Screening for subclinical deep-vein thrombosis

Most episodes of venous thromboembolism (VTE) are clinically silent; symptomatic events are merely the tip of the thromboembolism iceberg.1 The subgroup of events that evolve to develop clinical manifestations cannot accurately be predicted, and sudden death from pulmonary embolism (PE) may be the presenting feature. Consequently, various thromboprophylactic strategies are used to minimize the incidence of VTE in high‐risk cohorts (Table 1). While all of these reduce the frequency of deep‐vein thrombosis1 (DVT), a reduction in overall mortality has only been convincingly demonstrated with heparin.2 However, despite optimal prophylaxis, venographically‐demonstrated DVTs still supervene in an important minority.1,3–6 Moreover, not all high‐risk patients receive heparin, either because of contraindications, or because evidence of sustained benefit is lacking, as in acute ischaemic stroke.7 Given that the treatment of established VTE is highly effective,8 the concept of screening for subclinical DVT with initiation of treatment in selected cases, both in conjunction with optimal prophylaxis and in subgroups that do not receive heparin, also warrants consideration. However, there is much uncertainty about which, if any, subgroups should be screened, and whether this would have an important impact on clinical outcomes. In addition, it is unclear which screening tool should be used as all current modalities have shortcomings, in that they are either invasive, imperfectly sensitive or insufficiently evaluated. In this review, we discuss the clinical relevance of asymptomatic DVT, the potential benefits of early diagnosis and treatment, and the diagnostic tools available for this purpose, as well as highlighting important areas requiring further research. Prevalence of DVT in the absence of heparin prophylaxis in high‐risk patient groups2,9,10 aPost‐myocardial infarction or medical patients with additional risk factors. bMajor surgery >60; major surgery aged 40–60 with additional risk factors. Prevalence of DVT in the absence of heparin prophylaxis in high‐risk patient groups2,9,10 aPost‐myocardial infarction or medical patients with additional risk factors. bMajor surgery >60; major surgery aged 40–60 with additional risk factors. The prevalence of DVT in high‐risk cohorts such as general surgical, stroke and orthopaedic patients ranges from 19% to 84% (Table 1) in the absence of heparin prophylaxis.1,9,10 Only one‐third of these are proximal,1,2 in contrast to patients with symptomatic DVTs, in whom the majority are proximal.11 While most are asymptomatic12 and lyse or organize without causing symptoms,13 a minority will progress to clinically apparent disease. Isolated below‐knee DVT rarely produces substantial or fatal PE when cardiorespiratory reserve is adequate,14–16 but proximal propagation occurs in around 20% of cases17 and it is principally these which are of concern to the clinician. Studies investigating the natural history of untreated subclinical proximal DVTs in surgical and stroke patients in the anticoagulant era have been small, but have demonstrated a risk of clinical PE of some 40%.13,18 Pooled data from general surgical patients who acted as controls in trials of heparin thromboprophylaxis reveal prevalences of DVT (proximal and below knee), proximal DVT and fatal PE of 19%, 7% and 0.9%, respectively.19 If it is assumed that fatal PEs only arose from proximal DVT,20 these data suggest that the mortality associated with untreated asymptomatic proximal DVT might be some 12–13%. Although the precise risk in the current era is difficult to quantify, it is apparent that a significant minority of patients who develop subclinical proximal DVT will succumb to PE. Furthermore, as clinically‐evident PE is usually unheralded21 and is often lethal at presentation,22 fatal PE will continue to be an important initial presentation of subclinical VTE, even with optimal vigilance for subtle clinical manifestations. In addition, the morbidity of non‐fatal, clinically‐apparent VTE should not be overlooked, both in terms of potentially distressing symptoms and impaired or delayed rehabilitation. Asymptomatic DVT may also result in the post‐thrombotic syndrome (persistent pain and swelling, with or without ulceration). Siragusa et al.23 found an incidence of this disorder of 24%, compared to 4% in controls, after 2–4 years in patients who had subclinical DVT demonstrated by contrast venography (CV) following hip or knee surgery, despite at least 3 months anticoagulation. The risk was approximately 4‐fold higher in those with proximal compared to below‐knee thrombosis. A reduction in VTE‐related morbidity and mortality could be achieved either by preventing the formation of DVT, or by screening for DVT once formed, with initiation of treatment in selected cases to prevent progression, or a combination of both. However, primary prevention is obviously preferable to secondary prevention, and prophylaxis in high‐risk patients is the most cost‐efficient and clinically effective way to prevent the complications of subclinical DVT.1,24 Heparin prophylaxis in high‐risk patients is now a standard component of surgical care, a practice founded both on robust evidence of a reduced incidence of VTE and improved clinical end‐points. In an overview of 70 trials of prophylactic unfractionated heparin (UFH) in around 16 000 general, orthopaedic and urological surgery patients, DVTs and fatal PEs were reduced by two‐thirds, at the expense of a small excess of bleeds, and total mortality was reduced.19 Furthermore, low‐molecular‐weight heparin (LMWH) is at least as effective.25,26 The situation is, however, different in stroke patients. The incidence of asymptomatic DVT in the absence of prophylaxis is similar to that following orthopaedic surgery,9 although prophylaxis with UFH did not alter the likelihood of death or disability at 6 months in the International Stroke Trial.27 Although it may be argued that this may in part reflect the relative brevity of a 2‐week course of prophylaxis when many stroke patients remain immobile beyond this time frame, and that LMWH might have been more effective, heparin is no longer routinely recommended in these patients.7 More recently, a meta‐analysis of 15 000 internal medicine patients (excluding post‐stroke or myocardial infarction patients) has shown that heparin prophylaxis halved the risk of fatal PE though did not affect overall mortality, and that LMWH was as effective, but associated with a lower risk of bleeding.10 Aspirin reduces the risk of clinical VTE, including fatal PE, by at least one third in high‐risk medical, general surgical and orthopaedic patients, but has not been shown to reduce overall mortality.28,29 In the PEP trial, which recruited over 13 000 patients with hip fractures and over 4000 persons undergoing elective hip or knee arthroplasty, the dose of aspirin used was 160 mg o.d. Although major bleeds were increased by around a quarter in the hip fracture group, the absolute risk was small, and fatal bleeds were not increased.29 Combined data from over 40 000 patients has also shown that aspirin at a dose of 160 or 300 mg o.d. significantly reduces PE by 29% after acute ischaemic stroke.30 In a meta‐analysis of 11 high quality trials in a wide variety of surgical patients, graded compression elasticated stockings (GCS) reduced DVT by 68%, although their effect on PE and mortality were unclear.31 Unlike other prophylactic measures, they appear to be cost‐saving,32 and are routinely used in high‐risk surgical patients. Although there have been no large trials evaluating GCS in stroke patients, their effectiveness in this subgroup is currently being investigated in a multi‐centre trial. Based on surgical data, they have been recommended as standard care post‐stroke in the interim.33 Intermittent pneumatic compression devices help prevent DVT in general, orthopaedic and neurosurgical patients with a magnitude of effect comparable to that of heparin, and are therefore a useful alternative in patients at high risk of bleeding. However, there is less data with regard to PE and they have not been shown to reduce total mortality.1,34 They have also been shown to reduce both DVT and PE post‐stroke,35 although they are not ideally suited for the prolonged use often needed in these patients, and tend to be less well tolerated in this population than other prophylactic measures.36 There is currently a great deal of interest in novel antithrombotic agents, of which pentasaccharide and recombinant hirudin show the greatest promise. Both the synthetic pentasaccharide Org31540/SR90107A37 (a selective inhibitor of factor Xa) and the direct thrombin inhibitor desirudin38 (recombinant hirudin) have recently demonstrated superiority to the LMWH enoxaparin in preventing predominantly subclinical DVT following hip arthroplasty, with no added risk of bleeding. Further research is needed, particularly into whether or not clinical end‐points are improved compared to LMWH, and the cost‐effectiveness of newer agents, but if favourable, one or more of these agents may eventually supersede heparin as the thromboprophylactic of choice. Effective thromboprophylactic strategies are now routinely employed in most high‐risk patient groups. Frequently, more than one intervention is used and combinations of modalities may be more effective than agents used singly.9,39 However, the current approach in stroke patients may be suboptimal: heparin, the only intervention shown to reduce overall mortality in surgical patients,2 is no longer used, and although early use of aspirin and GCS are routine, the effect of the former is only modest and the latter are of unproven effectiveness in medical patients. The essence of any surveillance strategy would be the identification of proximal DVT in the expectation that anticoagulation at the pre‐symptomatic stage would prevent fatal PE. First convincingly demonstrated in 1960,40 anticoagulants are highly effective in reducing morbidity and mortality in VTE. In a recent overview of 21 studies of patients with symptomatic DVT, the risk of fatal PE during a 3‐month course of treatment was only 0.4%.6 In a study of 22 000 patients treated for DVT, the risk of all haemorrhage during hospitalization was 2.9%, and the incremental risk of major haemorrhage causing re‐hospitalization over the subsequent 3 months in the remainder, when compared to a cohort discharged after pneumonia or cellulitis, was 0.7%.41 Although haemorrhagic complications may be greater in the short term in some high‐risk patient subgroups, for example following acute ischaemic stroke,26 the ratio of benefit to risk of treating subclinical proximal DVT will usually be favourable. Despite heparin prophylaxis, asymptomatic DVT still occurs in 8–10% of general surgical patients,1,3 3–17% after major trauma,4 and 15–30% after hip or knee arthroplasty.1 Following orthopaedic surgery, new asymptomatic DVT also develops in 19–26% within 3–4 weeks of hospital discharge.5,6 Indeed, in the ICOPER study,42 which prospectively investigated 2110 unselected patients with acute PE, one‐third had received heparin prophylaxis. At first sight, these data raise the possibility that a combination of screening and standard prophylaxis might further improve outcome. This hypothesis was tested in a study investigating the effectiveness of pre‐discharge ultrasound (US) on clinical outcome in 1024 hip or knee arthroplasty patients following an average of 9 days warfarin prophylaxis. Patients were randomized to US or a sham procedure at discharge and DVTs were treated if confirmed venographically. Asymptomatic proximal DVTs were identified in 2.5% of the US group. However, the rate of clinical VTE at 3 months was the same in the intervention and control group at only 1%.43 Other studies in hip and knee arthroplasty patients confirm a 3‐month incidence of symptomatic VTE following an average of nine days in‐hospital prophylaxis of only 1–2%,44,45 despite an incidence of asymptomatic events several‐fold higher.5,6 In studies evaluating trauma patients screened with serial US in addition to standard prophylaxis, the prevalence of DVT has generally been around 5–6%.7,46–48 While surveillance has been advocated, there have been no large prospective trials validating this approach in terms of improved clinical outcomes in these patients. Routine screening in populations given adequate prophylaxis is therefore not currently recommended,1,9,20 though a much stronger case exists in high‐risk patients who have not received heparin prophylaxis.49 Acute stroke patients, in whom PE accounts for up to one quarter of early deaths,50 are an example of a group in whom the effectiveness of screening should be evaluated, with initial surveillance performed between days 7 and 10 post‐ictus, as most DVTs have formed within this time frame but the majority of PEs have yet to occur.51,52 Prospective studies are needed comparing clinical outcomes in patients systematically screened for subclinical proximal DVT, compared to usual care, although they would need to contain large numbers of patients to demonstrate a possible effect on overall mortality.53 Studies are also needed to evaluate whether subgroups with a particularly high prevalence of proximal DVT can be identified within cohorts of patients routinely receiving heparin prophylaxis, in whom selective screening might also be justified.54 For example, factors predicting DVT have been identified in trauma and hip arthroplasty patients,48,55 and a body mass index of 25 or greater has recently been identified as a key risk factor for post‐discharge VTE after hip arthroplasty.56 CV and US are the chief imaging modalities currently available to most clinicians, each of which have shortcomings as screening tools. These techniques, together with others that are primarily of research interest at present, but which might become important in due course, are discussed. This remains the gold standard for the investigation of symptomatic or asymptomatic lower limb DVT.14,20 However, the technique is invasive, is less sensitive when thrombosis is isolated to the pelvis, involves the use of ionising contrast material and may cause discomfort, allergic reactions, nephrotoxicity, skin necrosis and iatrogenic DVT.20 Therefore, although accurate, its use as a screening test in unselected patients is undesirable. US is now the initial investigation of choice in clinically‐suspected DVT,20 pooled data indicating a sensitivity and specificity of 96% and 98%, respectively for the diagnosis of symptomatic proximal DVT, although it is less sensitive for below‐knee and isolated pelvic DVT.57 It is, however, less useful for the detection of asymptomatic DVT. Combined data from 11 level 1 studies investigating the utility of various US techniques for the diagnosis of asymptomatic DVT in orthopaedic patients demonstrate a sensitivity of only 62% for proximal and 48% for below‐knee DVTs, and although the colour Doppler technique has theoretical advantages, there is no convincing evidence of its superiority to standard US.58–60 The specificity for proximal DVT was 97%, but the positive predictive value of an abnormal scan was only 66% because of low prevalence, suggesting that confirmation with CV should be considered before initiation of treatment.14,59 The reduced sensitivity is attributable to the fact that asymptomatic thrombi are more likely to be fresh, smaller and non‐occlusive than their symptomatic counterparts, and are therefore less likely to generate abnormal venous dynamics.58,61 In a study comparing US and CV in 100 neurosurgical patients, the sensitivity, specificity and positive predictive value of US for proximal DVT were 38%, 95% and 56% respectively.62 There are fewer data evaluating the utility of US as a screening test in other patient groups, and it is unclear whether its operating characteristics would be similar in these patients; for example, surgery‐related leg swelling might increase the technical difficulties of US after orthopaedic operations and therefore possibly diminish sensitivity.63 However, venographically‐demonstrated asymptomatic thrombi are not always a good surrogate marker for symptomatic events.43 It is possible that asymptomatic DVTs overlooked by US are low risk, and that despite a sensitivity which is only moderate, serial negative US examinations might yet be a useful marker of a subsequent low risk of clinical events. In support of this concept, US has been shown to be a powerful tool for predicting PE recurrence in patients with confirmed or suspected PE.64,65 For example, in patients without high clinical suspicion for PE, with non‐diagnostic ventilation perfusion scans, and negative over a 2‐week the risk of clinical VTE without treatment is only over the 3 even though a will have sustained a In addition, a data has shown that in patients with PE associated with scans, adequate cardiorespiratory reserve and negative serial US for DVT over the risk of subsequent fatal and PE is only and in the absence of This is despite the fact that US in patients with clinical PE less proximal DVT than Therefore, DVTs on serial US in patients with clinical PE are usually not associated with While it is possible that the same might when US is used as a screening this is and some currently the technique to be in this Studies are needed to evaluate the operating characteristics of US as a surveillance tool in high‐risk subgroups such as In addition, studies are to evaluate the utility of US as a screening in which the negative predictive value of or US studies for subsequent events is Studies performed to have either been small, or had low an a prevalence of DVT, to about the of negative US studies in this This technique on the of into with subsequent detection of increased over DVT. as a screening test to of it is now no longer used because of the risk of of with from pooled This increased venous in the of the proximal lower While sensitive for symptomatic proximal DVT less than it is not useful for the detection of or asymptomatic DVT.14,20 has the that it imaging of both lower detection of pelvic and if imaging during the same It can be used during and in the of or Pooled data from studies of venography in approximately 300 patients with clinically‐suspected DVT have shown that it to CV for the diagnosis of proximal DVT, with a sensitivity and specificity of and The technique not and is more sensitive than CV for the diagnosis of isolated pelvic There are fewer data in patients with symptomatic below‐knee DVT, though in one small study comparing to sensitivity and specificity were and 97%, has been for the diagnosis of asymptomatic DVT in only one in which it was to CV for proximal and pelvic in a group of patients with However, the technique is for small, which might its in asymptomatic direct imaging is a new of which in direct of as a high a of and While techniques have identified as a the of other imaging modalities and about a positive of The technique has recently been for the diagnosis of symptomatic et compared CV and in a prospective study of 100 patients with suspected DVT, with US used as the in had an overall sensitivity of and specificity of with for proximal and pelvic DVT of 96% and This technique has many as a screening for example of which is a more powerful of than its proximal but has not been in this However, similar characteristics might be in asymptomatic DVT, as it a positive of than the of on venous are a from the of the in venous Given that are increased in any in which is formed they are not a marker for but are highly particularly when with to the of They have a useful component in diagnostic as an Although cannot be used to a positive diagnosis of VTE, the concept that they might have utility as a screening tool is of it might be possible to a to a patient group, and also on the of an medical or surgical which patients with a high likelihood of proximal DVT, major DVT with in the This could a more use of imaging to confirm or significant DVT. A of such studies have been performed in a variety of high‐risk In orthopaedic patients, have generally because of the effect of surgery on However, in one study of hip arthroplasty and in patients following major or were for DVT et investigated stroke patients who on 25 days Patients were screened with US and within the same A of had positive and negative predictive for DVT of and and DVT in of patients. A of DVT surveillance after stroke should much as most DVTs within a but these data cannot be to patients in the acute because of potential by the effect of stroke on The of as a screening tool in this therefore These that in a screening an initial might significantly the of patients requiring However, should be that a in a study only to that subgroup and used, as each of the available have their operating In may identification of a subgroup with an approximately of DVT, which with the patients presenting with clinically‐suspected DVT in whom the diagnosis is Studies are needed to evaluate the utility of such an approach on clinical and its In an era of effective surveillance for asymptomatic DVT is rarely The effectiveness of screening for DVT in addition to standard prophylaxis has been in only one substantial trial, which did not show an in outcomes in a cohort of hip arthroplasty patients. While screening of high‐risk patients who have received adequate prophylaxis, particularly more than one thromboprophylactic intervention is used, may not be it may be possible to subgroups within these cohorts at particularly high risk of proximal DVT in whom screening would be A case for screening can be in patients in whom prophylaxis may be and a strategy for screening stroke patients, who do not routinely receive heparin prophylaxis and in whom PE to be an important cause of early mortality, should be a research The current choice of screening tool is between which is and for unselected or The sensitivity of US for asymptomatic DVT is only moderate, but examinations may reflect a lower associated and research is needed to evaluate whether or not negative studies still a low subsequent risk of clinical VTE. Further is needed in the of particularly direct as a screening If and given greater this technique might eventually alternative imaging modalities for this the of in a screening warrants further to