As a heightened risk of DVT begins immediately after hip fracture, blood clot formation can happen at any time, especially in elderly patients. Previous studies showed that the incidence of preoperative DVT in HF patients was between 6 and 62%, but no study assessed the admission rate of DVT in elderly Chinese HF patients [5, 18, 19]. Previous studies also ignored the fact that many elderly HF patients had already had venous thrombosis at admission. In our study, we revealed a high incidence of DVT in elderly Chinese HF patients at admission, which indicates that many elderly Chinese HF patients on admission already have DVT. Furthermore, the DVT could be divided into occult DVT and symptomatic DVT. The high prevalence of DVT on admission might be related to occult DVT which may have never caused a problem. However, few studies have focused on the long-term effects of occult DVT in elderly Chinese HF patients. In addition, unearthing occult DVT and early intervention might reduce the incidence of symptomatic DVT. In our study, both occult or symptomatic DVT were regarded as one type of DVT. In addition, previous history of venous thromboembolism and anticoagulant treatment were both exclusion criteria in our study included, so the time of thrombosis formation was from injury to admission and the DVT patients found in our study all had acute DVT. Currently, most doctors would treat elderly HF patients with some drugs or physical methods to prevent the occurrence of DVT. However, if elder HF patients was diagnosed with DVT at admission, doctors would have to start these patient with DVT treatment immediately rather than giving DVT prophylaxis. Thus, in elderly HF patients, early DVT identification is essential to limit late complications of DVT and prevent clot extension, acute PE, and recurrent thrombosis [20].
In our study, most of patients were diagnosed with distal DVT, which was consistent with a previous study [5]. However, the clinical relevance of distal DVT in elderly HF patients is disputed. We believe that patients with distal DVT also need active treatment, as there is no long-term follow-up study on the effect of distal DVT on the rehabilitation of HF patients, especially elderly Chinese patients.
In our study, age was an independent risk factor for DVT in elderly HF patients. This may have be caused either by an acquired prothrombotic state, the anatomical variations of lower veins and the thickening of venous valve cusps [21]. Some studies demonstrated that women have a higher risk of venous thrombosis than men, while others hold opposite view [21,22,23]. Our study identified being female as a risk factor for DVT in elderly HF patients, which is supported by a previous study [24]. Another study found that obesity was associated with DVT [25], however, our study, found no association between BMI and DVT in elderly HF patients. D-dimer and fibrinogen are indicators of coagulation function. In our study, high level of D-dimer and fibrinogen were associated with DVT in elderly HF patients. D-dimer is used to diagnose venous thromboembolism, including deep vein thrombosis and pulmonary embolism. However, there is controversy in determining the cut-off value of D-dimer because of its low diagnostic accuracy, especially in elderly patients [11, 26]. Prolonged immobilization and delays in fracture fixation were independent risk factors for DVT noted in previous studies [5, 24, 27]. The preoperative period of immobilization could be divided into the time from injury to admission and the period awaiting surgery [19, 28]. In our study, all elderly HF patients underwent venography or ultrasonography on admission. Five patients were diagnosed with DVT in the period awaiting surgery. This suggests that only a small proportion of DVT happens between admission and surgery. In our opinion, pharmacological or physical thromboprophylaxis greatly reduced the occurrence of DVT in the period awaiting surgery. In addition, our study showed that the time from injury to admission was associated with increased risk of DVT. Thus, clinicians should pay attention to not only the time awaiting surgery but also the time from injury to admission. In our view, improving the awareness of elderly HF patients to the need for hospitalization and shortening the time to hospitalization -transferring would shorten the time from injury to admission and decrease the risk of DVT. Taking all these risk factors into account, clinicians should know an individual’s risk of DVT and make a personalized treatment programme to prevent the incidence of pulmonary embolism. In our study, DVT was diagnosed by venography or colour Doppler ultrasonography. The diagnostic accuracy of colour Doppler ultrasonography might be lower than venography. However, in our study, an experienced radiologist could effectively improve the diagnostic accuracy of colour Doppler ultrasonography. Furthermore, superior examination could also decrease the misdiagnosis rate.
Currently, DVT diagnosis is mostly performed using venography, colour Doppler ultrasonography, and plasma markers. Venography was a gold standard to diagnose DVT, but was limited by its invasiveness and non-repeatability in clinical work [12]. Ultrasonography requires specialized equipment and experienced radiologists on 24-h duty, which is not available in some primary hospitals. The results of the ultrasonography can also be affected by the proficiency of the radiologist. Moreover, elderly HF patients would suffer from the pain caused by changing the positions of lower extremities. In addition, the waiting time for venography or ultrasonography would result in delays to surgery and increase the risk of perioperative complication [18, 29]. Some patients at low risk for DVT could be treated with early operation as soon as possible instead of waiting for venography or ultrasonography. A previous study examined the usefulness of erythema, lower limb swelling, tenderness, Homan’s sign, and calf diameter in predicting DVT, but did not find that any of these factors performed well in diagnosing DVT [30]. D-dimer is used as a diagnostic tool for DVT, however, the levels of D-dimer are influenced by many factors, such as inflammation, age, surgery, hospitalization, COPD, and other acute diseases [31,32,33]. In addition, there is controversy about the cut-off value of D-dimer for the diagnosis of DVT in elderly patients, especially in elderly trauma patients [15, 34]. Many recent studies have attempted to use age-adjusted cut-offs for D-dimer in the diagnosis of DVT in elderly outpatients to lower the incidence of pulmonary embolism [35, 36]. However, few studies focused on adjusted cut-off values for D-dimer in elderly trauma patients, especially in elderly HF patients. Receiver operating characteristic curves (ROC) are usually used to evaluate the diagnostic value of different methods: the larger the value of the area under curve (AUC), the better the diagnostic value. In our study, we calculated the AUC of all the independent risk factors, and found that D-dimer had the highest diagnostic value. However, the AUC of D-dimer was only 0.679. Thus, our study attempted to improve the DVT diagnostic system by using a new method, adjusted by independent risk factors including gender, age, time from injury to admission, D-dimer, and fibrinogen.
Previous studies mainly reported adjusted D-dimer cut-off values by age in elderly individuals. Han et.al reported age-adjusted D-dimer cut-off value by age × 0.01 μg/ml in elderly patients (> 50 years) [15]. In another study, they also found age-dependent reference intervals for D-dimer in elderly general population [26]. In the study of Schouten et al., they calculated adjusted D-dimer in elderly outpatients by the following formula:(0.1 × age (years) × D-dimer value (g/mL) [35]. Imai et al. calculated the D-dimer index of patients before total hip arthroplasty by 0.12 × age (years) + 0.45 × the D-dimer value (g/mL) [11]. Furthermore, the previous study also found that there was a markedly substantially increase of D-dimer and fibrinogen levels in fracture patients, especially in femoral fracture patients [37]. Using the results of multivariate logistic regression analysis, our study calculated the new DVT predictor from the following formula: the value of new predictor = 1.131× (female = 1, male = 0) + 0.071 × age (years) + 0.571 × time from injury to admission (days) + 1.028 × fibrinogen (g/L) + 0.123 × D-dimer (g/L). In our study, we reported a new calculated method to predict DVT (AUC: 0.852; 95% confidence interval: 0.806~0.898; p < 0.001). Figure 2 is a scatter dot plot of the new DVT predictor values of the DVT and non DVT groups. When we defined the cut-off value of the new predictor as 12.74 using Youden’s index, 116 of 248 patients had values under 12.74 with a sensitivity of 91.9% and a specificity of 66.7%. When the cut-off was 12.47, 95 of 248 patients had values under 12.47 with a sensitivity of 100%, a specificity of 54.6%, a negative predictive value (NPV) of 100% and a positive predictive value (PPV) of 48.4%. When its cut-off was 12.47, the new DVT predictor could be used as a diagnosis of exclusion, i.e., when the new DVT predictor value < 12.47, there was low risk of DVT, and surgery could be performed on elderly HF patients as soon as possible without the need for any further auxiliary examination. Our new method could improve the diagnostic value of D-dimer. The new DVT predictor had highest sensitivity among these predictor factors. For clinicians, with high sensitivity and high negative predictive value, new predictor could be conducted to exclude DVT diagnosis when the value of new DVT predictor was less than the cut-off value. New predictor estimated on admission could effectively identity the patients without DVT and ensure early operation without ultrasonography or venography. Furthermore, the new DVT predictor could be used to identify elderly Chinese HF patients with high-risk of DVT and could benefit from was helpful to take treatment of early anti-coagulation treatment and a shortened preoperative time. Follow-up studies by our group will attempt to use the new predictor in the diagnosis of DVT to shorten time from admission to surgery.
The present study had several limitations. First, our study was a retrospective analysis. Second, the sample size of enrolled patients was small, and all the patients came from same hospital. In addition, for some variables, small sample size may be underpowered to show significant differences. A multi-centre large sample study would be required to validate our findings. Third, most of the DVT patients were diagnosed using ultrasonography, which may have a lower accuracy than venography.