Julie Coope, Associate Director BRG, Simon Swift, surgeon and data expert, and Simon Jones PhD, Professor of Population Health at New York University, discuss how to harness data effectively
The growing volume of detailed data generated from routine healthcare processes presents not only a significant challenge for healthcare organisations, but also an unprecedented opportunity to improve how care is analysed, administered and tracked. Effectively harnessing this data is a complex task and requires robust and reliable collection processes.
The collection, curation and careful use of accurate data is crucial for tracking patient outcomes and ensuring long-term success. Julie Coope is a Registered General Nurse (UK) and Registered Nurse (US), as well as an associate director in BRG’s Global Health group addressing data collection and quality for health systems. She was recently joined by two experts: Simon Swift, a doctor, health economist, business leader and external advisor who worked in the National Health Service (NHS) and as an exited founder of a health data services company; and Simon Jones, PhD, a professor of population health in the Department of Population Health at New York University.
What key challenges do health systems face when trying to collect reliable healthcare quality data?
JULIE COOPE (JC): Data collection and quality management are foundational elements of the modern healthcare delivery and improvement process. The increasing complexity of health systems, coupled with the growing emphasis on evidence-based practice and value-based care, has made robust data collection and management more critical than ever. While some organisations and regions may do this better than others, nobody has mastered this.
SIMON SWIFT (SS): The desire to collect ‘reliable healthcare quality data’ is more complicated than most hospital executives realise. One challenge that faces healthcare organisations involves the variety of sources of data that exist. Some of this is structured data, but much remains unstructured in the form of text and images. A lot of effort is required to transform unstructured data into coded data, and we see increasing investment and support directed to these processes, including to digitise health records and automate processes, but much of this is still done by people, and people are fallible. That means that the data created is not always ‘reliable’.
SIMON JONES (SJ): I agree and would add that we need to ask ourselves two questions when we are interpreting and comparing data. First, what is the motivation of the person recording the data to be accurate? Second, do they use that information for their own purposes? Without ensuring that those collecting health data understand what they are recording, and that it has a use for those teams (as opposed to being fed into some health data collection system), there is a high likelihood that the information will not be as reliable as one might hope or expect.
Understanding data recording can be complicated to collect reliable health quality data; how can we encourage organisations to work towards a single source of truth (SSOT)?
JC: There is a desire for organisations to have one version of the truth, but the motivations and use of that SSOT vary by region and health system. For example, organisations in some regions are heavily incentivised by finance so their documentation focuses more on reimbursement. Other areas may be more heavily incentivised by quality or population health benchmarks. The value placed on enabling reimbursement may be influenced by the unique incentives and priorities in different regions or systems.
SS: It comes back to knowing who is recording this data, and why it is being recorded. For example, when a woman gives birth, it is common for there to be a certain amount of blood loss. Sometimes this is recorded in defined measurements, and other times it’s recorded in more general terms like minor, medium or major. This is only one variation, but there are many in medical procedures, and as a result you find variation in the terminology of data that has been recorded. This variation tends to create different truths.
SJ: Another challenge involves the need for organisations to simplify clinical coding. This can create variation in coding. If you think about a typical clinical consultation, there are often a couple of thousand words exchanged, and there can be complex interactions with multiple measurements taken. That entire interaction is then distilled into several codes which remove context and can lead to different sources of the truth.
SS: Correct. Another consideration is the impact financial incentives have on how consultations are recorded. For example, changing the complexity of how a clinical course is coded can increase or decrease fees and can lead to apparent variations in care which do not reflect clinical reality all the time.
How can healthcare organisations balance advanced data analysis (such as artificial intelligence and machine learning) with the need for foundational data literacy and standardised documentation for more reliable global comparisons?
SS: Many years ago, when I was running a function for the NHS, we collected a significant amount of data related to quality of care from hospitals in the UK. After careful analysis, we published the findings on a public-facing NHS website. Several hospitals were outliers, and the initial reaction by executives from those organisations was anger and frustration because they felt they had been singled out. However, soon after that initial reaction, board members from those same organisations reached out to learn more about what needed to be done to help address the issues highlighted in the benchmarking report. The lesson I learned was that surfacing information based on low-quality data could be a powerful incentive to get the attention and resources of leaders to improve their data collection efforts.
JC: Organisations can use such findings as an incentive as you mentioned, but there is a data literacy gap as well that makes understanding these global comparisons especially challenging. It requires a significant investment to find people who understand what the data means and to do the necessary inputs, collections and coding correctly. This requires engaging people across an entire organisation at all levels.
Given the disparities in resources and organisational maturity across different regions, how can we ensure meaningful comparisons of data or outcomes?
SS: This is a really difficult question to answer. I would start by looking to see if the data standards and classification systems used are comparable. What incentives or disincentives are there to use specific codes; and, at a fundamental level, what are they recording?
SJ: Another key consideration is understanding the workforce and defining the roles and responsibilities of those clinicians involved in treating and collecting patient data. The way in which these roles are defined and how care teams treat a patient varies widely. This variation inevitably leads to disparities in how organisations are resourced and in patient outcomes. One way to gain more context is to focus on collecting and reviewing qualitative data. What do patient satisfaction surveys tell us, what is being shared in social media, etc.?
JC: I agree that understanding incentives is crucial. For example, if there is an incentive to document a particular diagnosis or procedure in one region versus another will lead to disparities. Also, understanding what patient pathways and organisational structures are used to diagnose and care for a patient is crucial, because different regions address this very differently.
SS: An example of narrowing the focus would be to look at readmission rates. One would need to know many system factors to develop such a comparison, but if, for example, you look at readmission rates following primary coronary artery bypass surgery in women over a certain age, you’re less likely to have system factors to address. In addition, looking at readmission rates will engage the clinicians on the ground whose now ‘metaphorical’ pen is responsible for the data creation.
Summary
Accurate, comprehensive healthcare data can provide organisations with a strategic advantage and better position health systems for future success. By prioritising rigorous data collection and establishing a single source of truth, leaders can make more informed decisions, optimise resource allocation and better forecast future demands. Investing in robust data capabilities will ultimately deliver better patient outcomes and more sustainable services for your organisation.
CONTACT INFORMATION