Introduction
Individuals with schizophrenia show reduced cognitive performance compared to unaffected subjects in a wide range of domains (Fioravanti, Bianchi, & Cinti, Reference Fioravanti, Bianchi and Cinti2012), which is consistently associated with worse functional outcomes (Halverson et al., Reference Halverson, Orleans-Pobee, Merritt, Sheeran, Fett and Penn2019), regardless of age, gender, or illness chronicity (Fett et al., Reference Fett, Viechtbauer, Dominguez, Penn, van Os and Krabbendam2011). These deficits are present since the early stages of the disease, even in drug-naïve patients (Fatouros-Bergman, Cervenka, Flyckt, Edman, & Farde, Reference Fatouros-Bergman, Cervenka, Flyckt, Edman and Farde2014), and there is evidence of cognitive compromise since before diagnosis (Mollon & Reichenberg, Reference Mollon and Reichenberg2018). However, there is still controversy in the literature regarding the trajectory of cognitive change. Most studies report no further decline in cognition after the first episode (Szöke et al., Reference Szöke, Trandafir, Dupont, Méary, Schürhoff and Leboyer2008) even in never-medicated patients (Solís-Vivanco et al., Reference Solís-Vivanco, Rangel-Hassey, León-Ortiz, Mondragón-Maya, Reyes-Madrigal and de la Fuente-Sandoval2020), while others have raised the possibility of a deteriorating trajectory with aging (Fett et al., Reference Fett, Velthorst, Reichenberg, Ruggero, Callahan, Fochtmann and Kotov2020; Zanelli et al., Reference Zanelli, Mollon, Sandin, Morgan, Dazzan, Pilecka and Reichenberg2019). There is also great heterogeneity in the cognitive deficits in schizophrenia (Shmukler, Gurovich, Agius, & Zaytseva, Reference Shmukler, Gurovich, Agius and Zaytseva2015). Non-clinical studies show sex-differences in cognition; however, results for schizophrenia are inconsistent (Choleris, Galea, Sohrabji, & Frick, Reference Choleris, Galea, Sohrabji and Frick2018; Mendrek & Mancini-Marïe, Reference Mendrek and Mancini-Marïe2016). Recent literature has been trying to identify subgroups that might elucidate common mechanisms and risk factors, considering the variability within this population (Carruthers, Van Rheenen, Gurvich, Sumner, & Rossell, Reference Carruthers, Van Rheenen, Gurvich, Sumner and Rossell2019). Studies using data-driven techniques revealed subgroups of cognitively spared, intermediate cognitive impairments, and deficit subtypes (Green, Girshkin, Kremerskothen, Watkeys, & Quidé, Reference Green, Girshkin, Kremerskothen, Watkeys and Quidé2020). It is unclear, though, whether these findings result from diverse subgroups or only divisions in a linear continuum. Hence, adding evidence to this matter would have important implications for the clinical management of cognitive deficits in psychosis, in addition to broadening the understanding of the possible underlying mechanisms of psychopathology.
Schizophrenia and other psychotic disorders carry a genetic load related to cognitive performance and education attainment (Richards et al., Reference Richards, Pardiñas, Frizzati, Tansey, Lynham, Holmans and Walters2020) that is shared across different populations (Lam et al., Reference Lam, Chen, Li, Martin, Bryois, Ma and Huang2019). Nonetheless, a biocultural approach indicates that cultural patterns could influence neurobiology and inflammation in mental disorders (Shattuck, Reference Shattuck2019). Specifically, an important focus should be given to the effects of the socioeconomic status (SES), which is ‘a multidimensional construct comprising diverse socioeconomic factors’ (Braveman et al., Reference Braveman, Cubbin, Egerter, Chideya, Marchi, Metzler and Posner2005, p. 2879), commonly operationalized through education, income, and/or occupation variables (Farah, Reference Farah2017). SES is greatly predictive of several physical and mental health outcomes (Adler et al., Reference Adler, Boyce, Chesney, Cohen, Folkman, Kahn and Syme1994; Anderson & Armstead, Reference Anderson and Armstead1995), and has been linked to risk for psychosis (Kwok, Reference Kwok2014; Luo et al., Reference Luo, Zhang, He, Pang, Guo and Zheng2019). In non-clinical samples, SES is associated with cognitive ability and explains a significant proportion of shared environmental variance in twin-studies (Hanscombe et al., Reference Hanscombe, Trzaskowski, Haworth, Davis, Dale and Plomin2012; Hart, Petrill, Deater Deckard, & Thompson, Reference Hart, Petrill, Deater Deckard and Thompson2007). In schizophrenia, SES has been shown to impact cognition (Goldberg et al., Reference Goldberg, Fruchter, Davidson, Reichenberg, Yoffe and Weiser2011). Interestingly, a study from USA found a group by parental SES interaction (based on occupation and education) in predicting executive functioning, indicating that low SES was related to worse performance in individuals with schizophrenia, but not in controls (Yeo et al., Reference Yeo, Martinez, Pommy, Ehrlich, Schulz, Ho and Calhoun2014). Curiously, only a few studies have investigated these effects in diverse settings. In this context, Latin America arises as an opportunity for studying the impact of SES on people with schizophrenia's cognitive performance, while at the same time looking into an understudied region of the world.
Latin America is a multiethnic and multicultural continent formed from a mixture of different pre-colonial indigenous cultures, European colonizers (mostly Spanish and Portuguese), and African ethnic groups brought as slaves during the colonization. It is home to almost 650 million people – around 8.45% of the world's population (World Bank, 2019), and has highly urbanized regions, with an estimated 80% of the population living in cities (United Nations, 2018). Because of its historical economic and political instability, there are over 36 million Latin American immigrants living abroad (Bayona-i-Carrasco & Avila-Tàpies, Reference Bayona-i-Carrasco and Avila-Tàpies2020), making the study of this population relevant beyond the continent's geographical limits. Latin America has significant and evident economic inequalities (World Bank, n.d.). The inequities experienced by its population go beyond income and lead to social discrimination and exclusion from exercise of rights, autonomy, and access to opportunities and education (Abramo, Cecchini, & Ullmann, Reference Abramo, Cecchini and Ullmann2020; Neidhöfer, Serrano, & Gasparini, Reference Neidhöfer, Serrano and Gasparini2018). These result in major health inequalities, both in access and outcomes (Abramo et al., Reference Abramo, Cecchini and Ullmann2020), implicating in factors such as a reduction in life expectancy at birth (Bilal et al., Reference Bilal, Alazraqui, Caiaffa, Lopez-Olmedo, Martinez-Folgar, Miranda and Diez-Roux2019). Furthermore, people in this region face challenges linked to extreme violence, such as reduced life expectancy due to homicide in young people (Canudas-Romo & Aburto, Reference Canudas-Romo and Aburto2019). Such a different setting compared to highly industrialized western societies pose a valuable opportunity for the study of cognition, providing the opportunity to re-test previously known associations while also examining pending controversies under new light and different contexts, such as the effects of SES in cognition.
Therefore, we examined the cognitive performance of 1175 Latin American individuals with schizophrenia and healthy controls from five Latin American countries. We first focused on testing whether previously reported findings in schizophrenia and cognition, such as a global cognitive deficit, were replicated in such a different environment. We then sought to contribute to the discussions on the cognitive trajectory in schizophrenia by examining a possible differential association of age and cognition in our cross-sectional study. Additionally, we aimed to analyze how SES relates to cognitive performance in people with schizophrenia and healthy controls. Finally, we also explored whether patients clustered in different subgroups according to cognition.
Methods
Participants
This study is part of the ANDES network, which unites 15 groups from different countries across Latin America to promote science (Crossley et al., Reference Crossley, Guinjoan, Rivera, Jackowski, Gadelha, Elkis and Bressan2019). We included 1175 participants from five countries: Argentina, Brazil, Chile, Colombia, and Mexico, of which 864 were individuals diagnosed with schizophrenia or first-episode non-affective psychosis, and 311 were unaffected subjects. All participants were part of individual research projects that included cognitive and clinical assessments.
For a more complete description of sample features, we included in online Supplementary Table S1. each country's demographic characteristics, including population, Gross National Income (GNI) per capita, inequality measures (Gini coefficient), and Human Development Index (HDI) rating and rank. The per capita GNI (purchasing power parity) for the countries studied ranged from 12 896 to 219 722 011 USD, and Gini coefficients ranged from 40.6 to 53.3 (income distribution inequality). In terms of HDI (overall development), Chile and Argentina are ranked as ‘Very High’, while Mexico, Brazil, and Colombia are ranked ‘High’.
All centers recruited patients with schizophrenia or an associated diagnosis from clinical centers. Some centers also included healthy controls. All patients and controls participated willingly and voluntarily, and proper consent forms were signed. Their local ethical committee approved each site's project. Table 1 describes the inclusion and exclusion criteria used for patients and controls in each center.
Table 1. Sample characteristics by site
Assessments
Participants underwent a cognitive assessment through the MATRICS Consensus Cognitive Battery (MCCB). Because of differences in study design, some individuals completed only some of the battery subtests. We considered as SES the self-reported objective variables of personal education, parental education, and income, which are widely used in the neuroscience literature (Farah, Reference Farah2017). We collected additional data related to demographic (age, sex, and occupational status) and clinical factors (age of onset, illness duration, number of hospitalizations). Symptomatology in patients was assessed through the Positive and Negative Syndrome Scale (PANSS).
Data analysis
Statistical analyses were completed in R (version 4.0.2) and RStudio (version 1.3.1093). We transformed MCCB subtest raw scores into z-scores using the mean and standard deviation of the unaffected individuals. We used a reversed score of the Trail Making Test to maintain the direction of the other subtests. Then, we created a cognitive composite with the sum of z-scores divided by the number of subtests completed. Our first level of analysis was to compare patients and controls regarding demographic, socioeconomic, and cognitive data using independent samples t test and chi-square test when appropriate. We then investigated the relationship between cognitive performance and demographic, socioeconomic, and clinical data through linear regression models. Normality assumption was checked through visual inspections of histograms, qq-plots, and values of skew and kurtosis of the relevant variables. Skewed variables were log-transformed. Linear mixed-effects models with the site as a random effect were performed to confirm that linear models' findings were not due to site differences. Finally, we explored cognitive performance subgroups through hierarchical cluster analysis with the squared Euclidian distance and Ward linkage as the agglomeration procedure. The dendrogram's inspection was used as a criterion to establish the appropriate number of clusters to retain (Lima et al., Reference Lima, Rabelo-da-Ponte, Bücker, Czepielewski, Hasse-Sousa, Telesca and Rosa2019; Rabelo-Da-Ponte et al., Reference Rabelo-Da-Ponte, Lima, Martinez-Aran, Kapczinski, Vieta, Rosa and Czepielewski2020). We then compared the data-driven subgroups of cognitive performance regarding demographic, socioeconomic, and clinical variables using independent samples t test and χ2 test. Descriptive data were expressed as mean and standard deviation, and significance was set at p < 0.05, two-tailed.
Results
Cognitive performance, clinical and demographic factors
Table 2 presents the demographic, socioeconomic, clinical, and cognitive data. Patients had similar age than unaffected participants. The patient's group included more males and had less working or studying individuals. Patients also had fewer years of education, parents' years of education, and family income than controls. Regarding cognition, as expected, we found a worse performance of patients compared to controls in the cognitive composite (Fig. 1a; t(1128) = 18.588, p < 0.001, Cohen's d = 1.281), and in each subtest individually (Table 2., Fig. 1b).
Fig. 1. (a) Comparison of the MCCB cognitive composite (z-score) between healthy controls (HC) and individuals with schizophrenia (SZ). (b) Comparison of each subtest of the MCCB (z-score) between healthy controls (HC) and individuals with schizophrenia (SZ).
Table 2. Demographic, socioeconomic, clinical, and cognitive data
In patients, in linear models controlling for age, sex, and years of education, age of onset (F(4541) = 53.51, p < 0.001, t = 0.861, p = 0.390, β = 0.04) and illness duration (F(4565) = 56.46, p < 0.001, t = −1.645, p = 0.101, β = −0.09) were not related to the overall cognitive performance (online Supplementary Figure 1A–1B). Better cognitive performance was related to lower PANSS total score (F(4474) = 44.92, p < 0.001, t = −5.916, p < 0.001, β = −0.24) and its subscales (Positive: F(4475) = 42.67, p < 0.001, t = −5.275, p < 0.001, β = −0.21); Negative: F(4475) = 48.07, p < 0.001, t = −6.683, p < 0.001, β = −0.27; General: F(4475) = 39.57, p < 0.001, t = −4.263, p < 0.001, β = −0.17) (online Supplementary Figures S1C–S1F).
Effects of age and gender in patients and controls
We performed a linear regression model including the cognitive composite as the dependent variable, age, gender, and group as the independent predictors, and the interaction between age × group and gender × group (F(51,049) = 118.3, p < 0.001, Adj.R 2 = .358). We found that age (t = −7.430, p < 0.001, β = −0.34), gender (t = −2.200, p = 0.028, β = −0.10), and group (t = −8.464, p < 0.001, β = −0.63) were all independent predictors of cognitive performance, with no group by age (t = 0.950, p = 0.342, β = 0.08) or group by gender (t = 0.859, p = 0.391, β = 0.04) interactions. This indicated that aging was associated with a similar global cognitive decline in both patients and controls (Fig. 2a), and that male participants performed slightly better than females.
Fig. 2. Relationship between demographic and socioeconomic factors and MCCB cognitive composite (z-score) in healthy controls (HC) and individuals with schizophrenia (SZ), controlling for age and gender.
Exploring cognition in a relatively deprived setting: association with SES indicators of education, parental education, and income
We found a group by education interaction (t = 5.333, p < 0.001, β = 0.49) in a model with cognitive composite as the dependent variable, and the interaction between years of education and group as predictors, controlling for age and gender (F(5,817) = 177.1, p < 0.001, Adj.R 2 = 0.517), indicating that education was more important for the cognitive performance of patients than for controls (Fig. 2b). The same was found for parents' years of education (F(5,444) = 112.1, p < 0.001, Adj.R 2 = 0.553), where the interaction suggested a more significant effect for the patient's parents education (t = 2.839, p = 0.0047, β = 0.26), although in a less pronounced way than the patient's personal education (Fig. 2c).
Finally, a model with cognitive composite as the dependent variable, and income by group as predictors, controlling for age and gender (F(5,239) = 66.85, p < 0.001, Adj.R 2 = 0.574) found a group by income interaction (t = 4.471, p < 0.001, β = 0.25), indicating that patients with more income scored higher in cognitive performances, while income was not associated with cognitive performance in controls (Fig. 2d).
Investigating subgroups of cognitive performance
We conducted a hierarchical cluster analysis with only the patients with complete data considering the subtests z-scores of the MCCB. We identified two subgroups of cognitive performance: the first group had most patients (n = 388, 71.59%) and had lower middle performances than mean scores of healthy controls (z-scores between −1.05 and −0.19). The second subgroup (n = 154, 28.41%) showed performances considered as clinical deficits in all domains (z-scores below −1.5) (Fig. 3).
Fig. 3. Mean cognitive performance in each subtest z-scores between subgroups of individuals with schizophrenia (SZ) and healthy controls (HC).
Patients from the first subgroup were younger, had more personal and parental education, and had fewer years of illness, psychiatric hospitalizations, and symptoms (lower PANSS scores) than the second subgroup. Additionally, the first group had an increased estimated IQ in relation to the other group. There were no differences regarding age at onset and family income (Supplementary Table S2).
Effects of site
Linear mixed-effects models with the site as a random factor were performed to ascertain that all results previously mentioned were not due to site differences. No differences were found between models' results.
Discussion
This is the first large and representative study to characterize the cognitive deficits of schizophrenia in a Latin American population. Using this sample from an under-reported region of the world, we were able to confirm certain associations frequently found in the literature, as well as shed light on new controversies. We found that patients had worse cognitive performance than healthy controls, which was generalized across all cognitive domains. Age was an independent predictor of cognitive performance, and we did not find any evidence suggesting that this association was different in patients and controls. When we looked at the SES variables, we found that a deprived environment was related to worse cognitive impairments mostly in patients. Personal education, parental education, and income were significantly related to both groups' cognitive performance, but higher SES variables were associated with larger cognitive function increases in schizophrenia. These results might indicate a vulnerability of individuals with psychosis that could prompt patients to be more impacted by chronic exposure to social factors, as we can observe in poor and developing countries such as in Latin America. Finally, not all patients showed severe deficits, and the gravity of impairments was related to demographic, socioeconomic, and clinical variables. These findings are discussed below.
First, as expected, patients performed worse than healthy controls in all subtests of the MCCB and the cognitive composite. Cognition was also related to symptomatology, particularly negative symptoms. These results were expected since these findings have been widely reported in different regions of the world and remained robust over the decades (Schaefer, Giangrande, Weinberger, & Dickinson, Reference Schaefer, Giangrande, Weinberger and Dickinson2013). Nonetheless, there were no studies with representative data from Latin America reporting this outcome, which is why these results are important for a broader understanding of a diverse world. Further, as prior reports, we found high heterogeneity in the cognitive performance of individuals. Our data-driven clustering indicated that around 70% of patients presented performances considered as lower-middle compared to unaffected individuals. As indicated in the review by Green et al. (Reference Green, Girshkin, Kremerskothen, Watkeys and Quidé2020), a two-cluster solution has been found in previous studies. Our results suggest a smaller group for severe deficits (28.41% as opposed to 50% in other studies). A severe subtype appears to be common to all investigations of clusters, and several studies report differences in clinical and socioeconomic findings among groups (Green et al., Reference Green, Girshkin, Kremerskothen, Watkeys and Quidé2020), which supports our results. Moreover, the subgroups of cognitive performance seem to be present in related diagnoses such as bipolar disorder, indicating that these cognitive subgroups might not imply distinct profiles but possibly different stages of the same cognitive trajectory (Karantonis et al., Reference Karantonis, Rossell, Carruthers, Sumner, Hughes, Green and Van Rheenen2020).
Second, we observed in our sample that the SES variables of education, parental education, and income were all related to worse cognitive impairments mostly in patients, indicating that lower SES was associated to worse cognitive performance in schizophrenia, which is supported by the literature. A sample of individuals with schizophrenia from Australia (Wells et al., Reference Wells, Jacomb, Swaminathan, Sundram, Weinberg, Bruggemann and Weickert2020) divided into preserved, deteriorated, and compromised groups based on estimated premorbid IQ and current cognitive performance showed a SES difference. Additionally, the compromised patients (e.g. individuals with a significant decline from estimated premorbid IQ) showed greater childhood adversities and lower SES than the deteriorated patients (e.g. those with both current and estimated premorbid impairments) (Wells et al., Reference Wells, Jacomb, Swaminathan, Sundram, Weinberg, Bruggemann and Weickert2020), suggesting that social factors might impact both cognitive development and exposure to childhood adversity. Interestingly, there seems to be a difference between the observed and the expected global cognitive ability of patients, and even ‘normal’ performers (as indicated by normative data) might be impaired compared to their expected abilities (Hochberger et al., Reference Hochberger, Thomas, Joshi, Swerdlow, Braff, Gur and Light2020). This might suggest that, in addition to biological neurodevelopmental abnormalities (Czepielewski, Wang, Gama, & Barch, Reference Czepielewski, Wang, Gama and Barch2017), individuals at risk of severe mental disorders may also be more vulnerable to social factors during development. A study of the Philadelphia Neurodevelopmental Cohort (USA) (Gur et al., Reference Gur, Moore, Rosen, Barzilay, Roalf, Calkins and Gur2019) found that lower SES was related to both reduced performances in different cognitive domains and lower volume across brain regions, including white and gray matter, that was related to an accelerated neurodevelopment. In patients from Latin America (Crossley et al., Reference Crossley, Zugman, Reyes-Madrigal, Czepielewski, Castro, Diaz-Zuluaga and Bressan2021), we recently showed that income was related to total gray matter volume in unaffected individuals but not in psychosis patients. This potentially indicated that less brain vulnerability in patients (e.g. less gray matter loss) would be sufficient to become unwell in adverse environments, and that considering the upbringing of patients is critical to understanding schizophrenia's anatomy (Crossley et al., Reference Crossley, Zugman, Reyes-Madrigal, Czepielewski, Castro, Diaz-Zuluaga and Bressan2021). Although this may seem contradictory to the present report, we believe that these two Latin American findings indicate that environmental variables might be key to explaining disease outcomes. The difference between patients and controls in the impact received by deprived environments was seen in our sample related to cognitive performances, specially where less income and parental education were more related to worse impairments in cognition for patients. While the unexpected findings for unaffected individuals might be different from other studies in the literature, a diagnosis by SES interaction has been described related to executive functions in a sample from the USA (Yeo et al., Reference Yeo, Martinez, Pommy, Ehrlich, Schulz, Ho and Calhoun2014). Further, we highlight that these relationships have not been extensively studied in non-WEIRD populations.
Third, we found that patients did not show a steeper cognitive decline in the cognitive composite than controls. Only longitudinal studies following participants through aging can actually confirm this association. Nonetheless, although our data is cross-sectional, we have a wide range of ages in our large sample (13–66 years), which might bring an idea of a longitudinal profile. The cognitive aging in schizophrenia and related disorders is still unestablished (Czepielewski et al., Reference Czepielewski, Massuda, Panizzutti, Grun, Barbé-Tuana, Teixeira and Gama2018). Some evidence suggests early deficits with stable trajectories after the first episode (Sheffield, Karcher, & Barch, Reference Sheffield, Karcher and Barch2018). A longitudinal study by the Genetic Risk and Outcome of Psychosis (GROUP, Netherlands and Belgium) (Islam et al., Reference Islam, Habtewold, van Es, Quee, van den Heuvel, Alizadeh and Bruggeman2018) found five cognitive trajectories that remained stable after 3 and 6 years in individuals with SZ. However, others point to accelerated cognitive aging. The Suffolk County Mental Health Project (USA) (Fett et al., Reference Fett, Velthorst, Reichenberg, Ruggero, Callahan, Fochtmann and Kotov2020) showed that patients with a first psychotic hospitalization, after 18 years, presented declines in some cognitive domains that were clinically significant and larger than expected due to normal aging. This was similar to the Aetiology and Ethnicity in Schizophrenia and Other Psychoses study (AESOP, UK) (Zanelli et al., Reference Zanelli, Mollon, Sandin, Morgan, Dazzan, Pilecka and Reichenberg2019) that followed first-episode psychosis after 10 years and found a cognitive decline in specific domains after illness onset in patients with schizophrenia.
One possible argument for these inconsistencies may be related to the evidence that suggests that cognitive reserve might protect from accelerated cognitive aging declines (Van Rheenen et al., Reference Van Rheenen, Cropley, Fagerlund, Wannan, Bruggemann, Lenroot and Pantelis2020). Cognitive reserve refers to the brain's ability to reorganize itself to better cope with neuropathology, and is usually measured based on socio-behavioral proxies (e.g. IQ, education, and social activities) or other approaches (e.g. neuroimaging) (Stern, Reference Stern2002). In our sample, we found that education was more important for cognitive performance in patients than in controls, supporting this hypothesis. Years involved in formal learning is one of the main components of cognitive reserve. Individuals with increased cognitive reserve might show better functional and cognitive performances, in addition to better clinical outcomes (Herrero et al., Reference Herrero, Contador, Stern, Fernández-Calvo, Sánchez and Ramos2020). However, this process is greatly influenced by the fact that individuals who later develop schizophrenia show difficulties in acquisition of cognitive abilities during childhood and adolescence (Bora, Reference Bora2015), which might also explain the shorter years of study compared to unaffected individuals. The interaction between neurodevelopmental and neuroprogressive processes could elucidate the heterogeneity and distinct cognitive trajectories of the disorder (Reckziegel et al., Reference Reckziegel, Czepielewski, Hasse-Sousa, Martins, de Britto, de Lapa and Gama2021).
Fourth, our results revealed that increased positive, negative, and general psychopathology symptoms were all associated with worse cognitive performance, even after controlling for the effects of age and education. This might indicate that patients with increased psychopathology have worse disease trajectories associated with poorer cognitive performances (Islam et al., Reference Islam, Habtewold, van Es, Quee, van den Heuvel, Alizadeh and Bruggeman2018) since the overall symptomatology tend to persist over time (Haro et al., Reference Haro, Altamura, Corral, Elkis, Evans, Krebs and Nordstroem2018). A previous meta-analysis of cross-sectional studies (Ventura, Hellemann, Thames, Koellner, & Nuechterlein, Reference Ventura, Hellemann, Thames, Koellner and Nuechterlein2009) indicated that several cognitive domains were strongly related to negative symptoms, but not to positive symptoms, which was different to what we found in our study. However, the two different dimensions within positive symptoms have been shown to be differently related to cognitive deficits in schizophrenia. Specifically, disorganization symptoms have a moderate effect, while reality distortion has a weak effect in relation to cognition (Ventura, Thames, Wood, Guzik, & Hellemann, Reference Ventura, Thames, Wood, Guzik and Hellemann2010). Thus, the variability in symptom presentation might partially explain our findings regarding positive symptoms. Interestingly, these cognitive trajectories do not seem to be related to the age of onset (Islam et al., Reference Islam, Habtewold, van Es, Quee, van den Heuvel, Alizadeh and Bruggeman2018), as seen in our sample. Therefore, the relationship between clinical variables and cognition was partially supported by previous literature.
Our study had some limitations. We presented cross-sectional data from several sites with different inclusion criteria and patient profiles. However, we used a mixed-model approach to manage inter-site differences. We also did not control for possible confounders, such as duration of untreated psychosis, medication use, and other lifestyle factors. Moreover, we did not present T-scores from the MCCB domains. It should be noted that there are different findings regarding the interpretation of cognitive data when comparing different cultures with the same normative scores, even in well-developed and high-income countries (HIC) (Raudeberg, Iverson, & Hammar, Reference Raudeberg, Iverson and Hammar2019). Therefore, comparing patients with unaffected individuals from the same cultural and socioeconomic background is preferred – as conducted in our study.
There are different findings in the literature between low- and middle-income countries (LMIC) and HIC. Individuals from LMIC report higher stress sensitivity and prevalence of psychotic experiences (DeVylder et al., Reference DeVylder, Koyanagi, Unick, Oh, Nam and Stickley2016). Urbanicity does not seem to be related to increased risk for psychosis in developing countries, different from what has been described regarding HIC (DeVylder et al., Reference DeVylder, Kelleher, Lalane, Oh, Link and Koyanagi2018). Nonetheless, similarities have also been reported between LMIC and HIC findings, such as the generalized cognitive impairments in drug-naïve individuals with psychosis (Yang et al., Reference Yang, Ruiz, Mandavia, Grivel, Wong, Phillips and Stone2020), and the relationship between a more prolonged duration of untreated psychosis and poorer clinical improvement and increased functional disability (Farooq, Large, Nielssen, & Waheed, Reference Farooq, Large, Nielssen and Waheed2009). However, even though several Latin American countries focused on expanding of universal health coverage to reduce health access inequalities during the last decades (Atun et al., Reference Atun, de Andrade, Almeida, Cotlear, Dmytraczenko, Frenz and Wagstaff2015), there is still a significant concern regarding the treatment gap for mental disorders in LMIC (Kohn et al., Reference Kohn, Ali, Puac-Polanco, Figueroa, López-Soto, Morgan and Vicente2018), which might prevent better outcomes for these patients.
LMIC publish fewer scientific reports related to mental disorders compared to HIC, although the number has increased in the last decades (Large, Nielssen, Farooq, & Glozier, Reference Large, Nielssen, Farooq and Glozier2010). Those that exist are usually underpowered, mainly because of a lack of government research funding and research capacity. Therefore, the scientific community could greatly benefit from the ‘consolidation of more regional, interdisciplinary and international research networks’ (Forero, Trujillo, González-Giraldo, & Barreto, Reference Forero, Trujillo, González-Giraldo and Barreto2020), since ‘people working and living in LMICs are better placed to define issues of importance to their populations than are people living thousands of miles away in HICs – who often fund research based on their own interests’ (Beran et al., Reference Beran, Byass, Gbakima, Kahn, Sankoh, Tollman and Davies2017). Within this perspective, our study aimed to bring – from a collective effort – a contribution to the development of cognitive research in schizophrenia and related disorders in Latin American and other LMIC. Further research is needed combined with more practical measures from the scientific community, such as funding focused on LMIC and promoting the Schizophrenia International Research Society (SIRS) membership to represent the world's diversity (Gooding, Park, Dias, Goghari, & Chan, Reference Gooding, Park, Dias, Goghari and Chan2020).
In conclusion, this is the first large-scale study describing the cognitive characteristics of individuals with schizophrenia from Latin America and the possible impacts of SES on cognitive outcomes. As expected, patients showed general and heterogeneous cognitive deficits compared to unaffected individuals. Patients did not show evidence of accelerated cognitive aging; however, we found that they were most impacted compared to controls by a deprived environment as measured by SES variables. These findings indicate the need for public policy to protect children and youth from the effect of social adversities, especially for those at risk or experiencing schizophrenia or other severe mental disorders. The ANDES Network brings a unique chance to study psychosis in disadvantaged settings, which are frequently less represented in research publications. Future studies from the ANDES Network will further explore the effects of the environment on cognition to understand better the mechanisms involved in this crucial dysfunction.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0033291721002403.
Acknowledgements
The authors thank all the patients and their families for inspiration, support, and participation. Alfonso González-Valderrama, Carmen Paz Castañeda and Rubén Nachar thank all senior leadership, professionals, and staff from Instituto Psiquiátrico José Horwitz and Universidad Finis Terrae who were involved in this project. Rodolfo Solís-Vivanco, Francisco Reyes-Madrigal and Camilo de la Fuente-Sandoval thank Felipe Rangel-Hassey and Pablo León-Ortiz (Instituto Nacional de Neurología y Neurocirugía), and Alejandra Mondragón-Maya (Universidad Nacional Autónoma de México) for their help in data acquisition, and Ricardo Mora-Durán (Hospital Fray Bernardino Álvarez) for his help in recruitment.
Author contributions
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008.
Financial support
This project was supported by: ARGENTINA – Agencia de Promoción (PICT-2014-0633); BRAZIL – Fundação de Amparo à Pesquisa do Estado de São Paulo (Fapesp 2010/10788-6); CHILE – Agencia Nacional de Investigación y Desarrollo (ANID) Anillo ACT1414 and ACT192064, and FONDECYT 1180358 and 1200601; XII Concurso Nacional de Proyectos de Investigación y Desarrollo en Salud, FONIS 2015. Proyecto SA15I20058, Ministerio de Economia, Chile, Grant ICM P09-015F (Biomedical Neuroscience Institute); ANID FONDECYT 11150846, ANID - Millennium Science Initiative, grant ‘Millennium Nucleus to Improve the Mental Health of Adolescents and Youths, Imhay,’ and grant ‘Millennium Institute for Research in Depression and Personality, Midap’; FONDECYT Postdoctoral 3190790; COLOMBIA – PRISMA UNION TEMPORAL (UNIVERSIDAD DE ANTIOQUIA / HOSPITAL SAN VICENTE FUNDACIÓN), Colciencias-INVITACIÓN 990 de 3 de agosto de 2017, Codigo 99059634; MEXICO – Instituto Nacional de Neurología y Neurocirugía: Consejo Nacional de Ciencia y Tecnología, Mexico, (CONACyT) Grants No. 261987 and 261895, National Institutes of Health Grant No. R21 MH102374, CONACyT's Sistema Nacional de Investigadores.
Conflicts of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
