The Effect of Benzodiazepines on Catatonia: A Systematic Review and Meta‐Analysis

ABSTRACT

Introduction

Catatonia is a complex psychomotor syndrome associated with various psychiatric and somatic underlying disorders. Current treatment protocols include benzodiazepines and electroconvulsive therapy (ECT). Despite their widespread use, systematic evidence regarding efficacy and optimal benzodiazepine dosage remains limited. This systematic review and meta‐analysis aims to evaluate the effectiveness of benzodiazepines in treating catatonia.

Methods

A comprehensive search of PubMed, Embase, APA PsychInfo, and Cochrane central identified studies reporting benzodiazepine treatment outcomes in catatonia. Screening and data extraction were done by two independent researchers. A targeted meta‐analysis was performed on cases and studies that provided clear documentation of the Bush Francis Catatonia Rating Scale (BFCRS) scores before and after treatment, along with well‐defined dosage regimens. Finally, the relationship between dosage and BFCRS improvement was explored.

Results

Of 1336 screened articles, 53 met inclusion criteria. Most were case series using lorazepam (a benzodiazepine) to treat catatonia, with dosages ranging from 2 to 60 mg/day with a median dose of 8 mg/day. The overall remission rate was 55%; overall response was 77%. Treatment duration varied, but predominantly did not exceed 1 week. Our targeted analysis showed an overall effect of −11.88 points on the BFCRS when a benzodiazepine was used in the treatment of catatonia with a Cohen's d = −3.15. In a meta regression, no significant dose–response relationship was found.

Conclusion

Benzodiazepines are highly effective for treating catatonia, but optimal dosing remains unclear. The absence of a clear dose‐effect relationship highlights the need for individualized treatment strategies and well‐designed controlled studies to establish stronger evidence‐based treatment guidelines.

1. Introduction

Catatonia is a complex and sometimes life‐threatening syndrome associated with both psychiatric and somatic disorders. The prevalence of catatonia is estimated to be 9% among patients with a variety of mental and physical health conditions [1]. It consists of a broad cluster of different symptoms related to motor function, speech, behavior, affect, and autonomic activity. Symptom severity is often measured with the use of the Bush Francis Catatonia Rating Scale (BFCRS). If not adequately treated, catatonia can lead to severe and potentially fatal complications such as dehydration, malnutrition, kidney damage due to rhabdomyolysis, and pulmonary and cardiovascular problems [2].

Although the full pathophysiology of catatonia is not yet fully understood, an imbalance of various neurotransmitters appears to play a role, including those of the GABAergic system [3]. Research has indicated that in individuals with catatonia, a disruption can be found in the inhibitory function of GABA within several emotion‐related areas of the prefrontal cortex [3, 4]. This disruption is considered to be associated with a dysfunction in the balance between the GABA‐A and GABA‐B receptors [5]. In current guidelines, benzodiazepines are considered one of the first‐line treatments alongside electroconvulsive therapy (ECT) for catatonia [6]. Benzodiazepines are a class of drugs that bind to a specific site of the GABA‐A receptor. Once bound, they enhance the binding of GABA to the receptor, thereby increasing its inhibitory effects and potentially reducing symptoms of catatonia [7]. However, there is currently no clear treatment protocol for prescribing benzodiazepines in cases of catatonia.

To date, there is no clear consensus on the efficacy and the optimal dosage of benzodiazepines for the treatment of catatonia. This highlights the need for an updated systematic review of the existing literature to better understand the role of benzodiazepines in treating catatonia and to refine clinical practice guidelines. Therefore, we conducted a systematic literature search to assess the effectiveness of benzodiazepines in the treatment of catatonia, with the objective of providing a comprehensive overview of all available evidence.

1.1. Aim

This systematic review and meta‐analysis aims to evaluate the effectiveness of benzodiazepines in treating catatonia. Additionally, we aim to identify an optimal dosing regimen.

2. Methods

This systematic review was performed complying with the Preferred Reporting Items for Systematic reviews and Meta‐Analyses (PRISMA) guidelines [8]. The systematic review was registered with PROSPERO under the registration ID: CRD42023405682.

2.1. Search Strategy

A systematic literature search was performed in multiple databases (PubMed, Embase, APA PsycInfo, Cochrane central). The search contained “catatonia” (and related terms) and “benzodiazepines” (and related terms). The detailed search strategy is provided in the Supporting Information. The initial search was conducted in March 2023 and was updated in February 2024. No filters or limits were used and no restrictions were applied regarding the date of publication or publication type, ensuring a thorough screening of all relevant articles.

Titles and abstracts of the identified articles were independently screened by two researchers (L.B. and B.S.). Subsequently, the full text of potentially eligible articles was assessed by the same researchers. In both phases, the conflicts were discussed among researchers and a third researcher (E.v.E.) resolved remaining conflicts. If full texts were not available, efforts were made to contact the authors or the journals of the original publication. The screening process was facilitated using the Rayyan platform [9].

In this systematic review we included articles that were case series reporting on three or more cases, cohort studies, and trials. Excluded were studies published in a language other than Dutch or English, conference abstracts, articles describing animal research, and case reports describing fewer than three cases. Although zolpidem and zopiclone do not formally belong to the benzodiazepine group, studies involving these drugs were included in the review because their mechanism of action is relatively similar. Both drugs bind to the GABA‐A receptor, albeit at different subunits than traditional benzodiazepines, producing comparable central nervous system–depressant effects [10].

Data extraction was performed independently by two researchers (L.B. and B.S.) and compiled into a data sheet specifically created for this review. Extracted variables included: article type, study design, study aim, included participants, underlying disorders, co‐medication, first dose and total highest daily dose of benzodiazepine, assessment method (such as the BFCRS), response rate, remission rate, successive treatment and treatment setting. This review focuses on the effect of benzodiazepines; the effect of successive treatments with either medication or ECT is outside the scope of this review. If the successive treatment effect of ECT was described, studies are only included if the response to benzodiazepine treatment is described before successive treatment with ECT. In this way, non‐responders are also included in the analyses. The protocol initially specified the use of GRADE criteria for quality assessment; however, this was not feasible due to the overall low quality of the available evidence. Consequently, the methodological quality of evidence was independently assessed by both researchers using the Oxford Centre for Evidence‐Based Medicine (OCEBM) 2011 levels of evidence [11].

Response and remission rates were extracted from the data of the article if provided. If not provided, we have defined remission as a total resolution of symptoms or no longer meeting criteria of catatonia (i.e., BFCRS < 3). Response was defined as any partial response or reduction of symptoms of catatonia after administration of benzodiazepines.

2.2. Data Analysis

A full meta‐analysis was initially considered as the primary analysis method. However, due to the predominance of case series and low‐quality cohort studies, in addition to the significant variability in both methodologies and treatment protocols, conducting such an analysis was deemed unfeasible. To explore the potential correlation between treatment dosage and therapeutic outcome, we performed a targeted meta‐analysis. For this targeted analyses studies and cases were included that provided clear documentation of the BFCRS scores before and after treatment, along with well‐defined dosage regimens. For this reason, in some instances we have extracted individual cases from a larger case series if the necessary information was not provided for all cases or if the dosing regimen for individual cases varied.

2.3. Main Analysis and Sensitivity Analyses

The primary outcome for our targeted meta‐analysis was the change in BFCRS scores following intervention. Study weights were based on the variance of mean change scores. Since variances of mean change scores were not published, they were derived from equally supposed variances of pre‐ and post‐intervention scores, assuming a within‐subject correlation of r = 0.5. The choice of r = 0.5 is based on conventional assumptions, but additional sensitivity analyses were performed using alternative values (r = 0.3 and r = 0.7) to test the robustness of the results. If the variance of pre‐intervention scores were not published, it was assumed to be equal to an average of the variances of pre‐intervention scores found in other studies. This included the situation of case series, taking the result of each single case as the result of a separate study with a sample size of one. Additionally, sensitivity analysis was done, assuming different values for the variance. Meta‐analysis was conducted using a random‐effects model to account for heterogeneity. Effect sizes were calculated in terms of Cohen's d, defined as the mean difference in BFCRS scores before and after intervention, divided by the pre‐intervention standard deviation.

2.4. Meta‐Regression and Subgroup Analyses

Due to the lack of availability of randomized controlled trials, utilization of a typical meta‐regression, an evaluation based on the comparison of the effect in intervention groups with those in control groups (i.e., between group effects) was not possible. Consequently, in this study, we focused on within intervention group effects. Additionally, a meta‐regression was conducted to examine the relationship between dosage and BFCRS effect size. Studies with varying dosages within groups were excluded in a separate sensitivity analysis to prevent ecological bias. Subgroup analyses were also conducted to explore potential moderators of treatment effect.

2.5. Software and Statistical Tool

All analyses were conducted using Stata18 software for meta‐analysis, using random‐effects models, creating forest plots, and applying meta‐regression techniques [12].

3. Results

3.1. Search Results

The primary literature search was conducted in March 2023 and updated in February 2024. These searches combined provided 2416 results, of which 1337 were screened after removing duplicates. Of these 1337 articles, 136 were sought for retrieval to assess full text; of these articles, 53 studies met the eligibility criteria. The PRISMA flowchart for article selection is shown in Figure 1.

FIGURE 1.

PRISMA flow chart [8].

3.2. Characteristics of Included Studies

The 53 included articles were published between 1983 and 2023 and contained 19 case series reporting on a range between three and 10 cases. Additionally, we included 9 retrospective chart reviews or cohort studies and 24 prospective open label trials or prospective cohort studies. Only one RCT was included in this review. An overview of the characteristics of the included studies can be found in Table 1. The median included cases, even though the included articles ranged between three and 232, was 3.5. This indicates that the majority of the included studies presented a case series with a small number of included cases. As shown in Table 1, there are few studies with OCEBM level of evidence 2b, and most of them are classified as OCEBM level 4 [11].

TABLE 1.

Characteristics of included studies.

References	Origin	N	Study design	Underlying/comorbid disorder	First administrated dosage lorazepam or equivalent	Highest daily dosage lorazepam or equivalent	Outcome measure	Response (%)	Remission (%)	Level of evidence [11]
Aloysi et al. (2011) [13]	USA	4	Caseseries	Affective disorders	1 mg (PO/IV)	6 mg (PO)	Symptoms	100	0	4
Appiani et al. (2023) [14]	Argentina	22	Prospective cohort	Delirium	1–2 mg (IM)	2 mg (IM)	CAM‐ICU + BFCRS	27	n.r.	4
Benazzi (1991) [15]	Italy	4	Case series	Psychotic disorders	2 mg (IV/IM) a	14 mg (IV/IM)	Symptoms	100	n.r.	4
Bhattacharjee and Bandyopadhyay (2023) [16]	India	3	Case series	TB, neuro‐cysticercosis, pneumonia	2 mg (IM)	14 mg (IM)	BFCRS	0	0	4
Brelinski et al. (2009) [17]	France	3	Case series	SLE	2.5 mg (PO)	15 mg (PO)	Symptoms	100	n.r.	4
Bruijn and Blom (2010) [18]	Netherlands	3	Case series	Psychotic disorder	4 mg (IM)	60 mg (IM/PO)	Symptoms	100	100	4
Bush et al. (1996) [19]	USA	21	Open, prospective cohort	Psychotic + affective disorders, personality disorders, systemic illnesses	4 mg (PO/IM/IV)	8 mg (PO/IM/IV)	BFCRS	76	52	2b
Cottencin et al. (2007) [20]	France	12	Prospective case study	Diverse systemic illnesses, psychotic + affective disorders	2.5 mg/h (PO/IV)	20 mg (PO/IV)	Symptoms	100	83	4
Dutt et al. (2011) [21]	India	51	Retrospective chart review	Psychotic + affective disorders, organic illness	4 mg (PO/IM/IV)	12 mg (PO/IM/IV)	BFCRS	18	n.r.	4
England et al. (2011) [22]	USA	25	Retrospective chart review	Psychotic + affective disorders	n.r.	16 mg (n.r.)	Symptoms	96	28	4
Fricchione et al. (1983) [23]	USA	4	Case series	Affective disorder, encephalopathy, psychosis	1.5–2 mg (IV)	24 mg (PO/IV)	Symptoms	100	100	4
Hatta et al. (2007) [24]	Japan	41	Open label trial	Psychotic + affective disorders	3 mg (PO/IV) (daily dosage)	13.6 mg (PO/IV)	DSM IV	49	2	2b
Huang (2005) [25]	Taiwan	14	Prospective cohort	Psychotic disorders	2 mg (IM)	7.2 mg (IV)	DSM IV	100	n.r.	2b
Huang et al. (2013) [26]	Taiwan	12	Prospective open label	Affective disorders	2 mg (IM)	7.2 mg (IV)	DSM IV	100	100	2b
Hung & Huang (2006) [27]	Taiwan	7	Prospective open label	Affective disorders	2 mg (IM)	7.2 mg (IV)	DSM IV	100	100	2b
Jaimes‐Albornos & Serra‐Mestres (2013) [28]	Spain	7	Prospective cohort	Diverse somatic conditions	1 mg (PO)	10 mg (PO)	Symptoms	71	43	4
Johnson et al. (2022) [29]	India	3	Case series	Affective disorders and thyroid disease	6 mg (n.r.) (daily dosage)	14 mg (n.r.)	BFCRS	0	0	4
Kritzinger & Jordaan (2001) [30]	South Africa	9	Open prospective study	Psychotic + affective disorders, HIV, dementia	2 mg (IM)	2 mg (IM)	Symptoms	67	0	4
Layek et al. (2022) [31]	India	66	Cross sectional study	Psychotic + affective disorders, other	4 mg (PO/IM/IV) (daily dosage)	16 mg (PO/IM/IV)	DSM V + BFCRS	82	n.r.	2b
Lee et al. (2000) [32]	Australia	24	Prospective cohort	Psychotic + affective disorder, conversion, delirium	4 mg (PO/IM/IV) (daily dosage)	16 mg (PO/IM/IV)	BFCSI	79	67	2b
Lin & Huang (2013) [33]	Taiwan	21	Prospective cohort	Psychotic disorders	2 mg (IM)	8.25 mg (IV)	DSM‐IV and BFCRS	100	n.r.	2b
Lin et al. (2016) [34]	Taiwan	30	Prospective cohort	Psychotic + affective disorder, systemic illnesses	2 mg (IM)	8.25 mg (IV)	DSM‐IV	84	79	2b
Lin et al. (2017) [35]	Taiwan	21	Prospective cohort	General medical conditions or substance abuse	2 mg (IM)	8.25 mg (IV)	BFCRS	90	86	2b
Martényi et al. (1989) [36]	Hungary	4	Case report	Psychotic disorder	1 mg (IV)	6 mg (PO)	Symptoms	100	n.r.	4
Miles et al. (2019) [37]	USA	7	Prospective cohort	Down syndrome	2 mg (n.r.)	20 mg (PO)	BFCRS	100	14	4
Mishra (2023) [38]	India	3	Case series	Post Caesarian‐section	2 mg (IV)	2 mg (n.r.)	Symptoms	100	100	4
Nahar et al. (2017) [39]	India	36	Retrospective chart review	Postpartum psychosis and affective disorders	3 mg (IM/IV)	8 mg (n.r.)	BFCRS	75	50	4
Narayanaswamy et al. (2012) [40]	India	99	Retrospective cohort	Psychotic + affective disorders	2 mg (IV)	8 mg (n.r.)	Symptoms	32	32	4
Northoff et al. (1995) [41]	Germany	18	Prospective open label	Psychotic + affective disorders	1 mg (PO/IV)	10 mg (PO/IV)	Symptoms	67	n.r.	4
Northoff et al. (1998) [42]	Germany	22	Prospective open label	Psychotic + affective disorders	2 mg (PO/IV)	4 mg (PO/IV)	Rosebush + Lohr and Wiesniwski	59	59	2b
Payee et al. (1999) [43]	India	30	Prospective open label	Psychotic + affective disorders	3 mg (PO)	8 mg (PO)	BFCRS/symptoms	70	70	2b
Ramdurg et al. (2013) [44]	India	61	Retrospective chart analysis	Psychotic + affective disorders	4 mg (PO/IM/IV)	12 mg (PO/IM/IV)	Symptoms	67	n.r.	4
Rosebush et al. (1990) [45]	Canada	15	Prospective open label	Psychotic + affective disorders, organic brain disease	1 mg (PO/IM/IV)	2 mg (PO/IM/IV)	Symptoms	87	80	4
Rosebush & Mazurek (1996) [46]	Canada	5	Case series	Benzodiazepine withdrawal	1 mg (PO/IM)	3 mg (PO)	Symptoms	100	n.r.	4
Sakhardande et al. (2022) [47]	India	5	Case series	COVID‐19	2 mg (IV/PO)	8 mg (IV/PO)	BFCRS	100	40	4
Salam et al. (1987) [48]	USA	3	Case series	Psychotic + affective disorder	2 mg (IM)	5 mg (PO)	Symptoms	100	100	4
Salam and Kilzieh (1988) [49]	USA	5	Case series	Psychotic + affective disorder + organic underlying problem	2 mg (IM)	6 mg (IM/PO)	Symptoms	60	20	4
Schmider et al. (1999) [50]	Germany	17	Double‐blind crossover study	Psychotic + affective disorders	2 mg (SL)	6 mg (SL)	BPRS	59	n.r.	2b
Seethalakshmi et al. (2008) [51]	India	16	Prospective cohort	Psychotic + affective disorders, organic problem, dissociative disorder	1 mg (IV)	4 mg (IV)	BFCRS	100	75	2b
Seetharaman et al. (2021) [52]	India	4	Case series	Psychotic disorder or unknown	4 mg (IV)	12 mg (IV/PO)	BFCRS	100	33	4
Sharma et al. (2017) [53]	Spain	5	Case series	Dementia and/or depression	2 mg (PO)	2 mg (PO)	BFCRS	100	100	4
Smith et al. (2015) [54]	USA	4	Case series	Psychotic disorder or unknown	1 mg (PO/IV)	6.7 mg (PO)	BFCRS	100	100	4
Suchandra et al. (2021) [55]	India	57	Cross‐sectional observational study	Psychotic disorder and affective disorder	2 mg (IV)	4 mg (IV)	BFCRS	68	0	2b
Thamizh et al. (2016) [56]	India	4	Case series	Psychotic and/or affective disorder	4 mg (n.r./PO)	24 mg (n.r./PO)	BFCRS	50	50	4
Tibrewal et al. (2010) [57]	India	99	Retrospective chart analysis	Psychotic and/or affective disorder, or unknown	1 mg (IV)	6 mg (IV)	Symptoms	69	32	4
Tuerlings et al. (2010) [58]	Netherlands	26	Retrospective case study	Psychotic and/or affective disorder, drug abuse, dementia	2 mg (n.r.)	12 mg (n.r.)	Symptoms	21	13	4
Unal et al. (2017) [59]	Turkey	55	Prospective study	Psychotic and/or affective disorder, drug abuse, unknown	1 mg (PO)	15 mg (PO)	BFCRS	24	24	2b
Ungvari et al. (1994) [60]	China	18	Prospective study	Psychotic and/or affective disorder, personality disorder	2 mg (PO)	6 mg (PO)	Symptoms	100	22	4
Ungvari et al. (1999) [61]	China	18	RCT	Psychotic disorder	2 mg (PO)	6 mg (PO)	BFCRS	0	0	2b
Wetzel & Benkert (1988) [62]	USA	10	Case series	Psychotic and/or affective disorder	2.5 mg (PO/IV)	5 mg (PO/IV)	Symptoms	100	n.r.	4
White et al. (2015) [63]	USA	3	Case series	Intellectual disabilities	1 mg (PO)	18 mg (PO)	Kanner	67	33	4
Wilson et al. (2015) [64]	USA	232	Retrospective chart review	Diverse psychiatric and medical conditions, not further specified	n.r.	5.8 mg (n.r.)	BFCSI/BFCRS	96	85	4
Yassa et al. (1990) [65]	Canada	10	Open clinical trial	Psychotic and/or affective disorder	1 mg (PO/IM)	6 mg (PO/IM)	Symptoms	100	100	4

Note: Affective disorders can consist of major depression disorder and mania or depression in bipolar disorder; n reported is the number of participants in the study receiving benzodiazepines as treatment, this can differ from the total N of the reported study. It can also mean number of analyzed episodes of catatonia.

Abbreviations: BFCRS = Bush Francis Catatonia Rating Scale; BPRS = Brief Psychiatric Rating Scale; CAM‐ICU = confusion assessment method for the intensive care unit; DSM = Diagnostic and Statistical Manual of Mental Disorders; IM = intramuscular; IV = intravenous; n.r. = not reported; PO = per os (oral); SL = sublingual.

^a In studies, where “parenteral” was mentioned but not further defined, we mentioned IV/IM.

In 16 studies (30%), a benzodiazepine other than lorazepam was used to treat catatonia. The studies in which a different drug than lorazepam was used, or where another drug was used alongside or after lorazepam, the following alternative drugs are described: in nine (17%) studies diazepam was used, in four (8%) studies clonazepam was used, in two (4%) studies zolpidem was used, and in one (2%) study oxazepam was used. This could be alongside lorazepam or as an alternative when parenteral lorazepam was not available in the country of origin of the study. For the purpose of illustrating and comparing multiple studies in this review we have calculated the lorazepam equivalent dosage for the studies using a different type of benzodiazepine. We used an online dose conversion calculator [66]. The duration of drug administration varied widely between the administration of just one dose and multiple weeks of treatment. Most studies did not exceed 1 week of treatment.

3.3. Overall Effect

The highest daily dose varied between 2 and 60 mg with a median of 8 mg per day. We found an overall average response of 77% and average remission of 53%, calculated across all included studies, not taking sample size per study into account. Most studies reported either multiple routes of administration or a change in route of administration during the course of treatment. Among studies that maintained the same route of administration for all patients throughout the treatment period, the overall response rate to intramuscular injection was 31.3%, with 0% achieving remission (three studies, n = 34). For intravenous administration, the overall response rate was 79%, with 35.6% achieving remission (three studies, n = 172). Oral administration showed a 66.5% response rate, with 36.5% achieving remission (eight studies, n = 139).

3.4. Duration of Treatment

Treatment duration varied widely from 24 h to a maximum of 6 weeks. Most studies did not report the treatment duration at the maximum daily dosage. For the studies that did report on the treatment duration, this varied between hours and several weeks. However, for the majority of the studies that reported treatment duration, maximum study duration was no more than 1 week. Non responders often received ECT treatment that was described as successful in most studies but is not reported here as this is not the scope of this review. There were also patients that continued on maintenance lorazepam but were not further followed to observe response or remission.

3.5. Somatic and Psychiatric Underlying Conditions

When comparing studies that have included participants with only somatic underlying disorders as a diagnosis or classification with studies that have included participants with only psychiatric underlying disorders, we find a similar highest daily dose (median of 7.6 and 7.2 mg/day, respectively). However, the response rate in these groups varies, with an average 75% response in the studies reporting on patients with a somatic underlying cause of catatonia compared to an 81% response in the studies reporting on patients with only psychiatric underlying causes. Remission rates also differ between both groups and amount to an average of respectively 46% and 57%.

3.6. Meta‐Analysis

Data from a total of 20 (sub)studies and/or individual cases were extracted for use in the meta‐analysis, with sample sizes ranging from 1 (for the single cases) to 37 participants per study (see Table 2). The studies reported pre‐ and post‐intervention BFCRS scores, dosage information, and in some cases pre‐intervention standard deviations for BFCRS scores and dosages. For studies reporting a case series where individual data on the dosage of lorazepam and pre‐ and postintervention BFCRS scores were reported, we extracted individual cases for analysis. Studies that only provided BFCRS scores pre‐ and postintervention on a group level as well as the lorazepam dosage were included as a whole.

TABLE 2.

Data extracted and used in meta‐analysis.

Article	N	Mean BFCRS Pre	SD BFCRS Pre	Mean BFCRS Post	Mean Dosage	SD Dosage	Cohen's d	SE d
Bush et al. (1996) Group 1 [19]	16	17.7	6.8	2.07	3	1.6	−2.30	0.25
Bush et al. (1996) Group 2 [19]	5	17.4	3.5	13.08	5.5	2	−1.23	0.45
Johnson et al. (2022) C3 [29]	1	13	—	13	8	0	.	.
Lee et al. (2000) [32]	24	15.52	3.72 a	2.48	6	0	−3.50	0.20
Lin and Huang (2013) Group 1 [33]	13	8.9	2.8	0	2	0	−3.18	0.28
Lin and Huang (2013) Group 2 [33]	3	11.7	2.5	0	4	0	−4.68	0.58
Lin and Huang (2013) Group 3 [33]	2	11.5	3.5	0	8.25	0	−3.29	0.71
Lin and Huang (2013) Group 4 [33]	3	11.7	3.1	0	8.25	0	−3.77	0.58
Miles et al. (2019) C5 [37]	1	26	—	0	14	0	—	—
Sakhardande et al. (2022) C1 [47]	1	22	—	17	6	0	—	—
Sakhardande et al. (2022) C2 [47]	1	10	—	0	6	0	—	—
Sakhardande et al. (2022) C3 [47]	1	8	—	0	4	0	—	—
Seetharaman et al. (2021) C1 [52]	1	18	—	0	8	0	—	—
Seetharaman et al. (2021) C2 [52]	1	27	—	4	10	0	—	—
Seetharaman et al. (2021) C3 [52]	1	32	—	6	10	0	—	—
Sharma et al. (2017) [53]	5	16.3	4.5	0	2.2	3.1	−3.62	0.45
Suchandra et al. (2021) Group 1 [55]	37	12	—	6	2	0	—	—
Suchandra et al. (2021) Group 2 [55]	20	15	—	7.5	4	0	—	—
Thamizh et al. (2016) C2 [56]	1	12	—	0	10	0	—	—
Thamizh et al. (2016) C4 [56]	1	9	—	0	8	0	—	—

Abbreviations: — = not available; BFCRS = Bush Francis Catatonia Rating Scale; C1–C4 = single case 1–4; N = number of patients; SD = standard deviation; SE = standard error.

^a Lee et al. (2000) reported “The mean score on BFCRS on Day 0 was 15.52 ± 0.76.” We deduced from the report on age in the Supporting Information Appendix, that they reported “mean ± SEM (standard error of the mean),” implying SD = 3.72.

A random‐effects model was used to calculate the overall effect size across studies with three or more cases, with a known SD. The mean difference in BFCRS scores before and after intervention was −11.50 points (95% CI: −14.11 to −8.89), when standardized with the average SD of 4.0, corresponded to a Cohen's d of approximately 3, indicating a very large effect size. It is good to note that a negative difference in BFCRS score stands for an improvement in catatonic symptoms. The main forest plot (Figure 2) illustrates the effect estimates from each study, with an overall pooled effect size.

FIGURE 2.

Forest plot on summary effects on the Cohen's d.

Heterogeneity was assessed using the I2 statistic, which indicated a high degree of variability across studies (I ² = 86.85%, p < 0.001), suggesting substantial between‐study differences. Figure 3 shows the funnel plot from the main analysis. The asymmetry indicates a possible publication bias and also shows the high heterogeneity.

FIGURE 3.

Funnel plot assuming r = 0.5.

When performing a sensitivity analysis in which we have added two larger groups from the study of Suchandra et al. in which we imputed a presumed pre‐intervention SD value of 4. The overall effect size remained very large. The Cohen's d in this case is only slightly smaller with a value of −2.83, still indicating a very large effect size.

When performing sensitivity analyses assuming a within subject correlation of r = 0.3 instead of r = 0.5, the overall effect remains at Cohen's d − 3.14 (95% CI: −3.83 to −2.45). When assuming the within subject correlation of r = 0.7 we find an overall effect of Cohen's d − 3.17 (95% CI: −3.88 to −2.46).

3.7. Meta‐Regression on Dosage–Response Relationship

A meta‐regression was conducted to examine the relationship between dosage and effect in Cohen's d (Figure 4). The coefficient for dosage was −0.018 (p = 0.917), indicating no existing association between effect size and dosage. We have added a sensitivity analysis in which we included the study of Suchandra assuming a SD of 4 (Figure 5). In this regression we find a trend towards an association between higher dosages and higher effect. Coefficient −0.116, p = 0.496. This finding, however, is not statistically significant. Across the two meta‐regression models, the relationship between dosage and BFCRS improvement (as expressed in effect size Cohen's d) was not significant and there is no clear evidence that increasing the dose changes the treatment effect.

FIGURE 4.

Meta regression Cohen's d on dosage assuming r = 0.5. d = −3.07 − 0.018 × dosage, 95% of slope = (−0.349, 0.312), p‐value of slope = 0.913. BFCRS = Bush Francis Catatonia Rating Scale; Cohen's d = effect size on BFCRS; r = within‐subject correlation.

FIGURE 5.

Meta regression Cohen's d on dosage assuming SD = 4 and r = 0.5. Regression: d = −2.33 − 0.116 × dosage, 95% of slope = (−0.432, 0.200), p‐value of slope = 0.496. BFCRS = Bush Francis Catatonia Rating Scale; Cohen's d = effect size on BFCRS; SD = standard deviation; r = within subject correlation.

4. Discussion

Benzodiazepines, especially lorazepam, are widely used in the treatment of catatonia. However, the ideal dosage regimen and route of administration remain unclear. Due to the nature of the condition and the risk of progressing into malignant catatonia, adequate treatment is necessary to prevent major complications. Our systematic review and targeted meta‐analysis confirms the hypothesized high efficacy of benzodiazepines in the treatment of catatonia. To the best of our knowledge, this study is among the first to explore the high effect size of lorazepam in the treatment of catatonia, and to report the use of meta regression analysis to explore effective dosage of lorazepam treatment.

4.1. Lorazepam Is Effective in Treating Catatonia

We hypothesized that benzodiazepines, mainly lorazepam, would be effective in treating catatonia. Our results suggest a confirmation of this hypothesis with an average overall response rate of 77% and an average remission of 53%. In order to be able to give some context to our findings we chose to analyze a subset of the data where the required information for a meta‐analysis was available. We recognize the limitations to the available data and therefore the limitations of our analysis. Since all studies lack control groups, these findings describe a within‐group effect only. The meta‐analysis based on the subset of studies where the required information was available suggests a substantial reduction in BFCRS scores following intervention with benzodiazepines. The very large effect size (d = −3.15) of lorazepam in the treatment of catatonia further suggests that the intervention is highly effective, though this should be interpreted with caution. The large within group effect sizes, as shown in the forest plot, may partly be ascribed to the homogeneity of patient groups within the studies leading to small SD's of the severity of catatonic symptoms before the intervention and consequently to small standardizers to translate each difference into a standardized mean difference (Cohen's d). Additionally, it is important to consider the high between‐study heterogeneity in described effect sizes across analyzed studies and other differences like baseline severity, differences in treatment duration, and somatic and psychiatric comorbidity that are not accounted for as this may have influenced outcomes and limit the generalizability of these findings.

4.2. Optimal Dosing Regimen and Optimal Route of Administration Remain Unclear

Although consensus and evidence about the optimal dosage regimen and route of administration for the treatment of catatonia remain unclear, the studies in this review did correspond in the starting dosage. The most common dosage at the start of treatment was 2 mg. A number of studies started with a gift of 1 mg lorazepam after which they increased the dosage after 30 min to 1 h. These findings correspond with the widely used lorazepam challenge and titration approach.

Rogers et al. give the following recommendation for the treatment of lorazepam for patients with catatonia: “high doses above the licensed maximum may be necessary to achieve maximal effect” [6]. In this guideline an adequate trial is defined as achieving complete remission of symptoms, stopped titration due to side effects or a lorazepam dose of at least 16 mg per day. Our meta‐regression did not show a statistically significant association between dosage and effect size, though a weak trend of higher effectiveness with higher dosages was found when including a larger number of participants. Higher dosages than 9 mg have not been analyzed due to lacking data. Inclusion of studies that used higher dosages would be beneficial to study correlation between dose and effect. However, in the overall dataset we found a median highest daily dose of 8 mg, showing most of the articles reporting on benzodiazepine use for catatonia do not reach dosages of 16 mg per day, as recommended by Rogers et al. [6]. This underscores the complexity of adequate dosing of lorazepam and the need for individualized treatment plans. This means that the overall found response and remission rates might be an underestimation of the actual effect of lorazepam as higher dosages could possibly provide higher remission rates. It is also possible that, in clinical practice, higher doses are administered in cases of treatment resistance or when no alternative therapies are available.

The single randomized controlled trial was conducted in 1999 by Ungvari et al. and described a double blind cross‐over study in which participants were treated for long lasting catatonia with 6 weeks of lorazepam treatment and 6 weeks of placebo treatment [61]. Lorazepam was administered in a dosage of 6 mg per day. There was no significant effect of the benzodiazepine treatment on catatonia symptoms; this lack of effect might be caused by an inadequate dosage of lorazepam. Another hypothesis is that long lasting catatonic symptoms show a lower response to benzodiazepine treatment. Current literature lacks information based on randomized trials on the efficacy of higher doses of lorazepam or other benzodiazepines for reaching remission of catatonic symptoms. Duration of treatment was insufficiently reported among most of the included studies to perform an additional analysis on this aspect of treatment.

The route of administration was often insufficiently described, varied between cases, or was modified during the course of treatment. For example, treatment that was started parenterally was continued orally once possible. Clinically, parenteral administration is generally considered to be more effective than oral treatment. The limited data available for comparison support this assumption for intravenous administration but not for intramuscular use. Notably, the extracted data indicate substantially lower remission rates compared with the aggregated results from all studies. These findings should be interpreted with caution, as the observed differences may be influenced by bias within the included literature.

4.3. Influence of Underlying Disorder on Effectivity of Benzodiazepine Treatment

The large number of included studies allows us to compare the effect of benzodiazepines in catatonia associated with somatic and psychiatric underlying conditions. This distinction is often difficult to make and may not be entirely feasible. Nevertheless, reporting these findings may help to guide future research on this topic.

Our analysis suggests that studies reporting on cases following a somatic underlying condition show a lower response and remission rate than the studies reporting on cases with a diagnosed psychiatric condition underlying the catatonia. The reason for this difference remains unclear, as a shared neuro‐inflammatory process has been hypothesized across all forms of catatonia [67, 68, 69]. A potential hypothesis could be delayed recognition in general hospitals resulting in delayed treatment with worse treatment outcomes.

4.4. Strengths and Limitations

We aimed to present an overview of current literature as complete as possible. This led to the inclusion of data of 53 included studies in this systematic review, consisting of 19 case series, 9 retrospective chart reviews or cohort studies, 24 prospective open label trials or prospective cohort studies and one RCT. The only earlier systematic review on this subject that was performed by Pelzer et al. included 17 studies [70]. In contrast with Pelzer and colleagues, we chose to also include studies that did not use a standardized outcome measure. However, the quality of the included studies often did not exceed OCEBM level of evidence of 4, with 2b being the highest ranking. Furthermore, we only included studies published in the English or Dutch language, the omission of studies published in different languages may cause publication bias to be a factor in the results used in our analysis as it can be argued that positive findings are more often published in English than other languages.

Although the protocol prespecified the use of GRADE, its emphasis on randomized controlled trials rendered it unsuitable given the predominance of uncontrolled case series and observational studies in our dataset. We therefore applied the 2011 OCEBM levels of evidence, which offered a pragmatic and transparent framework for classifying study quality by design. While appropriate for datasets dominated by case series, this approach provides only a broad categorization and does not fully address the methodological limitations of the available evidence.

Because of the high heterogeneity in treatment duration, reported dosing regimens, routes of administration and use of standardized outcome measures, we have only been able to describe an overall response and remission rate which are not weighted and must be interpreted with caution. Additionally, the definition for response and remission were not always well described in the included articles. Especially the definition for response can vary between any response to treatment to a minimum of 50% reduction of symptoms. Of the 53 included studies, very few studies provided BFCRS scores before and after treatment with lorazepam or another benzodiazepine which is a notable gap in current literature on catatonia treatment. Even though most studies lacked a standardized outcome measure we were able to perform, what is to our knowledge, the first robust analysis of data on the efficacy of benzodiazepines as treatment for catatonia and dose related findings. Nonetheless, due to lack of RCT's we could not determine whether the large effect size found could be partially due to a placebo effect.

To provide a comprehensive review, we also included studies reporting on ECT treatment when outcomes following prior benzodiazepine therapy were available. We acknowledge, however, that not all ECT studies report benzodiazepine responses, which may introduce bias. Within our dataset, benzodiazepine treatment was frequently followed by improvement in catatonia in case reports and single‐arm open‐label studies, suggesting a potential treatment effect. However, studies that reported on patients that died due to catatonia are also integrated in this review [28]. The absence of control groups, combined with the likelihood that reports on patients with the most severe symptoms of catatonia were excluded due to early ECT referral or in rare cases even death may overestimate observed benefits and introduce substantial risk of bias. Also, studies that solely reported on ECT outcomes without describing prior benzodiazepine treatment effect were not included in this review. As a result, the apparent benefits should be interpreted with caution.

As mentioned earlier, lacking data, high heterogeneity of included studies, and the high number of assumed factors in the analyses are limitations to the conclusion we can derive from these analyses.

4.5. Future Research

We acknowledge the shortcomings of our analysis based on the available data. In order to be able to perform higher quality analyses in the future there is a need for reporting standardized outcome measures along with well described treatment protocols including dosing regimens and routes of administration in future publications in the field of catatonia research. Likewise, the lack of high‐quality randomized controlled trials focused on treatment is a significant limitation in the field of catatonia research. The acute nature of catatonia poses challenges to conducting research as complications can be severe and immediate treatment is often necessary, which makes planning for inclusion in a trial complex, but in our opinion not impossible.

Besides our recommendation for future randomized controlled research to develop an evidence‐based protocol for the treatment of catatonia with benzodiazepines, we also believe it is essential to further investigate the group of patients that do not respond to lorazepam or do not achieve remission of symptoms. The included studies display that 23%–45% of patients with catatonia do not, or do not sufficiently, respond to the first‐line treatment with benzodiazepines. One hypothesis we propose is that this could be linked to deficiencies in the GABAergic system, following the GABA‐model hypothesis of catatonia. While benzodiazepines primarily target GABA‐A receptors, a deficiency or dysfunction in GABA‐B receptors might contribute to the lack of response, as these receptors also play a critical role in modulating neuronal excitability and synaptic transmission [71]. In addition, an overall deficiency in GABA production or release could result in insufficient activation of GABA receptors, regardless of lorazepam administration, thereby reducing its therapeutic efficacy [72].

Further research is necessary to conclude if the differences in effect of lorazepam on somatic versus psychiatric underlying conditions that we see in this review can be reproduced, as studies included here show high heterogeneity. It is also important to further investigate neurobiological underlying differences. This could have implications for identification of, and treatment protocols for, different types of catatonia with various neurobiological underlying processes.

4.6. Implication and Conclusion

Considering our results, based on case series and cohort data incorporated in the meta‐analysis, we advise to continue the of use lorazepam in the treatment of catatonia with individually curated dosing regimens. The majority of studies provided a maximum dose up to a period of 1 week, after which the majority of non‐responding patients received ECT.

While benzodiazepines remain a cornerstone in the pharmacological treatment of catatonia, future research into both the neurobiological foundations of non‐response and optimizing treatment protocols, including trials on the optimal dosage of lorazepam, is crucial for improving treatment options and outcomes for patients with catatonia.

Author Contributions

L. Bot and B. Schotsman: conceptualization, data curation, formal analysis, investigation, methodology, validation, project administration, resources, visualization, writing – original draft, writing – review and editing. A. W. Hoogendoorn: methodology, formal analysis, investigation, supervision, writing – review and editing. C. H. M. Planting: software, resources, project administration. E. Oostra, B. W. C. Storosum, H. M. Heller, J. J. B. van der Vlugt, J. G. van Mill, L. J. C. A. Smarius, M. S. Oudijn, M. S. van Noorden, and S. M. M. Lange: conceptualization, resources, writing – review and editing. Y. D. van der Werf: conceptualization, writing – review and editing, supervision, project administration. E. van Exel: conceptualization, investigation, validation, methodology, writing – review and editing, supervision. All authors read and approved the final manuscript.

Funding

This research is funded by the Dutch Brain Foundation (Hersenstichting: DR‐2020‐00369). The funder has no specific role in the conceptualization, design, data collection, analysis, decision to publish, or preparation of the manuscript nor study protocol.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Data S1: acps70102‐sup‐0001‐PRISMA_2020_checklist.pdf. PRISMA 2020 checklist.

Data S2: acps70102‐sup‐0002‐Supinfo.pdf. PubMed 10‐03‐2023.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.