Detecting MCI and dementia in primary care: efficiency of the MMS, the FAQ and the IQCODE

Transcription

1 Family Practice 2012; 29: doi: /fampra/cmr114 Advance Access publication on 25 November 2011 The Author Published by Oxford University Press. All rights reserved. For permissions, please Detecting MCI and dementia in primary care: efficiency of the MMS, the FAQ and the IQCODE Isabel Cruz-Orduña a,b, *, José M Bellón c, Pedro Torrero d, Esperanza Aparicio d, Ana Sanz d, Nieves Mula d, Garbiñe Marzana d, Concepción Begué d, Dionisio Cabezón d and Javier Olazarán a,e a Alzheimer Disease Research Unit, CIEN Foundation, Carlos III Institute of Health, Alzheimer Center Reina Sofia Foundation, Madrid, b Department of Neurology, Infanta Leonor Hospital, Madrid, c Department of Statistics, Gregorio Marañón General University Hospital, Madrid, d Peña Prieta Primay Care Center, Madrid and e Hermanos Sangro Specialties Clinic, Department of Neurology, Gregorio Marañón General University Hospital, Madrid, Spain. *Correspondence to Isabel Cruz-Orduña, Centro Alzheimer Fundación Reina Sofía Fundación CIEN, C/Valderrebollo, 5-Complejo Alzheimer PAU de Vallecas, Madrid, Spain; icruz@fundacioncien.es Received 1 May 2011; Revised 19 October 2011; Accepted 31 October Objectives. To study the yield of three instruments for detection of patients with cognitive impairment in primary care. To investigate whether combining tests is better for detecting impairment than applying them separately. Methods. Seven primary care physicians (PCP) systematically recruited individuals aged over 49 years with a complaint or suspicion of cognitive impairment. The tests administered were the Mini-Mental State Test (MMS), the Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE) and the Pfeffer Functional Activities Questionnaire (FAQ). We calculated sensitivity, specificity and the area under the curve (AUC) and applied logistic regression analysis to determine the yield of the tests in combination. The gold standard was the clinical judgement of a neurologist based on a comprehensive assessment, which included a formal neuropsychological workup. Results. Of the 160 study patients, 90 (56%) had cognitive impairment (15 of these had dementia). The MMS had a sensitivity of 77% and a specificity of 70% in screening for cognitive impairment, with an AUC of Incorporation of the IQCODE increased the AUC to 0.86 (P = 0.01). As for dementia, the FAQ reached a sensitivity of 87% and a specificity of 82%, with an AUC of Incorporation of the MMS increased the AUC to 0.95 (P = 0.03). Conclusions. Cognitive impairment is probably underdiagnosed in primary care. The combination of the FAQ and the MMS had excellent performance for dementia detection; however, no satisfactory instrument or instrument combination could be found for cognitive impairment. Keywords. Dementia, Functional Activities Questionnaire, Informant Questionnaire on Cognitive Decline in the Elderly, mild cognitive impairment, Mini-Mental State Test, primary care. Introduction Early diagnosis of cognitive impairment should provide greater knowledge of relevant pathophysiology and makes it possible to prevent or anticipate progression and social and personal consequences. Hence, early detection of cognitive impairment in primary care could produce benefits at different levels, including early diagnosis, access to treatments, promotion of psychological strategies to deal with the disease and implementation of social services. 1 Mild cognitive impairment (MCI) is defined as a transitional zone between normal cognition and dementia. 2 The relevance of this heterogeneous syndrome derives from the fact that a substantial proportion of these patients develop dementia, particularly Alzheimer s disease. 3 Although the prevalence of MCI is high (3 27% in individuals aged over 60 years), 2,4 most patients with this condition and even those with dementia who attend primary care centres are not detected. 5 7 Consensus guidelines emphasized the need for early detection of cognitive impairment, with a role for primary care. 8 However, a major challenge is the observation that only a minority of primary care physicians (PCP) is convinced of the need of, or feels prepared for, early detection of cognitive impairment. 9 Cognitive screening, in those patients with suspicion or complaints, is a first step in the diagnosis of MCI and dementia, prior to further evaluation or referral. 10 Short tests of mental status and informant questionnaires have traditionally been used for this purpose in primary care. Their predictive capacity 401

2 402 Family Practice The International Journal for Research in Primary Care increases when they are combined However, whether applied separately or in combination, they have received little attention in the field of MCI. 14 The objective of the present work was to study the diagnostic yield of three instruments a brief test of mental status and two informant questionnaires applied separately and in combination to detect MCI and dementia in a primary care setting. Given that informant interviews provide additional data for tests of mental status (functional aspects, longitudinal perspective), we parted from the premise that combining a test of mental status with an informant interview would have a better diagnostic yield than any of the tests administered separately. Methods Study sample We systematically included all individuals who attended the seven medical clinics of the Peña Prieta Primary Care Center (Health District 1, Autonomous Community of Madrid) between 1 April 2000 and 31 October The inclusion criteria were as follows: - age >49 years, - any complaint or suspicion, raised by the patient, an informant or the PCP, related to cognition, cognition-related functions [i.e. performance of activities of daily living (ADLs)] or behaviour, of unknown aetiology, - a reliable informant was available, and - verbal consent to participate in the study from both the patient and the informant. Detection instruments The Mini-Mental State Test (MMS) 15 is the most widely used test for dementia detection and includes items associated with recent memory, calculation, language and visuospatial area. Scoring ranges from 0 (worst) to 30 (best). We used a validated Spanish-language version of the test. 16 The Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE) is a structured interview composed of 26 questions on the changes in cognitive function, instrumental and advanced ADL and patient personality and behaviour during the last 5 years. 17 It has proven useful for detecting early-stage dementia and is scored from 26 (best) to 130 (worst). We used a validated Spanish-language version of this instrument, 18 which may be self-administered by an informant after minimal instructions. It takes 15 minutes. The Functional Activities Questionnaire (FAQ) is a questionnaire on functional ADL that is administered to an informant. 19 It comprises 11 items and is scored from 0 (total independence) to 33 (total dependence). The Spanish-language version used in the present study has proven useful in clinical practice for the detection of cognitive impairment (i.e. MCI or dementia). 20 This instrument may be self-administered by the informant caregiver, after minimal instructions. It takes 5 minutes. Procedure PCP received two training sessions on the inclusion criteria and the rules for administering the detection instruments. The sessions included a discussion of fictitious cases and administration of the detection instruments to real patients. During recruitment, PCP systematically identified all patients who fulfilled the inclusion criteria and completed a short protocol including demographic variables, type of complaint (cognitive, cognitive functional, affective or behavioural), source of the complaint (patient, informant, both or physician s suspicion) and verbal consent. If the patient and the informant gave their consent, the detection instruments were applied and an appointment was made to carry out a formal neuropsychological workup some days later (details of the workup can be found elsewhere). 21 Cases who presented without an informant were also systematically registered, informed about the study and invited to participate with an informant. Patients who did not present with an informant and wished to undergo the formal neuropsychological evaluation and assessment by a neurologist were permitted to do so, although they were not included in the present study. Diagnosis of cognitive impairment Based on the clinical history, clinical examination and a formal neuropsychological examination, the diagnosis of cognitive impairment was made by a senior neurologist (JO) with no knowledge of the results of the detection instruments. The diagnoses and criteria applied were as follows: - No cognitive impairment (NCI): All neuropsychological tests were above the 10 percentile of published norms for individuals of a similar age and educational level. - Non-relevant cognitive dysfunction: Some tests were under the 10 percentile, although this was not deemed to be due to genuine cognitive deterioration, but rather to low premorbid cognitive level, low educational level, lack of motivation or sensory deficit. - MCI: Performance fell below the 10 percentile in at least one neuropsychological test. This was deemed of clinical relevance, although the criteria for dementia were not fulfilled Dementia: Performance fell below the 10 percentile in memory, at least one other cognitive domain displayed performance below the 10 percentile, cognitive deterioration was considered the cause of impairment in previous functional level and

3 MMS, FAQ and IQCODE in primary care 403 a confusion state was not present. 22 The Clinical Dementia Rating Scale was used to record the severity of dementia. 23 Statistical analysis We compared the clinical and demographic characteristics of the study patients with those who did not participate and of the participants according to their cognitive diagnosis. Non-parametric tests were used to compare quantitative variables (Mann Whitney and Kruskal Wallis) when the group was small or the variables were non-normally distributed. Qualitative variables were compared using the Pearson chi-square test and the Fisher exact test. For all comparisons and analyses, patients within the nonrelevant cognitive dysfunction group were included in the NCI group. All tests were two tailed and statistical significance was set at P < The detection yield of the three instruments was analysed using receiver operating characteristic (ROC) curves. The ROC curve was created using a nonparametric approach involving the representation of all the data points. This approach is the most suitable for continuous clinical data. 24 We determined the sensitivity, specificity, positive predictive value and negative predictive value, as well as the positive probability quotient to determine the optimum cut point, i.e. the point offering the highest degree of diagnostic accuracy. Differences between the areas under the curve (AUCs) were analysed using the Hanley and McNeil correction to compare diagnostic tests in the same sample. 25 Finally, a multivariate logistic regression analysis was performed to predict which test or combination of tests best revealed the presence or absence of cognitive impairment (i.e. MCI or dementia) or dementia. We used a stepwise approach, in which a variable was incorporated if P < 0.05 and extracted if P > 0.1. The independent variables were age, gender and level of education, as well as the score of the three detection instruments. All analyses of detection yield were carried out twice. First, according to the main diagnosis of interest, namely cognitive impairment (i.e. MCI or dementia) compared with no impairment, and second, according to the diagnosis of dementia compared with no dementia. Data were analysed using Statistical Package for Social Sciences version 15.0 (SPSS Inc, Chicago, IL). Results Demographic and clinical characteristics The PCP recruited 182 community-dwelling patients with a mean (SD) age of 72.2 years (8.9; range, 50 88) of whom most were women (70.9%). The study sample comprised 160 patients (87.9%): 13 patients were not included because they did not have an informant and nine refused to participate in some areas of the research. The only significant difference between the groups was that those patients who did not participate consumed a lower number of drugs (P = 0.020; data from the patients who did not participate are not presented). All the patients included underwent the formal neuropsychological exam. The average age of the informants was 54.3 (SD 16.7, range 18 91). A majority of these informants (69.4%) were women and the link was as follows: daughter or son (50.9%), spouse (31.5%) or other family link (17.6%). In half of the cases (49.5%), patient and informant lived together. The level of education of the informants was higher than that of patients: illiterate (2.6%), none/incomplete (20.0%), primary education (40.0%) or superior (37.4%). In 94.7% of the patients, the origin of the complaint was the patient him/herself, the informant or both. Only in the remaining 5.3% did the suspicion come from the primary care physician. The most common type of complaint or suspicion, both separately and in combination, was cognitive (95.0%), followed by behavioural (16.3%) and affective (10.6%). The content of cognitive complaints was distributed as follows: no cognitive complaints (5%), recent memory (46.9%), attention (18.7%), both (17.5%) and other content (11.9%). As for cognitive diagnosis, 62 patients (38.8%) had no cognitive impairment, 8 (5%) had a non-relevant cognitive abnormality and 90 (56.3%) had cognitive impairment. Of these, 75 patients (46.9%) had MCI and 15 (9.4%) had reached dementia (mild in 12 patients, moderate in 3). Table 1 shows the demographic and clinical characteristics of the patients included according to the degree of cognitive impairment. Older age was associated with a greater degree of impairment. The opposite pattern was observed in level of education and disease course. No differences were observed for the remaining variables. ROC curves for sensitivity and specificity of tests for diagnosis Figure 1 shows the AUCs of the three detection instruments for diagnosis of cognitive impairment (i.e. MCI or dementia; Fig. 1A) and for dementia (Fig. 1B). The best AUC for diagnosis of cognitive impairment was that of the MMS (0.82), whereas the best AUC for the diagnosis of dementia was that of FAQ (0.91). A wide overlap was observed between the confidence intervals (CIs) of the three instruments within each of the diagnoses of interest; the Hanley and McNeil correction confirmed that the differences between the areas were random. However, the three instruments were clearly superior in the diagnosis of dementia (Fig. 1A and B).

4 404 Family Practice The International Journal for Research in Primary Care TABLE 1 Demographic and clinical characteristics of the three study groups NCI (n = 70) MCI (n = 75) Dementia (n = 15) Age 69.2 (8.4) 73.7 (8.3) a 80.4 (4.8) ab <0.001 Gender (% women) Education (%) Illiterate None/incomplete Primary education Superior Duration of 4.4 (9.6) 2.4 (5.4) a 2.0 (1.3) symptoms (years) Medication (n) 2.6 (2.0) 3.0 (2.1) 3.7 (2.6) Chronic medical 2.1 (1.5) 2.2 (1.6) 2.0 (1.3) conditions (n) Psychiatric disorder (%) MCI, mild cognitive impairment; NCI, no cognitive impairment. Data are expressed as mean (SD) unless otherwise indicated. Age was compared using an analysis of variance. Duration of symptoms, medication and chronic medical conditions were compared using the Kruskal Wallis and Mann Whitney tests. Gender, education and psychiatric disorder were compared using the chi-square test. a Statistically significant differences in NCI group (P < 0.05). b Statistically significant differences in MCI group (P < 0.05). P As for the cut points for the best equilibrium between sensitivity and specificity and the different diagnostic indices, we observed a ceiling effect in the FAQ for detection of cognitive impairment: 17% of patients with cognitive impairment scored maximal functional independence 26 (Fig. 1). Regression analysis The results of the multivariate logistic regression models showed that the main predictive variable for the diagnosis of cognitive impairment was the MMS, although the IQCODE and gender also added predictive power to the model. Male gender was associated with a higher probability of impairment. As for the diagnosis of dementia, the FAQ was the main predictive variable, although the MMS and age also lent predictive value to the model (Table 2). Using the regression models obtained, we calculated the AUCs of the best combinations of the tests for the two diagnoses of interest. The AUC of the MMS + IQCODE + gender for the diagnosis of cognitive impairment (including dementia) was (95% CI: ). In the case of dementia, the combination of the FAQ + MMS + age enabled us to obtain an AUC of (95% CI: ). FIGURE 1 ROC curve for the detection of MCI or dementia (A) and dementia (B). For each test, the cutoff presenting the best trade-off between sensitivity and specificity is highlighted in bold. S, sensitivity (%); Sp, specificity (%); PPV, positive predictive value; NPV, negative predictive value; LR, likelihood ratio.

5 MMS, FAQ and IQCODE in primary care 405 TABLE 2 Logistic regression model for the detection of MCI or dementia and dementia Diagnosis Test b OR (95% CI) P Change in 2 log likelihood Significance of the change MCI or dementia MMS ( ) < <0.001 IQCODE ( ) Gender ( ) Constant Dementia FAQ ( ) <0.001 MMS ( ) Age ( ) Constant OR, odds ratio. Age, education and FAQ were excluded from the model of MCI or dementia. IQCODE: gender and educational level were excluded from the model of dementia. Discussion With the aim of providing PCP with useful tools for the detection of MCI and early dementia, we performed a study in which we systematically administered three detection instruments to patients who attended a primary care center with complaints or suspicion of cognitive impairment. All the PCP participated in the study and almost all the patients and their caregivers agreed to participate. This systematic search yielded a prevalence of cognitive impairment of 56% (47% MCI, 9% dementia). Since the vast majority of patients were recruited through their own complaints or through the complaint of an informant, these data show that cognitive complaints in primary care are relevant and lead us to suspect that cognitive impairment, especially MCI, is underdiagnosed. Several studies showing a close relationship between severity of dementia and its detection in primary care supports this impression. 5 7 In the univariate comparisons, age and low educational level were associated with a greater frequency of cognitive impairment (Table 1); this finding is consistent with the results of other studies. 14,27 However, in the regression analyses, a protective effect of (female) gender was observed, with no education effect (Table 2). A possible explanation for these results is that women could have attended doctor s office due to stressful situations or mild affective disorders more frequently than men. 28 Interestingly, an inverse association was observed between symptom duration and presence of cognitive impairment (Table 1). The experience gained throughout the study leads us to believe that this association arises because complaints from individuals with no impairment originate in slowly evolving stressful situations. Another possibility, though, is that people with progressive impairment tend to place the onset of their symptoms in a more recent time frame. The yield of the detection instruments was generally good for dementia but only acceptable for cognitive impairment (Fig. 1). This should be expected: the tools chosen were originally designed for the study of patients with confusional syndrome or established dementia. 15,17,19 The MMS was the best instrument for detection of cognitive impairment, as, in most cases, dementia is due to neurodegenerative processes in which cognitive functions are affected before ADL. 2 Incorporation of the IQCODE not the FA- Q added predictive value to the MMS. This was probably because the IQCODE, as well as functional ADL (e.g. using household appliances), includes advanced ADL (e.g. using a new appliance), in which the cognitive load is greater. 29 However, the clinical relevance of this finding is doubtful, given that the AUC only increased by 3.2%. In a study conducted in specialized setting, the IQCODE was more useful than the MMS for detecting patients with MCI who progressed to dementia (AUC 0.86). 13 The follow-up of our cohort will permit us to ascertain a potential role of the IQCODE for dementia prediction in MCI patients. The FAQ, on the other hand, proved extremely useful in the detection of dementia, with the caveat that this result was obtained with a small number of dementia patients. Larger studies, carried out in a clinical setting 20 and in the population, 30 provided similar results. Moreover, the inclusion of age, the MMS and the FAQ in the regression model yielded an AUC of 0.95, which seems difficult to surpass. For similar sensitivity values, the IQCODE had lower specificity than the FAQ in the diagnosis of dementia, probably due to the high degree of complexity of some of the items (Fig. 1). Therefore, the combination of a cognitive test with a scale of functional activities is confirmed as the best approach to screening for dementia. 11,12 Some cognitive tests, shorter than the MMS, have demonstrated superiority in the diagnosis of dementia in primary care Given the high performance of the FAQ and its feasibility, the combination of a very short cognitive test and the FAQ emerges as a promising way for the detection of dementia in primary care. In the present study, a satisfactory combination of instruments could not be found for the detection of MCI. The detection of MCI in primary care continues to be problematic and is probably affected by the variety of causes underlying this syndrome. 4 One possible

6 406 Family Practice The International Journal for Research in Primary Care solution would be to extend the range of cognitive tests. Another would be to manage them according to diagnostic suspicion. However, in both cases, the concepts of simplicity and low cost would be jeopardized. For the moment, the application of clinical judgement, based on medical history and supported by a short cognitive test or combination of cognitive tasks, preferably with psychometric criteria, continues to be the best method for selecting these types of patients and for their initial management. Declaration Ethical approval: none. Conflicts of interest: none. References 1 Woods RT, Moniz-Cook E, Iliffe S et al. Dementia: issues in early recognition and intervention in primary care. J R Soc Med 2003; 96: Petersen RC. Mild cognitive impairment: transition between aging and Alzheimer s disease. Neurologia 2000; 15: Mitchell AJ, Shiri-Feshki M. Rate of progression of mild cognitive impairment to dementia: meta-analysis of 41 robust inception cohort studies. Acta Psychiatr Scand 2009; 119: Busse A, Bischkopf J, Riedel-Heller SG, Angermeyer MC. Mild cognitive impairment: prevalence and incidence according to different diagnostic criteria. Results of the Leipzig Longitudinal Study of the Aged (LEILA75+). Br J Psychiatry 2003; 182: Valcour VG, Masaki KH, Curb JD, Blanchette PL. The detection of dementia in the primary care setting. Arch Intern Med 2000; 160: Löppönen M, Räihä I, Isoaho R et al. Diagnosing cognitive impairment and dementia in primary health care a more active approach is needed. Age Ageing 2003; 32: Zunzunegui Pastor MV, del Ser T, Rodríguez Laso A et al. Demencia no detectada y utilización de los servicios sanitarios: implicaciones para la atención primaria. Atención Primaria 2003; 31: Winblad B, Palmer K, Kivipelto M et al. Mild cognitive impairment beyond controversies, towards a consensus: report of the International Working Group on Mild Cognitive Impairment. J Intern Med 2004; 256: Bradford A, Kunik ME, Schulz P et al. Missed and delayed diagnosis of dementia in primary care: prevalence and contributing factors. Alzheimer Dis Assoc Disord 2009; 23: Fisher CA, Larner AJ. Frequency and diagnostic utility of cognitive test instrument use by GPs prior to memory clinic referral. Fam Pract 2007; 24: Mackinnon A, Khalilian A, Jorm AF et al. Improving screening accuracy for dementia in a community sample by augmenting cognitive testing with informant report. J Clin Epidemiol 2003; 56: Tierney MC, Nathan H, Geslani DM, Szalai JP. Contribution of informant and patient ratings to the accuracy of the mini-mental state examination in predicting probable Alzheimer s disease. J Am Geriatr Soc 2003; 51: Isella V, Villa L, Russo A et al. Discriminative and predictive power of an informant report in mild cognitive impairment. J Neurol Neurosurg Psychiatry 2006; 77: Petersen RC, Stevens JC, Ganguli M et al. Practice parameter: early detection of dementia: mild cognitive impairment (an evidence-based review). Report of the Quality Standards Subcommittee of the American Academy of Neurology. Neurology 2001; 56: Folstein MF, Folstein SE, McHugh PR. Mini-Mental State: a practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975; 12: Blesa R, Pujol M, Aguilar M et al. Clinical validity mini-mental state for Spanish speaking communities. Neuropsychologia 2001; 39: Jorm A, Korten A. Assessment of cognitive decline in the elderly by informant interview. Br J Psychiatry 1988; 152: Morales JM, González-Montalvo JI, Bermejo J, Del Ser T. The screening of mild dementia with a shortened Spanish version of the Informant Questionnaire on Cognitive Decline in the Elderly. Alzheimer Dis Assoc Disord 1995; 9: Pfeffer RI, Kurosaki TT, Harrah CH et al. Measurement of functional activities in older adults in the community. J Gerontol 1982; 37: Olazarán J, Mouronte P, Bermejo F. Validez clínica de dos escalas de actividades instrumentales en la enfermedad de Alzheimer. Neurología 2005; 20: Olazarán J, Torrero P, Cruz I et al. Mild cognitive impairment and dementia in primary care: the value of medical history. Fam Pract 2011; 28: American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision: DSM-IV-TR. Washington, DC: American Psychiatric Association, Hughes C, Berg L, Danziger W et al. A new clinical scale for the staging of dementia. Br J Psychiatry 1982; 140: Burgueño MJ, García-Bastos JL, Gónzalez-Buitrago JM. [ROC curves in the evaluation of diagnostic tests]. [Article in Spanish]. Med Clin (Barc) 1995; 104: Hanley J, McNeil B. A method of comparing the areas under receiver operating characteristic curves derived from the same cases. Radiology 1983; 148: McHorney CA, Tarlov AR. Individual-patient monitoring in clinical practice: are available health status surveys adequate? Qual Life Res 1995; 4: Letteneuer L, Launer LJ, Anderson K et al. Education and the risk for Alzheimer s disease: sex, makes a difference. EURODEM pooled analyses. Am J Epidemiol 2000; 116: Mant A, Broom DH, Duncan-Jones P. The path to prescription: sex differences in psychotropic drug prescribing for general practice patients. Soc Psychiatry 1983; 18: Reisberg B, Finkel S, Overall J et al. The Alzheimer s disease activities of daily living scale (ADL-IS). Int Psychogeriatr 2001; 13: Villanueva C, Bermejo F, Berbel A et al. Validación de un protocolo clínico para la detección de demencia en ámbito poblacional. Rev Neurol 2003; 36: Buschke H, Kuslansky G, Katz M et al. Screening for dementia with the memory impairment screen. Neurology 1999; 52: Böhm P, Peña-Casanova J, Gramunt N et al. Spanish version of the Memory Impairment Screen (MIS): normative data and discriminant validity. [Article in Spanish]. Neurologia 2005; 20: Carnero Pardo C, Sáez-Zea C, Montiel Navarro L et al. [Diagnostic accuracy of the Phototest for cognitive impairment, dementia]. [Article in Spanish]. Neurologia 2007; 22: Upadhyaya AK, Rajagopal M, Gale TM. The Six Item Cognitive Impairment Test (6-CIT) as a screening test for dementia: comparison with Mini-Mental State Examination (MMSE). Curr Aging Sci 2010; 3 (2):