Insight for diagnosis
A child's life-threatening misdiagnosis and a father's determination to help doctors better recognise rare symptoms in the future led to the formation of Isabel Healthcare. Sian Harris finds out how
When Isabel was three years old she caught chicken pox. This is a common enough thing to happen to a child but she didn’t recover in the same way that most other children do. In fact, she seemed to get worse. She had a fever and diarrhoea, and was not improving with paracetamol.
Her parents took her into the accident and emergency department of their local hospital, where staff reduced her fever before sending her home. ‘Each time we were told not to worry; it was just normal chicken pox,’ recalled her father, Jason Maude.
But Isabel did not get better; over the next few days her condition got worse. Within a couple of days she’d become very listless, with blisters on her groin and very low blood pressure. When her parents took her back to hospital she suddenly became delirious and a crash team had to be called to resuscitate her.
‘The clinicians had no idea how sick she was until it happened. This was almost the most frightening thing,’ remembers Maude. Isabel spent three weeks in intensive care, including having a heart attack. She was defibrillated 22 times and spent four weeks in the high-dependency unit.
Her chicken pox had brought on the complications of toxic shock syndrome and necrotising fasciitis, a flesh-eating bacteria.
Fortunately Isabel survived and is now a healthy teenager looking forward to starting university. But many who contract necrotising fasciitis do not recover or are left with serious long-term problems. Much of this comes down to the same problem that Isabel experienced: the failure of clinicians to recognise the symptoms of a rare condition or complication.
The seed of an idea
Although doctors are taught about very many things in medical school and keep up to date through reading case notes and other literature, they may not immediately think of a particular rare illness at the time that it is presented to them in a patient. Even if they remember that they’d read about something similar two years ago they may not immediately be able to find the required information.
‘We went to see the medical director where she was treated. He said that he lectures and writes papers on chicken pox but that it is difficult to get the message across. If you were a junior doctor, how would you recognise something that’s unusual? It’s a really difficult job,’ says Maude.
Maude and his family could have simply taken Isabel home, complained about her treatment and then got on with their lives. But the experience sowed a seed in Maude’s mind.
He had a background in finance. As head of equity research at Axa he’d had a team of analysts applying semantic techniques to financial data and information. He wondered if the same approach would be possible with medical information so that clinicians could search by symptoms and find the possibly rare conditions that they might otherwise have missed.
‘I thought if you have software that is clever you could apply it to medical knowledge in a more intelligent way. The medical information is all written down but you want to search by symptoms,’ he explained.
Maude and others set up a charitable trust to do just this, putting in their own money and using donated software. By 2001 the trust had a tool that worked – for paediatrics only, initially – and was made freely available on the internet. ‘We wanted to remove every possible barrier to doctors using it,’ he said.
However, as uptake grew they realised it could not be sustained on a charitable basis so, in 2004, the charity became a shareholder in a company called Isabel Healthcare. The company raised investment to launch an adult version of its tool in 2006. Today, said Maude, the tool works across all age groups and all diseases.
He describes the resource as a core database of medical symptoms – like ‘a medical Google’. It includes 10,000 diagnoses, 6,000 diseases and 4,000 drugs, from multiple sources.
The tool works using pattern recognition software, based on open-source software applied to a database of disease presentations. This approach means that users do not need to know specific keywords. Results are ranked and filtered to make them as relevant as possible – by, for example, age, gender and geography of the patient.
‘Users can put in any permutation and this makes it hard to test,’ Maude conceded. He said that the tool needs to come up with 10 good suggestions and, if even one of those is slightly out, then the tool will lose the credibility of the doctor. ‘We are constantly checking against cases, which is something we can only do over time.’
Balanced against this is a need for fast response times. He said that results can be delivered in less than three seconds over normal household broadband connections.
Today, the Isabel tool is used in clinical settings from primary care to specialists, to check that something hasn’t been missed in cases where a patient is not responding to treatment. Maude gave a recent example of the tool helping to identify the condition – and save the life – of a patient with learning difficulties who was not able to communicate their symptoms well. In other examples it can be used to identify illnesses where the most easily-recognised symptom is not present. He gave an example of somebody who had shortness of breath, nausea and fatigue – but no chest pain – and they turned out to have had a heart attack, a diagnosis suggested straight away by Isabel. ‘The crucial issue is recognition of a disease,’ he said.
But one of the challenges is that some in the medical community regard using such a tool as a threat, and have trouble imagining when they might use it. Maude argued that it should not be seen like this; its aim is to help. ‘Using Isabel can bring potentially relevant things to the fore of the mind. It’s like having colleagues around who can say, "Have you thought of this?"’
‘There is a huge body of knowledge constantly being churned out but doctors don’t need the knowledge when it suits somebody else to publish it; they need it when they have a patient with the disease. We deliver the knowledge when it is needed,’ he continued.
Of course, the information itself is an important part of the story too and Isabel Healthcare has agreements with the BMJ Group, Lippincott Williams & Wilkins, Dynamed and others. All versions of the product provide links to PubMed.
‘We can tailor the system so that you can go to the resources you normally have access to. If, for example, you subscribe to Dynamed we’ll provide you with Isabel Dynamed,’ he explained. ‘We don’t write any medical knowledge. That’s why we partner.’
Maude has high hopes for the tool: ‘I want every clinician in the world to have access to Isabel. It is also a way of making much better use of medical resources, for example where there is a shortage of doctors it is a way of spreading knowledge,’ he said.
And Isabel Healthcare still has a charitable focus to its business to help achieve these goals. ‘Our business is to sell the tool in developed countries and we also want to make it available to developing countries,’ said Maude. ‘We get people signing up for trials from all sorts of countries.’
The latest development is a patient version of the tool, which launched towards the end of 2012. The patient version is like the doctor version but, instead of linking to resources like Dynamed and Best Practices, it links to publicly-available resources such as Wikipedia.
‘There is tonnes of stuff out there but the question is how to make use of it,’ he explained. ‘There is the idea at the moment that patients should be informed but giving them lots of knowledge is only giving them half of what they need. Isabel is that missing link.’ The aim of the patient version, he said, is to equip patients to have more informed discussions with their doctors. He likened the potential of this tool to the cultural change that came about when the Bible was translated into English and became accessible to everybody, rather than just to those who could read Latin.
‘There are symptom checkers out there but they are very crude. We are putting in quite a lot of education and tips for patients,’ he continued. The patient tool will be made free to patients anywhere in the world – although there are controls on use to prevent doctors using the patient tool many times a day. The company also hopes to sell it to organisations such as pharmacies as a patient engagement tool. It could provide links straight to doctor finders, helping people to find doctors that specialise in their symptoms. Patients could also search by products that relate to their symptoms.
So what does the real Isabel think of the tool that bears her name?
According to her dad: ‘I think Isabel is very proud, especially when she hears stories about how the tool has changed someone’s life.’