Do age and education predict performance of older adults on the d2 Test?

The objective of this study was to investigate the association of age and education in the performance of cognitively preserved older adults in the d2 Sustained-Attention Test, and to compare the results of different age groups and levels of schooling in this instrument. The sample was composed of 211 adults, 60 years of age or older, who were not institutionalized, and who completed a sociodemographic questionnaire, the Mini Mental State Examination, the Geriatric Depression Scale (short form), and the d2 Test. Data analysis was conducted using descriptive statistics, partial correlations, multiple linear regression and one-way ANOVA. The results of partial correlations and multiple linear regression showed that age and years of schooling demonstrated significant associations with all d2 Test scores, with age being the predictive variable that showed the greatest influence on the performance of the older adults. Comparison of performance in the d2 Test among the six groups according to the distribution by age group (60-69 years and 70 years or more) and by levels of schooling (primary, secondary and higher) showed that younger adults with a higher level of schooling scored better on the d2 Test, suggesting the need for normative data studies for this population.


Introduction
Attention is a cognitive function that allows an individual to process a limited amount of information obtained and made available through the sensory organs, memory and other cognitive and physiological processes (Sternberg, 2006). There are different types of attention. One of them is sustained attention, defined as the ability of an individual to keep focus exclusively on only one stimulus and eliminate all others (Rueda & Monteiro, 2013).
Sustained attention may affect the memory performance of older adults (Parente & Taussik, 2002). Considering that sustained attention is directly related to memory performance (Kim et al., 2011), assessment of this function in a suitable manner, with relevant normative data for this population, is particularly relevant to (Fernandes & Santos, 2015;Neri & Yassuda, 2008). The literature supports this assumption (Carreiro et al., 2015;Rueda, 2010;Werlang, 2012). Therefore, this study aimed to investigate the association of age and education level with performance of older adults on the d2 Test, and to compare the performance of different age and educational groups.

Method Design
Cross-sectional study.

Participants
Participants were 211 older adults, recruited for convenience, from a metropolitan region in the South of Brazil, aged between 60 and 94 years (M = 69.83; SD = 6.77), with 1 to 24 years of formal schooling (M = 11.02; SD = 5.11); 87% were women (n = 183). The inclusion criteria were age 60 years or older, not being illiterate, and not being a resident of a long-term care facility. The exclusion criteria were: 1) scores suggestive of cognitive decline on the Mini Mental State Examination -MMSE (Folstein, Folstein, & McHugh, 1975, adapted by Chaves & Izquierdo, 1992 according to cutoff points for education suggested by Kochhann, Varela, Lisboa and Chaves (2010) for older adults from Southern Brazil (<22 for 1-5 years of study, <23 for 6-11, and <24 for 12 or more); 2) scores ≥6 on the Geriatric Depression Scale, short form -GDS-15 (Yesavage et al., 1982-1983, adapted by Almeida & Almeida, 1999; 3) presence of uncorrected primary sensory problems at the time of the evaluation (use of glasses or hearing aids, for example); 4) history of neurological and/or psychiatric disorders, investigated through self-report; and 5) failure to complete all study instruments.

Instruments
Sociodemographic Data. The sociodemographic data form collected the following variables: age, gender, marital status, education, and economic classification criteria (Associação Brasileira de Empresas de Pesquisa, 2015). It also investigated residential conditions, occupation, leisure activities, physical and mental health, medications used, smoking and drinking habits, physical activities and participation in social groups.
Mini Mental State Examination (MMSE). The MMSE is a cognitive screening instrument (Folstein, Folstein, & McHugh, 1975;Bertolucci et al., 1994). It was used with the purpose of excluding potential participants with scores that suggested dementia.
Geriatric Depression Scale, short form (GDS-15). The GDS-15 is a measure used to identify and quantify depressive symptoms in the elderly. Fifteen items compose the short version in Portuguese (Yesavage et al., 1982(Yesavage et al., -1983. This scale was used with the purpose of excluding potential participants with depressive symptoms, with a cutoff of 6 points (Paradela, Lourenço, & Veras, 2005).
d2 Test. The d2 is a cancellation test which measures the level of concentrated visual attention. It may be applied in individual or collective form (Brickenkamp,1962;Mattos & Júnior, 2010;Michels & Gonçalves, 2014;Rueda, 2011). This instrument is used in different contexts, such as driving school tests, organizations and in the clinical setting (Brickenkamp, 2002, Rueda, 2011. It is composed of 14 lines, each one with 47 symbols. The respondent has 20 seconds to find and mark three defined symbols among mixed symbols in each line (Brickenkamp, 2002;Michels & Gonçalves, 2014). Results of the d2 Test are mainly verified via five scores: 1) Total Number of Characters Processed (TN), which indicates the quantitative performance of the examinee and may indicate speed in activities that require attention; 2) Total Errors (E), which comprises the sum of omitted signs and wrong signs ticked, and evaluates sustained attention,; 3) Percentage of Errors (E%), defined by the formula (100 x E/TN), which indicates the association between speed and accuracy; 4) Total Correctly Processed (TN-E), the total number of correct answers (raw score minus total errors), which gives the overall performance; and 5) Fluctuation Rate (FR), which demonstrates fluctuation in the work rate of concentration capacity. This is checked on the line in which the respondent examined the highest number of items and on the line in which they examined the lowest number; the lowest number is then subtracted from from the highest. The distribution of errors (DE) committed in the first lines (1 to 4), in the middle lines (5 to 10) and in the final lines (11 to 14) can be analyzed (Brickenkamp, 2002).

Data Collection Procedure
The relevant institutional Ethics Committee approved the study (CAAE: 14769713.1.0000. 5336). Social groups for older adults in the Porto Alegre Metropolitan Area (universities or community centers, for example) were contacted. Participants were assessed individually in meetings that lasted approximately one hour, conducted between the years 2012 and 2013 at the social centers, in rooms free of visual and/or auditory distracting stimuli. Trained psychologists and psychology students conducted the interviews for data collection. The tests were corrected and analyzed by trained psychologists; d2 Test correction and analysis was computer-based.

Data Analysis Procedure
Descriptive statistics such as average scores, standard deviation, median and percentages were calculated. Data distribution was analyzed using the Kolmogorov-Smirnov test. Associations between the d2 Test scores (TN, E, E%, TN-E, and FR), age and years of study were investigated by partial correlations, controlling for education (years) and age (years). The strength of the associations was interpreted based on Cohen's 1988 classification, which defines values ≤ .2 as a small effect size, ≤ .5 as an medium effect size and ≥ .6 as a large effect size. To investigate potential predictive variables for performance in d2 Test scores (dependent variable), multiple linear regression analysis with the stepwise method was used. Age and years of study were defined as the predictor variables (independent variables). The presence of residues and collinearity was verified by the Durbin-Watson test.
One-way analysis of variance (ANOVA) with Scheffe's post-hoc test was performed to verify differences in performance in the d2 Test between two age groups (60-69 years and 70 years or more), further subdivided into three levels of education: primary (1-9 years of study), secondary (10-12 years of study) and higher (13 years or more of study). This classification was defined based on the Brazilian education system. Results with p < .05 were considered significant. Analyses were conducted using IBM SPSS Statistics for Windows, Version 22.0. Table 1 presents means, standard deviations, medians and ranges for MMSE, the GDS-15, and d2 Test scores in the overall sample. Results for the association and comparative analyses will be presented separately.

Analysis of association between d2 Test scores, age and education
The results of partial correlations correlation analysis are shown in Table 2. Age and education variables showed significant associations with all d2 Test scores, although the strength of the associations ranged from weak to moderate. Regarding age, results showed weak, negative correlations with d2 TN score and TN-E score. Associations with other d2 scores (E, E%, FR and distribution of errors) were weakly positive. Regarding education, results showed a moderate, negative association with the E% score. All other correlations were weak, some negative (E, FR and distribution of errors) and some positive (TN and TN-E). Table 3 presents the multiple linear regression models for each of the d2 Test scores. The lowest variance explained was obtained by the TN score model (6%), and the highest was obtained by the E% score model (20%). Although age and years of study contributed to all d2 Test scores, the standardized scores suggest that age was a predictor of greater influence on TN, E and FR performance, while years of study were more influential predictors of E% and TN-E performance.

Comparison analyzes among age groups stratified by educational levels
.140* -.196** Error distribution (lines 11 to 14) .145* -.192** a Partial correlation controlling for education (years) variable; b Partial correlation controlling for age (years) variable. * p ≤ .05; ** p ≤ .01; *** p ≤ .001. In the TN-E score, the group aged 60-69 with higher education scored significantly higher than the same age group with primary education alone (p post hoc = .004) and than the groups aged 70+ with primary (p post hoc ≤ .001) and secondary (p post hoc ≤ .001) education. Finally, in the FR score, the group aged 70+ with primary education alone obtained significantly lower scores than the groups aged 60-69 with secondary (p post hoc ≤ .005) and higher (p post hoc ≤ .001) education and than the group aged 70+ with higher education (p post hoc = .011). There were no significant differences in performance between groups in the E and error distribution scores.

Discussion
The purpose of this study was to investigate the association of age and education with performance on the d2 Sustained-Attention Test in older adults, and to compare the performance of different age groups, divided by schooling, in this population. The main finding was that age and education showed significant associations with all d2 Test scores. Thus, one can conclude that sustained-attention performance decreases as age increases, and that older adults with a higher level of education perform better.
There was a negative correlation between age and d2 Test net score, which suggests that, the more advanced the age, the worse the level of sustained attention. This finding is corroborated by previous studies which identified that, as age increases, processing speed slows down and the quality of responses may also decrease (Fernandes & Santos, 2015;Rueda et al., 2008;Rueda, Javier & Monteiro, 2013;Pesce et al., 2005), which probably leads to lower levels of attention.
Therefore, age had the highest influence on the performance of older adults in the d2 Test (TN, E, and FR). As age increases, there is a decrease in attention level due to natural aging of the brain, especially in the prefrontal cortex, which is the main area responsible for the attentional process (Custódio, Malaquias Júnior, & Voos, 2010). Thus, one may infer that, as age increases, the sustained attention level declines even in healthy older people.
Furthermore, it has been observed that education level is a protective factor for sustained attention. Participants with a higher level of education obtained higher scores in the d2 Test, corroborating these previous studies (Cecato, Fiorese, Bartholomeu, & Martinelli, 2011;Fernandes & Santos, 2015). In addition, education correlated positively with the net score. It is believed that people with more years of education perform better on this cancellation test because they are literate and familiar with letters (Brewster et al., 2014;Silva, Cardoso, & Fonseca, 2012). Other studies have observed that education works as a protective factor for cognitive capacities during the aging process (Argimon, 2002;Oliveira et al., 2015). Previous research demonstrated that the lower the level of education, the worse the cognitive performance, including on attention (Mello, Haddad,  & Dellaroza, 2011). A lower level of education is also associated with greater cognitive impairment (Argimon & Stein, 2005;Oliveira, et al., 2015). In other way, results pointed out that higher age and fewer years of education were associated to worse performance in sustained attention. These results suggest there is a need for better understanding of the influence of these variables on other cognitive abilities, since sustained attention, when altered, may impair memory and learning. The variables age and education may interfere significantly with attentional tasks, which is supported by the literature. The decrease of sustained attention with advancing age may be related to alterations that occur in the prefrontal cortex, which is one of the structures involved in the attentional process (Lezak et al., 2013).
The influence of education on healthy cognitive abilities in the elderly has been noted in several studies (Meijer, Van Boxtel, Van Gerven, Van Hooren, & Jolles, 2009;Strout & Howard, 2012), suggesting that years of education may be one of the components of the cognitive reserve (Stern, 2009(Stern, , 2012. Therefore, older adults who have completed more years of education may have lower odds of cognitive deterioration compared to those with fewer years of education (Contador, Bermejo-Pareja, Del Ser, & Benito-León, 2015).
Investigation of the influence of age and education on older adults' performance in the d2 Sustained-Attention Test and the gathering of normative data for this population showed how relevant and adequate this instrument was to assess sustained attention in the elderly, as its sensitivity is able to detect neurodevelopmental aspects and educational characteristics. The findings presented herein may contribute to the professional practice of psychological assessment by ensuring the adequacy and precision of this instrument for clinical use in the elderly population. It is known that studies investigating normative data for psychological instruments aim at increasing the evidence of validity and reliability for a determined population, as they consider the cultural, social and economic reality (Alchieri & Cruz, 2014;Fachel & Camey, 2000;Hutz, Bandeira, & Trentini, 2015).
We suggest that additional studies be conducted to obtain normative data about the performance of older adults in sustained attention tests, as the lack of studies focusing on the population over 60 is evident. Limitations of the present study include the small number of men in the sample and the fact that all participants were from Rio Grande do Sul, which limits the generalizability of data for other regions of the country.