Electroconvulsive Therapy Added to Non-Clozapine Antipsychotic Medication for Treatment Resistant Schizophrenia: Meta-Analysis of Randomized Controlled Trials

This meta-analysis of randomized controlled trials (RCTs) examined the efficacy and safety of the combination of electroconvulsive therapy (ECT) and antipsychotic medication (except for clozapine) versus the same antipsychotic monotherapy for treatment-resistant schizophrenia (TRS). Two independent investigators extracted data for a random effects meta-analysis and pre-specified subgroup and meta-regression analyses. Weighted and standard mean difference (WMD/SMD), risk ratio (RR) ±95% confidence intervals (CIs), number needed to treat (NNT), and number needed to harm (NNH) were calculated. Eleven studies (n = 818, duration = 10.2±5.5 weeks) were identified for meta-analysis. Adjunctive ECT was superior to antipsychotic monotherapy regarding (1) symptomatic improvement at last-observation endpoint with an SMD of -0.67 (p<0.00001; I2 = 62%), separating the two groups as early as weeks 1–2 with an SMD of -0.58 (p<0.00001; I2 = 0%); (2) study-defined response (RR = 1.48, p<0.0001) with an NNT of 6 (CI = 4–9) and remission rate (RR = 2.18, p = 0.0002) with an NNT of 8 (CI = 6–16); (3) PANSS positive and general symptom sub-scores at endpoint with a WMD between -3.48 to -1.32 (P = 0.01 to 0.009). Subgroup analyses were conducted comparing double blind/rater-masked vs. open RCTs, those with and without randomization details, and high quality (Jadad≥adadup analyses were Jadad<3) studies. The ECT-antipsychotic combination caused more headache (p = 0.02) with an NNH of 6 (CI = 4–11) and memory impairment (p = 0.001) with an NNH of 3 (CI = 2–5). The use of ECT to augment antipsychotic treatment (clozapine excepted) can be an effective treatment option for TRS, with increased frequency of self-reported memory impairment and headache. Trial registration CRD42014006689 (PROSPERO).

ECT has shown to be an effective and safe augmentation in TRS including clozapine-resistant schizophrenia [11,12]. It has been recommended for TRS by the Task Force Report of the American Psychiatric Association on ECT [13]. A number of randomized controlled trials (RCTs) [14][15][16][17][18][19][20][21][22][23][24] compared the efficacy and safety of ECT combined with antipsychotic medication other than clozapine with the same antipsychotic monotherapy. So far, the results have been conflicting.
A systematic review of 22 RCTs concluded that adjunctive ECT added to any type of antipsychotics could produce symptomatic improvement in TRS [25]. In another review of 12 RCTs Tharyan et al. [26] suggested that ECT combined with all types of antipsychotics may be an effective treatment for schizophrenia.
To the best of our knowledge no meta-analyses on ECT added to non-clozapine antipsychotics for TRS have been published. The present meta-analysis of RCTs evaluates the efficacy and safety of ECT added to non-clozapine antipsychotic medications for TRS, comparing this combination to antipsychotic monotherapy. In addition to international literature sources, Chinese databases that are not usually reviewed in the international literature were also searched.

Methods
This meta-analysis is based on the methodology recommended by the Cochrane Collaboration [27] and prepared according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [28] (S1 Fig). psychopathology of the Positive and Negative Syndrome Scale (PANSS) [29], or the Brief Psychiatric Rating Scale (BPRS) [30]. Key secondary outcomes included early (at 1 to 2 weeks) symptomatic improvement, PANSS positive, negative and general psychopathology sub-scores, response and remission defined by individual studies, patient-reported ADRs during the study period, and all-cause discontinuation.

Search methods
Major English (PubMed, PsycINFO, Embase, Cochrane Library databases, the Cochrane Controlled Trials Register) and Chinese databases (WanFang Database, Chinese Biomedical Database and China Journal Net) were searched from their inception until November 10, 2015 for RCTs using the following search terms: Electroconvulsive/Electroconvulsive therapy, schizophrenia, randomized controlled trial, placebo, and trial. The keywords were used in combination with the Boolean operators AND, OR, and NOT. The search was supplemented by using the "related article" function. Reference lists of eligible studies and relevant review articles were hand-searched. Authors were contacted for unpublished data if necessary.

Data extraction
Selection of studies, data extraction and synthesis and assessment of bias were conducted independently by two authors (WZ and XLC). All information was checked by another author (YQX). Inconsistencies were resolved by discussion.

Assessment of risk of bias
The methodological quality of RCTs was assessed by risk of bias [31] rating the method of random sequence generation (selection bias), allocation concealment (selection bias), blinding of participants and personnel (performance bias), blinding of outcome assessment (detection bias), incomplete outcome data (attrition bias), selective reporting (reporting bias) and other biases. Each domain was rated as "high risk", "unclear risk", or "low risk" [31]. In addition, the Jadad scale (range: 0-5) was used to assess study quality in five domains: "randomization," "double blinding," "description withdrawals and dropouts," "generation of random numbers," and "allocation concealment" [32]. RCTs were classified as high-quality when their Jadad total score was 3 and low quality when their Jadad score was <3 [33]. The GRADE system (S1 Table) was used to rate the quality of primary and secondary outcomes of the meta-analysis [34,35].

Data synthesis and statistical analyses
In order to combine studies, the random effects model [36] was used in all cases with the aid of the Review Manager Version 5.3 software (http://www.cochrane.org). Weighted or standard mean difference (WMD/SMD) and risk ratio (RR) with 95% confidence intervals (CIs) were calculated for continuous and dichotomous data, respectively. When RR was significant (p<0.05), the number needed to treat (NNT) or number needed to harm (NNH) were calculated. All statistical differences were considered significant when p<0.05.
In case of I 2 50% for the effect of primary outcomes, a sensitivity analyses was conducted after removing the three studies [14,19,23] that had an outlying effect size of SMD>-1.0. Furthermore, 5 comparative subgroup analyses were also conducted to identify potential moderators or mediators. These subgroup analyses were (1) Chinese vs. non-Chinese studies; (2) double blind/rater-masked vs. non-blinded studies; (3) duration of treatment <12 vs. 12 weeks; (4) mean number of ECT sessions <9 ECTs vs. 9 (since the number of ECT sessions was reported inconsistently, the median split of the number of recommended ECT sessions according to the ECT guidelines for schizophrenia in China were used, i.e., 6 to 12 sessions [37]; (5) high vs. low quality studies (3 vs. <3); and (6) co-starting vs. augmenting with antipsychotic. Since 2 RCTs [21,24] provided data only for key secondary outcomes, subgroup analyses of endpoint symptomatic improvement were conducted with the remaining 9 RCTs.
Additionally, a meta-regression analysis was performed involving sample size, trial duration, Jadad score, mean age, percentage of male patients and illness duration to identify potential moderators or mediators of the effect on endpoint symptomatic improvement using the STATA Version 12.0 software (http://www.stata.com).

Literature search
A total of 701 potentially relevant articles in the initial database search (698 trials) and other sources (3 trials) were ascertained. Eventually 11 RCTs met the selection criteria for meta-analysis (Table 1

Assessment of risk of bias
While 6 RCTs with a specific description regarding random sequence generation were rated as low risk, 6 RCTs were rated as high risk for allocation concealment. Masked assessors and double blindness were administered in 4 and 1 RCT, respectively. Regarding outcome data, 1 RCT reported loss to follow-up, but failed to use ITT analysis for incomplete outcome data. In addition, none of the studies registered their protocol, preventing formal assessment of reporting bias. Other biases were rated as low risk in all RCTs (S3 Fig).

Discussion
In this meta-analysis of 11 RCTs (n = 818) comparing add-on ECT to non-clozapine antipsychotics with antipsychotic monotherapy for TRS, the combination treatment was superior in terms of the primary and key secondary efficacy outcomes. Importantly, adjunctive ECT was both safe and well tolerated. In terms of the primary outcome measurement, adjunctive ECT was significantly superior to antipsychotic monotherapy with a medium effect size; according to Cohen [41] criteria, SMD of -0.2, -0.5, and -0.8 were defined as small, medium and large effect size, respectively. While the heterogeneity of the primary outcome result was 62%, the resultant effect size was still -0.44 and heterogeneity decreased to 0% after conducting a sensitivity analyses. The result was the same across all but one a priori defined subgroup analyses, namely non-Chinese studies. Subgroup analyses did not reveal a significant difference between the Chinese (SMD = -0.63, 95%CI:-0.93, -0.33) and non-Chinese studies (SMD = -0.92, 95%CI:-1.99, 0.14). In metaregression analyses, pre-specified baseline moderators or mediators of primary outcomes could not be identified. Publication bias was not detected by the funnel plot and Egger's test.
Adjunctive ECT was significantly superior to antipsychotic monotherapy with a moderate effect size of -0.58 after 1 to 2 weeks, which increased to the larger effect size of -0.67 after 8 weeks suggesting an overall reliability of the results. The majority of patients responded better to adjunctive ECT than to antipsychotic monotherapy: 50.9% vs. 32.9%, NNT = 6. The remission rates similarly favored adjunctive ECT: 21.0% vs. 9.5%, NNT = 8. While the optimal number of ECT sessions for schizophrenia remains unclear, 12 to 20 sessions have been shown to be adequate [42,43]. The number of ECT sessions were 14.2±8.0 (range = 7.6-36.0, median = 8.0) in ten studies. Thus, an ECT course of 14 sessions appears to be reasonable to target the symptoms of schizophrenia. Of note, in seven of the eleven studies, the electrode placement was reported. In all seven of these studies, bilateral placement was utilized.
In a recent meta-analysis on adjunctive ECT combined with any type of antipsychotics for TRS (22 RCTs), 11 RCTs focused on concurrent use of ECT with non-clozapine antipsychotics [25]. However, the authors did not examine the separate effect of this strategy. Another systematic review concluded that ECT may be an effective and safe augmentation strategy to clozapine in TRS, but the effects of ECT combined with non-clozapine were not examined [12].
Several limitations of this meta-analysis need to be mentioned. First, blinding methods for raters were only reported in 4 studies and only 7 of the 11 RCTs were rated as high quality. Second, there was statistical heterogeneity of endpoint symptomatic improvement resulting from methodological and clinical heterogeneity. This limitation was partly offset by one sensitivity analysis, 5 subgroup analyses and 7 meta-regression analyses to identify potential moderators or mediators of the effect on primary outcome. Third, most studies had relatively short duration (median = 8.0 weeks). Fourth, although serious adverse effects, such as confusion, could occur after ECT, apart from the ADRs mentioned above, no serious adverse effects were reported in the studies. Finally, few objective measures evaluated neurocognitive functioning. Only 2 RCTs [17,24] tested cognitive functions with the Wechsler Memory Scale without data for the control group in one RCT [24]; 1 RCT [14] applied the modified Mini-Mental State Examination; one RCT [16] used the Wisconsin Card Sorting Test.

Conclusions
This systematic review and meta-analysis of 11 RCTs with 818 patients concluded that ECT added to a non-clozapine antipsychotic medication for TRS is more effective than antipsychotic monotherapy. The ECT/non-clozapine antipsychotic combination is safe and tolerable. Future studies should examine clinical factors that predict efficacy and tolerability of ECT for patients with schizophrenia.