Richard Smith: Can the NHS ever manage to analyse data to improve patient care?

richard_smith_2014In order to improve patient outcomes the NHS badly needs to collect data on all that is happening, analyse it intelligently, and present the information to clinicians and managers in an understandable and actionable form. That was the central message from Keith McNeil, who recently became chief clinical information officer health and social care for the NHS, to the Cambridge Health Network last week.

Previously, he was a clinician for 30 years and chief executive of Addenbrooke’s. His message, as I cynically observed, was probably said 15 years ago (if not longer) but it hasn’t ever happened. Now it has to, insists McNeil, if, for example, England wants to improve from 26th out of 29 OECD countries for cancer outcomes to the top.

McNeil began his talk by describing how one time when he was at Addenbrooke’s the hospital was completely full, and he and clinicians had to meet every day to decide which patients would have to have their non-emergency operations postponed. Some of the patients had cancer, but McNeil and the others had to make decisions with inadequate information. “Sometimes we got it wrong.”

The National Information Board (NIB), which McNeil chairs, is leading on the programme to improve the use of data within the NHS. It’s doing it in partnership, and its primary job is to provide a platform to “plumb together” the many disjointed information systems within the NHS. Somebody in the audience pointed out that NIB has 28 members from the public sector and only four people representing patients and the private sector; the questioner added that he’d been at a meeting where the four were effectively isolated.

Marc Farr, director of information, East Kent Hospitals NHS Foundation Trust, said that one of the NHS’s problems was muddling up three different functions needed to produce improvement through data analysis. One function is to provide reliable information technology, the next need is to develop systems to gather data and standardise them, and the final function is to analyse the data to produce useful information. They are, he emphasised, different functions that need different skills, but all three are needed.

David Rose, who is now chief executive of Dr Foster and has been chief executive of several NHS hospitals, emphasised how the NHS gathers a massive amount of data but largely fails to use it intelligently. Energy is misplaced. In order of time and energy most is spent downloading and gathering data, followed by preparing reports, analysing data, and ultimately using the data to make decisions. It should, he argued, be the other way round. He described how the NHS has 6000 performance metrics for hospitals and 584 for primary care. These metrics are assembled into reports, which, he argued, “are not making a difference.”

Rose’s prescription was “cut out the crap,” standardise information to have a “single source of truth,” apply statistical rigour, and establish clear responsibility for whole system intelligence.

Somebody in the audience pointed out that the speakers had spoken as if the failure to use data to improve care was a “supply side problem,” but he thought that it was just as much—if not more—a “demand side problem”: clinicians and managers are not asking for intelligence from data. The questioner thought that patients might be more receptive.

Another member of the audience thought that the NHS has far too few statisticians and that it was almost impossible to find anybody within the NHS with whom you could have “a serious conversation about statistics.” We are, he said, “building a Ferrari but have no driver.” Statisticians, it was pointed out, can earn far higher salaries in banks or software companies than in the NHS. “We need to transform the workforce to make it numerate,” said somebody.

The audience agreed with McNeil’s vision of using data to improve patient care but seemed sceptical about the NHS getting the job done. One man said he was depressed because nobody had talked about gathering data on outcomes that matter to patients, while another said he was depressed because the story he had heard was of people busily gathering data to produce reports that nobody wants and can’t interpret anyway.

After the disaster of, when the NHS had to abandon its attempt to gather data on all patients, better and more concrete stories are needed on the benefits for the public and patients of better use of NHS data. Improving outcomes for patients with cancer might be a good starting point, and McNeil pointed out that the project of sequencing 100 000 genomes has the potential to improve outcomes—but only with high quality analysis of the massive amounts of data produced.

Richard Smith was the editor of The BMJ until 2004. 

Competing interests: None declared.