Objective

To validate the TNI-93 test in illiterate and low-educated subjects by setting cutoff scores to discriminate non-demented and demented subjects in a clinical setting (CESILL) and verifying the adequacy of these cutoff scores in a population-based study (AMI cohort).

Method

We used two study samples. First, a clinical setting (CESILL) comprising normal elderly participants and demented patients, mostly multicultural, low educated, or illiterate, was used to compute the cutoff scores of TNI-93 for the detection of dementia. Second, the AMI cohort, a population-based cohort of retired farmers living in a rural setting, was used as a replication study, to assess the detection properties of the cutoff scores in a different population composed mostly of low-educated older people.

Results

When combining the two scores, that is, free recall <6 or total recall <9, TNI-93 can detect dementia with a high sensitivity (87%) and specificity (96%), in the CESILL setting. These cutoff scores were roughly similar in the AMI cohort with high sensitivity (80% sensitivity) and specificity (81% specificity). In both study samples, the level of education had no effect on performance.

Conclusions

The TNI-93 appears to be a good test to detect dementia. The absence of a significant effect of education level on the performances makes the TNI-93 a tool of choice in the screening of dementia in illiterate/low-educated subjects.

Introduction

Illiteracy refers to the inability to read or write a simple message (UNO). Worldwide, one in five adults is still not literate (UNESCO, 2008). This frequency is higher, by about 20%, among the elderly who are also the most concerned by lower levels of education (UNESCO, 2008). In France, as in other Western countries, the illiteracy rate is about 10% of the total population (18–60 years; INSEE, 2013; UNESCO, 2008).

Low education and illiteracy are considered as major risk factors of developing Alzheimer's disease (AD) mainly in relation with a low cognitive reserve (Stern et al., 1994; Zhang et al., 1990). However, the cognitive evaluation of these subjects raises major difficulties due to the unsuitability of the tests used (Brucki, 2010). Most standard neuropsychological tests rely on reading and writing capacities and educational knowledge. In a comparative study of illiterate and educated subjects, educated subjects were found to outperform illiterates on all cognitive measures (Ostrosky, Ardila, Rosselli, López-Arango, & Uriel-Mendoza, 1998). Even the Mini-Mental State Examination (MMSE) is influenced by schooling and the cutoff scores are not valid in illiterate populations (Kalafat, Hugonot-Diener, & Poitrenaud, 2003). There is therefore a crucial need for adequate tools for the neuropsychological testing of illiterate and low-educated patients.

In a previous study, we designed and normalized a memory test (the TNI-93, which is the French acronym of “Test des Neuf Images du 93”, i.e., Nine Images test of the district of Seine-Saint-Denis) specially designed for illiterate and low-educated subjects (Dessi et al., 2009). The TNI-93 was standardized with outpatients at the “Centre d'Examen de Santé de Bobigny” (CES) aged over 60; 40% of the participants were women. The sample consisted of a majority of participants with a low level of education: more than 1 in 10 had never attended school and less than 1 in 3 had a late primary level of schooling (Dessi et al., 2009). Most participants memorized the nine images at the first attempt. There was no ceiling effect in free recall (FR) but a ceiling effect was found for total recall (TR). During the standardization study, the performances did not show any significant differences between participants depending on their grade level. There was a difference in performance between men and women only in FR, with women obtaining significantly better scores. Furthermore, this study showed that the TNI-93 presents many advantages for a low-educated population: administration of the test is rapid, the instructions are very simple, no written language is used and encoding is multimodal.

The goal of this study was to validate this test in illiterate and low-educated subjects by setting cutoff scores to discriminate non-demented and demented subjects in a clinical setting (CESILL); and verifying the adequacy of these cutoff scores in a population-based study (AMI cohort).

Methods

Material

The TNI-93 is an episodic memory test derived from the Memory Impairment Screening (MIS; Buschke et al., 1999) test and the Picture-Based MIS (Verghese et al., 2012). Subjects are asked to learn images using the principle of encoding specificity (Tulving & Thomson, 1973). The TNI-93 does not require the use of written language or the ability to speak French, and can therefore be administered in any language. The instructions are very simple and the duration of the test is around 5 min. During the encoding stage, nine images are displayed on the same sheet (a duck, a bike, a guitar, a carrot, an ear, a chair, a grape, a shoe, and a fork). The images were selected after a preclinical test conducted with patients at the CES during a medical checkup provided by the French National Health Service (Sécurité Sociale). The pictures were shown to a group of volunteers (different from those enrolled in the validation study) aged over 60 to ensure that all the images were easily recognized. A pool of black and white line drawings taken from Snodgrass & Vanderwart (1980) was chosen based on high familiarity items, which were all recognized and named by all the low-educated volunteers (most of whom had no educational qualifications). Each drawing in the set belonged to a different semantic category. A second pretest showed that choosing nine images (instead of 4 as in the MIS; Buschke et al., 1999, or 16 in the FCRST; Grober, Buschke, Crystal, Bang, & Dresner, 1988) offered a good compromise between ease of administration of the test and the lack of ceiling effect during the FR phase. Participants first complete a naming task according to the semantic category given by the examiner (animal, transportation, musical instrument, vegetable, part of the face, furniture, fruit, “something which is used to dress,” kitchen utensil; if the subject does not understand the name of this category, the examiner may repeat it with the following category: “an item used for eating”). After having masked the images, a cued recall (CR) is immediately proposed by giving the semantic category as a cue (the same cue as the one used during the encoding stage). If not all items are retrieved during the immediate CR, the images are displayed a second time with the same encoding procedure as before (a third time is possible if needed). Participants are then asked to count backwards 3 × 3 from 40, during 20 s (or to say the days of the week backwards for participants unable to count), as an interfering task. Finally, a FR is proposed during 2 min (score ranging from 0 to 9) followed by a  CR (score ranging from 0 to 9) for the forgotten items. Intrusions are reported for all the steps. TR is the sum of FR and CR (score ranging from 0 to 9).

Population

We used two study samples. First, a clinical setting (CESILL), comprising normal elderly participants and demented patients, both mostly multicultural, low educated, or illiterate, was used to compute the cutoff scores of TNI-93 for the detection of dementia. Second, the AMI cohort, a population-based cohort of elderly retired farmers living in rural areas, was used as a replication study to assess the detection properties of the cutoff scores in a different population composed mostly of low-educated older people.

Clinical Setting (CESILL)

CESILL is a clinical setting of 369 subjects aged 60 and more composed of 2 groups, 282 normal elderly participants recruited at the CES and 87 patients with dementia recruited at the Memory Clinic of Avicenne Hospital. All participants lived in the Seine-Saint-Denis district in France. Both groups were mostly multicultural, low educated and/or illiterate, and were either born in France, or are migrants from former French colonies (schooling in French) or migrants from other countries. The proficiency in French of the migrants was variable and most were bilingual or multilingual.

During the 32-month study period, the 282 normal subjects were consecutively included in this study. They underwent a medical checkup at the CES provided by the French National Health Service during which the TNI-93 test was administered. All the participants were examined by the neurologist of the center who excluded a diagnosis of dementia or psychiatric illness based on the clinical criteria of the DSM-IV (American Psychiatric Association, 2000). A memory complaint was either present or absent. Exclusion criteria included severe audiovisual loss or the presence of active medical, neurological, or psychiatric illness, which would interfere with completion of the study procedures. Cases with suspicion of memory disorders were not included in the group of normal subjects in the CESILL setting and were referred to the nearby Memory Clinic.

The pathological group of the CESILL setting consisted of 87 patients who were referred for cognitive disorders by their General Practitioner to the Memory Clinic of Avicenne Hospital. Patients lived in the same area as normal participants and had the same socioeconomic and cultural characteristics. The diagnosis of dementia and of AD was based on the DSM-IV criteria (American Psychiatric Association, 2000). In the patient group, the diagnosis was assessed by physicians blind to the results of the TNI-93.

Population-Based Cohort (AMI)

The AMI cohort is an epidemiological cohort on aging and health including 1,002 elderly farmers living in rural areas and randomly selected from the Farmer Health Insurance rolls. Selection criteria were: aged 65 years and older; living in a rural area in Gironde, South-Western France; retired from farming after at least 20 years of activity, and affiliated to the Health Insurance. The detailed methodology has been described elsewhere (Pérès et al., 2012). Briefly, the study started in 2007, with three follow-up visits conducted up to 2013. Visits were conducted at home by a neuropsychologist then by a geriatrician for all cases suspected of dementia, Parkinson's disease and depression (to confirm the diagnosis), or by a nurse for others. Informed and written consent were obtained at the beginning of the visit for all participants. Those individuals suspected of presenting neurocognitive disorders on the basis of the neuropsychologist's examination received an additional visit by a geriatrician who conducted a clinical examination to confirm or exclude the diagnosis of dementia and specify the etiology. Then, a case consensus conference attended by the geriatrician in charge of the visit and three other dementia specialists was conducted to finally confirm or exclude the diagnosis. The etiology was assigned according to the National Institute of Neurological and Communication Disorders and Stroke/AD and Related Disorders Association (NINCDS-ADRDA) criteria for AD, the National Institute of Neurologic Disorders and Stroke/“Association Internationale pour la Recherche et l'Enseignement en Neurosciences” (NINCDS-AIREN) criteria for vascular dementia and the Diagnostic and Statistical Manual of Mental Disorders Third Edition Revised (DSM-III-R) for Parkinson's disease dementia. The following data were collected during a 2-hr interview: age, gender, marital status, educational level, income, professional activities, social and material living environment, and currently used medication. The interviewer also assessed restriction in activities of daily living (ADL) through several scales, including basic ADL (Katz, Downs, Cash, & Grotz, 1970), Instrumental ADL (IADL; Lawton & Brody, 1969), and mobility (Rosow, Breslau, & Guttman, 1966). Anxiety state was assessed with the State Trait Anxiety Inventory (STAI; Spielberger, Gorshu, & Lushene, 1983), and depressive signs with the Centre for Epidemiologic Studies-Depression scale (CES-D; Radloff, 1977). A battery of neuropsychological tests was administered exploring subjective cognitive complaints according to the QPC scale (Thomas Antérion et al., 2003); the MMSE (Folstein, Folstein, & McHugh, 1975) for global cognitive performance; the Free and Cued Selective Reminding Test RL/RI-16 items (FCSRT; Grober et al., 1988) and the story recall subtest of the Wechsler memory scale (Wechsler, 1987) for episodic memory; the visual Delayed Matching-to-Sample task (Barbeau et al., 2004) for visual recognition; the Goblets test (Mokri et al., 2013) for visuo-spatial working memory; the Digit Symbol Substitution Task (Wechsler, 1981) for psychomotor speed; the Wechsler Similarities test (Wechsler, 1981) for abstract thinking; and the Isaacs Set Test (Isaacs & Akhtar, 1972) for verbal fluency. At the first follow-up visit, the TNI-93 was added to the neuropsychological battery.

Statistical Analysis

Age was dichotomized in 2 categories: below 70 versus 70 years and above. Educational level was defined in three categories: participants who had no schooling, those who had a low education level (i.e., primary school level with or without the French primary school certificate, corresponding approximately to 7 years of schooling), and those with a higher educational level (secondary and university degree). The effect of age, gender, and educational level on TNI-93 performances was assessed in the CESILL setting by t-tests and Kruskal–Wallis tests.

The ability of TNI-93 to detect dementia was assessed by calculating sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). The Youden index was used to identify the cutoff score for which the best compromise between sensitivity and specificity was achieved. The cutoff scores were calculated in the CESILL setting and then replicated in the AMI cohort.

Analyses were performed using SAS version 9.1 (SAS Institute, Inc., Cary, North Carolina).

Results

Sample Characteristics

General characteristics of the two samples are presented in Table 1.

Table 1.

Socio-demographic characteristics and TNI-93 scores of CESILL and AMI samples

 CESILL setting (N = 369) AMI cohort (N = 651) 
Mean age years (SD70.3  (7.2) 77.6  (6.2) 
 N  N  
Gender       
 Male 177  48.0 404  62.1 
 Female 192  52.0 247  37.9 
Educational level     
 No schooling 43  11.7 25  3.8 
 Low education 143  38.8 493  75.7 
 High education 183  49.6 133  20.4 
TNI-93 scores       
 FR mean score (SD6.6 min = 0; max = 9 (2.4) 6.8 min = 0; max = 9 (1.9) 
 CR mean score (SD1.7 min = 0; max = 7 (1.5) 1.9 min = 0; max = 9 (1.5) 
 TR mean score (SD8.4 min = 1; max = 9 (1.6) 8.7 min = 1; max = 9 (0.9) 
 CESILL setting (N = 369) AMI cohort (N = 651) 
Mean age years (SD70.3  (7.2) 77.6  (6.2) 
 N  N  
Gender       
 Male 177  48.0 404  62.1 
 Female 192  52.0 247  37.9 
Educational level     
 No schooling 43  11.7 25  3.8 
 Low education 143  38.8 493  75.7 
 High education 183  49.6 133  20.4 
TNI-93 scores       
 FR mean score (SD6.6 min = 0; max = 9 (2.4) 6.8 min = 0; max = 9 (1.9) 
 CR mean score (SD1.7 min = 0; max = 7 (1.5) 1.9 min = 0; max = 9 (1.5) 
 TR mean score (SD8.4 min = 1; max = 9 (1.6) 8.7 min = 1; max = 9 (0.9) 

FR = free recall; TR = total recall; CR = cued recall

The CESILL setting consisted of 369 participants (282 normal participants and 87 patients). The mean age was 70.3 years (SD 7.2), 48.0% of participants were men. Both in the CESILL and AMI samples, the number of years of education ranged from 0 to 14 years of education, 49.6% of participants had a “high level” of education (defined here as more than 8 years of education) and 11.7% had never attended school.

The TNI-93 was administered at the 2-year follow-up visit of the AMI cohort. Of the 731 eligible participants seen at this visit, 651 completed the TNI-93. Their mean age was 77.6 years (SD 6.2), 62.1% of the participants were men. Only 20.4% of participants had a “high level” of education and 3.8% had no schooling.

Effect of Age, Sex, and Educational Level on TNI-93 Performances

Table 2 shows the effect of age, gender, and education with TNI-93 performances in the CESILL setting and in the AMI cohort. In the CESILL setting, subjects with dementia have significantly lower performance than control subjects regardless of age, sex, and education level (p < .0001). As can be seen, age was significantly associated with both FR and TR performances (i.e., performances decreased with age). Gender was found to be associated with TR performances, men having higher scores than women. There was no significant effect of level of education on TNI-93 performances.

Table 2.

Effect of age, sex, and educational level on TNI-93 performances: (a) healthy controls versus patients with dementia for FR score in CESILL setting, (b) healthy controls versus patients with dementia for TR score in CESILL setting, (c) for overall in the setting CESILL, and (d) for all subjects in the AMI cohort

 TNI: healthy controls TNI: patients with dementia p-Value 
Mean SD Mean SD 
(a) CESILL setting: FR score 
Age      
 <70 years 7.86 0.93 3.73 2.22 <.0001 
 ≥70 years 7.53 1.18 2.46 2.10 <.0001 
Gender      
 Male 7.63 1.03 2.79 2.15 <.0001 
 Female 7.83 1.05 3.71 2.31 <.0001 
Educational level      
 No schooling 7.79 1.03 2.59 0.61 <.0001 
 Low education 7.61 1.11 2.51 0.83 <.0001 
 High education 7.81 1.00 2.39 0.88 <.0001 
(b) CESILL setting: TR score 
Age      
 <70 years 8.97 0.17 6.90 2.56 <.0001 
 ≥70 years 8.96 0.23 5.49 2.83 <.0001 
Gender      
 Male 8.97 0.22 5.90 2.79 <.0001 
 Female 8.97 0.17 6.80 2.67 <.0001 
Educational level      
 No schooling 8.96 0.19 4.41 1.86 <.0001 
 Low education 8.98 0.13 4.27 1.89 <.0001 
 High education 8.96 0.23 4.10 1.85 <.0001 
 TNI: healthy controls TNI: patients with dementia p-Value 
Mean SD Mean SD 
(a) CESILL setting: FR score 
Age      
 <70 years 7.86 0.93 3.73 2.22 <.0001 
 ≥70 years 7.53 1.18 2.46 2.10 <.0001 
Gender      
 Male 7.63 1.03 2.79 2.15 <.0001 
 Female 7.83 1.05 3.71 2.31 <.0001 
Educational level      
 No schooling 7.79 1.03 2.59 0.61 <.0001 
 Low education 7.61 1.11 2.51 0.83 <.0001 
 High education 7.81 1.00 2.39 0.88 <.0001 
(b) CESILL setting: TR score 
Age      
 <70 years 8.97 0.17 6.90 2.56 <.0001 
 ≥70 years 8.96 0.23 5.49 2.83 <.0001 
Gender      
 Male 8.97 0.22 5.90 2.79 <.0001 
 Female 8.97 0.17 6.80 2.67 <.0001 
Educational level      
 No schooling 8.96 0.19 4.41 1.86 <.0001 
 Low education 8.98 0.13 4.27 1.89 <.0001 
 High education 8.96 0.23 4.10 1.85 <.0001 
 TNI FR score TNI TR score 
Mean SD p-Value Mean SD p-Value 
(c) CESILL setting: overall 
Age       
 <70 years 7.49 1.62 <.0001 8.76 1.14 <.0001 
 ≥70 years 5.78 2.74  7.99 1.94  
Gender       
 Male 6.83 2.11 .16 8.59 1.29 .02 
 Female 6.48 2.63  8.19 1.87  
Educational level       
 No schooling 6.12 2.75 .33 7.77 2.37 .07 
 Low education 6.63 2.33  8.49 1.50  
 High education 6.79 2.36  8.44 1.49  
(d) AMI cohort: all subjects 
Age       
 <77 years 7.26 1.48 <.0001 8.87 0.60 .0003 
 ≥77 years 6.41 2.10  8.61 1.13  
Gender       
 Male 6.66 1.82 .002 8.75 0.80 .88 
 Female 7.13 1.88  8.73 1.06  
Educational level       
 No schooling 6.68 1.97 .55 8.72 0.84 .20 
 Low education 6.78 1.93  8.70 1.00  
 High education 7.09 1.51  8.88 0.41  
 TNI FR score TNI TR score 
Mean SD p-Value Mean SD p-Value 
(c) CESILL setting: overall 
Age       
 <70 years 7.49 1.62 <.0001 8.76 1.14 <.0001 
 ≥70 years 5.78 2.74  7.99 1.94  
Gender       
 Male 6.83 2.11 .16 8.59 1.29 .02 
 Female 6.48 2.63  8.19 1.87  
Educational level       
 No schooling 6.12 2.75 .33 7.77 2.37 .07 
 Low education 6.63 2.33  8.49 1.50  
 High education 6.79 2.36  8.44 1.49  
(d) AMI cohort: all subjects 
Age       
 <77 years 7.26 1.48 <.0001 8.87 0.60 .0003 
 ≥77 years 6.41 2.10  8.61 1.13  
Gender       
 Male 6.66 1.82 .002 8.75 0.80 .88 
 Female 7.13 1.88  8.73 1.06  
Educational level       
 No schooling 6.68 1.97 .55 8.72 0.84 .20 
 Low education 6.78 1.93  8.70 1.00  
 High education 7.09 1.51  8.88 0.41  

In the AMI cohort, age was also found to be significantly associated both with both FR (p < .0001) and TR (p = .0003) scores. No effect of education was observed for FR (p = .55) or for TR (p = .2) scores.

Properties of TNI-93 for Detecting Dementia

The Youden index was calculated to identify the best threshold value of FR and total scores for the detection of dementia. The cutoff values were calculated in the CESILL setting. They were calculated in the entire setting and for each of the three education groups. The same cutoff values were found for the whole setting and for the three groups: 6 for the FR score and 9 for the TR score.

Table 3 shows the sensitivity, specificity, PPV, and NPV obtained with these cutoffs in the entire setting. Using only the FR score, the sensitivity was 85%, the specificity 98%, the PPV 93% and the NPV 96%. When combining FR and TR scores, FR <6 or TR <9, the sensitivity was slightly higher (87%), and the specificity slightly lower (96%).

Table 3.

Sensitivity, specificity, PPV,  NPV, and Youden index of TNI-93 scores for the detection of dementia in the two samples (CESILL setting and AMI cohort) and in samples with MMSE score < or ≥20 in the AMI cohort

 CESILL setting AMI cohort 
FR < 6 TR < 9 FR < 6 or TR < 9 FR < 6 TR < 9 FR < 6 or TR < 9 
Sensitivity 85% 68% 87% 68% 54% 80% 
Specificity 98% 97% 96% 86% 90% 81% 
PPV 93% 88% 86% 32% 34% 29% 
NPV 96% 91% 96% 96% 95% 98% 
Youden index 0.83 0.65 0.83 0.53 0.44 0.60 
 CESILL setting AMI cohort 
FR < 6 TR < 9 FR < 6 or TR < 9 FR < 6 TR < 9 FR < 6 or TR < 9 
Sensitivity 85% 68% 87% 68% 54% 80% 
Specificity 98% 97% 96% 86% 90% 81% 
PPV 93% 88% 86% 32% 34% 29% 
NPV 96% 91% 96% 96% 95% 98% 
Youden index 0.83 0.65 0.83 0.53 0.44 0.60 
 Sample with MMSE score <20 (N = 41) Sample with MMSE score ≥20 (N = 606) 
FR < 6 TR < 9 FR < 6 or TR < 9 FR < 6 TR < 9 FR < 6 or TR < 9 
Sensitivity 81% 69% 92% 58% 42% 70% 
Specificity 80% 80% 67% 86% 90% 81% 
PPV 88% 86% 83% 19% 19% 17% 
NPV 71% 60% 83% 97% 96% 98% 
Youden index 0.61 0.49 0.59 0.43 0.32 0.51 
 Sample with MMSE score <20 (N = 41) Sample with MMSE score ≥20 (N = 606) 
FR < 6 TR < 9 FR < 6 or TR < 9 FR < 6 TR < 9 FR < 6 or TR < 9 
Sensitivity 81% 69% 92% 58% 42% 70% 
Specificity 80% 80% 67% 86% 90% 81% 
PPV 88% 86% 83% 19% 19% 17% 
NPV 71% 60% 83% 97% 96% 98% 
Youden index 0.61 0.49 0.59 0.43 0.32 0.51 

MMSE = Mini-Mental State Examination; PPV = positive predictive value; NPV = negative predictive value.

Thresholds were applied in the replication cohort (AMI cohort). The highest Youden index was obtained when combining the two TNI-93 scores (FR and TR scores). The respective values for sensitivity, specificity, PPV, and NPV were 80%, 81%, 29%, and 98%.

Properties of TNI-93 in Different Rates of Dementia

Mean MMSE scores of participants in the CESILL and AMI cohorts were, respectively, 20.3 (SD = 5.2) and 20.5 (SD = 3.9). Table 3 shows the sensitivity, specificity, PPV, NPV, and Youden index of TNI-93 scores for dementia detection in two samples of the AMI cohort according to the MMSE score: those participants with a MMSE score <20 and those with a score ≥20. In the group with a MMSE score ≥20, the TNI-93 scores of FR <6 or TR <9 score had a sensitivity of 92%, a specificity of 67%, a PPV of 83%, and a NPV of 83%. In patients with a MMSE score <20, the TNI-93 thresholds show less sensitivity and PPV (respectively, 70% and 17%) but a better specificity and NPV (respectively, 81% and 98%).

Discussion

The clinical diagnosis of dementia is mostly based on a clinical history of cognitive changes as evidenced by cognitive testing, and on their consequences on ADL (American Psychiatric Association, 2000; American Psychiatric Association, 2013; McKhann et al., 2011). Accurate neuropsychological assessment is essential in dementia diagnosis, and the most cost-efficient way for distinguishing age-related changes from pathological cognitive changes (Cullum et al., 2000).

Schooling has a major impact, not only on the ability to read and write, but also on how cognition will develop across life (Brucki, 2010). As underlined earlier, literacy and schooling influence performance on almost all cognitive tests and educated subjects outperform illiterates and low-educated subjects (Ardila & Rosselli, 1989; Ostrosky et al., 1998; Rosselli, Ardila, & Rosas, 1990), leading to the risk of mis- and over-diagnosis in less educated subjects (Teng, 2002). Moreover, the validity and reliability of a test used with individuals of different cultural or linguistic groups, who were not included in the standardization group, are questionable.

The TNI-93 is a new easy-to-administer tool designed to be used in different cultural contexts as well as with illiterate individuals. In a previous study, we showed that TNI-93 performances in normal subjects, aged 60 and more, were influenced by age but not by educational level (Dessi et al., 2009). In this study, we replicated this result in the CESILL setting composed of multicultural illiterate/low-educated urban participants with various levels of proficiency in French as well as in the AMI cohort group composed of French-speaking low-educated retired farmers living in rural areas. It is important to stress the absence of influence of education, as in the literature, in most memory tests, the effect of school attainment is considered to be greater than the effect of age (Ostrosky et al., 1998). This absence of educational level effect on TNI-93 performances reinforces its adequacy to assess a large variety of patients regardless of their educational background or absence of schooling.

CESILL is a clinical setting and AMI a community-based sample. One advantage of this study is that detection thresholds were determined in a clinical setting and then replicated in a community-based sample. Table 3 shows sensitivity, specificity, PPV, and NPV to screen dementia based on the two samples (clinical and replicate in a community cohort). Regarding its properties in the detection of dementia, the TNI-93 appears to be a very good test to screen dementia. Using the Youden index, we showed that it has a very good sensitivity (87%) and specificity (96%; Youden index: 0.83) when combining the TR and the FR scores (TR <9 or FR <6) in the CESILL setting and also in the AMI cohort (80% sensitivity and 81% specificity). For a clinical setting such as CESILL, this cutoff score with high PPV can be used to avoid false-positive cases when recruiting candidates for clinical trials. The TNI-93 can be a very interesting tool in the specialty clinic and it can be a good measure to screen dementia in primary care. However, this test does not seem to be appropriate in a community setting as suggested by the low PPV in the AMI community setting.

Table 3 shows that for patients who have no or only a slight cognitive decline (with MMSE score ≥20), the TNI-93 appears to be more specific than sensitive, whereas it is the opposite for patients who have a greater cognitive impairment (with MMSE score <20).

However, it is important to underline that this screening test does not provide any information regarding dementia etiology (e.g., AD or other types of dementia). The TNI was not designed to contribute to the differential diagnosis of different types of dementia but rather to determine whether the patient suffers from dementia or not. The question of whether the detection capabilities of the TNI are equivalent for all types of dementia could be an interesting issue to address in further studies. Likewise, TNI-93 has not been evaluated in the pre-dementia stages, so its properties to detect the earliest signs and to predict progression of MCI to dementia remain to be tested in future studies.

Therefore, the TNI-93 appears to be an education free test. Nonetheless, the question now is to what extent it can be considered as a culture free test. In this study, we have shown that the performances are identical in two different cultural groups (a multicultural illiterate/low-educated urban population with various levels of proficiency in French [CESILL setting], and a French-speaking low-educated retired farming population living in rural areas [AMI cohort]). This work is currently being extended to the French West Indies and India.

Funding

The AMI project was funded by AGRICA (CAMARCA, CRCCA, CCPMA PREVOYANCE, CPCEA, AGRI PREVOYANCE), la Mutualité Sociale Agricole (MSA) de Gironde, la Caisse Centrale de la Mutualité Sociale Agricole (CCMSA).

Conflict of Interest

None declared.

Acknowledgements

Study sponsors played no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript.

References

American Psychiatric Association
(
2000
).
Diagnostic and statistical manual of mental disorders
  (
4th ed. (text revised)
).
Washington, DC
:
Author
.
American Psychiatric Association
(
2013
).
Diagnostic and statistical manual of mental disorders
  (
5th ed.
).
Washington, DC
:
Author
. .
Ardila
,
A.
, &
Rosselli
,
M.
(
1989
).
Neuropsychological assessment in illiterates: Visuospatial and memory abilities
.
Brain and Cognition
 ,
11
,
147
166
.
Barbeau
,
E.
,
Didic
,
M.
,
Tramoni
,
E.
,
Felician
,
O.
,
Joubert
,
S.
,
Sontheimer
,
A.
, et al
. (
2004
).
Evaluation of visual recognition memory in MCI patients
.
Neurology
 ,
62
,
1317
1322
.
Brucki
,
S. M. D.
(
2010
).
Illiteracy and dementia
.
Dementia e Neuropsychologia
 ,
4
,
153
157
.
Buschke
,
H.
,
Kuslansky
,
G.
,
Katz
,
M.
,
Stewart
,
W. F.
,
Sliwinski
,
M. J.
,
Eckholdt
,
H. M.
, et al
. (
1999
).
Screening for dementia with the memory impairment screen
.
Neurology
 ,
52
,
231
238
.
Cullum
,
S.
,
Huppert
,
F. A.
,
McGee
,
M.
,
Dening
,
T.
,
Ahmed
,
A.
,
Paykel
,
E. S.
, et al
. (
2000
).
Decline across different domains of cognitive function in normal ageing: Results of a longitudinal population-based study using CAMCOG
.
International Journal of Geriatric Psychiatry
 ,
15
,
853
886
.
Dessi
,
F.
,
Maillet
,
D.
,
Metivet
,
E.
,
Michault
,
A.
,
Le Clésiau
,
H.
,
Ergis
,
A. M.
, et al
. (
2009
).
Assessment of episodic memory in illiterate elderly
.
Psychologie et Neuropsychiatrie du Vieillissement
 ,
7
,
287
296
. .
Folstein
,
M.
,
Folstein
,
S. E.
, &
McHugh
,
P. R.
(
1975
).
‘Mini-mental state’. A practical method for grading the cognitive state of patients for the clinician
.
Journal of Psychiatric Research
 ,
12
,
189
198
.
Grober
,
E.
,
Buschke
,
H.
,
Crystal
,
H.
,
Bang
,
S.
, &
Dresner
,
R.
(
1988
).
Screening for dementia by memory testing
.
Neurology
 ,
38
,
900
903
.
INSEE
(
2013
). Illettrisme: les chiffres. Exploitation par l'agence nationale de lutte contre l'illettrisme de l'enquête Information et vie quotidienne conduite en 2011-2012 [Illiteracy: data. Interpretation of the Information and daily living survey conducted in 2011-2012 by the French national agency against illiteracy]; http://www.anlci.gouv.fr/content/download/1372/29017/version/1/file/PLAQUETTE_CHIFFRES_JANVIER+2013.pdf, accessed 9/11/2015.
Isaacs
,
B.
, &
Akhtar
,
A. J.
(
1972
).
The set test: A rapid test of mental function in old people
.
Age and Ageing
 ,
1
,
222
226
.
Kalafat
,
M.
,
Hugonot-Diener
,
L.
, &
Poitrenaud
,
J.
(
2003
).
Standardisation et étalonnage français du « Mini Mental State » (MMS) version GRECO [French standardization of the Mini Mental State (MMS), GRECO's version]
.
Revue de Neuropsychologie
 ,
13
,
209
236
.
Katz
,
S.
,
Downs
,
T. D.
,
Cash
,
H. R.
, &
Grotz
,
R. C.
(
1970
).
Progress in development of the index of ADL
.
Gerontologist
 ,
10
,
20
30
.
Lawton
,
M. P.
, &
Brody
,
E. M.
(
1969
).
Assessment of older people: Self-maintaining and instrumental activities of daily living
.
Gerontologist
 ,
9
,
179
186
.
McKhann
,
G. M.
,
Knopman
,
D. S.
,
Chertkow
,
H.
,
Hyman
,
B. T.
,
Jack
,
C. R.
Jr.
,
Kawas
,
C. H.
, et al
. (
2011
).
The diagnosis of dementia due to Alzheimer's disease: Recommendations from the National Institute on Aging—Alzheimer's Association workgroups on diagnostic guidelines for Alzheimer's disease
.
Alzheimer's and Dementia
 ,
7
,
263
269
. .
Mokri
,
H.
,
Matharan
,
F.
,
Pérès
,
K.
,
Bouisson
,
J.
,
Dartigues
,
J.F.
,
Amieva
,
H.
, et al
. (
2013
).
The Goblets test: Norms in the elderly population and properties in the detection of cognitive impairment in elderly individuals selected from the general population
.
Revue de Neurologie
 ,
169
,
871
878
.
Ostrosky
,
F.
,
Ardila
,
A.
,
Rosselli
,
M.
,
López-Arango
,
G.
, &
Uriel-Mendoza
,
V.
(
1998
).
Neuropsychological test performance in illiterates
.
Archives of Clinical Neuropsychology
 ,
13
,
645
660
.
Pérès
,
K.
,
Matharan
,
F.
,
Allard
,
M.
,
Amieva
,
H.
,
Baldi
,
I.
,
Barberger-Gateau
,
P.
, et al
. (
2012
).
Health and aging in elderly farmers: The AMI cohort
.
BMC Public Health
 ,
12
,
558
. .
Radloff
,
S.
(
1977
).
The CES-D scale: A self-report depression scale for research in the general population
.
Applied Psychological Measurement
 ,
1
,
385
401
.
Rosow
,
I.
,
Breslau
,
N.
, &
Guttman
,
A.
(
1966
).
Health scale for the aged
.
The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences
 ,
21
,
556
559
.
Rosselli
,
M.
,
Ardila
,
A.
, &
Rosas
,
P.
(
1990
).
Neuropsychological assessment in illiterates II: Language and praxic abilities
.
Brain and Cognition
 ,
12
,
281
296
.
Snodgrass
,
J. G.
, &
Vanderwart
,
M.
(
1980
).
A standardized set of 260 pictures: Norms for name agreement, image agreement, familiarity, and visual complexity
.
The Journal of Experimental Psychology: Human Perception and Performance
 ,
6
,
174
215
.
Spielberger
,
C. D.
,
Gorshu
,
R. L.
, &
Lushene
,
R. D.
(
1983
).
Manual for the state-trait anxiety inventory (Form Y)
 .
Palo Alto
:
Consulting Psychologist Press
.
Stern
,
Y.
,
Gurland
,
B.
,
Tatemichi
,
T. K.
,
Tang
,
M. X.
,
Wilder
,
D.
, &
Mayeux
,
R.
(
1994
).
Influence of education and occupation on the incidence of Alzheimer's disease
.
The Journal of the Medical Association
 ,
271
,
1004
1010
.
Teng
,
E. L.
(
2002
).
Cultural and educational factors in the diagnosis of dementia
.
Alzheimer Disease and Associated Disorders
 ,
16
,
77
79
.
Thomas Antérion
,
C.
,
Ribas
,
C.
,
Honoré-Masson
,
S.
,
Berne
,
G.
,
Ruel
,
J. H.
, &
Laurent
,
B.
(
2003
).
Le questionnaire de plainte cognitive (QPC): Un outil de recherche de plainte suspecte d’évoquer une maladie d'Alzheimer [The cognitive complaint questionnaire (QPC): A tool to detect complaints raising the suspicion of Alzheimer's disease]
.
L'Année Gérontologique
 ,
17
,
56
65
.
Tulving
,
E.
, &
Thomson
,
D.
(
1973
).
Encoding specificity and retrieval processes in episodic memory
.
Psychological Review
 ,
80
,
352
373
.
UNESCO
(
2008
). Statistiques internationales sur l'alphabétisme: examen des concepts, de la méthodologie et des données actuelles [International statistics for illiteracy: study of current definitions, methods and data]; http://www.uis.unesco.org/Library/Documents/Literacyreport2008_fr.pdf, accessed 9/11/2015.
Verghese
,
J.
,
Noone
,
M. L.
,
Johnson
,
B.
,
Ambrose
,
A. F.
,
Wang
,
C.
,
Buschke
,
H.
, et al
. (
2012
).
Picture-based memory impairment screen for dementia
.
Journal of the American Geriatrics Society
 ,
60
,
2116
2120
.
Wechsler
,
D.
(
1981
).
WAIS manual
 .
New York
:
Psychological Corporation
.
Wechsler
,
D.
(
1987
).
Wechsler memory scale—revised edition. Manual
 .
New York
:
The Psychological Corporation
.
Zhang
,
M. Y.
,
Katzman
,
R.
,
Salmon
,
D.
,
Jin
,
H.
,
Cai
,
G. J.
,
Wang
,
Z. Y.
, et al
. (
1990
).
The prevalence of dementia and Alzheimer's disease in Shanghai, China: Impact of age, gender, and education
.
Annals of Neurology
 ,
27
,
428
437
.