Current issues of ACP Journal Club are published in Annals of Internal Medicine


Review: Chelation therapy is ineffective for peripheral arterial occlusive disease

ACP J Club. 1998 Mar-April; 128:38. doi:10.7326/ACPJC-1998-128-2-038

Source Citation

Ernst E. Chelation therapy for peripheral arterial occlusive disease: a systematic review. Circulation. 1997 Aug 5;96:1031-3.



To determine whether chelation therapy is effective for patients with peripheral arterial occlusive disease.

Data sources

Studies were identified using MEDLINE (1966 to 1996) and CISCOM (1996) databases, bibliographies of relevant papers, and suggestions from 6 experts and national societies.

Study selection

Studies were selected if they were randomized, placebo-controlled, double-blind trials of intravenous chelation therapy for peripheral arterial occlusive disease.

Data extraction

Data were extracted on sample characteristics, intervention, treatment schedule, patient follow-up, primary outcomes, and results. Outcomes included walking distances, ankle/arm blood pressure, transcutaneous oxygen tension, and angiogram results.

Main results

4 papers reported on 3 trials that involved patients with intermittent claudication. Meta-analysis was not done. None of the studies found intergroup differences for patients receiving chelation therapy compared with patients receiving placebo. A small trial of 10 men reported no intergroup differences in walking distance or ankle/arm blood pressure for patients who received 10 intravenous injections of disodium EDTA, 1.5 g, compared with patients who received placebo. A trial involving 153 patients allocated to 20 treatments of disodium EDTA, 3 g, or isotonic saline over 5 to 9 weeks found no intergroup differences for arteriogram or oxygenpressure at the end of treatment or after 6 months. A second paper analyzed data on a subgroup of 30 patients from the same trial and found a non-significant trend toward greater improvements in walking distance and peripheral arterial pressure among patients in the placebo group compared with the treatment group. 6 patients who received disodium EDTA showed clinical signs of hypocalcemia. A study of 32 patients who were allocated to 20 infusions of disodium EDTA, 3 g (n = 15), or to saline (n = 17) over 10 weeks found no intergroup differences for walking distances or ankle/arm blood pressure at 3-month follow-up. The proportion of patients who improved did not differ by group (60% vs 59%).


Intravenous chelation therapy offers no benefit to patients with peripheral arterial occlusive disease. 3 trials found no intergroup differences for patients who received EDTA compared with those who received placebo, and a subgroup analysis from 1 of the trials reported a trend toward greater improvements for patients who received placebo than for those who received EDTA.

Sources of funding: No external funding.

For article reprint: Dr. E. Ernst, University of Exeter, Department of Complementary Medicine, Postgraduate Medical School, 25 Victoria Park Road, Exeter EX2 4NT, England, UK. FAX 44-1392-430802.


Ernst suggests that a tenuous or even false rationale does not necessarily preclude clinical effectiveness and then examines the evidence to ascertain whether chelation is effective for peripheral arterial occlusive disease. It is quite misleading, however, to suggest that the results of this systematic review are as convincing and definitive about the lack of clinical effectiveness of chelation therapy for peripheral arterial disease as the author suggests.

Large improvements in the curability of vascular diseases are notoriously elusive. Reliable clinical research into whether a moderate difference exists between the effects of 2 treatments requires particularly strict control of systematic errors (biases) and random errors (chance) without undue data-dependent emphasis on the results of any particular subgroup or trial (1). Double-blind, placebo-controlled trials can control for systematic errors. Strict control of random error is simple in principle but difficult in practice because many patients must be randomly assigned. Small trials (or overviews of several small trials) cannot distinguish moderate benefit from no benefit.

It is unlikely that the trials included in this review had sufficient power to reliably exclude any major therapeutic benefit. However, evidence to suggest that this hypothesis might be worth testing in a larger clinical trial is also lacking.

Robert Clarke, MD
Radcliffe InfirmaryOxford, England, UK


1. Peto R. Statistics of chronic disease control. Nature. 1992;356:557-8.