The DemTect has a maximum transformed score of 18. The selection of this maximum score was random. For each subtest, transformation tables for two age groups (<60 years and ≥60 years) were provided for the first version of the DemTect. The maximum scores for each subtest range from 3 (word list, number transcoding, digit span) to 4 (verbal fluency) up to 5 (delayed recall). The decision on each maximum score was based on the subtests’ different sensitivities and specificities in a population of healthy control subjects, AD patients, and MCI patients [1, 2]. The age correction was necessary due to significant age effects in the control groups in both normative studies. Furthermore, an education correction is provided in the English version [2]. Here, it was defined that one point is added to the transformed total score in subjects with only basic education (≤11 years).

After much feedback from clinicians that the DemTect is frequently used in elderly patients aged 80 years or above, but also in young patients of 40 years or younger (with a wide range of clinical states), further normative work was done by our own group [13] that has lead to norms for the age groups “40−” and “80 + .” With these scores, the total score of the DemTect is now independent of the factor age for adult patients from young adulthood until old age. The relevance of the age correction is demonstrated in Fig. 8.2.

From the transformed total DemTect scores, it can be decided whether performance of the subject can be interpreted as age adequate (13–18 points), or whether MCI (9–12 points) or dementia must be suspected (≤8 points) (Table 8.2). Again, these scores were derived from the normative studies and show high sensitivity and specificity [1, 2].

It is important to emphasize that any interpretation from a screening tool must be preliminary; especially if a cognitive disorder is indicated, an elaborate neuropsychological examination is strongly recommended.

The administration time for the DemTect, including transformation of the raw scores and interpretation, is 8–10 min.

When patients are retested in follow-up examinations, explicit or implicit learning effects can occur when the same test versions are used. Thus, a parallel version of the DemTect, “DemTect B,” was developed [14].

Parallel versions of the five original DemTect subtests were designed (modifications are indicated in Table 8.1). The equivalence of the new and original subtests was analyzed in 80 healthy control subjects. There were no significant differences between the corresponding subtests of the two test versions except for the semantic verbal fluency task (category “supermarket” in DemTect and category “animals” in DemTect B) (Fig. 8.3). Thus, different algorithms for transforming raw scores into transformed scores were calculated for this subtest. For all other subtests, the transformation tables of the original DemTect can be used. Using this procedure, there were no significant differences between the transformed scores of the DemTect and DemTect B, including the total scores (max. 18 points, mean score 15.9, SD 1.9 in DemTect versus 15.5, SD 2.4 in DemTect B). Thus, the interpretation of specific score ranges of the DemTect could be adopted for DemTect B, and the total DemTect B can be regarded as equivalent to the DemTect.

Besides the two normative studies for the German and English version of the DemTect [1, 2], some other studies have demonstrated a high sensitivity and specificity of the tool (overview in Table 8.3) [15]. The sensitivity across all studies ranges between 83 and 100 % for AD patients, 67 and 86 % for patients with MCI or mild cognitive disorder, and was 90 % for vascular dementia (VaD) patients; the specificity ranged between 90 and 100 % [1, 2, 16–18]. In a validation of the DemTect with 18-fluoro-2-deoxyglucose positron emission tomography (18-FDG-PET), the ROC analysis showed an area under the curve (AUC) of 0.78 with a cutoff score of ≤13 (95 % CI 0.62–0.94; p = 0.006) [18].

The DemTect total transformed score is highly correlated with the MMSE (e.g. [2]; control group: p < 0.001, r = 0.43; AD group: p < 0.001, r = 0.55; MCI group: p < 0.01, r = 0.31). However, a regression analysis showed that although DemTect scores could be transformed into MMSE scores with the formula MMSE = 0.567 × DemTect score plus 19.997, DemTect scores only corresponded to MMSE scores higher than 20. This result reflects the fact that while the MMSE is a tool with which staging up to more severe stages of dementia is possible, the DemTect is a tool that is valuable for detecting and differentiating cognitive dysfunction when symptoms begin. Accordingly, the superiority of the DemTect compared to the MMSE regarding the sensitivity to assess early symptoms has been demonstrated [2, 16].

A good retest reliability with no significant differences in total transformed scores in 30 healthy controls which were tested two times with a time interval of 6 weeks (mean scores were 16.63 at t1 and 17.13 at t2) was shown [1].

Neural correlates of the DemTect’s five subtests regarding both gray matter brain atrophy and cerebral glucose metabolism were examined in 21 AD patients, 14 patients with frontotemporal lobar degeneration (FTLD), and 13 patients with subjective cognitive impairment (SCI) with structural magnetic resonance imaging (MRI) and F-18-fluorodeoxyglucose positron emission tomography (FDG-PET) by Woost et al. [19]. When all diagnostic groups were analyzed together, performance in the word list was positively correlated with glucose metabolism in the left temporal lobe. The number transcoding task was significantly related to glucose metabolism in a predominantly left lateralized frontotemporal network as well as a parietooccipital network including parts of the basal ganglia. Number transcoding was also associated with gray matter density in an extensive network including frontal, temporal, parietal and occipital areas. Working memory, tested with the digit span reverse, correlated with glucose metabolism in the left frontal cortex, the bilateral putamen, the head of caudate nucleus and the anterior insula. The only subtest for which no relationships with gray matter or glucose metabolism could be found was the supermarket task. Separate correlation analyses for the diagnostic groups partly verified or extended the correlates found for the overall sample analysis. The authors emphasize that their study serves as an external validation of the DemTect.