Research Profile - Cost-effective and compassionate

Researchers use computer simulation to propose a way to spare pregnant women from needless worry and unnecessary, costly procedures.

[ Back to main article ]

Mathematical modelling and incremental cost-effectiveness ratios might seem like subjects that could only warm the heart of a hard-core accountant.

However, to Dr. François Rousseau of Laval University, they are essential tools in helping doctors practise evidence-based personalized medicine.

“Physicians need to integrate a lot of information,” says Dr. Rousseau, who has an MD’s heart and a computer programmer’s brain. “It’s the art of medicine to a degree – using all the information available to make sure that the diagnosis and treatment fits the patient.”

The need to make a definitive diagnosis can be of life-altering importance, especially in prenatal care. At the present time, there are several different strategies for screening for disorders such as Down syndrome. Each strategy involves a different protocol for collecting “biomarkers” – biological indicators that can be measured to detect or monitor changes in a person's health. These biomarkers can be collected through such interventions as blood tests and ultrasounds.

Depending on the strategy used, “there is a huge difference in the number of cases that you can identify,” says Dr. Rousseau. “You want to minimize the number of false-positive or false-negative results and spare a lot of women a lot of worry.”

Beyond creating anxiety, an early false-positive result can lead to unnecessary and costly medical procedures such as chorionic villus sampling (removal of a piece of placenta to test for genetic defects) or an amniocentesis (drawing some fluid from the amniotic sac surrounding a fetus to check for abnormalities).

“Each time you do surgical intervention for a normal fetus in a false-positive case, there is a risk of losing that fetus,” says Dr. Rousseau.

To address the problem, Dr. Rousseau worked with a multidisciplinary team that included health economist Dr. Daniel Reinharz, geneticist Dr. Jean Gekas, and computer engineer Dr. Christian Gagné to create a computer simulation of 110,948 pregnancies (the exact number of pregnancies documented in Quebec in 2001). They analyzed outcomes from 19 different options possible through four basic screening approaches.

The results, first published in the prestigious British Medical Journal in 2009, showed that “contingent screening,” in which first trimester screening results are used to categorize women as high, intermediate or low risk, was the clear winner in terms of accuracy, safety and cost.

Dr. Rousseau’s goal is to make computer simulation models readily available to policy makers and health care managers so that they can make their decisions based on quantitative results and not anecdotal evidence or opinions.

In 2010, he and his Laval colleagues unveiled a computer simulator called SCHNAPS (SynCHroNous Agent- and Population-based Simulator) to analyze various options for screening, diagnosing, monitoring and treating diseases. SCHNAPS can create a virtual population to replicate the characteristics of any group of people that investigators, health care planners or policy makers want to study.

The team at Laval can tap into the awesome power of a supercomputer operated by CLUMEQ, a research consortium for high performance computing that includes McGill University and University of Quebec institutions.

“We decided to use the supercomputer three or four years ago because we test hundreds of different scenarios and we need to see quickly whether the results are robust,” says the CIHR-funded Dr. Rousseau, whose research also includes examining screening scenarios for osteoporosis, cystic fibrosis as well as some pharmacogenetic applications. “It allows us to run and rerun the simulations many times, changing the parameters to see whether it influences the results.”

"We can program in different parameters such as introducing a screening test that will pick up cases of disease earlier, with patients offered a specific treatment. So many people will accept this treatment, so many will accept other alternative treatments and still others will say, 'I don't want to get treated.' We will put all that into the simulator and see what the different patient trajectories will be."
– Dr. François Rousseau, Laval University