Comorbid Major Depressive Disorder in Schizophrenia: A Systematic Review and Meta-Analysis

Abstract

Comorbid major depressive disorder (MDD) in schizophrenia (SZ; SZ-MDD) has been identified as a major prognostic factor. However, the prevalence and associated factors of SZ-MDD have never been explored in a meta-analysis. All studies assessing the prevalence of SZ-MDD in stabilized outpatients with a standardized scale or with structured interviews were included. The Medline, Web of Science, PsycINFO, and Google Scholar databases were searched. Using random effects models, we calculated the pooled estimate of the prevalence of SZ-MDD. We used meta-regression and subgroup analyses to evaluate the potential moderators of the prevalence estimates, and we used the leave-one-out method for sensitivity analyses. Of the 5633 potentially eligible studies identified, 18 studies (n = 6140 SZ stabilized outpatients) were retrieved in the systematic review and included in the meta-analysis. The pooled estimate of the prevalence of SZ-MDD was 32.6% (95% CI: 27.9–37.6); there was high heterogeneity (I² = 92.6%), and Egger’s test did not reveal publication bias (P = .122). The following factors were found to be sources of heterogeneity: publication in or after 2015, the inclusion of patients from larger studies, the assessment tools, the inclusion of patients with substance use disorder or somatic chronic diseases, age, education level, the lifetime number of hospitalizations, and antidepressant use. Two-thirds of the extracted variables could not be explored due to an insufficient amount of published data. The prevalence of MDD is high among SZ individuals. Healthcare providers and public health officials should have an increased awareness of the burden of SZ-MDD.

Introduction

Major depressive disorder (MDD) is the most frequent mental disorder, and its impact on the global burden of disease and quality of life is increasing, especially in the wealthiest countries.¹ The development of antidepressants and talk therapies have changed the lives of countless patients. However, there remains great room for improvement, especially in schizophrenia (SZ) patients, for whom MDD was identified as the most frequent psychiatric comorbidity more than 30 years ago.² At the time, it was hypothesized that MDD may occur in more than 50% of patients with SZ.

Thirty years later, we still do not know the exact prevalence of MDD in SZ (SZ-MDD), despite numerous studies showing that it plays a major role in the prognosis of SZ, increases suicidality, increases weight gain, impairs quality of life, and impairs daily functioning independent of psychotic symptomatology.^3–5 Numerous studies have also examined the genetic overlap between MDD and SZ.⁶ Studies published in the last 3 decades have reported prevalences ranging from 16% to 69%,^7,8 and we recently identified MDD in approximately 30% of our national FondaMental Academic Center of Expertise for Schizophrenia (FACE-SZ) cohort.⁹ Such high rates may be explained by several factors. We have found that less than half of SZ patients with MDD receive antidepressants, and almost half of the patients treated with antidepressants remain unremitted.⁹ In other words, 3 out of 4 SZ patients with MDD do not receive successful treatment in contrast to the results of 3 meta-analyses indicating that antidepressants are effective for treating SZ-MDD patients.^10–12 This discrepancy may explain the high prevalences and is consistent with the lack of international consensus and guidelines for the diagnosis and treatment of SZ-MDD in the latest National Institute for Health and Care Excellence release.¹³ Altogether, these data suggest the need to improve the treatment of MDD in patients with SZ, and the first step is to identify the exact prevalence and the sources of the heterogeneity between studies.

The heterogeneity may be due to the absence of consensus on the categorial definition of SZ-MDD and the lack of a gold standard for evaluation.¹⁴ Some authors make no distinction between MDD with or without SZ and use structured clinical interviews, clinician-rated questionnaires, or self-reported questionnaires that have already been shown to contribute to heterogeneity when evaluating the prevalence of MDD. In an attempt to reduce this heterogeneity, the clinician-rated Calgary Depression Rating Scale for Schizophrenia (CDSS)¹⁵ was developed in the 1990s to specifically identify SZ-MDD patients by excluding symptoms that may be confounded with negative symptoms or antipsychotic secondary side effects, such as blunted affect or avolition.^16–19 However, this scale was not used in all recent studies, and the cutoff score to identify MDD varies across studies.

Beyond the assessment tool, many factors have been found to influence the prevalence of MDD with or without SZ, including sociodemographic data (female sex,²⁰ older age,²⁰ and unemployment²¹), addictions (tobacco, alcohol, and substance abuse),^22,23 physical health and biological disturbances (hypovitaminosis D,²⁴ metabolic syndrome,²⁵ chronic peripheral inflammation,²⁶ chronic physical pain,^27,28 and sleep disorders²⁹), treatments (first-generation antipsychotics,^30,31 lower adherence to treatment,³² and higher insight into illness³³).

In summary, as mentioned by Upthegrove et al,¹⁴ the current lack of evidence for the treatment of SZ-MDD is the result of a lack of sufficient investigation, leading to wide-ranging consequences for the illness and its prevention. MDD may, thus, be considered an important health inequality among SZ patients.³⁴

The primary objective of this systematic review and meta-analysis was to determine the prevalence of SZ-MDD in stabilized SZ community-dwelling/real-world outpatients. The secondary objective was to quantitatively evaluate associated factors, including assessment tools, sociodemographic data, clinical variables, treatments, physical health and biological variables, cognition, functioning, and quality of life that moderate estimates of SZ-MDD prevalence.

Methods

Literature Search Strategy

This meta-analysis was conducted in accordance with the Preferred Reporting Items for Systematic reviews and Meta-Analysis guidelines.³⁵ Systematic bibliographic searches were carried out according to the Cochrane methodology. This project was registered in PROSPERO (reference number CRD420181116475).

The search paradigm was based on the PubMed interface (Medline database) and adapted for 4 databases: Cochrane, ScienceDirect, PsycINFO, and Google Scholar. We only search for studies published in the last 3 decades (1990–2020) until January 18, 2020. The search paradigm was based on the following combination (depressive disorder, major[MeSH Terms]) AND (schizophrenia[MeSH Terms]). To ensure exhaustivity, the term “schizophrenia” was also combined with each identified depression assessment scale, namely, the CDSS,¹⁵ Montgomery and Åsberg Depression Rating Scale (MADRS),¹⁷ Hamilton Rating Scale for Depression,³⁶ Beck Depression Inventory,¹⁶ Center for Epidemiologic Studies-Depression Scale,³⁷ Hospital Anxiety and Depression Scale,³⁸ Patient Health Questionnaire (PHQ),¹⁹ Major Depression Inventory,³⁹ Maryland Trait and State Depression,⁴⁰ Zung Self-rating Depression Scale,⁴¹ and Geriatric Depression Scale.⁴² The full name and the abbreviation of each scale were used in the search paradigm to avoid overlooking any articles. In case of missing data, authors were contacted by email if possible.

The corresponding authors were asked to provide additional data that were not included in the original publications, and they were asked to provide any unpublished results.^43–47 The last search was carried out on January 18, 2020. The reference lists and bibliographies of relevant reviews and articles retrieved from the database searches were manually searched for additional eligible articles.

Eligibility

The inclusion criteria were as follows: (1) any language and date of publication; (2) original research papers; (3) observational studies (cohort studies or cross-sectional studies); (4) participants with a diagnosis of SZ or schizoaffective disorder according to the Diagnostic and Statistical Manual of Mental Disorders, any version, or the International Statistical Classification of Diseases and Related Health Problems; and (5) diagnosis of MDD based on a validated scale or structured interview.

The exclusion criteria were as follows: studies using a nonvalidated scale, studies using a scale without cutoff value (eg, Positive And Negative Syndrome Scale [PANSS] depressive subscore), and studies using a nonstructured interview for the diagnosis of MDD. The titles and abstracts were screened by 2 researchers (D.E.E. and G.F.). The full texts of manuscripts were then reviewed to determine whether a study would be included (D.E.E. and G.F.).

Data Extraction

The 82 extracted variables are presented in supplementary table S1. Two researchers (D.E.E. and G.F.) extracted data from the included studies in a systematic manner using a predesigned extraction form, which was based on the Joanna Briggs Institute Data Extraction Form for Prevalence and Incidence Studies.⁴⁸ Additional items relevant to the current study were also added. Each discrepancy in data extraction was examined by the first and last authors (D.E.E. and G.F.) to reach consensus.

Statistical Analyses

A random effects model was used to calculate the pooled estimate of the prevalence of SZ-MDD and its 95% CI. Heterogeneity between studies was quantified with the I² statistic.⁴⁹ Sensitivity analyses were conducted using the leave-1-out method. Publication bias was assessed graphically with a funnel plot and statistically with Egger’s test.⁵⁰ Finally, subgroup analyses for 9 binary and 1 categorial variable and univariate meta-regressions for 17 quantitative variables were used to evaluate factors that moderated the individual study estimates of the prevalence of SZ-MDD (supplementary table S1). Multivariate meta-regression was not performed due to an inadequate number of studies.⁵¹ The other variables (51 variables) were not explored due to insufficient data (reported in <4 studies). All analyses were carried out using Comprehensive Meta-Analysis software (version 3).

Role of the Funding Source

No drug manufacturing company was involved in the study design, the data collection, the data analysis, the data interpretation, the writing of the report, or the decision to submit the report for publication.

Results

Eighteen eligible studies including 6140 stabilized SZ outpatients (Birchwood et al⁵²; Bressan et al⁷; Sim et al⁵³; Maggini and Raballo⁴⁷; Conley et al⁴⁵; Gaha⁵⁴; Karadayı et al⁵⁵; Martín-Reyes et al⁴⁶; Ekinci et al⁵⁶; Majadas et al⁵⁷; Schennach et al⁴³; Jeyagurunathan et al⁸; Rossi et al⁴⁴; Chiang et al⁵⁸; Faugere et al⁵⁹; Fond et al [2018, 2019]; Arraras et al⁶⁰) were included in the quantitative analysis (flowchart shown in figure 1; study characteristics shown in table 1). The percentage of men was 64.1% in the studies in which the sex ratio was reported. One study with a small sample (n = 40) was excluded because of probable selection bias based on voluntary participation in a study targeting the link between cognitive ruminations and depressive symptoms.⁶¹

Fig. 1.

Flow chart.

Table 1.

Characteristics of studies (n = 18)

Study	Country	Sample size (n)	MDD (n)	MDD (%)	Scale	Mean age (years)	Sex ratio (M/W)
Birchwood et al77	United Kingdom	49	14	28.6	BDI	42.1	4.44
Bressan et al7	Brazil	80	13	16.3	DSMIV	38	1.22
Sim et al78	Singapore	66	11	16.7	DSMIV	28.2	1.06
Maggini and Raballo47	Italy	161	48	29.8	CDSS	35.11	2.03
Conley et al 200745	United States	2228	877	39.4	MADRS	41.8	1.6
Gaha79	Tunisia	80	21	26.3	MADRS	MD	MD
Karadayı et al80	Turkey	102	18	17.6	CDSS	31.49	1.49
Martin-Reyes et al46	Cuba	206	62	30.1	CDSS	40.75	1.31
Ekinci et al81	Turkey	100	35	35.0	CDSS	MD	1.86
Majadas et al82	Spain	90	28	31.1	CDSS	35	1.43
Schennach et al43	Germany	278	108	38.9	CDSS	34.77	1.41
Belvederi Murri et al83	Italy	89	35	39.3	CDSS	42.2	2.3
Jeyagurunathan et al8	Singapore	142	98	69.0	PHQ	40.52	1
Rossi et al44	Italy	921	262	28.5	CDSS	40.17	2.29
Faugère et al85	France	307	65	21.2	CDSS	35.74	2.23
Fond et al (2018)9	France	613	175	28.6	CDSS	32.3	2.83
Arraras et al86	Spain	141	62	43.9	CDSS	39.8	MD
Fond et al (2019)62	France	487	246	50.5	MINI	MD	MD
	Total	6140	2178

Note: BDI, Beck Depressive Inventory; CDSS, Calgary Depressive Scale of Schizophrenia; DSM, Diagnostic and Statistical Manuel of Mental Disorders; M, men; MADRS, Montgomery–Åsberg Depression Rating Scale; MD, missing data; MDD, major depressive disorder; MINI, Mini International Neuropsychiatric Interview; PHQ, Patient Health Questionnaire; W, women.

Only one study each was carried out in Africa (Tunisia), South America (Brazil), and Latin America (Cuba). The assessment of study quality is presented in supplementary tables S2 and 3. Only one study had very poor quality (quality score = 13) and 7 studies had high quality (quality score ≥25). None of the included studies were sponsored by the pharmaceutical industry.

Prevalence Estimates

The pooled estimate of the prevalence of SZ-MDD was 32.6% (95% CI: 27.9–37.6, figure 2). There was a significant amount of heterogeneity between studies (I² = 92.587, p < .001). The prevalence ranged from 16.3% to 69.0%, with the 2 highest prevalences being reported in a study including a homeless population in France (50.5%) and the only study using the PHQ questionnaire (69.0%; table 1).

Fig. 2.

Pooled prevalence of major depressive disorder in stabilized outpatients with schizophrenia in observational studies.

The funnel plot is shown in figure 3 and suggests evidence of small-study effects, but this finding was not supported by Egger’s test (p = .122). The results of the leave-1-out sensitivity analyses indicate that no single study had a disproportionate effect on the pooled estimate of the prevalence of SZ-MDD.

Fig. 3.

Funnel plot.

Subgroup Analyses (table 2)

Table 2.

Subgroup analyses. The study design is not listed because all studies were cross-sectional

	Yes		No
Variables	n	Estimated prevalence of SZ-MDD	n	Estimated prevalence of SZ-MDD	p-value
Study characteristics
Publication year (<2015 vs ≥2015)	8	39	10	27.3	.031
Western country (including North America and Europe)	10	33.8	8	30.8	.566
Inclusion of nonselected patients	11	28.6	7	39.2	.019
Multicentre recruitment	8	32.9	10	32.2	.897
Assessment tool a	3	46.6	15	30.2	.020
Clinical variables
Stabilization status (presence vs absence of a stabilization criterion in the inclusion criteria)	15	32.2	2	34.8	.750
Exclusion of patients with substance use disorder	9	27.5	9	38.2	.020
Treatments
Exclusion of patients using antidepressants	2	23.4	16	33.7	.180
Physical health and biological variables
Exclusion of patients with chronic somatic disease	10	26.6	8	40.8	.001

Note: n, number of studies; SZ-MDD, major depressive disorder in schizophrenia.

Statistically significant results (p < .05) are in bold.

^aAfter removing the single study including the homeless population. This variable had 3 modalities: structured interviews vs clinician-rated questionnaire vs self-reported questionnaire.

The results of the subgroup analyses are presented in table 2. The prevalence of SZ-MDD was found to be significantly higher in studies published in or after 2015 (39% vs 27.3%, p = .031). The prevalence was lower in studies with nonselected patients (28.6% vs 39.2%, p = .019). The prevalence was higher in studies using self-reported questionnaires (46.9% [3 studies]) than in studies using clinician rated-questionnaires (CDSS 30.1% [10 studies], MADRS (33.8% [2 studies]), and studies using structured clinical interviews (16.4% [2 studies], p < .001) after removing the only study including homeless patients, which had a high prevalence (50.5%).⁶² The prevalence was also higher in studies including patients with substance use disorders (38.2% vs 27.5%, p = .020) and patients with chronic somatic illnesses (40.8% vs 26.6%, p < .001). No significant association was found with the other variables (all p > .05).

Meta-Regression Analyses (table 3)

Table 3.

Univariate meta-regression analyses (only variables with ≥4 studies were included)

Variables	n	Beta	Lower limit	Upper limit	p-value
Sociodemographic data
Age	15	0.0637	0.0156	0.0112	.0094
Sex ratio (defined by number of men/number of women)	15	−0.1514	−0.4263	0.1236	.2806
Percentage of patients with high education levels (defined by university level or equivalent vs secondary level or lower)	4	−0.0507	−0.0702	−0.0311	.0010
Percentage of single patients	11	−2.9995	−0.646	0.4671	.0899
Percentage of unemployed patients	9	−0.0063	−0.0423	0.0297	.7306
Percentage of white patients	5	−0.0081	−0.0243	0.0082	.3322
Clinical variables
Age at illness onset	7	−0.0067	−0.1185	0.1051	.9062
Illness duration	9	0.0160	−0.0785	0.1104	.7406
Percentage of patients with schizoaffective disorder diagnosis	11	0.0183	−0.0058	0.0424	.1363
Number of lifetime hospitalizations	5	0.3068	0.0381	0.5755	.0252
Illness severity (PANSS total score)	4	0.0029	−0.0617	0.0675	.9303
Negative symptoms level (PANSS negative score or SANS score)	4	−0.0499	−0.2238	0.1241	.5741
Treatments
Percentage of patients treated with first-generation antipsychotics	6	0.0058	−0.0026	0.0143	.1750
Percentage of patients treated with second-generation antipsychotics	6	−0.0027	−0.0124	0.0071	.5919
Percentage of patients treated by antidepressants	6	0.047	0.0198	0.0742	.0007
Percentage of patients treated with mood stabilizers	4	0.0256	−0.0106	0.0618	.1663
Percentage of patients treated with anxiolytics	4	0.0009	−0.0151	0.0168	.9165

Note: Statistically significant results (p < .05) are in bold.

n, number of studies; PANSS, Positive and Negative Syndrome Scale.

The results of the meta-regression analyses are presented in table 3, and the figures for significant results are presented in supplementary figure S1. A higher prevalence of SZ-MDD was associated with older age (b = 0.064, 95% CI = 0.016–0.112, p = .009), a lower percentage of patients with high education levels (b = −0.051, 95% CI = −0.070 to −0.031, p < .001), a higher number of lifetime hospitalizations (b = 0.307, 95% CI = 0.038–0.576, p = .025), and a higher percentage of patients treated with antidepressants (b = 0.047, 95% CI = 0.020–0.074, p < .001). No significant association was found with the other variables (all p > .05).

Discussion

The results of the present study may be summarized as follows. Eighteen studies (n = 6140 SZ patients) were included. The pooled estimate of the prevalence of MDD in stabilized SZ outpatients was 32.6% (95% CI = 27.9–37.6). No publication bias was observed according to Egger’s test. There was high heterogeneity across studies. The following factors were significantly associated with the SZ-MDD prevalence: publication in or after 2015, the inclusion of nonselected patients, the assessment tools (structured interviews vs clinician-rated questionnaire vs self-reported questionnaire), the inclusion of patients with substance use disorder or somatic chronic diseases, older age, the percentage of patients with high education levels, the number of lifetime hospitalizations, and the percentage of patients treated with antidepressants. No significant association was found for the other variables analyzed herein. Two-thirds of the extracted variables could not be explored due to an insufficient number of studies reporting these items.

SZ-MDD was identified in almost one-third of SZ-stabilized outpatients, which is approximately 2–3 times higher than in the non-SZ population.^1,63 We also found that studies published in or after 2015 reported higher rates of SZ-MDD, which may suggest an increase in the prevalence of MDD in this population, similar to the increased prevalence of MDD in patients without SZ as reported by the World Health Organization.⁶⁴ Because MDD has been identified as a major prognostic factor of SZ, healthcare providers and public health officials should have an increased awareness of the burden of SZ-MDD. Our results highlight the need for the systematic screening and treatment of MDD in SZ outpatients.

Our results highlight the lack of consensus on the evaluation of SZ-MDD patients, as indicated by the discrepancies in the prevalence among studies using structured interviews, clinician-rated questionnaires, and self-reported questionnaires (even for assessments specifically designed to evaluate SZ-MDD). This lack of clear guidelines for diagnosing SZ-MDD is, thus, an important barrier for management and treatment based on current knowledge, evidence, and available treatments. The analysis of the effect of assessment tools showed that the prevalence of MDD was lower among patients evaluated by structured interviews than in those evaluated with a clinician-rated scale; there were similar rates between the CDSS and the MADRS (the CDSS was developed to specifically assess depressive symptoms in SZ). The prevalence was also higher among patients who were diagnosed using self-reported questionnaires. Self-reported tools and clinician-rated tools have shown high levels of agreement in patients without SZ.⁶⁵ This may be due to increased neuroticism in SZ, a personality trait that may induce a response bias with respect to self-reported depressive symptoms.⁶⁶ The interpretation of our results, thus, remains unclear: do SZ patients overestimate their depressive symptoms, or do clinicians miss depressive symptoms due to inadequate tools? The debate remains open and requires further investigations.

Among sociodemographic data, we have identified that studies carried out among older patients and patients with lower education levels reported higher rates of MDD. The older patients should be targeted; however, most of the samples had a mean age ranging from 28 to 42 years. Further studies should explore SZ-MDD in more specific age ranks (such as 18–25 and >50 years). In particular, MDD onset has been suggested to begin before SZ onset (even 4 years earlier),⁶⁷ which may explain the absence of an association between MDD prevalence and age at SZ onset or SZ duration in our results. It suggests that preventive interventions, such as psychoeducation, and lifestyle interventions, including tobacco cessation, dieting, and physical activity training, should also be carried out in younger patients to detect early depressive symptoms.⁶⁸ The lower prevalence of MDD in patients with higher education levels is consistent with studies in non-SZ populations.⁶⁹ Patients with higher education levels have more cognitive resources to cope with depressive symptoms. Higher education levels may also be a reflection of better care at home during childhood (eg, better diets, less abuse, and less neglect). This suggests that supplementary efforts should be made to prevent SZ-MDD in patients with lower levels of education. We had no data on patients’ income level, and thus, we could not determine if this association was due to higher financial resources, which is a limitation of our study. One of the most striking results regarding the sociodemographic data is the absence of an effect of sex on the prevalence. In individuals without SZ, MDD is 2–3 times more common in women than in men.⁷⁰ This suggests that the mechanisms of MDD onset in SZ patients may be partly specific and not influenced by hormones and that the increased prevalence of SZ-MDD patients may be mostly due to its increased prevalence in men. This may be consistent with previous findings suggesting dehydroepiandrosterone metabolism dysfunction in patients with SZ.⁷¹

Among clinical and physical health variables, our results suggest that substance abuse and chronic somatic illnesses may play a role in the increased prevalence of SZ-MDD. These associations have been described for individuals without SZ,^72,73 yet addictions and physical health remain insufficiently documented in SZ studies. For example, none of the studies reported the rate of tobacco smoking, but tobacco has been associated with an increased risk of MDD in the general population.⁷⁴ More than half of SZ patients are smokers or heavy smokers.⁷⁵ We also found no data regarding alcohol use disorder, metabolic syndrome, chronic pain, sexual dysfunctions, peripheral inflammation, or hypovitaminosis D, despite recent findings suggesting that these factors were associated with MDD.^12,76–79 We found no associations of SZ-MDD prevalence with global psychotic symptomatology, negative symptomatology, or stabilization status. This suggests that MDD should be considered a comorbidity rather than a dimension of SZ that could be connected to other dimensions as currently presented in the PANSS 5-factor structure.⁸⁰ We found that studies including patients with a higher number of lifetime hospitalizations reported a higher prevalence of SZ-MDD. Hospitalizations may be a severity marker and induce social withdrawal and autonomy loss, thereby leading to MDD; however, these results were not adjusted for age, which may be a key confounding factor. Hospitalizations may also play a role in MDD onset via psychotrauma, especially for hospitalization under restraint.⁸¹

Strengths

The number of participants (>6000 patients) and the large number of studies carried out in miscellaneous countries are strengths of the present article. No publication bias was detected, which strengthens the confidence in the findings. We have identified several sources of heterogeneity.

Limitations and Perspectives

Few studies have explored SZ-MDD, and they fail to answer very basic questions of real-life importance for SZ patients living with MDD. For example, why are there no sex differences in SZ-MDD, suggesting that male patients with SZ are at greater risk of MDD compared to women? Tobacco smoking is present in more than half of SZ patients, but is it a risk factor for SZ-MDD, and does smoking cessation improve MDD? How much do physical activity, diet, hypovitaminosis D, peripheral inflammation, folate deficiency, and metabolic syndrome participate in SZ-MDD onset and maintenance? We still do not know if the first MDD episode should be treated like the others in SZ, and the question is much more difficult given that the first MDD episode may occur far before SZ onset. Cognition, functioning, and quality of life are presumably impaired in SZ patients with MDD, yet the studies exploring these associations do not provide enough data to draw conclusions. Lifestyle factors (sleep, diet/nutrients/probiotics, and physical activity) may also have a strong impact on SZ-MDD and have not been reported in the included studies; these factors do have a strong impact on MDD in individuals without SZ.^82,84,85 Another external validity issue is that all studies recruited patients in the hospital or hospital-related ambulatory settings, which may induce a recruitment bias and limit the external validity of these findings to the whole population of patients with SZ. Our results also highlight the lack of data on MDD history (age at first MDD episode, lifetime number of MDD episodes, and the mean number of previous antidepressant treatments). Future studies should examine these factors. More studies should also be carried out in homeless SZ patients as homelessness is a probable risk factor for SZ-MDD.⁸⁶ The study with the highest prevalence (69%) used the PHQ, which has not been validated in SZ. Our results indicate that this study was an outlier, which suggests that the PHQ probably overestimates SZ-MDD prevalence and should not be used in future studies. Some studies may have presented depressive symptoms in a quantitative manner (without cutoff) and could be further analyzed, especially for studies presenting the PANSS depressive subscore (that has no cutoff).

Conclusion

SZ-MDD was identified in almost one-third of SZ-stabilized outpatients. Because MDD has been identified as a major prognostic factor of SZ, healthcare providers and public health officials should have an increased awareness of the burden of SZ-MDD. A consensus and clear guidelines for diagnosing SZ-MDD are needed to improve its care. Older and less-educated patients, substance abuse, and chronic somatic illness should be targeted in priority to guide screening, treatment, and prevention. Our results also highlight a lack of sufficient investigation on important research issues. Two-thirds of the potentially associated factors were not sufficiently explored, such as tobacco, alcohol, hypovitaminosis D, or inflammation associated with MDD in patients without SZ; thus, we could not draw conclusions on how these factors influence the prevalence of SZ-MDD. Further research is urgently needed to better understand how to screen, treat, and prevent SZ-MDD.

Funding

This study was funded by Hôpitaux Universitaires de Marseille, Fondation FondaMental, and Aix-Marseille Université.