The risk of VTE in relation to surgical procedures is still largely dependent on procedural risk and a simple RAM based on procedure and age is frequently used in practice (Geerts et al, 2008; Kearon & Hirsh, 2001; Salzman & Hirsh, 1982). Nevertheless, despite the availability of simple, cheap and effective prophylaxis, it is still underused in surgical patients (Kakkar et al, 2004; Yu et al, 2007). RAMs are not as well established for medical patients and only a minority currently receive prophylaxis (Amin et al, 2007; Cohen et al, 2008; Kahn et al, 2007; Tapson et al, 2007; Yu et al, 2007). Risk factors for VTE in medical patients have been identified from registries, epidemiological, case control, observational cohorts and control arms of randomised interventional studies. The risks of VTE and the benefits of prophylaxis have been reviewed and a comprehensive RAM derived for medical patients (Cohen et al, 2005). This relatively simple RAM combines medical conditions associated with VTE risk with individual patient risks, classified as evidence-based or consensus-based. The output from the RAM is either ‘yes’ or ‘no’ when deciding whether a medical patient should receive thromboprophylaxis.
Regardless of the approach to risk assessment, further refinements in defining at-risk patients are required as clinical practice continually changes and hence, the risk attributable to individual factors changes. Furthermore, how the various risk factors interact to determine risk is still unknown and so RAMs may yet be further simplified after multivariate analysis of large databases of patient demographics and outcomes in relation to hospital episodes. Hospital Episode Statistics (HES) is the national statistical data warehouse relating to care provided by NHS hospitals and for NHS hospital patients treated elsewhere and this may become a reliable resource that may enable identification of the majority of patients at risk by a minimum number of clinical risk factors. A large dataset would enable identification of risk factors and interaction by comparing regional or national patient and control cohorts.
There are reasons to prevent VTE other than death; namely morbidity from DVT and PE including long term complications, such as post-thrombotic syndrome and chronic pulmonary hypertension, risk of bleeding from therapeutic intensity anticoagulation, delayed discharge and re-admission. For example, the severity of hospital-acquired thrombosis and the complications due to having to administer therapeutic intensity anticoagulation were highlighted by a study in which 3·4% of patients died, with PE considered to be a major contributory factor in 40% of these, and 5% of all treated patients suffered gastrointestinal haemorrhage with one fatal cerebral bleed (Goldhaber et al, 2000). Despite many reasons for preventing hospital-acquired thrombosis the primary declared reason for risk assessment and prophylaxis in England, if not the UK, is prevention of hospital deaths due to PE (Department of Health 2008b). It is unlikely that reduction of hospital-acquired infection would have been achieved without prioritisation at the highest level of management within the NHS and appropriate allocation of resources. However, equally important was the ability to directly measure hospital-acquired infection by microbiological identification so that the effect of intervention could be monitored by measurement of the primary target, i.e. infection. In contrast measurement of fatal PE is problematic. At least 80% of fatal hospital-acquired emboli are not suspected before death and therefore the only way to monitor the primary outcome is by post mortem examination. However, due to procedural changes and public perceptions regarding post mortems it was predicted that post mortem rates would fall in the UK (Alikhan et al, 2004; Carr et al, 2002) and this has been confirmed. For example, post mortem rates at Addenbrooke’s Hospital have fallen by 95% since the post mortem study of fatal hospital-acquired thrombosis was reported (Baglin et al, 1997) (Fig 1). Therefore, the effort to reduce hospital-acquired thrombosis will be hindered by an inability to measure the effect of strategic interventions on the primary outcome of death. Instead, measurement of the surrogate marker of symptomatic non-fatal VTE will have to suffice. Events could theoretically be achieved by monitoring completed hospital episodes coded according to the International Classification of Disease (ICD 10th edition – I26 & I80 with Y83 identifying surgical intervention). However, inaccurate coding and overlap between primary and secondary coding may hamper this approach for some time in many hospitals. A likely, more reliable approach to monitoring, at least at present, is identification of patients referred for anticoagulant treatment within 3 months of hospitalisation. This process of identification can be supplemented with reports of positive ultrasounds and lung imaging from radiology departments. However, this method of monitoring is only likely to be informative if the majority of patients are referred to the same hospital for diagnosis and treatment of VTE as in which they were originally hospitalised. Thus, for many cities in the UK it is quite possible, indeed likely, that many patients will move between hospitals. Defining an episode of VTE as hospital-acquired requires validation at least by notes review and is likely to be most beneficial if linked by feedback to the initial clinical team. With this in mind a ‘Look Back’ programme was introduced at Addenbrookes’s Hospital in January 2008. Patients developing VTE in hospital or within 3 months of discharge are identified and the RAM, modelled on the Department of Health RAM (Department of Health 2008b), is retrospectively completed from the patient’s clinical record taken at the time of initial hospitalisation. It is then possible to determine if hospital-acquired VTE was the result of a failure of risk assessment, a failure to implement prophylaxis or a prophylaxis failure. This analysis is important as a report from a hospital with a high rate of thromboprophylaxis demonstrated that most fatal hospital-acquired thrombosis was due to failed prophylaxis rather than omitted prophylaxis (Goldhaber et al, 2000). In the Addenbrooke’s Look Back programme individual results are fed back to the responsible consultant and collectively results are incorporated into a hospital-wide dataset (Table III).