From ‘Big Data’ to Collective Wisdom – SIHI conference report

This year’s Southern Institute for Health Informatics (SIHI) conference, in Portsmouth on September 11th, was a hugely impressive event featuring a well-chosen array of excellent speakers. As one who has often struggled to get excited about the finer points of coding, this was a revelation: I was converted.

Just what can be achieved in the NHS when technology, culture and organisation are in harmony was breathtaking – examples included:

• reducing medication errors in a hospital closely tracked by significantly reducing mortality;
• how good data analysis complemented and directed inspection in the Keogh reviews;
• GIS and timeline visualisation enabling pattern determination to understand detailed infection dynamics;
• an improved early detection system for sepsis that is saving many lives by enabling earlier intervention.

As many speakers pointed out, technology is now rarely a limiting factor.  Computerisation can:

• improve patient safety;
• reduce discontinuities within & between organisations;
• reduce the unit cost of processes;
• remove tasks from humans;
• reduce waste.

However organisational and cultural issues often get in the way, such as:

• ignorance of the unit cost of processes;
• a passive approach to safety;
• organisational barriers, particularly to information sharing;
• unwillingness to use automatic identification & data capture (eg barcodes, RIFD);
• unwillingness of people to take decisions – examples given ranged from junior doctors to trust CEOs.

One speaker pointed out that “airlines don’t do RCTs” yet understand far better how to use technology to manage safety, improve customer experience and improve efficiency than the NHS.

The call for consistency and interoperability of data was strong.

A natural application of Big Data will be to pharmacogenomics in order to understand effects on a sufficient population of each genotype.

An approach that was clearly working well in one hospital was to treat all errors as important and avoidable, irrespective of whether they seemed important – technology helped greatly there (whence the much-reduced mortality referred to above).

There were lots of comments made, including from the floor, about coding problems – secondary care came in for particular criticism as clinicians typically have little visibility of the coding process, and there are perverse incentives, so data quality can be very poor – recording once, at the point of care delivery, should always be the objective. (One aside that lightened the mood in the conference was of a hospital where diagnosis of depression was very high until it was realised that the Read code for depression was the same as the short dialling code for the path lab.)

One presentation that particularly appealed to me was the use of Bayesian techniques to enable simple monitoring devices to measure vital signs – I’d never quite understood before how for example the camera on a smartphone could measure SpO2 so accurately without the closely-tuned twin semiconductor lasers that conventional pulse oximeters use. The same technique is being used to predict patient exacerbations, well in advance of them occurring, and even to segment patients on entry into those likely to have good and those likely to have poor outcomes, so that resource could be deployed accordingly.

There was general agreement that the sizable NHS England Technology Fund would be a significant help in encouraging trusts to use technology better – applicants were encouraged to focus on areas like improving processes (and less on aggregating data) and transfer of good practice.

The only drawback of the event to me was that it operated under Chatham House rules.  I began the event tweeting, interpreting said rules as avoiding quoting any risqué remarks.  However a discussion in the first break with one of the organisers made it clear that I should only attribute “motherhood & apple pie” statements so Twitter was turned off, and sadly I’ve not been able to give credit here to presenters, every one of whom was truly outstanding. (We’re hoping to get a few of them to present at the RSM’s Big Data event in June 2014)

Book early for next year!