Table 1.
General characteristics of the included studies in the systematic review of PTS prediction models.
| Study | Study designs | Population of Development (Sample size) | Study period | Predictors | Outcome | Internal validation | External validation, population (Sample size) |
|---|---|---|---|---|---|---|---|
| Tao Yu, 2022 [12] | Prospective Cohort | America (ATTRACT database) (550) | December 2009–December 2014 |
Extreme gradient boosting (XGBoost)： Diabetes mellitus, Baseline villalta Score, BMI, Previous VTE, High cholesterol, Weight, Treatment type. |
Developed and external validated four prediction model for PTS risk by machine learning. | 10 fold cross-validation | External validation, Chinese cohort (117). |
|
Logistic regression: Baseline villalta Score, Diabetes mellitus, BMI, Previous VTE, COPD, Treatment Type, High Cholesterol. | |||||||
|
Random forest: Weight，Baseline villalta Score, BMI, Diabetes mellitus, Inpatient qualify DVT, DVT leg, treatment type | |||||||
|
Gradient boosting decision tree (GBDT): Baseline villalta Score, Previous VTE, Diabetes mellitus, BMI, Weight, High Cholesterol, Treatment Type, | |||||||
| Lijun Zhu, 2022 [13] | Retrospective Cohort | China (518) | December 2018–December 2019 | Proximal DVT, Recurrent DVT, Age, Male sex, History of varicose veins. | Developed a prediction model for PTS after DVT. | 5 fold cross -validation. | None |
| Hao Huang, 2018 [14] | Retrospective Cohort | China (209) | January 2013–December 2014 | Iliac Vein Compression Syndrome, Occlusion, Residual Iliac-femoral vein thrombosis, Residual Femoral-Popliteal vein thrombosis, Insufficient Anticoagulation. | Developed of APTSD score prediction model for PTS risk in DVT patients. | Not reported | Temporal validation, Chinese cohort (102). |
| Jiantao Zhang, 2022 [15] | Prospective cohort | China (540) | June 2014–December 2016 | Ilio-femoral DVT, Active cancer, History of chronic venous insufficiency, Previous venous thromboembolism, Chronic kidney disease, Duration of compression therapy <6 months. | Developed a prediction model for PTS risk in DVT patients | Bootstrap | Temporal validation, Chinese cohort (268). |
| Peng Qiu, 2021 [16] | Retrospective, case-control study | China (210) | June 2016–June 2018 | The number of signs and symptoms, Male sex, Varicose vein history, BMI, Chronic DVT. | Developed a prediction model for PTS risk in DVT patients. Externally validated the SOX-PTS predictive model, and the SWITCO-PTS predictive model in their set. | Not reported. | Temporal validation, Chinese cohort (90). |
| Anat Rabinovich，2020 [17] | Prospective Cohort | America (ATTRACT database) (691) | December 2009–December 2014 | More extensive, BMI≥35, Baseline villalta score, Age. | Externally validated the SOX-PTS score for estimating the risk of developing PTS, moderate to severe PTS, and severe PTS, in patients with proximal DVT. | Bootstrap | This was an external validation study with model updates and the addition of an age variable. |
| Anat Rabinovich，2018 [18] | Prospective Cohort | Canada/America (SOX Trial database) (762) | June 2004–February 2010 | Iliac DVT, BMI≥35，Baseline villalta score. | Developed a prediction model for PTS after DVT. | Bootstrap | None |
| Marie Méan, 2018 [19] | Prospective Cohort | Switzerland (SWITCO65+ database) (276) | September 2009–December 2013 | Age>75 y, Concomitant antiplatelet/NSAID therapy, Multi-level thrombosis, Prior varicose vein surgery, Number of leg signs and symptoms. | Developed of prediction model for PTS risk in >65 y DVT patients. Externally validated the SOX-PTS predictive model in their set. | Bootstrap | None |
| Elham E. Amin, 2018 [20] | Prospective Cohort | Netherlands (451) | June 2003–June 2013 |
Baseline model: Age>56, BMI>30, Varicose veins, Smoking, Female sex, Iliofemoral thrombosis, History of DVT. Secondary model: Age>56, BMI>30, Varicose vein, Smoking, Residual vein obstruction. |
Developed a two-step model consisting of a model to be applied at baseline to predict the probability of developing PTS at 6 months, and a model to be applied at 6 months to predict the probability of PTS 24 months after initial thrombosis for those patients who did not develop PTS till then. | Bootstrap | External validation, Italy cohort (1107). |
| Tian'an Huang， 2022 [21] |
Retrospective Cohort | China (204) | June 2016–June 2018 | BMI＞24, Duration of disease＞14 days, History of varicose veins, Iliac DVT, Thrombus removal of level Ⅲ. | Developed a prediction model for PTS after DVT. | Bootstrap | None |