Why poor quality of ethnicity data should not preclude its use for identifying disparities in health and healthcare

Abstract

Background: Data of quality are needed to identify ethnic disparities in health and healthcare and to meet the challenges in governance of race relations. Yet concerns over completeness, accuracy and timeliness have been long-standing and inhibitive with respect to the analytical use of the data. Aims: To identify incompleteness of ethnicity data across routine health and healthcare datasets and to investigate the utility of analytical strategies for using data that is of suboptimal quality. Methods: An analysis by government office regions of ethnicity data incompleteness in routine datasets and a comprehensive review and evaluation of the literature on appropriate analytical strategies to address the use of such data. Results: There is only limited availability of ethnically coded routine datasets on health and healthcare, with substantial variability in valid ethnic coding: although a few have high levels of completeness, the majority are poor (notably hospital episode statistics, drug treatment data and non-medical workforce). In addition, there is also a more than twofold regional difference in quality. Organisational factors seem to be the main contributor to the differentials, and these are amenable-yet, in practice, difficult-to change. This article discusses the strengths and limitations of a variety of analytical strategies for using data of suboptimal quality and explores how they may answer important unresolved questions in relation to ethnic inequalities. Conclusions: Only by using the data, even when of suboptimal quality, and remaining close to it can healthcare organisations drive up quality.

Who comes back with what: a retrospective database study on reasons for emergency readmission to hospital in children and young people in England

“…This may be associated with incomplete or erroneous identifiers, thereby reducing the chance of linkage to a subsequent admission 16 28…”

Out-of-hospital cardiac arrest in South Asian and white populations in London: database evaluation of characteristics and outcome

“…First, a large proportion of patients were not assigned to an ethnic group, which reflects the reality of ethnic group coding in the NHS (and in most of Europe) 12 23. There is no reliable way to assess selection biases.…”

“…Separate codes (Z1–Z4) are used if patients are unable or unwilling to state their ethnicity. Recording ethnicity within the NHS is difficult and incomplete,12 and for obvious reasons, is particularly difficult during an emergency. Patients were categorised into one of two major ethnic groups—white (including white British, white Irish, and white from any other background) and South Asian (including Indian, Pakistani and Bangladeshi).…”

Responding to diversity: An exploratory study of migrant health policies in Europe

“…The objectives of the Catalan Immigration Master Plan for Health include the development of better data collection systems on migrant health and health service utilization. The National Health Service in England has long required collection of data on ethnicity in activity data, but the quality of recording is variable [32].…”