Clinical trial data quot;can predict individual treatment effects quot;
MedWire News: Data from randomized clinical trials can be used to predict treatment effects in individual patients, say researchers writing in the British Medical Journal.
They show that predictions may be based on existing validated risk scores, where available, or on models newly created to fit the trial data.
However they warn: "The value of such prediction of treatment effect for medical decision-making is conditional on the number willing to treat (NWT) to prevent one outcome event."
Frank Visseren (University Medical Center, Utrecht, The Netherlands) and colleagues investigated the feasibility of predicting treatment effects for individual patients based on data from trials.
As an example, they used data from the Justification for the Use of Statins in Prevention (JUPITER) trial, a randomized controlled study comparing rosuvastatin 20 mg daily versus placebo on the rate of cardiovascular events in 17,710 healthy individuals.
For each participant, Visseren's team calculated risk scores using three methods: the established Framingham and Reynolds risk scores, and a newly developed "optimal fit" model, based only on trial data. They then compared the benefit of rosuvastatin therapy in patients whose predicted treatment effect exceeded a decision threshold, versus the benefit of rosuvastatin therapy in either everyone or no-one.
The authors explain: "A theoretical advantage of [the optimal fit model] over using existing risk scores is that the model may be better calibrated to the population of interest. Furthermore, such a model is not based on the assumption that treatment effect increases linearly with baseline risk: modification of the treatment effect by patient characteristics can be tested and, if significant, included in the model."
The 10-year absolute risk reduction for cardiovascular events was 4.4% using the Framingham score, 4.2% using the Reynolds score, and 3.9% using the optimal fit model, report Visseren et al.
Importantly, prediction-based treatment was associated with a greater net benefit than treating either everyone or no-one. This was only the case if the decision threshold was set between 2% and 7%, however, equating to a NWT of between 15 and 50.
"Implementing trial results by treating all or no patients, expecting the treatment effect for everyone to be similar to the average treatment effect in the original trial, does not lead to optimal benefit," the researchers remark.
Noting that individualizing the prediction of treatment effects in clinical practice "is not necessarily" complicated, they conclude: "Applying this strategy in clinical practice leads to more selective treatment of patients who will benefit from treatment."
By Joanna Lyford