A systematic review and meta‐analysis of maintenance treatment for psychotic depression

Figures & data

Table 1. Overview of search strategy and collection of data items.

Electronic search	We conducted an electronic search of four databases (PubMed, Embase, PsycINFO, and Cochrane). The search string was made up of three search blocks separated by the word AND: one search block for the name of the disease, one search block for prevention and treatment, and one search block for relapse and recurrence. The search string used for the PubMed search is presented in Supplemental Appendix 2.
Inclusion criteria	The subjects needed to be in remission or to have responded to acute treatment prior to the start of the maintenance treatment. There needed to be more than five subjects in the intervention and control groups for a study to be included.
Exclusion criteria	Case reports, study protocols, reviews, studies examining only non-affective psychoses, post-partum psychoses and uncharacterized psychoses were excluded. Studies in any other language than English were excluded.
Study selection and collected data items	All abstracts from the initial search in December 2019 were screened independently by two authors (AA & CH). In case of disagreement, the article was included. The abstracts from the updated search in July 2021 were screened by one author (AA). The articles included for full-text analysis were assessed by two authors (AA & AN). From each article we extracted the following information: Patient characteristics Interventions Post-acute treatment status Definition of relapse and/or recurrence Number of events Drop-out rates General study characteristics Disagreements regarding the quality of the evidence were resolved by discussion between AA & AN.

Figure 1. PRISMA flow diagram of study selection.

Table 2. Quality assessment of individual studies.

Study	Randomization	Blinding	Withdrawals	Total score^b
Flint et al.	2	2	1	5/5
Meyers et al.	1^a	2	0	3/5
Wijkstra et al.	2	0	1	3/5
Bingham et al.	2	2	0	4/5
Navarro et al.	2	0	1	4/5

^aThe study was randomized, but the randomization method was not mentioned.

^bA total score of 0–2 points indicate a low-quality study, and a score of 3–5 points indicate a high-quality study.

Table 3. Patient characteristics, post-acute treatment status, and relapse definition of studies comparing AD + AP with AD monotherapy.

Study	Country	Number of participants in each treatment group^a	Age range and mean age (SD) in each medication group	Gender (%, female)	Required status after acute treatment	Definition of relapse
Flint et al.	United States and Canada	AD: 62 AD + AP: 64	- Age range: 18–85 years - AD: 55.7 (SD 14.9) - AD + AP: 55.0 (SD 15.1)	- AD: 66.1 - AD + AP: 57.8	Absence of hallucinations and delusions, and fulfilling either criteria 1 or 2 below were eligible for the stabilization phase (MMSE score ≥ 24 was also required after the stabilization phase to enter the maintenance phase): 1. HDRS^c score ≤ 10 2. HDRS score: 11–15 and ≥ 50% reduction of initial score, and graded as at least much improved on the Clinical Global Impression Scale	Fulfilling at least one of the following criteria: 1. Meeting symptomatic criteria for major depressive episode according to DSM-IV 2. HDRS score ≥ 18 3. SCID-rated psychosis 4. Having a suicide plan, attempting suicide, symptoms of mania or hypomania, or being hospitalized in a psychiatric unit
Meyers et al.	United States	AD: 13 AD + AP: 15	- Age range: 50–84 years - AD: 67.9 (SD 9.4) - AD + AP: 75.1 (SD 5.8)	- AD: 69.2 - AD + AP: 66.6	HDRS score < 10, SADS^d delusional score: 1 (indicating absence of delusions) and MMSE^e score > 24	Meeting criteria for major depression according to DSM-IV or developing delusional ideation
Wijkstra et al.	Netherlands	AD: 33^b AD + AP: 26	- Age range: 18–65 years - Imipramine: 52.0 (SD 9.6) - Venlafaxine: 53.7 (SD 6.8) - Venlafaxine and quetiapine: 49.5 (SD 11.5)	- Imipramine: 45.0 - Venlafaxine: 30.8 - Venlafaxine and quetiapine: 57.7	HDRS score ≤ 14 and a ≥ 50% decrease of the initial score	HDRS score > 14 or not maintaining a ≥ 50% decrease of the HDRS score (compared with prior acute treatment)

^aThe total number of participants, including potential drop-outs are presented.

^bThe patients receiving venlafaxine (n = 13) and imipramine (n = 20) are added together.

^cHDRS: Hamilton Depression Rating Scale; all studies used the 17 item version.

^dSADS: Schedule for Affective Disorders and Schizophrenia.

^eMMSE: Mini Mental State Examination.

Figure 2. Meta-analysis, random-effect analysis model of studies comparing AD + AP with AD monotherapy.

Table 4. GRADE scoring of studies comparing AD + AP with AD monotherapy.

Number of studies	Patients (events)	Final GRADE score	Design	Study design	Inconsistency	Indirectness	Imprecision	Publication bias	Large effect	Dose-response	Confounders
3	195 (56)	1/4	RCT	0	−1	0	−2	0	0	0	0

Table 5. Other differences and deficiencies of studies comparing AD + AP with AD monotherapy.

Study	Acute treatment	Maintenance treatment duration and follow-up time	Mean number of previous depressive episodes (SD)^b	Drop-out rate
Flint et al.	Sertraline and olanzapine	8 + 36 weeks^a	− Sertraline + olanzapine: 73.4% experiencing ≥ 2 episodes^c − Sertraline + placebo: 75.8% experiencing ≥ 2 episodes^c	9.5%
Meyers et al.	ECT	26 weeks	− Nortriptyline + perphenazine: 3.1. (1.9) − Nortriptyline + placebo: 2.6. (1.6)	Not reported
Wijkstra et al.	Same as during the maintenance phase	15 weeks	Imipramine: 1.0 (1.4) Venlafaxine: 0.4 (0.7) Venlafaxine + quetiapine: 0.9 (1.2)	10.2%

^aThe patients who responded/remitted after acute treatment continued to receive sertraline and olanzapine for 8 weeks in a stabilization phase. The patients who still responded/remitted were then (after randomization to receive either combination therapy or monotherapy) treated and followed-up for another 36 weeks. The monotherapy group had a 4 week taper-off period from olanzapine before changing to placebo.

^bNone of the studies reported whether the previous episodes had concurrent psychotic symptoms. None of the studies reported a statistically significant difference between the different treatment groups.

^cThe mean value was not reported, only the percentage of patients experiencing a minimum of two episodes was reported.

Data availability statement

All relevant data are presented in the article and supplementary material.