of HR change, there was a 1% change inmortality risk for hypertensive patients.19 Our meta-analysis showed that the practice of VSBE had theoverall effect of reducing resting HR by an average of 1.7beats/min. Therefore, VSBE is a useful intervention toreduce the resting HR of patients with cardiovasculardiseases.

Uncontrolled hypertension could lead to a higher risk ofheart failure, coronary heart disease, and major cardiovas-cular disease events.22 With every 3.6 mm Hg reduction ofmean BP, the relative risks of total cardiovascular events,strokes, coronary events, cardiovascular deaths, and totaldeaths were 0.86, 0.72, 0.91, 0.75, and 0.78 times, respec-tively, that of a 2.4 mm Hg reduction. These findingsindicated that reduction of BP is important for cardiovas-cular patients. In our results, the practice of VSBE couldreduce SBP by 6.36 mm Hg and DBP by 6.39 mm Hg onaverage. Therefore, VSBE could provide beneficial effectsof BP reduction for patients with hypertension and coronaryheart disease.

Only one study tested the effect of HR and BP at6 months among the 6 included studies. Three3,9,12 studiesreported the adherence to slow breathing exercises and pa-tients in 2 studies3,12 reported good adherence to the inter-vention. In the study by Modesti et al,9 the number ofpractice sessions (7 times/week requested) decreased from5.1 times/week at baseline to 3.3 times/week at 6 months.

Figure 3. VSBE versus control: effect on SBP.

Figure 4. VSBE versus control: effect on DBP.

152 The American Journal of Cardiology (www.ajconline.org)

Thus, monitoring of long-term adherence is required toobtain accurate information on the benefits of thisintervention.

The review was limited by the methodological quality ofthe 6 studies, which were suggested have a low to moderate

risk of bias. The blinded methods of allocation concealment,blinding of patients, therapists, and evaluators were notcompletely provided by the enrolled studies. Furthermore,we did not create funnel plots to show publication biasbecause the number of included studies was limited. Further

Miscellaneous/Slow Breathing Exercise and Meta-Analysis 153

limitations are that the SD of mean change before and afterthe intervention within the groups was transformed by aformula; the extracted data were not very accurate; and thesample was small, which could influence the interpretationof the results.

Disclosures

The authors have no conflicts of interest to disclose.

1. Bernardi L, Spadacini G, Bellwon J, Hajric R, Roskamm H, Frey AW.Effect of breathing rate on oxygen saturation and exercise performancein chronic heart failure. Lancet 1998;351:1308e1311.

2. D’Silva F, Vinay H, Muninarayanappa NV. Effectiveness of deepbreathing exercise (DBE) on the heart rate variability, BP, anxiety &depression of patients with coronary artery disease. NUJHS 2014;4:35e41.

3. Dixhoorn J. Cardiorespiratory effects of breathing and relaxation in-struction in myocardial infarction patients. Biol Psychol 1998;49:123e135.

4. Jones CU, Sangthong B, Pachirat O, Jones DA. Slow breathing trainingreduces resting blood pressure and the pressure responses to exercise.Physiol Res 2015;64:673e682.

5. Alderson P, Green S, Higgins J. Cochrane reviewers’ handbook 4.2. 2.Cochrane Libr 2004;1:1e59.

6. Kaushik RM, Kaushik R, Mahajan SK, Rajesh V. Effects of mentalrelaxation and slow breathing in essential hypertension. ComplementTher Med 2006;14:120e126.

7. Moseley AM, Herbert RD, Sherrington C, Maher CG. Evidence forphysiotherapy practice: a survey of the Physiotherapy Evidence Data-base (PEDro). Aust J Physiother 2002;48:43e49.

8. Foley NC, Teasell RW, Bhogal SK, Speechley MR. Stroke rehabili-tation evidence-based review: methodology. Top Stroke Rehabil2003;10:1e7.

9. Modesti PA, Ferrari A, Bazzini C, Costanzo G, Simonetti I, Taddei S,Biggeri A, Parati G, Gensini GF, Sirigatti S. Psychological predictorsof the antihypertensive effects of music-guided slow breathing.J Hypertens 2010;28:1097e1103.

10. Sundaram B, Maniyar PJ, Varghese JP, Singh VP. Slow breathingtraining on cardiorespiratory control and exercise capacity in persons

with essential hypertension e a randomized controlled trial. Indian JPhysiother Occup Ther 2012;6:17e21.

11. Wing RR. Behavioral treatment of obesity. Its application to type IIdiabetes. Diabetes care 1993;16:193e199.

12. Mourya M, Mahajan AS, Singh NP, Jain AK. Effect of slow- and fast-breathing exercises on autonomic functions in patients with essentialhypertension. J Altern Complement Med 2009;15:711e717.

13. La Rovere MT, Gnemmi M, Vaccarini C. Baroreflex sensitivity. ItalHeart J Suppl 2001;2:472e477.

14. Joseph CN, Porta C, Casucci G, Casiraghi N, Maffeis M, Rossi M,Bernardi L. Slow breathing improves arterial baroreflex sensitivity anddecreases blood pressure in essential hypertension. Hypertension2005;46:714e718.

15. Wan R, Weigand LA, Bateman R, Griffioen K, Mendelowitz D,Mattson MP. Evidence that BDNF regulates heart rate by a mechanisminvolving increased brainstem parasympathetic neuron excitability.J Neurochem 2014;129:573e580.

16. Spicuzza L, Gabutti A, Porta C, Montano N, Bernardi L. Yoga andchemoreflex response to hypoxia and hypercapnia. Lancet 2000;356:1495e1496.

17. Modesti PA, Ferrari A, Bazzini C, Boddi M. Time sequence of auto-nomic changes induced by daily slow-breathing sessions. Clin AutonRes 2015;25:95e104.

18. Cucherat M. Quantitative relationship between resting heart rate reductionand magnitude of clinical benefits in post-myocardial infarction: a meta-regression of randomized clinical trials. Eur Heart J 2007;28:3012e3019.

19. Paul L, Hastie CE, Li WS, Harrow C, Muir S, Connell JM,Dominiczak AF, McInnes GT, Padmanabhan S. Resting heart ratepattern during follow-up and mortality in hypertensive patients. Hy-pertension 2010;55:567e574.

20. Bohm M, Swedberg K, Komajda M, Borer JS, Ford I, Dubost-Brama A, Lerebours G, Tavazzi L. Heart rate as a risk factor in chronicheart failure (SHIFT): the association between heart rate and outcomesin a randomised placebo-controlled trial. Lancet 2010;376:886e894.

21. Fox K, Ford I, Steg PG, Tendera M, Robertson M, Ferrari R. Heart rateas a prognostic risk factor in patients with coronary artery disease andleft-ventricular systolic dysfunction (BEAUTIFUL): a subgroup anal-ysis of a randomised controlled trial. Lancet 2008;372:817e821.

22. Ettehad D, Emdin CA, Kiran A, Anderson SG, Callender T,Emberson J, Chalmers J, Rodgers A, Rahimi K. Blood pressurelowering for prevention of cardiovascular disease and death: a sys-tematic review and meta-analysis. Lancet 2016;387:957e967.