Introduction

Suicide is a major public health problem worldwide, with more than 800,000 deaths and about 16 million attempted suicides worldwide each year, and the incidence of suicide has steadily increased over the past 20 years [1,2,3]. A large number of acute suicidal depression patients are admitted to the emergency room each year [4]. In the absence of specific treatment for acute suicidal ideation, suicidal patients are often hospitalized only briefly to stabilize their condition and are discharged before psychopharmacological treatment shows efficacy. And for people with suicidal depression, only a third or less achieve remission, and less than half achieve 50% remission with typical first-line drugs [5]. For suicidal patients who need to be treated quickly, even before a thorough diagnosis is established, it is important to identify a drug that can provide rapid relief and reduce suicidal ideation.

Current treatments to reduce suicidal ideation and behavior occurrence, such as antidepressants, lithium, clozapine, psychotherapy, electroconvulsive therapy, and transcranial magnetic stimulation, are numerous, but are not immediately effective for all patients. Standard antidepressants have proven effective in treating suicidal ideation, but they take a relatively slow time to work, usually four to six weeks for optimal results. In addition, selective serotonin reuptake inhibitors warn of a potential increase in suicidal ideation in children and adolescents [6].

A growing body of research suggests that ketamine and esketamine may rapidly reduce suicidal ideation in depressed subjects at risk for suicide [7,8,9,10,11]. There have also been studies showing that even the most resistant patients can benefit from ketamine, and that medium to long-term treatment is effective for many patients [12].

The purpose of our study was to determine whether ketamine can rapidly improve the reduction of suicidal ideation behavior in patients with depression in the short term, and to explore the change in the intensity of its effect over time.

Methods

We performed network meta-analyses(NMA) and pairwise meta-analyses to compare the efficacy of ketamine in the reduction of suicidal ideation. NMA compares the efficacy of drugs at different time points. The effect intensity between the two can be directly compared, and the probability ranking prediction analysis and comparison of the strength of the therapeutic effect. A pairwise meta-analysis was performed to compare the effect of the first and last treatment in the repeated dosing group. The difference between the first and the last treatment was used to determine whether repeated dosing could bring greater benefits to ketamine treatment.

Search strategy

The retrieval strategy takes ketamine, suicide and depression as search terms, and searches their subject words and all entry terms. Combined with subject terms and entry terms, a comprehensive literature search was conducted in Cochrane Library, Embase, Web of science, CNKI, PubMed to retrieve all studies published in the last ten years up to August 1, 2022.

Study selection

Outpatients and inpatients aged between 18 and 70 years with depression as their primary psychiatric diagnosis. The included studies can be RCT studies or single-arm studies, which should meet the requirement that the enrolled patients have been assessed by strict relevant scales and have definite suicidal ideation (C-SSRS, CGI-SS-r, BSS, MADRS-SI, SSI part I). Studies with no baseline suicidal ideation rating data were excluded. Excluded were studies in which multiple therapies were used in combination with therapy (e.g., Physical therapy combined with drug therapy) and in which medication was changed during the study, as well as studies in which patients were enrolled with primary psychotic disorders (e.g., schizophrenia, schizoaffective disorder, etc.), dissociative disorders, pervasive developmental disorders, cognitive disorders, or anorexia nervosa.

We included different ketamine-related studies, including all elastic dose studies, which allow researchers to titrate to an adequate dose appropriate for individual patients. For fixed-dose studies, we included the target dose to the maximum dose based on the international consensus study on antipsychotic dosing [13]. Due to the limited number of included studies, for studies with single and repeated ketamine administration, we analyzed them together.

Network meta-analysis

Unlike paired meta-analyses, which allow one intervention to be compared with another on the basis of head-to-head data from randomized trials, NMA facilitate simultaneous comparisons of the efficacy or safety of multiple interventions that may not be directly compared in randomized trials [14, 15].

Quality assessment

Our analysis included all published RCTs and single-arm studies. For RCTs, we applied the Cochrane Risk of Bias tool 2 (RoB 2) to assess the risk of bias [16]. For single-arm studies, we applied the Methodological Index for Non-Randomized Studies (MINORS) score to assess the risk of bias [17]. Study quality was assessed independently by two investigators (ZS and DG), strictly following Cochrane RoB 2 and MINORS bias risk assessment method. If there are disagreements, resolve them through discussion.

Data extraction

The primary outcome measure of our study was the level of remission of suicidal ideation after medication in depressed patients with definite suicidal ideation. Therefore, we extracted all relevant binary and continuous outcome data from baseline to endpoint. However, due to the inconsistencies of the suicidal ideation assessment scales used in most studies, and relatively few studies meeting the criteria included, we have to give up the processing of the continuous outcome data. By redefining whether patients achieve effective remission of suicidal ideation, the results of binary data under different scales are unified (that is, the level of suicidal ideation of patients is reduced to (SSI) part I score < 2, QIDS-SR16 < 1,CGI-SS: 0 or 1, MADRS-SI ≤ 2 or SSI total score ≤ 3 were considered as effective mitigation, and the binary outcome data were extracted by judging whether effective mitigation was achieved or not). All the data is done and screened independently by the two authors (ZY S and DQ G), who cannot interfere with each other’s decisions. When there is disagreement between two authors, the third author (XL) clarifies and makes the final decision.

Data processing

Data transformation is carried out for studies containing data of multiple time nodes. For example, if a study has relevant data for a treatment at multiple time points, we will treat the data at each time point as a unique intervention. The previous study was converted into a multi-arm study on which we conducted the NMA (Supplementary Fig. 1).

Data analysis

The data were analyzed from 20 August 2022 to 10 September 2022. We performed pairwise meta-analysis and NMA in a Bayesian environment using Stata 15 and R 4.1.3 software. For pairwise meta-analysis and NMA, we flexibly selected fixed effects model and random effects model according to the heterogeneity between studies, and determined the heterogeneity of meta-analysis by I2 (threshold 50%). We extracted binary results in our analysis and calculated the effect size by relative ratio (RR) of the number of people with effective mitigation from different studies. The NMA synthesizes direct and indirect evidence of relevant therapeutic effects across all time nodes to provide an estimate of maximum efficacy. We compared the effects of different strategies using Surface under the Cumulative Ranking Curve (SUCRA), allowing for a comparative analysis of the intensity of drug action at each time node based on the surface under the cumulative ranking. SUCRA values range from 0 to 100%. The higher the SUCRA value, and the closer to 100%, the higher the likelihood that a therapy is in the top rank or one of the top ranks; the closer to 0 the SUCRA value, the more likely that a therapy is in the bottom rank, or one of the bottom ranks [15, 18]. The inconsistencies between direct evidence and indirect evidence were calculated by node splitting method. Publication bias was evaluated by funnel plot and Egger’s test. All statistical differences were considered significant when the p < 0.05.Subgroup analyses were performed to explore heterogeneity.

Results

Descriptions of the included studies

A total of 505 studies published before August 2022 were retrieved. After reading titles and abstracts, 369 studies were excluded. Reasons for included duplication, review articles, individual case reports, and animal experiments. 136 studies were included in the full text analysis, and 14 studies meeting the requirements were finally identified (Table 1). There were 1380 participants in 14 studies. Among them, 10 were RCT studies and 4 were single-arm studies. The mode of administration in these studies included both single and repeated doses. We conducted a pairwise meta-analysis of first-to-last treatment in six studies of repeated dosing to determine whether it is superior to a single dose. The PRISMA flowchart is shown in Fig. 1. The risk assessment of bias for RCTs and non-RCTs is shown in supplementary materials Tables 1 and 2.

Table 1 Included studies list.
Full size table
Fig. 1: PRISMA diagram.
figure 1

Flow diagram showing the study selection process.

Full size image

Network meta-analysis

We conducted a comprehensive comparison of the alleviating effect of suicidal ideation at different time points after the administration of different treatment strategies, obtained the direct and indirect comparison results between different treatment measures at each time point through the method of NMA, and conducted a comprehensive ranking comparison of the remission degree at different time points. A total of 14 studies with 1380 participants were included. The overall characteristics of the included studies are shown in Fig. 2.

Fig. 2: Overall efficacy network evidence figure.
figure 2

The size of the node corresponds to the number of participants assigned to each treatment. Treatments with direct comparisons are connected by a line; Its thickness corresponds to the number of tests compared. D1: Day 1, D3: day 3, D13: day 13, D26: day 26.

Full size image

After the first day of treatment, ketamine was significantly more effective than placebo (RR = 10.02, 95%CI = 4.24–23.68), esketamine (RR = 4.81, 95%CI = 1.56–14.84), and midazolam (RR = 3.14, 95%CI = 1.19–8.29). In addition, ketamine was significantly better than placebo at all the same time points. Ketamine D3 (RR = 2.89, 95%CI = 1.04–8.00), Ketamine D26 (RR = 4.29, 95%CI = 1.41–13.08). In the placebo group, the results on day 3 (RR = 0.20, 95%CI = 0.06–0.63) and day 26 (RR = 0.13, 95%CI = 0.06 to 0.27) were also significantly different from those on day 1. The effects of esketamine and midazolam were not significant at the same time points compared with placebo (Fig. 3). The probability ranking of the effects of different treatment measures on the remission level of suicidal ideation at multiple time nodes is shown in Fig. 4. Cumulative probability of different treatment measures affecting the remission level of suicidal ideation under multiple time nodes is shown in Supplementary Fig. 2. The pairwise comparison contributions of NMA and the risk of bias plots are shown in Supplementary Figs. 4 and 5.

Fig. 3: Network meta-analysis comparing the remission level of suicidal ideation with all treatment measures at different time points.
figure 3

Important results are shown in bold. D1: Day 1, D3: day 3, D13: day 13, D26: day 26.

Full size image
Fig. 4
figure 4

Probability ranking of the effects of different treatment measures on the remission level of suicidal ideation under multiple time nodes.

Full size image

By estimating SUCRA values, we ranked the remission degree of suicidal ideation for all treatment measures at different time points. The higher the SUCRA value, the greater the possibility of suicidality ideation remission, indicating that the treatment plan at this time node has a higher degree of impact on suicidality ideation remission.

For the remission of suicidal ideation in depressed patients, the 11 groups were ranked from highest to lowest as follows: KetamineD13 (99.1%), KetamineD26 (89.6%), EsketamineD26 (70.6%), KetamineD3 (70.0%), KetamineD1 (54.8%), PlaceboD26 (50.5%), MidazolamD3 (46.2%), PlaceboD3 (32.4%), MidazolamD1 (22.1%), EsketamineD1 (14.2%), PlaceboD1 (0.4%) (Supplementary Table 3).

Pairwise meta

In the repeated administration group (including ketamine and esketamine), there was a significant difference in the remission of suicidal ideation between the first day of treatment and the last treatment (RR = 0.56, 95%CI = 0.51–0.62). The overall results were heterogeneous (I2 = 81.2%, P = 0.000) and this heterogeneity persisted in the ketamine versus esketamine subgroups (I2 = 79.1%, P = 0.002). We then performed subgroup analyses according to race, and there was no longer heterogeneity(Fig. 5A, B). Egger’s publication bias test and the funnel plot are shown in Supplementary Figs. 5 and 6.

Fig. 5: Comparison of suicidal ideation between the first and last day of ketamine treatment with repeated dosing.
figure 5

A Meta-analysis of the remission effect of suicidal ideation between the first and the last treatment in the repeated administration group. B Meta-analysis of the effect of suicidal ideation remission after the first and last treatment in different race.

Full size image

Discussion

Although previous reviews and meta-analyses have provided valuable insights into the effects of specific medications on suicidal behavior in specific populations, none have been able to address the strength of early drug action in depressed suicidal patients [19,20,21,22,23]. Ketamine is a promising treatment option for patients with depression and rapid reduction of suicidal ideation in patients. The acute efficacy of ketamine for depression has been demonstrated in repeated RCTS [24, 25]. However, there has been very limited generalization of relevant RCT data in real practice. This has brought great resistance for many researchers to carry out relevant evidence-based medicine verification in this respect [12]. Therefore, we further processed the data to include data from multiple follow-up points in a single study to construct the model and NMA data at multiple time levels, which expanded our network model and made our results more robust.

Our study confirms the role of ketamine in the alleviation of suicidal ideation in depressed patients. However, due to the relatively small number of included studies, we included studies that included both outpatients and inpatients with depression, MDD and TRD, single and repeated ketamine doses for analysis. It is mainly aimed at depression patients with suicidal ideation. This needs further research and analysis in the future to distinguish the specific efficacy for different patient populations.

To our knowledge, this is the first NMA to simultaneously compare the effects of different treatment strategies at different time points. In addition, our pairwise meta-analysis of the first versus the last treatment difference results highlighted the additional benefit of repeated dosing in terms of resolution of suicidal ideation. This provides strong evidence for the choice of medication for acute suicidal ideation in patients with depression.

By using Stata software, we calculated and ranked the probability of alleviating suicidal ideation in different time nodes, as shown in Fig. 4. Through the observation and comparison of the relief efficacy at different time points in the placebo group, we understand the approximate recovery progress of the natural remission of suicidal ideation in depressed patients over time. To provide a rough assessment of the outcome by looking at the overall outcome of the probability ranking, ketamine treatment for depressive patients with severe suicidal ideation showed a higher degree of recovery on day 1 of suicidal ideation than on day 3 of placebo and a similar level as on day 26 of placebo. When the placebo group was used to assess remission after ketamine treatment as a natural remission level, the improvement in suicidal ideation after day 1 of ketamine treatment was similar to the level of natural recovery at about 26 days in non-users. However, this is not a rigorous calculation, and although SUCRA results showed such a ranking strength, there was no significant difference in NMA results between ketamine on day 1 and placebo on day 3 and 26. Therefore, we can only indirectly guess the strength of ketamine in this way, and the true strength of its effect needs to be confirmed by further research.

There are also interesting findings from our study. The reduction in suicidal ideation in the placebo group was significantly different on day 3 than on day 1. This represents a presumably natural process of recovery in depressed patients with suicidal ideation. This suggests that intervention in the acute phase of patients with suicidal ideation is necessary. The way of intervention is not limited to drugs, as long as patients can be reasonably and effectively controlled in the early stage, a considerable number of patients will achieve effective remission over time.

The difference in the intensity of ketamine and esketamine in alleviating suicidal ideation may be caused to some extent by different dosage forms. In the included studies, ketamine was mainly administered intravenously, with the dosage ranging from 0.1 to 1.0 mg/kg, and in most studies, 0.5 mg/kg. One randomized controlled trial showed that the efficacy of higher doses in antidepressants was similar to that of standard doses and was well tolerated [26]. Esketamine was mainly administered intranasally at a fixed dose of 84 mg. Studies have shown that the maximum plasma concentration can be reached within 1 min after intravenous injection [27]. When the antidepressant dose was given for more than 40 min, the concentration gradually decreased after the infusion ended [28]. Intramuscular administration peaks within 5–30 min, sublingual administration peaks within 45 min, and oral administration peaks within 20–120 min. Ketamine bioavailability varied widely among non-intravenous formulations, ranging from approximately 90% intramuscular, 50% intranasal, 29% sublingual, and 17% oral [29,30,31].

In terms of antidepressants, a number of meta-analyses in recent years have consistently concluded that a single intravenous administration of ketamine has a rapid antidepressant effect, reaching a peak within 24 h and lasting 3 to 7 days [32,33,34]. Repeated infusions achieve higher remission rates and extend the time to relapse of depression [35,36,37]. The results of our pairwise meta-analysis, at the same time, also demonstrate the additional benefit of repeated dosing in the remission of suicidal ideation. This underscores the importance of repeated dosing in the clinical setting for depressed patients with suicidal ideation. However, due to the limited number of included studies, we could not compare the differences in drug efficacy between different dosage forms. Considering drug utilization and pharmacokinetics, it is possible that this difference in drug administration has a similar effect on suicidal ideation and depression, and further research is needed in the future.

The NMA results showed that the optimal therapeutic effect was presented on the 13th day. This was only a rough calculation and prediction of the recovery of depressed patients after ketamine treatment based on the current limited research data, without assessing the remission of suicidal ideation at other treatment points in the middle, so we could only speculate that the optimal therapeutic effect was near the 13th day.

As for the recovery degree of the 13th day is significantly better than that of the 26th day, we believe there are mainly two reasons as follows. First, pharmacokinetic factors, after a single infusion, plasma levels of ketamine were not detected for 24 h, but very low levels of ketamine metabolites were detected for 3 consecutive days and 3 consecutive days, and repeated infusion reduced ketamine clearance [38, 39]. In addition to single administration studies, ketamine studies included in our study also included a considerable number of repeated infusion studies. Most of these studies had a treatment endpoint of about 13 days, which may be the main reason why the treatment results of 13 days were significantly better than those of 26 days. Second, it is possible that some patients relapsed after discontinuation of the drug. The study showed that repeated infusion may also prolong the time for depression relapse. The median time for relapse after the last ketamine infusion was about 18 days, and the level of suicidal ideation in patients was likely to change with the recurrence of depression.

There are also some limitations in our study. First of all, since there is no gold standard for the evaluation of suicidal ideation, most researchers adopt different methods for the diagnosis and evaluation of suicidal ideation, which also limits the integration of these research data and may also bring some bias to our research. Secondly, The sample size of the individual studies we included was low, which could also easily bias the results. As the study has been further processed by multi-arm research transformation and contains data of multiple time nodes, the sample size of some studies has been used several times, and the number of participants has been enlarged, which may have a certain impact on our study. However, we do not believe that this is a completely negative effect. The amplification of sample size may also make our network model more robust. In addition, besides the two-arm study, our NMA also includes three-arm and four-arm studies, which has a certain impact on the inspection and verification of the inconsistency of NMA and makes it very difficult to define the inconsistency. We did not do a good job of verifying that our NMA was inconsistent. The difficulty in defining inconsistency comes when we have 2- and 3-arm trial evidence, for example, AB, AC, BC, and ABC trials. Because inconsistency degrees of freedom (ICDF) corresponds to the number of independent loops, if a loop is formed from a multiarm trial alone, it is not counted as an independent loop and must therefore be discounted from the total ICDF. Thus, where there are mixtures of 2-arm and multiarm trials, our definition of inconsistency as arising in loops creates inherent technical difficulties that cannot, as far as is known, be avoided [40]. Observing through the funnel plot of the NMA, we consider its associated bias to be acceptable. We evaluated the randomized controlled trials included in (RoB 2). The results of our rigorous assessment of the risk of bias are shown in Supplementary Tables 1 and 2, and the overall bias of RCT studies is shown in Supplementary Fig. 7A, B. Two of the 10 studies on esketamine were from Janssen Pharmaceutical Companies, which may have created industry bias. All risk of bias was assessed by two study authors independently. If there were discrepancies between the two study authors, a third study author intervened to reassess the risk of bias. The opinions of all evaluated study authors were combined and the final decision was made by vote. We declare that none of the authors on our team have any conflicts of interest with ketamine products. For single-arm studies, we applied the MINORS. In this scale assessment, most of the information in the included articles was clearly reported and adequate. Little information reported but inadequate. Finally, we have to state that all the esketamine studies we included were from the same company, there was a lack of comparative studies from other institutions, and the number of studies was small. We remain suspicious of the results related to esketamine for the time being, which needs to be verified in more studies in the future.

The method of our NMA is innovative. It was an exploratory attempt. Although we cannot prove whether this method is reasonable enough at present, and many problems need to be further proved and explored in the future, it provides a new way of thinking for future NMA methods. Considering many aspects, we cannot completely rule out the existence of bias.

The purpose of our study was to explore the effect of ketamine on suicidal ideation in depressed patients. Our NMA provided a good explanation for the magnitude of reduction in suicidal ideation at different time points after ketamine use. This is of great reference value in how to use drugs more reasonably in clinical practice.

Ketamine’s advantage in rapidly alleviating suicidal ideation in patients with depression is consistent with previous studies [41, 42]. Our study provides strong research evidence to further support this research view. The rapid effect of ketamine in alleviating suicidal ideation may have important clinical implications, which can be the difference between life and death in suicidally depressed patients. The importance of long-term repeated ketamine administration is also illustrated in our study. This is of great significance for future drug development and clinical use.