Abstract

Information regarding cognitive abilities in earlier stages of life is essential to ascertain if and to what extent these may have declined. When unavailable, clinicians rely on estimate methods. One of the contemporary methods used worldwide combines performance on irregular word reading test with demographics since it has shown to provide reliable estimates of premorbid ability. Hence, a reading test portuguese irregular word reading test (TeLPI) was developed, filling an important gap in the neuropsychological evaluation of Portuguese speakers. Using 46 irregular, infrequent Portuguese words, TeLPI was validated against Wechsler Adult Intelligence Scale (WAIS)-III (N = 124), and regression-based equations were determined to estimate premorbid IQ considering TeLPI scores and demographic variables. TeLPI scores accounted for 63% of the variance of WAIS-III Full-Scale IQ, 62% of Verbal IQ, and 47% of Performance IQ and thus were considered valid for premorbid intelligence estimation.

Introduction

To characterize the extent to which an individual's cognitive abilities have declined, knowledge of cognitive performance in earlier stages of life is essential (Mackinnon, Ritchie, & Mulligan, 1999). In fact, the very concept of cognitive deficit assumes the existence of some previous normal or ideal level of functioning, against which patient outcomes can be compared and measured in a reliable and valid way (APA, 1998; Lezak, Howieson, & Loring, 2004; Mackinnon et al., 1999). However, such information is rarely available (Matsuoka, Masatake, Kasal, Koyama, & Kim, 2006), and therefore alternative methods for estimating premorbid ability (premorbid intelligence or IQ) must be used instead (Schoenberg, Lange, Marsh, & Saklofske, 2011). Several approaches for estimating premorbid IQ have been suggested. Some take into account qualitative data such as the individual's school and occupational records, family reports as well as socio-economic and educational levels (Crawford & Allan, 1997). Although reasonable findings can result from a qualitative estimation of premorbid intelligence (Baade & Schoenberg, 2004), there are several errors/deviations that may skew its accuracy (Kareken & Williams, 1994). In response to this problem, a variety of quantitative methods for estimating premorbid intelligence were developed (Franzen, Burgess, & Smith-Seemiller, 1997; Lezak et al., 2004) based on (a) resistant measures, (b) demographic equations, (c) reading tests, and (d) combined demographic and ability methods (e.g., Yates, 1956; Oklahoma Premorbid Intelligence Estimate-3: OPIE-3, Schoenberg, Scott, Duff, & Adams, 2002).

Premorbid estimation methods based solely on the current performance on resistant measures, such as the highest single subtest score on the WAIS (Wechsler Adult Intelligence Scale; Wechsler, 2008) or the WAIS vocabulary score, have fallen out of clinical use (Schoenberg et al., 2011). This can largely be attributed to research that has shown vocabulary tests, which require oral definitions and access to semantic word meaning, to be more vulnerable to brain damage (Del Ser, González-Montalvo, Martinez-Espinosa, Delgado-Villapalos, & Bermejo, 1997; Fuld, 1983) than verbal tests with briefer responses requiring only recognition or calling on practical experience (Lezak et al., 2004). Irregular word reading tests were subsequently developed (e.g., Del Ser et al., 1997; O'Carroll & Gilleard, 1986) with the rationale that, in cases of cognitive decline, the phonological component of language involved in reading aloud is better preserved than the semantic component, as phonology appears to be less dependent on the integrity of higher cognitive process than semantics (Bayles & Boone, 1982).

The reading paradigm has gained great acceptance in neuropsychological assessment and various instruments for estimating premorbid IQ have been developed worldwide: the National Adult Reading Test (NART; Nelson, 1982), the North American Adult Reading Test (Blair & Spreen, 1989), the American Adult Reading Test (Grober & Sliwinski, 1991), the French NART (Mackinnon et al., 1999), the Word Accentuation Test (WAT) in Spain (Del Ser et al., 1997), the Wechsler Test of Adult Reading (WTAR; The Psychological Corporation, 2001), the Japanese NART (JART; Matsuoka et al., 2006), the Swedish NART (NART-SWE; Rolstad et al., 2008), the Hopkins Adult Reading Test (HART; Schretlen et al., 2009), the Test of Premorbid Functioning (TOPF; NCS Pearson Corporation, 2009), and TOPF-UK (NCS Pearson Corporation, 2011).

The task on these instruments consists of reading aloud about 50 (depending on the instrument) irregular and infrequent native words graded by difficulty. The individual's result corresponds to the number of reading errors. Given that intelligent guesswork will not provide the correct pronunciation for each word in the test (i.e., pronunciation cannot be determined from spelling, due to the presence of irregular letter-sound pairings), it has been argued that performance on irregular words reading tests is most likely to depend on previous knowledge and not on the current cognitive capacity (Nelson, 1982; Nelson & Wilson, 1991).

A potential drawback to the current reading approaches lies in the possibility of a neurological insult that can also disrupt reading ability. Performance may also be affected by developmental language disorders since research has shown that the current reading ability is dependent on education and verbal abilities (The Psychological Corporation, 2001). Reading tests may furthermore be inappropriate for estimating premorbid intelligence in subjects with language lateralized brain dysfunction, reading disorder or lesser educated individuals (Schoenberg et al., 2011). Nevertheless, reading tests as the NART (Nelson & Wilson, 1991) are reported to be among the most reliable tests in clinical use (McGurn et al., 2004).

In estimating premorbid intelligence, an alternative to reading tests involving irregular words is the use of regression models (Barona, Reynolds, & Chastain, 1984; Crawford & Allan, 1997). Given that most of these models incorporate socio-demographic variables such as years of education, demographic predictions may be biased by social or individual conditions (e.g., developmental disorders or lack of motivation at school) which may undermine academic performance or employment and therefore underestimate premorbid ability. However, the biggest disadvantage of strictly demographic methods, when compared with those based on the oral reading of irregular words, is their lower accuracy in the estimation of premorbid intelligence in neurologically normal individuals (Schoenberg et al., 2011).

Contemporary methods combine performance on irregular word reading tests with demographics to predict WAIS-R, WAIS-III, or WAIS-IV indices since this combination provides more reliable estimates of premorbid ability than the use of either of them (e.g., Crawford, Nelson, Blackmore, & Cochrane, 1990) (cf., however for contradictory data: e.g., Blair & Spreen, 1989). The HART (Schretlen et al., 2009), the WTAR (The Psychological Corporation, 2001), and the TOPF-UK (NCS Pearson Corporation, 2011) are some examples. Another approach combines performance on selected WAIS-III subtests with demographics to predict premorbid WAIS-III Full-Scale IQ (FSIQ) (OPIE-3, Schoenberg, Duff, Scott, & Adams, 2003) and Verbal IQ (VIQ) and Performance IQ (PIQ) (OPIE-3P, Schoenberg, Duff, Dorfman, & Adams, 2004).

The aim of this study is to construct and validate a reading test portuguese irregular word reading test (TeLPI) comprised of Portuguese irregular words and to develop regression equations with significant demographic variables for WAIS-III FSIQ, VIQ, and PIQ (Wechsler, 2008). For these purposes, three steps were followed:

  • Construction of a reading test for subjects of different intellectual levels.

  • Validation of the test by relating it to another general intelligence test (WAIS-III), selection of the best items, and assessment of its reliability.

  • Determination of an empirical law (regression-based equations) to estimate the premorbid intellectual level (full and subscale scores on the WAIS-III) considering the reading test scores or other pertinent demographic variables.

Methods

Since reading tests that assess premorbid IQ are based on orthographic irregularities that are specific to a given language, the construction of a comparable test for the Portuguese population could not be accomplished by merely translating the items (words) of existing tests. Words in the target language must share properties with those used in these tests and must furthermore represent the native vocabulary. Hence, a preliminary study was conducted for the purpose of selecting a group of irregular words that were best suited for predicting IQ in a sample of healthy subjects. Having defined basic criteria for determining letter-sound irregularity, the first step consisted of selecting all of the eligible irregular words (e.g., ubiquidade “ubiquity”/guia “guide”/exame “exam”/caixa “box”) listed in the Portuguese lexical frequency database “Corlex” (Centro de Linguística da Universidade de Lisboa, 2003), from a total number of 16,210,438, thus obtaining 1,417 irregular words. Subsequently, all irregular words with a frequency rate on CORLEX above 27 (frequent and very frequent words) were eliminated and further refined linguistic criteria (elimination of technical jargon, or numerals, for instance) led to the final selection of 105 words considered suitable for the purposes of the experimental version of TeLPI (Alves, Simões, & Martins, 2010). In order to define coding criteria, a phonetic transcription (The International Phonetic Association, 2005) was provided for each word. Materials included a score sheet with the pronunciation criteria as well as a test book in which each word was presented separately and printed in a bold, 18-point font. In testing, TeLPI was introduced to the examinee as follows: “I will be showing you some words that I'd like you to read slowly out loud. Some words you may not recognize, but try reading them anyway.” For all participants, responses were recorded in digital audio format to ensure accurate scoring.

The 105 wordlist was applied on a sample of 130 healthy, community-dwelling Portuguese speakers that were born and had completed their formal education in Portugal, 16 years of age or older. Informed consent was obtained from each subject after the aim of the study was explained. To ensure that participants were cognitively healthy adults, the recruited subjects were interviewed by a psychologist with a standard questionnaire that included complete socio-demographic data, an inventory of current clinical health status, past habits, and medical history. Autonomy in daily living activities, motor, speech, audition, or vision disorders, of alcoholism history or substance abuse, neurological or psychiatric diseases, as well as of chronic unstable systemic disorders with impact in cognition, significant depressive complaints, and medication with possible impact in cognition (e.g., psychotropic or psycho-active drugs) were also checked in this interview since they represent possible exclusion criteria. In the case of older participants (subjects over 65 years), all information was always checked with their general practitioner, community center directors, and/or an informant, usually an individual in co-habitation or a close relative. All subjects underwent the 105 irregular word reading test, 9 of the 12 subtests of the Portuguese version of WAIS-III (Wechsler, 2008) that most correlate with FSIQ (i.e., Information, Vocabulary, Arithmetic, Comprehension, Similarities, Picture Completion, Block Design, Matrix Reasoning, and Symbol search) as well as the Mini-Mental State Examination (MMSE; Folstein, Folstein, & McHugh, 1975; Guerreiro, 1998), the Montreal Cognitive Assessment (MoCA; Nasreddine et al., 2005; Simões et al., 2008), and the Geriatric Depression Scale (GDS-30; Yesavage et al., 1983). Subjects scoring outside Portuguese cut-off scores on the MMSE (possible cognitive decline if MMSE ≤22 for subjects from 1 to 11 years of education and ≤27 for those with scholarship >11 years; Guerreiro, 1998; Guerreiro, Silva, & Botelho, 1994), on MoCA (Freitas, Simões, Alves, & Santana, 2011), and those who scored more than 20 on the GDS-30 (Yesavage et al., 1983) were excluded from further study. These procedures yielded a final sample group of 124 subjects. To select the items displaying the highest correlation with IQ, the correlation (Pearson's correlation) between reading scores was established (correct = 1, incorrect = 0) for each of the 105 Portuguese words. Portuguese words exhibiting the highest correlation with IQ were then selected to be included in the final version of TeLPI. The damaging effect of words on TeLPI's internal consistency was also taken into account in the final word selection.

In order to determine an empirical law, the estimated FSIQ, VIQ, and PIQ were regressed, based on the number of errors on the TeLPI and significant demographic variables. For test–retest reliability purposes, a sample subgroup of 60 subjects was re-examined with the same instruments used in the initial examination.

Results

After all exclusion criteria were applied, the validation sample included 124 subjects, 64 men (51.6%) and 60 women (48.4%), with an average age of 48.20 years (SD = 18.71; min = 16; max = 86). The average of years of education was 10.31 (SD = 4.375; min = 4; max = 20). Note that the educational level of the Portuguese population is typically low (mean education of the Portuguese population = 8.16; SD = 4.72; INE, 2011), given that 27.2% have ≤4 years of education. The mean MMSE score was 29.04 (SD = 1.185), ranging from 25 to 30, and the mean MoCA score was 26.53 (SD = 2.923), ranging from 15 to 30. The mean FSIQ was 109 (SD = 17.94; min = 66, max = 146), the mean VIQ was 110 (SD = 17.98; min = 63, max = 147), and the mean PIQ was 107 (SD = 16.69; min = 67, max = 149). The mean errors given on TeLPI (105 words) were 15.40 (SD = 11.60; min = 1, max = 62; Table 1). As observed in Table 1, descriptive statistics of the sample ≥25 years of age are also presented. Demographic statistics do not differ in great extent when comparing the total sample (≥16 years of age) to the ≥24 years of age subgroup sample. These data are relevant to the full understanding of the final TeLPI equations (see Results and Discussion sections). Descriptive statistics of the sample by age group are presented in Table 2.

Table 1.

Descriptive statistics of the sample

  Sample group
 
≥16 years of age ≥25 years of age 
N 124 105 
Age 
 Mean 48.20 53.29 
 Minimum–maximum 16–86 25–86 
 Standard deviation 18.71 15.57 
Years of schooling 
 Mean 10.31 10.10 
 Minimum–maximum 4–20 4–20 
 Standard deviation 4.37 4.60 
MMSE 
 Mean 29.04 28.97 
 Minimum–maximum 25–30 25–30 
 Standard deviation 1.18 1.19 
MoCA 
 Mean 26.53 26.14 
 Minimum–maximum 15–30 15–30 
 Standard deviation 2.92 2.95 
FSIQ 
 Mean 109 109 
 Minimum–maximum 66–146 66–146 
 Standard deviation 17.93 18.79 
VIQ 
 Mean 110 110 
 Minimum–maximum 63–147 63–147 
 Standard deviation 17.98 18.64 
PIQ 
 Mean 107 107 
 Minimum–maximum 67–149 67–149 
 Standard deviation 16.68 17.27 
TeLPI number of errors (105 words) 
 Mean 15.40 15.08 
 Minimum–maximum 1–62 1–58 
 Standard deviation 11.59 12.11 
TeLPI number of errors (46 words) 
 Mean 6.73 
 Minimum–maximum 0–37 0–37 
 Standard deviation 7.85 8.32 
  Sample group
 
≥16 years of age ≥25 years of age 
N 124 105 
Age 
 Mean 48.20 53.29 
 Minimum–maximum 16–86 25–86 
 Standard deviation 18.71 15.57 
Years of schooling 
 Mean 10.31 10.10 
 Minimum–maximum 4–20 4–20 
 Standard deviation 4.37 4.60 
MMSE 
 Mean 29.04 28.97 
 Minimum–maximum 25–30 25–30 
 Standard deviation 1.18 1.19 
MoCA 
 Mean 26.53 26.14 
 Minimum–maximum 15–30 15–30 
 Standard deviation 2.92 2.95 
FSIQ 
 Mean 109 109 
 Minimum–maximum 66–146 66–146 
 Standard deviation 17.93 18.79 
VIQ 
 Mean 110 110 
 Minimum–maximum 63–147 63–147 
 Standard deviation 17.98 18.64 
PIQ 
 Mean 107 107 
 Minimum–maximum 67–149 67–149 
 Standard deviation 16.68 17.27 
TeLPI number of errors (105 words) 
 Mean 15.40 15.08 
 Minimum–maximum 1–62 1–58 
 Standard deviation 11.59 12.11 
TeLPI number of errors (46 words) 
 Mean 6.73 
 Minimum–maximum 0–37 0–37 
 Standard deviation 7.85 8.32 

Notes: MMSE = Mini-Mental State Examination; MoCA = Montreal Cognitive Assessment; FSIQ = Full-Scale IQ; VIQ = Verbal IQ; PIQ = Performance IQ; TeLPI = Portuguese Irregular Word Reading Test.

Table 2.

Descriptive statistics of the sample by age group

  Age group
 
16–24 25–44 45–54 55–64 ≥65 Total 
N 19 29 26 25 25 124 
Years of schooling 
 Mean 11.53 11.34 9.73 9.48 9.64 10.31 
 Minimum 
 Maximum 17 20 17 17 18 20 
 Standard deviation 2.50 5.23 4.33 4.20 4.49 4.37 
FSIQ 
 Mean 110.37 107.14 110.31 104.72 115.92 109 
 Minimum 68 66 74 66 72 66 
 Maximum 129 146 143 146 145 146 
 Standard deviation 12.52 22.59 17.71 16.36 16.30 17.93 
VIQ 
 Mean 108.58 106.45 109.53 105.36 120.80 110 
 Minimum 69 63 73 70 83 63 
 Maximum 124 147 134 147 147 147 
 Standard deviation 14.10 21.97 17.24 14.84 15.93 17.98 
PIQ 
 Mean 110.95 106.93 109.73 102.68 106.88 107 
 Minimum 76 69 72 67 69 67 
 Maximum 132 137 149 136 143 149 
 Standard deviation 12.70 19.96 16.41 17.02 12.21 16.68 
  Age group
 
16–24 25–44 45–54 55–64 ≥65 Total 
N 19 29 26 25 25 124 
Years of schooling 
 Mean 11.53 11.34 9.73 9.48 9.64 10.31 
 Minimum 
 Maximum 17 20 17 17 18 20 
 Standard deviation 2.50 5.23 4.33 4.20 4.49 4.37 
FSIQ 
 Mean 110.37 107.14 110.31 104.72 115.92 109 
 Minimum 68 66 74 66 72 66 
 Maximum 129 146 143 146 145 146 
 Standard deviation 12.52 22.59 17.71 16.36 16.30 17.93 
VIQ 
 Mean 108.58 106.45 109.53 105.36 120.80 110 
 Minimum 69 63 73 70 83 63 
 Maximum 124 147 134 147 147 147 
 Standard deviation 14.10 21.97 17.24 14.84 15.93 17.98 
PIQ 
 Mean 110.95 106.93 109.73 102.68 106.88 107 
 Minimum 76 69 72 67 69 67 
 Maximum 132 137 149 136 143 149 
 Standard deviation 12.70 19.96 16.41 17.02 12.21 16.68 

Notes: FSIQ = Full-Scale IQ; VIQ = Verbal IQ; PIQ = Performance IQ.

In the sample of 124 healthy subjects, TeLPI scores exhibited high and significant correlations with FSIQ, r(122) = .753, p < .001, and VIQ, r(122) = .732, p < .001. Correlations with PIQ were also significant, although proving to be lower—r(122) = .655, p < .001. The highest WAIS-III subtest correlation occurred on Vocabulary, r(122) = .787, p < .001, and Information scores, r(122) = .716, p < .001. The correlations between the TeLPI and MoCA scores were significant and moderate, r(122) = .532, p < .001, and between the TeLPI and MMSE results, even if somewhat lower, were also significant, r(122) = .415, p < .001. The correlation between MMSE and MoCA, r(122) = .627, p < .001, is moderate and significant and correlations between MMSE and WAIS-III FSIQ, r(122) = .493, p < .001, and between MoCA and WAIS-III FSIQ, r(122) = .413, p < .001, though both significant are considered to be weak.

TeLPI items exhibiting a correlation with the total test score or with the WAIS-III scores <.400 were excluded. Also excluded were five words that had no discriminative power (zero-variance: words that all participants read correctly or incorrectly) and two words that had a damaging effect on the internal consistency of TeLPI (Cronbach's α difference of 0.150). Therefore, 59 words were eliminated leaving the final version of the TeLPI with 46 items. After the removal of these items, TeLPI maintained similar correlations (Table 3). The remaining 46 words were rearranged in ascending order of difficulty. The easiest word was correctly pronounced by 97.6% of the sample and the most difficult by 46.8%. The internal consistency (Cronbach's α) of the selected 46 items is 0.939, and thus considered excellent. Table 4 presents the reliability coefficients by age in the sample studied. The younger age group has a relatively weaker consistency that could be influenced by years of schooling (Table 2) but no significant differences were found between age groups, F(4, 119) = 1.26, p > .05, and therefore, this result could be related to TeLPI's specificity in assessing crystallized intelligence, which is largely maintained or even improved into old age (Hertzog, 2011).

Table 3.

Correlations of the TeLPI with external criteria (WAIS-III, MMSE, MoCA, and education)

 TeLPI (105 words) TeLPI (46 words) 
WAIS-III scores 
 FSIQ .753 .729 
 VIQ .732 .695 
 PIQ .655 .651 
 Vocabulary .787 .761 
 Information .716 .689 
 Comprehension .543 .513 
 Similarities .658 .624 
 Picture completion .472 .493 
 Block design .569 .490 
 Matrix reasoning .548 .540 
 Symbol search .472 .490 
 Arithmetic .482 .451 
MMSE .415 .431 
MoCA .532 .552 
Years of education .637 .600 
 TeLPI (105 words) TeLPI (46 words) 
WAIS-III scores 
 FSIQ .753 .729 
 VIQ .732 .695 
 PIQ .655 .651 
 Vocabulary .787 .761 
 Information .716 .689 
 Comprehension .543 .513 
 Similarities .658 .624 
 Picture completion .472 .493 
 Block design .569 .490 
 Matrix reasoning .548 .540 
 Symbol search .472 .490 
 Arithmetic .482 .451 
MMSE .415 .431 
MoCA .532 .552 
Years of education .637 .600 

Notes: WAIS = Wechsler Adult Intelligence Scale; FSIQ = Full-Scale IQ; VIQ = Verbal IQ; PIQ = Performance IQ; MMSE = Mini-Mental State Examination; MoCA = Montreal Cognitive Assessment. Correlation is significant at the .01 level (two-tailed).

Table 4.

Internal consistency reliability coefficients for the Portuguese validation sample by age

Age group Cronbach's α 
16–24 0.795 
25–44 0.954 
45–54 0.945 
55–64 0.901 
≥65 0.962 
Total 0.939 
Age group Cronbach's α 
16–24 0.795 
25–44 0.954 
45–54 0.945 
55–64 0.901 
≥65 0.962 
Total 0.939 

TeLPI test–retest reliability was examined on a sample of 60 subjects divided into two groups. The test–retest reliability of group 1 (N = 30), with a delay of approximately 4 months and an average of 143.57 days (SD = 42.51; min = 68; max = 200), was 0.95—r(28) = .95, p < .001. The second group (N = 30) was tested with a delay of 18 months and an average of 538.85 days (SD = 110.19; min = 201; max = 676) and presented a test–retest reliability of 0.98—r(28) = .98, p < .001.

In the TeLPI's final version (46 words), no significant gender effects were found, t(124) = 1.670, and age was also not significantly correlated with TeLPI scores, r(122) = .091. Consistent with these data are correlations between age and FSIQ, r(122) = .024, VIQ, r(122) = .145, and PIQ, r(122) = .146, that were also not significant for the 124 subjects of the sample. Years of education correlated significantly with performance on TeLPI, r(122) = .600, p < .001, as well as with FSIQ, r(122) = .661, p < .001, VIQ, r(122) = .662, p < .001, and PIQ, r(122) = .546, p < .001. As an additional validity test, a significant linear regression was obtained entering FSIQ scores, years of education, and age as predictors for TeLPI scores. FSIQ was a strong predictor, β = −0.729, t(121) = −11.75, p < .001, explaining, by itself, 52.7% of the variance in the TeLPI scores, adjusted R2 = .527, F(1, 122) = 138.22, p < .001, whereas years of education appeared to be a weaker predictor, β = −0.210, t(121) = −2.61, p < .001, accounting for only an additional 2% of this variance, adjusted R2= .549, F(2, 121) = 75.78, p < .001. As such, age did not result as a significant predictor—β = 0.064, t(120) = 1.00, p > .05.

To estimate the equivalent intellectual level measured by the WAIS-III, three significant stepwise linear regression equations were obtained from the sample of 124 healthy subjects with the two significant predictors found (TeLPI scores and years of education). Using FSIQ as a dependent variable, the regression equations revealed a significant model, β = 0.350, t(121) = 4.927, p < .001, predicting 60.3% of the variance of FSIQ, adjusted R2 = .603, F(2, 121) = 94.44, p< .001; 57% of VIQ, β = 0.383, t(121) = 5.19, p < .001 and adjusted R2 = .570, F(2, 121) = 82.49, p< .001; and 45.3% of PIQ, β = 0.243, t(121) = 2.92, p < .01 and adjusted R2 = .453, F(2, 121) = 51.98, p< .001. The equations are the following: Since the TeLPI is assumed to assess crystallized intelligence (presumably stable in adulthood), and given the first results and the lower values of Cronbach's α (0.795) in ages ranging from 16 to 24, we tried to enhance the percentage of explained variance by excluding subjects under 25 years of age from the sample, then computing new correlations and linear regressions with the remaining 105 subjects. Descriptive statistics of the sample ≥25 years of age is presented in Table 1. The regression equations for subjects 25 years and older increases explained variance from 60.3% to 63% of the FSIQ, β = 0.391, t(102) = 5.26, p < .001 and adjusted R2 = .633, F(2, 102) = 90.50, p< .001, from 57% to 62.3% of the VIQ, β = 0.434, t(102) = 5.77 p < .001 and adjusted R2 = .623, F(2, 102) = 87.05, p< .001, and from 45.3% to 47.2% of the PIQ, β = 0.272, t(102) = 3.02, p < .01 and adjusted R2 = .472, F(2, 102) = 47.54, p< .001, using TeLPI scores and years of education (Table 5) as predictors. These results are consistent with the fact that crystallized intelligence is reasonably stable in adulthood and after the completion of basic and formal school education (Hertzog, 2011). The regression equations are as follows:

  • TeLPI predicted WAIS-III FSIQ = 100.645 + (−1.165 × number of errors on the TeLPI) + (1.604 × number of school years completed);

  • TeLPI predicted WAIS-III VIQ = 100.987 + (−1.064 × number of errors on the TeLPI) + (1.576 × number of school years completed);

  • TeLPI predicted WAIS-III FSIQ = 104.945 + (−1.075 × number of errors on the TeLPI) + (0.926 × number of school years completed).

  • TeLPI predicted WAIS-III FSIQ = 102.046 + (−1.153 × number of errors on the TeLPI) + (1.534 × number of school years completed);

  • TeLPI predicted WAIS-III VIQ = 99.872 + (−1.017 × number of errors on the TeLPI) + (1.755 × number of school years completed);

  • TeLPI predicted WAIS-III FSIQ = 103.644 + (−1.031 × number of errors on the TeLPI) + (1.019 × number of school years completed).

Table 5.

Different regression equation possibilities for TeLPI

  Inserted variables N Explained variance
 
FSIQ (%) VIQ (%) PIQ (%) 
TeLPI (number of errors) 124 52.7 47.9 42 
TeLPI (number of errors) + years at school 124 60.3 57 45.3 
TeLPI (number of errors) 105 (≥25 years) 54 51 43 
TeLPI (number of errors) + years at school 105 (≥25 years) 63 62.3 47.2 
  Inserted variables N Explained variance
 
FSIQ (%) VIQ (%) PIQ (%) 
TeLPI (number of errors) 124 52.7 47.9 42 
TeLPI (number of errors) + years at school 124 60.3 57 45.3 
TeLPI (number of errors) 105 (≥25 years) 54 51 43 
TeLPI (number of errors) + years at school 105 (≥25 years) 63 62.3 47.2 

These models showed standard errors of the estimate (SEest) of 11.39 points for TeLPI predictors of FSIQ, 11.44 for VIQ, and 12.55 for PIQ. Using these equations, the predicted IQ for each individual ≥25 years of age was determined (n = 105). As presented in Table 6, the correlations (Pearson's r) between predicted and observed FSIQ, r(103) = .80, p< .001, VIQ, r(103) = .79, p< .001, and PIQ, r(103) = .69, p< .001, for subjects in this group were high and significant. The Pearson correlation between the predicted and the actual IQ scores ranged from .69 to .80, reflecting minimal “shrinkage” of predictive accuracy. The TeLPI scores predicted FSIQ scores and these were also significantly correlated with all of the nine subtests of the WAIS-III used. The paired-samples t-test revealed that estimates of all three pairs of predicted and observed FSIQ, t(104) = 0.089, p > .05, VIQ, t(104) = 0.153, p > .05, and PIQ, t(104) = −0.005, p > .05, scores based on TeLPI and years of education equations were not significantly different. These results demonstrate that predictive models based on the sample have minimal loss of fidelity.

Table 6.

Correlations between TeLPI predicted IQ and observed IQ in subjects ≥25 years of age (n = 105)

 TeLPI (46 words) 
Observed WAIS-III IQ's scores 
 FSIQ .800 
 VIQ .792 
 PIQ .693 
Observed WAIS-III subtests scores 
 Vocabulary .815 
 Information .743 
 Comprehension .642 
 Similarities .766 
 Picture completion .509 
 Block design .610 
 Matrix reasoning .611 
 Symbol search .509 
 Arithmetic .495 
 TeLPI (46 words) 
Observed WAIS-III IQ's scores 
 FSIQ .800 
 VIQ .792 
 PIQ .693 
Observed WAIS-III subtests scores 
 Vocabulary .815 
 Information .743 
 Comprehension .642 
 Similarities .766 
 Picture completion .509 
 Block design .610 
 Matrix reasoning .611 
 Symbol search .509 
 Arithmetic .495 

Notes: WAIS = Wechsler Adult Intelligence Scale; FSIQ = Full-Scale IQ; VIQ = Verbal IQ; PIQ = Performance IQ. Correlation is significant at the .01 level (two-tailed).

To examine the TeLPI's predicted accuracy in detail, individual estimates were examined (Table 7): the difference between actual WAIS-III FSIQ and TeLPI predicted FSIQ score is 0.097 points. 85% of the sample TeLPI predicted FSIQ fell within 1 SEest. The percentage of cases in which predicted TeLPI FSIQ fell within ±5, ±10, ±15, and ±20 points of their actual FSIQ, as well as difference in WAIS-III category classification (ranging from “extremely low” to “very superior”) between TeLPI FSIQ estimates and real FSIQ are also presented in Table 7. No wrong classifications were predicted by TeLPI in more than 20 points (or two descriptive categories of WAIS-III). Most estimating errors across IQ categories pertain to individuals with FSIQ above 120 and fewer to individuals with FSIQ under 89. TeLPI correctly accounted for 71% of the subjects' IQ within ±5 points of their actual FSIQ, 85% within ±10 points, 91% within ±15 points, and all of the subjects' IQ within ±20 points.

Table 7.

Descriptive statistics of differences between predicted and actual WAIS-III FSIQ scores (N = 105)

 Actual IQ (mean [SD]) Min/max Actual FSIQ Predicted IQ (mean [SD]) Min/Max Pred. FSIQ Mean differencea (SDPercent within ±5 points Percent within ±10 points Percent within ±15 points Percent within ±20 points Percent within the same categoryb Percent within previous/following categoryb Percent within two categoriesb under or above 
TeLPI 109.44 (18.793) 66/146 109.34 (14.948) 66/130 0.097 (11.284) 71 85 91 100 40 71 100 
Extremely low 67.60 (1.517) 66/69 76.13 (7.709) 67/84 −8.528 (6.291) 100 100 100 100 20 60 100 
Borderline 74 (2.000) 72/76 78.59 (8.956) 71/89 −4.588 (7.043) 100 100 100 100 67 100 100 
Low average 85.75 (3.012) 81/89 95.64 (14.991) 66/115 −9.888 (13.182) 88 88 88 100 63 100 
Average 101.83 (6.061) 91/109 106.10 (10.975) 75/129 −4.273 (10.429) 89 91 94 100 46 91 100 
High average 114.45 (2.686) 110/119 114.92 (6.668) 97/127 −0.467 (6.098) 86 95 100 100 64 100 100 
Superior 123.94 (2.461) 120/128 119.34 (6.852) 100/128 4.60278 (6.586) 56 89 100 100 44 94 100 
Very superior 138 (5.189) 132/146 122.09 (5.927) 110/130 15.908 (7.728) 36 36 100 57 100 
 Actual IQ (mean [SD]) Min/max Actual FSIQ Predicted IQ (mean [SD]) Min/Max Pred. FSIQ Mean differencea (SDPercent within ±5 points Percent within ±10 points Percent within ±15 points Percent within ±20 points Percent within the same categoryb Percent within previous/following categoryb Percent within two categoriesb under or above 
TeLPI 109.44 (18.793) 66/146 109.34 (14.948) 66/130 0.097 (11.284) 71 85 91 100 40 71 100 
Extremely low 67.60 (1.517) 66/69 76.13 (7.709) 67/84 −8.528 (6.291) 100 100 100 100 20 60 100 
Borderline 74 (2.000) 72/76 78.59 (8.956) 71/89 −4.588 (7.043) 100 100 100 100 67 100 100 
Low average 85.75 (3.012) 81/89 95.64 (14.991) 66/115 −9.888 (13.182) 88 88 88 100 63 100 
Average 101.83 (6.061) 91/109 106.10 (10.975) 75/129 −4.273 (10.429) 89 91 94 100 46 91 100 
High average 114.45 (2.686) 110/119 114.92 (6.668) 97/127 −0.467 (6.098) 86 95 100 100 64 100 100 
Superior 123.94 (2.461) 120/128 119.34 (6.852) 100/128 4.60278 (6.586) 56 89 100 100 44 94 100 
Very superior 138 (5.189) 132/146 122.09 (5.927) 110/130 15.908 (7.728) 36 36 100 57 100 

Notes: WAIS = Wechsler Adult Intelligence Scale; FSIQ = Full-Scale IQ.

aDifference = TeLPI predicted FSIQ − actual FSIQ.

bCategory = descriptive IQ category defined in the WAIS-III: ≤69 = extremely low, 70–79 = inferior, 80–89 = low average, 90–109 = average, 120–129 = superior, ≥130 = very superior.

Discussion

This study presents a Portuguese word reading test (TeLPI) specially designed to assess premorbid intelligence. An experimental version of TeLPI with 105 words was initially constructed but the list was then shortened to 46 items, corresponding to the words that most correlate with FSIQ (WAIS-III).

TeLPI final version has showed to have an excellent internal consistency, in line with other reading tests (e.g., HART, WTAR) that range from 0.80 to 0.97 (Schretlen et al., 2009; The Psychological Corporation, 2001). TeLPI scores seem to be stable since its test–retest reliability is high in both 4 and 18 months delay groups. Similar tests have also reported comparable test–retest reliabilities, such as the NART-R (0.98; Crawford, Parker, Stewart, Besson, & De Lacey, 1989; O'Carroll, 1987) and the WTAR (ranging from 0.90 to 0.94; The Psychological Corporation, 2001). The average performance on the TeLPI appears to be stable over time, indicating that previous exposure does not improve performance.

TeLPI also presents high correlations with FSIQ congruent with previous studies with other similar tests (e.g., NART, WAT, JART) that typically report moderate to high correlations from .40 to .88 (Del Ser et al., 1997; Matsuoka et al., 2006; Strauss, Sherman, & Spreen, 2006). Although correlations between WAIS-III FSIQ and TeLPI are considered high and those between MMSE and MoCA are, as expected, moderate, the weak correlations found between TeLPI and MMSE, on one hand, and TeLPI and MoCA, on the other, are worthy of some observations. The weak correlation between MMSE and WAIS-III FSIQ and between MoCA and WAIS-III FSIQ can give some insight into these results. In fact, while MMSE and MoCA, as brief cognitive screening tests, assess mental status of patients, WAIS-III and TeLPI (in healthy subjects) both assess intelligence. Note that correlation between MoCA and MMSE is n't higher since MoCA assesses different and more complex cognitive domains than MMSE, such as executive functions, visuospatial abilities, language, attention, concentration, and working memory (Freitas et al., 2011; Nasreddine et al., 2005).

Three regression equations were presented that can be used for accurate premorbid intelligence estimation. In this study, combining TeLPI performance with demographic information accounted for significantly more variance in FSIQ, VIQ, and PIQ than performance on TeLPI alone, especially in healthy subjects 25 years of age or higher. Nevertheless, a set of regression equations for subjects ≥16 years of age were also presented given their potential usefulness in clinical settings.

Although other studies involving reading tests have showed a significant improvement of premorbid IQ estimation when demographic variables such as race, age, or gender are taken into consideration in the regression formulas (e.g., Rolstad et al., 2008; Schretlen et al., 2009), only the variable years of education has showed to be significant in the models derived from the TeLPI validation sample. Race was not considered in the present study due to the lack of variability in the sample (100% of the validation sample was Caucasian) and probably because WAIS-III FSIQ is age corrected, the amount of variance age contributed to was n't significant. The relationship between age, education, and IQ has been markedly studied (Schoenberg et al., 2011) but this relation was n't found in the current analysis, as it has been in studies involving other instruments (e.g., The Psychological Corporation, 2001). Significant demographic variables (years of education, in the case of TeLPI) explained an additional 4.2%–11.3% of variance in IQ scores, beyond that explained by TeLPI performance alone totalizing 63% of explained variance. These results are in accordance with NART results (Crawford et al., 1990; Mathias, Bowen, & Barret-Woodbridge, 2007) and with those obtained by other international reading tests (e.g., Rolstad et al., 2008; Schretlen et al., 2009) and are in opposition to the Blair and Spreen (1989) and Bright, Jaldow, and Kopelman (2002) findings, reporting the lack of significant improvement of the accuracy of IQ predictors in adding demographic variables to regression equations.

Our data show that TeLPI scores are a reliable measure for the estimation of FSIQ and VIQ, but are poorer in predicting PIQ. This finding is similar to those seen in previous studies with the NART (Nelson & Wilson, 1991) and its various international adaptations (e.g., Matsuoka et al., 2006). Nevertheless, all three regression equations represent good estimation measures.

In comparison with other reading tests used worldwide, one of the advantages of the TeLPI is related to floor and ceiling effects that constrain the range of IQ scores predicted. Table 7 displays the theoretical ranges of the TeLPI regression formulas. However, when utilizing the TeLPI and other reading tests, some caution must be taken, since the use of regression procedures is limited in terms of range of predicted scores. Whereas the equations derived for other reading test such as the NART-R (Blair & Spreen, 1989) cannot predict IQ scores <80.2 and >120 or the NART-SWE (Rolstad et al., 2008), with a range of prediction of approximately 90–125, our equations can predict a wider range of premorbid IQ, ranging from 66 to 131. These results are similar to those found by Schretlen and colleagues (2009) with the HART. The range of the HART is 71–130 and, in comparison, the TeLPI is able to predict a slightly larger proportion in the lower IQ range, even if approximately the same proportion in the higher range. A similar tendency is observed in the 75.9–124.1 range of the JART (Matsuoka et al., 2006). In the original NART study (Nelson, 1982), the regression equations yielded a possible predicted FSIQ range from 68.6 to 130.6, results that are very similar to our own findings. Schoenberg and colleagues (2002) also reported ranges in the OPIE-3 between 50.7 and 131.6 that vary with the algorithm used. Note that the most extreme scores by healthy adults using TeLPI formulas were 66–130 for FSIQ, 69–131 for VIQ, and 70–122 for PIQ, and so predictions outside these ranges are not empirically justified by the current study. A further limitation of this study is that, in the sample collected, the WAIS-III FSIQ mean is nearly a standard deviation above the mean (Table 1) which can suggest that the TeLPI could be less effective in predicting premorbid IQ for subjects with low IQ.

Data presented in Table 7 reveal that, overall, the proportion of individuals classified within ±5 points of their actual IQ by TeLPI FSIQ estimates is considered good, the exception being IQ >120 that is drastically enhanced in the superior range (120–129) when a difference of ±10 points is considered. Similarly, these results have also been found with other instruments of premorbid estimation, such as the OPIE-3 (Schoenberg et al., 2002), with less accurate estimations in the superior ranges of IQ. TeLPI FSIQ estimates within ±10 points of the actual IQ (85%) were similar to the ones found with OPIE-3 (75%–93%) and greater than those predicted by WTAR (70.4%) or the combined WTAR-demographics approach (73.4%; The Psychological Corporation, 2001).

A more thorough analysis on the TeLPI FSIQ estimates reveals that in the superior ranges (including “superior” and “very superior”) none of the subjects' IQ was overestimated but in nine cases IQ was underestimated. This tendency is not seen in the other ranges where general overestimation of IQ is observed. Psychometric restrictions of this nature are known to influence premorbid intelligence instruments that use regression formulas, particularly in cases of extreme scores (Schoenberg et al., 2011; Veil & Koopman, 2001).

One possible criticism to TeLPI estimations is that the SEest associated with IQ predictions were higher than similar tests reflecting less accurate predictions than those reported by Blair and Spreen (1989) for the NART-R or the OPIE-3 (Schoenberg et al., 2002). High SEest have also been reported by Schretlen and colleagues (2009) regarding the HART, which includes NART-R items in its final version. Also note that the sample IQ scores were prorated using nine subtests of WAIS-III rather than the full WAIS-III (12 subtests). Although VIQ and PIQ are very similar and trustworthy, even when the available prorating option is used for calculating the IQs by five verbal and four performance subtests for the IQ estimations, the total measurement error could be greater than reported and should be considered as another limitation of the present study.

Evaluating premorbid IQ is an important step in neuropsychological assessment and has obvious potential in clinical settings. TeLPI predicted FSIQ is likely to be a useful method for estimating premorbid IQ, providing a measure against which a patient's current performance can be compared. Although other data sources, particularly academic records, may offer additional information from which premorbid cognitive functioning can be inferred (Baade & Schoenberg, 2004), premorbid IQ estimation instruments offer enhanced reliability in the diagnosis of cognitive deterioration. The TeLPI is easy to apply, short, well tolerated, exhibits excellent concurrent validity, and is, overall, valid for premorbid intelligence estimation in a normal population, filling an important gap in the neuropsychological evaluation of adult Portuguese speakers aged 25–86. Developing research involving the use of the TeLPI with clinical samples (including Mild Cognitive Impairment and Alzheimer's disease) will further confirm the validity of TeLPI's regression formulas in cognitive decline samples. A normative study is also being currently carried out as to allow for test performance interpretation in comparison with norms regarding a reference group that is representative of the Portuguese population (Alves, Simões, Martins, Freitas, & Santana, 2011). Independent validation of the prediction equations is also an essential issue to be addressed in the future.

Funding

This research was supported by the Portuguese Foundation for Science and Technology (SFRH/BD/37748/2007).

Conflict of Interest

None declared.

References

Alves
L.
Simões
M. R.
Martins
C.
Avaliação da Inteligência Pré-Mórbida: Desenvolvimento da versão experimental do Teste de Leitura de Palavras Irregulares (TeLPI) para a População Portuguesa [Premorbid Intelligence Assessment: Experimental version of the Irregular Word Reading Test (TeLPI)]
Psychologica
 , 
2010
, vol. 
52
 
3
(pg. 
295
-
311
)
Alves
L.
Simões
M. R.
Martins
C.
Freitas
S.
Santana
I.
Teste de Leitura de Palavras Irregulares (TeLPI): Estudo de validação clínica em sujeitos com declínio cognitivo [The Irregular Word Reading Test (TeLPI): Validation study in cognitive decline subjects]
 , 
2011
XV Conferência Internacional Avaliação Psicológica: Formas e Contexto [XV International Conference of Psychological Assessment], Lisboa, Portugal
American Psychological Association
Guidelines for the evaluation of dementia and age-related cognitive decline
American Psychologist
 , 
1998
, vol. 
53
 
12
(pg. 
1298
-
1303
)
Baade
L. E.
Schoenberg
M. R.
A proposed method to estimate premorbid intelligence utilizing group achievement measures from school records
Archives of Clinical Neuropsychology
 , 
2004
, vol. 
19
 (pg. 
227
-
243
)
Barona
A.
Reynolds
C. R.
Chastain
R.
A demographically based index of premorbid intelligence for the WAIS-R
Journal of Consulting and Clinical Psychology
 , 
1984
, vol. 
52
 
5
(pg. 
885
-
887
)
Bayles
K. A.
Boone
D. R.
The potential of language tasks for identifying senile dementia
Journal of Speech and Hearing Disorders
 , 
1982
, vol. 
47
 (pg. 
210
-
214
)
Blair
J. R.
Spreen
O.
Predicting premorbid IQ: A revision of the National Adult Reading Test
The Clinical Neuropsychologist
 , 
1989
, vol. 
3
 (pg. 
885
-
887
)
Bright
P.
Jaldow
E.
Kopelman
M. D.
The National Adult Reading Test as a measure of premorbid intelligence: A comparison with estimates derived from demographic variables
Journal of the International Neuropsychological Society
 , 
2002
, vol. 
8
 
6
(pg. 
847
-
854
)
Centro de Linguística da Universidade de Lisboa
Corlex
 , 
2003
 
Retrieved February 10, 2008, from www.clul.ul.pt
Crawford
J. R.
Allan
K. M.
Estimating premorbid WAIS-R IQ with demographic variables: Regression equations derived from a UK sample
The Clinical Neuropsychologist
 , 
1997
, vol. 
11
 
2
(pg. 
192
-
197
)
Crawford
J. R.
Nelson
H. E.
Blackmore
L.
Cochrane
R. H.
Estimating premorbid intelligence by combining the NART and demographic variables: An examination of the NART standardization sample and supplementary equations
Personality and Individual Differences
 , 
1990
, vol. 
11
 
11
(pg. 
1153
-
1157
)
Crawford
J. R.
Parker
D. M.
Stewart
L. E.
Besson
J. A. O.
De Lacey
G.
Prediction of WAIS IQ with the National Adult Reading Test: Cross validation and extension
British Journal of Clinical Psychology
 , 
1989
, vol. 
28
 (pg. 
267
-
273
)
Del Ser
T.
González-Montalvo
J.
Martinez-Espinosa
S.
Delgado-Villapalos
C.
Bermejo
F.
Estimation of premorbid intelligence in Spanish people with the Word Accentuation Test and its application to the diagnosis of dementia
Brain and Cognition
 , 
1997
, vol. 
33
 (pg. 
343
-
356
)
Folstein
M. F.
Folstein
S. E.
McHugh
P. R.
“Mini-Mental State”. A practical method for grading the cognitive state of patients for the clinician
Journal of Psychiatric Research
 , 
1975
, vol. 
12
 (pg. 
189
-
198
)
Franzen
M. D.
Burgess
E. J.
Smith-Seemiller
L.
Methods of estimating premorbid functioning
Archives of Clinical Neuropsychology
 , 
1997
, vol. 
12
 
8
(pg. 
711
-
738
)
Freitas
S.
Simões
M. R.
Alves
L.
Santana
I.
Montreal Cognitive Assessment (MoCA): Normative study for the Portuguese population
Journal of Clinical and Experimental Neuropsychology
 , 
2011
, vol. 
33
 
9
(pg. 
989
-
996
)
Fuld
P. A.
Reisberg
B.
Psychometric differentiation of the dementias: An overview
Alzheimer's disease: The Stanford reference
 , 
1983
New York
Free Press
(pg. 
201
-
210
)
Grober
E.
Sliwinski
M.
Development and validation of a model for estimating premorbid verbal intelligence in the elderly
Journal of Clinical and Experimental Neuropsychology
 , 
1991
, vol. 
13
 (pg. 
933
-
949
)
Guerreiro
M.
Contributo da Neuropsicologia para o estudo das Demências. Dissertação de doutoramento [Contribution of Neuropsychology to the study of dementia]
 , 
1998
Unpublished doctoral dissertation, Faculdade de Medicina da Universidade de Lisboa
Guerreiro
M.
Silva
A. P.
Botelho
M. A.
Adaptação à população portuguesa da tradução do “Mini Mental State Examination” (MMSE) [Adaptation to the Portuguese population of the “Mini Mental State Examination” (MMSE)]
Revista Portuguesa de Neurologia
 , 
1994
, vol. 
1
 pg. 
9
 
Hertzog
C.
Sternberg
R. J.
Kaufman
S. B.
Intelligence in adulthood
The Cambridge handbook of intelligence
 , 
2011
New York
Cambridge University Press
(pg. 
174
-
190
)
Instituto Nacional de Estatística
Portal de Estatísticas Oficiais [Portuguese Official Statistics]
 , 
2011
 
Retrieved from http://www.ine.pt
Kareken
D. A.
Williams
J. M.
Human judgment and estimation of premorbid intellectual function
Psychological Assessment
 , 
1994
, vol. 
6
 
2
(pg. 
83
-
91
)
Lezak
M. D.
Howieson
D. B.
Loring
D. W.
Neuropsychological assessment
 , 
2004
4th ed.
Oxford
Oxford University Press
Mackinnon
A.
Ritchie
K.
Mulligan
R.
The measurement properties of a French language adaptation of the National Adult Reading Test
International Journal of Methods in Psychiatric Research
 , 
1999
, vol. 
8
 
1
(pg. 
27
-
38
)
Mathias
J. L.
Bowen
S. C.
Barret-Woodbridge
M.
Accuracy of the Wechsler Test of Adult Reading (WTAR) and National Adult Reading Test (NART) when estimating IQ in a healthy Australian sample
Australian Psychologist
 , 
2007
, vol. 
9
 
1
(pg. 
83
-
88
)
Matsuoka
K.
Masatake
U.
Kasal
K.
Koyama
K.
Kim
Y.
Estimation of premorbid IQ in individuals with Alzheimer's disease using Japanese ideographic script (Kanji) compound words: Japanese version of the National Adult Reading Test
Psychiatry and Clinical Neurosciences
 , 
2006
, vol. 
60
 (pg. 
332
-
339
)
McGurn
B.
Starr
J. M.
Topfer
J. A.
Pattie
A.
Whiteman
M. C.
Lemmon
H. A.
, et al.  . 
Pronunciation of irregular words is preserved in dementia, validating premorbid IQ estimation
Neurology
 , 
2004
, vol. 
62
 (pg. 
1184
-
1186
)
Nasreddine
Z.
Phillips
N. A.
Bédirian
V.
Charbonneau
S.
Whitehead
V.
Collin
I.
, et al.  . 
The Montreal Cognitive Assessment, MoCA: A brief screening tool for Mild Cognitive Impairment
American Geriatrics Society
 , 
2005
, vol. 
53
 (pg. 
695
-
699
)
NCS Pearson Corporation
Manual for the test of pre-morbid functioning (TOPF)
 , 
2009
San Antonio, TX
Author
NCS Pearson Corporation
Test of Premorbid Functioning—UK version
 , 
2011
Oxford
Author
Nelson
H. E.
National Adult Reading Test: Test manual
 , 
1982
Upton Park, UK
NFER-Windsor
Nelson
H. E.
Wilson
J.
National Adult Reading Test (NART): Test manual
 , 
1991
2nd ed.
London
NFER-Windsor
O'Carroll
R. E.
The inter-rater reliability of the National Adult Reading Test (NART): A pilot study
British Journal of Clinical Psychology
 , 
1987
, vol. 
26
 (pg. 
229
-
230
)
O'Carroll
R. E.
Gilleard
C. J.
Estimation of premorbid intelligence in dementia
British Journal of Clinical Psychology
 , 
1986
, vol. 
24
 (pg. 
157
-
158
)
Rolstad
S.
Nordlund
A.
Gustavsson
M. H.
Eckerstrom
C.
Klang
O.
Hansen
S.
, et al.  . 
The Swedish National Adult Reading Test (NART-SWE): A test of premorbid IQ
Scandinavian Journal of Psychology
 , 
2008
, vol. 
49
 (pg. 
577
-
582
)
Schoenberg
M. R.
Duff
K.
Dorfman
K. D.
Adams
R. L.
Differential estimation of verbal intelligence and performance intelligence scores from combined performance and demographic variables: The OPIE-3 verbal and performance algorithms
The Clinical Neuropsychologist
 , 
2004
, vol. 
18
 (pg. 
266
-
276
)
Schoenberg
M. R.
Duff
K.
Scott
J. G.
Adams
R. L.
An evaluation of the clinical utility of the OPIE-3 as an estimate of premorbid WAIS-III FSIQ
The Clinical Neuropsychologist
 , 
2003
, vol. 
17
 (pg. 
308
-
321
)
Schoenberg
M. R.
Lange
R. T.
Marsh
P.
Saklofske
D. H.
Kreutzer
J. S.
Deluca
J.
Caplan
B.
Premorbid intelligence
Encyclopedia of clinical neuropsychology
 , 
2011
New York
Springer Science
(pg. 
2004
-
2010
)
Schoenberg
M. R.
Scott
J. G.
Duff
K.
Adams
R. L.
Estimation of WAIS-III intelligence from combined performance and demographic variables: Development of the OPIE-3
The Clinical Neuropsychologist
 , 
2002
, vol. 
16
 (pg. 
426
-
438
)
Schretlen
D. J.
Winicki
J. M.
Meyer
S. M.
Testa
S. M.
Pearlson
G. D.
Gordon
B.
Development, psychometric properties, and validity of the Hopkins Adult Reading Test (HART)
The Clinical Neuropsychologist
 , 
2009
, vol. 
23
 (pg. 
926
-
943
)
Simões
M. R.
Freitas
S.
Santana
I.
Firmino
H.
Martins
C.
Nasreddine
Z.
, et al.  . 
Montreal Cognitive Assessment (MoCA): Versão final portuguesa [Montreal Cognitive Assessment (MoCA): Final portuguese version]
 , 
2008
Coimbra, Portugal
Serviço de Avaliação Psicológica, Faculdade de Psicologia e de Ciências da Educação da Universidade de Coimbra
Strauss
E.
Sherman
E. M. S.
Spreen
O.
A compendium of neuropsychological tests: Administration, norms, and commentary
 , 
2006
3rd ed.
New York
Oxford University Press
The International Phonetic Association
International phonetic alphabet
2005
 
The Psychological Corporation
Wechsler Test of Adult Reading (WTAR): Test manual
 , 
2001
San Antonio, TX
Author
Veil
H. O. F.
Koopman
R. F.
The bias in regression-based indices of premorbid IQ
Psychological Assessment
 , 
2001
, vol. 
13
 
3
(pg. 
356
-
368
)
Wechsler
D.
Escala de Inteligência de Wechsler para Adultos – Terceira edição (WAIS-III) [Wechsler Adult Intelligence Scale-Third Edition]
 , 
2008
Lisboa, Portugal
Cegoc
Yates
A.
The use of vocabulary in the measurement of intelligence deterioration: A review
Journal of Mental Science
 , 
1956
, vol. 
102
 (pg. 
409
-
440
)
Yesavage
J. A.
Brink
T. L.
Rose
T. L.
Lum
O.
Huang
V.
Adey
M. B.
, et al.  . 
Development and validation of a geriatric depression screening scale: A preliminary report
Journal of Psychiatry Research
 , 
1983
, vol. 
17
 (pg. 
37
-
49
)