This research evaluated how well three specific prediction models could identify patients at risk for deep vein thrombosis (DVT) among those admitted with acute stroke. The team analyzed data from 1,270 patients treated at the Stroke Center of South China Hospital of Shenzhen University. They compared the Shen Xiaofang, Lu Qiufang, and Xi Pan models using standard statistical measures like accuracy and predictive values.
The analysis revealed that the Lu Qiufang model generally outperformed the others. It achieved an accuracy of 82.1% and correctly identified 84.9% of patients who did not have DVT. All three models showed a negative predictive value exceeding 90%, meaning they were very good at ruling out DVT when the test was negative. The overall incidence of DVT in this group was 17.08%.
Despite the promising numbers, experts caution that this was a single-center validation study. The results might not hold true in different healthcare settings or with different patient populations. Further multicenter studies are needed to confirm if these models work reliably across diverse regions before they can be widely adopted for routine clinical use.