Validation and Reliability of The Frontal Assesment Battery (FAB) in Turkish

Validation and Reliability of The Frontal Assesment Battery (FAB) in Turkish Nilgün TUNÇAY, Gul KAYSERİLİ, Erhan ESER, Yaşar ZORLU, Berna Binnur AKDEDE, Görsev YENER Tepecik Training and Research Hospital, Neurology Clinic, Izmir, Türkiye Dokuz Eylül University, Department of Neurological Sciences, Izmir, Türkiye Celal Bayar University, Department of Public Health, Manisa, Türkiye Dokuz Eylül University, Department of Psychiatry, Izmir, Türkiye Istanbul Kultur University, BEYINMER, Istanbul, Türkiye


INTRODUCTION
It is useful to assess the frontal lobe functions and to determine executive function disorders in the diagnosis, differential diagnosis and monitoring of degenerative diseases including the subcortical structures.However, clinical assessment of the frontal lobe functions is difficult and the application of the tests used in clinical practice requires some time.Therefore, less time-consuming tests are required (4 . 5. 14).
The Frontal Assessment Battery (FAB), designed by Dubois et al., is a simple, easily applicable, non-time-consuming (about 10 minutes) bedside test that is sensitive to frontal lobe functional disorders (5) .This study aims to determine the validity and reliability of FAB in Turkish society and examined to its applicability in Turkish-speaking patients.Secondly, the study aims to determine the validity of Frontal Assessment Battery in neuropsychiatric diseases causing frontal functional disorders such as Parkinson's Disease (PD), Alzheimer's disease (AD) and Schizophrenia.To determine the reliability of this study, the FAB results were compared with those of the Stroop Test, which is sensitive to frontal lobe functions; and the frequently used the Mini-Mental State Examination (MMSE), which assesses the general cognitive functions and has proven validity in Turkish population.

Subjects
As the sample size should be at least fivefold or even ten-fold the number of items in validity and reliability studies, this study includes 96 neuropsychiatric subjects (33 subjects with Alzheimer type dementia, 30 subjects with idiopathic Parkinson's and 31 subjects with Schizophrenia ) and 94 healthy subjects for the 6-item FAB.

Subjects with Alzheimer's Disease
The participants were chosen consecutively from the subjects who were followed in both the Department of Neurology of the Faculty of Medicine of Dokuz Eylül University and the Dementia Polyclinic of the Neurology Clinic of Tepecik Training and Research Hospital and who were diagnosed with dementia according the DSM-IV and with probable dementia according to the NINCDS-ADRDA criteria (6) .

Subjects with Schizophrenia
The study included 31 subjects who were followed in the Schizophrenia Polyclinic of the Psychiatry Clinic of the Faculty of Medicine of Dokuz Eylül University and who were diagnosed with schizophrenia according to the DSM IV.The patients were assessed by the policlinic physician using the clinical assessment scale.

Healthy Subjects
Healthy control subjects were chosen voluntarily among the relatives of patients participating in the study that could response to notices positively, live independently in the society and have no prior history of a neuro-psychiatric disease.
Inclusion criteria consisted from having no prior history of a psychiatric or neurological disease for the control group participants; being a primary school graduate or above for all healthy participants and patients in the study; a score of "0" in the Clinical Dementia Rating (CDR) Scale; good general health with no prior history of a neurological or psychiatric disease.

Translation of the Test into Turkish
Written permissions were received from the authors to use and translate FAB into Turkish.The steps followed in the translation process are as follows: 1. Forward Translation: The directives of the test were translated into Turkish by 9 people with advanced understanding of English.
2. Combining the forward translations (forming the consensus version): Different translations were combined by two independent experts.

Back-Translation:
The consensus version was then back-translated into English by another person fluent in both of the languages and was then compared conceptually with the original text.
Before starting to FAB validation, a pilot study was conducted with 40 voluntary participants.Majority of the 20 of these voluntary participants did not give the expected answer to the question "what is the common facet of table and chair".Remaining 20 voluntary participants were asked "what is the common facet of coffee table, chair and armchair"; an increased success rate was observed.The authors were then consulted on whether it was appropriate to use the question "…" instead of the question "…".The authors were proposed not to change the original test, so therefore the originality of the test was preserved, even though the scores in healthy elderly were better when similarity was asked among three items.
4. The latest Turkish version of FAB, a socio-demographic questionnaire, the MMSE and the Stroop Test were applied to patients and healthy participants.

Psychometric analyses were carried out.
This study was approved by the ethical committee of Dokuz Eylül University, Faculty of Medicine.Prior to the application, informed consent was obtained from patients, healthy controls and the relatives of subjects with dementia.

Application
A short demographic questionnaire, the MMSE, the Stroop Test and the FAB were applied to patients and healthy participants.The tests were conducted by a physician.The FAB consists of 6 items, each of which was evaluated with a score between 0 and 3.The maximum possible score is 18 and higher scores indicate better performance.
Inter-judge reliability was conducted with another psychologist who was trained in neuropsychology.

Psychometric Analyses
The SPSS program (version 13.0 for Windows) was used for the statistical analyses.

Reliability Analyses
Internal consistency, inter-rater reliability (n=22) and test-retest reliability (n=20) were examined.Cronbach's alpha was used to test the internal consistency and Intraclass Correlation Coefficient -ICC was used for inter-judge reliability and test-retest reliability.In both tests, a score close to 1 indicates higher reliability.

Validity Analyses
Two basic approaches were used in examining the validity of the Turkish version of FAB. b.Known-groups validity is sometimes defined as "sensitivity analysis".It aims to test and detect the actual variance between the groups which are previously expected (assumed) to be hypothetically different.The test is assumed to function well in case it detects and confirms such variance.Student's t-test was used in paired comparisons while one-way analysis of variance (ANOVA) was used in comparisons with more than two categories, and Tukey-B test was used in the post-hoc analysis.

RESULTS
Table 1 shows the socio-demographic characteristics of the 94 subjects and the 92 controls included in this study.No significant difference was observed between subjects and controls in terms of age (p:0.877),gender (p:0.074) and educational levels (p:0.063).
Table 2 shows the FAB, Stroop Test and MMSE scores of the study group.All three instruments showed significant differences in all scores between the total neuropsychiatric subject groups and control groups except for the score for the sixth category of FAB (p<0.05).In all scores, subject groups had lower scores than control groups.

1.Reliability Results
Table 3 shows internal consistencies of FAB scores in various frontal lobe diseases and in the control group.As shown in the table, high alpha values were obtained in total frontal lobe diseases and subjects with AD.
While medium-level internal consistency values were obtained in the subjects with schizophrenia and in the control group, low level internal consistency values (alpha values) were obtained in subjects with PD.
Test-retest values of the FAB were presented with intraclass correlation coefficients (Figure 1).As shown in the figure, statistically significant consistency (r= 0.89; 95% CI 0.72-0.95) was observed in the test results conducted monthly.A very high intra-rater reliability rate was also found (r=1).

1.Validity Results
Table 4 shows the concurrent validity of FAB and MMSE.Assessment of the subjects showed a positive correlation between FAB and MMSE.Similarly, high scores were obtained in subjects with AD while medium level scores were determined in subjects with schizophrenia and the control group, and relatively low correlations were established in subjects with PD.
Table 5 shows concurrent validity of FAB and Stroop test.First of all, a negative correlation was determined between the FAB and Stroop sub-test scores, as expected.However, there is a high negative correlation between the FAB score and particularly the error number sub-score of the Stroop test in the control group; and all subject groups except for those with schizophrenia.While there is a negative correlation between the FAB score and error-recovering sub-score in patients with PD, the relationships between the FAB scores and other sub scores of Stroop test in the control group were found to be significant, though a higher negative correlation was expected.
Tables 6,7 and 8 show the "Known-groups (construct) validity" of the Turkish version of FAB in the categories of age, gender and education, respectively.As shown in Table 6, FAB score decreases with age.A similar tendency was observed in both subject and control groups, especially among participants aged above 65 years.
According to Table 7, female subjects achieved significantly lower FAB scores than male cases; no such gender difference was observed in the control group.
Table 8 shows FAB scores by education groups.Significant differences were determined among the sub-categories of education in both subject and control groups.Test scores show a tendency to increase with higher education level.There is also a break in the FAB scores between the group of primary graduates and the group educated at high school level and above.

DISCUSSION
This study examined in detail the applicability of the Frontal Assessment Battery (FAB) in Turkish population.FAB internal consistency coefficient was found to be sufficient in the group with neuropsychiatric diseases (Cronbach's alpha=0.73)and to be at a medium level in the healthy control group (Cronbach's alpha=0.52).The test-retest reliability was found to be very high (r=0.89),as in previous studies (1,4,14,17) .The Cronbach's alpha value was found 0.78 in the original study by Dubois et al., 0.70 in a study by Nakaaiki et al. and 0.77 in a study of Mok et al (5,13,14) .
In the present study, internal consistency was 0.73 in the case group with Alzheimer's disease, 0.28 in the subject group with Parkinson's disease and 0.66 in the subject group with Schizophrenia.The reason for the low internal consistency in the group with Parkinson's disease is the wider range of FAB scores in this group.Some studies conducted with Single Photon Emission Computed Topography (SPECT) showed that a wider range of FAB scores could be related to the disruption in the left parietal lobe and supramarginal gyrus together with the disruption in the frontostriatal circuit (3,12) .A SPECT study by Matsui et al. included two groups of Parkinson's patients with similar MMSE and different FAB scores (9.8 and 15.3); hypoperfusion areas were observed on the parietal, temporal and frontal lobes (12) .
Total FAB scores of the neuropsychiatric patient group with frontal lobe syndrome were lower than those of the healthy individuals.This finding indicates that the accuracy of this test is good.
FAB scores in healthy participants and participants with neuropsychiatric diseases were found to be positively correlated with education and negatively correlated with age.These findings support those of previous studies and emphasize the importance of these two factors in neurocognitive assessment (2,11,13,14) .
This study found positive correlation between MMSE and FAB scores in healthy individuals, in the whole neuropsychiatric disease group considered as having frontal lobe syndrome, and in the Alzheimer's and Schizophrenia sub-groups, but found no such significant correlation in the patients with Parkinson's disease.The study by Dubois et al. (5) found no high correlation between FAB and MMSE scores but a positive correlation was reported by Kugo et al. (10) in cases with dementia; by Castiglioni et al. (4) in cases with Azheimer's disease and frontotemporal dementia; by Lima et al. (11) in healthy individuals and cases with PD; and by Beato et al. (2) in healthy individuals.This could be explained by the weak distinctiveness of FAB as welll as the assessment of some frontal functions through MMSE.However, the relationship of the cognitive functions of FAB with general measurements and further data, such as sub-group analysis of MMSE, are required to clarify the relationship between FAB and MMSE.Correlation was established between FAB and only the error number sub-score of the Stroop test (except for patients with Schizophrenia).Given these results, the Stroop test may not be particularly appropriate for cross correlation.Because frontal lobe activities occur across a very widespread area and it is also thought that both tests cannot assess the same localization (3,8,16) .
Stroop performance is considered to support the parallel data processing model mostly in the left frontal region and to be closely related with the anterior cingulate cortex (ACC) and the orbital parts of the prefrontal cortex.The monitoring study by Yoshida et al. shows that FAB particularly reflects the functions of the left precentral and bilateral callosomarginal regions (18) .Revision of these correlations in large groups of patients with a functional monitoring study may be suggested.
The cases included in our study were chosen from neuropsychiatric disease groups that show frontal lobe syndrome.
The FAB results show that there is a significant difference between the neuropsychiatric case group and the control group.AD was determined as the most remarkable disorder among those included in the study.Although also different from the control group in both PD and Schizophrenia groups, these two groups were found to be compatible to each other.
The sixth FAB item (peripheral autonomy) did not function in both control and case groups.The reason could be the absence of cases with extensive frontal lobe disorder.Grasping reflex disorder and, concordantly, a peripheral autonomy subtest disorder, were determined in only one case with advanced AD. FAB scores showed significant increase after the level of secondary education.This indicates that elementary school level is not sufficient for the development of frontal functions.
Table 9 shows the demographic characteristics of individuals from the control groups of previous studies and the control group of this study.The study by Dubois et al. (5) found the highest FAB scores, while the control group of the study by Mok et al. (13) had the lowest score.One reason of this situation may be the relatively low educational level and the advanced age of the Mok group.It was found that age and educational level did not have an effect on the FAB scores in the studies by Dubois et al. (5) and Kugo et al. (10) and that FAB was positively correlated with educational level and negatively correlated with age in the studies by Mok et al. (13) , Lima et al. (11) and in the present study.Another possible reason the differing results may be linguistic and cultural differences; for example, low FAB scores were reported in the control cases of studies conducted in China and Japan (10,13) .FAB scores are higher in the control group of the study conducted in French, which is the language in which the test was originally developed (5) .This could result from the use of a different system of conceptualizing in Turkey, like many other Far East countries (10,12,13,14,15,18) .Conceptualization problems also occurred in our study group.Potential cultural difference was found to be important in the Turkish version of FAB.In the present study, no significant difference was found in the similarity subtest, except for the cases with AD and the control group.
In the present study, elderly controls in particular, and controls with low educational level, answered "for eating" to the question about similarity of table and chair, which was not accepted as correct.
The answer "equipment for studying" was accepted in the study conducted in Japan (10) .This may be the reason for the absence of a score difference between the case and healthy control groups, because a relationship was also established between conceptualization sub-test and education in healthy individuals.In Turkish society, many healthy elderly individuals with low educational level responded "for eating" and were not able to give the expected correct answer.
This study did not take into account whether schizophrenia cases were positively or negatively symptomatic and any potential effects of their drug therapy; this could be regarded as one of the limitations of this study.Another limitation is the exclusion of the clinical stage of the disease in cases with AD, PD and schizophrenia.It may be more appropriate to make clinical staging, to make comparison with larger groups and to assess FAB together with SPECT.
In conclusion, the FAB scores of the neuropsychiatric patients with a possible frontal lobe syndrome involvement were lower compared to the healthy population.This study concludes that the Turkish version of FAB could be used as a reliable scale for the assessment of frontal lobe functions in various neuron-psychiatric diseases and especially in Alzheimer's disease.FAB will contribute to the monitoring and scanning process of frontal lobe functions as it is easily applicable and not time-consuming compared to other tests.We believe that the present study indicates opportunities for more comprehensive studies to be conducted for this purpose.
Parkinson's Disease Subjects were followed in both the Department of Neurology of the Faculty of Medicine of Dokuz Eylül University and the Movement Disorders Polyclinic of the Neurology Clinic of Tepecik Training and Research Hospital and they were diagnosed with PD according to the Hughes criteria.
a. Convergent Validity b.Known-groups Validity a.In the convergent validity approach, non-parametric correlations (Spearman's Rho) of simultaneously applied FAB and Stroop Tests and FAB and MMSE tests were calculated.It is expected that these comparisons would indicate a relationship, because the Stroop test assesses frontal lobe functions, as does the FAB.

Table 1 .
Socio-demographic characteristics of the study group.

Table 2 .
FAB, Stroop Test and MMSE scores of the study group.

Table 3 .
Internal consistency of the FAB.

Table 4 .
Concurrent validity of FAB and MMSE

Table 5 .
Concurrent Validity of FAB and Stroop Tests.
**Pearson Correlation coefficients (all comparisons are significant at the level of p <0.01) **Pearson Correlation coefficients (all comparisons are significant at the level of p <0.05) Figure 1.Relationship between FAB test-retest results (r=0.89)

Table 6 .
Average FAB scores by age groups (whole study group).

Table 7 .
Average FAB scores by gender (whole study group).

Table 8 .
Average FAB scores by education level (whole study group).

Table 9 .
Results of related studies