Data is Everywhere but Nowhere Around Me! wrap up

What another amazing evening with a vibrant discussion on such a key topic.  We’d like to thank our new sponsor Nous Group for their support and, of course, our panellists, moderated by Prof. Mei Krishnasamy, University of Melbourne:

  • Prof. Andrew Wilson – Chief Medical Officer for the State of Victoria;
  • Kira Leeb – Executive Director, Health & System Performance Reporting, Victoria Agency for Health Information (VAHI); and
  • Mike Roberts – Clinical Academic Lead for Population Health, UCLPartners, UK and Clinical lead for NHS COPD and Asthma Audit, Royal College of Physicians, UK.

Data is always such an all-encompassing topic and the insight provided by the panel was impressive and detailed.  Here’s the wrap up:

  1. People need data that is relevant and understandable.

The key is it needs to be both relevant and understandable for the specific audience, whether clinician, chief executive or consumer.  Clinicians need to see the connection between the data they create and the data they receive.  Currently, in Australia, this connection seems to be limited.

  1. Patient Reported Outcome Measures (PROM’s) tell a story at scale for the system.

The UK’s experience with PROM’s is mainly surgically-related with the key issue being it’s a discrete measure (a point in time or before vs. after) but not great at measuring a continuum of care for an individual’s pathway.  Ultimately, it’s not a great care metric.  However, over a period of time, it can start to inform or make an impact on actual practice as more of people’s metrics are collected and able to be analysed and compared.

  1. Patients need to be supported to ask questions.

Patients need to be supported to ask questions of the clinicians related to their own situation and expectations.  For example, when considering knee surgery, a patient could ask the surgeon if 4 key things will change for them as a result of the operation.  If the surgery can’t deliver better kneeling, perhaps the patient should elect not to proceed if that was a key improvement expectation.

  1. Clinical Quality Registries (CQR’s) need a strategy in Australia.

CQR’s are expensive and difficult to run and, certainly in Australia, there is patchy application.  As with all data, the general rule of “rubbish in, rubbish out” applies, even for CQR’s.  Regardless of the source, it’s what you do with the data that is really critical.  The NHS audits, run for many years, have highlighted data hasn’t really changed treatment patterns but linking the clinical audit to improvement programs has demonstrated real results.

  1. UK experience: clinical audits coupled with improvement can drive significant change.

Clinical audit linked to improvement is the UK way and it appears to be generating significant success.  There are currently 26 national audits around the NHS and here are some key findings:

  • Reduced number of metric collected, tracked and focused on (from 60 to 6) and provided real-time feedback;
  • Added “best practice tariff” that provides standard payout of 85% for general achievement but increases to 100% if the hospital successfully completes the 2 key items shown to be most impactful;
  • Tariff is returned to clinical unit which allows clinicians to negotiate with hospital senior executives for business case support and funding investment.

This has shown remarkable results (28% improvement) in changing treatment patterns.  However, it’s starting to plateau; so, additional work is needed to modify and tweak the 6 current metrics.

The UK is also starting to consider self-sustaining audits which means organisations will need to identify how best to deliver the data going forward.

Final thoughts for Australia

Clinicians and the government both need to take a proactive role within the system to bring data, from all aspects, to all players in health.  Clinicians also need to own their role in the generation of the data and certainly be suspicious of data but not dismissive.  The coupling of audit to focus improvement efforts would be a great first step.