Screening for cognitive impairment in older African-Caribbeans

Transcription

1 Psychological Medicine, 2000, 30, Cambridge University Press Printed in the United Kingdom Screening for cognitive impairment in older African-Caribbeans G. RAIT, M. MORLEY, A. BURNS, R. BALDWIN, C. CHEW-GRAHAM AND A. S. ST LEGER From the Department of Old Age Psychiatry, Withington Hospital, School of Primary Care and Department of Public Health, University of Manchester and Department of Psychiatry, Manchester Royal Infirmary, Manchester ABSTRACT Background. There are increasing numbers of older African-Caribbeans in the United Kingdom. Screening instruments are commonly used in the detection of cognitive impairment, but have not been assessed within this population. This study aimed to develop culturally modified versions of screening instruments for cognitive impairment (Mini-Mental State Examination () and Abbreviated Mental Test ()) and to determine their sensitivity and specificity in the diagnosis of dementia. Methods. The instruments were modified using a process involving a community group of African- Caribbeans and an academic group of health professionals. They were used in a two-stage study involving community resident African-Caribbeans aged 60 years or over in inner-city Manchester, comparing the screening instruments against a computerized diagnostic interview. Results. One hundred and thirty people completed the study. The results for the largest subgroup, the Jamaicans (N 96) were analysed. Effects of gender, age and education on the and scores were evaluated. The correlations between the screening instruments and diagnostic interview were highly significant (P 0 001). At appropriate cut-offs both screening instruments demonstrated high sensitivity and acceptable specificity levels. Conclusions. A defined process with lay input has assisted in producing culturally modified versions of the and that perform well compared with a diagnostic interview, if an appropriate cut-off is used. They are easy to administer and acceptable to older African-Caribbean people. The results need to be viewed within the limitations of the current study. INTRODUCTION There are increasing numbers of older African- Caribbeans in the United Kingdom (OPCS, 1993). People from ethnic minority backgrounds face barriers to mental health care, including the lack of culturally sensitive diagnoses and management in health and social services (Manthorpe & Hettiaratchy, 1993; Rait et al. 1996; Bhugra, 1997). The detection and diagnosis of dementia is complex and affected by Address for correspondence: Dr Greta Rait, Department of Primary Care and Population Sciences, Royal Free & University College Medical School, Holborn Union Building, Archway Campus, Highgate Hill, London N19 3UA. 957 people s health beliefs and access to information, knowledge and attitudes in primary care, and access to, and the approach of secondary services. Screening instruments can assist health care professionals to recognize dementia. The Mini-Mental State Examination (; Folstein et al. 1975) and Abbreviated Mental Test (; Hodkinson, 1972) are widely used questionnaires, but they have not been validated in the African-Caribbean population in the UK. This study aimed to develop culturally modified versions of the and, and determine their sensitivity and specificity by comparing them with a diagnostic computerized interview.

2 958 G. Rait and others METHOD Screening instruments The and are used in research, and clinically in primary and secondary care. They have been validated internationally with different ethnic groups (Escobar et al. 1986; Salmon et al. 1989; Rocca et al. 1992; Hooijer et al. 1992; Ganguli et al. 1995). In the UK there is limited research on their performance in immigrant groups e.g. South Asian Gujarati speakers (Lindsay et al. 1997). Diagnostic instrument The GMS-AGECAT is a computerized diagnostic interview for detecting mental health problems in older people (Copeland et al. 1986). It is designed to provide a standardized clinical decision-making process that records the presence of psychiatric symptoms at six levels (0 5). A case is a score of three or more. The rating of organic case correlates highly with dementia. GMS-AGECAT has been used with the African- Caribbean community in the United Kingdom (McCraken et al. 1997). Adaptation of screening instruments The participation of lay people from an African- Caribbean background was an integral part of the process to adapt the instruments. Content and technical equivalences (Flaherty et al. 1988) were assessed for each item on the screening instruments. Content equivalence ensured that each item (question or statement) retained its meaning in the study culture. Technical equivalence ensured that when the screening instruments were used they were applied in a culturally sensitive manner. Two separate groups were involved in this process: an academic working group and a community working group. The academic group included psychiatrists, general practitioners (GPs), a psychologist and a public health physician. The community group included social and voluntary workers of an African-Caribbean background, who worked with older African- Caribbeans and those with mental health problems. The community group reviewed each item and discussed cultural appropriateness and suggested suitable substitutions. These were discussed with the academic group to ensure the same concepts were still being tested. Content equivalence There were concerns from the community group that there may be a wide range of educational attainment among older African-Caribbeans. The main modifications involved those items where education and literacy were significant. These were writing a sentence, copying a drawing and spelling World backwards. For those who were not literate the alternatives were saying a complete sentence and saying the days of the week backwards. In view of the concerns about lower educational attainment, the attention item was simplified to counting backwards from ten to one and the construct item involved copying a less complex figure (diamond in a box). There was concern over the item relating to the date of the First World War, as this was not thought to be a familiar date. Suggestions included the date of independence of the island of origin or the date of the Second World War and these were included. Technical equivalence The community group suggested having a broader explanation of the project aims, a clearer introduction to the use of each screening instrument and an explanation of the more abstract or complicated items. The suggestions were incorporated into the interviewers training manual. Pilot study The ability to recruit subjects, likely response rate and acceptability of the modified screening instruments were assessed in a pilot study based in an inner-city practice in central Manchester. Out of 27 subjects, identified by practice staff from the general practice list, of people aged 60 years and over, 20 agreed to participate (74%). All, but one, completed the interview and found it acceptable. All subjects repeated the address as Bradford rather than Bedford. This was changed for the main study as subjects in the North of England were more familiar with a northern town. Two subjects interviewed (10%) were not African-Caribbean (White and African). Sampling General practices in areas with high proportions of African-Caribbeans were sent an introductory

3 Cognitive impairment screening in older African-Caribbeans 959 letter. Nine out of 16 practices on the Health Authority s list agreed to participate. Five declined because of workload and two gave no reason. These practices were similar in size and number of partners to the participating practices. None of the practices in the study had a register of ethnic minority patients. Subjects were identified by a computer search for patients aged 60 and over, followed by a manual search by practice staff for those thought to be of African- Caribbean origin. Contacts were established with local community leaders, and day-care organizers, to facilitate participation and identify people not on GPs lists or missed by the above method. Subjects were only excluded if known to be terminally ill, or have severe physical or mental illness that would effect their ability to be interviewed, or were not community resident. Interviewers Two interviewers were employed to administer the screening instruments in a standardized manner. Training was via videos, practical demonstrations and observation in the field. A training manual facilitated reliable rating. Interrater reliability was high (98%) and measured by interviewers assessing standard videotaped interviews. Main study The study had ethical approval and was conducted in two stages. The subjects were community resident African-Caribbeans in Manchester, aged 60 years of age or over. In stage one, eligible subjects were contacted by letter and interviewed at home. They were screened for cognitive impairment, using the and, by the interviewers. Demographic data was collected, including selfassigned ethnicity from Census categories, place of birth, education and physical illness. In stage two, subjects were interviewed at home using the GMS-AGECAT. This was administered by G. R., who was blind to the results of the screening interview. The general practice notes of those subjects who completed both stages were examined for documentation of mental health assessments and problems in the previous year. Analysis Data were analysed using SPSS 7.5 for Windows (SPSS, 1997). The mean scores were calculated for the and for different ages and educational attainment. The effects of gender, age and education were estimated using linear regression. Correlation coefficients were calculated between the scores on the (out of 30) with the score on the GMS-AGECAT (out of 5), and between the (out of 10) and the GMS-AGECAT. A receiver operated characteristic (ROC) curve was produced for each instrument. This allowed for the choice of most appropriate cut-off and estimation of sensitivity and specificity levels for each. RESULTS Response In stage one, 227 subjects were contacted and 160 subjects agreed to participate (71%). Of the 160 subjects screened by the interviewers, 12 were of inappropriate ethnicity for the study (8%) and three had moved out of the district. Of the 145 eligible subjects, 130 subjects (90%) agreed to participate in stage two. Of the 15 who declined the second interview, 10 subjects (7%) felt their mental health did not warrant further investigation and five subjects were physically too unwell (3%). All had scored within the normal range for the screening instruments. The majority of subjects completed both interview stages within 6 days (90%). Demographics A large number of subjects lived alone (40%). Most of the remaining lived with a spouse or partner (32%), or other family members (25%). All the subjects were registered with a GP and most had attended the practice in the past year (94%). About a third (31%) reported a physical disability that they felt regularly affected their ability to perform daily tasks. Over a half of subjects had diagnosed hypertension (55%) and over a third had diabetes (37%), with about a third having both (30%). Most were literate in English (96%). The mean numbers of years in formal education was 9 (range 0 17 years; S.D. 2 84). The largest subgroup, of those who described themselves as Black Caribbean, was composed

4 960 G. Rait and others Age group (years) Table 1. Mean scores by age group Subjects N All ages 26 1 (25 2 to27 0) 8 1 (7 8 to8 3) (25 3 to28 7) 8 3 (7 6 to9 0) (24 9 to27 8) 8 1 (7 5 to8 7) (24 9 to27 9) 8 4 (7 9 to8 8) (23 1 to28 2) 7 4 (6 0 to8 7) (16 3 to29 5) 7 4 (6 5 to8 4) 9 Table 2. Education (years) Mean scores by education level Subjects N (20 1 to30 2) 7 8 (6 0 to9 8) (23 3 to27 4) 8 3 (7 9 to8 7) (25 0 to27 4) 7 8 (7 2 to8 4) (27 1 to29 3) 8 5 (7 9 to9 1) 12 Gender Table 3. Mean scores by gender Subjects N Male 26 1 (24 6 to27 6) 8 4 (8 0 to8 8) 48 Female 26 1 (25 0 to27 2) 7 7 (7 7 to8 2) 48 of subjects who had been born in and migrated from Jamaica (N 96). There were equal numbers of males to females. The mean age for Jamaicans was 69 years (range years; S.D. 6 3). Instruments Results were calculated for the largest ethnic group in the study, the Jamaicans (N 96). The mean scores for the and are presented by age group (Table 1), educational level (Table 2) and gender (Table 3). The scores for the screening instruments did not follow statistically normal distributions. Log transformation was performed before the regression analysis. Age, years of education or gender did not have an effect on the score. Age did have an inverse effect on the score at a level of significance 0 05 (P 0 013). The strengths of the relationships between the GMS-AGECAT and each screening test were compared using the Spearman correlation coefficient. The correlations obtained were statistically highly significant (r 0 47, P 0 001); (r 0 59, P 0 001). Sensitivity False positive rate False positive rate FIG. 1. Receiver Operating Characteristic (ROC) curve for and (, boxed value cut-off score). Appropriate cut-off points can be identified for the and using the ROC curves (Fig. 1). Examples of appropriate cut-offs with 95% confidence intervals for the are 27 (sensitivity 100% (54 to 100); specificity 69% (60 to 78)) and 26 (sensitivity 83% (76 to 91); specificity 78% (69 to 86)). Examples for the are 8 (sensitivity 100% (54 to 100); specificity 83% (76 to 91)) and 7 (sensitivity 67% (57 to 76); specificity 94% (90 to 99)). Six cases consistent with an organic diagnosis (N 6; 6%) were detected by diagnostic interview. DISCUSSION Principal findings Cultural modification of screening instruments requires the input of lay people of the appropriate cultural background. The modified and do not appear to be affected by education or gender. They performed well compared with diagnostic interview. Subjects found the instruments acceptable and interviewers found them easy to administer. Only

5 Cognitive impairment screening in older African-Caribbeans 961 brief training was required, making them suitable for use by different groups of health care professionals. However, these results need to be viewed in the context of working cross-culturally with established instruments and factors associated with sample selection. Sample Ethnicity African-Caribbean is a heterogeneous term encompassing people from different islands with separate cultural identities (Hutchinson & McKenzie, 1995). The majority of older adults in the UK and in Manchester are from Jamaica. Our sample was representative in that it was mainly Jamaican. At interview, subjects indicated their own ethnic background, with analysis based on this. Instruments may perform differently in different cultures (e.g. cultural contrast between those from Jamaica, Trinidad or Barbados). In order to increase validity data was analysed by ethnic group. A sensitivity analysis of the data was performed and could not confidently predict whether other ethnic groups had similar results to the Jamaican group. Identification There are no easy methods to identify people from an African-Caribbean background. Studies have enumerated populations by alternate doorknocking and creating local registers (Richards et al. 1996) or using multiple methods including Family Health Services Authority lists, information from community groups and snowballing i.e. obtaining further contacts via subjects (McCraken et al. 1997). There are advantages and limitations to these approaches. Our study used GPs lists and community links. Distinguishing African-Caribbean subjects by surname is difficult. The primary method of sample identification was by general practice staff. This resulted in the inclusion of people who were not African-Caribbean (8%) and presumably excluded eligible subjects. We attempted to be over-inclusive and use community links to contact those who were missed. We could have supplemented our methods with a door-knocking exercise. This has disadvantages, including the difficulties of obtaining accurate information (Richards et al. 1996). For the purposes of assessing screening tests a strictly random sample is not required. The sample appears to have covered a sufficient range of subjects to make the results applicable to the general population. Recruitment and response Over 90% of the sample was recruited via their GP. This approach worked well in this group, although lay community members had anticipated a lower response rate. The advice from the community group of African-Caribbeans, particularly about how to introduce and explain the study, and what questions to be prepared for, may have optimized the study s acceptability. This has implications for staff training with regard to cultural awareness. The overall response rate (60%) was acceptable. It could be suggested that the non-responders may be from important groups such as the mildly cognitively impaired or depressed. After the first interview most people were keen to participate in the second interview. This may have been due to the culturally sensitive manner in which interviews were conducted and the time spent explaining the project aims directly to the subjects. Demographic characteristics Almost three-quarters of subjects lived alone or with a spouse of similar age. They may need to have access in the future to formal support services, such as social services. Policy makers will need to plan suitable services responsive to the needs of older African-Caribbeans. Many have chronic medical conditions, such as hypertension and diabetes. The health services need to be organized to provide acceptable and appropriate care. Performance of screening tests Culture-specific instruments detect particular cultural presentations in a defined population. These presentations may be missed by conventional instruments, such as the. Factors such as language, literacy and ethnicity have been shown to have an effect (Escobar et al. 1986; Tombaugh & McIntyre, 1992). Education did not have an effect in our study, although age had an effect on the. The advantages of using existing instruments are, that they are readily available and health professionals are, to varying degrees, familiar with their use. A study of older African-Caribbeans has shown that the widely used Geriatric Depression Scale

6 962 G. Rait and others performed well compared to a culture specific instrument (Abas et al. 1998) in screening for depression. However, modifying current instruments to make them culture-fair may still result in them missing culture-specific presentations. The evaluation process using academic and community groups, and analysing the test item equivalences aimed to eliminate some cultural bias. The advisory panel of community members commented on liaising with, and interviewing older African-Caribbeans. Their suggestions with regard to content were incorporated into the modified screens, advice about gaining rapport and emphasizing full explanations was used in the study. The sensitivity and specificity of the instruments will have been facilitated by this. The administration of the instrument in a culturally sensitive way may be as important a factor as the content of the scale itself. Although the modified and have been validated in this population two points need to be emphasized. First, the small number of cases (N 6) make the results less valid. Secondly, the cut-off scores may have to be altered. For the modified a score of seven or less has a high sensitivity (100%) and specificity (83%). For the modified a score of 25 would give a reasonable sensitivity and specificity, but a higher cut-off, a score of 26, would give a 100% sensitivity, but only a 69% sensitivity. The choice of cut-off would depend on available resources to deal with false positives and the problems associated with people being in the false negative group. Use of GMS-AGECAT A significant limitation is the absence of a gold standard diagnostic instrument for older African-Caribbeans. The GMS-AGECAT provides a diagnostic guide, rather than a true gold standard. However, it allows for comparative work. The use of the GMS-AGECAT, and may lead to high agreement solely because they originate from a similar cultural background, with overlapping questions. Ideally, the diagnostic stage should use a combination of modified GMS-AGECAT and a culturally sensitive clinical assessment. This is a problem that is faced whenever a study like this is conducted. We tried to overcome cultural bias by using the community group to advise on modifications to the screening instruments. Another way of assessing the validity of the GMS-AGECAT in a different cultural setting would be to compare the rate of decline of cases and controls. Detection of cognitive impairment Six cases of dementia (6%) were detected by diagnostic interview. This was not an epidemiological study. The nature of the sample means that the findings should not be extrapolated as estimates of dementia prevalence. This figure is slightly lower than in a recent study (8%) with similar numbers of older Caribbeans (McCraken et al. 1997). The GPs notes suggested that only one out of the six cases of dementia was known to their GP. The other cases had consulted four or more times in the preceding year, but there were no records of relevant symptoms or mental health status. This suggests that there remains a need for training in detecting dementia when dealing with ethnic minorities. Dementia is associated with hypertension. Four of the six case subjects and over half (55%) of all subjects interviewed had known hypertension. Future work Ageing in society The cultural context of ageing has to be considered in detecting dementia. If a culture removes responsibility from older people, the symptoms may be less obvious, or tolerated by the family or carers as a part of the ageing process. This will influence whether individuals or families present to doctors and whether they discuss the symptoms. Qualitative work is necessary to gain an understanding of the cultural presentations of dementia and attitudes to illness and management in older African- Caribbeans. From this targeted information could help to inform people about dementia and improve health professionals knowledge of dementia in this community. Screening Commander and colleagues (1997) found an under-recognition of mental health disorders in Black subjects in primary care. The use of screening instruments is not advocated on a population basis, and may be more sensitive when patients are suspected of having problems (Iliffe et al. 1994). This area needs further research, particularly within primary care.

7 Cognitive impairment screening in older African-Caribbeans 963 Strategies for the detection of dementia in older people need to be addressed at a local and national level. This will involve training practice nurses and GPs within primary care, and nursing and medical staff in secondary care. The involvement of lay people from ethnic minority backgrounds is crucial to the development of culturally appropriate and responsive mental health services. We would like to thank the interviewers (T. Osundeko and A. Webb), U. Parker who conducted the audit and B. Farragher for statistical advice. We would like to acknowledge the assistance of the GPs and staff, and all the subjects who generously gave us their time. This study was funded by the NHS Research and Development Fund (Northern and Yorkshire Region). REFERENCES Abas, M., Phillips, C., Carter, J., Walker, S., Banerjee, S. & Levy, S. (1998). Culturally sensitive validation of screening questionnaires for depression in older African-Caribbean people living in South London. British Journal of Psychiatry 173, Bhugra, D. (1997). Setting up psychiatric services: cross-cultural issues in planning and delivery. International Journal of Social Psychiatry 43, Commander, M., Sashi Dharan, S., Odell, S. & Surtees, P. (1997). Access to mental health care in a inner-city health district. II: Association with demographic factors. British Journal of Psychiatry 170, Copeland, J., Dewey, M. & Griffiths, H. (1986). A computerised psychiatric diagnostic system and case nomenclature for elderly subjects: GMS and AGECAT. Psychological Medicine 16, Escobar, J., Burnam, A., Karno, M., Forsythe, A., Landsverk, J. & Golding, J. (1986). Use of the in a community population of mixed ethnicity. Journal of Nervous and Mental Disease 174, Flaherty, J., Gaviria, F., Pathak, D., Mitchell, T., Wintrob, R., Richman, J. & Birz, S. (1988). Developing instruments for crosscultural psychiatric research. Journal of Nervous and Mental Disease 176, Folstein, M., Folstein, S. & McHuges, P. (1975). Mini-Mental State: a practical method for grading the cognitive state of patients for the clinician. Journal of Psychiatric Research 12, Ganguli, M., Ratcliff, G., Chandra, V., Sharma, S., Gilby, J., Pandev, R., Belle, S., Ryan, S., Baker, C., Seaberg, E. & Dekosky, S. (1995). A Hindi version of the : the development of a cognitive screening instrument for a largely illiterate population in India. International Journal of Geriatric Psychiatry 10, Hodkinson, H. (1972). Evaluation of a mental test score for assessment of mental impairment in the elderly. Age and Ageing 1, Hooijer, C., Dinkgreve, M., Jonker, C. & Lindebloom, J. (1992). Short screening tests for dementia in the elderly population: I. A comparison between S,, MSQ and SPMSQ. International Journal of Geriatric Psychiatry 7, Hutchinson, G. & McKenzie, K. (1995). What is Afro-Caribbean? Implications for psychiatric research. Psychiatric Bulletin 19, Iliffe, S., Mitchley, S., Gould, M. & Haines, A. (1994). Evaluation of the use of brief screening instruments for dementia, depression and problem drinking among elderly people in general practice. British Journal of General Practice 44, Lindsay, J., Jagger, C., Mlynik-Szmid, A., Sinorwala, A., Peet, S. & Moledina, F. (1997). The Mini-Mental State Examination in an elderly immigrant Gujarati population in the United Kingdom. International Journal of Geriatric Psychiatry 12, McCraken, C., Boneham, M., Copeland, J., Williams, K., Wilson, K., Scott, A., McKibben, P. & Cleave, N. (1997). Prevalence of dementia and depression among elderly people in Black and ethnic minorities. British Journal of Psychiatry 171, Manthorpe, J. & Hettiaratchy, P. (1993). Ethnic minority elders in the UK. International Review of Psychiatry 5, Office of Population, Censuses and Surveys (1993). Census Ethnic Groups and Country of Birth. Vol 1 2. HMSO: London. Rait, G., Burns, A. & Chew, C. (1996). Age, ethnicity and mental illness: a triple whammy. British Medical Journal 313, Richards, M., Brayne, C., Forde, C., Abas, M. & Levy, R. (1996). Surveying African-Caribbeans elders in the community: implications for research on health and health service use. International Journal of Geriatric Psychiatry 11, Rocca, W., Bonaiuto, S., Lippi, A., Luciana, P., Pistarelli, T. & Grandinetti, A. (1992). Validation of the Hodkinson Abbreviated Mental Test as a screening instrument for dementia in an Italian population. Neuroepidemiology 11, Salmon, D., Reikkinen, P., Katzman, R., Zhang, M., Jin, H. & Yu, E. (1989). Cross-cultural studies of dementia: a comparison of performance in Finland and China. Archives of Neurology 46, SPSS (1997). SPSS for Windows Release 7.5. SPSS Inc.: Chicago, IL. Tombaugh, T. & McIntyre, N. (1992). The Mini-Mental State Examination: a comprehensive review. Journal of the American Geriatric Society 40,