See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/236281601

Efficacy and Safety of Ephedra and Ephedrine for Weight Loss and Athletic
PerformanceA Meta-analysis

Article in JAMA The Journal of the American Medical Association · March 2003
DOI: 10.1001/jama.289.12.1470

CITATIONS READS
30 1,783

9 authors, including:

Mary L Hardy Sally C. Morton

University of California, Los Angeles Arizona State University
15 PUBLICATIONS 1,178 CITATIONS 216 PUBLICATIONS 24,949 CITATIONS

SEE PROFILE SEE PROFILE

Marika Suttorp Booth Shannon Rhodes

RAND Corporation University of California, Los Angeles
115 PUBLICATIONS 11,436 CITATIONS 65 PUBLICATIONS 5,954 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Systematic Review of Complex Interventions View project

Teryl CQI View project

All content following this page was uploaded by Marika Suttorp Booth on 26 May 2014.

The user has requested enhancement of the downloaded file.

 REVIEW

Efficacy and Safety of Ephedra

and Ephedrine for Weight Loss
and Athletic Performance
A Meta-analysis
Paul G. Shekelle, MD, PhD Context Ephedra and ephedrine sometimes are used for weight loss or enhanced
Mary L. Hardy, MD athletic performance, but the efficacy and safety of these compounds are uncertain.
Sally C. Morton, PhD Objective To assess the efficacy and safety of ephedra and ephedrine used for weight
loss and enhanced athletic performance.
Margaret Maglione, MPP
Data Sources We searched 9 databases using the terms ephedra, ephedrine, ad-
Walter A. Mojica, MD, MPH verse effect, side effect, efficacy, effective, and toxic. We included unpublished trials
Marika J. Suttorp, MS and non–English-language documents. Adverse events reported to the US Food and
Drug Administration MedWatch program were assessed.
Shannon L. Rhodes, MFA
Study Selection Eligible studies were controlled trials of ephedra or ephedrine used
Lara Jungvig, BA for weight loss or athletic performance and case reports of adverse events associated with
James Gagné, MD such use. Eligible studies for weight loss were human studies with at least 8 weeks of
follow-up; and for athletic performance, those having no minimum follow-up. Eligible

U
SE OF EPHEDRINE ALKALOID– case reports documented that ephedra or ephedrine was consumed within 24 hours prior
containing products to pro- to an adverse event or that ephedrine or an associated product was found in blood or
mote weight loss or to en- urine, and that other potential causes had been excluded. Of the 530 articles screened,
52 controlled trials and 65 case reports were included in the adverse events analysis. Of
hance athletic performance more than 18000 other case reports screened, 284 underwent detailed review.
has garnered a great deal of recent at-
tention, due in part to a number of well- Data Extraction Two reviewers independently identified trials of efficacy and safety
of ephedra and ephedrine on weight loss or athletic performance; disagreements were
publicized adverse events reportedly as-
resolved by consensus. Case reports were reviewed with explicit and implicit methods.
sociated with the use of ephedra- or
ephedrine alkaloid–containing prod- Data Synthesis No weight loss trials assessed duration of treatment greater than 6
months. Pooled results for trials comparing placebo with ephedrine (n=5), ephedrine
ucts.1-3 These reports have led several
and caffeine (n=12), ephedra (n=1), and ephedra and herbs containing caffeine (n=4)
groups to ask the US Food and Drug Ad- yielded estimates of weight loss (more than placebo) of 0.6 (95% confidence interval,
ministration (FDA) to ban the produc- 0.2-1.0), 1.0 (0.7-1.3), 0.8 (0.4-1.2), and 1.0 (0.6-1.3) kg/mo, respectively. Sensitiv-
tion and sale of ephedra products.4 Ad- ity analyses did not substantially alter the latter 3 results. No trials of ephedra and ath-
vocates counter that ephedra is safe and letic performance were found; 7 trials of ephedrine were too heterogeneous to syn-
effective.5 The US Department of Health thesize. Safety data from 50 trials yielded estimates of 2.2- to 3.6-fold increases in
and Human Services requested this syn- odds of psychiatric, autonomic, or gastrointestinal symptoms, and heart palpitations.
thesis of available evidence regarding ef- Data are insufficient to draw conclusions about adverse events occurring at a rate less
ficacy and safety of ephedra use to clarify than 1.0 per thousand. The majority of case reports are insufficiently documented to
allow meaningful assessment.
the existing state of the science on
ephedrine alkaloids. The National In- Conclusions Ephedrine and ephedra promote modest short-term weight loss (⬇0.9
stitutes of Health will use this informa- kg/mo more than placebo) in clinical trials. There are no data regarding long-term weight
loss, and evidence to support use of ephedra for athletic performance is insufficient.
tion to guide an expanded research ef-
Use of ephedra or ephedrine and caffeine is associated with increased risk of psychi-
fort to better understand the safety of atric, autonomic, or gastrointestinal symptoms, and heart palpitations.
ephedrine alkaloids.6 JAMA. 2003;289:1537-1545 www.jama.com

Author Affiliations are listed at the end of this article. Practice Center–RAND, 1700 Main St, PO Box 2138,
See also p 1568 and Patient Page. Corresponding Author and Reprints: Paul G. Shekelle, Santa Monica, CA 90407-2138 (e-mail: shekelle
MD, PhD, Southern California Evidence-based @rand.org).

©2003 American Medical Association. All rights reserved. (Reprinted) JAMA, March 26, 2003—Vol 289, No. 12 1537

Downloaded From: http://jama.jamanetwork.com/ on 02/25/2013

 EPHEDRA AND EPHEDRINE FOR WEIGHT LOSS AND ATHLETIC PERFORMANCE

METHODS Trials assessing athletic perfor- We assessed the possibility of pub-

Identification and Synthesis mance were too heterogeneous to sup- lication bias by evaluating a funnel plot
of Controlled Trials port statistical pooling. For the weight of effect sizes for asymmetry, which re-
Since ephedrine is believed to be the al- loss and adverse events analyses, we sults from the nonpublication of small,
kaloid most responsible for the effects abstracted the relevant outcome data negative trials. Because graphical evalu-
of herbal ephedra, we searched for con- (weight loss, occurrence of an adverse ation can be subjective, we also con-
trolled trials of ephedrine or ephedra event). Weight loss outcomes were ducted an adjusted rank correlation
that assessed either weight loss or ath- pooled in clinically similar groups of test12 and a regression asymmetry test.13
letic performance in humans. trials. We first verified that weight loss We conducted all analyses using
We searched the following elec- was linear over the range of time for Stata14 except for the exact analyses,
tronic databases in March and April which data were available by compar- which were conducted using StatXact,15
2001: MEDLINE (1965-2001), EMBASE ing pooled monthly rates of weight loss with P⬍.05 considered significant. A
(1974-2001), BIOSIS (1969-2001), Al- based on 2- month, 3- month, and more detailed explanation of these meth-
lied and Complementary Medicine Da- 4-month data separately. We then pooled ods is available at http://www.rand.org
tabase (AMED) (1984-2001), MANTIS monthly weight loss effect sizes using a /health/epc.
(1880-2000), the Cochrane Con- random-effects model.7 We also pooled Because the randomized trials had,
trolled Clinical Trials Register Data- percentage weight loss from baseline in even in aggregate, low statistical power
base, and International Pharmaceuti- the treatment groups for comparison to assess the possibility of rare adverse
cal Abstracts (1970-2001), PASCAL with other weight loss products. events, we searched for and reviewed
(1973-2001), and SCISEARCH (1974- We tested for a dose effect using a case reports of adverse events associ-
1989, 1990-2001). Search terms in- meta-regression model,8 defining an ated with use of ephedrine or ephedra.
cluded ephedra, ephedrine, adverse effect, ephedrine low dose as 10 to 20 mg/d; Case reports cannot be considered, ex-
side effect, efficacy, effective, and toxic. The medium dose as 40 to 90 mg/d; and cept in extraordinary circumstances, to
specific search strategies are displayed high dose as 100 to 150 mg/d. be conclusive evidence of a cause-and-
in an Appendix of detailed methods We conducted sensitivity analyses to effect relationship. Case reports can,
available at http://www.rand.org/health determine the robustness of our conclu- however, be useful to establish the po-
/epc. We then searched the reference lists sions. One sensitivity analysis divided the tential for a causal relationship. We as-
of each retrieved article. There was no trials into those with less than 20% at- sessed all reports for ephedrine and
language restriction. We also searched trition compared with all others. A sec- ephedra in the FDA MedWatch files as
for unpublished or “gray” literature by ond analysis dropped an included trial of September 30, 2001, all case reports
posting announcements in the jour- by Moheb et al9 that presented data in we identified in published literature, and
nals Phytomedicine and Herbalgram and abstract form and required sample-size a very large file of symptoms reported
by writing to numerous regulatory of- assumptions. A third analysis assessed to a manufacturer of ephedra-contain-
ficials and herbal experts in other coun- only those trials scoring 3 or greater on ing dietary supplements. All case re-
tries seeking unpublished evidence. the Jadad scale of quality,10,11 which gives ports were screened and we reviewed in
Two trained reviewers indepen- points based on randomization, blind- detail all reports of death, myocardial in-
dently reviewed titles, abstracts, and ar- ing, and description of withdrawals, and farction/acute coronary syndromes/
ticles to identify controlled trials in hu- ranges from 0 (poor) to 5 (good). In other cardiac symptoms, cerebrovascu-
mans assessing weight loss or athletic other settings a threshold of 3 has been lar accident/transient ischemic attack/
performance. Disagreements were re- shown to be related to bias.11 We also per- other neurologic symptoms, seizure, and
solved by consensus. Articles meeting formed sensitivity analyses where rel- serious psychiatric symptoms.
these inclusion criteria were then re- evant by dropping trials that included We searched for documentation that
viewed in duplicate to abstract data re- other medications (aspirin, syneph- (1) an adverse event had occurred; (2)
garding study design and execution, par- rine) with ephedrine. The sensitivity the subject had consumed ephedra or
ticipants, therapies, and outcomes analysis based on study quality was speci- ephedrine within 24 hours prior to the
(abstracted variables are available at fied a priori; all others were post hoc af- adverse event or that a toxicological ex-
http://www.rand.org/health/epc). To as- ter examining the relevant studies. amination had revealed ephedrine or
sess efficacy in weight loss, we included For the safety analysis, we esti- one of its associated products in the
only trials with at least 8 weeks’ dura- mated the pooled odds ratio (OR) for blood or urine; and (3) an adequate in-
tion of treatment, because shorter dura- clinically related groups of symptoms vestigation had evaluated for and ex-
tions of treatment are thought to be less using exact conditional inference meth- cluded other potential causes. Cases
informative. To assess adverse events, we ods because of the rarity of events. To meeting all these criteria were labeled
included all identified controlled trials assess dose response, we conducted a “sentinel events.” Cases meeting the
for either indication. Articles were not stratified analysis with the dose strata first 2 criteria that had other possible
masked for author or journal. defined as for the efficacy analysis. causes of the event were labeled “pos-
1538 JAMA, March 26, 2003—Vol 289, No. 12 (Reprinted) ©2003 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ on 02/25/2013

 EPHEDRA AND EPHEDRINE FOR WEIGHT LOSS AND ATHLETIC PERFORMANCE

sible sentinel events.” We used the clini- was 11% at 4 months. A sensitivity proached statistical significance
cal judgment of expert clinicians to as- analysis of only those trials scoring 3 (P = .05). Neither graphical assess-
sess whether other causes had been or greater on the Jadad scale yielded a ment nor statistical tests yielded evi-
adequately evaluated and excluded. pooled estimate of effect substantially dence of publication bias (P = .45 for
lower than that yielded by the main rank test; P=.82 for regression test).
RESULTS analysis (weight loss rate, 0.2 kg per Ephedrine and Caffeine vs Pla-
FIGURE 1 illustrates the flow of litera- month); this difference (compared with cebo. We identified 12 trials that as-
ture reviewed. We identified 550 ar- the main analysis) was statistically sig- sessed the effect of ephedrine and caf-
ticles. These included 3 clinical trials that nificant (P=.049). All of these trials had feine vs placebo for weight loss,* with
were unpublished at the time our re- an attrition rate of greater than 20% and results reported at between 2 and 4
view was undertaken; one has since been therefore no sensitivity analysis on at- months’ duration of treatment. Seven
published. We were unable to obtain 20 trition could be performed. A final sen- were described as randomized, double-
articles; from their titles and keywords sitivity analysis dropping the trial by blind, placebo-controlled trials. The
none appeared to be clinical trials of Moheb et al9 did not materially change random-effects pooled estimate of the
ephedrine or ephedra. these results. In our dose analysis, only rate of weight loss was 1.0 kg per month
high doses of ephedrine produced a above weight lost with placebo (95% CI,
Efficacy: Weight Loss weight loss that was statistically sig- 0.7-1.3) (Figure 2). The pooled aver-
We identified 44 controlled trials (51 nificantly greater than zero, and the age percentage weight loss in the ephed-
articles) that assessed ephedra, ephed- difference in weight loss between me-
rine, or ephedrine and other com- dium-dose trials and high-dose trials ap- *References 16, 19, 21-24, 26, 27, 29-31.
pounds used for weight loss. Of these,
18 trials were excluded from pooled Figure 1. Flow of Reviewed Literature
analysis because they had a duration of
treatment shorter than 8 weeks; an- 550 Articles Requested
452 Identified by Library Search
other 6 trials were excluded for other 64 Identified by Outreach to Expert
reasons (eg, insufficient data for meta- 34 Identified in Reference Lists

analysis; crossover design without pro-

vider outcomes at the crossover point). 20 Not Found

For the remaining 20 trials, we clas-

530 Articles Screened
sified the comparisons made into 6 cat-
egories: ephedrine vs placebo, ephed-
400 Excluded
rine and caffeine vs placebo, ephedrine 149 Topic
and caffeine vs ephedrine alone, ephed- 101 Subject
68 Population
rine vs another active weight loss phar- 36 Study Design: Descriptive
maceutical, ephedra vs placebo, ephe- 30 Study Design: Review/Meta-analysis
9 Study Design: Not RCT, CCT, Case
dra with herbs containing caffeine vs Series, or Case Report
placebo. We found no trials of ephe- 4 Not an Article
3 Duplicate of Article Already Included
dra vs another active weight loss
therapy. Details of the 19 placebo-
controlled studies are presented in 71 Case Reports or 52 Randomized Controlled Trials or Controlled
TABLE 1. Case Series Clinical Trials (59 Articles)

Ephedrine vs Placebo. We identi-

fied 5 trials containing 6 comparisons
6 Duplicate Reports 8 Athletic Performance 44 Weight Loss Trials
that assessed the effect of ephedrine vs Excluded Trials
placebo.9,22,25,28,29 All 5 were described
as randomized, placebo-controlled trials
with results reported at 3 or 4 months’ 65 Included in Adverse
Events Analysis
52 Trials Evaluated
for Adverse Events
24 Excluded From Weight
Loss Meta-analysis
duration of treatment. The random- Analysis 18 Follow-up < 8 wk
6 Other∗
effects pooled estimate of the rate of
weight loss was 0.6 kg per month above
52 Included in Adverse 20 Included in Weight
weight lost with placebo (95% confi- Events Analysis† Loss Meta-analysis
dence interval [CI], 0.2-1.0) (FIGURE 2).
The pooled average percentage weight RCT indicates randomized cointrolled trial; CCT, controlled clinical trial.
loss in the ephedrine-treated patients, *Various reasons (see “Methods” section).
†2 studies had no placebo group and therefore did not contribute to the odds ratio meta-analysis.
compared with pretreatment weight,
©2003 American Medical Association. All rights reserved. (Reprinted) JAMA, March 26, 2003—Vol 289, No. 12 1539

Downloaded From: http://jama.jamanetwork.com/ on 02/25/2013

 EPHEDRA AND EPHEDRINE FOR WEIGHT LOSS AND ATHLETIC PERFORMANCE

rine- and caffeine-treated patients, com- with placebo of 0.9, 0.9, and 1.4 kg/mo were too few trials to perform any sen-
pared with pretreatment weight, was for low, medium, and high doses, re- sitivity analysis.
11% at 4 months. A sensitivity analy- spectively) but these differences were Ephedrine vs Another Active Weight
sis of only those trials that scored 3 or not statistically significant. Neither Loss Pharmaceutical. We identified 2
greater on the Jadad scale yielded a re- graphical assessment nor statistical tests trials that compared ephedrine vs an-
sult similar to that yielded by the main revealed publication bias (P = .30 for other active weight loss therapy.26,32 Both
analysis. Another sensitivity analysis of rank test; P=.12 for regression test). took place in Denmark and did not have
only those trials that had less than 20% Ephedrine and Caffeine vs Ephed- sufficient statistical power to detect even
attrition also yielded a similar result. A rine. We identified 3 trials that in- a 30% difference between groups. One
third sensitivity analysis dropping the cluded comparisons that assessed a trial compared dexfenfluramine with a
trial by Moheb et al9 did not change the combination of ephedrine and caf- combination of ephedrine and caf-
primary analysis. Neither did sensitiv- feine vs ephedrine alone.9,22,29 All 3 had feine.32 At 15 weeks of treatment there
ity analyses that dropped trials that in- attrition rates of greater than 20%. The was no statistically significant differ-
cluded synephrine or aspirin. In our random-effects pooled estimate of ence in weight loss between groups.
dose analysis, there was a trend to- weight loss was 0.4 kg per month above The second trial compared the Elsinore
ward increased weight loss with higher weight lost with ephedrine alone (95% pill (a prescription containing ephed-
doses (weight loss greater than that lost CI, 0.02-0.7) (data not shown). There rine and caffeine) with diethylpropion

Table 1. Placebo-Controlled Trials Used in Meta-analysis

Average Weight Loss,
Sample Size, No. (SD), kg
Jadad Trial
Source Scale Score Intervention (Dose, mg) Duration Placebo Intervention Placebo Intervention
Astrup et al,16 1992 2 Ephedrine (60) and caffeine (600) 8 wk 8 8 8.4 (2.9) 10.1 (1.0)
Boozer et al,17 2002 5 Ephedrine from ma huang (86.4) 24 wk 84 83 2.6 (3.2) 5.3 (5.0)
and caffeine from kola nut (196)
Boozer et al,18 2001 5 Ephedrine from ma huang (77.4) 8 wk 32 35 0.8 (2.4) 4.0 (3.4)
and caffeine from guarana (300)
Buemann et al,19 3 Ephedrine (60) and caffeine (600) 8 wk 16 16 7.1 (2.4) 8.4 (2.4)
1994
Colker et al,20 2001 2 Ephedrine from ma huang 8 wk 12 14 0.49 (2.35) 2.56 (2.35)
(unknown dose) and Coleus
forksohlli (unknown dose)
Daly et al,21 1993 2 Ephedrine (150), caffeine (150), and 8 wk 15 14 0.7 (2.2) 2.2 (2.3)
aspirin (330)
Donikyan 4 Ephedrine from ma huang (72) and 12 wk 94 92 1.4 (2.7) 3.4 (3.1)
(unpublished data, chromium picolinate
2002) (450 µg)
Greenway et al 2 Ephedrine from ma huang (72) and 12 wk 20 20 0.8 (2.6) 3.9 (4.0)
(unpublished data) caffeine from unspecified
herb (210)
Jensen et al,22 1980 1 Ephedrine (100) 16 wk 17 24 0.5 (4.7) 7.9 (4.7)
Ephedrine (100) and caffeine (275) 23 9.4 (4.7)
Kalman et al,23 2000 3 Ephedrine (40), caffeine (400), 8 wk 14 16 2.1 (2.4) 3.1 (2.4)
synephrine (10), and aspirin (30)
Kettle et al,24 1998 0 Ephedrine (20) and caffeine (200) 6 mo 45 45 12.8 (6.7) 15.6 (7.1)
Lumholtz et al,25 1980 2 Ephedrine (120) 18 wk 63 63 4.0 (5.3) 9.5 (5.3)
Malchow-Moller 3 Ephedrine (60) and caffeine (150) 12 wk 33 49 4.1 (3.5) 8.1 (3.5)
et al,26 1981
Moheb et al,9 1998 2 Ephedrine (150) 12 wk 32 64 6.2 (3.5) 8.8 (3.5)
Ephedrine (150) and caffeine (150) 64 8.9 (3.5)
Molnar et al,27 2000 4 Ephedrine (30-60) and caffeine 20 wk 16 16 0.5 (4.3) 7.9 (6.0)
(100 for 1 wk)
Pasquali et al,28 1985 3 Ephedrine (75) 3 mo 21 19 8.7 (3.5) 8.7 (2.4)
Ephedrine (150) 22 10.2 (3.5)
Quaade et al,29 1992 3 Ephedrine (60) 24 wk 45 45 13.2 (6.6) 14.3 (5.9)
Ephedrine (60) and caffeine (600) 45 16.6 (6.8)
Roed et al,30 1980 3 Ephedrine (120) and caffeine (300) 12 wk 69 69 5.2 (3.5) 10.0 (3.5)
Van Mil and Molnar,31 1 Ephedrine (60) and caffeine (600) 20 wk 16 16 1.5 (8.1) 8.7 (5.7)
2000

1540 JAMA, March 26, 2003—Vol 289, No. 12 (Reprinted) ©2003 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ on 02/25/2013

 EPHEDRA AND EPHEDRINE FOR WEIGHT LOSS AND ATHLETIC PERFORMANCE

(Tenuate).26 At 12 weeks of treatment, comes, and is not described below.33 as oxygen consumption, time to exhaus-
there was no statistically significant dif- The remaining 7 trials were not appro- tion, or carbon dioxide production, but
ference in weight loss between groups. priate for pooled analysis because they the combination of ephedrine and caf-
Ephedra vs Placebo. We identified involved different types of exercise feine consistently demonstrated a 20%
a single trial that assessed ephedra vs (power and endurance) and different to 30% increase in performance. The
placebo (L.A. Donikyan, unpublished outcome measures, so they are dis- single trial of strength training did show
data, 2002). This trial was a random- cussed here individually. We found no an improvement in muscle endurance
ized, double-blind, parallel-group as- trials assessing the effects of herbal but only on the first of 3 repetitions. In
sessment of Metab-O-LITE (Rexall Sun- ephedra on athletic performance. the only trial to test the effects of ephed-
down, Boca Raton, Fla), which contains Six trials by Bell and colleagues34-39 as- rine and caffeine on thermal regulation,
ephedra and other compounds but caf- sessed the exercise capacity of small no increase in temperature was re-
feine or herbs that contain caffeine. The groups of healthy male participants (all ported. Nausea and vomiting were re-
duration of the trial was 3 months. This trials included 24 participants or fewer) ported in a third of the participants given
trial reported a rate of weight loss of 0.8 and are summarized in TABLE 2. These ephedrine at a dose of 1 mg/kg with 5
kg per month greater than weight lost trials reported that neither caffeine nor mg/kg of caffeine, but not in any of those
with placebo for the group receiving ephedrine alone had significant effects on given a lower dose of 0.8 mg/kg of ephed-
ephedra (95% CI, 0.4-1.2) (Figure 2). parameters of exercise performance such rine and 4 mg/kg of caffeine.
This trial had 17% attrition.
Ephedra with Herbs Containing Caf-
feine vs Placebo. We identified 4 trials Figure 2. Monthly Weight Loss in Placebo-Controlled Trials
(3 published17,18,20 and 1 unpublished
Favors Favors
[F. Greenway et al, unpublished data]) Intervention Placebo
that assessed the effect of ephedra with
Ephedrine vs Placebo
herbs containing caffeine, with results
Jensen et al,22 1980
reported at between 2 and 4 months’ Lumholtz et al,25 1980
duration of treatment. All 4 were de- Moheb et al,9 1998
scribed as randomized placebo- Pasquali et al,28 1985
Pasquali et al,28 1985
controlled trials. The pooled random- Quaade et al,29 1992
effects estimate of the rate of weight loss Combined
was 1.0 kg per month above weight lost
with placebo (95% CI, 0.6-1.3) (Fig- Ephedrine and Caffeine vs Placebo
ure 2). The pooled average percent Astrup et al,16 1992
Buemann et al,19 1994
weight loss in the ephedra-treated pa- Daly et al,21 1993
tients, compared with pretreatment Jensen et al,22 1980
weight, was 5.2% at 4 months. A sen- Kalman et al,23 2000
Kettle et al,24 1998
sitivity analysis with only those trials Malchow-Moller et al,26 1981
scoring 3 or more on the Jadad scale Moheb et al,9 1998
yielded a result similar to the main Molnar et al,27 2000
Quaade et al,29 1992
analysis. All these studies assessed me-
Roed et al,30 1980
dium doses of ephedra and therefore no Van Mil and Molnar,31 2000
analysis stratified by dose was pos- Combined
sible. Neither graphical assessment or
statistical tests revealed any evidence of Ephedra vs Placebo
Donikyan (Unpublished Data, 2002)
publication bias (P = .73 for rank test;
P=.23 for regression test).
Ephedra With Herbs Containing Caffeine vs Placebo
Boozer et al,18 2001
Efficacy: Athletic Performance
Boozer et al,17 2002
We found 8 published controlled Colker et al,20 2001
trials33-40 of the effects of synthetic Greenway et al (Unpublished Data)
Combined
ephedrine on athletic performance;
most were crossover designs and all but
1 also included caffeine. One study as- –4.0 –3.0 –2.0 –1.0 0 1.0

sessed the effect of ephedrine and ex- Weight Loss, kg

ercise training on basal metabolic rate,
did not report athletic performance out- Sizes of boxes correspond to sample sizes. Error bars indicate 95% confidence intervals.

©2003 American Medical Association. All rights reserved. (Reprinted) JAMA, March 26, 2003—Vol 289, No. 12 1541

Downloaded From: http://jama.jamanetwork.com/ on 02/25/2013

 EPHEDRA AND EPHEDRINE FOR WEIGHT LOSS AND ATHLETIC PERFORMANCE

One other trial was identified. It re- ness,” “insomnia,” “difficulty sleep- meta-analysis. The OR will slightly
ported no statistically significant im- ing,” “increased sweating,” “increased overestimate the risk ratio for these
provement in a battery of tests of physi- perspiration,” and “sweating;” (3) up- events, because they occur in 10% to
cal function including oxygen uptake, per gastrointestinal symptoms, ie, those 20% of participants. Participants in ac-
measures of endurance and power, re- described in the original clinical trials tive treatment had between 2.2 and 3.6
action time, hand-eye coordination, as “nausea,” “vomiting,” “abdominal increased odds for the adverse events
speed, and self-perceived exertion.40 pain,” “upset stomach,” “heartburn,” of psychiatric symptoms, autonomic
and “gastroesophageal reflux;” (4) pal- hyperactivity, upper gastrointestinal
Safety Assessment: pitations, ie, those symptoms de- symptoms, and heart palpitations.
Randomized Trials scribed in the original clinical trials as There was a trend toward an increase
Two of 52 trials were excluded from the “palpitations,” “irregular heartbeat,” of similar magnitude in the report of hy-
OR analysis because they did not have “loud heartbeat,” “heart pounding,” and pertension, but this increase was not
a placebo group.32,41 There were nu- “increased or stronger heartbeat;” (5) statistically significant. In our dose
merous symptoms reported as ad- tachycardia, ie, those symptoms de- analysis, there was a trend toward
verse events in the remaining 50 trials scribed in the original clinical trials as higher risk of adverse events with
of ephedra and ephedrine. We grouped “tachycardia” and “slightly elevated higher doses of ephedrine, but data were
clinically similar symptoms into 7 cat- heart rate;” (6) hypertension, ie, those sparse and these differences were not
egories: (1) psychiatric symptoms, ie, symptoms described in the original statistically significant (for example, ad-
those described in the original clinical clinical trials as “hypertension,” “in- justed ORs of autonomic hyperactiv-
trials as “euphoria,” “neurotic behav- creased systolic blood pressure,” and ity were 2.7 and 10.4 for medium- and
ior,” “agitation,” “neuropsychiatric,” “increased diastolic blood pressure;” high-dose ephedrine, respectively, but
“depressed mood,” “giddiness,” “irri- and (7) headache. the 95% CIs overlapped). It is not pos-
tability,” and “anxiety;” (2) auto- TABLE 3 presents the pooled esti- sible to estimate the degree to which
nomic hyperactivity symptoms, ie, mate of the OR for each of those ad- caffeine contributes to these results, be-
those described in the original clinical verse events for which there were cause there were too few trials of ephed-
trials as “tremor,” “twitching,” “jitteri- sufficient data to justify attempting rine alone or ephedra alone to support

Table 2. Summary of Exercise Trials by Bell and Colleagues

Compounds Received by
Source Comparison Groups Exercise Type Results
Bell et al,34 1998 Placebo Cycle ergometer trials to Ephedrine and caffeine significantly
Ephedrine (1 mg/kg) exhaustion increased time to exhaustion
Caffeine (5 mg/kg) compared with placebo. Heart rate
Ephedrine (1 mg/kg) and during exercise was significantly
caffeine (5 mg/kg) increased for ephedrine and caffeine,
caffeine groups.
Bell and Jacobs,35 1999 Placebo Canadian Forces Warrior Test Run times for ephedrine and caffeine trial
Ephedrine (75 mg) and (3.2-km run wearing 11 kg of were significantly faster than those
caffeine (375 mg) equipment) for control and placebo trials.
Bell et al,36 1999 Placebo Treadmill walking at 50% VO2 Ephedrine and caffeine did not
Ephedrine (1 mg/kg) and peak, 40°C climate, 30% significantly change tolerance times
caffeine (5 mg/kg) relative humidity when compared with placebo.
Ephedrine and caffeine did not affect
skin or rectal temperature, sweat
rate, or sensation of thermal comfort.
Bell et al,37 2000 Placebo Cycle ergometer trials to A lower dose of ephedrine and caffeine
Caffeine (5 mg/kg) and exhaustion at 85% VO2 peak resulted in ergogenic effect similar in
ephedrine (0.8 mg/kg) magnitude as reported previously
Caffeine (4 mg/kg) and with a higher dose, while fewer
ephedrine (1 mg/kg) adverse effects were reported.
Caffeine (4 mg/kg) and
ephedrine (0.8 mg/kg)
Pasternak et al,38 1999 Placebo Three supersets of leg press and Ephedrine and ephedrine and caffeine
Ephedrine (0.8 mg/kg) bench press, to exhaustion increased muscular endurance, but
Caffeine (4 mg/kg) only in the first set. Systolic blood
Ephedrine (0.8 mg/kg) and pressure was increased with
caffeine (4 mg/kg) ephedrine and with ephedrine and
caffeine.
Bell et al,39 2001 Placebo Two different cycle ergometer Ephedrine improved performance during
Ephedrine (1 mg/kg) tests, with 1 to exhaustion at Wingate test of anaerobic power.
Caffeine (5 mg/kg) 125% VO2 peak Caffeine increased time to
Ephedrine 1 (mg/kg) and exhaustion in second test.
caffeine (5 mg/kg)

1542 JAMA, March 26, 2003—Vol 289, No. 12 (Reprinted) ©2003 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ on 02/25/2013

 EPHEDRA AND EPHEDRINE FOR WEIGHT LOSS AND ATHLETIC PERFORMANCE

stratified analyses. The 1 trial of herbal Safety Assessment: Case Reports tailed review. We identified 2 deaths,
ephedra without herbs containing caf- We reviewed 71 case reports in the pub- 3 myocardial infarctions, 9 cerebrovas-
feine reported a statistically signifi- lished literature, 1820 case reports in cular accidents, 3 seizures, and 5 psy-
cant 2-fold increase in gastrointestinal the FDA MedWatch files, and 15 951 chiatric cases as sentinel events with
symptoms. cases reported to 1 manufacturer of prior ephedra consumption; and 3
If we consider all 1706 patients in the ephedra-containing dietary supple- deaths, 2 myocardial infarctions, 2 cere-
treatment groups of all 52 trials, these ments. The majority of case reports are brovascular accidents, 1 seizure, and 3
trials have at least 80% power to dis- insufficiently documented to make an psychiatric cases as sentinel events with
tinguish a serious adverse event rate of informed judgment about a relation- prior ephedrine consumption. We iden-
1.0 per thousand or higher. Thus, ship between the use of ephedrine- or tified an additional 43 and 7 cases as
though we observed no serious ad- ephedra-containing dietary supple- possible sentinel events with prior ephe-
verse events in any of these trials, the ments and the adverse event in ques- dra and ephedrine consumption, re-
relatively small aggregated sample size tion. From the unpublished reports, 284 spectively. About half of sentinel events
across the treatment groups in all trials concerned serious adverse events and occurred in persons aged 30 years or
limits the strength of our conclusions. had sufficient evidence to warrant de- younger (TABLE 4). Full details are

Table 3. Summary of Meta-analysis of Adverse Events Reporting in Controlled Trials

Adverse Events

Sample Size, No. No.

Adverse Event* No. of Studies Placebo Intervention Placebo Intervention Pooled OR (95% CI)
Psychiatric symptoms 8 273 351 16 59 3.64 (1.91-7.31)
Autonomic hyperactivity 13 365 587 39 138 3.37 (2.19-5.31)
Heart palpitations 11 386 563 18 51 2.29 (1.27-4.32)
Hypertension 5 257 305 3 7 2.19 (0.49-13.34)
Upper GI symptoms 10 432 568 46 88 2.15 (1.39-3.38)
Headache 5 123 185 8 16 1.64 (0.62-4.68)
Tachycardia 1 45 90 0 6 NR
Abbreviations: CI, confidence interval; GI, gastrointestinal; NR, not reported; OR, odds ratio.
*A study may contribute adverse events to more than 1 category.

Table 4. Summary of Adverse Events With Consumption of Ephedra or Ephedrine

Myocardial Cerebrovascular
Infarction/Other Accident/ Psychiatric
Death Cardiac Other Neurologic Seizure Symptoms

Demographics Ephedra Ephedrine Ephedra Ephedrine Ephedra Ephedrine Ephedra Ephedrine Ephedra Ephedrine
Total events, No.
Sentinel 2 3 3 2 9 2 3 1 5 3
Possible sentinel 9 3 9 1 11 2 7 0 7 1
Events by sex, No.
Women
Sentinel 1 1 0 2 5 2 3 1 2 1
Possible sentinel 3 1 4 1 8 0 5 0 4 0
Men
Sentinel 1 2 3 0 4 0 0 0 3 2
Possible sentinel 6 2 5 0 3 2 2 0 3 1
Events by age, No.
13-30 y
Sentinel 2 2 2 1 3 2 2 0 3 2
Possible sentinel 5 0 1 1 2 1 3 0 5 0
31-50 y
Sentinel 0 1 1 1 5 0 1 1 2 0
Possible sentinel 4 2 7 0 5 1 3 0 1 1
51-70 y
Sentinel 0 0 0 0 1 0 0 0 0 1
Possible sentinel 0 1 1 0 4 0 1 0 1 0

©2003 American Medical Association. All rights reserved. (Reprinted) JAMA, March 26, 2003—Vol 289, No. 12 1543

Downloaded From: http://jama.jamanetwork.com/ on 02/25/2013

 EPHEDRA AND EPHEDRINE FOR WEIGHT LOSS AND ATHLETIC PERFORMANCE

available at http://www.ahrq.gov/clinic psychiatric symptoms, autonomic symp- mal meta-analysis at all. Fourth, the
/evrptfiles.html#ephedra. toms, upper gastrointestinal symp- generalizability of the results to the use
toms, and heart palpitations. It is not pos- of ephedra or ephedrine in a general
COMMENT sible to separate out the effect caffeine population may not be valid. The
We found sufficient evidence to con- may contribute to these events. As most weight loss trials frequently involved
clude that the short-term use of ephed- ephedra-containing dietary supple- medical screening to detect preexist-
rine at high doses, or of ephedrine and ments also contain herbs with caffeine, ing conditions, such as heart disease,
caffeine, ephedra, and ephedra with this issue may be moot. We also found that may predispose subjects to an in-
herbs containing caffeine, promotes a number of case reports of serious ad- creased risk of adverse events. Such pa-
weight loss in selected patient popula- verse events that occurred in young tients were then excluded from the
tions. These results satisfy, with 1 ex- adults without apparent causes. Similar trials. Whether patients taking ephe-
ception, published criteria42 for the observations have been made by oth- dra or ephedrine without such medi-
evaluation of weight loss products, as ers2 and raise the possibility that there cal screening have a similar risk of ad-
they are randomized, blinded, placebo- may be a causal relationship between verse events is unknown. Finally, the
controlled, and resulted in a weight loss ephedrine or ephedra use and rare seri- weight loss trials as a group had lim-
of at least 5% of pretreatment weight. ous adverse events. A recent case- ited duration of treatment times and
The criterion not met is duration of control study reported an adjusted OR thus we cannot draw conclusions about
treatment, as all trials but 3 reported less of 3.59 (95% CI, 0.70-18.35) for hem- the association between ephedra and
than 6 months duration of treatment. orrhagic stroke in persons taking more weight loss over longer and more clini-
Trials assessing 1 year of treatment are than 32 mg/d of ephedra. We note that cally relevant intervals than about 4
considered desirable, as are trials as- we found no studies of ephedra where months, nor about what happens to
sessing what happens to weight after the participants consumed less than 32 mg/d weight after the ephedra or ephedrine
weight loss product is discontinued. of ephedra.53 In addition, another re- is stopped.
We can also conclude that caffeine cent report linked ephedra use to an in- With these limitations in mind we
adds additional efficacy to ephedrine in creased risk of adverse events reported found sufficient evidence to conclude
promoting weight loss, and the effects to poison control centers, relative to other that ephedrine- and ephedra-contain-
of ephedrine and caffeine and ephedra herbal products.54 ing dietary supplements have modest
with or without herbs containing caf- Our study has several limitations. short-term benefits with respect to
feine are approximately equivalent, each Many of the trials had methods prob- weight loss and have harms in terms of
resulting in about 0.9 kg per month of lems. However, not all trials did, and a 2- to 3-fold increase in psychiatric
weight loss above that lost with pla- our sensitivity analyses of the higher- symptoms, autonomic symptoms, up-
cebo, extending out to 4 months. To help quality trials supported our main analy- per gastrointestinal symptoms, and
put these data in context, placebo- sis with respect to ephedrine and caf- heart palpitations. More serious ad-
controlled studies of some FDA- feine and with respect to ephedra. Only verse effects from ephedra use cannot
approved weight loss pharmacothera- for ephedrine alone did we find evi- be excluded at a rate less than 1.0 per
pies have reported losses 2.7 to 4.5 kg dence that higher-quality trials re- thousand, and case reports raise the
greater than losses with placebo over 6 ported smaller effects, as has been re- possibility that a causal relationship
to 12 months for patients taking sibutra- ported in meta-analyses of other clinical with serious adverse events may exist.
mine43-46 or orlistat47-51; or 7.2 kg greater questions.11 Second, while we did not We did not find sufficient evidence to
than with placebo at 9 months for pa- find any evidence of publication bias, support the use of ephedra for enhanc-
tients taking phentermine.52 There are this does not mean that it does not ing athletic performance.
no data regarding weight loss beyond 6 exist. There is no way of ever knowing
Author Affiliations: Southern California Evidence-
months’ use of ephedrine or ephedra. if all trials conducted have been iden- based Practice Center–RAND, Santa Monica, Calif
Regarding athletic performance, we tified. We made reasonable efforts to (Drs Shekelle, Morton, Mojica, and Mss Maglione,
Suttorp, Rhodes, and Jungvig); Cedars-Sinai Medical
were unable to identify any controlled identify and did identify some unpub- Center, Los Angeles, Calif (Dr Hardy); Greater Los An-
trials of ephedra. The few identified lished data. Third, we did not observe geles Veterans Affairs Healthcare System (Dr Shek-
elle); and Department of Family Medicine, University
trials of ephedrine did not study it as significant heterogeneity among the of Southern California, Los Angeles (Dr Gagné).
used by the general population, that is, weight loss trials but acknowledge that Author Contributions: Study concept and design:
repeated use. Therefore, the effect of the ␹2 test of heterogeneity is under- Shekelle, Hardy, Morton, Maglione, Suttorp.
Acquisition of data: Shekelle, Hardy, Morton,
ephedra or ephedrine as it is used to powered. We used a random-effects ap- Maglione, Mojica, Suttorp, Rhodes, Gagné.
promote enhanced athletic perfor- proach and sensitivity analysis to try to Analysis and interpretation of data: Shekelle, Hardy,
Morton, Maglione, Suttorp, Jungvig.
mance is unknown. incorporate and understand possible Drafting of the manuscript: Shekelle, Morton, Suttorp,
We found sufficient evidence to con- heterogeneity. The heterogeneity Rhodes, Jungvig.
Critical revision of the manuscript for important in-
clude that ephedrine and ephedra are among the athletic performance trials tellectual content: Shekelle, Hardy, Morton, Maglione,
associated with 2 to 3 times the risk of prevented us from conducting a for- Mojica, Suttorp, Jungvig, Gagné.

1544 JAMA, March 26, 2003—Vol 289, No. 12 (Reprinted) ©2003 American Medical Association. All rights reserved.

Downloaded From: http://jama.jamanetwork.com/ on 02/25/2013

 EPHEDRA AND EPHEDRINE FOR WEIGHT LOSS AND ATHLETIC PERFORMANCE

Statistical expertise: Shekelle, Morton, Suttorp. RAND, Santa Monica, Calif, under Agency for Health- Acknowledgment: We thank Birgit Danila, MA, for
Obtained funding: Shekelle, Hardy, Morton, Maglione. care Research and Quality contract 290-97-0001. translation of Danish trials, and Elizabeth Roth for pro-
Administrative, technical, or material support: Shekelle, Disclaimer: The authors of this article are respon- gramming advice. We also thank the members of our
Maglione, Mojica, Suttorp, Rhodes, Jungvig. sible for its contents. No statement in this article should Technical Expert Panel who advised us on the project
Study supervision: Shekelle, Morton, Maglione, Mojica. be construed as an official position of the Agency for and reviewed our findings. Dr Shekelle was a Senior
Funding/Support: This research was performed by the Healthcare Research and Quality or of the US De- Research Associate of the Veterans Affairs Health Ser-
Southern California Evidence-based Practice Center– partment of Health and Human Services. vices Research and Development Service.

REFERENCES
1. Abrahamson A. Wheeler used supplement. Los lipids and lipoproteins during a 4.2 MJ/day diet. Int J preserves the ergogenic effect. Aviat Space Environ
Angeles Times. August 11, 2001; Sports section: Obes Relat Metab Disord. 1994;18:329-332. Med. 2000;71:415-419.
D1, D9. 20. Colker CM, Swain MA, Lynch L. A pilot study 38. Pasternak H, Jacobs I, Bell D. Effects of ingesting
2. Haller CA, Benowitz NL. Adverse cardiovascular and evaluating the effects of an ephedrine and forskolin- caffeine and ephedrine on muscular endurance. Can
central nervous system events associated with di- based product on body weight and body composi- J Appl Physiol. 1999;24:471.
etary supplements containing ephedra alkaloids. N Engl tion in overweight, healthy women [abstract]. J Am 39. Bell DG, Jacobs I, Ellerington K. Effect of caffeine
J Med. 2000;343:1833-1838. Coll Nutr. 2001;20:98. and ephedrine ingestion on anaerobic exercise perfor-
3. Carey B. Risks of ephedra usage in spotlight. Los 21. Daly PA, Krieger DR, Dulloo AG, Young JB, Lands- mance. Med Sci Sports Exerc. 2001;33:1399-1403.
Angeles Times. August 27, 2001; Health section: berg L. Ephedrine, caffeine and aspirin. Int J Obes Relat 40. Sidney KH, Lefcoe NM. The effects of ephedrine
D11, D13. Metab Disord. 1993;17(suppl 1):S73-S78. on the physiological and psychological responses to
4. Public Citizen. Public Citizen Calls On FDA To Ban 22. Jensen KB, Dano P, Draeby N, Hansen SH, Kans- submaximal and maximal exercise in man. Med Sci
Dietary Supplements Containing Ephedra. Septem- trup J. Elsinore tablets and ephedrine as slimming agents Sports. 1977;9:95-99.
ber 5, 2001. Available at: http://www.citizen.org [in Danish]. Ugeskr Laeger. 1980;142:1499-1501. 41. Toubro S, Astrup A. Randomised comparison of
/pressroom/release.cfm?ID = 654. Accessibility veri- 23. Kalman DS, Colker CM, Shi Q, Swain MA. Ef- diets for maintaining obese subjects’ weight after ma-
fied February 17, 2003. fects of a weight-loss aid in healthy overweight adults. jor weight loss. BMJ. 1997;314:29-34.
5. Marsa L. A tug of war in a larger battle: ephedra is Curr Ther Res Clin Exp. 2000;61:199-205. 42. Bray GA. Evaluation of drugs for treating obe-
now under intense scrutiny. Los Angeles Times. Sep- 24. Kettle R, Toubro S, Astrup A. Ephedrine/ sity. Obes Res. 1995;3(suppl 4):425S-434S.
tember 2, 2002; Health section:D10, D12. caffeine enhances abdominal fat loss in females. Int J 43. Fujioka K, Seaton TB, Rowe E, et al. Weight loss with
6. Department of Health and Human Services. HHS Obes Relat Metab Disord. 1998;22(suppl 3):S264. sibutramine improves glycaemic control and other meta-
announces plans to study ephedra. HHS News. June 25. Lumholtz IB, Thorsteinsson B, Wamberg T, et al. bolic parameters in obese patients with type 2 diabetes
14, 2002. Available at: http://www.hhs.gov/news Ephedrine in the treatment of obesity [in Danish]. mellitus. Diabetes Obes Metab. 2000;2:175-187.
/press/2002press/20020614.html. Accessibility veri- Ugeskr Laeger. 1980;142:1487-1490. 44. Gokcel A, Karakose H, Ertorer EM, Tanaci N, Tu-
fied February 17, 2003. 26. Malchow-Moller A, Larsen S, Hey H, Stokholm KH, tuncu NB, Guvener N. Effects of sibutramine in obese
7. DerSimonian R, Laird N. Meta-analysis in clinical Juhl E, Quaade F. Ephedrine as an anorectic: the story female subjects with type 2 diabetes and poor blood glu-
trials. Control Clin Trials. 1986;7:177-188. of the “Elsinore pill.” Int J Obes. 1981;5:183-187. cose control. Diabetes Care. 2001;24:1957-1960.
8. Berkey CS, Hoaglin DC, Mosteller F, Colditz GA. 27. Molnar D, Torok K, Erhardt E, Jeges S. Safety and 45. Smith IG, Goulder MA. Randomized placebo-
A random-effects regression model for meta- efficacy of treatment with an ephedrine/caffeine mix- controlled trial of long-term treatment with sibutra-
analysis. Stat Med. 1995;14:395-411. ture. Int J Obes Relat Metab Disord. 2000;24:1573- mine in mild to moderate obesity. J Fam Pract. 2001;
9. Moheb MA, Geissler CA, Lancer K. Effect of ephed- 1578. 50:505-512.
rine, caffeine, and aspirin, in combinations of weight 28. Pasquali R, Baraldi G, Cesari MP, et al. A con- 46. Wirth A, Krause J. Long-term weight loss with
loss in obese women [abstract]. Int J Obes Relat Metab trolled trial using ephedrine in the treatment of obe- sibutramine. JAMA. 2001;286:1331-1339.
Disord. 1998;22(suppl 3):S264. sity. Int J Obes. 1985;9:93-98. 47. Van Gaal LF, Broom JI, Enzi G, Toplak H, for the
10. Jadad AR, Moore RA, Carroll D, et al. Assessing 29. Quaade F, Astrup A, Breum L, Toubro S, Hein P. Orlistat Dose-Ranging Study Group. Efficacy and tol-
the quality of reports of randomized clinical trials: The effect of an ephedrine/caffeine combination as a erability of orlistat in the treatment of obesity. Eur J
is blinding necessary? Control Clin Trials. 1996;17: supplement to a weight-reducing diet [in Danish]. Clin Pharmacol. 1998;54:125-132.
1-12. Ugeskr Laeger. 1992;154:1258-1263. 48. Hill JO, Hauptman J, Anderson JW, et al. Orli-
11. Moher D, Pham B, Jones A, et al. Does quality of 30. Roed P, Hansen PW, Bidstrup B, Kaern M, Helles stat, a lipase inhibitor, for weight maintenance after
reports of randomised trials affect estimates of inter- A, Petersen KP. Elsinore banting tablets [in Danish]. conventional dieting: a 1-y study. Am J Clin Nutr. 1999;
vention efficacy reported in meta-analyses? Lancet. Ugeskr Laeger. 1980;142:1491-1495. 69:1108-1116.
1998;352:609-613. 31. Van Mil E, Molnar D. Drug treatment in obese 49. Karhunen L, Franssila-Kallunki A, Rissanen P, et
12. Begg CB, Mazumdar M. Operating characteris- adolescents [abstract]. Int J Obes. 2000;24(suppl 1): al. Effect of orlistat treatment on body composition
tics of a rank correlation test for publication bias. Bio- S184. and resting energy expenditure during a two-year
metrics. 1994;50:1088-1101. 32. Breum L, Pedersen JK, Ahlstrom F, Frimodt-Moller weight-reduction programme in obese Finns. Int J Obes
13. Egger M, Davey Smith G, Schneider M, Minder J. Comparison of an ephedrine/caffeine combination and Relat Metab Disord. 2000;24:1567-1572.
C. Bias in meta-analysis detected by a simple, graphi- dexfenfluramine in the treatment of obesity. Int J Obes 50. Micic D, Ivkovic-Lazar T, Dragojevic R, et al. Or-
cal test. BMJ. 1997;315:629-634. Relat Metab Disord. 1994;18:99-103. listat, a gastrointestinal lipase inhibitor, in therapy of
14. Stata [computer program]. Release 7.0. College 33. Oksbjerg N, Meyer T, Hvid-Jacobsen K. Aerobic obesity with concomitant hyperlipidemia [in Serbo-
Station, Tex: Stata Corp; 2001. training increases the effect of ephedrine on energy Croatian]. Med Pregl. 1999;52:323-333.
15. StatXact 4 [computer program]. Version 4.0. Cam- expenditure [in Danish]. Ugeskr Laeger. 1986;148: 51. Muls E, Kolanowski J, Scheen A, Van Gaal L. The
bridge, Mass: Cytel Software Corp; 2000. 2021-2024. effects of orlistat on weight and on serum lipids in obese
16. Astrup A, Buemann B, Christensen NJ, et al. The 34. Bell DG, Jacobs I, Zamecnik J. Effects of caffeine, patients with hypercholesterolemia. Int J Obes Relat
effect of ephedrine/caffeine mixture on energy ex- ephedrine and their combination on time to exhaus- Metab Disord. 2001;25:1713-1721.
penditure and body composition in obese women. Me- tion during high-intensity exercise. Eur J Appl Physiol 52. Munro JF, MacCuish AC, Wilson EM, Duncan LJ.
tabolism. 1992;41:686-688. Occup Physiol. 1998;77:427-433. Comparison of continuous and intermittent anorec-
17. Boozer CN, Daly PA, Homel P, et al. Herbal ephe- 35. Bell DG, Jacobs I. Combined caffeine and ephedrine tic therapy in obesity. BMJ. 1968;1:352.
dra/caffeine for weight loss. Int J Obes Relat Metab ingestion improves run times of Canadian Forces War- 53. Morgenstern LB, Viscoli CM, Kernan WN, et al.
Disord. 2002;26:593-604. rior Test. Aviat Space Environ Med. 1999;70:325-329. Use of ephedra-containing products and risk for hem-
18. Boozer CN, Nasser JA, Heymsfield SB, Wang V, 36. Bell DG, Jacobs I, McLellan TM, Miyazaki M, Sabis- orrhagic stroke. Neurology. 2003;60:132-135.
Chen G, Solomon JL. An herbal supplement contain- ton CM. Thermal regulation in the heat during exer- 54. Bent S, Tiedt TN, Odden MC, Shlipak MG. The
ing Ma Huang-Guarana for weight loss. Int J Obes cise after caffeine and ephedrine ingestion. Aviat Space relative safety of ephedra compared with other herbal
Relat Metab Disord. 2001;25:316-324. Environ Med. 1999;70:583-588. products. Available at: http://www.acponline.org
19. Buemann B, Marckmann P, Christensen NJ, As- 37. Bell DG, Jacobs I, McLellan TM, Zamecnik J. Re- /journals/annals/ephedra.htm?hp. Accessibility veri-
trup A. The effect of ephedrine plus caffeine on plasma ducing the dose of combined caffeine and ephedrine fied February 17, 2003.

©2003 American Medical Association. All rights reserved. (Reprinted) JAMA, March 26, 2003—Vol 289, No. 12 1545

DownloadedViewFrom:
publicationhttp://jama.jamanetwork.com/
stats on 02/25/2013