INTRODUCTION
Learning and memory are cognitive functions necessary for independent daily living across the lifespan (Strauss et al., Reference Strauss, Sherman and Spreen2006). In particular, neuropsychological assessments of learning and memory aim to measure the cognitive abilities of registering, storing, and retrieving new information. Although multiple neuropsychological instruments are available to assess verbal and visual learning and memory processes [e.g., Rey Auditory Verbal Learning Test (Rey, Reference Rey1958), Selective Reminding Test (Buschke, Reference Buschke1973; Buschke & Fuld, Reference Buschke and Fuld1974), California Verbal Learning Test (Delis et al., Reference Delis, Kramer, Kaplan and Ober1987, Reference Delis, Kramer, Kaplan and Ober2000), Continuous Visual Memory Test (Trahan & Larrabee, Reference Trahan and Larrabee1988), or the Brief Visuospatial Memory Test–Revised (Benedict, Reference Benedict1997)], two of the most widely used instruments to measure learning and memory abilities worldwide are the Rey–Osterrieth Complex Figure (ROCF) Test (Rey, Reference Rey1941) and the Hopkins Verbal Learning Test–Revised (HVLT-R; Benedict et al., Reference Benedict, Schretlen, Groninger and Brandt1998; Brandt & Benedict, Reference Brandt and Benedict2001).
The ROCF uses an asymmetrical stimulus to measure cognitive performance via demonstrated abilities to recall visual information (Fastenau, Reference Fastenau1996). Two conditions commonly utilized in the ROCF to evaluate memory abilities include the Immediate and Delayed Recall trials (Shin et al., Reference Shin, Park, Park, Seol and Kwon2006). The ROCF can be used to evaluate visual-based learning and memory in the context of dementia, traumatic brain injury, or other neurological disorders (e.g., schizophrenia, Huntington’s disease, Korsakoff’s syndrome; Shimamura et al., Reference Shimamura, Salmon, Squire and Butters1987; Silverstein et al., Reference Silverstein, Osborn and Palumbo1998; Tierney et al., Reference Tierney, Nores, Snow, Fisher, Zorzitto and Reid1994). Support for reliability and validity of the ROCF has been established in past research with a wide variety of samples (e.g., pediatric, adult, geriatric; Berry et al., Reference Berry, Allen and Schmitt1991; Fastenau et al., Reference Fastenau, Denburg and Hufford1999; Waber & Holmes, Reference Waber and Holmes1985). More specifically, the ROCF has received support for adequate inter-rater, alternate form, test-retest, and internal consistency reliability (Berry et al., Reference Berry, Allen and Schmitt1991). Its notable psychometric support has contributed to its wide usage. The ROCF has been used worldwide in countries such as Argentina, Bolivia, Canada, Chile, Colombia, Cuba, Denmark, Ecuador, El Salvador, Guatemala, Honduras, Italy, Mexico, New Zealand, Paraguay, Peru, Spain, and the United States (e.g., Ardila & Rosselli, Reference Ardila and Rosselli1994; Fernando et al., Reference Fernando, Chard, Butcher and McKay2003; Galindo & Cortes, Reference Galindo, Cortes, Knight and Kaplan2003; Rivera et al., Reference Rivera, Perrin, Morlett-Paredes, Galarza-del-Angel, Martinez, Garza and Aliaga2015a; Strauss et al., Reference Strauss, Sherman and Spreen2006; Vogel et al., Reference Vogel, Stokholm and Jorgensen2012).
The HVLT-R is an auditory-based measure of learning and memory involving a list of words (Benedict et al., Reference Benedict, Schretlen, Groninger and Brandt1998; Brandt, Reference Brandt1991). Although the HVLT-R has six alternate forms, commonly utilized trials include the Hopkins Total Recall, Hopkins Delayed, and Hopkins Recognition forms (Benedict et al., Reference Benedict, Schretlen, Groninger and Brandt1998). The HVLT-R is also used to detect memory impairments associated with dementia, brain injury, HIV/AIDS, or other neurological disorders (Cysiqu et al., Reference Cysique, Jin, Franklin, Morgan, Shi, Yu and Ake2007; Kuslansky et al., Reference Kuslansky, Katz, Verghese, Hall, Lapuerta, LaRuffa and Lipton2004). Some research suggests it is best used for elderly people suspected of having dementia (Shapiro et al., Reference Shapiro, Benedict, Schretlen and Brandt1999). It has received support for test–retest reliability, construct validity, and concurrent validity in past research with adults and geriatric populations (Benedict et al., Reference Benedict, Schretlen, Groninger and Brandt1998; Shapiro et al., Reference Shapiro, Benedict, Schretlen and Brandt1999). The HVLT-R has also been used globally in countries such as Argentina, Brazil, Bolivia, Cameroon, Chile, China, Colombia, Cuba, El Salvador, Guatemala, Honduras, India, Mexico, Paraguay, Peru, South Africa, and the United States (Arango-Lasprilla et al., Reference Arango-Lasprilla, Rivera, Garza, Saracho, Rodríguez, Rodríguez-Agudelo and Martínez2015; Benedict et al., Reference Benedict, Schretlen, Groninger and Brandt1998; Cysique et al., Reference Cysique, Jin, Franklin, Morgan, Shi, Yu and Ake2007; Hoare et al., Reference Hoare, Westgarth-Taylor, Fouche, Combrinck, Spottiswoode, Stein and Joska2012; Kanmogne et al., Reference Kanmogne, Kuate, Cysique, Fonsah, Eta, Doh and McCutchan2010; Rivera et al., Reference Rivera, Olivera Plaza, Quijano, Calderón Chagualá, De los Reyes Aragón, Utria Rodríguez, Arango-Lasprilla, Arango-Lasprilla and Rivera2015b; Yepthomi et al., Reference Yepthomi, Paul, Vallabhaneni, Kumarasamy, Tate, Solomon and Flanigan2006).
Although these measures were originally normed with English-speaking samples, normative data now exist for Spanish speaking adults for both the ROCF and HVLT-R (Arango-Lasprilla et al., Reference Arango-Lasprilla, Rivera, Garza, Saracho, Rodríguez, Rodríguez-Agudelo and Martínez2015; Arango-Lasprilla & Rivera, Reference Arango-Lasprilla and Rivera2015; Cherner et al., Reference Cherner, Suarez, Lazzaretto, Fortuny, Mindt and Dawes2007; Rivera et al., Reference Rivera, Perrin, Morlett-Paredes, Galarza-del-Angel, Martinez, Garza and Aliaga2015a, Reference Rivera, Olivera Plaza, Quijano, Calderón Chagualá, De los Reyes Aragón, Utria Rodríguez, Arango-Lasprilla, Arango-Lasprilla and Rivera2015b). Standardized normative data for diverse samples are necessary to validly assess memory outside of the United States and reduce the risk for misinterpretation of scores. For example, the risk of score misinterpretation is high when using an improper normative sample as a comparison. In addition to adequate, representative normative data, another important point of concern to reduce misinterpretation of scores on neuropsychological assessments is to determine the frequency and determinants of low test scores among healthy individuals.
Multivariate base rates (MVBRs) of low scores allow neuropsychologists to simultaneously interpret large amounts of data in different populations. When a battery of assessments is completed, chances increase dramatically for individuals to have one or more low scores on any individual test (Binder et al., Reference Binder, Iverson and Brooks2009; Brooks et al., Reference Brooks, Strauss, Sherman, Iverson and Slick2009, Reference Brooks, Sherman and Iverson2010, Reference Buschke2017). Thus, clinicians who are tasked with interpreting a large amount of clinical data must determine whether or not results reflect cognitive impairment (i.e., true positive) or a low score in an otherwise healthy individual (i.e., false positive). In addition, factors such as age, education, and gender tend to alter MVBRs and increase the prevalence of low scores among samples (Brooks & Iverson, Reference Brooks and Iverson2010; Schretlen et al., Reference Schretlen, Testa, Winicki, Pearlson and Gordon2008). Thus, MVBRs are an additional interpretation tool that can be used to improve the accuracy of identifying cognitive impairments and reduce misdiagnosing deficits where there are none. MVBRs have been developed for adult clinical samples to evaluate impairments such as amnestic mild cognitive impairment, Alzheimer’s disease (Oltra–Cucarella et al., Reference Oltra-Cucarella, Sánchez-SanSegundo, Lipnicki, Sachdev, Crawford and Pérez-Vicente2018), and mild neurocognitive disorder (Holdnack et al., Reference Holdnack, Tulsky, Brooks, Slotkin, Gershon, Heinemann and Iverson2017). Although MVBRs have been examined in English-speaking White adult populations (e.g., Brooks et al., Reference Brooks, Sherman and Iverson2010; Schretlen et al., Reference Schretlen, Testa, Winicki, Pearlson and Gordon2008), to date no study has tested MVBRs with Spanish-speaking populations.
Demographic and culture-related factors have shown to influence low scores (e.g., Brook et al., Reference Brooks, Sherman and Iverson2010, Reference Brooks, Holdnack and Iverson2017). As such, it is expected that MVBRs change from culture to culture. The present study aims to fill this gap in the literature by examining MVBRs among a Spanish-speaking adult Latino sample across 11 countries and the commonwealth of Puerto Rico who completed the ROCF and HVLT-R to assess their learning and memory capacities. Developing MVBRs among Spanish-speaking Latino individuals will allow for improved clinical interpretation of their neuropsychological performance to reduce the likelihood of over-diagnosing cognitive deficits. The goal of the present study is to develop and present the base rates of low scores on the ROCF and HVLT-R in a table that can facilitate interpretation of test scores to maintain an adequate false-positive rate when these two assessments are administered in a battery together (e.g., Brooks et al., Reference Brooks, Sherman and Iverson2010). It was hypothesized that the prevalence of low scores on the ROCF and HVLT-R, as determined using MVBRs, will exceed the expected prevalence rates found when interpreting a single score in isolation.
METHODS
Participants
The sample consisted of 5402 healthy individuals who were recruited from Argentina, Bolivia, Chile, Colombia, Cuba, El Salvador, Guatemala, Honduras, Mexico, Paraguay, Peru, and Puerto Rico. The demographic characteristics (i.e., age, education, and sex) by country can be found in Table 1.
Table 1. Sample distribution by country, age, education, and sex
To be eligible for study participation, individuals must have met the following requirements: (a) were between 18 and 95 years of age, (b) were born and currently live in the country where the protocol was conducted, (c) spoke Spanish as their native language, (d) had completed at least 1 year of formal education, (e) were able to read and write at the time of evaluation, (f) scored ≥23 on a Spanish version of the Mini-Mental State Examination (Folstein et al., Reference Folstein, Folstein and McHugh1975; Villaseñor-Cabrera et al., Reference Villaseñor-Cabrera, Guàrdia-Olmos, Jiménez-Maldonado, Rizo-Curiel and Peró-Cebollero2010), (g) scored ≤4 on a Spanish version of the Patient Health Questionnaire–9 (PHQ-9; Kroenke et al., Reference Kroenke, Spitzer and Williams2001), and (h) scored ≥90 on the Barthel Index (Mahoney & Barthel, Reference Mahoney and Barthel1965).
A self-report questionnaire was administered to collect information about the participants’ medical history and health status. Participants were determined to be ineligible if they reported or endorsed the following: (a) medical services received for diagnosed neurological or psychiatric conditions, (b) daily consumption and/or use of an illicit substance, (c) history of chronic disease (e.g., diabetes mellitus), (d) regular use of pain or other medications that may impact cognitive functioning, and/or (e) severe visual and/or hearing deficit. All participants were community volunteers who did not receive financial compensation for participation.
Measures
Rey–Osterrieth Complex Figure
The examiner administered the ROCF Figure A, which included the Copy portion, Immediate Recall after a 3-min delay, and then the Delayed Recall 30 min after the copy trial. The Spanish-language ROCF manual scoring guidelines were followed (Rey, Reference Rey2009). The ROCF includes 18 elements, and the maximum score for each of the two tasks (Immediate and Delayed Recall) is 36. Two points are given when the element is correctly reproduced; 1 point is given when the reproduction is distorted, incomplete but placed properly, or complete but placed poorly; and .5 point is credited when the element is distorted or incomplete and placed poorly. A score of 0 is given when the element is absent or is not recognizable (Osterrieth, Reference Osterrieth1944). The ROCF is one of the 10 most commonly used tests by clinicians and researchers from 16 Latin American countries (Arango-Lasprilla et al., Reference Arango-Lasprilla, Stevens, Morlett Paredes, Ardila and Rivera2017).
Hopkins Verbal Learning Test–Revised
The HVLT-R list used in the present study was Form 5 because pilot testing supported that all words included on the list were known, used, and represented the same meaning in each participating country. Form 5 contains a list of 12 semantically related words in three categories (i.e., professions, sports, and vegetables). Three trials of successive learning are presented, in which the list of 12 words is read to the participant, and the correct answers of each learning trial are recorded. Total Recall is the sum of words recalled correctly in the three trials. After 20–25 min, the Delayed Recall and recognition phase occurs, where the subject is asked to recall all the words that they can remember from the initial list (Benedict et al., Reference Benedict, Schretlen, Groninger and Brandt1998; Brandt, Reference Brandt1991). HVLT-R has received support for adequate psychometric properties with Spanish-speaking populations (Guàrdia-Olmos et al., Reference Guàrdia-Olmos, Rivera, Peró-Cebollero, Arango-Lasprilla, Arango-Lasprilla and Rivera2015b).
Procedure
The participants completed the ROCF and HVLT-R as part of a large battery of neuropsychological tests. For further information regarding the study’s procedure, see Arango-Lasprilla and Rivera (Reference Arango-Lasprilla and Rivera2015) and Guàrdia-Olmos et al. (Reference Guàrdia-Olmos, Rivera, Peró-Cebollero, Arango-Lasprilla, Arango-Lasprilla and Rivera2015b). The University of Deusto’s (Bilbao, Spain) Ethics Committee approved this study as the coordinating institution.
Statistical Analyses
Sample Size
The accuracy of the total sample size by country was established using classical estimation assuming infinite (i.e., very large) population sizes (Arrufat et al., Reference Arrufat, Olmos and Blanxart1999), where the case of maximum uncertainty was assumed (π = 1 – π = .5) with a confidence interval of 95%. The maximum error of sample sizes ranges from .063 to .049.
Demographic Variables’ Effect on Neuropsychological Performance
The effects of demographic variables on ROCF (Immediate and Delayed Recall) and HVLT-R (Total Recall, Delayed Recall, and Recognition) scores were evaluated by means of multiple linear regression analyses. The full regression models included the following as predictors: age, age2, level of education, sex, and all two-way interactions between these variables. Age was centered (= calendar age—mean age in the sample by country) before computing the squared age term to avoid multicollinearity (Kutner et al., Reference Kutner, Nachtsheim, Neter and Li2005). Use of the squared age term allows for determination of potential linear or quadratic (e.g., curvilinear) effects of age on test scores. Education was dummy coded into a variable of 0 and 1: 1 if the participant had >12 years of education and 0 if the participants had 1–12 years of education (Guàrdia et al., Reference Guàrdia, Jarne, Pena-Casanova and Gil2005; Peña-Casanova et al., Reference Peña-Casanova, Blesa, Aguilar, Gramunt-Fombuena, Gómez-Ansón, Oliva and Martínez-Parra2009), and Sex was dummy coded as Male = 1 and Female = 0. Independent variables that were not statistically significant in the multiple regression model were removed from the model, and the reduced model was fitted again. In the stepwise model-building procedure, no predictor was removed as long as it was also included in a higher order term in the model (Aiken et al., Reference Aiken, West and Reno1991). The full regression model can be formally described as:
$ {y_i} = {{\rm{B}}_0} + {{\rm{B}}_1} \cdot {(Age - {\overline x_{Age\,by\,\,country}})_i} + {{\rm{B}}_2} \cdot (Age - {\overline{ x}_{Age\,by\,\,country}})_i^2 + {{\rm{B}}_3} \cdot {(Level\,Education)_i} + {{\rm{B}}_4} \cdot Se{x_i} + {{\rm{B}}_k} \cdot Interaction{s_i} + {\varepsilon _i}$
. A Bonferroni-corrected alpha-level of .005 (=.05 / 9) was used. The model assumes that the residuals εi are normally distributed with mean 0 and variance
$\sigma _\varepsilon ^2$
, i.e.,
${\varepsilon _i}\ ∼N(0,\sigma _\varepsilon ^2)$
. For all multiple linear regression models, the following assumptions were evaluated: (a) multicollinearity [evaluated by computing the Variance Inflation Factor (VIF), which should not exceed 10, and by computing the collinearity tolerance values, which should not exceed 1], (b) homoscedasticity (evaluated by grouping the participants into quartiles of the predicted test scores and applying Levene’s test on the residuals), (c) normality of the standardized residuals (evaluated by conducting the Kolmogorov–Smirnov test), and (d) the existence of influential values (evaluated by computing the maximum Cook’s distance).
Calculation of Adjusted Z-score
Adjusted Z-scores for each raw score were calculated using the information provided in each final regression model in a three-step procedure (Rivera & Arango-Lasprilla, Reference Rivera and Arango-Lasprilla2017; Van Der Elst et al., Reference Van Der Elst, Van Boxtel, Van Breukelen and Jolles2006a, Reference Van Der Elst, Van Boxtel, Van Breukelen and Jolles2006b): (1) The expected test score (Ŷi) is computed based on the fixed effect parameter estimated of the established final regression model: Ŷi = B0 + B1X1i + B2X2i + ... + BKX2i. (2) To obtain the residual value ei , a subtraction between the raw score of the neuropsychological test Yi and the predicted value (Ŷi) previously calculated was performed as shown in the following formula: ei = Yi - εi. (3) Using the residual standard deviation (SDe ) value provided by the regression model, residuals were standardized: zi = ei/SDe. This three-step process was applied to each score (ROCF Immediate Recall, ROCF Delayed Recall, HVLT-R Total Recall, HVLT-R Delayed Recall, and HVLT-R Recognition) separately for each country.
Multivariate Base Rates
The exact percentile corresponding to the Z-score previously calculated was obtained using the standard normal cumulative distribution function (if the model assumption of normality of the residuals was met in the normative sample), or via the empirical cumulative distribution function of the standardized residuals (if the standardized residuals were not normally distributed in the normative sample). Percentiles that are routinely used in clinical practice or research as indicator of low performance were analyzed in this study: (a) below the 25th percentile, (b) below the 16th percentile, (c) below the 10th percentile, (d) below the 5th percentile, and (e) below the 2nd percentile.
The prevalence below each of these percentiles was calculated. This base-rate analysis was calculated to involve examination of learning and memory performance on the five Z-scores (ROCF Immediate Recall, ROCF Delayed Recall, HVLT-R Total Recall, HVLT-R Delayed Recall, and HVLT-R Recognition) simultaneously, not each score in isolation. All the analyses were performed using SPSS version 23 (IBM Corp., 2015).
RESULTS
The assumptions of multiple linear regression analysis were largely met for all final models. There was no multicollinearity (i.e., the VIF values in all final models did not exceed 3.143, and thus well below the threshold value of 10 that is indicative of multicollinearity; collinearity tolerance values did not exceed the value of 1) nor influential cases (i.e., the maximum Cook’s distance value was .493). Levene’s test suggested homoscedasticity in all countries except for the models of Argentina and Paraguay in HVLT-R Delayed Recall, and for Paraguay in HVLT-R Total Recall. In ROCF Immediate and Delayed Recall, homoscedasticity was not met in all countries except for Paraguay in ROCF Immediate Recall, and for Chile, Cuba, El Salvador, Guatemala, Honduras, Mexico, Peru, and Puerto Rico in ROCF Delayed Recall. Standardized residuals of the models were normally distributed in all countries (as evaluated with the Kolmogorov–Smirnov test) except for the HVLT-R Recognition and the ROCF Immediate Recall in Argentina, Chile, Colombia, Cuba, Guatemala, Honduras, Mexico, Peru, and Puerto Rico.
Table 2 shows the final regression models for each score (ROCF Immediate Recall, ROCF Delayed Recall, HVLT-R Total Recall, HVLT-R Delayed Recall, and HVLT-R Recognition) and country. The amount of variance explained in scores ranged from 1.8% (in Cuba on the HVLT-R Recognition score) to 44.9% (in Paraguay on the HVLT-R Delayed Recall).
Table 2. Beta coefficients and R 2 for each score and country
The base rates of low test scores on the memory and learning performance are presented in Table 3. Between 57.3% (El Salvador) and 64.6% (Bolivia) of the sample have at least one of the five scores below the 25th percentile, and between 36.3% (Paraguay) and 49.3% (Bolivian and Cuba) scored below the 16th percentile on one or more scores. Moreover, between 27.1% (El Salvador) and 33.9% (Puerto Rico) scored below the 10th percentile on at least one of the five scores, and between 24.8% (Paraguay) and 33.9% (Puerto Rico) scored below the 5th percentile on one or more scores. Finally, between 5.9% (Chile, El Salvador, and Peru) and 10.3% (Argentina) scored below the 2nd percentile on at least one of the five scores.
Table 3. Cumulative proportion of adults with the specified number of adjusted learning and memory low scores below the specified percentile cutoff by country
An example will be provided to facilitate the interpretation of Table 3. For example, in Colombia, 64.3% of the sample have at least one of the five scores below the 25th percentile, 46.7% below the 16th percentile, 32.6% below the 10th percentile, 18% below the 5th percentile, and 7.5% below the 2nd percentile. The same results are represented visually in Figure 1.
Fig. 1. Cumulative proportion of Colombian adults with the specified number of adjusted learning and memory low scores below the specified percentile cutoff.
Additionally, the reader can find in Appendices A1–A12 in the Supplementary Material the base rates of low test scores on the memory and learning performance for each country divided by age, sex, and education.
DISCUSSION
The recent collection and publication of normative data for the ROCF and HVLT-R based on N = 5402 Hispanic adults across 11 countries and Puerto Rico represent a leap forward for neuropsychological assessments with Spanish-speaking populations. Without adequate normative data to use with adults from Argentina, Bolivia, Chile, Colombia, Cuba, El Salvador, Guatemala, Honduras, Mexico, Paraguay, Peru, and Puerto Rico, it is likely that memory impairments have been over-diagnosed by clinicians at an alarming rate (e.g., Cherner et al., Reference Cherner, Suarez, Lazzaretto, Fortuny, Mindt and Dawes2007). The presentation of MVBRs for these Spanish-based normative scores represents another step forward in the interpretation of these data and the ethical progression towards lowering the rates of misdiagnosed memory impairments.
The results of the present study supported our hypothesis. When considering MVBRs for the five scores from the ROCF and HVLT-R, the obtained prevalence rates of low scores far exceeded the theoretical prevalence rates based on a Gaussian distribution for a single score. For example, having one or more memory scores <16th percentile (i.e., one standard deviation below the mean) occurred in 27–34% of Spanish-speaking adults, which is much higher than the theoretical base rate of <16%. If a clinician wanted to maintain a prevalence rate of <16% for low scores when considering MVBRs, then this would be achieved by interpreting three or more scores <16th percentile (i.e., found in 7.3–12.5% across the countries). One or more “impaired” memory scores, when impaired is defined as a score falling below the 2nd percentile (more than two standard deviations below the mean), was found in 6–10% of Spanish-speaking adults. Again, maintaining the desired <2% prevalence rate could be achieved by having two or more scores <2nd percentile, rather than just one. A clear advantage of using MVBRs when interpreting multiple scores simultaneously is that they adjust for the inflation of prevalence rates that exceed conventional expectations when only interpreting a single score (Binder et al., Reference Binder, Iverson and Brooks2009; Brooks et al., Reference Brooks, Iverson and White2009, Reference Brooks, Sherman and Iverson2010; Schretlen et al., Reference Schretlen, Testa, Winicki, Pearlson and Gordon2008).
The results of this study from a large Spanish-speaking population are consistent with the literature on MVBRs of memory scores using North American English-speaking samples. Prior studies considering English-speaking samples of adults (Brooks et al., Reference Brooks, Holdnack and Iverson2011, Reference Brooks, Iverson, Holdnack, Holdnack, Drozdick, Weiss and Iverson2013) and older adults (Brooks et al., Reference Brooks, Iverson, Holdnack and Feldman2008, Reference Brooks, Iverson and White2009, Reference Brooks, Holdnack and Iverson2011, Reference Brooks, Iverson, Holdnack, Holdnack, Drozdick, Weiss and Iverson2013) have also shown high rates of low memory scores when multiple scores are considered simultaneously. On the WMS-IV (Wechsler, Reference Wechsler2009) in adults aged 18–69 years, having one or more low scores (≤16th percentile) was found in 29% of the sample when considering four primary indexes and in 51% when considering six scores from the primary subtests. In a large sample of older adults aged 55–87 years, having one or more low scores (≤16th percentile) on the WMS-IV was found in 64% of individuals when considering eight scores from primary memory subtests. And finally, in another large sample of older adults aged 55–79 years, having one or more low scores (≤16th percentile) on the NAB Memory Module was found in 55% when considering the 10 primary scores. Clearly, having low scores is not necessarily atypical, and the base rates of low scores increase as the number of scores increases; therefore, adjusting interpretation using MVBRs will help clinicians to minimize misdiagnosis (Brooks et al., Reference Brooks, Iverson and White2007, Reference Brooks, Iverson, Feldman and Holdnack2009).
The base rates of low memory scores did not differ across the broad levels of education in this Spanish-speaking sample. Although prior MVBR studies with English-speaking samples have often shown that base rates are higher among those with fewer years of education—for example, one or more low WMS-IV scores (≤16th percentile) was found in 84% of adults with eight or fewer years of education but in only 37% with 16 or more years of education (Brooks et al., Reference Brooks, Iverson, Holdnack, Holdnack, Drozdick, Weiss and Iverson2013)—those with 1–12 years of schooling had roughly equivalent rates of low memory scores compared to those with more than 12 years of education in the present study. One potential reason for the absence of differences in the present study was that the standard scores were adjusted using the regression for education (in addition to age and sex). When scores are adjusted for education, different base rates of low scores across education levels become minimal and nonsignificant (Brooks et al., Reference Brooks, Iverson, Holdnack, Holdnack, Drozdick, Weiss and Iverson2013).
These results should be interpreted in the light of the following limitations: (a) In this study, the MVBRs were calculated for two of the most commonly used tests to measure learning and memory processes; however, we do not know if these results are similar or different when other memory tests are used. (b) The number of tests used in the present study was five, and thus it is possible that to the extent that more scores from other memory tests were included, these results could even be lower. (c) The present study was conducted with a large Spanish-speaking population from 11 Latin American countries and Puerto Rico, and for this reason it is not possible to generalize these results to those countries outside of the present sample or those whose language is not Spanish (e.g., Brazil). (d) It is possible that the low scores found in this study could be explained by some variables that were not measured or not considered when carrying out the study, such level of bilingualism and the quality of education, among others. (e) Education was used as a dummy coded, dichotomous variable (i.e., 12 or >12 years of education), and as such, future studies should include education as a continuous variable. Finally, (f) the sample was not stratified by intellectual level, which has been shown to be associated with different base rates of low scores on cognitive measures (Brooks et al., Reference Brooks, Iverson, Holdnack and Feldman2008, Reference Brooks, Iverson and White2009, Reference Brooks, Holdnack and Iverson2011, Reference Brooks, Iverson, Holdnack, Holdnack, Drozdick, Weiss and Iverson2013; Guàrdia-Olmos et al., Reference Guàrdia-Olmos, Peró-Cebollero, Rivera and Arango-Lasprilla2015a; Rivera & Arango-Lasprilla, Reference Rivera and Arango-Lasprilla2017). Future research will consider MVBRs in Latino samples with varying levels of intellectual abilities.
Having access to normative data from Argentina, Bolivia, Chile, Colombia, Cuba, El Salvador, Guatemala, Honduras, Mexico, Paraguay, Peru, and Puerto Rico is an advancement for neuropsychologists who assess Spanish-speaking adults. Knowing the MVBRs of commonly used memory scores in this large sample will improve the interpretation of these normative data. Consistent with the literature with English-speaking adults, it is also common for Spanish-speaking adults to have higher rates of low memory scores when multiple scores are being interpreted. Thus, the presence of low scores may not necessarily indicate an impairment. MVBRs are an interpretive tool that clinicians should not ignore, but instead should use judiciously and with clinical judgment.
ACKNOWLEDGMENTS
Brian Brooks acknowledges partial salary funding from the Canadian Institutes for Health Research (CIHR) Embedded Clinician Researcher Salary Award ().
CONFLICT OF INTEREST
Authors declare no conflicts of interest except Brian Brooks.
CONFLICT STATEMENT FOR BRIAN BROOKS
Dr. Brooks reports the following conflicts of interest: co-author of the Child and Adolescent Memory Profile (ChAMP, Sherman and Brooks, Reference Sherman and Brooks2015, PAR Inc.), Memory Validity Profile (MVP; Sherman and Brooks, Reference Sherman and Brooks2015, PAR Inc.), and Multidimensional Everyday Memory Ratings for Youth (MEMRY, Sherman and Brooks, Reference Sherman and Brooks2017, PAR Inc.), and he receives royalties for the sales of these tests; coeditor of the Pediatric Forensic Neuropsychology textbook (2012, Oxford University Press) and receives royalties for the sales of this book; previously been provided with free test credits from CNS Vital Signs as an in-kind support for his research.
SUPPLEMENTARY MATERIALS
To view supplementary material for this article, please visit http://doi.org/10.1017/S135561771900050X.
