INTRODUCTION

Deep vein thrombosis (DVT) often shows as edema and limb pain leading to pulmonary thromboembolism (PTE). Additionally, DVT has significant associated morbidity and mortality rates.

Its annual incidence is 50 to 100 cases per 100 000 inhabitants (¹). DVT is more common among women aged 20 to 40 years, though in the 54 to 60 age group, it is more common in men (¹).

Its pathophysiology is explained by Virchow's triad: hypercoagulability, venous stasis, and endothelial damage. DVT main risk factors include cancer, acute illness, surgery, trauma, immobility, obesity, infectious or inflammatory diseases, hormonal therapy, pregnancy, and thrombophilia (¹).

OBJECTIVES

This study endpoint is to compare two diagnostic strategies in terms of costs and effectiveness:

Strategy #1: perform D-dimer analysis on all patients with suspected DVT and conduct a Doppler echocardiography on patients testing positive.

Strategy #2: use Wells' criteria on all patients with suspected DVT and perform D-dimer analysis on those with low probability, and a Doppler echocardiography on those with moderate or high probability, and on those with a positive D-dimer test.

The study secondary endpoint is to analyze spending at our center regarding the diagnosis of DVT, and identify factors associated with its presentation.

MATERIALS AND METHODS

This was a prospective study of patients with a 6-month history of suspected DVT in the lower extremities from May through October 2019 treated in our unit. The following variables were analyzed: bed rest, paralysis or paresis, recent surgery, active cancer, thrombophilia, history of DVT/PTE, pregnancy, clinical signs of edema, slurring, increased collateral circulation, alternative diagnosis, results of Wells' criteria, presence of associated PTE, D-dimer results, venous Doppler echocardiography results, and generated costs.

Costs were calculated based on spending and categorized into emergency care, D-dimer test reagent, and venous Doppler echocardiography, obtained from the of the community Official Bulletin (BOCYL) and the hospital billing unit.

The statistical analysis was performed using the SPSS program. The chi-square test was used to compare two dichotomous qualitative variables, and Fisher's exact test was used to analyze a dichotomous qualitative variable against another quantitative variable. Sensitivity, specificity, positive and negative predictive values were calculated to analyze the effectiveness of the two strategies. The costs incurred per patient and the costs that would have been generated with the two proposed strategies were analyzed too.

RESULTS

A total of 249 patients were included in the study, with a mean age of 70 years (ranging from 21 to 95), 46.59 % of whom (116) had DVT.

Compared with those without DVT, patients with DVT showed a statistically significant presence of the following factors: male gender, cancer, pain, edema, slurring, PTE, obesity, and fewer alternative diagnoses (Table I).

Table I. Characteristics of the patients (overall suspected DVT) with DVT vs those without it. 

The overall costs generated by the diagnosis or exclusion of DVT amounted to €47,934.63 (Table II), which resulted in €192.5 per patient among the 249 individuals from the sample.

Table II. Cost derived from diagnosing/ruling out DVT in our center across a 6-month period. 

After applying Wells' criteria, 116 patients (46.58 %) had low chances of developing DVT, while 133 (53.41 %) had moderate-to-high chances. Among the former with low clinical pretest probabilities of DVT, only 11 (9.4 %) actually had DVT compared with 105 (78.94 %) of those with moderate-to-high pretest probabilities.

Regarding the primary endpoint, only patients who had undergone D-dimer testing (n = 144) were included to guarantee a homogeneous sample in both strategies (Fig. 1). The costs for each strategy were:

Figure 1. Distribution of both strategies to assess their corresponding costs. 

In strategy #1:

144 × €101.41 (emergency care for suspected DVT) + 144 × €3.85 (D-dimer for all suspicions) + 138 × €88.86 (Doppler echocardiography only for positive D-dimer suspicions) = €14,603.04 + €554.4 + €12,262.68 = €27,420.12 → €190.41 per patient.

In strategy #2:

144 × €101.41 (emergency care for suspected DVT) + 71 × €88.86 (Doppler echocardiography only for moderate-to-high chances) + 73 × €3.85 (D-dimer for low probability) + 67 × €88.86 (Doppler echocardiography for low probabilities with positive D-dimer testing) = €14,603.04 + €6,309.06 + €281.05 + €5,953.62 = €27,146.77 → €188.51 per patient.

Regarding the effectiveness study of both strategies, strategy #1 (Table III), based on the direct use of D-dimer, achieved sensitivity rates of 100 % (all 60 patients with DVT had a positive D-dimer tests), specificity rates of 7.1 % (6 out of the 84 patients without DVT, had a negative D-dimer test), positive predictive values of 43 % (60 out of the 138 patients with positive D-dimer tests had DVT), and negative predictive values of 100 % (none of the 6 patients with negative D-dimer tests had DVT).

Table III. Contingency table of strategy #1. 

Strategy #2 (Table IV), based on pretest clinical probabilities (Wells' criteria), achieved sensitivity rates of 88.3 % (53 out of 60 patients with DVT had moderate-to-high chances), specificity rates of 78.57 % (66 out of 84 patients without DVT had a low probabilities), positive predictive values of 74 % (53 out of 71 patients with moderate-to-high probability had DVT), and negative predictive values of 90 % (66 out of 73 patients with low probability did not have DVT).

Table IV. Contingency table of strategy #2. 

DISCUSSION

The total annual cost of thromboembolic disease is estimated to be 1.5 to 12.2 trillion euros in Europe and 7 to 10 trillion dollars in the United States (¹-³). This represents a significant economic burden for all health care systems.

The use of diagnostic algorithms, specifically the combination of clinical probabilities and D-dimer testing before performing Doppler echocardiography, has proven cost-effective (⁴).

In our country, Fuentes et al. (⁵) conducted a cost-effectiveness study of the diagnostic process regarding DVT, with characteristics that are less comparable to ours, such as having been conducted in the primary care instead of the hospital setting, different exclusion criteria, and varying costs associated with diagnostic tests and emergency care, since the study was conducted in two different health care systems. Nonetheless, they also concluded that the use of diagnostic algorithms regarding DVT is cost-effective.

The sensitivity of D-dimer testing is approximately 95 %, with negative predictive values of 99 % to 100 %, and specificity rates of 35 % to 55 % (⁶). In our sample, we obtained a sensitivity rate and negative predictive value of 100 %. However, the specificity rate < 7.1 % was probably due to the characteristics of the sample and its inclusion criteria, with a low number of patients and negative D-dimer results treated by our unit. Although the specificity rate is the result of initial screening in other units, if the algorithm recommended had been followed, it would have been cero. Specificity is expected to be higher in the overall population.

The sensitivity of Doppler echocardiography has been reported to be 97 % for proximal DVT and 73 % for calf veins (⁴,⁷). In our sample, we assumed a sensitivity rate of 100 % because a positive result on the Doppler echocardiography was considered indicative of the presence of DVT.

The presence of false negatives in the Wells' criteria (for low probabilities) has been described to be around 5 % (⁸). In our sample, it was slightly higher, and 9.4 % of the patients with low probabilities developed DVT.

Table V illustrates a comparison between the two strategies analyzed:

Table V. Comparison between effectiveness and cost between the two strategies. 

The specific weight of strategy #1, identified by the D-dimer result, is its high sensitivity and negative predictive value. However, its low specificity and positive predictive value mean that its indiscriminate use leads to more diagnostic tests (Doppler echocardiography) being performed on patients without DVT, resulting in more health care spending.

Strategy #2 uses pretest clinical probability (Wells' criteria) initially, which marks the result of the D-dimer testing, is consistent with the diagnostic algorithm recommended by the current clinical practice guidelines (¹,⁹). This explains why it showed better results in our study, with higher effectiveness and lower costs.

Strategy #2 turned out to be 2 % more cost-effective than the spending generated at our center. Even the erroneous strategy #1 also turned out to be more cost-effective (1 %), maybe because analysis and Doppler echocardiography testing of 6 patients with negative D-dimer results were requested.

Despite the existence of these algorithms, the routine clinical practice does not always abide to them, leading to additional costs and burdens on the health care system, primarily due to unnecessary diagnostic tests being performed, due to either a certain lack of awareness of such algorithms or the use of defensive medicine. In our sample, we could have achieved a 2 % reduction in health care expenses per patient.

CONCLUSIONS

The use of diagnostic algorithms (in the case of DVT, the initial clinical probability stratification [Wells' criteria] combined with D-dimer analysis in cases of low probability) is a cost-effective strategy to select patients eligible for Doppler echocardiography.

Greater awareness is needed on the importance of abiding by the diagnostic process of DVT among the various health professionals involved, among whom vascular surgeons are often only the final step.