Rapidly-administered short forms of the Wechsler Adult Intelligence Scale 3rd edition

Transcription

1 Archives of Clinical Neuropsychology 22 (2007) Abstract Rapidly-administered short forms of the Wechsler Adult Intelligence Scale 3rd edition Alison J. Donnell a, Neil Pliskin a, James Holdnack b, Bradley Axelrod c, Christopher Randolph d, a Department of Psychiatry, University of Illinois at Chicago, Chicago, IL, United States b Harcourt Assessment, Inc., San Antonio, TX, United States c Psychology Section, Detroit Veterans Administration Medical Center, Detroit, MI, United States d Department of Neurology, Loyola University Medical Center, Maywood, Illinois, United States Accepted 20 June 2007 Although the Wechsler Full Scale IQ (FSIQ) is a common component of most neuropsychological evaluations, there are many clinical situations where the complete administration of this battery is precluded by various constraints, including limitations of time and patient compliance. These constraints are particularly true for dementia evaluations involving elderly patients. The present study reports data on two short forms particularly suited to dementia evaluations, each requiring less than 20 min of administration time. One of the short forms was previously validated in dementia for the WAIS-R [Randolph, C., Mohr, E., & Chase, T. N. (1993). Assessment of intellectual function in dementing disorders: Validity of WAIS-R short forms for patients with Alzheimer s, Huntington s, and Parkinson s disease. Journal of Clinical and Experimental Neuropsychology, 15, ]; the second was developed specifically for patients with motor disabilities. These short forms were validated using the WAIS-III normative standardization sample (N = 2450), neurologic sample (N = 63), and matched controls (N = 49), and a separate mixed clinical sample (N = 70). The results suggest that each short form provides an accurate and reliable estimate of WAIS-III FSIQ, validating their use in appropriate clinical contexts. The present data support the use of these short forms for dementia evaluations, and suggests that they may be applicable for the evaluation of other neurological and neuropsychiatric disorders that involve acquired neurocognitive impairment National Academy of Neuropsychology. Published by Elsevier Ltd. All rights reserved. Keywords: Wechsler Adult Intelligence Scale; Dementia; Short form 1. Introduction Measurement of general intellectual ability, although essentially a composite score reflecting ability across multiple neurocognitive domains, remains a core component of psychological and neuropsychological evaluations. The Wechsler Adult Intelligence Scale (WAIS), in its various editions, has been the most popular measure for assessing intelligence in the United States (Harrison, Kaufman, Hickman, & Kaufman, 1988; Matarazzo & Herman, 1984; Rabin, Barr, & Burton, 2005). The most recent revision, Wechsler Adult Intelligence Scale Third Edition (WAIS- III), extends the upper age normative data assessing the intellectual abilities of adults ages 16 through 89 (Wechsler, Corresponding author. ABPP-CN, 1 East Erie, Suite 355, Chicago, IL 60611, United States. Tel.: ; fax: address: crandol@lumc.edu (C. Randolph) /$ see front matter 2007 National Academy of Neuropsychology. Published by Elsevier Ltd. All rights reserved. doi: /j.acn

2 918 A.J. Donnell et al. / Archives of Clinical Neuropsychology 22 (2007) ) and is considered more useful for an elderly population than past editions (Wymer, Rayls, & Wagner, 2003). However, several practical issues commonly interfere with administering the WAIS-III to an older population. First, patient endurance often dictates shorter exams, especially in a cognitively compromised elderly population where patients can become fatigued, frustrated, and agitated. The WAIS-III manual reports that the average administration time for the 11 WAIS-III subtests that yield the three IQ scores requires approximately 75 min (Wechsler, 1997), although administration time of the WAIS-III in clinical populations may be as long as 90 min (Ryan, Lopez, & Werth, 1998). Second, neuropsychologists are under increasing pressure to be time- and cost-efficient and with time constraints dictated by third party reimbursement for assessments, the full WAIS-III often becomes impractical to administer. Consequently, an abbreviated version of the WAIS-III that reduces the evaluation time without significantly impacting the reliability or validity of the Full Scale IQ (FSIQ) score for a compromised elderly population would be valuable. Shortened versions of intellectual tests in a dementia evaluation are useful for obtaining a quick measurement of FSIQ to compare to estimates of premorbid function for the purpose of gauging intellectual decline. In a situation where assessment is time limited, a briefer evaluation of intelligence in older adults would also allow more time for the assessment of other neuropsychological functions to take place. The use of shortened versions of Wechsler scales has been ongoing since 1956 (Doppelt, 1956). Although a number of short forms exist for the WAIS and several have been validated specifically for the WAIS-III, few have been objectively evaluated with an older cognitively compromised sample. One criterion for an acceptable abbreviated version is that the total testing time should be reduced by at least 50% (Levy, 1968). Resnick and Entin (1971) also proposed that an intelligence quotient (IQ) obtained from an abbreviated version should significantly correlate with the Full Scale IQ, and according to Nunnally (1978), have reliabilities equal to or higher than The four-subtest short form proposed by Kaufman (1990) for the WAIS-revised (WAIS-R) has one of the shortest administration times (less than 20 min) and correlated 0.96 with the FSIQ in the standardization sample of the WAIS-R. This short-form tetrad of subtests is composed of Arithmetic, Similarities, Picture Completion, and Digit Symbol-Coding. Because of its high reliability, short administration time, and practical utility in terms of the information obtained for a dementia evaluation (i.e., arithmetic computation, abstraction, attention, and processing speed), Randolph, Mohr, and Chase (1993) examined the psychometric properties of the Kaufman tetrad in a dementia sample and found that it actually underestimated the FSIQ in their sample. As such, Randolph et al. revised Kaufman s scaling table in order to correct for this underestimation. Results indicated that the revised scaling table produced a high correlation with the WAIS-R FSIQ (r = 0.96) and corrected for the underestimation. In fact, 72% of the entire sample was within five points of their actual IQ whereas only 52% were within this range using the original Kaufman scaling. The present study was undertaken in order to validate this revised version of the Kaufman short-form for the WAIS-III. Therefore, the goal of the present study was to examine the revised scaling of the short form Kaufman tetrad (SF1) for the WAIS-III in healthy and compromised older adult samples. Additionally, a second short form was validated that substituted Letter-Number Sequencing and Symbol Search for Arithmetic and Digit Symbol-Coding. The second short form (SF2) was developed to aid the clinician working with patients having significant motor problems or when working with subjects that have low educational attainment and subsequently may have poorly developed arithmetic skills. Additionally, this study examined SF1 estimated FSIQs as compared to the Wechsler Test of Adult Reading (WTAR: The Psychological Corporation, 2001) premorbid predicted FSIQ to examine the anticipated change in FSIQ with cognitively compromised samples. We chose to use the full WAIS-III normative sample to validate these short forms for normals across the full age range (16 89), as these short forms may also prove useful in neuropsychological evaluations of younger patients under certain circumstances. We hypothesized that the Randolph et al. revised scaling of SF1 would prove valid for the WAIS-III, and that SF2 would prove to be a reliable alternative. 2. Method 2.1. Participants Participants were members of the WAIS-III/WMS-III standardization sample (Wechsler, 1997), which contains both healthy controls (N = 2450) and specific clinical samples. A group of older healthy adults (N = 49) were extracted

3 A.J. Donnell et al. / Archives of Clinical Neuropsychology 22 (2007) Table 1 Demographic information for the matched control and clinical samples Matched controls (n = 49) Neurologic sample (n = 63) Mixed VA sample (n = 70) Age 63.0 (S.D. = 15.5) 64.1 (S.D. = 14.5) 53.7 (S.D. = 15.0) Gender Female Male Race/ethnicity Caucasian African American Hispanic Other Education 8 years years years years years from the standardization sample and were matched on age, education, gender, and ethnicity to the clinical groups identified from the standardization sample as the groups of interest. The clinical groups included those diagnosed with degenerative neurological disorders of Alzheimer s disease, Huntington s disease, and Parkinson s disease (neurologic sample). There were no significant differences between the neurologic sample and the matched control group on any demographic variables. A second sample of clinical cases from a veteran s medical center (mixed VA sample) was also included and comprised of heterogeneous neurological and psychiatric disorders, such as Alzheimer s, seizure, stroke, vascular dementia, Huntington s, depressive disorder, schizophrenia, alcohol abuse, bipolar, and personality disorders (N = 70). All participants in the WAIS-III/WMS-III standardization project underwent informed consent procedures. The mixed VA sample data were obtained during clinical evaluations predating HIPAA legislation; these WAIS-III data have been previously reported and were re-analysed for the present study. Table 1 provides the demographic data for matched control and clinical samples Procedure The WAIS-III and WTAR were administered according to the standardized procedures outlined in their respective manuals (Wechsler, 1997b, c; The Psychological Corporation, 2001). All of the subtests were administered in order. Therefore, all subjects completed the full version of the WAIS-III and not just the proposed short forms. All WTAR predictions of premorbid FSIQ used the WTAR + demographics tables in the WTAR manual. Pearson correlation coefficients were used to compare short form and full WAIS-III FSIQ scores. 3. Results 3.1. Healthy standardization sample SF1 Results indicated that in the normal standardization sample the SF1 estimated FSIQ correlated highly (r = 0.93) with the actual FSIQ. The SF1 estimated FSIQ and the actual FSIQ means for the standardization sample were both 100.3, resulting in no difference. Seventy-two percent of the standardization sample s SF1 estimated FSIQ was within 5 points of their actual FSIQ and 94% was within 10 points. It is important to note that for the healthy samples, it was necessary to adjust the sum of scaled scores by subtracting 2 points from the total of Similarities, Picture Completion, Arithmetic and Digit Symbol to estimate the SF1 FSIQ. The Appendix provides the table to convert the sum of scaled scores to the estimated SF1 FSIQ.

4 920 A.J. Donnell et al. / Archives of Clinical Neuropsychology 22 (2007) Table 2 Means and standard deviations of SF1, Actual FSIQ, and WTAR Difference SF1 FSIQ Actual FSIQ Difference WTAR-P SF1-WTAR-P Standardization sample * (14.5) (14.5) 0.0 (5.3) Matched controls * (16.3) (17.8) 0.4 (5.8) (12.3) Neurologic sample 86.1 (13.0) 86.2 (11.5) 0.1 (4.3) (10.5) 18.1 Mixed VA sample 83.0 (10.3) 83.3 (11.1) 0.3 (4.5) Note: SF1: Short-Form 1; FSIQ: Full Scale IQ; WTAR-P: Wechsler Test of Adult Reading predicted premorbid FSIQ. * In the standardization and matched control samples, it is necessary to adjust the sum of scaled scores by subtracting 2 points to the total of Similarities, Picture Completion, Arithmetic and Digit Symbol before looking up the FSIQ equivalent. Wechsler Adult Intelligence Scale-3rd edition Copyright 1997 by The Psychological Corporation. Reproduced by Permission. All Rights Reserved Clinical samples SF1 Results indicated that the SF1 estimated FSIQ correlated highly (r = 0.95) with the actual FSIQ for the neurologic sample. The SF1 estimated FSIQ average for the neurologic sample was 86.1 and the actual FSIQ average was 86.2, resulting in a negligible mean difference of 0.1. Eighty-three percent of the neurologic sample estimated IQ was within 5 points of their actual IQ and 97% was within 10 points of their actual IQ. The mean WTAR-predicted premorbid FSIQ for the neurologic sample was 104.2, resulting in a mean difference from their SF1 estimated IQ of 18.1 and reflected the expected drop from their premorbid IQ. The actual mean WAIS-III FSIQ for the subset of healthy normals matched to the neurologic sample on demographic variables was 108.4, and the SF1 estimated FSIQ for this group was The WRAT predicted FSIQ for this healthy subgroup was Therefore, the SF1 estimated FSIQ for the neurologic sample was a reliable estimate of actual current FSIQ, and both measures reflected a drop from premorbid FSIQ of approximately 18 points based upon WTAR prediction, and of approximately 22 points based upon comparison to demographically-matched controls. For the mixed VA sample, results indicated that the SF1 estimated FSIQ also correlated highly (r = 0.91) with the actual FSIQ. The SF1 estimated FSIQ average for the mixed VA sample was 83.3 and the actual FSIQ average was 83.0, resulting in a mean difference of only 0.3. Eighty percent of the mixed VA sample estimated IQ was within 5 points of their actual IQ and 97% was within 10 points. Tables 2 and 3 provide means, standard deviations, reliability, and percentages for SF Healthy samples Short Form 2 Results indicated that in the standardization sample the SF2 estimated FSIQ correlated highly (r = 0.89) with the actual FSIQ. The SF2 estimated FSIQ average was and actual FSIQ was 100.3, resulting in a mean difference of 0.2. Fifty-nine percent of the standardization sample s SF2 estimated FSIQ was within 5 points of their actual FSIQ and 87% was within 10 points. For the healthy samples, it was necessary to adjust the sum of scaled scores by subtracting 3 points from the total of Similarities, Picture Completion, Letter-Number Sequencing, and Symbol Search to estimate the SF2 estimated FSIQ. Appendix A provides the table to convert the sum of scaled scores to the estimated SF2 FSIQ. Table 3 SF1 compared to Actual FSIQ Correlation SF1 and FSIQ % within 5 pts or less % within 10 pts or less % within 15 pts or less Standardization sample Matched controls Neurologic sample Mixed VA sample Note: SF1: Short-Form 1; FSIQ: Full Scale IQ.

5 Table 4 Means and Standard Deviations of SF2, Actual FSIQ, and WTAR A.J. Donnell et al. / Archives of Clinical Neuropsychology 22 (2007) SF2 FSIQ difference Actual FSIQ Difference WTAR-P SF2-WTAR-P Standardization sample * (14.6) (14.5) 0.2 (7.1) Matched controls * (15.9) (17.8) 2.1 (7.5) (12.3) 1.2 Neurologic sample 86.0 (13.3) 86.2 (11.5) 0.2 (5.8) (10.5) 18.2 Mixed VA sample 86.7 (10.5) 83.3 (11.1) 3.4 (4.3) Note: SF2: Short-Form 2; FSIQ: Full Scale IQ; WTAR-P: Wechsler Test of Adult Reading predicted premorbid FSIQ. * In the standardization and matched control samples, it is necessary to adjust the sum of scaled scores by subtracting 3 points to the total of Similarities, Picture Completion, Letter-Number Sequencing and Symbol Search before looking up the FSIQ equivalent. Wechsler Adult Intelligence Scale-3rd edition Copyright 1997 by The Psychological Corporation. Reproduced by Permission. All Rights Reserved. Table 5 SF2 compared to actual FSIQ Correlation SF2 and FSIQ % within 5 pts or less % within 10 pts or less % within 15 pts or less Standardization sample Matched controls Neurologic sample Mixed VA sample Note: SF2: Short-Form 2; FSIQ: Full Scale IQ Clinical samples Short Form 2 Results indicated that for the SF2 estimated FSIQ correlated highly (r = 0.90) with the actual FSIQ for the neurologic sample. The SF2 estimated FSIQ average for the neurologic sample was 86.0, resulting in a mean difference of 0.2 from their actual FSIQ of Seventy percent of the neurologic sample estimated IQ was within 5 points of their actual IQ and 97% was within 10 points of their actual IQ. As stated previously, the mean WTAR- predicted premorbid FSIQ for the neurologic sample was 104.2, resulting in a mean difference from their SF2 estimated IQ of 18.2 and again reflected the expected drop from their premorbid IQ. The SF2 estimated FSIQ for the subset of healthy normals matched to the neurologic sample on demographic variables was 106.3, resulting in a mean difference of 2.1. For the mixed VA sample, results indicated that for the SF2 estimated FSIQ also correlated highly (r = 0.92) with the actual FSIQ. The SF2 estimated FSIQ average for the mixed VA sample was 86.7, resulting in a mean difference of 3.4 from their actual mean of Sixty-seven percent of the mixed VA sample estimated IQ was within 5 points of their actual IQ, and 97% was within 10 points. Tables 4 and 5 provide means, standard deviations, reliability, and percentages for SF2. 4. Discussion Neuropsychological assessments of older adults with suspected dementia or neurological compromise typically evaluate multiple cognitive domains. The evaluation of current intelligence is often important in that it can provide information about the extent of cognitive decline when compared to predicted premorbid performance. The WAIS-III FSIQ is a well-understood composite measure of general cognitive functioning and a common component of most neuropsychological and psychological evaluations. There are, however, many clinical situations where the complete administration of this battery is precluded by various constraints, including limitations of time and patient compliance. Therefore, the use of a valid, reliable, and quick evaluation of intelligence is beneficial in these circumstances. The short forms used in this study meet the acceptable criteria for abbreviated versions of measuring intelligence. Both short forms can be rapidly administered, in less than 20 min, which reduces the total testing time of intelligence by more than 75%. Additionally, both short forms are significantly correlated with the Full Scale IQ and have reliabilities generally above The SF1 appears to provide a slightly more accurate estimate of FSIQ and assesses clinical domains of interest (i.e., arithmetic computation, abstraction, attention, and processing speed), for a dementia evaluation. Therefore, the

6 922 A.J. Donnell et al. / Archives of Clinical Neuropsychology 22 (2007) use of SF2 is recommended only for those people with substantial motoric compromise while SF1 is recommended for most other clinical populations. It should be noted that the Symbol Search task included in SF2 also has a motoric component, although the motor demands of this task are quite minimal and unlikely to impact significantly upon test performance. The WTAR + demographics prediction of premorbid FSIQ in our neurological sample studied resulted in a slight underprediction, in comparison to the observed FSIQ of the matched control group. This underprediction was small, however (approximately 3 points), and consistent with the frequently reported observation that reading pronounciation performance does tend to decline somewhat with advancing dementia. The use of the WTAR is obviously the best methodology for predicted premorbid WAIS-III FSIQ, as these measures were co-normed, and clinicians can be reasonably assured that use of this methodology will result in a conservative estimate of intellectual decline in dementia evaluations employing the WTAR and these short forms (or full WAIS-III testing). That is, estimates of premorbid FSIQ using the WTAR in a clinical sample will be most accurate in cases of minimal acquired impairment, and will tend to slightly underestimate premorbid FSIQ with advancing dementia. Estimates of the magnitude of decline will be affected accordingly, with a tendency to slightly underestimate the magnitude of decline in cases of more advanced dementia. This is, if anything, preferable to error in the opposite direction (overestimating decline, and perhaps committing a type I error in making a diagnosis of dementia in the absence of acquired cognitive decline). In sum, the use of these short forms will be helpful in minimizing frustration and fatigue on the part of the patient, providing a reliable estimate of current intellectual status (and decline from predicted premorbid level using the WTAR). This will result in greater efficiency in neuropsychological assessment of elderly or more impaired patients, also allowing the inclusion of other neuropsychological tests critical to the evaluation of dementia within a single focused test session. Appendix A Scaling tables for Short Form 1 and Short Form 2 Short Form 1 Short Form 2 Sum SS IQ 90% CI 95% CI Percentile Sum SS IQ 90% CI 95% CI Percentile

7 A.J. Donnell et al. / Archives of Clinical Neuropsychology 22 (2007) Appendix A (Continued ) Short Form 1 Short Form 2 Sum SS IQ 90% CI 95% CI Percentile Sum SS IQ 90% CI 95% CI Percentile > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > >99.9 Wechsler Adult Intelligence Scale 3rd edition Copyright 1997 by The Psychological Corporation. Reproduced by Permission. All Rights Reserved. References Axelrod, B. N. (2001). Administration duration for the Wechsler Adult Intelligence Scale-III and Wechsler Memory Scale-III. Archives of Clinical Neuropsychology, 16, Doppelt, J. E. (1956). Estimated the full scale score on the Wechsler Adult Intelligence Scale from scores on four subtests. Journal of Consulting Psychology, 20,

8 924 A.J. Donnell et al. / Archives of Clinical Neuropsychology 22 (2007) Harrison, P. L., Kaufman, A. S., Hickman, J. A., & Kaufman, N. L. (1988). A survey of tests used for adult assessment. Journal of Psychoeducational Assessment, 6, Kaufman, A. S. (1990). Assessing adolescent and adult intelligence. New York: Allyn and Bacon Inc., pp Levy, P. (1968). Short-form tests: A methodological review. Psychological Bulletin, 69, Matarazzo, J. D., & Herman, D. O. (1984). Base rate data for the WAIS-R: Test-retest stability and VIQ-PIQ differences. Journal of Clinical Neuropsychology, 6, Nunnally, J. (1978). Psychometric Theory (2nd ed.). New York: McGraw-Hill. Rabin, L. A., Barr, W. B., & Burton, L. A. (2005). Assessment practices of clinical neuropsychologists in the United States and Canada: A survey of INS, NAN, and APA Division 40 members. Archives of Clinical Neuropsychology, 20, Randolph, C., Mohr, E., & Chase, T. N. (1993). Assessment of intellectual function in dementing disorders: Validity of WAIS-R short forms for patients with Alzheimer s, Huntington s, and Parkinson s disease. Journal of Clinical and Experimental Neuropsychology, 15, Resnick, R. J., & Entin, A. D. (1971). Is an abbreviated form of the WISC valid for Afro-Americans? Journal of Consulting and Clinical Psychology, 36, Ryan, J. J., Lopez, S. J., & Werth, T. R. (1998). Administration time estimates for WAIS-III subtests, scales, and short forms in a clinical sample. Journal of Psychoeducational Assessment, 16, The Psychological Corporation. (2001). Wechsler Test of Adult Reading: Manual. San Antonio, TX: The Psychological Corporation. Wechsler, D. (1997). Wechsler Adult Intelligence Scale Third Edition. San Antonio, TX: The Psychological Corporation. Wechsler, D. (1997b). WAIS-III/WMS-III technical manual. San Antonio TX: The Psychological Corporation. Wechsler, D. (1997c). Wechsler Memory Scale Revised. San Antonio, TX: The Psychological Corporation. Wymer, J. H., Rayls, K., & Wagner, M. T. (2003). Utility of a clinically derived abbreviated form of the WAIS-III. Archives of Clinical Neuropsychology, 18,