Patient outcomes: Are we acting on the data?

The experts agree: Health systems can get better if they study the patient data, to see what produces the best outcomes. But in practice, questions abound. How do you avoid information overload? Do you gather information from the patients alone, or also from their family? Should hospital budgets be linked in some way to this information?

“We’re a long way from being able to package data in a way that is meaningful and actionable,” says Ian Forde, senior health policy analyst at the Organisation for Economic Cooperation and Development.

At a Science|Business Healthy Measures conference in London December 16th, he and other experts discussed the many challenges in analysing and acting on data about patient outcomes. The conference was hosted by the University of Warwick Medical School. (See prior reports from this conference here and here.)

A case study in success

The potential benefits of working with patient data were highlighted by an international study of hysterectomy outcomes, led by Meghana Pandit, consultant gynaecologist, chief medical officer and deputy chief executive of University Hospitals Coventry and Warwick (UHCW).

Over 32,000 cases of hysterectomies carried out in the UK, US and Australia were analysed in the study. It concluded that while newer, less invasive vaginal or laparoscopic procedures are safer overall, they do carry a risk of complications. Where these occur, there is a higher risk of haemorrhage and readmission to hospital than with the older, abdominal procedures.

As a result of the research, UHCW set up morbidity scorecards for every single surgical speciality. Groups of surgeons can look at their figures monthly. “We can benchmark why someone is doing better ... and support those who are not doing so well,” Pandit said.

The hysterectomy study drew on data collected by the healthcare informatics company Dr Foster, as part of its Global Comparators programme. As a member of the programme UHCW is able to compare its performance to other hospitals around the world. Pandit said one result, unrelated to the hysterectomy research, was the discovery that UHCW was an outlier in terms of treating heart attacks. The response was to train paramedics and to send people with a suspected heart attack straight for a CAT scan on arrival at A&E.

The problem of Alzheimer’s

A success story is always welcome, but different diseases pose different problems. Take Alzheimer’s, for which no new drug  has been approved since Aricept, from Pfizer, was registered in 1996.

In the light of a litany of expensive failures, three years ago the charity Alzheimer’s Disease UK (ADUK) reassessed its research strategy. “We changed the way we do research and got into early stage discovery,” said Alison Evans, head of impact and innovation at ADUK. In 2014 the charity launched a campaign to raise £100 million to invest in the full spectrum of dementia research from fundamental neuroscience to prevention.

Drugs that failed in development have shown a trend towards efficacy and it is thought that by the time Alzheimer’s symptoms are overt, the damage is irreversible. There is now a considerable effort to identify subjects in the early stages of disease, to get the right patients into clinical trials and demonstrate disease modification.

However, for any drug that does receive approval, the effect is likely to be modest and to play out over many years. Despite the huge economic and social costs of Alzheimer’s, that will make it difficult to demonstrate a clear cost benefit and to get reimbursement. ARUK is working on new assessment methodologies. “We have to make sure patients get access – so is the value that patients see the same as what the system values?” Evans said.

Actionable data

Since 2003, the OECD has been collecting Health Care Quality Indicators which are widely regarded as one of the world’s most reliable sources of comparable data in healthcare. To date, this has covered hard statistics of costs, activity and inputs, but now a new generation of health statistics will include measures of patient outcomes, said the OECD’s Forde. “The data are also more granular, so they are not just looking at national averages, but variations within countries,” he said. One illustration of the potential value of such data: survival rates following a heart attack can vary seven-fold within one country.

“We want to make the data as actionable as possible, so we culled indicators by one third recently,” Forde said. The OECD is collaborating with the European Commission and the World Health Organisation-Europe on the development of stronger country assessments and inter-country comparisons, to facilitate debate and help member states make use of data to improve healthcare systems. At present “only a small portion” of data collection is concerned with outcomes, which is a new field of work for the Commission, said Federico Paoli, a policy coordinator in the European Commission’s Directorate-General for Health and Food Safety.

In addition to data collection, the Commission, with OECD and WHO Europe, is working on how to apply data once it is gathered in and analysed. “We are mainly addressing policy makers to support them to use existing outcomes indicators,” Paoli said.

Obstacles to applying outcomes data

First and foremost, some experts say, healthcare systems have to switch their basic model from funding activity – in which providers get more money if a patient comes to a hospital five times rather once to complete a single episode of care – to paying for outcomes. The problem is finding a way to link payment to outcomes without creating false incentives. As Forde noted, the National Health Service in the UK has a national scheme for relating payment to outcomes in hip replacement. The question is, “is it working? Does the average hospital find the data useful?”

In the UK, some outcomes-based contracts have been awarded, in which 20 per cent of the funding is subject to long-term outcomes. Some of these contracts have not worked very well, though the approach is still in its infancy.

Collecting outcomes measures

If outcomes measures are hard to define, they are also hard to collect. “This is particularly true for patients with co-morbidities who are in the community,” said Forde. “How do you survey them in a sensitive way and then see how to attribute wellbeing to the healthcare system?

That is a big issue in dementia, where carers are often the source of data, Evans of the Alzheimer’s foundation said. “Do you measure carer’s views? Physically, how do you get the data?” she said.

For Kiran Patel, medical director for NHS England, the problem is not so much that outcomes are hard to measure, but that healthcare systems do not know how to assess them. “Alzheimer’s is a good example, carers can contribute a lot more than we use them for at the moment,” Patel said.

A further difficulty relates to the number of different systems in which data relating to a single individual may reside, making it hard to build a complete, integrated picture. Sally Smith, chief operations officer of My Clinical Outcomes, which provides a web-based platform for the collection of patient-reported outcomes, believes such digital technologies are the only practical means of collecting data and then providing something back to the patients. “That lowers the barrier. You give them a chart and information relating to their condition, tell them about care and how it matters. That improves the quality of care,” Smith said.

While the Commission’s Paoli agreed information technology helps, there remain issues of data entry, ensuring there is compatibility between systems, and having common definitions. A further concern is the timeliness of data. If datasets are not aligned time-wise, country-to-country and intra-country comparisons are undermined, while there are obvious risks in framing policy around old data. “We’ve got to reduce the time from data collection and analysis, to distribution,” Paoli said.

Healthcare professionals are already charged with significant data collection duties. To gain their cooperation in collecting outcomes data, UHCW’s Pandit said the data has to be meaningful and align with their values. “You’ve got to go back constantly to say this is the result; this is how it will benefit you and your patients,” she said.