Addressing Ethnic Diversity in Health Outcome Measurement
'Addressing Ethnic Diversity in Health Outcome Measurement: a systematic and critical review of the literature'
By Dr. Gary S. Collins and Professor Mark R D Johnson
The majority of the commonly used health outcome measures were developed in English speaking countries (such as the UK or US) and more often than not intended for use among what were seen as ethnically homogenous (generally ‘White’) populations. Concern has arisen about the suitability of such measures in populations of different ethnic or cultural background (including UK-resident Black and minority ethnic populations).
The objective of this review is to explore whether any of the more routinely used health outcome measures, which tend to have development roots either in the US or the UK can be routinely used and have comparable meaning or suitability among UK Black and minority ethnic groups (following translation or cultural adaptation). The review will explore whether factors such as illiteracy and educational attainment are barriers to acceptability or effectiveness of such measures. The review will also highlight the extent to which these measures have been accepted overseas in the country of origin of UK migrant populations (largely East and South Asia) and whether they can be used in their present state (culturally adapted) or used as building blocks to develop UK versions more suitable for UK-resident minority ethnic groups.
A systematic search for published papers between 1995 and 2003 (inclusive) was carried out using Applied Social Sciences Index and Abstract (ASSIA), British Nursing Index (BNI), Cochrane Library, Cumulative Index to Nursing and Allied Health Library (CINAHL), EMBASE, Health Management Information Consortium (HMIC), Medline (PubMed), PsycINFO, Science Citation Index (SCI), Social Science Citation Index (SSCI). Concerted efforts were made to reduce any publication bias by trawling the lesser-established and local databases in India and Pakistan to retrieve articles published in journals not routinely indexed in the more established databases. Thus, in addition to the above databases, IndMed (http://indmed.nic.in) PakMediNet (http://www.pakmedinet.com) and MEDLIP were also searched within the specified time frame. Due to language difficulties, we found no easily accessible Chinese medical literature index to search in English. To overcome this, we relied on the previous mentioned databases, and reference lists (and following references through the ISI Cited Reference Lookup service). Where possible we contacted authors and carried out Internet searches (through Google and Scirus) to uncover any other articles relating to translations and administration of the instruments under review, which were not identified in any of the databases listed above.
Studies that either translated, culturally adapted or administered the Edinburgh Postnatal and Depression Scale (EPDS), EuroQoL EQ-5D, General Health Questionnaire (GHQ), Hospital Anxiety and Depression Scale (HADS), London Handicap Scale (LHS), Mini-Mental State Examination (MMSE), Nottingham Health Profile (NHP), Medical Outcomes Short Form-36 (SF-36), Sickness Impact Profile (SIP) and the World Health Organisation Quality of Life Questionnaire (WHOQOL) in subjects with reference to UK-resident Black and minority ethnic groups (Indian, Pakistani Bangladeshi, Chinese, African-Caribbean). Focus was directed towards studies that translated or adapted the measures, either for use in the UK or outside the UK, but with reference to UK BME groups. Studies administering any of the instruments were highlighted in order to identify the extent to which such measures have been accepted.
Data collection and analysis
The paucity of any sizeable datasets (at individual-patient level) and heterogeneity of the smaller data sets prohibited any meaningful secondary analysis of data (to establish if responses differ among cultural groups compared to the White population) and thus this segment of the review was excluded after consultation and agreement with the expert panel.
A total of 402 articles met criteria for review, of which the bulk concerned translation into Chinese (Cantonese or Mandarin). The majority of the studies involved administration, translation / cultural adaptation or validation of the GHQ (n=134; 33.3%), MMSE (n=88; 21.9%) and the SF-36 (n=71; 17.7%). Fifty-six studies (13.9%) were based within the UK either partially or fully – fifteen of these (26.8%) concerned translation or cultural adaptation of the instruments. The overwhelming majority of papers retrieved were developed, validated and administered outside the UK. Little attention has been addressed towards the needs of Black African-Caribbean groups; only the MMSE received attention in terms of the needs of Black African-Caribbean people. A number of studies highlighted increased levels of illiteracy and low educational attainment and developed instruments to cater for those who would normally be excluded due to inability to read and write a language. Again, a number of papers culturally adapting the MMSE addressed such concerns, given the literacy problems in elderly populations, both within the UK and overseas, which have a direct bearing on the ability to complete the MMSE. Very few authors queried whether overall concepts of health status measurement or quality of life were an appropriate concept to measure in any of the target populations.
Very few of the measures within the scope of this review have undergone detailed and thorough linguistic and conceptual validation so that if administered in a study alongside the original versions (English), confidence in the ability to compare results would be questionable. Instruments such as the LHS, NHP and the SIP have largely been ignored in preference for other measures. This may be due to length of the measures, in that the SIP for example comprises 136 items and is generally more favoured in the US and has been used less extensively in UK-based studies. The LHS is relatively new instrument and uptake popularity has yet to be realised, and thus only one study was identified which translated this instrument for use in one of the BME groups (Chinese). Translations and cultural adaptations involving the more established instruments, such as the GHQ, HADS, MMSE and SF-36 were more common.
Implications for practice
1. There is still unconvincing evidence either way to support or discourage the use of health outcome measures in Britain’s BME population.
2. It is likely that the needs of Britain’s BME population will constantly evolve, with younger members following educational trajectories different from those of their parents. New minority groups will however continue to develop through migration.
3. Research outside the UK indicates there is some evidence to support the use of health outcome measures, most notably among Chinese speaking subjects, yet there is little evidence to support the acceptability of many of the instruments this review focussed on for use among the Britain’s South Asian population.
4. Threshold or cut-off values taken as indicators of psychiatric caseness or cognitive impairment in screening instruments (GHQ, HADS, MMSE) should be ascertained prior to a study in comparable populations (in terms of language and culture) instead of adopting recommended instrument scores, generally assumed from the white indigenous population.
Recommendations for further research
1. There is a need for a transparent and detailed documentation of the development, translation and cultural adaptation of instruments, alongside psychometric testing, for languages spoken among Britain’s BME population.
2. Research is needed into the effect of acculturation and its influence on health status, including how it affects the way in which a subject responds to aspects of psychological well-being.
'Addressing Ethnic Diversity in Health Outcome Measurement' is a draft for review. You can view the full draft paper pdf - (915 KB)