Education did not interact with major depression on performance of memory tests in acute southern Brazilian in patients

The relationship of cognitive function to depression in older adults has become a topic of extensive clinical interest and research. Objective To analyze association between cognitive/memory performance,Major Depression, and education in 206 inpatients from the Psychiatry and Internal Medicine Departments. Methods Patients were evaluated by the Mini Mental State Examination, a battery of memory tests, and the Montgomery-Åsberg Depression Rating Scale. Depression patients comprised 45 severe and 42 mild/moderate, according to the Montgomery-Asberg scale. The effect of psychoactive drugs was recorded (30% drug-free). Education was measured in years. Cognitive/memory tests assessed five domains: general mental functioning, attention, sustained attention/working memory, learning memory (verbal), and remote memory. An index for memory impairment was created (positivity: 50% of tests below cutoff). Results The chief effect on worse performance was Major Depression for the domains (age and education adjusted) of attention, learning, remote memory, and general functioning. For the domain “sustained attention and working memory”, only severely depressed patients differed from the medical controls (p=.008). Education showed an independent effect on test performances. No interaction between depression and educational status was observed.We also observed an independent effect of psychoactive drugs on some cognitive/memory domains. Logistic Regression showed Major Depression as the main risk for cognitive impairment. Conclusions These data demonstrated association of Major Depression with impaired cognitive performance independent of educational attainment or psychiatric medications.

A revival of interest in testing patients with depression on a wide range of neuropsychological tasks has occurred in the last decade, provoking a growing awareness that, akin to other psychiatric and neurologic disorders, mood disorders may be associated with a distinctive pattern of cognitive impairment 16 . However, these impairments are seldom quantified. An attempt to establish a profile of neuropsychological deficits for clinically depressed patients was carried out by means of a metaanalysis published in 1997 17 . This meta-analysis analyzed investigations published between 1975 and 1996 and took into consideration several methodological criteria. The findings suggested a diffuse impairment of brain function. A more recent review targeted the role of the dorsal and ventral aspects of the prefrontal cortex and interactions between affect, motivation and cognitive function in depression 16 .
Among demographic variables, measures of impact of various cultural aspects are complex, especially in subsets of different cultures within the same population. Education can be considered as an element of culture 18 and includes literacy and schooling.
Formal education is the most significant element in culture, and both have significant effects on cognition 19 . Education has an important influence on cognitive test performance, whereby groups with higher levels of education perform better on most neuropsychological tests [19][20][21][22][23][24][25] . An implication of this influence is the need for careful evaluation of any psychometric or psychological test or scale in subsets of a population.
We hypothesized that educational attainment would be an interaction factor for depression to significantly affect cognition. The main goal of the present study was the analysis of performance in cognitive tests in currently depressed patients, comparing this with medical inpatients, and to evaluate the impact of educational attainment, age and gender.

Methods
The study was carried out using a cross-sectional design. We selected patients admitted to the Psychiatric Unit, during the first 48 hours after admission, who met DSM-IV criteria for a current major depressive episode (major depressive disorder). At the same time, patients admitted to the Internal Medicine Unit were evaluated for the study (comprising the comparison group). Inclusion criteria for these patients were presence of acute illnesses without global systemic disturbances and being highly functional before hospital admission, whereas exclusion criteria were presence of any psychiatric or neurologic disease and use of psychoactive drugs. The WHO Self-Report Questionnaire -SRQ 26,27 screened mental disorders among these patients. Eight positive questions was the cutoff for mental disorder among women, and seven among men 27 . Controls also did not meet criteria for Major Depression (DSM IV).
Psychoactive drugs for Major Depression patients administered during the last month were classified into four categories: none, antidepressants (mostly selective serotonin reuptake inhibitors), and antidepressants with other psychiatric drug (benzodiazepine, lithium, and neuroleptics). Use of benzodiazepines within 6 hours before interview was also an exclusion criteria. Severely depressed patients were distributed according to categories of drug use as 34% (N=15) none, 38% (N=17) antidepressants, and 29% (N=13) antidepressants with other psychiatric drug. The mild/moderate patients were 35% (N=15) none, 36% (N=15) antidepressants, and 26% (N=11) antidepressants with other psychiatric medication. There was no significant statistical difference between the two groups (chi-square=0.347; p=0.963).
All participants were assessed by the Montgomery-Äsberg Depression Rating Scale 28,29 . Educational attainment was given in years. The neuropsychological battery included tests that assessed five general domains: general mental functioning, attention, sustained attention and working memory, learning memory (verbal), and remote memory. General mental functioning was measured with the Mini-Mental State 30,31 . Attention was assessed with the word span 31,32 , while sustained attention and working memory with the both digit span and immediate recall of the Wechsler's Logical memory test 32 . Learning was measured by the delayed retrieval of the word list and Logical memory 32 . Remote memory was assessed with the Major Public Events, Famous Faces and Autobiographic data tests 33,34 .
We developed an index for the evaluation of cognitive impairment through an epidemiological strategy that assesses tests in parallel to enhance diagnostic power (sensitivity and specificity) 31 . For the index, we applied cutoffs to tests, and analyzed a combination of 50% of positive results as the outcome.
The sample consisted of 206 inpatients, 87 from the Psychiatry Unit and 119 from the Medical Unit. This sample size was sufficient to detect a difference of 20% (with an error of 5%) in attention test performance (OR=3 and N=65 in each group) between depressed and healthy comparison subjects 35 . Table 1 presents demographic characteristics of sample. The depression group included 65 women and 22 men, with age range from 19 to 76 years (mean ± standard deviation, 43.13±11.63) and mean education 8.12± 10.82 years (1 to 19). The Montgomery-Åsberg depression rating scale presented mean ± standard deviation, 30.24±11.85 for the forty-two patients with mild/ moderate symptoms (<30) and 45 with severe (≥30) symptoms. The medical group consisted of 71 women and 48 men, mean age 45.83±9.50 (20 to 78), years of education 7.05±3.58 (1 to 16), Montgomery-Åsberg 5.08±4.58 (mild symptoms), and the Self-Report Questionnaire 3.16±1.59.
The study was approved by the Ethics Committee for Medical Research at Hospital de Clinicas de Porto Alegre, and was conducted according to the principles established in the Helsinki declaration. Patients signed an informed consent after the nature of all procedures had been fully explained, and patient confidentiality was maintained.

Statistical analysis
Groups were first compared on demographic and clinical variables by using analyses of variance (one-way ANOVA), chi-square analyses, and Student t tests.
The analyses of neuropsychological test data were carried out in a hierarchical fashion. First, all test scores were converted to z scores, corrected according to standards from external normative study groups (N=87, age range= 19-76). Domain scores were then calculated by averaging the z scores of the primary measure for each test within each domain (general mental functioning, attention, sustained attention and working memory, learning memory [verbal], and remote memory). Domain scores were input into a multivariate analysis of variance (MANOVA) comparing three groups. Educational attainment was recoded to a two-level factor (″ 7 [incomplete first grade education] and >7 years [at least complete first grade education]) for the MANOVA interaction analysis. Age entered the equation as a covariant. The main effect of gender was tested but since no significant impact alone, or as interaction was observed it is not presented.
Logistic Regression was used to determine main multivariate association with learning/memory impairment. For Logistic Regression, the following parameters are presented: B (regression coefficient) S.E. (an estimate of the standard deviation for the error terms in regression), Wald, Odds Ratio (OR) and the 95% Confidence Interval (CI) with lower and upper limits. Table 2, shows mean±SD of tests classified into cognitive/memory domains. The comparisons between groups were adjusted for age. Age correlated with Mini-Mental (B= -0.043; p=0.001), word span (B= -0.013; p=0.002), delayed recall of the word list (B= -0.027; p=0.002) and Logical memory (B= -0.029; p=0.005), famous faces (B= -0.081; p=0.0001), autobiographical data (B= -0.020; p=0.001), and Montgomery-Asberg depression rating scale (B=0.162; p=0.002) (MANOVA covariance: withinsubject effect for the whole sample).

Assessment of depression effect
Diagnosis of depression presented an effect upon the Mini Mental (p=0.0001), Word span (p=0.001), the delayed recall of the Word list (p=0.001), Logical memory    Table 2).
Education showed an independent effect on tests performances (Table 4 and Figure 2). No interaction between depression and educational status was observed.

Use of anti-depressives and antipsychotics and interaction of variables
An additional analysis was carried out with depressed patients alone, severe and mild/moderate, psychoactive drugs and education as independent variables. Dependent variables were cognitive/memory tests and age as covariant. The effect of education was the same as observed above, as was correlation of age with tests. The scores on delayed recall of the word list and on Mini Mental were higher among drug-free depressed patients, than those who were taking antidepressants with other psychiatric medications (p=0.001 and p=0.002, respectively). The patients who were taking antidepressants alone also showed higher test scores than those an antidepressants with other psychiatric medications (p=0.005 and p=0.038, respectively). Drug-free patients showed higher scores on immediate and delayed recall of logical memory than patients taking antidepressants (p=0.001) or antidepressants with other psychiatric medications (p=0.027 and p=0.001, respectively) ( Table 3). Effect of severity of depression was similar to that presented in Table 2.
The analysis of domains (sum of individual test z scores under definition) showed that attention, learning, remote memory, and general mental functioning were impaired in both severe and mild/moderate depressed patients compared to medical inpatients (age and education adjusted) (Figure 1). For the domain "sustained attention and working memory", only severely depressed patients differed from the medical controls (p= 0.008). Severely depressed patients significantly differed from the mild/moderate on domains "Remote memory" and "General mental functioning" (p=0.024 and p= 0.016, respectively) ( Figure 1). The severe patients presented the worst performances.

INDEX (50% of positive tests) -logistic regression
For this model the independent variables age, Montgomery-Asberg depression scale, education, sex, and diagnostic status were used in the analysis.

Discussion
We aimed to evaluate performance on cognitive tests in a group of clinically depressed patients comparing with a group of cognitively normal medical inpatients, analyzing impact of education. Depression showed a significant effect upon cognition as well as education, but no interaction was observed between them. Age correlat-  36 . Cognitive tasks may be sensitive to the effects of some antidepressants 37 and most of our patients were under the effect of such medications. In our sample, we observed a significant effect of psychoactive drugs upon cognitive performance in general mental functioning, learning memory and sustained attention domains. However, a proportion (30%) of our patients was drug-free and was uniformly distributed between severe and mild/moderate groups, as were the other classes of drug use. We carried out analyses, controlled for drug effect, and cognitive/memory performances demonstrated the same independent effect from depression and education.
The effect of severity of depression was observed on five tests (corresponding to five different domains) in this sample. Although the effect of severity of depression on test performance has been measured in many studies by examining the correlation between depression rating scales, especially Hamilton´s, and test scores, the findings have been conflicting. Some studies reported no correlation between performance and severity of depression [38][39][40][41][42] , while others demonstrated this relationship 7,36,[43][44][45] . Correlations could be sensitive to patient selection because the Hamilton Depression scale may be confounded by severe    28,29 are the most extensively used observer instruments world-wide in clinical and psychopharmacological depression research to assess severity of depression after a categorical diagnosis has been ascertained 47 .
The MADRS is increasingly employed in clinical research because earlier studies had suggested the scale could be superior to the traditional HAMD 17 with respect to sensitivity to change 30,48 and other psychometric characteristics 49 .
Education has a significant influence on cognitive test performance. According to our findings, education can be an important confounder in establishing cognitive deficits related to depression. Groups with higher levels of education perform better on most neuropsychological tests [20][21][22][23] . On the other hand, low educational attainment may be responsible for false-positive responses in cognitive assessment. The impact of education associated to presence of diseases on cognitive tests or batteries has been extensively evaluated, even among subjects with lower attainment. There is extensive evidence that low education levels are linked to an indirect index of lower reserve capacity (i.e., a risk factor) which reduces the threshold for neuropsychological abnormality 50 .
Our study emphasized the independent effect of lower education and of diagnosis of depression. The applicability of neuropsychological tests and their performance in countries where rates of illiteracy and low socioeconomic levels are high, as is the case in Brazil, remains a very important issue to be debated. The sample was drawn from a city in which socioeconomic and educational characteristics are different from the majority of the other large Brazilian cities. This may suggest that similar investigations carried out in these locations could serve to demonstrate the practical problems of cross-cultural testing.