Correspondence

Enoxaparin for the Prevention of Venous Thromboembolism

N Engl J Med 2000; 342:136-137January 13, 2000

Article

To the Editor:

Samama et al. (Sept. 9 issue)1 compared enoxaparin with saline placebo for the prevention of venous thromboembolism in acutely ill medical patients. I was surprised that the study drug was compared with a placebo, since venous thromboembolism is a potentially life-threatening illness and since those receiving the placebo would have a higher incidence of morbidity. The authors state, “The use of a placebo group was considered to be ethically justifiable because the incidence of venous thromboembolism among such patients had not been established, there was no established method of thromboprophylaxis for these patients, and patients with a very high risk of venous thromboembolism were excluded.” This statement does not make sense to me.

First, with regard to the incidence of thromboembolism, Samama et al. state that they assumed an incidence of 15 percent, and the results in their placebo group confirm this assumption. Thus, they knew very well what the incidence of thromboembolism was. Second, use of unfractionated heparin is a standard method of prophylaxis against this disease, so it would have been better to compare enoxaparin with this standard and to prevent any harm to patients in a placebo group. Indeed, one of the authors' references2 is a meta-analysis of comparisons of low-molecular-weight heparin with unfractionated heparin, placebo, and dextran for the prevention of venous thromboembolism. The conclusions were that both forms of heparin were better than dextran or placebo and that a clear decision as to which form of heparin was better could not be made. Unfortunately, because of the use of a placebo group in the study by Samama et al., approximately 27 more patients had deep-vein thrombosis than would have if all groups had received prophylaxis, 14 more had hemorrhage, and 4 more died during the treatment period. This is a high price to pay to avoid a direct comparison with the standard therapy.

Moreover, Samama et al. state that of 1102 patients enrolled in the study, 1073 were included in the analysis of safety. What happened to the others? They should have been included since the authors mention in the Methods section that they applied the intention-to-treat principle. The authors also state that “one patient in the 40-mg group died, but the hemorrhage was not considered to be related to treatment (it was characterized by massive hemoptysis due to bronchial carcinoma).” All bleeding must be considered a complication of the study drug.

Mitchel B. Sosis, M.D., Ph.D.
5804 Parade Field Way, Lansdale, PA 19446

2 References

1. 1

   Samama MM, Cohen AT, Darmon J, et al. A comparison of enoxaparin with placebo for the prevention of venous thromboembolism in acutely ill medical patients. N Engl J Med 1999;341:793-800
   Full Text | Web of Science | Medline

2. 2

   Leizorovicz A, Haugh MC, Chapuis FR, Samama MM, Boissel JP. Low molecular weight heparin in prevention of perioperative thrombosis. BMJ 1992;305:9913-9920
   CrossRef

To the Editor:

In the sole figure included in their report on the efficacy and safety of prophylactic enoxaparin in acutely ill, hospitalized medical patients, Samama et al. imply with their graph and explicitly state in their figure legend that 40 mg of enoxaparin reduced the risk of death relative to the risk associated with placebo. According to the authors' own statistical analysis, this conclusion is not true. The authors report a relative risk of death in the 40-mg enoxaparin group of 0.83, with a 95 percent confidence interval of 0.56 to 1.21 and a P value of 0.31. Therefore, no reduction in the risk of death was identified. To be correct, the authors should have stated that there was no statistically significant difference between the 40-mg group and the placebo group in the risk of death.

James M. Hynson, M.D.
University of California, San Francisco, School of Medicine, San Francisco, CA 94143

To the Editor:

Samama and colleagues concluded that prophylactic treatment with 40 mg of enoxaparin per day, given subcutaneously, safely reduces the risk of venous thromboembolism in patients with acute medical illnesses. Analysis of the primary outcome shows that the number of patients who would need to be treated1 to prevent one venous thromboembolic event by using 40 mg of enoxaparin per day rather than placebo (number needed to treat) was 11. However, 62 percent of the prevented events were instances of distal deep-vein thrombosis. The absolute reduction in the risk of pulmonary embolism was 1.1 percent, and the number needed to treat to prevent one episode of pulmonary embolism was 90. Analysis of the incidence of severe hemorrhage (major or fatal) in both the 40-mg enoxaparin group and the placebo group shows that the number needed to harm (the number of patients who would have to receive enoxaparin for 1 patient to have severe hemorrhage) was 125. Thus, about 300 hospitalized medical patients would need to be treated with 40 mg of enoxaparin per day to prevent three cases of pulmonary embolism, but two cases of severe hemorrhage would result.

Enrique J. Calderon, M.D.
Jose M. Varela, M.D.
Miguel A. Gonzalez de la Puente, M.D.
Virgen del Rocio University Hospital, 41013 Seville, Spain

1 References

1. 1

   Laupacis A, Sackett DL, Roberts RS. An assessment of clinically useful measures of the consequences of treatment. N Engl J Med 1988;318:1728-1733
   Full Text | Web of Science | Medline

Author/Editor Response

The authors reply:

To the Editor: We cannot agree with Dr. Sosis's comment that unfractionated heparin is a standard treatment. At the time we planned our study, some physicians were using thromboprophylaxis in general medical patients, but there was no solid scientific evidence to support this practice. Controlled studies of unfractionated heparin have been small, and on the basis of their results it is difficult to make recommendations about its use. Studies of mortality have had various methodologic limitations, and the results are conflicting, as we indicated in our article. Consensus in this area is not definitive, and recommendations are controversial.1,2

Also in response to Dr. Sosis: the frequency of venous thromboembolism was not precisely known and was anticipated to be 15 percent on the basis of the hypothesis that our population would be at moderate risk, according to venographic data from studies of surgical patients (these data are unavailable for medical patients). We wanted to evaluate whether the response to therapy would be the same as it is for surgical patients, and indeed, for the lower dose of enoxaparin (20 mg), it was not. Hence, we included a group that received placebo rather than low-dose unfractionated heparin. In contrast to Dr. Sosis's contention, we considered it unethical to treat patients with a potentially harmful treatment such as anticoagulants when the benefit had not been clearly established. We agree that analysis according to the intention to treat is the primary method of analysis, but to evaluate unwanted events, an analysis of data on all patients who received at least one dose of therapy was legitimate. Although treatment may have contributed to the massive hemoptysis in the patient with bronchial carcinoma, the investigator who performed the blinded assessment reported that this episode was unrelated to the study treatment. Autopsy findings were consistent with neoplastic vessel erosion.

In response to Dr. Hynson: the legend to Figure 1 neither implies nor states that enoxaparin reduced the risk of death relative to that associated with placebo. However, we agree that this issue could have been clarified by adding a P value, consistent with the legend and the Results section, to the figure. Although data on the risk of death constituted neither a primary nor a secondary efficacy outcome, this was a major prespecified safety outcome.

In response to Dr. Calderon and colleagues: the number needed to treat and the number needed to harm are other ways to express absolute differences, but they should be presented with confidence intervals. They are meaningful only for statistically significant end points. In the calculations of Dr. Calderon and colleagues, the confidence intervals would include a negative number of patients to treat.

Meyer Michel Samama, M.D.
Hôtel Dieu, 75181 Paris CEDEX 04, France

Alexander Thomas Cohen, M.D.
Guy's, King's, and St. Thomas' School of Medicine, London SE5 9RS, United Kingdom

Alain Leizorovicz, M.D.
University of Lyons, 69424 Lyons CEDEX 03, France

for the Prophylaxis in Medical Patients with Enoxaparin Study Group

2 References

1. 1

   Nicolaides AN. Prevention of venous thromboembolism: European Consensus Statement. Int Angiol 1997;16:3-38
   Web of Science | Medline

2. 2

   Clagett GP, Anderson FA Jr, Heit J, Levine MN, Wheeler HB. Prevention of venous thromboembolism. Chest 1995;108:Suppl:312S-334S
   CrossRef | Web of Science | Medline