Website of the Department of Health

Background

1. The following is a discussion document which has been circulated by DH Statistics Division to Government departments and others with an interest in ethnic issues. Comments on content and the approach being adopted are very welcome.

DH position

2. There is very strong support within the department, from Ministers and officials and also from data providers, for collection of ethnic data on workforce and clients. This is a top priority, across the range of activity for which DH is responsible.

3. It has been agreed within DH's Statistics Division that, in principle, such data should be collected on a basis consistent with the 2001 census categories; with a transition period for datasets where ethnic data are already being collected in whatever form (generally based on 1991 census categories, or variants of them).

4. Office for National Statistics (ONS) Census Division have tested a number of definitions of categories and their ordering on the census form; it has been clear for some time that, whatever the ethnic categories finally chosen, they were likely to be significantly different from those used in 1991, not least because of the increasing importance of the "mixed parentage" category, which has no counterpart in the 1991 census.

5. Earlier this year the census White Paper confirmed what the ethnic categories will be, in the 2001 census. These are assumed to be more or less final, though changes cannot wholly be ruled out between now and November when the formal Census legislative Order is made. In particular the White Paper clarifies the position on the mixed category where there was some confusion as to what the "other" group comprised.

Implications for DH central collections

6. In the light of the Census White paper, DH needs to decide what the ethnic categories should be collected. Ideally it should apply to ALL statistical collections in DH. This is to avoid having ethnic categories varying according to the type of provision being measured - different categories as between Social Services and education/housing, for example.

7. It is proposed that the framework "Ethnic Classifications" should be used for this purpose. It is consistent with the draft table summaries currently being proposed by ONS.

8. Since many authorities who supply data to DH have asked for clarification of exactly which ethnic groups the categories contain, we have given the more detailed DRAFT listing which ONS Census staff will be using to classify individuals in the census. We understand that, before finalizing this draft, ONS has asked departments to specify any particular categories for which they may wish to have data from the Census. [There would be considerable interest in DH in having certain European categories for which there is a "health dimension", for example].

National versus local data requirement

9. The national ethnic framework is much less detailed than is probably appropriate at the local level for local management and monitoring purposes. But it is flexible enough to enable authorities to use more detailed categories if they so wish, by "pulling out" - that is subdividing - the national ethnic categories into the more detailed ones required locally. This should enable authorities to collate their ethnic data to any level of detail they require, whilst at the same time enabling them to reaggregate their data on a consistent basis, to national totals.

Data handling implications

10. There are some categories where there is a simple read across from the 1991 Census ethnic categories to the 2001 ones, for example, Indian, Pakistani, Bangladeshi and Chinese. For the remainder, and this includes the category of "white", there is no such simple read across. In these circumstances, data providers will have a major task in checking and possibly updating existing records of staff and clients.

11. DH will have to decide for each central collection the timetable for changing over to the new ethnic categories. In practice, a separate decision will probably be needed for each dataset, dependent upon the amount of consultations needed with policy branches, other data users and data suppliers. An overall implementation plan, to include all current and anticipated new datasets, will be needed.

12. For each central data collection we will need to decide whether

13. Case (a) seems most appropriate for "stock type" data and (b) for "flow type" data. For example, we might decide that all NHS/Social Services workforce records need to be reclassified to the new ethnic categories, but that ethnic data on hospital episodes (HES) data should simply switch over to the new ethnic categories from some (preannounced) date.

Task in prospect

14. The scale of the task in prospect can hardly be underestimated. Thus there are over 1 million staff in the health service, and a further 1 million in social services, of which perhaps 30% are employed by Local Authorities. There are about 11 million HES records each year, for inpatients alone (outpatients would at least double this). Getting ethnic data for all these groups is a major undertaking, on any examination of the facts.

Missing data

15. It should also be noted that, where ethnic data are supposedly collected already, in many cases such records are frequently wrong or incomplete because of non-response (para 17) or some other reason. This problem will add to the task of updating records. For example, Local Authorities Social services departments - who are very strong supporters of the principle of ethnic data collection - have asked, on practical grounds, for more than the usual six months notice, in respect of the change to the new ethnic categories because of the gaps in existing records.

Notice period

16. Given the scale of most of the datasets in question, and the incompleteness of many of the ethnic records within them, a notice period of at least a year is appropriate for any change in ethnic categories (or indeed anything else). This is true both for NHS and Social Services datasets.

Recording of ethnicity

17. In the case of staff, it is assumed that ethnic data will be self-reported. In the case of clients, it is less easy to see who should be responsible for deciding - the case of children is particularly tricky in this context. Unless we have a clear-cut set of rules for how ethnicity data are to be collected administratively, the consistency of reporting will be open to doubt.

Non response

18. A further problem is that many staff and clients are unwilling to categorize themselves on an ethnic basis. This underscores the need for the Department to explain the context in which the data are to be collected and used - this includes the need to achieve better data linkage between datasets than is generally the case currently.

19. The bottom line is that if any group is underreported, for whatever reason, then there is the distinct danger that resources allocated to that group nationally will be less than is required. Social Services staff fully appreciate this point, when it is made to them. But the argument does need to be made, and at a senior level.

Liaison with OGD

20. DH will need to consult with OGD and with LA representatives, CRE and REU etc on "DH ethnic framework", to ensure that it is the same as, or consistent with, ethnic frameworks proposed for use in these other contexts. A meeting of interested parties may be needed to iron out any possible problems.

Religion

21. One further point which may need agreement is whether we will need a field for data on religion along the lines of that in the census, and if so, for what categories. For example, it might be appropriate for certain client groups, but possibly not for NHS/Social Services workforce.

"Map" of DH collections of ethnic origin information

Ethnic Classifications

It is understood that in the 2001 census, ethnicity categories are to be coded as follows:

Outline "Map" of Ethnic Origin Information in Statistical Collections (Actual or in Planning) and Activities