Doctors need reliable ways to spot patients at risk for blood clots after surgery. Lung cancer patients face a higher danger of these clots, known as venous thromboembolism. A recent look at twenty different prediction models tried to measure how well these tools work. The combined results showed an average accuracy score of 0.85. However, the range of scores was huge, going from 0.66 to 0.99. This means some models work well while others do not.
The review found that most of the studies used were done in single centers and looked back at past data. These design choices create a high risk of bias. Because of this, the tools cannot be trusted to guide care for every patient. The differences between the models were so large that they cannot be used together in a standard way.
Safety issues were not reported in the review. The main problem is that the evidence is too weak to support using these models in daily practice. Doctors should not rely on existing tools for lung cancer patients undergoing surgery until better data is available.