Abstract
Background:
There are no short valid instruments to evaluate cognitive status in severe Alzheimer’s disease (AD) patients in the Spanish language.
Objective:
To validate the Spanish version of the Baylor Profound Mental Status Examination (BPMSE-Sp).
Methods:
The Baylor Profound Mental Status Examination (BPMSE) was translated to Spanish and back translated. Validation was conducted in 100 patients with severe probable AD with a Mini-Mental State Examination (MMSE) <12. We assessed internal consistency (Cronbach’s alpha), concurrent validity (Pearson’s correlations) with the MMSE, Severe Impairment Battery (SIB), Neuropsychiatric Inventory Short Form (NPI-Q) and the Functional Assessment Staging and reliability.
Results:
The mean age of patients was 84.9; 74% were female; 64% were institutionalized. The mean MMSE was 5.6; the mean BPMSE-Sp was 13.6; the mean BPMSE-Sp behavior was 1.2; the mean SIB was 42.2; and the mean NPI-Q was 4.7. BPMSE-Sp presented good internal consistency (Cronbach α= 0.84). There were significant correlations between the BPMSE-Sp and MMSE (r = 0.86, p < 0.001), and between the BPMSE-Sp and SIB (r = 0.92, p < 0.001). Inter-rater and test-retest reliability were in both cases excellent, ranging between 0.96 and 0.99 (p < 0.001). BPMSE-Sp had fewer floor and ceiling effects than the MMSE.
Conclusions:
The BPMSE-Sp is a valid tool for use in daily practice and research in the evaluation of cognitive function of patients with severe AD.
INTRODUCTION
Spain is one of the fastest aging societies in the world. It is estimated that 19% of the population will be aged 65 or older by 2020 and 32% by 2050 [1]. The prevalence of dementia is estimated at around 2.8% in 75–79 year olds and up to 19.1% in those older than 90 years [2]. Alzheimer’s disease (AD) is the most common type of dementia, and severe AD accounts for approximately 28% of that [3]. In Spain, there are approximately 400,000 AD patients [4]. The Spanish population is increasingly aging; therefore, a scale for an assessment of cognitive function in severe AD is necessary.
The majority of the scales used to measure cognitive function in AD have limited sensitivity for the assessment of patients with a severe degree of impairment. Staging scales like Clinical Dementia Rating Scale [5], the Global Deterioration Scale [6], and the Functional Assessment Staging (FAST) [7] have the disadvantage that they are not sensitive to progression variability in various cognitive and behavioral domains. The Mini-Mental Status Examination (MMSE) [8], the instrument for the assessment of cognitive function, contains significant verbal information and has floor effects, with questionable results, especially in those with major problems of sensory and language disorders [9, 10]. Others, like the Severe Cognitive Impairment Profile [11], the Severe Impairment Battery (SIB) [12], and the Test for Severe Impairment [13] have disadvantages in requirements such as time length, trained examiners, and specific material for their realization.
Doody et al., designed the Baylor Profound Mental Status Examination (BPMSE), a scale for staging and monitoring severe and profound AD patients, modeled after the MMSE [14]. The BPMSE has the advantage that it can be administered in just 5 minutes without requiring an informant and does not require extensive training of the examiner. It also has good psychometric properties in severe AD patients, presenting high correlation with MMSE (r = 0.76). It has been validated in Korean [10] and Danish [15]. The Korean team concluded that the BPMSE-Ko was a time-efficient and useful tool to measure cognitive function in patients with severe AD, and the Danish team concluded that the BPMSE was a very stable and strong scale and was recommended as a severity measurement for the cognitive performance of patients suffering from severe dementia.
However, there are a lack of reliable and valid short tools in Spanish to examine the cognitive state of a patient with severe cognitive impairment. Therefore, we conducted our study to validate and adapt the BPMSE for a Spanish population and also to examine its psychometric properties.
MATERIAL AND METHODS
Translation
The BPMSE was translated into two versions in Spanish, performed by bilingual staff, whose mother language was Spanish, from the Alzheimer’s Disease and Memory Disorders Center of the Baylor College of Medicine (Houston). These translations were revised by two physicians, using a version adapted to Castilian-Spanish, who understood the purpose of the questionnaire to ensure conceptual translation [16]. Back-translations of all these versions were performed by two graduates of English philology, different from the first bilingual translators. The research team met to resolve differences and ensure cross-cultural adaptation. The consensus on the latest version was performed with the principal author of the original version of the questionnaire, to warrant the quality control of the translation. The final version was back-translated by two other bilingual individuals in Spain, and the two versions were compared. With this methodology we ensured that the characteristics and structure of the original BPMSE were maintained in the Spanish version. The final Spanish version of the BPMSE (BPMSE–Sp) is shown in Supplementary Material 1.
Participants
The study included 100 patients with probable AD (NINCDS/ADRDA criteria). Inclusion criteria were: age 65 and older, residing in Albacete, Spanish, diagnosed with severe probable AD (MMSE <12 points), living at home or institutionalized, and with the presence of usual caregiver for more than 2 hours/day who could provide reliable information. All had brain imaging, neurological evaluation and routine laboratory tests that excluded other possible causes of dementia. Exclusion criteria were: clinically significant neurological or psychiatric conditions (e.g., global aphasia, blindness, quadriplegia, and head injury), alcohol or drug abuse. Participant’s tutors or legal representatives signed informed consent prior to their participation in the study. The protocol complied with the Helsinki Declaration. The Ethics Committee of the Health Care Area of Albacete and the Clinical Research Committee of the Complejo Hospitalario Universitario of Albacete, Albacete, Spain, approved the study.
Validation
Subjects were included between September and December 2014. Several cognitive performance assessments were conducted the same day to determine the concurrent validity: MMSE, SIB, Neuropsychiatric Inventory Short Form (NPI-Q), and FAST. All scales were used in their Spanish versions [17–20]. Finally, a subsample of 50 patients was selected for an inter-rater and test-retest reliability study. Two trained physicians consecutively administered the scales again to 50 participants to calculate inter-rater reliability, and test-retest reliability was calculated by having the same physician repeat the interview 1 or 2 days later for each of the 50 participants.
Demographic and psychological characteristics
Demographic data were collected on age, gender, and education level. The dementia severity was assessed through these following cognition scales: MMSE, SIB, and FAST. These scores were compared with scores of BPMSE-Sp cognitive subscale. Also, presence and severity of behavioral problems were measured with the NPI-Q [19], comparing the score with that of the BPMSE-Sp behavior subscale.
Baylor Profound Mental Status Examination
The BPMSE assesses four areas of cognition: language, orientation, attention, and motor skills. It also allows the evaluator to rate presence or absence of 10 problem behaviors and makes two qualitative observations of social interaction and communication. It has the advantage that it can be administered in just 5 minutes, without requiring an informant or requiring extensive training of the examiner.
The BPMSE total cognitive subscale has a score between 0 (maximum cognitive impairment) and 25 (best cognitive score for severe dementias). BPMSE behavior subscale score has a score between 0 (no behavioral disturbances) and 10 (all behavioral disturbances). In the BPMSE communication and social interaction scales, the evaluator can describe the baseline characteristics and, at later time, using a 7-point Likert scale, can rate changes in these two items compared to baseline: 1 (marked improvement) and 7 (marked worsening).
Data analysis
First, a descriptive analysis of the sample was performed, including central and dispersion parameters. The validity of the BPMSE-Sp was tested using factor analysis, using a principal component methodology based on correlation analysis. Suitability was tested using the Bartlett’s test of sphericity and the Kaise-Meyer-Olkin measure of sampling adequacy. Principal components and the explained variance proportions were determined, and finally the model was checked using Varimax rotation. Pearson correlation were used to compare the total scores and the scores for each of the four BPMSE-Sp cognitive components. Concurrent validity was tested using Pearson correlations between BPMSE-Sp-cognitive and the Spanish versions of MMSE, SIB, and FAST. We also compared the NPI-Q with the BPMSE-Sp behavior subscale.
Internal consistency was assessed by computing the coefficient α. The reliability of all scales, and BPMSE-Sp subcomponents, were examined using the intraclass correlation coefficient.
To compare the ceiling and floor effect of the BPMSE-Sp with the MMSE, the percentage of participants who obtained the highest and lowest scores on each scale was calculated. All data were stored and analyzed using SPSS 17.0 (SPSS, Inc., Chicago, IL) for Windows.
RESULTS
Table 1 shows demographic characteristics of the sample. The mean time for BPMSE-Sp administration was five minutes per patient.
Validity
Factor analysis of the function components, identified six factors that accounted for 77.2% of the variance (p < 0.001): Language (factor 1) accounted for 45.8% of the variance, motor skills (factors 2 and 4) accounted for another 14.9% , orientation (factor 3) accounted for 6.8% , and attention (factors 5 and 6) for an additional 9.7% . Suitability was demonstrated by a significant Bartlett’s test of sphericity (p < 0.001). Significant correlations were found among all the BPMSE-Sp components, as seen in Table 2 (p < 0.001).
Correlation coefficients between BPMSE-Sp and MMSE, SIB, and FAST were measured by Pearson correlation (Table 3). We also compared BPMSE-Sp behavior subscale with NPI-Q (Table 3). Figure 1 presents the linear association between BPMSE-Sp and the rest of the test employed for validation. There was a strong linear association between MMSE and BPMSE-Sp (r = 0.86) (Fig. 1A), and between SIB and BPMSE-Sp (r = 0.92) (Fig. 1B). However, the linear relationship between BPMSE-Sp behavior subscale with the NPI-Q Severity (Fig. 1C) and NPI-Q Distress (Fig. 1D) was very low (r = 0.19).
Reliability
The coefficient α, reflecting intercorrelations for items on the BPMSE-Sp, was 0.84 (p < 0.001), indicating good internal consistency. Inter-rater and test-retest reliability were excellent, as shown in Table 4, ranging between 0.96 and 0.99.
Ceiling and floor effects
The highest MMSE score in our sample was 11 points; this score was detected in 17 participants. In these patients, BPMSE-Sp scores ranged between 19 and 25, and only in two patients (11.8% ) the maximal scoring was reached (BPMSE-Sp = 25). The lowest MMSE score (0 points) was detected in 18 patients. In these participants, BPMSE-Sp scores ranged from 0 to 2, and in 12 patients (66.7% ) the lowest score was reached.
DISCUSSION
Scales previously validated in Spanish to assess dementia have many disadvantages in severe AD patients: they need a long time, trained examiners, or specific material. Furthermore, there are no short and accessible tools to assess severe AD patients. Our study confirms that the Spanish version of the BPMSE-Sp (cognitive subscale) is a valid and reliable measure of cognitive assessment in patients with this dementia stage. Our data demonstrated good construct and concurrent validity, as well as an excellent inter-rater and test-retest reliability.
Other BPMSE validation studies have only used the BPMSE cognitive subscale [10, 15], not including the BPMSE behavior subscale. The Korean study justified this issue arguing that their main objective was to compare the BPMSE-Ko with the SIB, and the SIB only measures cognitive domains. In our study, we compared the BPMSE-Sp behavior subscale with NPI-Q, with fair results. The poor relationship between BPMSE-Sp behavior subscale and NPI-Q (r = 0.19) could be justified because BPMSE-Sp behavior subscale information is obtained by the subjective impression of the examiner during the interview, while the NPI-Q is obtained by questions to the primary caregiver, without the subjective influence of the examiner. We did not test the two descriptive BPMSE questions concerning communication and social interactions, because we did not have a Spanish comparison scale for these items. Although these two questions were not validated, we translated and included them in the final adapted scale.
The inter-rated reliability of the BPMSE-Sp was similar to those presented in the Korean and Danish versions [10, 15], but we obtained better test-retest reliability. In our test-retest reliability methodology, the second assessment was conducted in the initial 48 hours after the first evaluation, while the Korean version was tested again after 5 weeks, and the Danish version after 1 week. Cognitive conditions may vary within very short time frames in severe dementia patients.
Regarding construct validity, all BPMSE-Sp components presented high correlation coefficients, except for the motor components. This fact could be explained by the high number of patients with no education included in the study (32% ), many of whom were probably illiterate. This could explain good scores on language components, but low scores on motor components like “write your name.” The Korean and Danish validations had good correlation between language-motor components, but their participants had a higher educational level than ours. Korean participants presented a mean of 5.6 years of education, and 96% of Danish participants had 7 years of education or more.
From the ceiling and floor effects results we could conclude that, probably, the BPMSE-Sp has fewer ceiling and, in a lower degree, floor effects than the MMSE. But a study with a larger sample is necessary to confirm this.
The majority of the scales used to measure cognitive function focus on the diagnosis of AD in mild degree. But we cannot forget the importance of evaluating cognitive status even in the advanced stages of dementia: this will be useful for analyzing the evolution,prognosis, and response to treatment of our patients.
In conclusion, the cognitive part of BPMSE-Sp has been shown to be useful to measure cognitive function in Spanish patients with severe AD. BPMSE-Sp is faster and easier than MMSE or SIB, making it an effective and efficient tool for everyday medical practice.
