Evidence from Indian studies on safety and efficacy of therapeutic transcranial magnetic stimulation across neuropsychiatric disorders- A systematic review and meta-analysis

ABSTRACT

Repetitive transcranial magnetic stimulation (rTMS) is potentially effective as an augmentation strategy in the treatment of many neuropsychiatric conditions. Several Indian studies have been conducted in this regard. We aimed to quantitatively synthesize evidence from Indian studies assessing efficacy and safety of rTMS across broad range of neuropsychiatric conditions. Fifty two studies- both randomized controlled and non-controlled studies were included for a series of random-effects meta-analyses. Pre-post intervention effects of rTMS efficacy were estimated in “active only” rTMS treatment arms/groups and “active vs sham” (sham-controlled) studies using pooled Standardized Mean Differences (SMDs). The outcomes were ‘any depression’, depression in unipolar/bipolar depressive disorder, depression in obsessive compulsive disorder (OCD), depression in schizophrenia, schizophrenia symptoms (positive, negative, total psychopathology, auditory hallucinations and cognitive deficits), obsessive compulsive symptoms of OCD, mania, craving/compulsion in substance use disorders (SUDs) and migraine (headache severity and frequency). Frequencies and odds ratios (OR) for adverse events were calculated. Methodological quality of included studies, publication bias and sensitivity assessment for each meta-analyses was conducted. Meta-analyses of “active only” studies suggested a significant effect of rTMS for all outcomes, with moderate to large effect sizes, at both end of treatment as well as at follow-up. However, except for migraine (headache severity and frequency) with large effect sizes at end of treatment only and craving in alcohol dependence where moderate effect size at follow-up only, rTMS was not found to be effective for any outcome in the series of “active vs sham” meta-analyses. Significant heterogeneity was seen. Serious adverse events were rare. Publication bias was common and the sham controlled positive results lost significance in sensitivity analysis. We conclude that rTMS is safe and shows positive results in ‘only active’ treatment groups for all the studied neuropsychiatric conditions. However, the sham-controlled evidence for efficacy is negative from India.

Conclusion

rTMS is safe and shows positive results in “only active” treatment groups for all the studied neuropsychiatric conditions. However, the sham-controlled evidence for efficacy is negative from India.

INTRODUCTION

Almost one in seven Indians is affected by mental health morbidity of varying severity.[1] Psychiatric contribution to the total disease burden has almost doubled in the last two decades.[1] WHO has estimated that the burden of mental health problems in India is 2443 disability-adjusted life years (DALYs) per 100 00 population.[2]

Treatment resistance affects up to 60% of patients with psychiatric disorders and is associated with increased healthcare burden and costs up to ten-fold higher relative to patients.[3] Moreover, psychosocial interventions may not be readily available, scalable, and may not be beneficial in our Indian context.[4] Consequently, there has been a pressing need to explore novel therapeutic interventions. Non-invasive brain stimulation (NIBS) like transcranial magnetic stimulation (TMS) has been proposed as a promising intervention strategy for neuropsychiatric disorders.[5] TMS has immediate effects on neural excitability (also after effects) which makes it a potentially apposite therapeutic tool for mental disorders.[5,6]

Since 2008, the US Food and Drug Administration (FDA) has approved many repetitive transcranial magnetic stimulation (rTMS) equipment for indications such as MDD, OCD, migraine, and smoking cessation, exclusively based on Western data.[7] Moreover, recently appraised guidelines on the therapeutic efficacy of rTMS in psychiatric disorders have been put forth.[8] The Indian Psychiatric Society guidelines in this regard have also been framed.[9] Also, there is an exponential growth in the research on rTMS from India over the last 25 years, and many open-label and randomized controlled trials (RCTs) of TMS have been conducted in India across a large number of mental disorders.[10] However, the evidence from these studies has not been synthesized. Therefore, we performed a systematic review and meta-analysis of RCTs and non-controlled studies of rTMS across a broad range of neuropsychiatric conditions.

METHODS

This study was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines.

Eligibility criteria

Studies were included if they met the following criteria: (1) randomized, sham-controlled trials or non-controlled/open label trials or quasi-experimental studies or retrospective studies assessing the effect of rTMS (2) including children, adolescents and/or adults (3) diagnosis of a mental health condition using standardized diagnostic criteria or standardized diagnostic tools and (4) use of standardized scales assessing core symptoms or task performance. Besides the core psychiatric diagnoses, studies on migraine, fibromyalgia and chronic pain disorders, essential tremors, etc., were also intended to be included.

As our meta-analysis was intended to be an inclusive one, we included studies where another adjunctive intervention such as cognitive behavioral therapy or cognitive training was also delivered along with rTMS. Studies were excluded if there were less than five patients in the active rTMS treatment arm/group.

Search strategy and selection of studies

We did a scoping search of the literature using PubMed database before conducting a comprehensive literature search to know extant of existing literature in the context of the study objective. A scoping search revealed that meta-analysis with the same aim as ours has not been conducted.

For the purpose of meta-analysis, all types of articles published till July 15, 2022, were searched in the following databases: PubMed, Web of Science, and Directory of Open Access Journals (DOAJ). Clinical Trials Registry India was also searched as an attempt to include any possible unpublished data. The key words used for the search strategy were (TMS OR rTMS OR iTBS OR Transcranial magnetic stimulation OR repetitive transcranial magnetic stimulation OR Theta burst stimulation AND India). The filters of “Humans” and “Country” were used in PubMed and Web of Science, respectively, to refine the search results. Two authors (SKT and SMG) independently carried out the search in different databases. The initial screening was carried out by going through the titles and abstracts of the search results using Rayyan web. Cross x was also looked for to complete the search process.

After exclusion of 127 duplicate articles, we extracted 435 articles, which were further assessed for abstract screening and inclusion in the review. About 43 were review articles or meta-analyses, 14 were background articles, 74 articles neither addressed TMS nor psychiatry diagnosis or outcome, 46 articles had different outcome that addressed TMS but outcome was not measurement of symptoms related to psychiatry, 42 articles had different population group which did not have a psychiatry diagnosis, 16 articles had different intervention, i.e. study population with psychiatry diagnosis but not TMS as intervention (13 tDCS, 2 DBS, and 1 Yoga), 67 articles that were only trial registrations (13 protocol registrations with data published as separate articles, 14 trials are ongoing/data unpublished and inaccessible full text, 40 trials had either unrelated outcome/diagnosis or intervention), 3 were not Indian studies, and 1 was guidelines. The resultant 129 studies were further screened for eligibility and full text, of which 77 articles were excluded, with reasons mentioned in PRISMA flowchart [Figure 1], finally allowing us with 52 articles to include in the analysis.

Figure 1.

PRISMA flowchart showing the inclusion and exclusion of articles in the analysis

Outcome measures

While preintervention and post-intervention scores of core as well as secondary symptoms (such as depression in schizophrenia/OCD) symptoms in each disorder were the primary outcome for the meta-analysis of rTMS treatment arms from all the studies, change in core symptoms in each disorder was the primary outcome for the meta-analysis of sham-controlled rTMS trials.

Data extraction

Means, standard deviations, and sample size of each treatment arm were extracted from three time points, wherever available—preintervention and two post-intervention endpoints (end of treatment and one follow-up (2 weeks to 3 months, whichever is later)) from active treatment arms/groups and sham treatment arms. Adverse effects reported in each of the 52 studies were jotted. Also rTMS intervention parameters, i.e. stimulation type (high frequency (HF, ≥ Hz), low frequency (LF, ≤1 Hz), iTBS or cTBS), stimulus frequency, stimulation target, TMS equipment making, coil type, target location identification method, intensity, number pulses in each session, number of sessions, duration of intervention, and method of sham stimulation were reviewed.

As four studies either did not match with the corresponding studies in that diagnostic subgroup or did not report means and standard deviations, they were not included for the estimation of pooled SMD. However, these studies were retained for the assessment of safety of rTMS, and all other relevant information was noted.

Risk of bias assessment

Risk of bias was independently assessed by two investigators (SMG and SKT). The included studies were assessed for risk of bias using appropriate tools based on the type of study design. For the purpose of assessment, the studies were classified as randomized control studies and non-randomized studies. The Cochrane risk of bias assessment tool (ROB) was used for the randomized control studies, and a graph as well as summary of authors’ judgment of each included study was synthesized using RevMan 5.4.1 (The Nordic Cochrane Centre, The Cochrane Collaboration, Copenhagen). The risk of bias was done in the domains of randomization, allocation, blinding, outcome assessment, and reporting which were summarized and represented in a graph. The Newcastle–Ottawa Scale[11] was used for assessing the quality of non-controlled studies. As a part of risk of bias assessment, the CTRI website was also searched for the protocols of the included studies to ensure whether the all the specified study outcomes were reported consistently or not.

Statistical analysis

All analyses were conducted using the Review Manager (RevMan), Version 5.4.1, when two or more eligible studies on the same outcome were available. Data were grouped by disorder (depression, schizophrenia, obsessive compulsive disorder, etc.) and outcome (symptoms/symptom domains) and pooled using random effects models based on standardized mean difference (SMD). For sham-controlled studies, the change in mean was estimated as pre-mean minus post-mean; the change in standard deviation (SD) was estimated using the formula:

√((SD_{preintervention})²+ (SD_endpoint)² – 2* r * SD_{preintervention} * SD_endpoint), we considered “r” (correlation coefficient) =0.4, as a conservative estimate.

Heterogeneity was assessed using the I² statistic, which specifically estimates the proportion of total variability due to between-study heterogeneity. To examine sources of heterogeneity in symptom outcomes among TMS trials, subgroup analyses were conducted based on stimulation type and stimulation target. Only two such subgroup analyses (LF-Left dorsolateral prefrontal cortex (DLPFC) for unipolar/bipolar depression and LF-supplementary motor area (SMA) for obsessive compulsive disorder (OCD)) were conducted as only they were deemed to have sufficient studies for the subanalysis. SMD values of 0.2–0.5 were considered small, values of 0.5–0.8 were medium, and values >0.8 were considered large.

For assessment of adverse events, all the reported events were tabulated. Frequencies with reference to the total sample size and odds ratios (OR) with reference to the event occurrence in the sham control group were calculated, for each of the adverse effect. The modified Haldane–Anscombe correction was used for calculation of OR when value of any of cells was zero.

Sensitivity analysis and publication bias

Sensitivity analyses were conducted excluding studies rated as “high” on the risk of bias assessment and “low” on quality scoring using the Newcastle–Ottawa Scale. Publication bias was assessed visually via funnel plots.

RESULTS

Study characteristics

A total of 52 articles, which reported 52 studies, were selected for analysis. See Table 1 and Supplementary material-I for details of study characteristics. Majority of the studies included schizophrenia (n = 13) followed by OCD (n = 11), depressive disorders (n = 10), migraine (n = 7), substance use disorders (n = 5), mania (n = 2), panic disorder (n = 1), fibromyalgia (n = 1), chronic tension-type headache (CTTH, n = 1), and essential tremors (n = 1). Except studies on migraine, which gave a washout period before rTMS treatment and used rTMS as monotherapy for prophylaxis of migraine, all other studies used rTMS as an augment along with the ongoing pharmacological or pharmacological + psychotherapeutic/cognitive treatments.

Table 1.

Study characteristics

Study	Condition	Sham (S)/Priming (P)/Active (A)/TAU (T)/Non (N)-controlled	Sample size		Stimulation type	Target
			Active rTMS arms (n=56)	Sham/No rTMS control (n=31; sham=29)
Goyal et al. 2007[12]	Schizophrenia	S	5	5	HF (10 Hz)	LDLPFC
Bagati et al. 2009[13]	Schizophrenia	S	20	20	LF (1 Hz)	LTPC
Praharaj et al. 2009[14]	Mania	S	21	20	HF (20 Hz)	RDLPFC
Mishra et al. 2010[15]	ADS	S	30	15	HF (10 Hz)	RDLPFC
Sarkhel et al. 2010[16]	OCD	S	21	21	HF (10 Hz)	RDLPFC
Jhanwar et al. 2011[17]	Depression	N	19	-	HF (10 Hz)	LDLPFC
Nongpiur et al. 2011[18]	Depression	P	18	-	LF (1 Hz) + Theta frequency (5-7 Hz) priming	RDLPFC
			19	-	LF (1 Hz) + Sham priming	RDLPFC
Lingeswaran 2011[19]	Depression	S	9	14	HF (10 Hz)	LDLPFC
Kumar et al. 2011[20]	OCD	N	12	-	LF (1 Hz)	SMA
Ray et al. 2011[21]	Depression	S	20	20	HF (10 Hz)	LDLPFC
Misra et al. 2012[22]	Migraine	N	51	-	HF (8 Hz)	LMC
Misra et al. 2013[23]	Migraine	S	50	50	HF (10 Hz)	LMC
Mishra et al. 2015[24]	ADS	A (Right vs Left)	10	-	HF (10 Hz)	RDLPFC
			10	-	HF (10 Hz)	LDLPFC
Pathak et al. 2015[25]	Mania	S	13	13	HF (20 Hz)	RDLPFC
Ray et al. 2015[26]	Schizophrenia	P	20	-	LF (1 Hz) + HF (6 Hz) priming	LTPC
			20	-	LF (1 Hz) + Sham priming	LTPC
Garg et al. 2016[27]	Schizophrenia	S	20	20	High (theta) frequency (5-7 Hz)	Cerebellar vermis
Jha et al. 2016[28]	Depression	N	20	-	HF (20 Hz)	LDLPFC
Kalita et al. 2016[29]	Migraine	A (3 vs. 1 sessions)	39	-	HF (10 Hz)	LMC
			37	-	HF (10 Hz)	LMC
Tikka et al. 2017[30]	Schizophrenia	S	8	7	cTBS (50 Hz)	RIPL
Arumugham et al. 2018[31]	OCD	S	19	17	LF (1 Hz)	SMA
Kumar et al. 2018a[32]	OCD	N	25	-	LF (1 Hz)	LOFC
Kumar et al. 2018b[33]	Panic disorder	N	13	-	HF (20 Hz)	LDLPFC
Kumar et al. 2018c[34]	Migraine	N	14	-	HF (20 Hz)	LDLPFC
Verma et al. 2018[35]	Depression	N	22	-	HF (15 Hz)	LDLPFC
Mattoo et al. 2019[36]	CTTH	S	15	15	LF (1 Hz)	RDLPFC
Sahu et al. 2019[37]	Migraine	S	20	21	iTBS (50 Hz)	LDLPFC
Singh et al. 2019[38]	OCD	N	79	-	LF (1 Hz)	SMA/LOFC
Baliga et al. 2020[39]	Depression	N	83	-	iTBS+cTBS (50 Hz)	LDLPFC + RDLPFC
Kumar et al. 2020[40]	Schizophrenia	S	50	50	HF (20 Hz)	LDLPFC
Raikwar et al. 2020[41]	ADS	S	30	30	HF (10 Hz)	LDLPFC
Sharma et al. 2020[42]	Post-stroke depression	S	47	49	LF (1 Hz)	Contralesional PFC
Singh et al. 2020[43]	Schizophrenia	S	15	15	HF (20 Hz)	LDLPFC
Tanwar et al. 2020[44]	Fibromyalgia	S	45	41	LF (1 Hz)	RDLPFC
Agrawal et al. 2021[45]	OCD	N	5	-	LF (1 Hz)	SMA
Basavaraju et al. 2021[46]	Schizophrenia	S	30	30	iTBS (50 Hz)	Cerebellar vermis
Dutta et al. 2021[47]	OCD	S	18	15	iTBS (50 Hz)	LOFC
Chauhan et al. 2021[48]	Schizophrenia	S	19	17	iTBS (50 Hz)	Cerebellar vermis
Gupta et al. 2021a[49]	Schizophrenia	T	20	19	LF (1 Hz)	LTPC
Gupta et al. 2021b[50]	ADS	T	50	50	iTBS+cTBS (50 Hz)	RDLPFC + LDLPFC
Kalita et al. 2021[51]	Migraine	A (rTMS alone vs rTMS with amitriptyline)	37	-	HF (10 Hz)	LMC
			39	-	HF (10 Hz)	LMC
Kumar et al. 2021[52]	Migraine	S	10	10	HF (10Hz)	LMC
Shere et al. 2021[53]	Depression	N	23	-	iTBS+cTBS (50 Hz)	LDLPFC + RDLPFC
Syed et al. 2021[54]	OCD	N	12	-	iTBS (50 Hz)	BL dmPFC
Tyagi et al. 2021[55]	Schizophrenia	S	30	29	cTBS (50 Hz)	BLTPC
Ankit et al. 2022[56]	ODS	S	20	20	cTBS (50 Hz)	ROFC
Batra et al. 2022[57]	Essential tremors	N	10	-	cTBS (50 Hz)	LMC
Joshi et al. 2022[58]	OCD	S	13	11	LF (1 Hz)	SMA
Mallik et al. 2022[59]	Bipolar depression	S	11	8	cTBS (50 Hz)	RDLPFC
Reddy et al. 2022[60]	OCD	N	15	-	HF (20 Hz)	ACC + MPFC
Shah et al. 2022[61]	Migraine	S	36	36	LF (1 Hz)	LDLPFC
			36		HF (8 Hz)	LDLPFC
Vidya et al. 2022[62]	OCD	P	15	-	LF (1 Hz) + HF (6 Hz) priming	SMA
			15		LF (1 Hz) + Sham priming	SMA
Thirthalli et al. 2022[63]	Schizophrenia	S	15	13	HF (20 Hz)	RDLPFC + LDLPFC
Total			1448	701

rTMS=Repetitive transcranial magnetic stimulation; OCD=Obsessive compulsive disorder; ADS=Alcohol dependence syndrome; CTTH=Chronic tension-type headache; HF=High frequency; LF=Low frequency; cTBS=Continuous theta burst stimulation; iTBS=Intermittent theta burst stimulation; LDLPFC=Left dorsolateral prefrontal cortex; RDLPFC=Right dorsolateral prefrontal cortex; LTPC=Left temporoparietal cortex; SMA=Supplementary motor area; LMC=Left motor cortex; RIPL=Right inferior parietal lobule; LOFC=Left orbitofrontal cortex; ROFC=Right orbitofrontal cortex; PFC=Prefrontal cortex; dmPFC=Dorsomedial prefrontal cortex; ACC=Anterior cingulate cortex; MPFC=Medial prefrontal cortex

While 37 of the 52 studies were RCTs and 15 were non-controlled studies that included rTMS as an intervention in one or either of the arms. Out of the 37 RCTs, 32 studies were sham-controlled studies. Three studies had active control arms with rTMS being the intervention in both the experimental and the control arms. One study had both active and a sham-controlled arm. Of the 32 sham-controlled studies, 3 studies investigated the effect of “priming,” and the sham group received sham priming along with active rTMS stimulation.

While 39 studies used the conventional high- or low-frequency stimulation, 13 studies investigated the effects of theta burst stimulation (TBS). Studies on migraine stimulated the motor cortex. While the majority of the studies used the Magstim make (n = 29), Magventure (n = 10), Neurosoft (n = 2), EBNeuro (n = 1), and Medicaid (n = 1) were the other equipment makes used. Other studies did not report the rTMS equipment make. While majority of the studies used the conventional figure-of-eight coil, two studies used the H coil and one study used the double-cone coil for deeper stimulation. In the conventional target, dorsolateral prefrontal cortex (DLPFC, n = 26) was stimulated by the majority of studies, supplementary motor area (SMA, n = 6), orbitofrontal cortex (OFC, n = 4), temporo-parietal cortex (TPC, n = 3), cerebellar vermis (n = 3), anterior cingulate cortex + medial prefrontal cortex (n = 1), and dorsomedial prefrontal cortex (n = 1) too were investigated as the target for stimulation. Three studies used bilateral stimulation. Twenty studies used the conventional “5cm” rule for localizing the target location for stimulation. While 16 studies used the international 10-20 system of EEG electrode placement method, magnetic resonance imaging (MRI) and single-photon emission computed tomography (SPECT)-based neuronavigation was used by four and one studies, respectively. The intensity of stimulation ranged between 70 and 120% of resting motor threshold (RMT). Number of pulses delivered in a single session ranged between 500 and 3000, commonest being 1200 (n = 12). While the majority of studies used once daily sessions, 3 studies used the intensive—2 session/day protocol. For sham stimulation, 14 studies used sham coils and 13 studies employed the angling method (angle range: 45-90°).

Risk of bias

While 14 of the 15 non-controlled studies were deemed to be of “moderate” quality, one study was deemed to be of “low” quality. Eleven (29.7%) and five (13.5%) out of 37 RCTs were deemed to be at “high” and “unclear” risk of bias. See Supplementary Material-II, Supplementary Table II and Figure 1 for details.

Meta-analysis- Efficacy

Twenty-nine studies were deemed “true sham”- controlled studies, i.e. without any active rTMS stimulation in the sham group and were included for the “active vs. sham” meta-analysis that reported sham-controlled effects. Treatment arms of all other studies were considered for the pre–post-intervention effects of active rTMS stimulation. Four studies[28,36,44,50] out of 52 could not be included for meta-analysis of efficacy as they did not report means and standard deviations.

For assessment of effect on depression, we derived data not only from unipolar or bipolar depressive disorder groups, but also from secondary/comorbid depression present with OCD, schizophrenia, migraine, and panic disorder; one study assessed post-stroke depression. For schizophrenia, positive symptoms, negative symptoms, total psychopathology, auditory hallucinations, and overall cognitive functions were analyzed as separate outcomes. For studies on substance use disorders, craving/compulsion was the outcome chosen.

The intervention efficacy meta-analyses were broadly divided into “active only,” which included only the active intervention treatment arms/groups, and “active versus sham.” For each of these two broad divisions, two subdivisions are made for- i) preintervention to end of treatment and ii) preintervention to follow-up. Follow-up data was not available for all studies that were included. Meta-analysis for preintervention to follow-up could only be performed for some studies [Supplementary Material-III].

Active only/preintervention to end of treatment—Depression

Active rTMS was found to be significantly effective as an augment in the treatment of depression [Table 2]. Improvements in “any depression,” depression in unipolar or bipolar depressive disorders and depression associated with schizophrenia had large effect sizes. Improvement in depression associated with OCD had a moderate effect size. These meta-analyses showed significant heterogeneity among the included study results.

Table 2.

Meta-analysis (active treatment arms) preintervention to end of treatment Depression (21 treatment arms)

Study	Effect
	SMD	Confidence interval	n	Weighted % any (Sub)
Any Depression	Pooled SMD 1.57 (1.19-1.95); K=22; n=487;Z=8.07; P<.001; Heterogeneity I2=84%; P<.001
Depression (Unipolar + Bipolar)	Pooled SMD 2.28 (1.67-2.89); K=10; n=225; Z=7.36; P<.001; Heterogeneity I2=81%; P<.001
Depression (HF LDLPFC)	Pooled SMD 2.41 (1.45-3.36); K=4; n=70; Z=4.96; P<.001; Heterogeneity I2=76%; P<.01
Jhanwar et al. 2011	2.02	1.22-2.81	19	4.6 (10.7)
Lingeswaran 2011	2.22	0.98-3.45	9	3.6 (8.5)
Nongpiur et al. 2011 (with priming)	3.73	2.60-4.85	18	3.8 (9.1)
Nongpiur et al. 2011 (without priming)	2.89	1.95-3.82	19	4.3 (10.0)
Ray et al. 2011	4.48	3.27-5.68	20	3.6 (8.7)
Verma et al. 2018 (MADRS)	1.24	0.14-2.34	8	4.2 (9.2)
Verma et al. 2018 (HAMD)	2.24	1.26-3.21	14	3.9 (9.8)
Baliga et al. 2020	1.46	1.12-1.80	83	5.5 (12.6)
Shere et al. 2021	2.34	1.59-3.09	24	4.7 (11.0)
Mallik et al. 2022 (bipolar depression)	0.83	-0.05-1.71	11	4.4 (10.3)
Depression comorbid with OCD	Pooled SMD 0.73 (0.25-1.21); K=5; n=88; Z=3.00; P=<.01; Heterogeneity I2=57%; P=0.05
Sarkhel et al. 2010	1.40	0.72-2.09	21	4.8 (20.4)
Arumugham et al. 2018	0.08	-0.55-0.72	19	4.9 (21.6)
Dutta et al. 2021	0.43	-0.24-1.09	18	4.9 (20.9)
Vidya et al. 2022 (with priming)	1.10	0.32-1.87	15	4.6 (18.2)
Vidya et al. 2022 (without priming)	0.73	-0.01-1.47	15	4.7 (18.2)
Depression comorbid with Schizophrenia	Pooled SMD 0.88 (0.20-1.55); K=4; n=100; Z=2.55; P<.05; Heterogeneity I2=79%; P<.01
Garg et al. 2016	0.56	-0.07-1.19	20	4.9 (24.8)
Basavaraju et al. 2020	1.91	1.29-2.52	30	5.0 (25.1)
Singh et al. 2020	0.39	-0.33-1.12	15	4.7 (23.3)
Tyagi et al. 2022	0.62	0.10-1.14	30	5.2 (26.8)
Depression comorbid with Panic disorder
Kumar et al. 2018b	2.02	1.05-3.00	13	4.2
Depression comorbid with Panic disorder
Kumar et al. 2018c	1.31	0.49-2.14	14	4.5
Post-stroke depression
Sharma et al. 2020	2.10	1.60-2.61	47	5.2

OCD=Obsessive compulsive disorder; HF=High frequency; LDLPFC=Left dorsolateral prefrontal cortex; SMD=Standardized mean difference; K=Number of studies; n=Number of samples; MADRS=Montgomery and Asberg Depression Rating Scale; HAMD=Hamilton Depression Rating Scale

A subgroup analysis also showed that HF rTMS targeting the left DLPFC was significantly (large effect size) effective for reducing depression scores in unipolar/bipolar depressive disorders. However, heterogeneity was significantly large.

For studies including depression comorbid with panic disorder, migraine, and post-stroke depression, meta-analyses could not be conducted as there was only one study each. The evidence from these single studies was positive and showed large effect sizes, too.

Active only/preintervention to end of treatment—Schizophrenia

Active rTMS was found to be significantly effective as an augment in the treatment of schizophrenia symptoms [Table 3]. Improvements in positive symptoms, negative symptoms, total psychopathology and auditory hallucinations in schizophrenia were all significant with large effect sizes. These meta-analyses showed significant heterogeneity. Improvement in the overall cognitive function was significant with a moderate effect size and without heterogeneity.

Table 3.

Meta-analysis (active treatment arms) preintervention to end of treatment Schizophrenia (12 treatment arms)

Study	Effect
	SMD	Confidence interval	n	Weighted %
Positive Symptoms	Pooled SMD 0.97 (0.55-1.39); K=9; n=202; Z=4.56; P<.001; Heterogeneity I2=73%; P<.001
Positive Symptoms (Primary outcome)	Pooled SMD 1.13 (0.59-1.67); K=7; n=137; Z=4.13; P<.001; Heterogeneity I2=75%; P<.001
Garg et al. 2016	0.32	-0.31-0.94	20	11.8
Ray et al. 2015 (with priming)	1.98	1.21-2.75	20	10.3
Ray et al. 2015 (without priming)	1.07	0.41-1.74	20	11.3
Tikka et al. 2017	0.69	-0.33-1.71	8	8.2
Kumar et al. 2020	0.56	0.16-0.96	50	13.9
Singh et al. 2020	0.45	-0.28-1.18	15	10.8
Chauhan et al. 2021	0.93	0.26-1.60	19	11.3
Gupta et al. 2021	2.44	1.61-3.28	20	9.7
Tyagi et al. 2022	0.66	0.14-1.18	30	12.8
Negative symptoms (PANSS-NS & SANS)	Pooled SMD 1.35 (0.91-1.80); K=12; n=297; Z=5.97; P=<.001; Heterogeneity I2=82%; P<.001
Negative symptoms (only PANSS-NS)	Pooled SMD 1.05 (0.66-1.43); K=9; n=202; Z=5.38; P=<.001; Heterogeneity I2=67%; P<.01
Negative symptoms (PANSS-NS & SANS, Primary outcome)	Pooled SMD 1.49 (0.95-2.03); K=8; n=219; Z=5.41; P=<.001; Heterogeneity I2=83%; P<.001
Garg et al. 2016	0.41	-0.22-1.03	20	8.7 (11.8 only PANSS-NS)
Ray et al. 2015 (with priming)	1.93	1.16-2.69	20	8.1 (10.2 only PANSS-NS)
Ray et al. 2015 (without priming)	1.62	0.89-2.34	20	8.3 (10.7 only PANSS-NS)
Tikka et al. 2017	0.37	-0.63-1.36	8	7.0 (8.0 only PANSS-NS)
Basavaraju et al. 2020 (SANS)	3.42	2.61-4.23	30	7.8
Kumar et al. 2020 (SANS)	1.36	0.92-1.79	50	9.5
Kumar et al. 2020	1.18	0.75-1.60	50	9.6 (14.2 only PANSS-NS)
Singh et al. 2020 (SANS)	2.25	1.31-3.19	15	7.2
Singh et al. 2020	1.45	0.63-2.27	15	7.8 (9.6 only PANSS-NS)
Chauhan et al. 2021	0.78	0.11-1.44	19	8.5 (11.4 only PANSS-NS)
Gupta et al. 2021	1.47	0.77-2.18	20	8.3 (10.9 only PANSS-NS)
Tyagi et al. 2022	0.33	-0.18-0.84	30	9.2 (13.2 only PANSS-NS)
PANSS total scores	Pooled SMD 1.40 (0.77-2.04); K=9; n=202; Z=4.34; P=<.001; Heterogeneity I2=87%; P<.001
Garg et al. 2016	0.43	-0.20-1.05	20	11.7
Ray et al. 2015 (with priming)	2.75	1.86-3.64	20	10.5
Ray et al. 2015 (without priming)	3.43	2.42-4.44	20	9.9
Tikka et al. 2017	0.48	-0.52-1.48	8	10.0
Kumar et al. 2020	1.21	0.78-1.64	50	12.5
Chauhan et al. 2021	1.09	0.40-1.78	19	11.4
Gupta et al. 2021	2.69	1.81-3.57	20	10.6
Thirtalli et al. 2022	0.46	-0.26-1.19	15	11.3
Tyagi et al. 2022	0.52	0.00-1.03	30	12.2
Auditory Hallucinations	Pooled SMD 1.79 (0.81-2.77); K=5; n=120; Z=3.58; P=<.001; Heterogeneity I2=90%; P<.001
Bagati et al. 2009	4.42	3.23-5.61	20	17.0
Ray et al. 2015 (with priming)	1.92	1.16-2.69	20	20.0
Ray et al. 2015 (without priming)	1.73	1.00-2.47	20	20.2
Tyagi et al. 2022 (AVHRS)	0.58	0.06-1.09	30	21.4
Tyagi et al. 2022 (PSYRATS-AH)	0.84	0.32-1.37	30	21.4
Cognition	Pooled SMD 0.45 (0.10-0.81); K=3; n=64; Z=2.53; P<.05; Heterogeneity I2=0%; P=0.85
Chauhan et al. 2021 (SCoRS)	0.45	-0.19-1.10	19	29.7
Thirtalli et al. 2022 (Bhatia Battery)	0.29	-0.43-1.01	15	23.8
Tyagi et al. 2022 (ScoRS)	0.54	0.02-1.06	30	46.4

PANSS=Positive and negative syndrome scale; NS=Negative symptoms; PS=Positive symptoms; SANS=Scale for assessment of negative symptoms; AVHRS=Auditory Vocal Hallucination Rating Scale; PSYRATS-AH=Psychotic Symptom Rating Scales - Auditory Hallucination; ScoRS=Schizophrenia Cognition Rating Scale; SMD=Standardized mean difference; K=Number of studies; n=Number of samples

A subgroup analysis also showed only for studies where the primary outcome was positive symptoms or negative symptoms, rTMS was found to be associated with significant improvement with large effect sizes, however with significantly large heterogeneity.

Active only/preintervention to end of treatment—OCD

Active rTMS as an augment was found to significantly improve the core symptoms of OCD with larger effect sizes [Table 4]. A subgroup analysis also showed that LF rTMS targeting the SMA was also significant with a large effect size. However, heterogeneity was significantly large across both these analyses.

Table 4.

Meta-analysis (active treatment arms) baseline to post-intervention Obsessive compulsive disorder (12 treatment arms), mania (2 treatment arms), substance use disorder (5 treatment arms), headache (6 treatment arms), panic disorder (1 treatment arm), and essential tremors (1 treatment arm)

Study	Effect
	SMD	Confidence interval	n	Weighted %
Obsessive Compulsive Disorder	Pooled SMD 1.24 (0.85-1.63); K=12; n=190; Z=6.25; P<.001; Heterogeneity I2=65%; P<.001
Obsessive Compulsive Disorder (LF-SMA)	Pooled SMD 1.33 (0.80-1.86); K=6; n=79; Z=4.95; P<.001; Heterogeneity I2=52%; P<.001
Sarkhel et al. 2010	0.62	0.00-1.24	21	10.0
Kumar et al. 2011	1.30	0.41-2.20	12	7.8
Arumugham et al. 2018	0.44	-0.20-1.09	19	9.8
Kumar et al. 2018a	1.32	0.70-1.93	25	10.0
Singh et al. 2019	1.09	0.42-1.75	20	9.6
Agrawal et al. 2021	1.11	0.27-2.50	5	5.0
Dutta et al. 2021	0.47	-0.20-1.13	18	9.6
Syed et al. 2021	0.93	0.08-1.78	12	8.2
Joshi et al. 2022	2.11	1.12-3.10	13	7.2
Reddy et al. 2022	3.28	2.14-4.42	15	6.2
Vidya et al. 2022 (with priming)	1.64	0.80-2.48	15	8.2
Vidya et al. 2022 (without priming)	1.60	0.77-2.44	15	8.3
Mania	Pooled SMD 2.43 (0.53-4.33); K=2; n=34; Z=2.50; P=<.05; Heterogeneity I2=88%; P<.01
Praharaj et al. 2009	3.41	2.43-4.38	21	49.4
Pathak et al. 2015	1.47	0.59-2.35	13	50.6
Substance use disorders (craving/compulsion)	Pooled SMD 7.09 (3.91-10.28); K=5; n=100; Z=4.37; P<.001; Heterogeneity I2=94%; P<.001
Alcohol Dependence Syndrome (craving/compulsion)	Pooled SMD 7.80 (3.43-12.18); K=4; n=80; Z=3.50; P<.001; Heterogeneity I2=96%; P<.001
Mishra et al. 2010	9.51	7.68-11.34	30	20.0
Mishra et al. 2015 (right)	3.91	2.30-5.52	10	20.3
Mishra et al. 2015 (left)	3.57	2.05-5.09	10	20.4
Raikwar et al. 2021	14.68	11.91-17.44	30	18.5
Ankit et al. 2022	4.59	3.37-5.82	20	20.8
Migraine- severity	Pooled SMD 2.15 (0.76-3.54); K=6; n=167; Z=3.03; P<.01; Heterogeneity I2=96%; P<.001
Misra et al. 2012	5.77	4.87-6.67	51	16.4
Kumar et al. 2018c	1.55	0.69-2.41	14	16.5
Sahu et al. 2019	2.04	1.26-2.82	20	25.4
Kumar et al. 2021	2.11	0.97-3.25	10	16.7
Shah et al. 2022 (LF)	0.27	-0.20-0.73	36	17.3
Shah et al. 2022 (HF)	1.31	0.80-1.83	36	17.2
Migraine- frequency	Pooled SMD 1.73 (1.20-2.26); K=6; n=167; Z=6.38; P<.001; Heterogeneity I2=74%; P<.01
Misra et al. 2012	1.73	1.27-2.19	51	20.8
Kumar et al. 2018c	0.97	0.18-1.76	14	15.9
Sahu et al. 2019	2.08	1.30-2.87	20	16.0
Kumar et al. 2021	3.78	2.21-5.35	10	7.8
Shah et al. 2022 (LF)	1.01	0.52-1.50	36	20.3
Shah et al. 2022 (HF)	2.00	1.43-2.57	36	19.2
Panic disorder
Kumar et al. 2018b	1.90	0.95-2.85	13	100
Essential tremors
Batra et al. 2022 (FTMRS)	0.22	-0.66-1.10	10	100
Batra et al. 2022 (TETRAS)	0.31	-0.57-1.19	10	100

HF=High frequency; LF=Low frequency; SMA=Supplementary motor area; SMD=Standardized mean difference; K=Number of studies; n=Number of samples; FTMRS=Fahn-Tolosa-Marin Tremor Rating Scale; TETRAS=Tremor Research Group Essential Tremor Rating Scale

Active only/preintervention to end of treatment—Mania

Active rTMS as an augment was found to significantly improve mania with larger effect sizes [Table 4]. Heterogeneity was found to be significantly large.

Active only/preintervention to end of treatment—Substance use disorders

Three studies with alcohol dependence syndrome and one with opioid dependence syndrome were included for the analysis. Active rTMS as an augment was found to significantly improve craving in substance use disorders with larger effect sizes but large heterogeneity [Table 4]. A subgroup analysis also showed that the effect of rTMS was significant with large effect sizes in reducing craving in alcohol dependence syndrome. Heterogeneity in the subgroup analysis too was significantly large.

Active only/preintervention to end of treatment—Migraine

Both headache severity and frequency were found to significantly reduce with active rTMS prophylaxis with larger effect sizes but with large heterogeneity [Table 4].

For studies including panic disorder, essential tremors, and meta-analyses could not be conducted as there was only one study each. The evidence from these studies, however, was positive.

Active only/preintervention to follow-up—Depression

Active rTMS was found to be significantly effective as an augment in the treatment of depression even until follow-up (of 2 weeks to 12 weeks) [Table 5]. Improvements in “any depression,” depression in unipolar or bipolar depressive disorders, and depression associated with OCD were had large effect sizes. Improvement in depression associated with schizophrenia had a moderate effect size. The meta-analyses for depression with OCD showed no heterogeneity; rest of them showed significant heterogeneity.

Table 5.

Meta-analysis (active treatment arms) preintervention to follow-up Depression (11 treatment arms)

Study	Effect
	SMD	Confidence Interval	n	Weighted % any (Sub)
Any Depression	Pooled SMD 1.38 (0.71-2.05); K=11; n=190; Z=4.04; P<.001; Heterogeneity I2=89%; P<.001
Depression (Unipolar + Bipolar)	Pooled SMD 2.53 (1.40-3.66); K=4; n=72; Z=4.39; P<.001; Heterogeneity I2=83%; P<.001
Nongpiur et al. 2011 (with priming)	3.95	2.78-5.12	18	7.9 (22.9)
Nongpiur et al. 2011 (without priming)	3.00	2.05-3.96	19	8.6 (24.9)
Shere et al. 2021	2.34	1.59-3.09	24	9.2 (26.8)
Mallik et al. 2022 (only bipolar depression)	0.99	0.09-1.88	11	8.8 (25.5)
Depression comorbid with OCD	Pooled SMD 1.00 (0.57-1.43); K=3; n=48; Z=4.58; P=<.001; Heterogeneity I2=0%; P=0.50
Dutta et al. 2021	1.15	0.44-1.87	18	9.3 (36.4)
Vidya et al. 2022 (with priming)	1.23	0.44-2.02	15	9.1 (29.6)
Vidya et al. 2022 (without priming)	0.64	-0.09-1.38	15	9.2 (34.0)
Depression comorbid with Schizophrenia	Pooled SMD 0.39 (-0.62-1.40); K=3; n=80; Z=0.76; P=0.45; Heterogeneity I2=90%; P<.001
Garg et al. 2016	0.86	0.21-1.51	20	9.5 (32.3)
Basavaraju et al. 2020	-0.58	-1.10- -0.07	30	9.8 (34.0)
Tyagi et al. 2022	0.92	0.39-1.46	30	9.7 (33.8)
Depression comorbid with Migraine
Kumar et al. 2018	1.27	0.45-2.09	14	9.0

OCD=Obsessive compulsive disorder; SMD=Standardized mean difference; K=Number of studies; n=Number of samples

Active only/preintervention to follow-up—Schizophrenia

Active rTMS was found to be significantly effective as an augment in the treatment of various symptom domains of schizophrenia even until follow-up (of 2 weeks to 6 weeks) [Table 6]. Improvements in positive symptoms, negative symptoms, total psychopathology, auditory hallucinations, and cognitive functions in schizophrenia were all significant with large effect sizes. Except for cognitive functions, all meta-analyses showed significant heterogeneity.

Table 6.

Meta-analysis (active treatment arms) preintervention to follow-up Schizophrenia (12 treatment arms)

Study	Effect
	SMD	Confidence Interval	n	Weighted %
Positive Symptoms	Pooled SMD 1.11 (0.61-1.60); K=6; n=159; Z=4.36; P<.001; Heterogeneity I2=75%; P<.01
Garg et al. 2016	0.37	-0.26-0.99	20	16.8
Ray et al. 2015 (with priming)	2.54	1.69-3.39	20	13.6
Ray et al. 2015 (without priming)	1.28	0.60-1.97	20	15.9
Kumar et al. 2020	0.65	0.25-1.05	50	19.9
Chauhan et al. 2021	1.06	0.38-1.75	19	15.9
Tyagi et al. 2022	1.09	0.55-1.64	30	17.9
Negative symptoms (PANSS-NS & SANS)	Pooled SMD 1.74 (1.09-2.39); K=8; n=239; Z=5.24; P=<.001; Heterogeneity I2=89%; P<.001
Negative symptoms (only PANSS-NS)	Pooled SMD 1.42 (0.78-2.06); K=6; n=159; Z=4.34; P=<.001; Heterogeneity I2=84%; P<.001
Garg et al. 2016	0.61	-0.03-1.25	20	12.7 (17.0 only PANSS-NS)
Ray et al. 2015 (with priming)	2.77	1.88-3.66	20	11.4 (14.6 only PANSS-NS)
Ray et al. 2015 (without priming)	2.37	1.54-3.19	20	11.7 (15.2 only PANSS-NS)
Basavaraju et al. 2020 (SANS)	3.77	2.91-4.64	30	11.5
Kumar et al. 2020 (SANS)	1.69	1.23-2.14	50	13.5
Kumar et al. 2020	1.51	1.06-1.96	50	13.5 (18.6 only PANSS-NS)
Chauhan et al. 2021	1.03	0.34-1.71	19	12.5 (16.6 only PANSS-NS)
Tyagi et al. 2022	0.55	0.04-1.07	30	13.2 (18.0 only PANSS-NS)
PANSS total scores	Pooled SMD 1.89 (1.05-2.73); K=6; n=159; Z=4.39; P=<.001; Heterogeneity I2=89%; P<.001
Garg et al. 2016	0.57	-0.07-1.20	20	16.8
Ray et al. 2015 (with priming)	3.57	2.54-4.60	20	15.0
Ray et al. 2015 (without priming)	3.80	2.73-4.88	20	14.7
Kumar et al. 2020	1.58	1.13-2.03	50	18.3
Chauhan et al. 2021	1.49	0.76-2.22	19	16.8
Tyagi et al. 2022	0.88	0.35-1.41	30	17.9
Auditory Hallucinations	Pooled SMD 2.10 (1.02-3.18); K=4; n=100; Z=3.83; P=<.001; Heterogeneity I2=89%; P<.001
Ray et al. 2015 (with priming)	3.35	2.36-4.34	20	23.1
Ray et al. 2015 (without priming)	3.04	2.11-3.98	20	23.6
Tyagi et al. 2022 (AVHRS)	0.97	0.43-1.51	30	26.8
Tyagi et al. 2022 (PSYRATS-AH)	1.32	0.76-1.88	30	26.8
Cognition	Pooled SMD 0.91 (0.49-1.33); K=2; n=49; Z=4.27; P<.001; Heterogeneity I2=0%; P=0.78
Chauhan et al. 2021 (ScoRS)	0.83	0.17-1.50	19	39.3
Tyagi et al. 2022 (ScoRS)	0.96	0.42-1.49	30	60.7

PANSS=positive and negative syndrome scale; NS=negative symptoms; PS=positive symptoms; SANS=scale for assessment of negative symptoms; AVHRS=Auditory Vocal Hallucination Rating Scale; PSYRATS-AH=Psychotic Symptom Rating Scales - Auditory Hallucination; ScoRS=Schizophrenia Cognition Rating Scale; SMD=standardized mean difference; K=number of studies; n=number of samples

Active only/preintervention to follow-up—OCD

Active rTMS as an augment was found to significantly improve the core symptoms of OCD with larger effect sizes even until follow-up [Table 7]. Heterogeneity was low and not significant.

Table 7.

Meta-analysis (active treatment arms) preintervention to follow-up Obsessive compulsive disorder (4 treatment arms), substance use disorder (2 treatment arms), and headache (4 treatment arms)

Study	Effect
	SMD	Confidence Interval	n	Weighted %
Obsessive Compulsive Disorder	Pooled SMD 1.86 (1.32-2.41); K=4; n=73; Z=6.73; P<.001; Heterogeneity I2=44%; P=0.15
Kumar et al. 2018a	1.34	0.73-1.96	25	32.9
Dutta et al. 2021	1.67	0.90-2.45	18	26.6
Vidya et al. 2022 (with priming)	2.28	1.33-3.22	15	20.9
Vidya et al. 2022 (without priming)	2.55	1.55-3.54	15	19.6
Substance use disorder	Pooled SMD 6.17 (1.07-11.28); K=2; n=60; Z=2.37; P<.05; Heterogeneity I2=97%; P<0.001
Mishra et al. 2010	3.62	2.78-4.46	30	51.1
Raikwar et al. 2020	8.83	7.12-10.54	30	48.9
Migraine- severity	Pooled SMD 2.07 (1.48-2.66); K=4; n=95; Z=6.91; P<.001; Heterogeneity I2=56%; P=0.08
Misra et al. 2012	2.72	2.18-3.27	51	32.9
Kumar et al. 2018c	1.90	0.99-2.82	14	21.7
Sahu et al. 2019	1.70	0.96-2.43	20	26.7
Kumar et al. 2021	1.66	0.61-2.71	10	18.7
Migraine- frequency	Pooled SMD 2.19 (0.72-3.66); K=3; n=44; Z=2.92; P<.01; Heterogeneity I2=85%; P<.01
Kumar et al. 2018c	0.97	0.18-1.76	14	37.1
Sahu et al. 2019	1.75	1.01-2.49	20	37.5
Kumar et al. 2021	4.60	2.78-6.42	10	25.4

SMD=Standardized mean difference; K=number of studies; n=Number of samples

Active only/preintervention to follow-up—Substance use disorders

Two studies, both with alcohol dependence syndrome, were included for the analysis. Active rTMS as an augment was found to significantly improve craving in substance use disorders with larger effect sizes but significant heterogeneity, even until follow-up [Table 7].

Active only/preintervention to follow-up—Migraine

Both headache severity and frequency were found to significantly reduce until follow-up (1–3 months) with active rTMS prophylaxis with larger effect sizes [Table 7]. While the heterogeneity was not significant for headache severity, it was significant for headache frequency.

Active versus Sham/preintervention to end of treatment

The sham-controlled improvement of rTMS was only significant in migraine prophylaxis—for both headache severity and frequency. This improvement though was associated with significant heterogeneity. None of the other outcome measures showed sham-controlled improvement in any of the other disorders. Heterogeneity was not significant for any depression, unipolar and bipolar depression, depression associated with OCD and schizophrenia, positive and negative symptoms of schizophrenia, total psychopathology and cognitive dysfunction, and obsessive compulsive symptoms in OCD Tables 8–10 and Figure 2.

Table 8.

Meta-analysis (active vs sham treatment arms) preintervention to end of treatment Depression (12*2 treatment arms)

Study	Effect
	SMD	Confidence Interval	n	Weighted % Any (Sub)
Any Depression	Pooled SMD 0.24 (-0.09-0.57); K=12; n=245/243; Z=1.43; P=0.15; Heterogeneity I2=66%; P<.001
Depression (Unipolar + Bipolar)	Pooled SMD 0.89 (-0.68-2.45); K=3; n=40/42; Z=1.11; P=0.27; Heterogeneity I2=90%; P<.001
Lingeswaran 2011	0.12	-0.72-0.96	9/14	7.1 (33.5)
Ray et al. 2011	2.46	1.62-3.30	20/20	7.1 (33.5)
Mallik et al. 2022 (only Bipolar depression)	0.07	-0.84-0.98	11/8	6.6 (33.0)
Depression comorbid with OCD	Pooled SMD 0.08 (-0.30-0.45); K=3; n=58/53; Z=0.40; P=0.69; Heterogeneity I2=0%; P=0.75
Sarkhel et al. 2010	0.24	-0.36-0.85	21/21	9.1 (37.8)
Arumugham et al. 2018	-0.10	-0.75-0.56	19/17	8.7 (32.5)
Dutta et al. 2021	0.05	-0.63-0.74	18/15	8.4 (29.7)
Depression associated with Schizophrenia	Pooled SMD 0.08 (-0.20-0.35); K=5; n=100/99; Z=0.53; P=0.59; Heterogeneity I2=0%; P=0.80
Goyal et al. 2007	0.23	-1.01-1.48	5/5	4.5 (5.0)
Garg et al. 2016	0.29	-0.33-0.91	20/20	9.0 (20.0)
Basavaraju et al. 2020	-0.14	-0.64-0.37	30/30	10.1 (30.2)
Singh et al. 2020	-0.09	-0.81-0.62	15/15	8.1 (15.1)
Tyagi et al. 2022	0.21	-0.30-0.72	30/29	10.1 (29.6)
Post-stroke depression
Sharma et al. 2020	0.24	-0.43-0.37	47/49	11.1

OCD=Obsessive compulsive disorder; SMD=Standardized mean difference; K=Number of studies; n=Number of samples; MADRS=Montgomery and Asberg Depression Rating Scale; HAMD=Hamilton Depression Rating Scale

Table 10.

Meta-analysis (active vs sham treatment arms) preintervention to end of treatment Obsessive compulsive disorder (4*2 treatment arms), mania (2*2 treatment arms), substance use disorders (3*2 treatment arms), and headache (4+3 treatment arms)

Study	Effect
	SMD	Confidence Interval	n	Weighted %
Obsessive Compulsive Disorder	Pooled SMD 0.21 (-0.20-0.62); K=4; n=71/64; Z=1.00; P=0.32; Heterogeneity I2=30%; P=0.23
Sarkhel et al. 2010	0.15	-0.46-0.75	21/21	29.8
Arumugham et al. 2018	-0.20	-0.85-0.46	19/17	26.8
Dutta et al. 2021	0.20	-0.49-0.88	18/15	25.2
Joshi et al. 2022	0.94	0.08-1.79	13/11	18.3
Mania	Pooled SMD 0.57 (-0.37-1.51); K=2; n=34/33; Z=1.18; P=0.24; Heterogeneity I2=71%; P=0.06
Praharaj et al. 2009	0.07	-0.70-0.83	13/13	47.7
Pathak et al. 2015	1.03	0.37-1.68	20/20	52.3
Substance use disorders (craving/compulsion)	Pooled SMD 0.73 (-0.18-1.64); K=3; n=80/65; Z=1.56; P=0.12; Heterogeneity I2=85%; P<.01
Alcohol dependence syndrome	Pooled SMD 1.00 (-0.49-2.48); K=2; n=60/45; Z=1.31; P=0.19; Heterogeneity I2=91%; P<.001
Mishra et al. 2010	1.78	1.05-2.51	30/15	31.5 (48.5% for ADS)
Raikwar et al. 2021	0.26	-0.25-0.77	30/30	35.2 (51.5% for ADS)
Ankit et al. 2022	0.22	-0.40-0.84	20/20	33.3
Headache- severity	Pooled SMD 1.04 (0.25-1.82); K=4; n=102/103; Z=2.59; P<.05; Heterogeneity I2=84%; P<.001
Sahu et al. 2019	1.85	1.11-2.59	20/21	24.2
Kumar et al. 2021	1.48	0.47-2.50	10/10	20.4
Shah et al. 2022 (LF)	0.09	-0.37-0.55	36/36	27.9
Shah et al. 2022 (HF)	0.95	0.46-1.44	36/36	27.6
Headache- frequency	Pooled SMD 1.67 (0.61-2.74); K=4; n=102/103; Z=3.07; P<.01; Heterogeneity I2=90%; P<.001
Sahu et al. 2019	1.93	1.17-2.68	20/21	26.0
Kumar et al. 2021	3.88	2.28-5.48	10/10	17.8
Shah et al. 2022 (LF)	0.30	-0.16-0.76	36/36	28.3
Shah et al. 2022 (HF)	1.41	0.89-1.93	36/36	27.9

HF=High frequency; LF=Low frequency; SMD=Standardized mean difference; K=Number of studies; n=Number of samples

Figure 2.

Forest Plots depicting the results of meta-analyses for sham-controlled studies for depression (unipolar/bipolar), schizophrenia (negative symptoms and auditory hallucinations), obsessive compulsive disorder, migraine (headached severity and frequency)

Table 9.

Meta-analysis (active vs sham treatment arms) preintervention to end of treatment Schizophrenia (9*2 treatment arms)

Study	Effect
	SMD	Confidence Interval	n	Weighted %
Positive Symptoms	Pooled SMD 0.06 (-0.17-0.30); K=6; n=142/138; Z=0.53; P=0.60; Heterogeneity I2=0%; P=0.96
Positive Symptoms (primary outcome)	Pooled SMD 0.14 (-0.18-0.46); K=4; n=77/73; Z=0.88; P=0.38; Heterogeneity I2=0%; P=0.93
Garg et al. 2016	0.08	-0.54-0.70	20/20	14.3
Tikka et al. 2017	0.19	-0.82-1.21	8/7	5.3
Kumar et al. 2020	0.00	-0.39-0.39	50/50	35.8
Singh et al. 2020	-0.13	-0.84-0.59	15/15	10.7
Chauhan et al. 2021	0.00	-0.65-0.65	19/17	12.9
Tyagi et al. 2022	0.27	-0.25-0.78	30/29	20.9
Negative symptoms (PANSS-NS & SANS)	Pooled SMD 0.08 (-0.19-0.35); K=9; n=237/233; Z=0.57; P=0.57; Heterogeneity I2=49%; P=0.05
Negative symptoms (only PANSS-NS)	Pooled SMD 0.16 (-0.07-0.40); K=6; n=142/138; Z=1.37; P=0.17; Heterogeneity I2=0%; P=0.96
Negative symptoms (PANSS-NS & SANS, primary outcome)	Pooled SMD 0.07 (-0.22-0.36); K=8; n=229/226; Z=0.49; P=0.63; Heterogeneity I2=55%; P<.05
Garg et al. 2016	0.15	-0.47-0.78	20/20	10.5 (14.3 only PANSS-NS)
Tikka et al. 2017	0.17	-0.84-1.19	8/7	5.4 (5.3 only PANSS-NS)
Basavaraju et al. 2020 (SANS)	-0.83	-1.36- -0.30	30/30	12.4
Kumar et al. 2020 (SANS)	0.39	0.00-0.79	50/50	15.7
Kumar et al. 2020	0.31	-0.08-0.71	50/50	15.7 (35.5 only PANSS-NS)
Singh et al. 2020 (SANS)	0.33	-0.39-1.05	15/15	8.8
Singh et al. 2020	0.13	-0.58-0.85	15/15	8.8 (10.8 only PANSS-NS)
Chauhan et al. 2021	-0.03	-0.68-0.62	19/17	9.9 (12.9 only PANSS-NS)
Tyagi et al. 2022	0.05	-0.46-0.56	30/29	12.8 (21.2 only PANSS-NS)
PANSS total scores	Pooled SMD -0.04 (-0.32-0.24); K=6; n=142/136; Z=0.29; P=0.77; Heterogeneity I2=22%; P=0.27
Garg et al. 2016	0.19	-0.44-0.81	20/20	15.9
Tikka et al. 2017	-0.01	-1.03-1.00	8/7	6.9
Kumar et al. 2020	0.19	-0.20-0.58	50/50	30.1
Chauhan et al. 2021	-0.09	-0.74-0.57	19/17	14.6
Thirtalli et al. 2022	0.09	-0.65-0.83	15/13	11.9
Tyagi et al. 2022	-0.61	-1.13- -0.08	30/29	20.6
Auditory Hallucinations	Pooled SMD 1.02 (-0.28-2.31); K=3; n=80/78; Z=1.54; P=0.12; Heterogeneity I2=93%; P<.001
Bagati et al. 2009	2.77	1.88-3.67	20/20	31.1
Tyagi et al. 2022 (AVHRS)	0.19	-0.32-0.71	30/29	34.5
Tyagi et al. 2022 (PSYRATS-AH)	0.26	-0.26-0.77	30/29	34.5
Cognition	Pooled SMD -0.11 (-0.46-0.25); K=3; n=64/59; Z=0.60; P=0.55; Heterogeneity I2=0%; P=0.47
Chauhan et al. 2021 (ScoRS)	-0.21	-0.86-0.45	19/17	29.3
Thirtalli et al. 2022 (Bhatia Battery)	-0.44	-1.20-0.31	15/13	22.3
Tyagi et al. 2022 (ScoRS)	0.10	-0.41-0.61	30/29	48.4

PANSS=Positive and negative syndrome scale; NS=Negative symptoms; PS=Positive symptoms; SANS=Scale for assessment of negative symptoms; AVHRS=Auditory Vocal Hallucination Rating Scale; PSYRATS-AH=Psychotic Symptom Rating Scales - Auditory Hallucination; ScoRS=Schizophrenia Cognition Rating Scale; SMD=Standardized mean difference; K=Number of studies; n=number of samples

Active versus Sham/preintervention to follow-up

Meta-analysis for sham controlled studies for the preintervention to follow-up could only be performed for only a few outcomes. Significant effect of rTMS at follow-up was shown only for alcohol dependence syndrome, that too with no heterogeneity [Table 11]. All other analysis revealed no significant effect. Except for analysis of migraine studies, all other outcomes showed that the heterogeneity was not significant.

Table 11.

Meta-analysis (active vs sham treatment arms) preintervention to follow-up Schizophrenia (5*2 treatment arms), alcohol dependence syndrome (2*2 treatment arms), and headache (3*2 treatment arms)

Study	Effect
	SMD	Confidence Interval	n	Weighted %
Positive Symptoms	Pooled SMD 0.14 (-0.17-0.44); K=4; n=119/116; Z=0.87; P=0.38; Heterogeneity I2=25%; P=0.26
Garg et al. 2016	-0.23	-0.85-0.39	20/20	19.3
Kumar et al. 2020	0.04	-0.35-0.43	50/50	37.4
Chauhan et al. 2021	0.14	-0.51-0.80	19/17	17.8
Tyagi et al. 2022	0.55	0.03-1.07	30/29	25.4
Negative symptoms (PANSS-NS & SANS)	Pooled SMD 0.06 (-0.38-0.50); K=6; n=199/196; Z=0.26; P=0.80; Heterogeneity I2=78%; P<.001
Negative symptoms (only PANSS-NS)	Pooled SMD 0.26 (-0.03-0.54); K=4; n=119/116; Z=1.74; P=0.08; Heterogeneity I2=17%; P=0.30
Garg et al. 2016	0.19	-0.43-0.81	20/20	15.2 (18.7 only PANSS-NS)
Basavaraju et al. 2020 (SANS)	-0.83	-1.36- -0.30	30/30	16.4
Kumar et al. 2020 (SANS)	0.39	0.00-0.79	50/50	18.5
Kumar et al. 2020	0.55	0.15-0.95	50/50	18.4 (38.1 only PANSS-NS)
Chauhan et al. 2021	-0.12	-0.78-0.53	19/17	14.7 (17.1 only PANSS-NS)
Tyagi et al. 2022	0.12	-0.39-0.63	30/29	16.8 (26.1 only PANSS-NS)
PANSS total scores	Pooled SMD 0.22 (-0.06-0.50); K=4; n=119/116; Z=1.54; P=0.12; Heterogeneity I2=13%; P=0.33
Garg et al. 2016	0.07	-0.55-0.69	20/20	18.4
Kumar et al. 2020	0.41	0.01-0.81	50/50	39.5
Chauhan et al. 2021	-0.26	-0.92-0.39	19/17	16.5
Tyagi et al. 2022	0.35	-0.17-0.86	30/29	25.6
Schizophrenia- Cognition	Pooled SMD 0.11 (-0.34-0.55); K=2; n=49/46; Z=0.47; P=0.64; Heterogeneity I2=15%; P=0.28
Chauhan et al. 2021 (ScoRS)	-0.17	-0.83-0.48	19/17	39.8
Tyagi et al. 2022 (ScoRS)	0.29	-0.22-0.80	30/29	60.2
Schizophrenia- Depression	Pooled SMD 0.19 (-0.15-0.53); K=3; n=64/66; Z=1.08; P=0.28; Heterogeneity I2=0%; P=0.45
Garg et al. 2016	0.37	-0.26-0.99	20/20	29.4
Chauhan et al. 2021	-0.17	-0.83-0.48	19/17	26.8
Tyagi et al. 2022	0.29	-0.23-0.80	30/29	43.7
Alcohol dependence syndrome	Pooled SMD 0.47 (0.07-0.87); K=2; n=60/45; Z=2.31; P<0.05; Heterogeneity I2=0%; P=0.68
Mishra et al. 2010	0.57	-0.06-1.20	30/15	39.6
Raikwar et al. 2021	0.40	-0.11-0.91	30/30	60.4
Headache- severity	Pooled SMD 2.51 (-0.09-5.12); K=3; n=80/81; Z=1.89; P=0.06; Heterogeneity I2=97%; P<0.001
Misra et al. 2013	5.18	4.35-6.01	50/50	33.3
Sahu et al. 2019	1.21	0.54-1.88	20/21	33.7
Kumar et al. 2021	1.15	0.19-2.12	10/10	32.9
Headache- frequency	Pooled SMD 2.79 (0.02-5.56); K=2; n=30/31; Z=1.97; P=0.05; Heterogeneity I2=89%; P<0.01
Sahu et al. 2019	1.49	0.79-2.19	20/21	54.1
Kumar et al. 2021	4.32	2.59-6.05	10/10	45.9

PANSS=positive and negative syndrome scale; NS=negative symptoms; PS=positive symptoms; SANS=scale for assessment of negative symptoms; ScoRS=Schizophrenia Cognition Rating Scale; SMD=standardized mean difference; K=Number of studies; n=Number of samples

Safety/Adverse events

The highest frequency for any adverse event associated with rTMS was headache (8.9% for active rTMS; 3.36% for sham rTMS), followed by scalp pain/discomfort and facial pain. The frequency of serious adverse events due to active rTMS-seizures and affective switch was rare (<0.5% for each). Compared to sham, the odds of active rTMS to precipitate adverse events were higher for all the adverse events [Table 12]. See supplementary material-IV for adverse events reported in each of the included study.

Table 12.

Adverse effects

Adverse effect/event	Active (n=1065)	Sham (n=565)	Odds ratio
Seizure	4 (0.37%)	0	4.79
Affective switch	5 (0.47%)	0	5.87
Headache	95 (8.92%)	19 (3.36%)	2.81
Scalp Pain/discomfort	57 (5.35%)	6 (1.06%)	5.27
Dizziness/Giddiness	18 (1.69%)	7 (1.24%)	1.37
Facial pain	50 (4.69%)	7 (1.24%)	3.93
Neck pain/contraction	7 (0.66%)	0	8.01
Drowsiness	6 (0.56%)	0	6.94
Insomnia	4 (0.38%)	0	4.79
Nightmare	1 (0.09%)	0	1.59
Twitching (facial/temporal)	2 (0.19%)	0	2.66
Low hearing	1 (0.09%)	0	1.59
Rhinorrhea	1 (0.09%)	0	1.59
Lacrimation	1 (0.09%)	0	1.59
Discontinuation due to adverse effects	16 (1.50%) 4 (25%)- Seizure 5 (31.25%)- Switch 7 (43.75%)- Headache/scalp pain

Sensitivity analysis and publication bias

Sensitivity analysis conducted by removing the studies deemed to have low quality and to be at high risk of bias suggested to have not a significant impact, except for active vs. sham meta-analysis for migraine studies. The significant effect of rTMS in the prophylaxis of migraine on the headache severity was lost, when the study with high risk of bias was excluded. Both studies included for meta-analysis of effects of rTMS on mania were deemed to be on high risk of bias. See supplementary material-V for sensitivity analysis.

Funnel charts of each of the analysis (supplementary material-VI) showed that publication bias may be significant for “any depression,” schizophrenia (positive symptoms, negative symptoms, total psychopathology, and auditory hallucinations) and substance use disorder, for the pre–post-intervention meta-analyses and for schizophrenia (positive symptoms and auditory hallucinations), mania, substance use disorder, and migraine (headache severity and frequency).

DISCUSSION

Efficacy

Our meta-analyses show positive sham-controlled evidence for the use of rTMS only for migraine (headache severity and frequency) at end of treatment and for craving in alcohol dependence at follow-up. Outcomes for all other disorders such as depression, schizophrenia, and obsessive compulsive disorder were not significant. However, meta-analyses of “active only” studies suggested a significant effect of rTMS for all outcomes, with moderate-to-large effect sizes, both at end of treatment and at follow-up.

While the result on migraine prophylaxis is supported by other meta-analysis,[64] positive evidence for anti-craving effect in alcohol dependence syndrome is not supported.[65] However, both these positive results lost significance in the sensitivity analysis. We could not find any significant effect of rTMS on depression, positive symptoms including auditory hallucinations and negative symptoms of schizophrenia, and OCD in contrast to a recent meta-analysis across various mental disorders including studies from over worldwide.[5] The intervention complexity, complexity in the assessment, lack of homogeneity and more importantly, lack of sufficient number of studies could be the possible explanation.

Majority of the Indian studies on rTMS were on schizophrenia for treatment of various symptom domains followed by depression/bipolar disorder compared to recent meta-analysis on TMS which observed half the studies worldwide were on either depression or bipolar disorder followed by studies on schizophrenia.[5] The rTMS studies on anxiety disorders and cigarette smoking were sparse in India.

Although studies included in our review involve a wide range of population and diagnostic categories, rTMS was received as an adjuvant treatment along with psychotropic medications or psychotherapy that suggests the possible improvement in core psychiatry symptom severity could be due to additive pharmacodynamic effect or secondary augmentation of the ongoing management rather than primary role of TMS only. At the same time, the therapeutic advantage of TMS could not be refuted due to the fact that most of the studies included were treatment resistant or population who does not improve with existing evidence-based treatments.

Among patients with depression, current analysis indicates the significant improvement in depression with active rTMS only with large effect size but substantial heterogeneity. The high-frequency stimulation of LDLPFC was more effective as observed from the subgroup analysis with large effect size. The number of studies that used threshold or suprathreshold stimulus intensity had positive evidence, and the number of pulses per session ranges from 500 to 3000 with total of 10–20 sessions given over 2–4 weeks found useful in reducing symptoms of depression. The evidence from current analysis was contrary to the previous literature and FDA approved recommendations of rTMS for treatment-resistant depression.[66] This could be due to the significant methodological and clinical heterogeneity that was observed across the studies. We observed that there were no studies in India that tried to assess the role of TMS in suicidality neither as primary outcome nor secondary outcome suggesting the need for future studies in this area.

For schizophrenia, the most common sites of stimulation varied between low-frequency left temporo-parietal cortex for positive symptoms/auditory hallucinations[67] to high-frequency left DLPFC for negative symptoms/positive/overall symptoms with positive evidence.[68] Although a few studies used high-frequency theta burst stimulation targeting cerebellar vermis for negative symptoms, only one study found positive evidence.[27] The studies that delivered minimum of 600–2000 pulses/session with at least 10 sessions over a duration of minimum 2 weeks at either threshold or suprathreshold stimulus intensity had resulted better improvement. The meta-analysis and subgroup analysis showed significant improvements in positive symptoms, negative symptoms, auditory hallucinations, and cognition in schizophrenia with large effect sizes within active arm only both at the end of intervention and follow-up with substantial to high heterogeneity but not when compared to sham control. These findings go against with the recent international literature.[5,67,68] However, any specific recommendations are not plausible about the optimal stimulus parameters given the wide variation in the number of pulses delivered per session and in the duration of treatment and as subgroup analysis could not be performed based on these parameters.

The evidence of rTMS for OCD from the current analysis suggests rTMS either as an early augmentation or augmentation in treatment-resistant patients. There was significant improvement in OCD with larger effect sizes with low-to-moderate heterogeneity but only in the active arm from pre-to-post-intervention and until follow-up but not when compared to sham. The finding of LF rTMS targeting the SMA can significantly improve the outcomes of OCD supported by a recent meta-analysis,[69] but the other TMS protocols targeting ACC, Medial PFC, and OFC that showed positive results do warrant future research using deeper stimulation techniques and coils. The number of pulses per session ranged from 800 to 2000 using threshold or suprathreshold intensity with a total of 10–20 sessions delivered over 2 weeks to 4 weeks except for iTBS where a total of 10 sessions (600 pulses per session) were delivered over 5 days. As subgroup analysis could not be performed on stimulus parameters, we cannot emphasize any standard protocol.

The rTMS was not effective in improvement of craving in substance use disorders when compared to sham, but a significant effect of rTMS on alcohol carving only at follow-up with large effect size was noted. Contrasting target locations were used in the included studies. Although the recent meta-analysis conducted only on alcohol craving did not support the role of rTMS,[65] a previous meta-analysis had suggested potential role of high-frequency rTMS of left DLPFC on reducing overall substance craving.[70]

We emphasize that there was significant effect of rTMS on migraine prophylaxis that includes improvement of both headache severity and frequency compared to sham; the effect however could not be sustained at follow-up. The excitatory stimulation (high-frequency rTMS or iTBS) of left motor cortex was the common mode and site of stimulation in all of the studies that observed positive findings, which correspond to the finding from a recent meta-analysis.[64]

We could not find Indian studies on dementia, smoking cessation, generalized anxiety disorder, post-traumatic stress disorder, attention-deficit hyperactivity disorder, autism spectrum disorder, etc., where evidence for the positive effects of rTMS is emerging elsewhere. Perhaps, we did identify some ongoing studies on autism and cannabis use disorder from CTRI registry.

Safety and tolerability

With respect to the safety and tolerability, rTMS is considered as a safe non-invasive brain stimulation technique with no serious adverse effects apart from known side effects of headache and local site discomfort as observed in most of the studies. The reasons for discontinuation of treatment were headache and affective switch, followed by seizures. The odds of having seizure and affective switch were 4.7 and 5.8 times higher with TMS compared to sham. This necessitates the need for caution while considering the TMS with close monitoring for possible adverse effects and further research on safety of TMS with focus on TMS-EEG-based studies. As observed from the current evidence, TMS has either promising role in improvement of cognition or without any deterioration in cognitive functions particularly memory implies the possible lack of cognitive side effects as seen with ECT.[49]

Feasibility of intervention/setting

As we found that only patients who were admitted or willing to continue inpatient management during the course of TMS sessions were included in most of the studies, these findings diminish the applicability or utility of TMS in other than inpatient settings although the TMS does not involve administration of anesthesia or close monitoring as compared to electroconvulsive therapy. The complexity of the population, intervention, comparator (sham), and outcomes of the review could have resulted in the significant statistical heterogeneity for most of the outcomes measured in the current review.

Strengths and limitations

This is the first review and meta-analysis done across various psychiatric diagnoses for providing evidence for Indian population that can further help in developing guidelines of rTMS in psychiatry that is regionally relevant. Although the analysis was based on available 52 studies that were included, we were able to synthesize the quantitative data from only the nearly half of the available studies with targeted values of the outcomes that warrants more future studies with rigorous methodology. The risk of bias summary graph also suggests the possibility of publications or reporting bias as observed from the funnel plots as about 25% of the studies had selective reporting of data. The number of studies that fall under detection bias was higher compared to other domains of risk of bias indicating the chances of overestimation of results or the winner’s curse. The available number of RCTs on some disorders like anxiety disorders, substance use disorders, and childhood/adolescent disorders like ADHD were either relatively very few or not studied respectively indicating the need for future research on those disorders. The intervention complexity in terms of different TMS and sham protocols that have been used with different stimulus parameters, durations, and setting could have resulted varied results including the chances of placebo effect. The measurement of outcomes was not homogenous as evident from the different scales used in different studies that might be reflected in the statistical heterogeneity observed in the current study. Though we tried to be inclusive with thorough search of all the relevant literature, there might still be a chance that inaccessible/incompletely retrieved articles might have been missed as there was considerable number of studies that were unpublished based on CTRI search results. The follow-up durations also varied across the included studies which further could limit the conclusive evidence on longing effects of TMS on the symptom domains measured.

CONCLUSION

The overall findings from our series of meta-analyses suggest a positive, yet less sensitive, role of rTMS for migraine prophylaxis and alcohol craving but a still and silent role in other neuropsychiatric conditions that necessitates the need for more studies as well as to foster rigorous methodology in future studies from India as its superiority over sham cannot be ascertained yet. Parenthetically, we also suggest taking a parachute approach to the evidence of current analysis if there was strong empirical evidence on effects and utility of rTMS role in some psychiatry diagnosis as often poor resources in terms of availability of infrastructure, skilled person, and cost particularly can impede the large RCT’s with rigorous approaches for newer treatments like rTMS.[71]

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.

SUPPLEMENTARY MATERIAL

I. Study characteristics

Study	Sham (S)/Priming (P)/Active (A)/TAU (T)/Non (N)-controlled	Sample size		TMS make	Target location method	Coil type	Intensity (% of RMT)	Number of pulses	Sessions/Duration	Sham type	Drop outs/Analysis
		Active rTMS arms (n=56)	Sham/No rTMS control (n=31; sham=29)
Goyal et al. 2007	S	5	5	Magstim Rapid	5cm rule	Figure-of-eight	110	1000	10/2W	Angling the coil 45	No drop
Bagati et al. 2009	S	20	20	Magstim Rapid R	10-20 EEG	Figure-of-eight	90	NR	10/2W	Angling the coil 45	No drop
Praharaj et al. 2009	S	21	20	Magstim Rapid	5cm rule	Figure-of-eight	110	800	10/10D	Angling the coil 45	No drop
Mishra et al. 2010	S	30	15	Magstim Rapid	5cm rule	Figure-of-eight	110	1000	10/10D	Sham coil	10 drop/PP
Sarkhel et al. 2010	S	21	21	Magstim Rapid	5cm rule	Figure-of-eight	110	800	10/2W	Angling the coil 45	No drop
Nongpiur et al. 2011	P	18 19	- -	Magstim Rapid	5cm rule	Figure-of-eight	100; P: 90 110	900; P: 400 900	10/2W	Sham coil	3 (2+1) Lost in follow-up/PP
Jhanwar et al. 2011	N	19	-	NR	5cm rule	Figure-of-eight	110	625	20/4W	NA	2 drop/ITT
Ray et al. 2011	S	20	20	Magstim Rapid	5cm rule	Figure-of-eight	90	1200	10/10D	Angling the coil 45	1 (1+0) discontinued/PP
Lingeswaran 2011	S	9	14	Magstim Standard Rapid	5cm rule	Figure-of-eight	100	500	12/2W	Angling the coil 90	6 (3+3) Drop/PP
Kumar et al. 2011	N	12	-	Magstim Rapid sqr	10-20 EEG	Figure-of-eight	100	1000	15/3W	NA	NA
Misra et al. 2012	N	51	-	Magstim Rapid	MT hotspot	Figure-of-eight	70	584	3/6D	NA	No drop
Misra et al. 2013	S	50	50	Magstim Rapid sqr	MT hotspot	Figure-of-eight	70	600	3/6D	NA	5 (3+2) discontinued/PP
Mishra et al. 2015	A (Right vs Left)	10 10	- -	Magstim Rapid	5cm rule	Figure-of-eight	110	1000	10/10D	NA	No drop
Pathak et al. 2015	S	13	13	Magstim Rapid	5cm rule	Figure-of-eight	110	800	10/10D	Angling the coil 45	No drop
Ray et al. 2015	P	20 20	- -	Magstim Rapid	10-20 EEG	Figure-of-eight	100; P: 100 100	1200; P: 600 1200	10/2W	Sham coil	5 (3+2) discontinued/PP
Kalita et al. 2016	A (3 vs 1 sessions)	39 37	- -	NR	MT hotspot	Figure-of-eight	70	600	3/6D	Sham coil	22 (13+9)/PP
Garg et al. 2016	S	20	20	Magstim Rapid	10-20 EEG	Double cone	100	600	10/2W	Sham coil	7 (4+3) discontinued/PP
Jha et al. 2016	N	20	-	Magstim super Rapid	SPECT guided	Figure-of-eight	110	800	20/4W	NA	2 drop/ITT
Tikka et al. 2017	S	8	7	Magstim Rapid sqr	sMRI guided	Figure-of-eight	80	900	10/2W	Sham coil	5 (2+3) discontinued/PP
Arumugham et al. 2018	S	19	17	Magventure MagPro R100	10-20 EEG	Figure-of-eight	100	1200	18/3W	Sham coil	2 (1+1) discontinued/PP
Kumar et al. 2018	N	25	-	Magstim Rapid sqr	10-20 EEG	Figure-of-eight	110	1200	20/4W	NA	NA
Kumar et al. 2018	N	14	-	Magstim Rapid sqr	5cm rule	Figure-of-eight	110	1000	15/3W	NA	NA
Verma et al. 2018	N	22	-	NR	5cm rule	Figure-of-eight	110	3000	15/3W	NA	NA
Mattoo et al. 2019	S	15	15	Neurosoft, NeuroMS/D	5cm rule	Figure-of-eight	110	1200	20/4W	Angling the coil 90	No drop
Singh et al. 2019	N	79	-	Magstim Rapid sqr	10-20 EEG	Figure-of-eight	110	1200	20/4W	NA	NA
Sahu et al. 2019	S	20	21	Magstim Rapid sqr	5cm rule	Figure-of-eight	80	600	10/2W	Angling the coil 90	11 dropped/PP
Tanwar et al. 2020	S	45	41	Magventure MagPro R100	5cm rule	Figure-of-eight	90	1200	20/4W	Angling the coil 45	4 dropped/PP
Kumar et al. 2020	S	50	50	Magstim Rapid sqr	5cm rule	Figure-of-eight	100	2000	20/4W	Sham coil	7 (4+3) dropped/ITT
Singh et al. 2020	S	15	15	Magstim Rapid sqr	5cm rule	Figure-of-eight	100	2000	20/4W	Sham coil	4 (3+1) dropped/ITT
Sharma et al. 2020	S	47	49	Magstim Rapid sqr	10-20 EEG	Figure-of-eight	110	750	10/2W	Angling the coil 90	4 (3+1) dropped/ITT
Basavaraju et al. 2021	S	30	30	Magventure MagPro X100	sMRI guided	Figure-of-eight	100	600	10/5D	Sham coil	11 lost to follow-up/ITT
Kalita et al. 2021	A (rTMS alone vs rTMS with Amitriptyline)	37 39	- -	Magstim Rapid sqr	MT hotspot	Figure-of-eight	70	600	3 + 3/6D + 3M	NA	7 (4+3) dropped/ITT
Dutta et al. 2021	S	18	15	Magventure MagPro R30- TB booster	10-20 EEG	Figure-of-eight	80	600	10/5D	Sham coil	5/PP
Chauhan et al. 2021	S	19	17	Magventure MagPro R30- TB booster	10-20 EEG	Figure-of-eight	80	600	10/5D	Sham coil	4 (2+2) dropped/ITT
Kumar et al. 2021	S	10	10	Neurosoft, NeuroMS/D	MT hotspot	Figure-of-eight	70	600	10/2W	Angling the coil 90	No drop
Gupta et al. 2021	T	20	19	Magstim Rapid sqr	10-20 EEG	Figure-of-eight	100	1200	10/2W	NA	1 drop/PP
Gupta et al. 2021	T	50	50	NR	NR	NR	90	600	NR	NA	No drop
Tyagi et al. 2021	S	30	29	Magventure MagPro R30- TB booster	10-20 EEG	Figure-of-eight	80	600	20/2W	Sham coil	9 (5+4) dropped; ITT
Thirthalli et al. 2022	S	15	13	Magventure MagPro R30	sMRI guided	Figure-of-eight	100	2000	NR	NR	NA
Ankit et al. 2022	S	20	20	NR	10-20 EEG	Figure-of-eight	80	900	14/2W	Angling the coil 90	No drop
Vidya et al. 2022	P	15 15	- -	Magventure MagPro R30	10-20 EEG	Figure-of-eight	100; P: 80 100	1200; P: 600 1200	10/2W		3 drop/PP
Joshi et al. 2022	Y	13	11	Medicaid MediStim (MS-30)	10-20 EEG	Figure-of-eight	100	1600	20/23D	Sham coil	4 dropped/PP
Mallik et al. 2022	Y	11	8	Magventure MagPro R30- TB booster	10-20 EEG	Figure-of-eight	80	600	15/5D	B-65 sham coil	2 drop/ITT
Reddy et al. 2022	N	15	-	PowerMag 100, Brainsway	In-built	H-7dTMS	100	2000	10/2W	NA	No drop
Batra et al. 2022	N	10	-	Magstim Rapid sqr	MT hotspot	Figure-of-eight	80	1200	1	NA	No drop
Kumar et al. 2018	N	13	-	Magstim Rapid	5cm rule	Figure-of-eight	110	1000	20/4W	NA	No drop
Shah et al. 2022	S	36 36	36	Magstim Rapid	5cm rule	Figure-of-eight	70 70	500 584	3/6D 3/6D	30%RMT; 100 pulses	No drop
Syed et al. 2021	N	12	-	MagPro, Magventure	sMRI/10-20 EEG	D-B80 coil	90-100	600	6 to 26/NR	NA	2
Baliga et al. 2020	N	83	-	NR	7cm rule	NR	90-100	1800+1800	4 to 38/NR	NA	4
Agrawal et al. 2021	N	5	-	NR	NR	NR		1200	20/4W	NA	No drop
Shere et al. 2021	N	23	-	NR	7cm rule	NR	80	1800+1800	10/10D	NA	2
Raikwar et al. 2020	S	30	30	ATES MEDICA, EBNeuro	5cm rule	Figure-of-eight	120	800	10/10D	Switch off; circular coil on	No drop
Total		1448	701

II. Quality/Risk of Bias Assessment

NRS (Non-Randomized Studies)	Selection Domain				Comparability Domain Comparability/confounding	Outcome/Exposure Domain
	Representativeness of cases/cohorts	Selection of comparative arm	Ascertainment of treatment	Demonstration that outcome not present at start of study		Assessment of outcome	Follow-up long enough	Adequacy of follow-up	Total score	Quality	Quality (AHRQ Standards)
	Jhanwar VG 2011	★		★		★		★	★	★	6.00	moderate	Poor
Kumar N 2011	★		★			★	★	★	5.00	moderate	Poor
Misra UK 2012	★		★	★			★	★	5.00	moderate	Poor
Jha S 2016	★		★	★		★	★	★	6.00	moderate	Poor
Kumar S 2018_Migraine	★		★				★	★	4.00	moderate	Poor
Kumar S 2018_OCD	★		★			★	★	★	5.00	moderate	Poor
Kumar S 2018	★		★			★	★	★	5.00	moderate	Poor
Verma R 2018	★		★			★	★	★	5.00	moderate	Poor
Singh S 2019	★		★		★	★	★	★	6.00	moderate	Fair
Baliga SP 2020	★		★		★			★	4.00	moderate	Fair
Shere SS 2021	★		★	★		★	★	★	6.00	moderate	Poor
Agrawal A 2021			★	★				★	4.00	moderate	Poor
Syed FA 2021	★		★	★		★			3.00	low	Poor
Reddy S 2022	★		★	★		★	★		5.00	moderate	Poor
Case control study	Definition of case adequate	Representativeness	Selection of control	Definition of control	Comparability	Ascertainment of exposure	Same method of ascertainment	Non-response rate
Batra D 2022	★	★	★	★	★			★	6.00	moderate	Fair

SUPPLEMENTARY TABLE II

Quality Scoring as per scale-(High quality:7-9, Moderate quality:4-6, Low quality:0-3)

Except 1 study that falls under low quality, remaining has moderate quality

Converting the Newcastle-Ottawa scales to AHRQ standards

3 studies are having fair quality and remaining 12 studies has poor quality

(Good quality: 3 or 4 stars in selection domain AND 1 or 2 stars in comparability domain AND 2 or 3 stars in outcome/exposure domain

Fair quality: 2 stars in selection domain AND 1 or 2 stars in comparability domain AND 2 or 3 stars in outcome/exposure domain

Poor quality: 0 or 1 star in selection domain OR 0 stars in comparability domain OR 0 or 1 stars in outcome/exposure domain)

Risk of Bias assessment according to Cochrane ROB2 Summary of evidence and graph

High quality studies: 10, Fair quality: 14, Poor quality: 13

Converting the Cochrane Risk of Bias Tool to AHRQ Standards

Good quality: All criteria met (i.e. low for each domain) or one criteria unclear but unlikely to effect the outcome based on judgment of the reviewer

Fair quality: One criterion not met (i.e. high risk of bias for one domain) or two criteria unclear, and the assessment that this was unlikely to have biased the outcome, and there is no known important limitation that could invalidate the results

Poor quality: One criterion not met (i.e. high risk of bias for one domain) or two criteria unclear, and the assessment that this was likely to have biased the outcome, and there are important limitations that could invalidate the results OR Two or more criteria listed as high or unclear risk of bias

III. All Studies with available follow-up data

Authors	Year	Condition	Outcomes	Included for meta-analysis (data of 2 or more studies available)
Mishra et al	2010	ADS	Craving
Nongpiur et al	2011	Depression * 2	Depression
Misra et al	2012	Migraine	Headache severity
Ray et al	2015	Schizophrenia * 2	Positive symptoms
			Negative symptoms
			Total psychopathology
			Auditory hallucinations (AVHRS)
Garg et al	2016	Schizophrenia	Positive symptoms
			Negative symptoms
			Total psychopathology
			Depression (CDSS)
Kumar et al	2018	OCD	YBOCS
Kumar et al	2018	Migraine	Frequency
			Severity
			MIDAS
			HAMD
Sahu et al	2019	Migraine	Frequency
			Severity
			MIDAS
Kumar et al	2020	Schizophrenia	Positive symptoms
			Negative symptoms (SANS)
			Negative symptoms (NS)
			Total psychopathology
Basavaraju et al	2021	Schizophrenia	Negative symptoms (SANS)
			Depression (CDSS)
Dutta et al	2021	OCD	OCS (YBOCS)
			Depression (HAMD)
Chauhan et al	2021	Schizophrenia	Positive symptoms
			Negative symptoms
			Total psychopathology
			Cognition
Kumar et al	2021	Migraine	Frequency
			Severity
Tyagi et al	2022	Schizophrenia	Positive symptoms
			Negative symptoms
			Total psychopathology
			Auditory hallucinations (AVHRS)
			Auditory hallucinations (PSYRATS-AH)
			Depression (CDSS)
			Cognition
Mallik et al	2022	Bipolar Depression	Depression (HAMD)
Vidya et al.	2022	OCD * 2	OCS (YBOCS)
			Depression (HAMD)
Shere et al.	2021	Depression	CDRS
Raikwar et al	2020	ADS	ACQ
Sham-controlled studies with available follow-up data
Authors	Year	Condition	Outcomes	Included for meta-analysis (data of 2 or more studies available)
Mishra et al	2010	Alcohol dependence syndrome	Craving
Misra et al	2012	Migraine	Headache severity
Garg et al	2016	Schizophrenia	Positive symptoms
			Negative symptoms
			Total psychopathology
			Depression (CDSS)
Sahu et al	2019	Migraine	Frequency
			Severity
Kumar et al	2020	Schizophrenia	Positive symptoms
			Negative symptoms (SANS)
			Negative symptoms (NS)
			Total psychopathology
Basavaraju et al	2021	Schizophrenia	Negative symptoms (SANS)
			Depression (CDSS)
Dutta et al	2021	OCD	OCS (YBOCS)
			Depression (HAMD)
Chauhan et al	2021	Schizophrenia	Positive symptoms
			Negative symptoms
			Total psychopathology
			Cognition
Kumar et al	2021	Migraine	Frequency
			Severity
Tyagi et al	2022	Schizophrenia	Positive symptoms
			Negative symptoms
			Total psychopathology
			Auditory hallucinations (AVHRS)
			Auditory hallucinations (PSYRATS-AH)
			Depression (CDSS)
			Cognition
Mallik et al	2022	Bipolar depression	Depression (HAMD)
Raikwar et al	2021	ADS	Craving

IV. Adverse effects

Study	Year	Headache		Scalp pain/discomfort/irritation		Dizziness		Seizure	Switch	Other		Drop
		A	S	A	S	A	S
Goyal et al	2007	2	0	0	0	0	0	0	0	0	0	0
Bagati et al	2009						Not reported					0
Praharaj et al	2009	6	0	0	0	1	0	0	0	First session anxiety (5)	0	0
Mishra et al	2010	5	0	1	0	0	0	1	0	First session anxiety (4)	0	2 (1 seizure; 1 scalp pain)
Sarkhel et al	2010	3	0	0	0	0	0	0	0	0	0	0
Nongpiur et al	2011	2		4		0		0	0	0		0
		1		3		0		0	0	Temporal Twitch (1)		0
Jhanwar et al	2011	4		4
Ray et al	2011	1	0	5	0					Low hearing (1)		2 (scalp pain & fear)
Lingeswaran	2011	1	0									1 (headache)
Kumar et al	2011	0		2					1			1 (switch)
Misra et al	2012	3		0		0				Rhinorrhoea (1)
Misra et al	2013									Drowsiness (1) Facial pain (50)	Facial pain (7)
Mishra et al	2015									Nightmare + insomnia (1)
Pathak et al	2015	2	0
Ray et al	2015						Not reported
Kalita et al	2016						Not reported
Garg et al	2016	5	0							Drowsiness (1)
Jha et al	2016	3
Tikka et al	2017	2	2	2	2
Arumugham et al	2018						Not reported
Kumar et al	2018	3	0	4	0
Kumar et al	2018	0	0	0	0
Verma et al	2018						Not reported
Mattoo et al	2019						Not reported
Singh et al	2019	10	0	13	0
Sahu et al	2019						Not reported
Tanwar et al	2020	2	0			0	1			Neck pain (2)
Kumar et al	2020	0	0	0	0	0	0	1				1 (seizure)
Singh et al	2020	3	4	5	4		0	1		Facial twitch (1)		1 (seizure)
Sharma et al	2020	0	0	0	0	0	0	1				1 (seizure)
Basavaraju et al	2021	0	0	0	0	0	0	0	2	Neck contraction (1)		2 (switch)
Kalita et al	2021						Not reported
Dutta et al	2021	3	1
Chauhan et al	2021	5	2
Kumar et al	2021	0	0
Gupta et al	2021	0	0
Gupta et al	2021						Not reported
Tyagi et al	2022	5	4
Thirthalli et al	2022						Not reported
Ankit et al	2022						Not reported
Vidya et al	2022	0
		0
Joshi et al	2022	5	4
Mallik et al	2022	1	1
Reddy et al	2022	4		5						Insomnia (3)
Batra et al	2022						Not reported
Kumar et al	2018	1		2
Shah et al	2022					10	6
						7
Syed et al	2021	1		1								1 (scalp pain and headache)
Baliga et al	2020	0		2								2 (scalp pain)
Agrawal et al	2021	0	0
Shere et al	2021	8		3					2	Neck pain (4), lacrimation (1)		2 (switch)
Raikwar et al	2020	4	1	1	0					Drowsiness (4)

V. Sensitivity Analysis

Outcome	Pre-post	Active-Sham
Any depression	SMD 6.83 (P<.001); I2=82%	SMD 1.26 (P=0.21); I2=72%
	Excluding: Nongpiur et al., Sarkhel et al., Shere et al., Goyal et al	Excluding: Sarkhel et al., Goyal et al
Depression (unipolar, bipolar)	SMD 5.60 (P<.001); I2=79%	-
	Excluding: Nongpiur et al., Shere et al	-
Depression in OCD	SMD 2.53 (P=0.01); I2=31%	SMD 0.11 (P=0.91); I2=0% (Significance of effect reduced from P<.001-<.01)
	Excluding: Sarkhel et al	Excluding: Sarkhel et al
Depression in schizophrenia	-	SMD 0.46 (P=0.64); I2=0%
	-	Excluding: Goyal et al
Schizophrenia (Positive symptoms)	SMD 4.12 (P<.001); I2=59%	-
	Excluding: Gupta et al	-
Schizophrenia (Negative symptoms)	SMD 5.53 (P<.001); I2=83%	-
	Excluding: Gupta et al	-
Schizophrenia (Total symptoms)	SMD 3.83 (P<.001); I2=87%	SMD 0.38 (P=0.71); I2=37%
	Excluding: Gupta et al., Thirthalli et al	Excluding: Thirthalli et al
Schizophrenia (Cognition)	SMD 2.46 (P<.01); I2=0%	SMD 0.06 (P=0.95); I2=0%
	Excluding: Thirthalli et al	Excluding: Thirthalli et al
OCD	SMD 6.16 (P<0.001); I2=65%	SMD 0.84 (P=0.40); I2=53%
	Excluding: Sarkhel et al	Excluding: Sarkhel et al
Substance use disorder- craving	SMD 3.68 (P<0.001); I2=94%	SMD 1.22 (P=0.22);I2=0% Heterogeneity lost significance
	Excluding: Mishra et al. 2010	Excluding: Mishra et al. 2010
Headache- severity	SMD 2.56 (P=0.01); I2=97% (Significance of effect reduced from P<.001-<.05)	SMD 1.95 (P=0.05); I2=79% Significance of effect lost
	Excluding: Sahu et al	Excluding: Sahu et al
Headache- frequency	SMD 5.40 (P<.001); I2=77%	SMD 2.36 (P=0.02); I2=91% (Significance of effect reduced from P<.01-<.05)
	Excluding: Sahu et al	Excluding: Sahu et al