DASH Score for predicting recurrent VTE

Clinical background

After a first unprovoked venous thromboembolism (VTE), the risk of recurrence is high, approaching 30% within five years of stopping treatment. At the same time, extended anticoagulation carries a bleeding risk that increases over time. The decision whether to continue or stop anticoagulation after the initial treatment period therefore requires an individual trade-off between recurrence risk and bleeding risk. The DASH score was constructed to quantify the risk of recurrence and thereby identify patients in whom the risk is low enough for stopping treatment to be considered. The instrument rests on variables readily available in clinical practice and requires no advanced imaging or genetic testing, but it does require the D-dimer to be measured after anticoagulation has been stopped.

Calculating DASH

DASH (D-dimer, Age, Sex, Hormonal therapy) is calculated as:

$$\text{DASH} = 2 \times D + A + M - 2 \times H$$

where:

• $D = 1$ if the D-dimer is abnormal after anticoagulation has been stopped, otherwise $0$
• $A = 1$ if age is $\le 50$ years, otherwise $0$
• $M = 1$ if male, otherwise $0$
• $H = 1$ if a woman on hormonal therapy at the time of the VTE, otherwise $0$

The score ranges from $-2$ to $4$. The only variable requiring a test is the D-dimer, which must be measured after anticoagulation has been stopped, usually after at least three months of treatment.

The derivation cohort consisted of 1,818 patients with a first unprovoked VTE treated with a vitamin K antagonist for at least three months, pooled from seven prospective studies [1]. Optimism-corrected Cox regression coefficients were used to identify the four predictors, and the model was internally validated using bootstrap methods. The modelled outcome was symptomatic recurrent VTE after treatment was stopped.

Interpretation in practice

The central interpretation is that a score of 1 or less identifies a patient whose risk of recurrence is low enough for anticoagulation to be stopped. In the derivation cohort this corresponded to an annual recurrence risk of 3.1% (95% CI 2.3 to 3.9), and about half of all patients with unprovoked VTE fell into this group [1].

These annual recurrence risks apply to patients treated with a vitamin K antagonist. In patients treated with a direct oral anticoagulant (DOAC), the D-dimer can be measured after a shorter washout period, but validation of DASH in this population is more limited [3].

Validation and performance

An external validation of DASH was carried out in a retrospective cohort of 827 patients with proximal deep vein thrombosis or pulmonary embolism, of whom 100 (12.1%) had a documented recurrence [3]. The proportion of patients classified as low risk (DASH $\le 1$) was higher than in the derivation cohort (66.3% versus 51.6%), suggesting that the model may tend to classify a larger proportion of patients as low risk in clinical practice than was the case in the original cohort. The cumulative recurrence risk at one year for DASH $\le 1$ was 3.6%, in line with the model's prediction. Calibration, measured as the slope between observed and predicted cumulative incidence at two years, was 0.71 (95% CI 0.51 to 1.45), indicating acceptable agreement.

Discrimination was, however, age-dependent. The c-statistic was 0.72 for patients under 65 years but fell to 0.54 for patients over 65 years, that is, hardly better than chance [3]. In older patients, the recurrence risk exceeded 5% per year even in the lowest risk group, meaning that DASH cannot reliably identify low-risk patients in this age group.

A post-hoc analysis of the PADIS-PE study, which included 371 patients with unprovoked pulmonary embolism only, showed a c-statistic of 0.60 (95% CI 0.53 to 0.66) for DASH [4]. In this population it was only the HERDOO2 score, not DASH, that identified patients with a recurrence risk below 3% per year. DASH therefore performed less well in a cohort consisting solely of patients with pulmonary embolism than in the original mixed VTE cohort.

A systematic review of prognostic models for recurrent VTE identified three established instruments: DASH, the Vienna prediction model and HERDOO2 [2]. The review judged that DASH and Vienna had been developed with strong methodology, but that all three models carried at least a moderate risk of bias, mainly because of insufficient external validation at the time of the review. The predictors consistently important across the models were sex, the site of the index event and the D-dimer.

Limitations

DASH applies only to patients with a first unprovoked VTE. It should not be used after provoked VTE, where the recurrence risk is lower and governed by the provoking factor, nor in cancer-associated thrombosis, where different risk models and different treatment recommendations apply.

The model was developed in patients treated with a vitamin K antagonist. The D-dimer is measured after anticoagulation has been stopped, and it is uncertain how well DASH performs in patients treated with a DOAC, in whom the D-dimer may fall more quickly after stopping.

The most important limitation is its age dependence. In patients over 65 years, DASH in practice loses its discriminatory ability, and the recurrence risk is high enough even at the lowest score that stopping treatment cannot be recommended on the basis of the instrument [3]. This is particularly problematic since age is itself a risk factor for recurrence, but DASH assigns points only for young age ($\le 50$ years) and has no component penalising old age.

A further practical problem is that DASH requires anticoagulation to be stopped before the D-dimer is measured. This means that the decision to stop has in practice already been made by the time the instrument can be applied in full, which makes DASH more a verification tool than a decision aid in the strict sense. The model therefore cannot replace a clinical assessment weighing patient preference, bleeding risk and comorbidity.

References

1. Tosetto A, Iorio A, Marcucci M et al. Predicting disease recurrence in patients with previous unprovoked venous thromboembolism: a proposed prediction score (DASH). J Thromb Haemost. 2012;10(6):1019-25. PMID: 22489957
2. Ensor J, Riley RD, Moore D et al. Systematic review of prognostic models for recurrent venous thromboembolism (VTE) post-treatment of first unprovoked VTE. BMJ Open. 2016;6(5):e011190. PMID: 27154483
3. Tosetto A, Testa S, Martinelli I et al. External validation of the DASH prediction rule: a retrospective cohort study. J Thromb Haemost. 2017;15(10):1963-1970. PMID: 28762665
4. Raj L, Presles E, Le Mao R et al. Evaluation of venous thromboembolism recurrence scores in an unprovoked pulmonary embolism population: a post-hoc analysis of the PADIS-PE trial. Am J Med. 2020;133(8):e406-e421. PMID: 32333853