“Professor Predict” and his algorithm for calculating your heart attack risk and treatmentby Donna Chisholm
Rod Jackson is professor of epidemiology at the University of Auckland, but in some circles, he’s simply “Professor Predict.”
“We know what causes it, how to prevent it and how to treat it. If you’re going to get any disease, a heart attack is the one to have if you survive it, because the treatment is amazingly good.”
Jackson is professor of epidemiology at the University of Auckland, but in some circles, he’s simply “Professor Predict.” His research team’s data from more than 500,000 patients treated by GPs in primary care is about to be incorporated in new treatment guidelines for cardiovascular disease. Currently, the algorithms doctors use to guide patients as to what treatment to have and when to start it are based on heart-disease risk scores from patients in the long-running Framingham study in the US. But Jackson has found that those scores overestimate risk by nearly 100% when compared with the more recent data from the New Zealand cohort. That’s largely because coronary death rates have been plummeting since the late 1960s.
The existing guidelines say doctors should discuss drug treatment with non-symptomatic patients when their risk of a heart attack or stroke is 10% or higher over the next five years. The threshold was reduced in 2014 from 15%, when statins became less expensive, but other countries, including the US and UK, now have lower thresholds. The UK’s is down to 10% over 10 years, and the US’s is 7.5% over 10 years – the equivalent of 5% and 3.75% over five years. US guidelines also say statins should be considered for patients with a 10-year risk of between 5% and 7.5%.
Jackson, who’s one of those advising the Ministry of Health on the new guidelines, believes patients at a 5% five-year risk level should be told that the benefits of drug treatment – say a blood-pressure-lowering medication or a statin – would outweigh the risks, but the benefits would be quite small until the risk reaches about 10%. “Most patients will prefer not to take pills at a 5% risk, but others who are very risk-averse may choose to.
“What we can do, which the rest of the world hasn’t done, is rather than making 5% five-year risk a drug-treatment threshold, we should tell patients it’s the point of equipoise when the benefits of drugs begin to outweigh the risks and use the opportunity to initiate a discussion about the multiple ways patients can also reduce their risk without drugs.”
With New Zealand comparatively under-treating in relation to the UK and US, the fact our current prediction algorithm overestimates risk means patients are probably getting similar treatment to those in the UK and US anyway.
The approach to risk assessment based on Jackson’s Predict model is a far cry from what was happening when he started his research in 1982. “We used to predict risk by just measuring blood pressure. When statins came along, we independently measured lipids as a way of predicting risk. But the combination of factors is a much better predictor, and that’s so important because the benefit of treatment is directly proportionate to risk.”
Jackson says in the 1980s, he realised definitions of “hypertension” were extremely fluid over time. “I decided the medical profession had no idea what they were talking about. It was a bit of a shock to the system. When I went to medical school, my clinical teachers told me they had been taught hypertension was a reading of greater than 100 diastolic. When I was there, we were talking 95. When I qualified, it was down to 90 and now, if you ask students, they’d say 80.”
Jackson says his recognition of the knowledge gaps and the implications that had for treatment launched his career in public health. He developed his first relatively crude multi-risk algorithm chart for treatment when he was asked to produce guidelines for the Ministry of Health in 1992. In 1993, it was published in the prestigious British Medical Journal.
“It was revolutionary, because it says there are some people with a blood pressure of 160 systolic who you won’t give drugs to today, but there are people with blood pressures of 130 systolic that you will. That’s what was counter-intuitive to a lot of people. We’d always been taught that there’s hypertension and there’s not hypertension – you treat people above a certain blood pressure and leave them if it’s below.”
In practice, that meant a woman with a blood pressure of 160 systolic at the age of 50 and no other risk factors had a 2% chance of having some kind of heart event in five years. “If I lower her blood pressure, I could lower that risk from 2% to 1.5%, but that’s a pretty small benefit. She might be overweight, drinking too much and probably not very fit, so initially I’d say to her, ‘We need to lower your blood pressure, but we don’t need to start drugs today; we can try other things. It’s quite likely at some stage you’ll need drugs, but why don’t we try something else first?’”
He says half the doctors at the time “got it immediately and the other half thought it was sacrilegious – particularly hypertension specialists”.
The early charts largely weighted all risk factors equally, but now information about patient outcomes allows them to be rated according to their importance. For example, a man who smokes increases his risk by about 75%; a woman who smokes doubles her risk. The computerised Predict charts, which are used by about a third of the country’s GPs, have produced a gold mine of data that’s unique in the world, says Jackson.
Ministry of Health national programme manager Karen Evison told the Listener there’s little that’s contentious about Jackson’s work and an associated Heart Foundation evidence update, so the main part of the new guidelines still to be decided is the percentage risk at which doctors should start talking with patients about treatment and what that conversation should cover. Some feedback suggested the start point should be 6, 7 or 8%, rather than 5%.
Evison says guidelines are sometimes wrongly interpreted as an instruction to start treatment, rather than advice to have a conversation about treatment options, which may not include drugs for lower-risk people.
This article was first published in the September 17, 2016 issue of the New Zealand Listener.
Ian McEwan’s tale of human-robot love links emotional and artificial intelligence in intriguing ways, writes Charlotte Grimshaw.Read more
The chemical residues on fruit and vegetables are not dangerous, but rinsing is still advisable.Read more
A three-month trial at Christchurch Hospital saw remarkable results.Read more
Until recently, the Auckland War Memorial Museum’s buildings were highly dysfunctional, says John Glen, the museum’s head of building infrastructure.Read more
More than 230 tonnes of plastic including straws, bags and toothbrushes found on Australian islands.Read more
Violent extremists are often depicted as “lone wolves”. But this belies the broader psychological, social and digital contexts in which they act.Read more
Seeing an NZ flag flying at a neo-fascist rally in Germany prompted David Hall to ask why violent radicalisation was affecting even his fellow Kiwis.Read more