Sensitivity and specificity of the Bamberg Dementia Screening Test’s (BDST) full and short versions: brief screening instruments for geriatric patients that are suitable for infectious environments

In a large sample of 1295 participants, it could be shown that the BDST shows high sensitivity and sufficient specificity regarding the detection of mild dementia. Thereby, the BDST cutoff score of ≤ 37/50 for mild dementia obtained in a smaller sample of a previous study [22] could be cross-validated.

The full version of the BDST proved superior to the MMSE in detecting dementia. This can be ascribed mainly to higher sensitivity values. For instance, the sensitivity of the MMSE regarding mild forms of dementia is .76, indicating that roughly a quarter of patients might remain undetected, while the corresponding sensitivity value for the BDST of .93 is significantly higher.

Even a shortened BDST version consisting of only the first two subtests was shown to be comparable concerning diagnostic quality while surpassing the MMSE with respect to brevity (16 vs. 30 items), the necessity of test materials (no additional test materials needed vs. pencil, watch and two extra sheets of paper needed) and testing time (less than two minutes vs. seven to eight minutes). Notably, the first subtest, requiring less than 1 min of administration time, already proved to detect possible moderate dementia with sufficient sensitivity and specificity.

Based on the sensitivity and specificity values found in this study, high negative predictive values would be expected even if the proportion of dementia cases were up to 40% in the population screened with the BDST.

Table 4 suggests that the BDST_s has higher sensitivity, but lower specificity than the MMSE, meaning it leads to more false-positive cases. While the PPVs in Table 4 are excellent (which may be a result of its high sensitivity but surely is also due to the fact that there are many more dementia cases than CNT cases in our sample), the NPVs, like for the MMSE, are not satisfactory. Therefore, the BDST_s should be used as a first-stage screening tool. In case the BDST_s is positive, patients can be referred to a specialist for a more thorough diagnosis. On the other hand, if the BDST_s is negative, the remaining subtests of the BDST should also be administered, at least in settings where many dementia cases are to be expected (see also Fig. 2 on this topic).

It can be argued that by lowering cutoffs, a better balance between SN and SP could be obtained. However, lowering the cutoff to ≤ 36/50 decreases sensitivity from .93 to .88 and increases specificity from .82 to .85, which would lead to an overall worse diagnostic accuracy. We also placed more emphasis on a high detection rate (despite risking a higher false-positive rate), so we decided to keep the cutoff at ≤ 37/50.

These results justify an adaptive algorithm for application as illustrated in Fig. 3. Any mistake in the first subtest (naming of animals) indicates moderate dementia so that the testing can be terminated, and the patient should be referred to psychiatrists and neuropsychologists.

Otherwise, if in the second subtest there is more than one error in the recall of the animals named in the first subtests (for example two animals could only be recognized but not freely recalled), either mild or moderate dementia can be suspected and the test session could also be terminated at this point.

Nevertheless, as the diagnostic quality of the BDST is higher than that of the BDST_s, the full BDST should be administered whenever possible. However, if there is too little time to perform the complete test, taking 1 or 2 min for the abbreviated versions has proven to be better than performing no screening at all. However, especially in settings with high prevalence of true positive cases, a positive BDST_s score is highly likely to be truly positive, but a negative BDST_s does not mean very much. As NPVs are low, clinicians should be cautious about over-interpreting a negative test of the BDST_s, especially if the score is close to the cutoff point.

The BDST was also able to prove sufficient sensitivity and specificity not only for Alzheimer’s but also for other types of dementia. Notably, different test profiles matching theoretical expectations for participants suffering from either Alzheimer’s or other types of dementia became apparent. As expected, the first subgroup performed worse in the subtests affecting verbal and visual memory while the latter group showed more pronounced deficits in the subtest covering cognitive flexibility. However, as can be seen in Fig. 1, the numerical differences in the scores are very small, and it should be noted that for the other three subtests, no significant differences could be found, which might be caused by the heterogeneity of the non-Alzheimer group. It was not our purpose, though, to develop a screening that can distinguish between Alzheimer’s disease and other forms of dementia but to analyze whether the BDST can detect people with dementia, even when they do not suffer from Alzheimer’s disease.

The total scores of the BDST, MMSE, and CERAD-Plus were highly intercorrelated. This was also true for the scores of CERAD-Plus and BDST subtests with similar content. This is remarkable because of the considerably different modes of administration and test behavior that make the BDST administrable even for bedridden patients. These findings provide additional evidence regarding construct validity of the screening measure introduced in this paper.

Another strength of the test is that—although cognitively healthy control patients experienced virtually no difficulties with the completion of the tasks—no ceiling effects for the BDST occurred. The average BDST score was more than two standard deviations below the BDST’s maximum score, whereas the average MMSE score obtained in our sample was only about one standard deviation below MMSE’s maximum score. Furthermore, no floor effects for subjects suffering from moderate dementia could be detected, as these participants still reached an average score of 20. Thus, the BDST allows for testing a wide range of performance levels of both weakly and strongly impaired individuals, which makes this test particularly suitable for geriatric patients. Furthermore, administration of the BDST does not require investigators to ask considerably basic questions about spatial and temporal orientation, which might be experienced as humiliating and stressful [37].

Also, educational level, gender, or age have shown no substantial effects on test performance, independently of cognitive status, which itself is affected by educational level and age.

It should not go unmentioned that the advantages of the BDST over the MMSE also apply to many alternative tests for dementia and cognitive impairment like the Montreal Cognitive Assessment (MoCA, [38]), the Addenbrooke’s Cognitive Examination III (ACE-III, [39]) or the Test Your Memory (TYM [40]), which have also been shown to outperform the MMSE and to have similar or even better sensitivity and specificity values. The BDST is shorter than most other screenings and can be administered even in cases where conventional screenings cannot be used. For example, if there is minimal time to administer a screening, the TYM, which has excellent diagnostic quality [41], could be a better choice, because, as patients are to fill in the test themselves, administration time is minimized. However, participants must be able to read, write, and draw into relatively small gaps in the test form. For example, a clock face must be drawn into a circle with a diameter of about 3.5 cm (1.2 in). For participants that are bedridden and too physically ill to meet these requirements, the short and full versions of the BDST could therefore be a viable alternative to the TYM.

It should be noted, however, that poor test performance in cognition tests per se could be due to other reasons than dementia, for example delirium, diabetes, depression, or pain [42]. Hence, prior to using cognitive screening tests, these conditions should be excluded.

Generally, neither the BDST nor any other dementia screening test can replace detailed neuropsychological testing and other diagnostic procedures such as cranial CT or MRI scans, and others. Especially in populations where dementia cases are rare, even screening tests with good sensitivity and specificity have low positive predictive values, i.e., subjects who in reality do not have dementia are to be expected to have falsely positive test results. This is why many authors argue that in clinical settings routine dementia screening is not feasible [43]. To overcome this issue, we recommend that dementia screenings should be combined with a quick and valid screening for delirium, e.g., the 4AT [44], and that the term “dementia” should be avoided until careful diagnostics have been carried out.

It should be pointed out that the data obtained for the BDST in this study and presented in this paper was drawn from a clinical sample of patients referred for neuropsychological testing. Although this might be a valid setting for the BDST in many cases (e.g., in a geriatric ward, a quick assessment might be very helpful), this has led to a high proportion of participants with the target condition (dementia), which possibly affected the PPV and NPV estimates.

Furthermore, by selecting the positivity cutoff after performing the test using ROC analyses, we have increased the risk of too optimistic accuracy estimates. However, it should be noted that this effect is strongest in smaller samples. As can be seen in Table 4, for the short and the long version of the BDST, the determined cutoff scores for mild dementia stay the same, regardless of whether the whole sample of participants with dementia or the two subsamples of participants with Alzheimer’s and other forms of dementia are considered.

Although it should be noted that the cutoff score of ≤ 37/50 is a replication of the cutoff score that was determined for a different, albeit smaller sample in a former study [22], our findings have to be cross-validated, desirably in a population-based sample. Furthermore, the development of parallel versions of the BDST would aid in the long-term evaluation of its usefulness in measuring cognitive decline in the course of illness.

When considering the clinical benefits of the BDST test, one has to realize that the identification rate of dementia reporting in hospital records ranges between 26 and 70% in European and North American countries [45‐48], which means that many patients suffering from dementia are overlooked. As no prior preparation or test materials are needed and the BDST is easy to administer and evaluate, this test might help to reduce the barriers of using neuropsychological screening tests in geriatric wards of general hospitals or residents of nursing homes. In the optional BDST smartphone app, which can be obtained from the corresponding author, the user is guided through administration, automated scoring, and interpretation of the results and provided with an automatic suggestion for a medical report text. This could help to avoid potential adverse events like falls or poor hydration, which are more likely to occur in patients suffering from dementia and affect their outcome, for example with regard to institutionalization, mortality, and length of stay [49]. Thus, screening measures like the BDST could help to improve outcome and reduce costs of stays in geriatric wards of general hospitals. The fact that no test utensils (pencils, test forms, written instructions, etc.) have to be exchanged between the examiner and the person being tested makes the BDST especially suitable for dementia screenings in infectious environments or remote telemedical testing via video calls. Thus, in rural regions, where no specialized personal is available, a more timely diagnosis of dementia could be initiated.

In summary, the data presented here yields evidence that “true bedside” measures like the BDST may qualify as valid screening measures for the diagnosis of dementia. Especially in geriatric settings, such measures might help to avoid unwanted effects on the health of overlooked dementia patients with minimal effort.