Meta-analysis of the Efficacy and Safety of Repetitive Transcranial Magnetic Stimulation (rTMS) in the Treatment of Depression.

BACKGROUND: Repetitive transcranial magnetic stimulation (rTMS) is a new type of physiotherapy technology that has been widely used in the research of depression. Although many clinical trials have found that compared to the placebo interventions, rTMS has a significant effect on the improvement of depressive symptoms, the outcomes remain inconsistent due to differences in rTMS treatment frequency, parameter settings, and site for stimulation. AIMS: This study systematically evaluated the safety and efficacy of rTMS combined with antidepressants for the treatment of depression in Chinese and English randomized, double-blind and sham controlled trials and explored the possible related factors affecting the efficacy and safety.
METHODS: We used keywords "depression" and "transcranial magnetic Stimulaton" as filters to search for the Clinical Randomized Controlled Trials (RCTs) of rTMS treatments for depression both in Chinese electronic databases: Wan fang, Wellpresi, and China Knowledge Network and in English electronic databases: PubMed, Web of Science, Embase, PsycINFO, Cochrane Library (total 8 databases) up to January 5, 2017; assessed the quality of the included studies with Cochrane risk of bias assessment tool; and according to the trial groups performed statistical analysis of the efficacy and safety presented in the included studies with RevMan5.3 software.
RESULTS: A total of 9798 articles were retrieved, and finally, 29 studies were included in this study, with a total sample size of 1659, in which the sample size of the study groups was 838, and the control group sample size was 821. After Meta-analysis, we found that treatment combined rTMS with antidepressants improves depressive symptoms in patients with depression (SDM=-0.84, 95%CI=-1.19 ~ -0.48). Based on the Cochrane risk bias Assessment tool, an assessment of the bias of the included studies was conducted, one of which was assessed as having a "high risk of bias" and others as "impossible to judge". None of the included studies reported significant adverse events, and Meta-analysis showed no statistically significant differences in dropout rate between the two groups (RR=1.27, 95%CI: 0.75~2.12, Z=0.89, p=0.37).
CONCLUSION: treatment that combined rTMS with antidepressant medication for depressive symptoms has a certain therapeutic advantage versus the placebo controls, demonstrated slight side effects, and attained good acceptability, but the differences between trials remained relatively large. Clinical trials with large sample sizes are required for further exploration of the possible related factors affecting the efficacy.

1. Introduction

Depression is a clinically common form of mental illness characterized by depressive mood and / or loss of interest and accompanied by mental disease with somatic and neurophysiological symptoms.[ WHO reports that depression is one of the major risk factors for years of disability.[ It is predicted that by 2020, depression will jump from 4th place to the 2nd leading cause of global burden of disease.[ The pathogenesis of depression in not yet clear, and the treatment for depression is still mainly pharmaceutical; however, many patients treated with pharmacotherapy do achieve ideal outcomes. There still remains a significant portion of patients (20%-30%) who despite having received sufficient dosage and completed the prescribed course of treatment still do not see a total alleviation of depressive symptoms. Although new antidepressants continue to emerge, the side effects of medication therapy are still not completely avoidable.[

With the development of imagining technology, research findings show that patients with depression may have organic brain damage. This phenomenon indicates that the pathology of depression is probably related to organic brain damage. Fortunately, thanks to the introduction of a series new techniques in neural modulation, advancements have been made in the treatment of depression. Among these modulation techniques is repetitive transcranial magnetic stimulation (rTMS). Developed in the mid 1980s, the technique is a bio-stimulation that affects and changes the function of the brain. By making use of the time varying magnetic field to act on the cerebral cortex and creating an induced current in the cerebral cortex that alters the action potential of cortical neurons, rTMS is a biological stimulation that affects brain metabolism and neuronal electrical activity. Based on the mechanism of TMS, the induced pulses of current can depolarize neurons and when applied repetitively (an approach known as rTMS) can modulate cortical excitability through altering the parameters of stimulation[ to repair white brain matter or neurologic damage, thus attaining therapeutic effects.

Repetitive transcranial magnetic stimulation can be divided into high-frequency stimulation (5-20Hz) and low-frequency stimulation (≤1Hz). Depending on the frequency, the high frequencies can increase cortical excitability, and the low-frequency suppresses excitability.[ Recently, rTMS and fMRI (functional magnetic resonance image, fMRI) were combined to identify cognitive-related brain areas [ responsible for executing cognitive tasks. And with the development of technology, deep transcranial magnetic stimulation has gradually become an effective treatment for mental illness[ Repetitive transcranial magnetic stimulation has been shown to be effective for the treatment of affective disorders such as depression in many randomized controlled studies,[ but most of the sample size in these studies was relatively small. As a result, general consistent conclusions cannot be drawn across these studies.[ Clinicians and patients believe that rTMS is a way to treat depression, but there is still a need for more evidence to support the determination of optimal parameter settings for treating depression. Thus, in this study, we compare the efficacy of antidepressants combined with rTMS treatment versus sham controlled rTMS in treating patients with depression.

2. Methods

2.1 Literature screening and retrieval strategy

In this study, we used the keywords: “抑郁”(depression) and “经颅磁刺激”(TMS) to retrieve articles from the Chinese databases: Chinese National Knowledge infrastructure (CNKI), Wang Fang Data, and China Science and Technology Journal Database (CSTJ); and used the keywords: “depress*”, “transcranial magnetic stimulation”, “TMS”, “rTMS” to retrieve from the following English language databases: Embase, PubMed, the Cochrane Library, Web of Science, PsycInfo. We searched for Randomized Control Trials (RCTS) that study the efficacy and safety of rTMS in the treatment of depression, with the date of publication on or before 5 January 2017.

2.2 Inclusion and exclusion criteria

This study included the randomized sham controlled studies of the efficacy and safety of RTMS in the treatment of depression and evaluated the efficacy and safety of the combination of RTMS and antidepressants in the treatment of depression.

2.2.1 Objective of study

All subjects that participated in the study groups were classified according to one of the following psychiatric diagnostic standards: International Classification of Diseases (ICD) [, Diagnostic and Statistical Manual of Mental Disorders (DSM) [, or the third edition of the Chinese Mental Illness Diagnostic Standard (CCMD-3).[

2.2.2 Included study types

The included studies were randomized controlled trials in which the study group used rTMS intervention and the control group used rTMS sham coils or flipped stimulation coils at a certain angle to achieve the sham stimulus effect. In the outcome, the extent of improvement and side effects in the patients with depression was measured. The research program design types are as follows: ① left high frequency stimulation VS. left high frequency sham stimulation; ② right low frequency stimulation VS right low frequency sham stimulation; ③ left high frequency stimulation (combined with medication treatment) VS left high frequency pseudo-stimulation (combined with medication treatment); ④ right low-frequency stimulation (combined with medication treatment) VS right low-frequency sham stimulation (combined with medication treatment / psychotherapy).

2.2.3 Exclusion criteria

Studies with the following contents were excluded:

(1) Experimental studies using animals; (2) senile depression, postpartum depression, post-traumatic stress disorder with depression; (3) review and case report studies; (4) repeatedly published studies; (5) improvement of non-depressive symptoms, such as, changes in cortical excitability, change in cerebral hemodynamic characteristics, or cognitive functions etc. at treatment outcome as the primary outcome indicators; (6) using blank control as controlled group or studies involving electroconvulsive therapy; (7) studies with unspecified randomization methods and cross-sectional design were excluded.

2.3 Literature screening and data extraction

Two researchers used the same inclusion and exclusion criteria to screen the literature retrieved from the electronic databases. We used the following screen and extraction process: (1) Check for duplicates from the retrieved articles. (2) Titles and abstracts of the retrieved articles were separately screened by two researchers to exclude those articles unrelated to this study. (3) the full text of remaining articles was read to further screen out articles according to listed inclusion and exclusion criteria. (4) Any disagreements about whether articles shoul be included or excluded were discussed among the two researchers, in the case where no consensus could be reached, a third senior research was consulted to make the final determination (see Figure 1 for study flowchart). The included information extraction form was developed by Wei Yanyan. The two researchers extracted the research data separately, and the extracted information included categories such as study authors, year of publication, sample size, true stimulus frequency, stimulus site, stimulus intensity (% of resting motor threshold), sham stimulation mode, and treatment cycle.

Figure 1.

Literatures screening flowchart

2.4 Risk of Bias assessment

A risk of bias assessment was carried out for all RCTs included in this study according to the guidelines put forth by the Cochrane Collaboration Network. The assessment mainly includes the following seven aspects: (1) random sequence generation (selection bias); (2) allocation concealment (selection bias); (3) Blinding of the subjects and the researcher (implementation bias); (4) Blindness of measurement of outcomes (measurement bias); (5) Integrity of the results (attribution bias); (6) Selective reporting of outcomes (reporting bias); (7) Other bias. All risky information included in this study was evaluated separately by two investigators and was discussed and agreed to by a third researcher in cases of disagreement.

2.5 Outcome Measures

Primary outcome measures: Assessment of efficacy of rTMS in treating the depressive symptoms of patients with depression

The outcome measures included in this study were score assigned with 1st priority in the study: Hamilton Depression Rating Scale (HDRS) scores measured before and after the intervention, Montgomery Asberg Depression Rating Scale (MADRS) score before and after the intervention of rTMS as the second priority score, and Beck Depression Inventory (BDI) score change before and after rTMS intervention as the third priority score.

Secondary Outcome Measures: Improvement in overall function, side effects, safety, and tolerability of treatment.

To assess the improvement of overall function of patients with depression after rTMS intervention, we used mainly the scores of Brief Psychiatric Rating Scale (BPRS) and Global Assessment of Functioning (GAF) scores to calibrate the change. Safety was assessed by comparing the differences in adverse reactions between the two groups. The comparison included the general adverse reactions such as headache, nausea, and insomnia and serious adverse reactions such as epilepsy. The acceptability of rTMS treatment was compared by the dropout rate between the two groups during the treatment courses.

2.6 Statistical Analysis

Data were analyzed using the Revman 5.3 statistical software, and heterogeneity was assessed using the χ2 test. When all studies met the statistical homogeneity (p> 0.1, I <50%), we used the fixed effects model for Meta-analysis of the treatment effect and side effects; otherwise, we employed the random effects model for Meta-analysis and took the source of heterogeneity into consideration. For the combined effect analysis, we used Standardized Mean Deviation (SMD), Relative Risk (RR) and its 95% CI. The final calculated result was shown in the Forest Plot. Cochrane was used for risk assessment and funnel plot for observing publication bias. At the same time, Stata12.0 linear regression method was designated to detect funnel chart symmetry.

3. Results

3.1 Literature screening process

Using the search strategy specified in above, we retrieved from 5 English databases and 3 Chinese databases a total of 9798 related articles. Endnote Document Management Software was used for exclusion screening, and the following studies were excluded based on the following: duplicate study- 4,125 studies; articles with irrelevant research purposes- 4,849 studies; did not meet inclusion criteria- 824 studies; unknown process in grouping or without randomized sham controlled trials- 45 studies; and repeatedly published- 2 studies[ and duplicate reports from the results of 2 master’s theses.[ In addition, a study was excluded because only the lowest, highest, and median scores for the Hamilton Depression Inventory score for TMS interventions were given, leaving the mean and standard deviation unspecified as well as the side effects and dropout rate unreported.[ In the end 29 articles were included in this systematic review.[

3.2 Characteristics of included studies

All subjects included in this study were diagnosed with depression with one of the following diagnostic criteria: DSM-IV, CCMD-3, or ICD-10. Three studies were with the subjects that met the diagnostic criteria for refractory depression, and in many cases, the parameters setting in the rTMS treatment were the high-frequency stimulus applied on the left hemisphere. Four of the studies used 1 Hz of low-frequency stimulus over the right hemisphere,[ and in a 2010 article, the stimulus frequency 5 Hz and 20 Hz were utilized alternately to perform interventions,[ but to reach equilibrium with the sham controlled group, the subjects included in the sham control were also equally distributed using the frequencies of 5 Hz and 20 Hz. In Xie et al. (2015) 30% resting motor threshold was used, the intensity of the stimulation in all other studies was controlled within the range of 80%-120% of resting motor threshold. In all the included studies, the shortest treatment period was 2 weeks, and the longest was 8 weeks. Twelve studies used sham coil as a means [ to setup the sham controlled group; in the remaining studies, the coil was rotated 45, 90, or 180 degrees to achieve the effect of sham therapy, but in George et al., how the sham stimulus control was achieved was not specified.[ During the entire course of rTMS treatment, all subjects maintained the original type or dose of medication therapy or received a specific dose of medication therapy after a period of evaluation.

3.2.1 Quality of the included studies

In the literature screening process, the studies with unspecified conditions for randomized grouping or with high risk in random grouping were excluded; therefore, in quality assessment of the included studies (see figure 2), all the included studies were presented with conditions depicting the randomized grouping and were rated as “Low risk”. Five studies qualified their randomized allocation concealment,[ and the selection bias was rated as “Low risk.” 11 studies used blind methodology with their experimenters and researchers[ and performance bias was rated as “Low risk.” One study was selective in reporting their results,[ the reporting bias was rated as “High risk”. Studies with unclear information were rated as having “Unclear risk”. Figure 3 is a funnel plot that incorporates the trials studying the efficacy of the therapy that uses medication combined with rTMS in the treatment of depression. The existence of an asymmetrical trend may due to publication bias or other causes.

Figure 2.

Risk of bias assessment of 29 included studies based on Cochrane Collaboration tool

Figure 3.

Funnel plot to identify the presence of potential publication bias in 29 included studies on rTMS combined with antidepressant medication in treating depression

3.3 Treatment effect

Of the 29 included studies, the primary outcome measures were the Hamilton Depression Symptom Inventory (HAMD) score before and after the intervention with 6 studies using 21 items on the HAMD scale; 3 studies using 24 items on the HAMD scale; and the remaining studies using 17 items on the HAMD scale. The heterogeneity of the included studies was high (χ = 293.24, I = 90%); therefore, the random effects model was used for meta-analysis. The results show that efficacy of the rTMS combined with antidepressant therapy in treatment of depression is significantly higher than the sham stimulation group (SMD = -0.84, 95% CI: -1.19 ~ -0.48), and the difference was statistically significant (Z = 4.65, p< 0.01) See Figure 4. According to the GRADE score, as the main outcome measure, i.e. the improvement in symptoms of depression in rTMS interventions, the overall quality level of evidence is “moderate” as shown in Table 2.

Figure 4.

Meta-analysis forest plot showing efficacy of rTMS combined with antidepressant medication treatment versus sham control treatment in treating depression

Table 2.

GRADE quality of evidence assessment of individual outcome indicators for the efficacy of rTMS combined with antidepressant medication therapy in the treatment of depression

Outcome indicator	No. of sample cases in the included studies	heterogeneity		Model of analysis	Group effect value		Estimated value	95% Confidence interval	GRADE quality of evidence
		I2	p		Z	p
Treatment effect	1659	90%	<0.01	Random effect model	4.65	<0.01	0.84(SMD)	-1.19,-0.48	Moderate
Side effect	1353	38%	0.06	Fixed effect model	4.62	<0.01	1.96(RR)	1.47,2.61	Moderate
Drop-out rate	882	0%	0.82	Fixed effect model	0.89	0.37	1.27(RR)	0.75,2.12	Moderate

SMD: standardized mean difference; RR: relative risk;

GRADE: The Grading of Recommendations Assessment, Development and Evaluation

3.4 Subgroup analysis

According to the sites of stimulation (the left hemisphere and right hemisphere) the studies are divided into subgroups. The results of subgroup analysis were χ = 518.84, I = 96% and χ = 7.65, I = 48%. The heterogeneity results were χ = 529.07, I = 95%, p<0.01 (see Figure 5), suggesting greater heterogeneity with the left hemisphere stimulation site. According to the administered frequencies of stimulation the studies were divided into two groups: a group with high-frequency stimulation >1 Hz and a group with low-frequency stimulation ≤1Hz, and the sub-group analysis results were χ = 489.56, I = 95% and χ = 7.65, and I= 61% respectively. The combined heterogeneity results were χ = 499.37 and I= 94%, p<0.01 (see Figure 6). Subgroup analyzes were performed according to the duration of the treatment course (i.e. treatment course ≤4 weeks and> 4 weeks). The subgroup analysis results were χ = 471.26, I= 95% and χ = 9.62, I = 58% Post hoc heterogeneity resulted in χ = 502.28, I= 94%, p <0.01 (see Figure 7). Subgroup analyzes were performed over the differences between studies published in Chinese-language journals and studies published in English-language journals. The subgroup analyzes showed χ = 203.52, I = 91%, χ = 290.18, and I= 97%, respectively. The combined heterogeneity was χ = 499.37, I = 94%, p <0.01 (see Figure 8).

Figure 5.

Subgroup analysis forest plot of stimulation on the left hemisphere versus stimulation on the right hemisphere

Figure 6.

Forest plot of subgroup analysis of high frequency stimulation vs low frequency stimulation

Figure 7.

Forest plot showing Subgroup analysis of course of treatment≤4 weeks VS course of treatment>4 weeks

Figure 8.

Forest plot showing subgroup analysis of efficacy in English studies vs the efficacy in Chinese Studies

3.5 Heterogeneity Meta-regression

Given that heterogeneity may be due to the differences in the severity, age, and prescript stimulations parameters of the subjects, linear regression was used to assess the relationship between heterogeneity and baseline depression, age of participants, and stimulation parameters. Baseline HAMD scores, intensity of stimulation, frequency of stimulation, and stimulation regimens were included as factors in the regression model to assess the effect on heterogeneity. Baseline HAMD scores and regression analysis of age alone showed P values of 0.993 and 0.142, suggesting that the severity and age of patients with baseline depression were not a contributing factor to heterogeneity. Then the stimulation intensity, stimulation frequency and stimulation treatment course were included in the regression model to get the p value of 0.052, 0.536 and 0.047 respectively. The intensity and stimulation treatment course may be related factors causing heterogeneity. Among the two factors, when the course of treatment was put into the regression model, that explained 12.8% of the variation in heterogeneity.

3.6 Meta-analysis of adverse reactions

None of the included studies reported serious adverse effects. Twenty of the studies reported their subjects experienced slight discomfort including: headache, pain in the stimulation site, muscle tension, dizziness, loss of interest et cetera. Of the 690 subjects in the true stimulation treatment group, 319 reported discomfort, and 108 of 663 subjects in the sham controlled group reported discomfort. The included studies were statistically homogenous (χ = 25.60, p= 0.06, I= 38%), thus a statistical analysis using the fixed effects model was performed. The results showed that rTMS combined with antidepressants in the treatment of depression has a higher incidence rate of side effects, RR = 1.96, 95% CI: 1.47 ~ 2.61. (Figure 9)

Figure 9.

Forest plot showing side effects of rTMS combined with antidepressant medication treatment for depression

3.7 Meta-analysis of dropout rate

Twelve included studies reported participant withdrawal, and meta-analysis of the withdrawal cases data was performed. The results showed good homogeneity among the studies (χ = 6.76, p = 0.82, I= 0), and were analyzed using the fixed effects model. There were no significant differences between the two groups (27 cases in the stimulus group and 22 cases in the sham controlled group), the difference was not statistically significant (RR = 1.27, 95% CI: 0.75-2.12, Z = 0.89).

4. Discussion

4.1 Main findings

Although pharmacotherapy is still the most commonly used treatment for depression, rTMS treatment for patients with refractory depression is an available option. The results of this study show that rTMS treatment of depression has a higher incidence rate of side effects, because the included studies use selfreporting methods to collect data on side effects from the subjects and seldom use scales for quantitative assessment. Also, the side effects disappeared shortly after treatment.

Although there are many meta-analyzes on the efficacy of rTMS in the treatment of depression, most of them are confined to the English literature. The present study focused on the efficacy of rTMS versus the sham control in the treatment of depressive symptoms. Compared with the previous meta-analyses, this study has larger sample size that consists of 29 studies and a total sample size of 1659 subjects and included Chinese literature, of which 10 studies were randomized controlled trials published in Chinese, and the sample size of 571 cases in these Chinese studies accounted for a certain percentage of the total sample size. The quality of evidence of GRADE for the primary outcome measure (treatment effect) was “moderate,” and the study of rTMS in combination with drug therapy for depression requires further improvement in the quality of studies; side effects and dropout rates to show the acceptability of using rTMS to treat patients with depression.

4.2 Limitations

Although all enrolled studies employed randomized grouping and blind methods in evaluation, the study outcomes show that heterogeneity among the included studies was high. Heterogeneity was analyzed by using regression model and subgroup analysis, etc. The stimulus frequency, stimulus intensity and duration of treatment courses were set to the regression model, and the results showed that duration of the treatment course may be one of the factors causing heterogeneity. Similarly, there may be other factors, such as the subjects’ course of disease and number of stimulus train, determining heterogeneity.

4.3 Implications

Treatment combined rTMS with antidepressants pharmacotherapy is an important option for clinicians in treating depression. Especially for some refractory cases of depression, rTMS is a feasible option for consideration. However, affecting the treatment, there are many parameters, such as the intensity of the stimulus, frequency of the stimulus train, the site for stimulation, or even the course of treatment. Testing and optimizing these parameters settings and as much as exploring the maintenance effect of rTMS after treatment still depends on the yet to come representative randomized clinical trials.

Table 1.

Basic information of the included studies

No	study	Diagnostic criteria	N(M/F)	Age(M±SD)	Site for stimulation	Frequency	Magnitude (%MT)	Course of therapy (week)	Sham stimulation	Combined with medication (Y/N)
			rTMS group	Sham stimulation group
1	George 1997	DSM-IV	7(1/6) 42.4(15.47)	5(0/5) 41.0(8.28)	Left DLPFC	20 Hz	90	4	45o	Y
2	Klein 1999	DSM-IV	36(7/29) 60.5(15.1)	34(10/24) 58.9(18.3)	Right prefrontal area	1 Hz	110	2	45o	Y
3	Berman 2000	DSM-IV	10(8/2) 45.2(9.54)	10(6/4) 39.4(10.81)	Left DLPFC	20 Hz	80	2	45o	Y
4	George 2000		20(7/13) 42.2(10.8)	10(4/6) 48.5(8)	Left prefrontal cortex	5/20 Hz	100	2	45o	Y
5	Garcia 2001	DSM-IV	11(5/6) 43.2(13.1)	11(5/6) 45.0(18.3)	Left DLPFC	20 Hz	90	2	90o	Y
6	Kauffmann 2004	DSM-IV	5(NA) (NA)	7(NA) (NA)	5cm anterior to the Right Motor Cortex	1 Hz	110	2	45o	Y
7	Rumi 2005	DSM-IV	22(3/19) 39.3(12.8)	24(4/20) 38.9(8.8)	Left DLPFC	5 Hz	120	4	Sham coil	Y
8	Avery 2006	DSM-IV	35(14/21) 26.2(12.3)	33(17/16) 25.4(11.7)	Left DLPFC	10 Hz	110	3	90o	Y
9	Januel 2006	DSM-IV	11(2/9) 38.64(11.16)	16(4/12) 37.19(11.67)	Right DLPFC	1Hz	90	4	Sham coil	Y
10	Loo 2007	DSM-IV	19(11/8) 45.7(15.0)	19(9/10) 49.8(2.5)	Left DLPFC	10Hz	110	6	Sham coil	Y
11	Reardon 2007	DSM-IV	155(69/86) 47.9(11.0)	146(72/74) 48.7(10.6)	Left DLPFC	10 Hz	120	6	Sham coil	Y
12	Mogg 2008	DSM-IV	29(13/16) 55(18.0)	30(9/21) 52(15.5)	Left DLPFC	10 Hz	110	2	Sham coil	Y
13	Schutter 2009	DSM-IV	17(7/10) 44.4(11.8)	17(10/7) 43.8(12.5)	Right parietal cortex	2 Hz	90	2	Sham coil	Y
14	George 2010	DSM-IV	92(34/58) 47.7(10.6)	98(36/62) 46.5(12.3)	Left prefrontal cortex	10 Hz	120	2	NA	Y
15	Lingeswaran 2011	DSM-IV	9(3/6) 34(10.5)	14(6/8) 37.2(11.8)	Left DLPFC	10 Hz	100	2	90o	Y
16	Ray 2011	ICD-10	20(15/5) 36.75(12.27)	20(17/3) 31.25(9.28)	Left DLPFC	10 Hz	90	2	45o	Y
17	Huang 2012	DSM-IV	28(9/19) 32.77(7.28)	28(8/20) 31.35(7.39)	Left DLPFC	10 Hz	90	2	90o	Y
18	XIE 2015	ICD-10	35(12/23) 65.3(5.1)	26(8/18) 64.7(4.2)	Left DLPFC	10Hz	30	4	Mock-coil	Y
19	Zhang 2011	DSM-IV	14(11/3) 50.8(13.3)	14(9/5) 43.8(13.9)	Left DLPFC	10 Hz	110	4	180o	Y
20	Wang 2012	CCMD-3	20(15/5) 34.85(13.71)	20(14/6) 36.75(16.70)	Left DLPFC	15 Hz	110	4	180o	Y
21	Li 2013	CCMD-3	15(9/6) NR	15(8/7) NR	Left DLPFC	10 Hz	100	4	90o	Y
22	Wang 2013	DSM-IV	30(14/16) 37.68(8.13)	29(13/16) 38.13(7.79)	right DLPFC	1 Hz	100	4	90o	Y
23	Fang 2014	DSM-IV	24(9/15) 41.63(11.02)	24(10/14) 44.58(12.36)	left DLPFC	10 Hz	80	2	Sham coil	Y
24	Yuan 2014	DSM-IV	30(9/21) 34.81±9.74	30(11/19) 36.76±17.79	left DLPFC	20 Hz	110	6	Sham coil	Y
25	Xu 2014	CCMD-3	30(16/14) 35.4(8.6)	30(15/15) 36.2(8.3)	left DLPFC	10 Hz	80	6	90o	Y
26	Hu 2015	CCMD-3	35(20/15) 36.0(7.2)	35(19/16) 35.6(7.5)	left DLPFC	1-20 Hz	80-110	4	Sham coil	Y
27	Shi 2015	ICD-10	42(19/23) NR	42(21/21) NR	left DLPFC	10 Hz	100	4	90o	Y
28	Xiao 2015	ICD-10	30(12/18) 31.6(10.2)	30(11/19) 32.9(14.2)	left DLPFC	10 Hz	80	4	Sham coil	Y
29	Liang 2016	DSM-IV	30(15/15) 36.60(5.75)	30(13/17) 36.45(5.71)	left DLPFC	10 Hz	NA	8	Sham coil	Y

Remarks: N: number of subjects included in a study; M: Mean; SD:Standard deviation; DLPFC: Dorsolateral Prefrontal Cortex; MT: Motor Threshold; Y: Yes; N: No; NA: Not Applied