Using routinely collected data

1
Using routinely collected data

• Dr Colin Fischbacher
• Information Services Division
• NHS National Services, Scotland

2
Talk outline

• Potential role of routine data
• The current situation
• Primary care
• Secondary care
• Other sources
• Some conclusions

3
Why use routine data?

• Considerable existing investment in data
  collection
• Very large datasets with universal coverage
• Marginal cost of adding ethnic group is
  relatively small
• May increase profile of ethnic health inequalities

4
The vision a routine system for collecting
ethnic information

• Based on self-definition
• Collected with informed consent
• Agreed categories, compatible with census
• Collected once (probably in primary care) and
  transferable using standard data format
• Collected by fully trained staff
• Records everything needed for appropriate care
  (origin, religion, language, other?)

5
What could this provide?

• monitoring uptake of services
• targeting services
• ensuring care is appropriate
• policy development
• performance management

DOH. Collecting ethnic category data. Oct 2001
6
Drivers in Scotland

• supporting patient focussed care
• demonstrating compliance with legal obligations
• investigating ethnic variations in health and
  health care provision

7
Recent developments in Scotland

• Fair for all audit
• ISDs Equality and Diversity Information
  Programme
• Diversity audit of health databases

8
The current situationprimary care

• information on ethnicity not collected in
  Scotland
• largely incomplete in England?
• some local efforts based on mailshots
• as in Liverpool 58 coded
• but very intensive (70p/patient)

9
Issues in primary care

• ethnicity as a clinical condition rather than a
  demographic characteristic
• ensuring both standardisation and flexibility for
  data collection
• availability of standard Read codes
• ability to share data

10
The 2006 GMS contract

• Includes 1 point in organisational domain for the
  collection of ethnic group information
• . . . but does this include electronic recording?
• . . . and do GPs think it worthwhile?
• . . . effect on awareness of the issue?

11
The current situationsecondary care

• Field for ethnicity in Scottish SMR databases
  since 1996
• recorded in only 9.2 of hospital episodes (46.2
  in one Ayrshire acute trust)
• until 2004 used ethnic grouping that did not
  match the Census categories
• Although not mandatory it is strongly
  recommended that these items be completed
  whenever the information is available

12
The current situationsecondary care

• Ethnic monitoring in England
• workforce (1991) patients (1995)
• guidelines, training for staff
• Completeness of HES variable

13
High HES completeness

• Blackledge (BMJ 2003) reported higher incidence
  of heart failure among South Asians in Leicester
  using HES data
• based on self reported coding for ethnicity
  coverage thorough validated using name search
  methods

14
Lower HES completeness

• London Health Observatory review of data 1997/8
  2000/01 found
• 37-38 ethnic group not known
• Valid codes in around 66 of cases

http//www.lho.org.uk/Download/3nhjq2aa2pnxxbmora3
szquu/live/8907/EHIP_Update_4.doc
15
Ethnicity data incompleteness in HES () by
Government Office Regions, England, 2003/04
SourceIndications of Public Health in the
English Regions Number 4 Ethnicity and Health.
Association of Public Health Observatories, Oct
2005
16
Issues with HES

• How are the data collected?
• Can we ensure self-definition?
• How often do systems default to white?
• How well trained are staff?
• Is the situation changing?

17
Other databases

• Cancer registration (ethnic group available in
  18 in Scotland)
• Child health (CHI database in Scotland) median
  22 complete
• Health Visitor/District Nurse databases
• Lothian database said to be 100 complete
• Scottish Birth Record
• Diabetes registers
• GUM clinics (94 recorded)

18
Issues for other databases

• All incomplete
• Few or none match current census categories

19
Another routine source

• Two health questions in the Census

20
Some other issues

• Data standards (NCDDP in Scotland)
• www.datadictionary.scot.nhs.uk
• Ambivalence among clinicians
• Persistence of race as a clinical issue, eg in
  relation to haemo-globinopathies and diabetes

21
. . . some other issues

• low salience in areas with small ethnic minority
  populations
• informing patients about the reasons for
  collecting these data
• ethnicity in the context of the wider diversity
  agenda

22
Conclusions

• clear policy commitment
• clear legal drivers
• information vacuum
• (with some variations)
• ideal solution some way off
• may be a place for interim approaches