Introduction
Schizophrenia (Sz) is a complex brain disorder characterized by a wide array of symptoms, including delusions, hallucinations, disorganized behavior, and negative symptoms that together represent the diagnostic criteria for this unique disease (American Psychiatric Association, 2013). Although cognitive deficits have been long considered a core feature of SZ, they are not included in the DSM-5 or the ICD-10 criteria (Keefe & Fenton, Reference Keefe and Fenton2007). Impairment in a broad range of cognitive domains has been consistently reported in individuals with SZ, including attention, working memory, verbal learning and memory, and executive functions (Bowie & Harvey, Reference Bowie and Harvey2006). These deficits are moderate to severe and are present before the onset of frank psychosis (Davidson et al., Reference Davidson, Reichenberg, Rabinowitz, Weiser, Kaplan and Mark1999). In fact, different studies of neuropsychological functioning in ultra-high-risk subjects demonstrated cognitive impairments that were intermediate between healthy controls and patients with first-episode of psychosis (FEP) (Hambrecht, Lammertink, Klosterkötter, Matuschek, & Pukrop, Reference Hambrecht, Lammertink, Klosterkötter, Matuschek and Pukrop2002; Hawkins et al., Reference Hawkins, Addington, Keefe, Christensen, Perkins, Zipurksy and McGlashan2004). Thus, neuropsychological deficits after a FEP appeared to remain stable over time (Bozikas & Andreou, Reference Bozikas and Andreou2011; Sánchez-Torres et al., Reference Sánchez-Torres, Moreno-Izco, Lorente-Omeñaca, Cabrera, Lobo, González-Pinto and Balanzá-Martínez2018).
The developmental origins of health and disease (DOHaD) paradigm represents a framework that posits that environmental perturbations during the perinatal period or early life may lead to adverse health outcomes in adulthood (Gluckman & Hanson, Reference Gluckman and Hanson2006; Hanson & Gluckman, Reference Hanson and Gluckman2014). Among this environmental factors, obstetric complications (OCs) have been found to increase the risk of a plethora of chronic diseases, including mental health conditions such as psychosis (Cannon, Jones, & Murray, Reference Cannon, Jones and Murray2002; Davies et al., Reference Davies, Segre, Estradé, Radua, De Micheli, Provenzani and Fusar-Poli2020; Garcia-Rizo & Bitanihirwe, Reference Garcia-Rizo and Bitanihirwe2020; Garcia-Rizo, Fernandez-Egea, Bernardo, & Kirkpatrick, Reference Garcia-Rizo, Fernandez-Egea, Bernardo and Kirkpatrick2015; O'Donnell & Meaney, Reference O'donnell and Meaney2017; Radua et al., Reference Radua, Ramella-Cravaro, Ioannidis, Reichenberg, Phiphopthatsanee, Amir and Fusar-Poli2018). Due to its heterogeneity, OCs could be at least described as difficulties during pregnancy and delivery with different outcomes (Mezquida et al., Reference Mezquida, Fernandez-Egea, Treen, Mané, Bergé, Savulich and Garcia-Rizo2018), as suggested by Cannon (Cannon et al., Reference Cannon, Jones and Murray2002). Different difficulties during the perinatal period have long-lasting consequences on the cognitive profile of the general population, not only in childhood (Stålnacke, Tessma, Böhm, & Herlenius, Reference Stålnacke, Tessma, Böhm and Herlenius2019), but also, during adulthood (Flensborg-Madsen & Mortensen, Reference Flensborg-Madsen and Mortensen2017). Early prenatal and postnatal periods are extremely sensitive for the subsequent development of neuronal function and the programming of later behavior (Krugers & Joëls, Reference Krugers and Joëls2014). For instance, preterm birth correlates with the later cognitive outcomes (Baron & Rey-Casserly, Reference Baron and Rey-Casserly2010; Mathewson et al., Reference Mathewson, Chow, Dobson, Pope, Schmidt and Van Lieshout2017). Similarly, hypoxic insults during mid- to late pregnancy have been reported to have a sex-specific effect on intelligence quotient (IQ) (Anastario, Salafia, Fitzmaurice, & Goldstein, Reference Anastario, Salafia, Fitzmaurice and Goldstein2012). As with brain morphology, cognition is also affected by the specific timing of the insult during brain development (Gee & Casey, Reference Gee and Casey2015).
Specifically, patients with SZ show a higher prevalence of complications during the perinatal period (e.g. infections (Brown, Reference Brown2006), maternal stress (Khashan et al., Reference Khashan, Abel, McNamee, Pedersen, Webb, Baker and Mortensen2008) which beyond increasing the risk of psychiatric diagnosis also promote a diverse array of detrimental outcomes ranging from metabolic dysfunction (Garcia-Rizo & Bitanihirwe, Reference Garcia-Rizo and Bitanihirwe2020) and neuroimaging anomalies (Costas, Garcia-Rizo, Bitanihirwe, & Penades, Reference Costas, Garcia-Rizo, Bitanihirwe and Penades2020) to cognitive and clinical psychopathology (Mezquida et al., Reference Mezquida, Fernández-Egea, Treen, Mané, Bergé, Savulich and García-Rizo2021). Indeed, these perinatal events might act as ‘scars’ which affect the brain in different pathways and so different abnormalities in functionality or structure might be expected (Insel, Reference Insel2010). Thus, evidence suggests that, within patients with psychosis or related disorders, there is a relationship between the presence of OCs and poor cognitive performance in different domains (Borkowska & Rybakowski, Reference Borkowska and Rybakowski2002; Brown et al., Reference Brown, Vinogradov, Kremen, Poole, Bao, Kern and McKeague2011, Reference Brown, Vinogradov, Kremen, Poole, Deicken, Penner and Schaefer2009; Ellman, Yolken, Buka, Torrey, & Cannon, Reference Ellman, Yolken, Buka, Torrey and Cannon2009; Gilvarry et al., Reference Gilvarry, Takei, Russell, Rushe, Hemsley and Murray2000, Reference Gilvarry, Russell, Jones, Sham, Hemsley and Murray2001; Holthausen et al., Reference Holthausen, Wiersma, Sitskoorn, Hijman, Dingemans, Schene and van den Bosch2002; Ochoa et al., Reference Ochoa, Huerta-Ramos, Barajas, Iniesta, Dolz, Baños and Usall2013; Torniainen et al., Reference Torniainen, Wegelius, Tuulio-Henriksson, Lonnqvist, Suvisaari, Lönnqvist and Suvisaari2013; Yurgelun-Todd & Kinney, Reference Yurgelun-Todd and Kinney1993), even though results are heterogeneous. In fact, these cognitive disturbances, which sometimes are present at the time of onset of the first symptoms (Bora & Murray, Reference Bora and Murray2014) might be somehow related to the presence of perinatal events.
In this context, our aim was to systematically review the available studies investigating the relationship between OCs and cognition in SZ and related psychotic disorders, using meta-analytic methods whenever possible. We hypothesized that OCs are related to even lower cognitive performance in patients diagnosed with SZ.
Materials and methods
Design
Data for the systematic review and meta-analyses were gathered in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Protocol (PRISMA-P) checklist (Moher, Liberati, Tetzlaff, & Altman, Reference Moher, Liberati, Tetzlaff and Altman2009) (see online Supplementary Material 1). The review process has been registered with the international prospective register of systematic reviews (PROSPERO): 2018 CRD42018094238.
Literature search strategy
A systematic and computerized literature search was conducted in different scientific databases including Medline via Pubmed, EMBASE via OVID, SCOPUS, and the Cochrane Library.
The search strategy was performed using key terms relating to ‘obstetric complications’, ‘cognition’ and ‘schizophrenia’ or ‘other psychotic disorders’. Reference lists of included articles were hand-searched for additional studies. A snowballing approach also was applied to identify additional studies meeting the inclusion criteria in the reference lists of studies that met the inclusion criteria. The full search strategy is described in online Supplementary Material 2.
Inclusion and exclusion criteria
The present study aimed to identify the literature focusing on the association between OCs and cognitive impairment in individuals with SZ and other psychotic disorders. We only included studies in which cognitive domains were evaluated in SZ patients with (cases) and without (controls) OCs. Included studies were published in English or Spanish, and on populations aged over 7 years old. We excluded commentaries, editorials, expert reviews, case reports, case series, or literature that did not provide novel information/research. No restrictions were placed on settings or time frames.
Papers published until the end of January 2022, and fulfilling the inclusion criteria were considered. If a study did not explicitly report data (e.g. expert opinions, comments, and editorials), we contacted the main authors to retrieve and ensure the quality of data.
Data extraction and quality assessment
Two independent reviewers (SA and MG) evaluated the identified articles. After all, potentially relevant studies were identified from each respective database; duplicates were removed using Mendeley software. Titles and abstracts of the remaining studies were screened. Firstly, MG independently screened studies with titles and abstracts and excluded those irrelevant to the research question. Whether a study's title suggested that it may contain relevant data/content, the abstract was assessed. Secondly, SA and MG assessed abstracts and excluded those which did not meet the inclusion criteria. Finally, whether the abstract indicated that the study provided information of relevance with regard to the effects of OCs on the cognition of SZ subjects, the full-text article of the study was read in its entirety to determine its eligibility. Then the suitability of these remaining full-text articles was subsequently evaluated, and so unrelated or nonapplicable studies were excluded following inclusion and exclusion criteria. During the process, discrepancies between the two reviewers were resolved by discussion with a third reviewer (CGR).
As with the eligibility phase, quality assessment was conducted by two authors (BB and CGR) using the Newcastle-Ottawa Scale (NOS) (Wells, Shea, O'Connell, & Peterson, Reference Wells, Shea, O'Connell and Peterson2000). This tool is divided into three sections that evaluate three quality parameters (selection, comparability, and outcome) divided across seven specific items.
Exposure and outcome measures
Exposure to OCs was defined as a wide range of events such as pregnancy complications, abnormal fetal growth, and delivery complications, which have already been related with an increased risk of SZ (Cannon et al., Reference Cannon, Jones and Murray2002). Since different methodologies have been described along the literature to describe the presence of a wide range of OCs, authors decided to include OCs data as a dichotomous variable: presence or absence of OCs. Additionally, seven studies have already been divided the groups according to the presence or absence of OC (Borkowska & Rybakowski, Reference Borkowska and Rybakowski2002; Brown et al., Reference Brown, Vinogradov, Kremen, Poole, Deicken, Penner and Schaefer2009, Reference Brown, Vinogradov, Kremen, Poole, Bao, Kern and McKeague2011; Gilvarry et al., Reference Gilvarry, Takei, Russell, Rushe, Hemsley and Murray2000, Reference Gilvarry, Russell, Jones, Sham, Hemsley and Murray2001; Ochoa et al., Reference Ochoa, Huerta-Ramos, Barajas, Iniesta, Dolz, Baños and Usall2013; Yurgelun-Todd & Kinney, Reference Yurgelun-Todd and Kinney1993). Whether OCs data was reported on a continuous variable through scatter plot, we used a tool called WebPlotDigitizer to extract graphed data (Drevon, Fursa, & Malcolm, Reference Drevon, Fursa and Malcolm2017). Then, we split the continuous variables using the cut-off of OCs according to the previous definition (Cannon et al., Reference Cannon, Jones and Murray2002). Besides, we combined two means and standard deviations into one mean and standard deviation when the authors reported different measures of dispersions for subjects with SCZ but with distinctive OCs (Altman, Machin, Bryant, & Gardner, Reference Altman, Machin, Bryant and Gardner2000; Higgins et al., Reference Higgins, Thomas, Chandler, Cumpston, Li, Page and Welch2019).
The primary outcome was cognitive performance. It was grouped considering the most evaluated and affected domains in SZ (Bowie & Harvey, Reference Bowie and Harvey2006), as follows:
(1) Verbal memory: Assessed with the California Verbal Learning Test (CVLT) (Delis, Kramer, Kaplan, & Ober, Reference Delis, Kramer, Kaplan and Ober1987) or ‘Test Aprendizaje Verbal España-Complutense’ [Complutense Spanish Verbal Learning Test] (TAVEC) (Benedet & Alexandre, Reference Benedet and Alexandre1998).
(2) Working memory: Assessed with the Letter-Number Sequencing Subtest of the Wechsler Adult Intelligence Scale (WAIS) (Wechsler, Reference Wechsler1997) or Wechsler Intelligence Scale for Children (WISC) (Corral, Arribas, Santamaría, Sueiro, & Pereña, Reference Corral, Arribas, Santamaría, Sueiro and Pereña2005) and Auditory N-back (2-back) test (Keefe & Fenton, Reference Keefe and Fenton2007).
(3) Executive functions: Evaluated using the interference score from the Stroop test (Golden, Reference Golden1978; Macleod, Reference Macleod1991) and Wisconsin Card Sorting Test (WCST) (Grant & Berg, Reference Grant and Berg1948).
(4) Verbal fluency: Assessed with verbal fluency (category) and Thurstone word fluency test (Thurstone, Reference Thurstone1938).
(5) Processing speed: Tested with the Trail Making Test, form A (TMT-A) (Reitan & Wolfson, Reference Reitan and Wolfson1995) and the Symbol Search subtest from WAIS.
(6) Attention: Tested with the Continuous Performance Test (CPT) (Conners, Reference Conners2002) and the Digit Span Forward subtest of WAIS.
(7) IQ: Assessed with the National Adult Reading Test (NART) (Nelson, Reference Nelson1982) and with the Vocabulary subtest of WAIS/WISC-IV.
Statistical analyses
When adequate, a meta-analysis was performed using R software (version 4.1.0) and RStudio (version 1.4.1103) with packages ‘meta’ (version 4.17) and ‘metafor’ (version 2.4). We calculated effect sizes (Hedges' g) of cognitive domains within each included study comparing the means and standard deviation in subjects with SZ and OCs (cases) v. individuals with SZ without OCs (controls). The values are expressed as the mean Hedges' g across studies along with 95% confidence interval (CI). We calculated Hedges' g because it is appropriate for assessing bias in small sample size (Hedges & Olkin, Reference Hedges and Olkin1985). A negative Hedges' g value means a poor performance of subjects with SZ exposed to OCs compared to nonexposed subjects with SZ. The effect size of Hedges' g was interpreted as follows: 0.2 = small, 0.5 = medium, and 0.8 = large. The neuropsychological assessments are highly heterogeneous; because of this, we will consider a random-effects model rather than the fixed-effect model, since the random-effects model is more conservative.
Additionally, we reported the I 2 statistic to quantify the proportion of between-study variability. Values of I 2 of 25%, 50%, and 70% were considered as small, moderate, and high proportions of heterogeneity, respectively (Higgins, Thompson, Deeks, & Altman, Reference Higgins, Thompson, Deeks and Altman2003). We tested homogeneity using the Q-statistic(X 2). The funnel plot was used for visual identification of bias or systematic heterogeneity. When there is no publication bias, all studies would lie symmetrically around the striped line (pooled effect size). After that, we performed Eggers' test to determine statistically the presence of publication bias (Egger, Smith, Schneider, & Minder, Reference Egger, Smith, Schneider and Minder1997). We considered a p value <0.05 significant in all statistical analysis.
Results
Study selection
The electronic database search yielded 4166 articles (see Fig. 1). Following the article screening of titles and abstracts, 22 full-text articles were obtained for detailed eligibility assessment. Full-text copies of those articles were obtained and when it was feasible, the authors were contacted directly for unpublished data and additional information. In this respect, five authors were contacted via e-mail and one author provided retrieved nonpublished data from the publication (Ochoa et al., Reference Ochoa, Huerta-Ramos, Barajas, Iniesta, Dolz, Baños and Usall2013). Following this process, 10 articles were included for the qualitative assessment, and eight of which underwent meta-analysis (Borkowska & Rybakowski, Reference Borkowska and Rybakowski2002; Brown et al., Reference Brown, Vinogradov, Kremen, Poole, Bao, Kern and McKeague2011, Reference Brown, Vinogradov, Kremen, Poole, Deicken, Penner and Schaefer2009; Ellman et al., Reference Ellman, Yolken, Buka, Torrey and Cannon2009; Gilvarry et al., Reference Gilvarry, Takei, Russell, Rushe, Hemsley and Murray2000. Reference Gilvarry, Russell, Jones, Sham, Hemsley and Murray2001; Holthausen et al., Reference Holthausen, Wiersma, Sitskoorn, Hijman, Dingemans, Schene and van den Bosch2002; Ochoa et al., Reference Ochoa, Huerta-Ramos, Barajas, Iniesta, Dolz, Baños and Usall2013; Torniainen et al., Reference Torniainen, Wegelius, Tuulio-Henriksson, Lonnqvist, Suvisaari, Lönnqvist and Suvisaari2013; Yurgelun-Todd & Kinney, Reference Yurgelun-Todd and Kinney1993), while 12 were excluded due to a reason (see online Supplementary Material 3).
Fig. 1. PRISMA flow diagram. Presentation of the procedure of literature searching and selection with numbers of articles at each stage.
Quality assessment of the included studies showed that three studies were of high quality (score of >7 out of 9) and the remaining seven were of moderate quality (scoring 5 to 7 out of 9) (see Table 1). Inter-rater reliability between the reviewers on the NOS was evaluated by means of Cohen's kappa coefficient (κ), where the strength of κ was interpreted as follows: <0.20 poor; 0.21–0.40 fair; 0.41–0.60 moderate 0.61–0.80 substantial; 0.81–1.00 almost perfect (Altman, Reference Altman1991). The overall interassessment agreement for total NOS scores was moderate (κ = 0.57).
Table 1. List and principal characteristics of included studies in the systematic review and meta-analysis
Cases, Schizophrenia spectrum disorder patients affected with an obstetric complication; Controls, Schizophrenia spectrum disorder patients without an obstetric complication; SZ, Schizophrenia; DSM, Diagnostic and Statistical Manual of Mental Disorders; SADS-L, Schedule for Affective Disorders and Schizophrenia, lifetime version; RDC, Research Diagnostic Criteria; FEP, First-episode psychosis; WCST, Wisconsin Card Sort Test; NART, National Adult Reading Test; TVFT, Thurstone's Verbal Fluency Test; TMT, Trail Making Test; CPT, Continuous Performance Task; CVLT, California Verbal Learning Test; WAIS, Wechsler Adult Intelligence Scale; LNS, Letter-Number Sequencing; FIQ, full scale IQ; PIQ, performance IQ; VIQ, verbal IQ; WISC, Wechsler Intelligence Scale for Children; TAVEC, ‘Test Aprendizaje Verbal España-Complutense’ [Complutense Spanish Verbal Learning Test]; TAVECI, ‘Test Aprendizaje Verbal España-Complutense Infantil’ [Complutense Spanish Verbal Learning Test for Children]; WMS, Wechsler Memory Scale; OC + , Patients who had been exposed to obstetric complications; OC-, Patients who had not been exposed to obstetric complications; BW, Birth Weight; HBW, High Birth Weight; LBW, Low Birth Weight; OC, Obstetric complication.
a Newcastle-Ottawa Scale.
Characteristics of included studies
A total of 693 subjects with SZ or related disorders were included across the different studies (range: 10–174), of which 37.95% (n = 263) presented with an OC. The main psychiatric diagnostic criteria applied were the Diagnostic and Statistical Manual of Mental Disorders, 4th Edition (DSM-IV). In terms of neurocognitive assessment, the most frequently evaluated domains were executive functioning (n of studies = 7) (Borkowska & Rybakowski, Reference Borkowska and Rybakowski2002; Brown et al., Reference Brown, Vinogradov, Kremen, Poole, Bao, Kern and McKeague2011, Reference Brown, Vinogradov, Kremen, Poole, Deicken, Penner and Schaefer2009; Holthausen et al., Reference Holthausen, Wiersma, Sitskoorn, Hijman, Dingemans, Schene and van den Bosch2002; Ochoa et al., Reference Ochoa, Huerta-Ramos, Barajas, Iniesta, Dolz, Baños and Usall2013; Torniainen et al., Reference Torniainen, Wegelius, Tuulio-Henriksson, Lonnqvist, Suvisaari, Lönnqvist and Suvisaari2013; Yurgelun-Todd & Kinney, Reference Yurgelun-Todd and Kinney1993), followed by attention (n = 5) (Brown et al., Reference Brown, Vinogradov, Kremen, Poole, Deicken, Penner and Schaefer2009; Ellman et al., Reference Ellman, Yolken, Buka, Torrey and Cannon2009; Holthausen et al., Reference Holthausen, Wiersma, Sitskoorn, Hijman, Dingemans, Schene and van den Bosch2002; Ochoa et al., Reference Ochoa, Huerta-Ramos, Barajas, Iniesta, Dolz, Baños and Usall2013; Torniainen et al., Reference Torniainen, Wegelius, Tuulio-Henriksson, Lonnqvist, Suvisaari, Lönnqvist and Suvisaari2013) and processing speed (n = 5) (Brown et al., Reference Brown, Vinogradov, Kremen, Poole, Bao, Kern and McKeague2011; Ellman et al., Reference Ellman, Yolken, Buka, Torrey and Cannon2009; Holthausen et al., Reference Holthausen, Wiersma, Sitskoorn, Hijman, Dingemans, Schene and van den Bosch2002; Ochoa et al., Reference Ochoa, Huerta-Ramos, Barajas, Iniesta, Dolz, Baños and Usall2013; Torniainen et al., Reference Torniainen, Wegelius, Tuulio-Henriksson, Lonnqvist, Suvisaari, Lönnqvist and Suvisaari2013). When exploring how OCs were measured, four articles used an objective measure such as birth weight (Torniainen et al., Reference Torniainen, Wegelius, Tuulio-Henriksson, Lonnqvist, Suvisaari, Lönnqvist and Suvisaari2013) or maternal mother infection antibody testing (Brown et al., Reference Brown, Vinogradov, Kremen, Poole, Bao, Kern and McKeague2011, Reference Brown, Vinogradov, Kremen, Poole, Deicken, Penner and Schaefer2009; Ellman et al., Reference Ellman, Yolken, Buka, Torrey and Cannon2009), while health records/registries, or patient/mother interview/scales were used in the other articles. See Table 1 for further details.
Qualitative synthesis of outcomes
Verbal memory
Two studies, one focusing on early onset of psychosis patients with a FEP (Ochoa et al., Reference Ochoa, Huerta-Ramos, Barajas, Iniesta, Dolz, Baños and Usall2013) and another with chronic SZ (Brown et al., Reference Brown, Vinogradov, Kremen, Poole, Bao, Kern and McKeague2011), described that SZ patients who had experienced OCs performed worse in this domain compared to controls. A separate study of FEP and a second episode of psychosis patients did not find differences in OCs between those with normal cognition and those with cognitive impairment (Holthausen et al., Reference Holthausen, Wiersma, Sitskoorn, Hijman, Dingemans, Schene and van den Bosch2002).
Working memory
Two of four studies (Ochoa et al., Reference Ochoa, Huerta-Ramos, Barajas, Iniesta, Dolz, Baños and Usall2013; Torniainen et al., Reference Torniainen, Wegelius, Tuulio-Henriksson, Lonnqvist, Suvisaari, Lönnqvist and Suvisaari2013) described an association between OCs and impaired working memory, while another (Brown et al., Reference Brown, Vinogradov, Kremen, Poole, Bao, Kern and McKeague2011) reported a trend for an association between maternal infection and the composite score for working memory (p = 0.051). Contrary, one study did not observe any such differences (Brown et al., Reference Brown, Vinogradov, Kremen, Poole, Deicken, Penner and Schaefer2009).
Executive function
Four of the seven studies that investigated the relationship between OCs and executive functioning showed an association between impaired executive performance and a history of OCs (Borkowska & Rybakowski, Reference Borkowska and Rybakowski2002; Brown et al., Reference Brown, Vinogradov, Kremen, Poole, Deicken, Penner and Schaefer2009; Ochoa et al., Reference Ochoa, Huerta-Ramos, Barajas, Iniesta, Dolz, Baños and Usall2013; Torniainen et al., Reference Torniainen, Wegelius, Tuulio-Henriksson, Lonnqvist, Suvisaari, Lönnqvist and Suvisaari2013). The three remaining studies did not observe such association (Brown et al., Reference Brown, Vinogradov, Kremen, Poole, Bao, Kern and McKeague2011; Holthausen et al., Reference Holthausen, Wiersma, Sitskoorn, Hijman, Dingemans, Schene and van den Bosch2002; Yurgelun-Todd & Kinney, Reference Yurgelun-Todd and Kinney1993).
Verbal fluency
None of the four studies that evaluated this domain reported differences between those patients with and without a history of OCs (Brown et al., Reference Brown, Vinogradov, Kremen, Poole, Bao, Kern and McKeague2011, Reference Brown, Vinogradov, Kremen, Poole, Deicken, Penner and Schaefer2009; Gilvarry et al., Reference Gilvarry, Russell, Jones, Sham, Hemsley and Murray2001; Holthausen et al., Reference Holthausen, Wiersma, Sitskoorn, Hijman, Dingemans, Schene and van den Bosch2002).
Attention
The Five studies that evaluated attention described different results. attention found that patients with OCs had worse performance when evaluated by forward digits (Ellman et al., Reference Ellman, Yolken, Buka, Torrey and Cannon2009; Torniainen et al., Reference Torniainen, Wegelius, Tuulio-Henriksson, Lonnqvist, Suvisaari, Lönnqvist and Suvisaari2013) or CPT (Ochoa et al., Reference Ochoa, Huerta-Ramos, Barajas, Iniesta, Dolz, Baños and Usall2013), while the other studies did not report any significant differences (Brown et al., Reference Brown, Vinogradov, Kremen, Poole, Deicken, Penner and Schaefer2009; Holthausen et al., Reference Holthausen, Wiersma, Sitskoorn, Hijman, Dingemans, Schene and van den Bosch2002)
Processing speed
Five articles assessed this domain. Two of them found evidence that OCs were associated with worse performance in processing speed (Ochoa et al., Reference Ochoa, Huerta-Ramos, Barajas, Iniesta, Dolz, Baños and Usall2013; Torniainen et al., Reference Torniainen, Wegelius, Tuulio-Henriksson, Lonnqvist, Suvisaari, Lönnqvist and Suvisaari2013), while the remaining three studies did not find such effects (Brown et al., Reference Brown, Vinogradov, Kremen, Poole, Deicken, Penner and Schaefer2009; Ellman et al., Reference Ellman, Yolken, Buka, Torrey and Cannon2009; Holthausen et al., Reference Holthausen, Wiersma, Sitskoorn, Hijman, Dingemans, Schene and van den Bosch2002).
Intellectual quotient
While Torniainen et al., found that there was not a significant difference between those patients with and without OCs (Torniainen et al., Reference Torniainen, Wegelius, Tuulio-Henriksson, Lonnqvist, Suvisaari, Lönnqvist and Suvisaari2013), another article described a higher IQ for those who were not exposed (Ochoa et al., Reference Ochoa, Huerta-Ramos, Barajas, Iniesta, Dolz, Baños and Usall2013). In a third study that used a prospective birth cohort with 111 psychotic patients, it was found that those exposed to influenza B had a nonsignificant reduction in IQ (as measured by WISC vocabulary subscale) (Ellman et al., Reference Ellman, Yolken, Buka, Torrey and Cannon2009).
Quantitative synthesis of outcomes
The eight studies included in the meta-analysis comprised of 401 participants [median (interquartile range) participants per study, 39 (24.5–57.5)]. A total of 276 (90.49%) were male, but three studies did not report the frequencies of male subjects with SZ (Brown et al., Reference Brown, Vinogradov, Kremen, Poole, Bao, Kern and McKeague2011, Reference Brown, Vinogradov, Kremen, Poole, Deicken, Penner and Schaefer2009; Gilvarry et al., Reference Gilvarry, Takei, Russell, Rushe, Hemsley and Murray2000).
SZ subjects with OCs showed poor verbal memory performance in the random-effect model compared to those without OCs [Hedges' g = −0.89 (95% CI −1.41 to −0.37), p < 0.001] (Fig. 2). Similarly, a worse neuropsychological performance in working memory was found in individuals with SZ and with OCs [Hedges' g = −1.47 (95% CI −2.89 to −0.06), p = 0.01].
Fig. 2. Forest Plot of Cognitive Differences Between Patients With Schizophrenia With and Without Obstetric complications. Subj, Subjects; *p < 0.05: *** p < 0.001.
In relation to random-effect models, there were no significant differences in executive function [Hedges' g = 0.10 (95% CI −0.19 to 0.39)], verbal fluency [Hedges' g = −0.32 (95% CI −0.79 to −0.15)], intellectual functioning [Hedges' g = −0.45 (95% CI −1.22 to 0.31)], processing speed [Hedges' g = 0.19 (95% CI −0.45 to 0.83)], and attention (Hedges' g = 0.24 (95% CI −0.17 to 0.64)) between subjects with a history of OCs and without OCs.
We detected high heterogeneity (I 2 > 70%) in two of the evaluated cognitive domains: working memory [X 2(2) = 10.74, p < 0.01; I 2 = 81% (95% CI 42%–94%)], and intellectual functioning [X 2(2) = 10.21, p < 0.01; I 2 = 80% (95% CI 38%–94%)]. Nevertheless, other neuropsychological functions showed moderate (I 2 > 50% and I 2 < 70%) or low (I 2 < 50%) heterogeneity as executive function [X 2(4) = 8.69, p = 0.07; I 2 = 54% (95% CI 0%–83%)], processing speed [X 2(2) = 4.23, p = 0.12; I 2 = 53% (95% CI 0%–86%)], verbal memory [X 2(1) = 0.07, p = 0.79; I 2 = 0%], verbal fluency [X 2(2) = 0.88, p = 0.64; I 2 = 0% (95% CI 0%–90%)], and attention [X 2(3) = 4.92, p = 0.18; I 2 = 39% (95% CI 0%–79%)]. There was no evidence of publication biases as indicated by visual inspection (online Supplementary Material 4a) and by the Egger test for small study effects (online Supplementary Material4b).
Discussion
The main findings of the present meta-analysis indicate that patients with SZ and OCs have cognitive deficits in working memory and verbal memory, in comparison to those SZ patients nonexposed to OCs and besides the few number of compared articles. To the best of our knowledge, this is the first systematic review and meta-analysis of the association between OCs and cognitive function in individuals with SZ and other psychotic disorders.
Although cognitive impairments across several domains have been reported in patients with SZ, our findings suggest that OCs were related to working and verbal memory. These cognitive domains have been considered a core cognitive deficit in SZ and have been proposed as endophenotypic markers (Park & Gooding, Reference Park and Gooding2014) that are present at early phases of the illness, in ultra-high risk subjects and first episode psychosis patients (Brown et al., Reference Brown, Vinogradov, Kremen, Poole, Bao, Kern and McKeague2011; Sheffield, Karcher, & Barch, Reference Sheffield, Karcher and Barch2018). Indeed, verbal memory deficits may be an important risk marker for the development of schizophrenia-spectrum psychotic disorders (Lencz et al., Reference Lencz, Smith, Mclaughlin, Auther, Nakayama, Hovey and Cornblatt2006). Interestingly, cognitive difficulties have a great implication on the functionality of subjects with SZ while the literature suggests that cognitive dysfunction is associated with prominent functional impairment, which involves social, occupational, and independent living activities (Van Winkel et al., Reference Van Winkel, Myin-Germeys, De Hert, Delespaul, Peuskens and Van Os2007). Verbal memory and working memory impairments have been also described as an important predictor of poor psychosocial functioning and everyday life (Faerden et al., Reference Faerden, Barrett, Nesvåg, Friis, Finset, Marder and Melle2013; Fett et al., Reference Fett, Viechtbauer, de Dominguez, Penn, van Os and Krabbendam2011; Green, Kern, Braff, & Mintz, Reference Green, Kern, Braff and Mintz2000; Hubacher et al., Reference Hubacher, Weiland, Calabrese, Stoppe, Stöcklin, Fischer-Barnicol and Penner2013; Puig et al., Reference Puig, Penadés, Gastó, Catalán, Torres and Salamero2008; Tolman & Kurtz, Reference Tolman and Kurtz2012).
Unlike our results, four of the six included studies (Borkowska & Rybakowski, Reference Borkowska and Rybakowski2002; Brown et al., Reference Brown, Vinogradov, Kremen, Poole, Deicken, Penner and Schaefer2009; Ochoa et al., Reference Ochoa, Huerta-Ramos, Barajas, Iniesta, Dolz, Baños and Usall2013; Torniainen et al., Reference Torniainen, Wegelius, Tuulio-Henriksson, Lonnqvist, Suvisaari, Lönnqvist and Suvisaari2013) found an association between OCs and impaired executive function. Nevertheless, some aspects must be considered: first of all, the breadth of the concept of executive functions, and therefore, the large number of different assessment tools and their heterogeneity. Thus, while we have centered our analysis on inhibitory control (Stroop) and cognitive flexibility (WCST), there are others aspects of the executive functioning domain that remain unexplored (i.e. reasoning, problem solving, planning) (Diamond, Reference Diamond2013). Therefore, it might be more suitable to discuss techniques that assess different components of executive functions rather than functions per se. In consonance with our findings, none of the incorporated studies found significant results related to OCs and verbal fluency, neither phonological nor semantic. Interestingly, when evaluating figural fluency, it was significantly reduced in patients exposed to OCs compared to unexposed case subjects (Brown et al., Reference Brown, Vinogradov, Kremen, Poole, Bao, Kern and McKeague2011, Reference Brown, Vinogradov, Kremen, Poole, Deicken, Penner and Schaefer2009).
As regards to verbal memory, studies with both first and second episode of the psychosis population found no differences between subjects and controls (Holthausen et al., Reference Holthausen, Wiersma, Sitskoorn, Hijman, Dingemans, Schene and van den Bosch2002; Ochoa et al., Reference Ochoa, Huerta-Ramos, Barajas, Iniesta, Dolz, Baños and Usall2013), while in chronic SZ patients opposite results were found (Brown et al., Reference Brown, Vinogradov, Kremen, Poole, Bao, Kern and McKeague2011). A plausible explanation could be that, while we had focused our analysis in immediate memory, different results could have been obtained assessing the learning curve or delayed memory.
Opposite results have been found for processing speed: while some studies did not find significant results (Brown et al., Reference Brown, Vinogradov, Kremen, Poole, Deicken, Penner and Schaefer2009; Ellman et al., Reference Ellman, Yolken, Buka, Torrey and Cannon2009; Holthausen et al., Reference Holthausen, Wiersma, Sitskoorn, Hijman, Dingemans, Schene and van den Bosch2002), others described that OCs were related to slower performances (Ochoa et al., Reference Ochoa, Huerta-Ramos, Barajas, Iniesta, Dolz, Baños and Usall2013; Torniainen et al., Reference Torniainen, Wegelius, Tuulio-Henriksson, Lonnqvist, Suvisaari, Lönnqvist and Suvisaari2013). One possible explanation might be that certain factors, such as antipsychotic medication, relapse, severity of illness, among others, could influence processing speed (Knowles, David, & Reichenberg, Reference Knowles, David and Reichenberg2010). For that reason, studies in naïve FEP patients in which the effect of some confounding factors such as chronicity or the influence of antipsychotic treatment are avoided, can be especially relevant. In a study on 62 FEP patients (both, early onset and adult onset) that evaluated OCs and were grouped according to their neurodevelopment contribution, patients with a higher prevalence of early environmental events presented a worse processing speed assessment evaluated with the TMT-A (Ochoa et al., Reference Ochoa, Huerta-Ramos, Barajas, Iniesta, Dolz, Baños and Usall2013). Moreover, the difference in sample size could interfere in the results as well as how OCs were assessed. For example, Brown et al., assessed OCs by exposure to maternal infection (Brown et al., Reference Brown, Vinogradov, Kremen, Poole, Bao, Kern and McKeague2011, Reference Brown, Vinogradov, Kremen, Poole, Deicken, Penner and Schaefer2009), while others by birth weight or Lewis and Murray scale (Ochoa et al., Reference Ochoa, Huerta-Ramos, Barajas, Iniesta, Dolz, Baños and Usall2013).
In regard to attention, our results did not find significant differences between patients with and without history of OCs. Previous literature has found conflicting results, some of them described worse performance on attention in patients with a history of an OC (Ochoa et al., Reference Ochoa, Huerta-Ramos, Barajas, Iniesta, Dolz, Baños and Usall2013; Torniainen et al., Reference Torniainen, Wegelius, Tuulio-Henriksson, Lonnqvist, Suvisaari, Lönnqvist and Suvisaari2013) while others have identified similar performance between groups (Brown et al., Reference Brown, Vinogradov, Kremen, Poole, Deicken, Penner and Schaefer2009). Even so, the complexity of the domain (viz., focused, selective, divided, alternating attention) as well as the heterogeneity of the tests might explain this result (Coubard, Reference Coubard2015).
Taking into account the existing published meta-analyses that provide support for the presence of OCs as a risk factor for SZ later in life (Cannon et al., Reference Cannon, Jones and Murray2002; Davies et al., Reference Davies, Segre, Estradé, Radua, De Micheli, Provenzani and Fusar-Poli2020; Radua et al., Reference Radua, Ramella-Cravaro, Ioannidis, Reichenberg, Phiphopthatsanee, Amir and Fusar-Poli2018) and that our results suggest that OCs have been associated with verbal memory and working memory performance in patients with SZ, an exhaustive assessment of OCs would be recommended to tailor specific early intervention strategies. This point is relevant, because verbal memory has been extensively demonstrated as the best predictor of psychosocial functioning, so alterations in this domain are indicative of inadequate or very poor functioning in the community (Penadés & Gastó, Reference Penadés and Gastó2010). Therefore, patients with OCs and FEP should be carefully followed up, and appropriate rehabilitation approaches could be designed.
Limitations
Several limitations should be taken into consideration when interpreting our results. These should be conceptualized according to the heterogeneity of the sample studied and the scarce number of retrieved articles. Besides, considering the small sample size of some of the manuscripts included, a main concern stems from the concept of OCs. Although in some studies OCs are grouped together (Gilvarry et al., Reference Gilvarry, Takei, Russell, Rushe, Hemsley and Murray2000; Reference Gilvarry, Russell, Jones, Sham, Hemsley and Murray2001), literature and a critic review of the scales used (Cannon et al., Reference Cannon, Jones and Murray2002) suggest at least the need to differentiate between OCs during pregnancy and delivery with different outcomes (Mezquida et al., Reference Mezquida, Fernandez-Egea, Treen, Mané, Bergé, Savulich and Garcia-Rizo2018), or even to differentiate between the intrauterine period, with difficulties during pregnancy or difficulties related to abnormal fetal growth or development (Cannon et al., Reference Cannon, Jones and Murray2002). Another important limitation which could have biased our results, is the timing of the event and the sex of the fetus (Ellman et al., Reference Ellman, Murphy, Maxwell, Calvo, Cooper, Schaefer and Brown2019), with different responses to environmental exposure (Al-Qaraghouli & Fang, Reference Al-Qaraghouli and Fang2017). The intensity and duration of the adverse event as well as the time of exposure (duration and critical period) might also drive different outcomes (Graignic-Philippe, Dayan, Chokron, Jacquet, & Tordjman, Reference Graignic-Philippe, Dayan, Chokron, Jacquet and Tordjman2014).
Although several cognitive domains are assessed in the context of the present study, some cognitive domains were not investigated at all (i.e. visual learning and memory or social cognition) or others were investigated by a smaller number of studies (i.e. just two studies included data on verbal memory).
Conclusion
Our results suggest that there is consistent evidence for a significant relationship between the presence of OCs and worse cognitive performance in two specific domains widely replicated to be altered in SZ, such as working memory and verbal memory, while a trend towards significance was observed in relation to IQ. However, no further statistical significance was described, probably due to the heterogeneity of the main risk factor studied, OCs. Nevertheless, these results confirm the effect of the perinatal period in the later development of cognitive disturbances in psychosis–present even at onset– and highlight a feasible common pathophysiological pathway (Bock, Wainstock, Braun, & Segal, Reference Bock, Wainstock, Braun and Segal2015). However, as there were a limited number of studies to compare, further studies which specifically study the characteristics of perinatal stressful events are required to fully understand its impact on the cognitive function of patients affected by psychosis.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0033291722002409.
Acknowledgements
Dr S Amoretti has been supported by a Sara Borrell contract (CD20/00177), funded by Instituto de Salud Carlos III (ISCIII) and co-funded by European Social Fund ‘Investing in your future’ and thanks the support of the Secretaria d'Universitats I Recerca del Departament d'Economia i Coneixement (2017 SGR 1365), the CERCA Programme, and the Departament de Salut de la Generalitat de Catalunya for the PERIS grant SLT006/17/00345.
Dr N Verdolini thanks the BITRECS project, which has received funding from the European Union's Horizon 2020 research and innovation program under the Marie Skłodowska-Curie grant agreement No 754550 and from ‘La Caixa’ Foundation (ID 100010434), under the agreement LCF/PR/GN18/50310006.
Dr E Vieta thanks the support of the Instituto de Salud Carlos III; the CIBER of Mental Health (CIBERSAM); the Secretaria d'Universitats i Recerca del Departament d'Economia i Coneixement (2017 SGR 1365).
Dr C Garcia-Rizo thanks the support to the Instituto de Salud Carlos III and the Spanish Ministry of Economy and Competitiveness (PI14/00753, PI20/00661) and the General Evaluation Branch and the European Regional Development Fund (FEDER).
We thank Dr Ochoa for providing the requested data for our research.
Author contribution
Dr Amoretti analyzed and collected data while writing the manuscript in close collaboration with FD Rabelo-da-Ponte. FD Rabelo-da-Ponte verified and performed the analytical calculations and with Dr Amoretti wrote to the final version of the manuscript. Dr M Garriga conceived the study design, analyzed and collected data, and contributed to the final interpretation of the findings. M Florencia Forte contributed to the final interpretation of the findings. while provided a critical review and developed the theoretical formalism. Dr R Penadés, Dr E Vieta, Dr E. Parellada, Dr J Antoni Ramos-Quiroga, Dr CS Gama and Dr N Verdolini provided critical review and developed the theoretical formalism. Dr B Bitanihirwe analyzed and collected data while providing a critical review and developed the theoretical formalism. Dr C Garcia-Rizo conceived of the presented idea, analyzed data, and supervised the progression and findings of this work. All authors discussed the results and contributed to the final manuscript.
Financial support
This research received no specific grant from any funding agency, commercial or not-for-profit sectors.
Conflict of interest
Dr M Garriga has received financial support for CME activities and travel funds from the following entities (unrelated to the present work): Ferrer, Lundbeck, Janssen.
Dr E Vieta has received grants and served as a consultant, advisor or CME speaker for the following entities (unrelated to the present work): AB-Biotics, Abbvie, Angelini, Celon, Dainippon Sumitomo Pharma, Ferrer, Gedeon Richter, GH Research. Janssen, Lundbeck, Otsuka, Sage, Sanofi-Aventis, and Takeda.
Dr E Parellada has received financial support for CME activities and travel funds from the following entities (unrelated to the present work): Janssen-Cilag, GlaxoSmithKline, Ferrer, ADAMED.
Dr J Ramos-Qurioga has received financial support for CME activities and travel funds from the following entities (unrelated to the present work): Eli-Lilly, Janssen-Cilag, Novartis, Shire, Takeda, Bial, Shionogui, Lundbeck, Almirall, Almirall, Braingaze, Sincrolab, Medice and Rubió, Raffo, Actelion, Ferrer, Oryzon, Roche, Psious.
Dr N Verdolini has received financial support for CME activities and travel funds from the following entities (unrelated to the present work): Angelini, Janssen-Cilag, Lundbeck, Otsuka.
Dr C Garcia-Rizo has received financial support for CME activities and travel funds from the following entities (unrelated to the present work): Adamed, Angelini, Janssen, and Otsuka.
The rest of the authors report no biomedical financial interests or potential conflicts of interest.
