#OpenDataSavesLives: Session 37 - Health Inequalities

It’s great to see that our events continue to grow with 60+ attendees. This session follows on from our last event on the same topic in October. Here is the write-up and recording from the last session, where we discussed the current state of inequality in access and outcomes in healthcare. This time we looked more closely at the challenges of collecting, sharing and processing ethnicity data to understand health inequalities.

Speakers:

• Sarah Scobie, Deputy Director of Research, the Nuffield Trust
• Dr Marc Farr, Chief Analytical Officer EKHUFT & Co-founder of ODSL

Sarah Scobie, Deputy Director of Research, Nuffield Trust

The Nuffield Trust are a health think tank undertaking research, assessing evidence and engaging with policy makers to improve health and social care. They have a huge portfolio of ongoing work into health inequalities looking at specific demographic groups and trying to understand the wider factors that influence access and outcomes, with a particular focus on elective care. They’re also looking at how data quality, including how ethnicity is encoded in equality data, can be improved and made more reproducible.

Challenges of ethnicity analysis:

Sarah explained some of the challenges of handling ethnicity data to unwarranted variation in treatment. But why is this analysis so challenging? Variations themselves are complex, as ethnic differences in condition prevalence, access to treatment and outcomes are the result of multiple factors. There is a question about deprivation, with significant links between health variation and socioeconomic background. We all know that engaging with the health service can take some effort and persistence and can be particularly difficult for those with challenges in their lives that must take a higher priority than health.

One issue is the incomplete, inconsistent and inaccurate coding of ethnicity. For example, inpatient, outpatient and A&E systems encode ethnicity differently within and between trusts, with inconsistencies between hospital data and census making analysis and comparison against national populations more challenging. There aren’t just differences in how ethnic groups are encoded but also systemic bias caused by miscategorisation by healthcare professionals. Some of Sarah’s research unveiled that more than a third of patients from a minority ethnic group were miscategorised as ‘Other’ ethnicity, and ethnicity data was often missing entirely.

You can take a closer look at Sarah’s research and methodology on the Nuffield Trust website. All of their code is available on Github. We ran a session on COVID and ethnicity in the midst of the pandemic that touched on how ethnicity is encoded in the health service - see notes from the session here.

Links from the Nuffield Trust

• Childhood obesity: is where you live important?

  This new Nuffield Trust analysis looks at the association that neighbourhoods, communities and their characteristics have with obesity in children.

• Ethnicity coding in English health service datasets

  This report looks at the quality and consistency of ethnicity coding within health datasets.

• The elective care backlog and ethnicity

  How did the fallout from the pandemic affect people across different ethnic groups, and was the impact of those cancelled procedures spread evenly?

• Quality and Inequality

  How have inequalities in the quality of care changed over the last 10 years?

• The Nuffield Trust Github Repository

• 
  

Links from the audience

• Understanding Patient Data

  Putting people at the centre of decisions about patient data

• Understanding Patient Data

  Insights from Black & South Asian people on patient data

• Ethnicity data: data quality resources

  Methods and quality reports, blog posts and other resources on ethnicity data quality issues.

• Ethnic group, national identity and religion

  A guide for the collection and classification of ethnic group, national identity and religion data in the UK.

We had some excellent discussion around how we maximise the impact of this work and allow others to pick up and build on the progress they have made. Sarah raised the matter of getting the key messages from their research in front of senior decision makers - they are too busy to read a 50 page report, so how do we share the most important message so change can be made nationally?

This, I think, is where the value of informal networks and the web is clear. By convening through a platform like ODSL, sharing code and methodology on Github, connecting with others and sharing progress, we are providing more contact points for feedback, collaboration and reducing duplication of effort. This becomes especially important where there are multiple actors, like NHS trusts, all working towards the same goal of reducing inequalities but adopting different approaches and working in silos, finding different solutions to the challenges of improving data quality and systemic bias toward ethnic groups.

Marc Farr, co-founder of Open Data Saves Lives and Chief Analytical Officer, EKHUFT

We need to start working together in figuring out how to code some of the analysis on inequalities data. Knowing where to start in performing this analysis can be a challenge for trusts, therefore some regions are establishing some governance to offer guidance on how to report and start to address inequalities. An Inequalities and Unwarranted Variation Committee has been set up at EKHUFT for this purpose who have been developing some constitutional standards for a more systematic approach to measuring inequalities. We’ll follow up and signpost on their progress and approach for others to benefit over the next few months.

Unfortunately, Daniel Hayes was unable to join us on the day but he’s done some fantastic high profile work on inequalities at UHCW. Their research highlights two challenges to overcome as health services recover from the impact of the pandemic. The first is ensuring that restoration doesn’t increase inequalities, and the slide below alludes to how this might happen. The second challenge then, is ensuring that restoration post-COVID helps to reduce inequalities instead of widening the gap.One of the challenges of service recovery after the pandemic is ensuring that we improve rather than worsen inequalities.Credit: University Hospitals Coventry and Warwickshire NHS Trust.

Conclusion

For me, there were two themes that emerged from the session. The first is around methods, data and code. Where is the starting point for conducting analysis on inequalities, how do trusts ensure they have the skills and resources to address them, and how do they report and present their research? We’re running our next event on RAP for Health Inequalities on the 10th May. One of the reasons for this is to provide a platform for conversations around code, because getting to this level of detail raises questions around data quality, information governance and methodology that otherwise would be left to organisations to figure out in silos and slow progress.

The second question seems to be one around information sharing. Sarah raised an interesting point that once the research is done and the findings presented, how do we then communicate this to those who can make decisions, and inform change? People are busy, so how do we inform the right people in the right way? This is where informal networks, community and working in the open are especially valuable. We’ll keep coming back to this over the next year and as the ODSL programme evolves.Our next event will be on the 10th May - don’t forget to register on Eventbrite. And as always, we’d love to hear from you if you are interested or working in this space.A huge thank you to our sponsors and supporters - their funding makes a huge difference and keeps the programme running.