A data analyst, a QPI manager, and a doctor, walk into a bar but cannot agree on a table. Their disagreement is not surprising if we understand their contrasting perspectives on data.
All three are seeking the same goal, to improve the delivery of health care to their community. All three are interested in looking at data as one way of achieving this goal.
Policymakers seek data that identifies (at least) a minimum acceptable level of healthcare and warns of any significant deterioration in those standards. Hospital administrators seek to ensure that healthcare is cost-effective and equitable.
QPI managers seek to support statutory reporting obligations and identifying areas for improvement. The clinician seeks to improve professional standards and patient outcomes. Each perspective has its strengths and limitations.
The clinician is trained in evidence-based medicine and prefers data derived from well-constructed randomised-controlled trials and peer-reviewed.[1] He is aware that much of the published research findings are probably false.[2]
He is suspicious of clinical data collected by non-clinical professionals for nonclinical purposes. Most of us have a wealth of clinical experience yet lack formal training in data analytics, biomedical statistics, or business management.
The data analyst is trained to understand complexities of data and the myriad of statistical methods and is convinced of their utility [3] but has limited clinical experience or training to determine their priority or significance. The hospital manager sits somewhere in the middle of these two. With no formal training in either clinical or statistical methods she relies on both.
To benefit from all these varied perspectives, we need collaboration. Alliance in our skills, honesty in our own ignorance, and trust in each other. For this reason, VAHI has taken the brave step to appoint a senior clinician for 12-months to works with their team.
As that senior clinician, what I have I learnt so far? First there is an amazing group of dedicated professionals working with large complex clinical datasets that are, or will be, of incredible benefit to the delivery of healthcare. Their desire is to improve reporting of clinical data back to ‘the coalface’; to the clinicians and managers, politicians and public servants, who make the myriad of decisions that affect health care delivery in Victoria.
Reliable data are now available from numerous sources - administrative (coding) data, clinical registries, statutory organisations, and statistical bureau. The strength of these data is in their comprehensive nature and high quality. They cannot answer all our questions, but they can provide insights we could not otherwise secure.
Data are powerful. Reliable data tell us about the real world; which may be different to my personal perception of reality.[1] I am repeatedly surprised that my perception of reality is sometimes misleading or incomplete when we look at comprehensive data. But even the best data has their limits. Data show us what we did rather than what we should do. Like any new technology, data are ambiguous in the sense that they do not tell us how to use it well.
One way to harness data effectively is through collaboration. Maybe then we can all sit down and drink together?
Dr Graeme Duke has recently joined VAHI as Data Analytic Fellow, in addition to his roles as Deputy Director, Eastern Health Intensive Care Services in Melbourne, and clinical lead for intensive care research. To get in touch with Graeme about how we can collaborate to use data more effectively, contact [email protected].
References
- Collins R, Bowman L, Landray M, Peto R. The Magic of Randomization versus the Myth of Real-World Evidence, N Engl J Med 2020; 382: 674-678
- Ioannidis, JPA. Why Most Published Research Findings Are False. Plos Med 2005;2(8):e124. Available from: http://doi.org/10.1371/journal.pmed.1004085.
- Aylin P, Bottle A, Majeed A. Use of administrative data or clinical databases as predictors of risk of death in hospital: comparison of models. BMJ 2007;334(7602):1044–1044. Available from: http://www.bmj.com/content/334/7602/1044