Re: NICE’s recommendations for thromboembolism are not evidence based
I thank the wide interest in my Personal View, the format of which did not allow detailed examination of the evidence and further positive suggestions. The polemical style has perhaps seemed to polarise views more than need be. I note the general support from day to day clinicians.
Lifeblood have made detailed comments which I would like to respond to and amplify. Firstly I do not doubt the individual sincerity of Lifeblood members and NICE members. That is not the way Pharma works. Rather they use the 'food, flattery and friendship plus funding' route so eloquently described by Peter Grant.
A further good example of the work of Pharma is paradoxically given within Lifeblood's response. My inability to understand where the 2005 claims of 25,000 preventable deaths was because the reference now given where it originated was published two years later in 2007 (The VITAE study, Cohen et al, 2007)! The paper declares that ‘The VITAE study was funded by sanofi-aventis. The model structure and underlying data were approved by an independent expert board. Editorial assistance was provided by sanofi-aventis’. Detailed analysis of the VITAE study shows that the claims that 60,000 people in the UK die of VTE each year is based on just one epidemiological study from France and an algorithm which purports to describe the relationship between DVT (symptomatic and undiagnosed) and PE (detected and otherwise). The study then uses four layers of projection to achieve its inflated claim. As Lifeblood's response indicates, the 25,000 prevented deaths would require prophylaxis to be 66% effective at preventing death from VTE which it clearly is not. Whilst I agree that diagnosis of PE is difficult, death is a fairly robust end-point.
The central claim of the policy is that 25,000 deaths will be prevented. This key claim cannot be maintained and should be dropped. Only direct high quality meta-analysis can give us the answer to whether deaths can be prevented - NICE's own study showed that in medical patients it probably cannot with current medications. Projections and data from unrelated post-mortem studies are interesting but do not show that death can be prevented.
DVT, principally asymptomatic, is prevented but that is not the claim of NICE. I could support a campaign that advertised itself as 'preventing 25,000, largely asymptomatic, DVTs', but that is not quite so catchy.
Then what about harms - a recent further meta-analysis of prophylaxis in medical patients (Ann Intern Med2011;155:602-15) suggests that for every 1000 medical and stroke patients given LMWH, 3 symptomatic PEs will be prevented but at the expense of 4 major bleeds with no conclusive effect on death. Lifeblood emphasise the reduction in suffering from preventing DVT whilst ignoring the 'suffering' from major bleeds some of which are intra-cerebral. It should also be pointed out that trials only recruited patients with lo risk of bleeding so arguing that careful case selection will prevent these bleeds is superfluous - the trial data with an excess of bleeding apply to low risk populations.
It would be interesting to read the quality of patient information leaflets relating to the decision to accept prophylaxis and how they express these data.
NICE should review their guidance in the light of this new evidence and the widespread concern about the policy amongst clinicians. My central thesis is that the potential of preventing deaths from VTE will almost certainly not be fulfilled using current recommendations and that we should begin a search for new strategies from a blank sheet using a wider range of expert input.
If we can accept that at current state of knowledge, lmwh may not prevent VTE, what other strategies could be used? these are strategies I conceived with just 10 minutes thought:
• Routine scanning of all patients admitted at some pre-determined point (3-5 days seems reasonable) to look for asymptomatic DVT and giving treatment dose after that. This would ensure that treatment was targeted at the highest risk patients but with a risk that it would be too late for some
• Further studies of high dose aspirin in high risk patients given that aspirin seems to show promise in some studies but because it is not a patent protected drug has been poorly studied in this context
• A trial with death or symptomatic PE as end point - sample size calculation (http://www.sealedenvelope.com/power_binary.php) suggests that with 90% certainty and 0.05 significance level, just 68,424 patients would need to be randomized to test the hypothesis that prophylaxis improves survival from 96% (hospital survival in Stein) to 95.5%. If survival advantage is better, 18,058 patients would be required in each arm to examine improvement from 96% to 95%. If the calculations put forward by NICE etc are used, 15% of patients in the trials died and 10% of deaths were due to VTE and treatment was 50% protective, this would imply a reduction in mortality from 15% to 14.25%, then 107274 patients would need to be randomized but if survival was improved by 1% to 86% then 52112 would be required.
• Further studies of the best length of treatment for LMWH. Many of the studies (e.g Prevent which examined effectiveness of Dalteparin) used treatment until discharge or 14 days of treatment as a minimum even if discharged before. Currently the policy being implemented means that in reality patients get treatment only whilst an in-patient and mean length of stay is 5.5 days – we may well be implementing an ineffective regime at present
• Various strategies based on early mobilization, foot exercises, early discharge etc could be assessed and whilst having cost implications they would have few side effects. Similarly, approaches to reduce admissions rates and offer ambulatory care at home could be effective. Length of stay has reduced considerably since the main studies were conducted.
• Concentrate on better diagnosis of VTE. Currently there are clearly some patients dying of PE after presenting to doctors and being misdiagnosed. Better training, more routine use of risk scores and D-dimer tests in primary and secondary care, lower thresholds for performing CT pulmonary angiography and Doppler of legs and better monitoring of respiratory rate and hypoxia in primary care could all be useful strategies. One starting point would be an analysis of the deaths of patients where PE has been missed.
• Many medical patients are readmitted within 90 days and many have ongoing risk for VTE irrespective of admission. In the high risk patients (e.g. cancer, chronic inflammation) consideration should be given to long term prophylactic strategies
I would like the debate on VTE prophylaxis to move forward. I hope that my article, through its polemical approach, has initiated new debate which will be accepted by NICE and all interested parties can move forward together on this important subject
Competing interests: as before