WAIS-IV short form applied to a mixed neurological Swedish clinical sample

Abstract

The Wechsler Adult Intelligence Scale, fourth version (WAIS-IV), is a frequently used instrument for neuropsychological assessment. The aim was to assess the degree of conformity between the Scandinavian adaptation of the WAIS-IV and a short form of this scale (SF) in a mixed sample of neurological diagnoses. The SF comprised Block Design, Similarities, Digit Span, Arithmetic, Information, Digit Symbol (in the WAIS-IV named Coding), and Picture Completion, the latter here replaced by Matrix Reasoning. The sample consisted of 150 patients and included multiple sclerosis (n = 27), brain tumor (n = 15), traumatic brain injury (n = 60) and vascular brain damage (n = 48). There was a lack of congruence between the WAIS-IV and the SF in the entire sample, revealing selectively significantly higher scores for the SF on Full Scale IQ (FSIQ) and Processing Speed Index (PSI). On a diagnostic group level, the discrepancies were as follows: in the traumatic brain injury group on FSIQ, Verbal Comprehension Index (VCI) and PSI, in the vascular damage group the FSIQ was significantly higher on SF compared to the WAIS-IV. Since the results revealed several mismatches between the SF and the WAIS-IV, except for the MS and traumatic brain tumor group, there is a lack of interchangeability between these two sets of tests. Thus, generally the SF cannot be recommended as a substitute for the WAIS-IV in this type of mixed Swedish neurological sample. The small sample sizes make the generalizability of this study limited.

Keywords:

• : short form
• WAIS-IV
• cultural adaptation
• clinical applicability

Neuropsychological assessment is a time-consuming process, and ever since the publication of” A Brief Binet-Simon Scale” for children (Doll, 1917) research has aimed to abbreviate and reduce its extensiveness through a wide plethora of short forms (Rabin, 1943; Úbeda et al., 2016; Denney et al., 2015). For an extensive review of short forms of the Wechsler Adult Intelligence Scale (WAIS), see Lindau and Najström (2019). According to the manual of the current Scandinavian version of the American scale (Wechsler, 2008), the WAIS-Fourth Edition (WAIS-IV), the testing time varies between clinical and non-clinical groups. In the Scandinavian adaptation of the WAIS-IV the administration time for the standardization sample aged 16:00-74:11 years is found to range from 125 to 160 minutes, and in an elderly sample 75:00-90:11 from 140 to 200 minutes. Add to this the clinician's expenditure of time for scoring, analysis of test profiles and communication of the results.

The abbreviation (SF) of the Wechsler Adult Intelligence Scale-Revised (WAIS-R) by Ward (1990) have been mentioned by several authors as particularly useful (e.g. Strauss et al., 1996; Pilgrim et al., 1999; Meyers et al., 2013; Girard et al., 2010; Ryan et al., 1996; Callahan et al., 1997; Bulzacka et al., 2016). This brief measure encompasses seven subtests: Block Design, Similarities, Digit Span, Arithmetic, Information, Digit Symbol (in the WAIS-IV named Coding), and Picture Completion, allowing for a reduction of testing time by about 50% (Meyers et al., 2013). Later studies have confirmed the clinical validity and reliability of the SF for the subsequent editions of the WAIS: WAIS-Third Edition (WAIS-III) and the WAIS-IV (Pilgrim et al., 1999; Meyers et al., 2013).

The SF was derived from the American version of the WAIS-R, and likewise, the above-mentioned studies of its psychometric properties were tested on American samples. However, as has been thoroughly discussed and stressed in Lindau and Najström (2019), neuropsychological test batteries cannot automatically be dislodged from one cultural context to another. Instead, it is necessary to ensure the applicability of the test content, as well as the norms and abbreviations of a scale, to the cultural context in which the battery is to be used. In Lindau and Najström (2019) the statistical validity of the SF for the WAIS-IV was examined on a Swedish non-clinical sample consisting of 261 individuals ranging in age from 18 to 73 years. In the entire sample, the results were contradictory, showing conformity between Full Scale IQ (FSIQ) of the original version of WAIS-IV and the prorated IQ-values derived from SF, but lacking consistency at the index level. Thus, according to the authors the utility of the SF in a non-clinical sample is not confirmed. However, in clinical SF studies the SF has been used with good results (Pilgrim et al., 1999; Meyers et al., 2013). Due to this contradiction, and since the SF has not yet been tested in any Swedish clinical sample, the aim was to assess the usefulness of the SF compared to the original WAIS-IV for neurological, diagnostic purposes.

The following question was addressed:

• How well does the FSIQ and index estimates of the SF correspond to the regular WAIS-IV measures in a Swedish clinical sample of mixed neurological diagnoses?

Method

Sample and inclusion criteria

The entire sample encompassed 150 patients with an even distribution between females and males, devided in the following diagnostic groups: multiple sclerosis (MS), (n = 27), brain tumor (n = 15), traumatic brain injury (n = 60), and vascular damage (n = 48). The medical conditions were diagnosed according to the requirements of the Swedish version of the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10-SE). This is a representative sample for neurologic conditions often undergoing neuropsychological examination (Zucchella et al., 2018; Ramirez-Aubone & Lee, 2019). The neuropsychological inclusion criterion was a completed evaluation with the ten core tests of the WAIS-IV using the Scandinavian adaptation of the WAIS-IV (Wechsler, 2010). Demographic information is shown in Table 1.

Table 1 Demography for the entire sample and diagnostic groups.

Clinical sample	All	SD	MS	SD	Tumor	SD	Trauma	SD	Vasc.	SD
N	150		27		15		60		48
F/M, n	79/72		22/5		7/8		24/37		26/22
Age at testing, x̄	43.3	13.5	41.5	11.8	39.1	15.1	40.7	14.6	49	10.8

Recruitment

The patients were recruited from Region Gävleborg: Neurologmottagningen in Gävle and Rehabiliteringsmedicin in Sandviken, Sweden. In order to access neuropsychological data, two approaches were used: (1) the patients were asked to participate in the study in connection with regular clinical examinations during the period June 2017 – April 2018, and (2) through journal searches in psychologist archives. Patients identified in the journal archives were contacted by mail and telephone. In both recruitment methods, patients were asked for informed written consent to use their data for the research study in accordance with the demands of the Regional Ethics Review Board in Stockholm.

Multiple sclerosis

The MS group encompassed 27 patients: 22 with relapsing-remitting MS, one patient with primary progressive MS and four with secondary progressive MS. All patients afflicted by relapsing-remitting MS were on regular disease modifying medication.

Brain tumor

This diagnostic group comprised 15 patients. The diagnoses were made after brain imaging, biopsy and histological analyses. Tumors such as glioma, meningioma, and astrocytoma were localized to different parts of the brain: frontally, parietally and temporally, mostly in the left hemisphere (eight of 15 cases). Epilepsy as a consequence of the brain tumors sometimes occurred.

Traumatic brain injury

This group included a total of 61 patients, whereof six had clear left-sided injuries, five clear right-sided and 50 patients bilateral lesions or injuries that were not visible on brain imaging, or where brain imaging information was missing, although this latter category was in minority. One patient suffered at the time of the neuropsychological examination from several disorders affecting the brain and was excluded from the statistical analyses.

The traumatic brain injuries consisted of hemorrhages, contusions, brain edema or ischemia, appearing in scattered areas in the four lobes of the cerebrum, and/or in cerebellum and sometimes even subdurally or subarchnoidally. No medical classification in mild/moderate/severe traumatic injury was available for this group, but primarily patients with moderate to severe traumatic brain injuries were referred to the clinics in question. As a result of the traumatic episode, some of the patients also suffered from epilepsy.

Vascular brain damage

This group comprised 48 patients with the following types of brain injuries: twelve left sided infarctions, 14 right sided infarctions, seven infarctions where the localizations were diffuse, five left sided intracerebral hemorrhages, two right sided intracerebral hemorrhages, one hemmorrhage without clear localization and seven subarachnoid hemorrhages (right- and left sided). The infarctions and hemorrhages were found in the frontal, parietal, temporal and occipital areas as well as in cerebellum. Further, some patients had epilepsy as a sequelae of the vascular episode.

Neuropsychological assessment

The test framework of the WAIS-IV consists of a measure of overall cognitive ability, FSIQ, and four indices, each representing a cognitive domain. FSIQ is a composite score deduced from the sum of scaled scores of the ten core tests, and the indices are composite scores of these subtests' scaled scores, distributed on the cognitive domains where they belong: Verbal Comprehension Index (VCI), Perceptual Reasoning Index (PRI), Working Memory Index (WMI) and Processing Speed Index (PSI). The core tests included in VCI are Similarities (SI), Vocabulary (VC) and Information (IN); in PRI: Block Design (BD), Matrix Reasoning (MR) and Visual Puzzles (VP); in WMI: Digit Span (DS) and Arithmetic (AR) and in PSI Symbol Search (SS) and Coding (CD). Further, five supplemental subtests are intended as substitutions for the core tests or to gain in-depth understanding of the test profile. These supplemental subtests Comprehension (VCI), Figure Weights and Picture Completion (PRI), Letter-Number Sequencing (WMI) and Cancellation (PSI) were not used in the present study.

To obtain the full WAIS-IV results, the ten core subtests were administered following standard procedures (Wechsler, 2010). The assessments were made by experienced licensed neuropsychologists and intern psychologists under the supervision of trained licensed neuropsychologists. In the WAIS-IV, all subtests in the SF (Information, Digit Span, Arithmetic, Similarities, Picture Completion, Block Design and Coding) except Picture Completion are core tests. In the current study the SF was applied with Matrix Reasoning replacing Picture Completion according to the WAIS-IV manual (Wechsler, 2010).

Prorating scaling for the estimation of SF FSIQ and indices

The FSIQ for the SF and its indices VCI, PRI, and PSI were estimated according to the formulae for proration (Table 2) in the WAIS-IV (2010) manual (p. 102-103 and Table A.8, Appendix A). This method is commonly used in studies of short forms of the WAIS (e.g. Bulzacka et al., 2016). The WMI was excluded from the proration scaling, since it comprises the same tests in the SF as in the WAIS-IV

Table 2 Formulas for proration of SF index values based on estimated SF scaled scores.

Index	Formulas
FSIQ	(Block Design sc^a+ Similarities sc + Digit Span sc + Matrix Reasoning sc + Arithmetic sc + Information + Coding) x 10/7
VCI	(Information sc + Similarities sc) x 3/2
PRI	(Block Design sc + Matrix Reasoning sc) x 3/2
PSI	Coding sc x 2
WMI	Identical in SF to WAIS-IV

a sc, scaled score.

Statistical analyses

Histogram and Kolmogorov-Smirnov test were used to assess normality in the entire sample. The analyses did not result in any significant value, p=.200, and thus the data was considered as approximately normally distributed, allowing for the use of parametric statistics.

The comparisons between the original WAIS-IV measures and the SF in the entire sample and in the diagnostic subgroups were made by paired samples t-test. WMI was excluded from the paired samples t-tests, since this index is identical in the WAIS-IV to the SF. The level of significance was set at p<.05.

For all statistical analyses the “IBM SPSS Statistics” version 25, was used.

Ethical approval

The study was approved by the Regional Review Board in Stockholm, Ref: 2017/549-31.

Results

When checking for outliers (Brace, Snelgar, & Kemp, 2012) six were identified. Five individuals scored too high relative the main pattern of the WMI index. For the sixth individual, data was too low on this index as well as on the WAIS-IV FSIQ and SF FSIQ. All these outlier values were deleted.

WAIS-IV versus SF – entire sample

The entire group’s results on the neuropsychological assessment with the WAIS-IV and the SF are shown in Table 3. All measures revolved around one standard deviation below normal average FSIQ, that is100 points, according to the WAIS-IV (Wechsler, 2010) intelligence scale. In both cases, the SF estimates were slightly higher than the WAIS-IV measures. The standard deviations from the mean cognitive performance levels were consistently high.

Table 3 WAIS-IV and SF FSIQ and composite scores, Mean and Standard Deviation in total sample and diagnostic groups.

Measure	Total sample	MS	Brain tumor	Head trauma	Vascular
N	150	27	15	60	48
WAIS-IV FSIQ	86.09 (15.90)	90.61 (12.27)	88.00 (16.92)	89.27 (14.45)	79.92 (17.90)
SF FSIQ	86.68 (16.65)	90.74 (13.02)	88.71 (18.71)	89.98 (15.27)	80.67 (18.46)
WAIS-IV VCI	88.24 (16.31)	93.70 (14.48)	84.47 (18.62)	90.47 (15.01)	84.38 (17.76)
SF VCI	88.72 (18.67)	93.43 (15.74)	85.40 (20.87)	91.60 (18.08)	84.29 (19.91)
WAIS-IV PRI	90.91 (17.15)	95.57 (14.23)	93.20 (22.02)	94.90 (16.09)	83.50 (16.54)
SF PRI	91.15 (18.01)	96.26 (16.21)	92.53 (24.51)	95.37 (16.24)	83.83 (17.21)
WAIS-IV PSI	83.72 (16.61)	88.61 (14.27)	83.33 (15.50)	86.50 (16.25)	78.77 (17.82)
SF PSI	84.80 (17.51)	90.00 (15.69)	85.00 (15.66)	87.17 (17.14)	80.38 (18.91)

Paired samples t-test for the entire group indicated significant differences between the WAIS-IV FSIQ and the SF FSIQ as well as between the WAIS-IV PSI and the SF PSI, see Table 4. There were no significant differences between the WAIS-IV and the SF on the indices VCI and PRI.

Table 4 Paired sample t-test between the WAIS-IV and the SF for the entire sample.

Pair	M^a	SD^b	t-value	df^c	p
WAIS-IV FSIQ - SF FSIQ	−.593	2.499	−2.908	149	.004
WAIS-IV VCI - SF VCI	−.477	4.020	−1.458	149	.147
WAIS-IV PRI - SF PRI	−.238	4.777	−.613	149	.541
WAIS-IV PSI - SF PSI	−1.086	6.188	−2.157	149	.033

a M, mean difference.

b SD, standard deviation.

c df, degrees of freedom.

WAIS-IV versus SF – diagnostic group level

Multiple sclerosis

The results of the assessments with the WAIS-IV and the SF in the MS group varied between 1 SD below average and average. Compared to the measures of the original WAIS-IV the SF estimates were mostly slightly higher (Table 3).

The analyses showed no significant differences in the MS group between the WAIS-IV and the SF FSIQ and their indices.

Brain tumor

Mean values of both test batteries on the FSIQ and the indices are shown in Table 3. The SF values were slightly higher than the WAIS-IV values, though about one SD below average normative scores on the intelligence scale. The results of the paired samples t-test between the WAIS-IV and the SF FSIQ and index measures in the brain tumor group indicated no significant differences.

Traumatic brain injury

Cognitive performance levels in the traumatic brain injury group ranged between low average FSIQ and average (Wechsler, 2010), with slightly higher values for the SF measures compared to the WAIS-IV measures (Table 3). There were significant differences between the WAIS-IV and the SF FSIQ as well as between the two batteries’ measurements of VCI. No differences between the remaining indices were found (Table 5).

Table 5 Paired samples t-test between WAIS-IV and SF in the traumatic brain injury group.

Pair	M^a	SD^b	t-value	df^c	p
WAIS-IV FSIQ - SF FSIQ	−.717	2.591	−2.142	59	.036
WAIS-IV VCI - SF VCI	−1.133	4.343	−2.021	59	.048
WAIS-IV PRI - SF PRI	−.467	4.876	−.741	59	.461
WAIS-IV PSI - SF PSI	−.667	5.219	−.989	59	.327

a M, mean value.

b SD, standard deviation.

c df, degrees of freedom.

Vascular brain damage

The results indicate that the mean cognitive performance scores in the vascular group varied between 1 and 2 SD below average FSIQ (Wechsler, 2010), (Table 3).

There was a significant difference in the vascular brain damage group between the WAIS-IV FSIQ and the SF FSIQ estimate, where the SF score was significantly higher than its WAIS-IV counterpart (Table 6). Apart from that, no significant differences were found.

Table 6 Paired samples t-test of WAIS-IV and SF in the vascular brain damage group.

Pair	M^a	SD^b	t-value	df^c	p
FSIQ WAIS-IV- FSIQ SF	−.750	2.392	−2.172	47	.035
WAIS-IV VCI - SF VCI	.083	4.277	.135	47	.893
WAIS-IV PRI - SF PRI	−.333	3.986	-.579	47	.565
WAIS-IV PSI - SF PSI	−1.604	8.039	−1.382	47	.173

a M, mean value.

b SD, standard deviation.

c df, degrees of freedom.

Summary of results

The paired sample t-test of the WAIS-IV and the SF measures of the entire sample resulted in non-significant results regarding VCI and PRI indices and significant differences concerning FSIQ and PSI. The results in the diagnostic groups were dispersed. In the vascular group FSIQ differed, but not the indices; the head trauma group differed in FSIQ and VCI but not in PRI and PSI; in the MS and brain tumor groups no significant differences were found between the WAIS-IV and SF.

Discussion

The aim was to investigate the degree of correspondence between the SF (Ward, 1990) and the Scandinavian adaptation of the full WAIS-IV (Wechsler, 2010) in a Swedish mixed neurological sample with cognitive disturbances. The comparisons showed an irregular pattern, revealing conformity as well as a significant lack of it, in the entire sample as well as in the different diagnostic groups. The approach to analyze the entire sample, with mixed neurological diagnoses, was thought to mimic the clinical situation, where a variety of neurological disorders with symptomathologies affecting the cognition usually appear, and where the task for the clinical neuropsychologist is to make a differential diagnostic statement (Zucchella et al., 2018; Ramirez-Aubone & Lee, 2019). The main finding of the analyses is, that the SF cannot be regarded as an adequate counterpart of the WAIS-IV in a Swedish clinical sample of mixed neurological conditions. This observation contradicts Meyers et al. (2013), who found that Ward’s formula Ward (1990) with high reliability and validity could be applied to clinical groups. However, the clinical groups in Meyers et al. (2013) differed from the present study, both in terms of the number of diagnoses included (16 diagnostic entities whereof several encompassing just one case) and types of diagnoses. In the study by Meyers et al. (2013) the majority of patients were diagnosed with cognitive disorder not otherwise specified (NOS), developmental disorders such as learning disabilities and attention-deficit disorder/attention deficit hyperactivity disorder or primary mental health diagnoses.

Regarding the SF FSIQ and index estimates, and their WAIS-IV counterparts for each of the diagnostic groups, the analyses generated several contradictory results. It is doubtful whether the SF is reliable with respect to traumatic brain injury and vascular disorder, since in the traumatic brain injury group the FSIQ and VCI differed, and in the vascular group there was a difference between the two FSIQ measures. However, there were no significant differences between the two test sets in the MS and brain tumor groups respectively. Hence, the SF could be plausible in groups with these diagnoses.

Different factors may have contributed to the lack of overall consistency between the WAIS-IV and the SF in the diagnostic groups. A potential contributing factor might have been the sample size differing between the diagnostic groups. With this said, in the present study the analyses were based on within-group comparisons and not between-groups making sample size a minor issue. Nevertheless, the differing and somewhat meagre sample sizes might compromise the generalizability of the results. Furthermore, variations in sample sizes reflect the clinical reality, where different neurological diagnoses are variously common, which may also mirror their epidemiological spread.

Another possible contributing factor was the heterogeneity of the groups, and that the groups were not analyzed in different types, such as different variants of MS, brain tumor, head trauma, and vascular incidents. Subdivisions were not created as some of the diagnostic groups were too small for such a classification to be relevant: MS (n = 27) and brain tumor (n = 15), and over all, the medical records were too vague in regard to lesion localization/disease variants, which implied that there was no satisfactory basis for a meaningful classification to be made from. This discussion concerning the accuracy and characteristics of the subsamples do not, from a statistical point of view, have any relevance since the aim of the study was to compare the degree of correspondence between the WAIS-IV and the SF based on a within-group comparison.

From a cultural point of view, when non-culturally adapted neuropsychological tests are used for diagnostic purposes, the risk is high for misdiagnosis, in some cases up to 51.9% false positives (Daugherty et al., 2017). This highlights the necessity of a study of the SF applied to the Scandinavian-adapted version of the WAIS-IV to be performed on a Swedish clinical group, rather than automatically applying the findings by Meyers et al. (2013) to a Swedish cultural context.van Ool et al. (2018) recommend the use of short forms of the WAIS-IV for research or for screening of cognitive impairment in epilepsy or other neurological diseases, but advises against it in the routine investigations of clinical patients. To this should be added, that if the SF is considered as a screening instrument it could entail a risk of overestimation of performance levels, increasing the probability of Type II errors, e.g., failing to discover a diagnosis, even though one exists. This tendency of SF overestimating performance levels can be seen in the present study, as well as in Lindau and Najström (2019).

Limitations of the study

A downscaling of the ordinary WAIS-IV to SF entails that FSIQ and indices are measured with fewer subtests, which in turn could have been analysed by examining the internal consistency of the SF, for example with Cronbach's alpha. According to Smith et al. (2000), a more thorough examination would have been to administer both the WAIS-IV and the SF to one and the same set of patients, and then, through a special procedure, correct for learning effects. Smith et al. (2000) emphasizes this approach as one of the best methods for determining the reliability and validity of a short form. Achieving full reliability and validity is according to Smith et al. (2000) psychometrically even more difficult when developing a short form than an original test, given the fewer number of tasks in a short form. The data used in this study was derived from clinical assessments where the psychologists who administrated the testing had to tailor the sessions according to the resources of each patient. As a consequence the testing order may have differed.

Future research

When neuropsychological results are included in the diagnostic work, they usually form part of a puzzle consisting of many clinical pieces that must be added to make a diagnosis. Future studies should put interest in comparing the effectiveness of the full WAIS-IV and the SF to imaging data and biomarkers, in regard to the diagnoses included in the present study.

Conclusions

Since the results revealed several mismatches between the SF and the WAIS-IV, except for the MS and brain tumor group, there is a lack of interchangeability between these two measures. Thus, generally, the SF cannot be recommended as a substitute for the WAIS-IV in this type of mixed neurological sample. For a short form to be considered as useful in neuropsychological clinical practice it is necessary that its psychometric qualities are generally satisfactory, not as in the present study, selectively. This taken into account, the study could not confirm the results of Ward (1990) and Meyers et al. (2013) in this Swedish neurological sample.

Acknowledgments

Neurologmottagningen in Gävle and Rehabiliteringsmedicin in Sandviken (Sweden) are thanked for approving the conduct of the study at their clinics, as well as the licensed Psychologists at these clinics for providing assistance during the data collection phase. A special thank is addressed to the patients for their valuable contributions. This article is a revision of a work previously composed as a specialist thesis within the framework of the Swedish Psychological Association, Lundberg, Milena: Kortform av WAIS-IV tillämpad på ett svenskt kliniskt urval av neurologiska diagnoser för diagnostik av kognitiv dysfunktion. The thesis can be requested online from https://www.psykologforbundet.se/Specialistutbildning/Om-specialistutbildningen/specialistarbetet2/.

Additional information

Funding

The financial support provided by Centrum för Kompetensutveckling inom Vård och Omsorg (CKVO) at Stockholm University, Sweden, has been of great help in the carrying through of the research.