1. Adding Heparin to Aspirin Reduces the
Incidence of Myocardial Infarction and
Death in Patients With Unstable Angina
A Meta-analysis
Allison Oler, MD; Mary A. Whooley, MD; Jacqueline Oler, PhD; Deborah Grady, MD, MPH
Objective.p=m-Toestimate the risk of myocardial infarction (MI) and death in pa- lows time for endogenous fibrinolysis to
tients with unstable angina who are treated with aspirin plus heparin compared with occur. In theory, adding heparin to aspi¬
patients treated with aspirin alone. rin should reduce intracoronary obstruc¬
Data Sources.p=m-Studieswere retrieved using MEDLINE, bibliographies, and tion, improve coronary blood flow, reduce
consultation with experts. myocardial ischemia, and ultimately de¬
crease cardiac morbidity and mortality in
Study Selection.p=m-Onlypublished trials that enrolled patients with unstable an- patients with unstable angina.9 Several
gina, randomized participants to aspirin plus heparin vs aspirin alone, and reported randomized clinical trials have demon¬
incidence of myocardial infarction or death were included in the meta-analysis. strated a trend toward reduced risk of
Data Extraction.p=m-Patientoutcomes including MI or death, recurrent ischemic death or nonfatal myocardial infarction in
pain, and major bleeding during randomized treatment; revascularization proce- patients with unstable angina treated with
dures after randomization; and MI or death during the 2 to 12 weeks following ran- aspirin plus intravenous heparin com¬
domization were extracted by 2 authors, 1 of whom was blinded to the journal, in- pared with patients treated with aspirin
stitution, and author of each study. alone.7·10"14 However, it has not been es¬
Data Synthesis.p=m-Sixrandomized trials were included. The overall summary tablished definitively that the combina¬
relative risk (RR) of MI or death during randomized treatment was 0.67 (95% con- tion of aspirin plus heparin is superior to
fidence interval [CI], 0.44-1.02) in patients with unstable angina treated with aspirin aspirin alone. We performed a meta-analy¬
sis of published randomized trials to de¬
plus heparin compared with those treated with aspirin alone. The summary RRs for termine whether treatment with intra¬
secondary endpoints in patients treated with aspirin plus heparin compared with venous heparin and aspirin is more
those treated with aspirin alone were 0.68 (95% CI, 0.40-1.17) for recurrent ische- effective than treatment with aspirin
mic pain; 0.82 (95% CI, 0.56-1.20) for MI or death 2 to 12 weeks following random- alone in preventing MI or death in pa¬
ization; 1.03 (95% CI, 0.74-1.43) for revascularization; and 1.99 (95% CI, 0.52-7.65) tients with unstable angina.
for major bleeding. We found no statistically significant heterogeneity among indi-
vidual study findings. METHODS
Conclusions.p=m-Ourfindings are consistent with a 33% reduction in risk of MI or Literature Review
death in patients with unstable angina treated with aspirin plus heparin compared We performed a literature search us¬
with those treated with aspirin alone. The bulk of evidence suggests that most pa-
tients with unstable angina should be treated with both heparin and aspirin.
ing the MEDLINE database (January
1966 to September 1995) with the key¬
JAMA. 1996;276:811-815 words "aspirin," "heparin," and "unstable
angina." The search was not restricted to
UNSTABLE ANGINA, ranging from of patients admitted to the hospital with citations in the English-language litera¬
progressive angina to angina at rest, unstable angina progress to myocardial ture. In addition, a manual search was
results from intracoronary plaque dis¬ infarction (MI) within 2 weeks of diag¬ done using reference lists from identified
ruption causing increased stenosis and, nosis.2·3 One-year mortality of patients articles and consultation with experts.
in some cases, intermittent thrombosis.1 with unstable angina ranges from 5% to Studies included in the meta-analysis
Prospective studies have found that 12% 14% with approximately half of these met the following criteria: (1) a random¬
deaths occurring within 4 weeks of di¬ ized clinical trial; (2) eligible participants
agnosis.4 In patients with unstable an¬ were admitted to the hospital with the
From the Departments of Medicine (Drs A. Oler,
Whooley, and Grady) and Epidemiology and Biostatis- gina, aspirin reduces the risk of throm¬ diagnosis of unstable angina or non-Q-
tics (Dr Grady), University of California, San Francisco, bosis by inhibiting platelet aggregation wave myocardial infarction; (3) partici¬
School of Medicine; the General Internal Medicine and decreases the risk of cardiac death pants were assigned either to intrave¬
Section, San Francisco Veterans Affairs Medical Cen- or nonfatal MI by 30% to 51% 7 nous heparin and aspirin or to aspirin
ter (Drs Whooley and Grady); and the Department of
Quantitative Methods, Drexel University, Philadelphia, Heparin binds to antithrombin III and alone; and (4) the incidence of myocar¬
Pa (Dr J. Oler). induces a conformational change that re¬ dial infarction (prolonged chest pain as¬
Reprints: Deborah Grady, MD, MPH, General Inter- sults in rapid inhibition of thrombin.8 This sociated with Q waves or persistent ST
nal Medicine Section, San Francisco Veterans Affairs
Medical Center, 111A1, 4150 Clement St, San Fran- inhibition of thrombin prevents propaga¬ changes on electrocardiogram and/or a
cisco, CA 94121. tion of an established thrombus and al- 2-fold increase over baseline creatine
Downloaded from www.jama.com by guest on April 16, 2010

2. kinase levels with elevated MB fractions) Table 1.—Characteristics of 6 Randomized Trials of Aspirin Plus Heparin vs Aspirin Alone to Prevent
or death while on randomized treatment Myocardial Infarction and Death in Patients Admitted to the Hospital With Unstable Angina
was reported. Clinical trials were in¬
Goal Partial
cluded regardless of whether they in¬ Aspirin Thromboplastin Duration of
Source Blinding Dose, mg Time Heparin Therapy, d
cluded a heparin placebo in the aspirin
only group or the patients had been tak¬ Theroux et al, 1988'° Participants and 325 twice daily 1.5-2 X normal
investigators
ing aspirin prior to hospital admission. RISC Group, 19907 None 75 daily Not stated
Data Extraction Cohen et al, 1990" None 80/325 dally*
Cohen et al, 1994" Participants 162.5 daily 2 x normal 3-4
Two of us (A.O. and M.A.W.) inde¬
Holdright et al, 1994" Participants 150 daily 1.5-2 x normal
pendently reviewed each study that met Gurfinkel et al, 1995" Participants and 200 daily 2 normal 5-7
the inclusion criteria. One of us ab¬ investigators
stracted data in an unblinded fashion;
the other was blinded to journal, year of *Aspirin dose was 80 mg/d in the heparin plus aspirin group and 325 mg/d in the aspirin only group.
publication, authors, and institution. We
evaluated each study with regard to pa¬
tient selection, blinding, and adequacy measure, or estimated overall RR, was RESULTS
of randomization, and recorded the dos¬ calculated using the DerSimonian and
Study Selection
age and duration of each treatment. Dis¬ Laird model which uses a random effects
crepant findings between reviewers model to incorporate variance between An initial search using MEDLINE, ref¬
were settled by discussion. study findings in a weighted average of erence review, and expert consultation
Heparin was given as a continuous in¬ rate ratios.18·19 Approximate 95% CIs were yielded 135 citations. Of these, 19 were
fusion in 5 of 6 studies1044 with the rate obtained on the natural log transforma¬ randomized trials.7·9"14·22^3 Amongthese tri¬
adjusted to achieve a goal partial throm- tion scale and the limits reexpressed us¬ als, we found 8 that enrolled participants
boplastin time (PTT) of 1.5 to 2 times ing the natural antilog transformation. A with unstable angina, randomized par¬
normal (Table 1). In the sixth study, hep¬ 2 statistic summing the squared devia¬ ticipants to aspirin or heparin plus aspi¬
arin was administered as intermittent in¬ tion of each ofthe study natural log trans¬ rin, and reported the risk of MI or death
travenous boluses and a goal PTT was not formations ofthe OR from their weighted during randomized treatment.7·9"14·80 One
stated.7 The duration of heparin therapy average was used to test whether a sum¬ of the 8 trials was excluded because we
was 2 to 7 days in all studies. Aspirin was mary OR was appropriate for each end- could not determine whether all patients
given orally in the doses shown in Table point. These calculations were substan¬ were treated with aspirin.30 Another was
1 and continued indefinitely. tiated using the log linear fits discussed excluded because only a portion of the
Incidence of MI or death during ran¬ next, and obtained using BMDP statis¬ patients in the study received aspirin.9
domized treatment was abstracted as the tical software.20 Thus, 6 trials with a total of 1353 patients
primary outcome. In addition, data were The primary endpoint in this meta- met all inclusion criteria for the meta-
abstracted on incidence of recurrent is¬ analysis was MI or death while on ran¬ analysis.7·10"14 Characteristics of these tri¬
chemie pain (anginal chest pain with is¬ domized treatment. Relationships among als are shown in Table 1.
chemie ST-T changes on electrocardio¬ the 3 dichotomous factors—myocardial All 6 trials excluded participants who
gram) and major bleeding (bleeding infarction or death, treatment regimen, had evolving Q-wave MI on admission;
requiring transfusion or fall in hemoglo¬ and study—were explored further by fit¬ coronary artery bypass grafting within
bin by at least 20 g/L) during randomized ting a hierarchical log-linear model to the 12 months prior to admission; a contra¬
treatment. We also abstracted data on cell counts.21 This approach models the indication to aspirin or anticoagulation;
MI or death between discontinuation of natural log cell frequencies as linear com¬ or had already been anticoagulated at ad¬
randomized treatment and 12 weeks fol¬ binations of main effects, and second-or¬ mission. Unstable angina was diagnosed
lowing randomization, and revasculariza¬ der and third-order interactions. A like¬ using the following criteria: (1) recent on¬
tion procedures (angioplasty or surgical lihood ratio 2 statistic was used to test set (less than 1 month) of prolonged or
bypass) during the 12 weeks following the statistical significance of the third- recurrent chest pain suggestive of myo¬
randomization. order interaction parameter, effectively cardial ischemia7·11"14; (2) pain of increas¬
a test for heterogeneity in the ORs across ing severity, at rest or with minimal ef¬
Statistical Analysis studies. Likelihood ratio x2s were exam¬ fort7·10"14; and (3) last episode of pain
Rate ratios from each study were used ined to identify the most parsimonious occurring within 48 hours of admission.10"14
as measures of effect in these analyses. "best" fit to the cell counts. For these Patients thought to be having non-Q-
The total numbers of patient outcomes data, the likelihood ratio 2 statistic pro¬ wave MI on admission were included un¬
in both the aspirin and aspirin plus hep¬ vided a test of the statistical significance der the definition of unstable angina.
arin groups were recorded in 2 X 2 tables. of the second-order interaction param¬
To improve bias and precision proper¬ eter, which measured the variation in MI Summary Effect on Ml and Death
ties, 0.5 was added to every cell in any and death rate across treatment regimen, The findings of each of the 6 trials
table containing a zero.15 Relative risks controlling for study. demonstrated a trend toward improved
(RRs) with 95% confidence intervals We examined potential for publica¬ outcome using aspirin plus heparin com¬
(CIs) were calculated individually for tion bias using the correlation between pared with aspirin alone, but none of the
each study.16 number of subjects and RR in each study. findings reached statistical significance
The summary measure of effect in the If small studies with negative results (Table 2). The incidence of MI or death
meta-analysis for each endpoint is the were less likely to be published, then during heparin therapy was 7.9% (55/
odds ratio (OR).17 Since the sample cross- the correlation between number of sub¬ 698) in participants treated with aspirin
product has a highly skewed distribution, jects and RR would be high. If publi¬ plus heparin, and 10.4% (68/655) in par¬
its natural log transformation is appro¬ cation bias does not exist, then no sig¬ ticipants treated with aspirin alone. The
priate for purposes of estimation and hy¬ nificant correlation between number of summary RR of MI or death during ran¬
pothesis testing.15 The summary effect subjects and RR would be evident. domized treatment was 0.67 (95% CI,
Downloaded from www.jama.com by guest on April 16, 2010

3. Table 2.—Incidence of Myocardial Infarction or Death and Relative Risk of Myocardial Infarction or Death
During Treatment With Heparin From 6 Randomized Trials in Patients Admitted to the Hospital With Favors
Unstable Angina Heparin Favors
Source Plus Aspirin Aspirin
Myocardial Infarction or Death, No.* (%)
Plus RR Therouxetal, 198810
Source Aspirin Aspirin Heparin (95% Cl)t
Theroux et al, 19 4/121 (3) 2/122 (2) 0.50 (0.18-2.66) RISC Group, 19907
RISC Group, 19907 7/189(4) 3/210(1) 0.39(0.18-1.47) Cohen et al, 199011
Cohen et al, 1990" 1/32 (3) 0/37 (0) 0.29 (0.06-6.87) Cohen et al, 199412
Cohen et al, 19941: 9/109(8) 4/105(4) 0.46(0.24-1.45) Holdright et al, 199413
Holdright et al, 1994'= 40/131 (31) 42/154(27) 0.89(0.66-1.29) Gurfinkel et al, 199514
Gurfinkel et al, 1995" 7/73(10) 4/70 (6) 0.60 (0.29-1.95)
55/698 (8) 0.67 (0.44-1.02) Summary Relative Risk
Summary 68/655(10)
~r ™f
"Number of patients who had a myocardial infarction or died during randomized treatment/total number 0.01 0.1 1 10
of participants assigned to the treatment group.
¡ RR indicates relative risk; and CI, confidence interval for myocardial Infarction or death in the aspirin plus heparin Relative Risk
compared with the aspirin group.
^Summary RR estimate and 95% CI from meta-analysis.
Relative risk of myocardial infarction or death dur¬
0.44-1.02; P=.06) in patients treated with RR for recurrent ischemie pain was 0.68 ing hospitalization.
aspirin plus heparin compared with those (95% CI, 0.4-1.17; for test of heteroge¬
treated with aspirin alone (Figure 1). neity, P=.08) in patients treated with but none of the trials was large enough
The result of a test for heterogeneity aspirin plus heparin compared with those to demonstrate a statistically significant
was not statistically significant (P=.78). treated with aspirin alone. benefit. Combining data from these 6
Incidence of MI or death from dis¬ Five studies10"14 reported the number trials using formal meta-analytic tech¬
continuation of randomized therapy to of patients in each treatment group who niques resulted in a summary RR esti¬
12 weeks after randomization was re¬ underwent coronary artery bypassgraft¬ mate of 0.67 (95% CI, 0.44-1.02), sug¬
ported in 4 studies (Table 3).7·10·12·13 Based ing or percutaneous transluminal coro¬ gesting a 33% reduction in MI or death
on data from these 4 studies, the risk of nary angioplasty within 12 weeks after during heparin therapy in patients
late MI or death was 12.4% (73/591) in randomization (Table 3). Of participants treated with heparin plus aspirin com¬
participants treated with aspirin plus treated with aspirin plus heparin, 23.8% pared with patients treated with aspirin
heparin and 13.8% (76/550) in partici¬ (116/488) underwent revascularization, alone. This result closely approached sta¬
pants treated with aspirin alone. The compared with 23.4% (109/466) of those tistical significance.
summary RR for MI or death from dis¬ treated with aspirin alone. The summary In 1 trial,13 31% (40/131) of participants
continuation of randomized therapy to RR for revascularization procedures was in the aspirin group and 27% (42/154) of
12 weeks after randomization was 0.82 1.03 (95% CI, 0.84-1.43; for test of het¬ participants in the aspirin plus heparin
(95% CI, 0.56-1.20; for test of heteroge¬ erogeneity, P=.41) in patients treated group developed MI or died during ran¬
neity, =.76) in patients treated with with aspirin plus heparin compared with domized treatment. This incidence of MI
aspirin plus heparin compared with those those treated with aspirin alone. or death was markedly higher than the
treated with aspirin alone. Of note, 1 of Overall, 0.4% (3/655) of patients treated average incidence of 5.3% (28/254) in the
these 4 studies continued anticoagula- with aspirin alone and 1.5% (10/698) of aspirin groups and 2.4% (13/544) in the
tion therapy with warfarin in the hep¬ patients treated with aspirin plus hep¬ aspirin plus heparin groups reported in
arin plus aspirin group after heparin was arin developed major bleeding during the 5 other trials. The explanation for this
discontinued.12 Removing the findings randomized treatment. Based on data increased risk of death or MI is unclear.
of this trial from the summary results from all 6 trials, the summary RR for It is possible that participants in this trial
did not significantly alter the summary major bleeding was 1.89 (95% CI, 0.66- were sicker at baseline, or that the in¬
RR estimate. 5.38; for test of heterogeneity, =.68) vestigators used a more sensitive defini¬
in patients treated with aspirin plus hep¬ tion of MI. When we performed a post
Summary Effects on Ischemie Pain, arin compared with those treated with hoc meta-analysis excluding this trial, the
Revascularization Procedures, and summary RR was 0.45 (95% CI, 0.23-0.89),
aspirin alone (Table 3).
Major Bleeding suggesting a 55% reduction in MI or death
Recurrence of ischemie pain was re¬ COMMENT in patients treated with aspirin plus hep¬
ported in 5 of the 6 studies (Table 3).1014 Unstable angina is an important medi¬ arin compared with patients treated with
The within study effect of heparin plus cal problem associated with a high in¬ aspirin alone. This result was statistically
aspirin compared with aspirin alone for cidence of MI and death. Many physi¬ significant, but because it was a post hoc
recurrent ischemie pain was heterog¬ cians routinely treat patients with analysis, the findings should be viewed
eneous across studies (for test of het¬ unstable angina with both aspirin and with caution.
erogeneity, P=.02). Heterogeneity was heparin, but adding heparin to aspirin Depending on whether the 1 trial with
primarily due to 1 trial11 that used 80 mg has not been demonstrated to reduce a markedly high incidence of MI or death
of aspirin daily in the group receiving the incidence of MI and death compared is excluded, the average risk of MI or
heparin and 325 mg of aspirin daily in with treatment with aspirin alone. death in the combined studies is either
the group not receiving heparin (Table We found 6 randomized trials of treat¬ 5.3% or 10.4% for patients treated with
1). With this study excluded from the ment with aspirin plus intravenous hep¬ aspirin alone. If we assume that adding
analysis, the incidence of recurrent is¬ arin compared with treatment with as¬ heparin to aspirin reduces the risk of MI
chemie pain during randomized treat¬ pirin alone in patients with unstable or death by 33% to 55% (depending on
ment was 17.3% (78/451) in participants angina. Each of these trials reported a whether the 1 trial was excluded), the
treated with aspirin plus heparin, and trend toward decreased risk of MI or absolute risk reduction of MI or death
22.6% (98/434) in participants treated death in patients with unstable angina during treatment is 2.9% to 3.4% for
with aspirin alone, and the summary treated with both heparin and aspirin, patients with unstable angina treated
Downloaded from www.jama.com by guest on April 16, 2010

4. Table 3.—Relative Risk of Recurrent Ischemie Pain and Major Bleeding During Treatment With Heparin, pared with the 0.4% incidence of major
Relative Risk of Revascularization Procedures Following Randomization, and Relative Risk of Myocardial bleeding reported in patients treated
Infarction or Death During the 2-12 Weeks Following Randomization in Patients With Unstable Angina* with aspirin alone does not justify with¬
RR (95% CI) RR (95% CI) RR (95% CI) RR (95% CI) holding heparin from patients with un¬
of Recurrent of Major of CABG of Ml or Death stable angina.
Source Ischemie Pain Bleeding or PTCA at 2-12 wk
Several limitations ofthis meta-analy¬
Theroux étal, 198810 0.64(0.41-1.24) 0.97(0.75-1.28)
1.98(0.50-10.63) 0.50(0.25-1.60) sis deserve comment. The summary RR
RISC Group, 19907_NA_NB_NA_0.77 (0.46-1 63) may be overestimated if publication bias
Cohen et al, 1990" NB NA
1.98 (1.01-4.19)t 0.59(0.93-2.67) makes it more likely that studies show¬
Cohen étal,
1994'2_0.36(0.23-0.82) 7.06(0.46-135.15) 0.78(0.47-1.57) 0.75(0.41-1.80) ing benefit are published while those
Holdright et al, 1994'3 0.74(0.52-1.17) 0.85(0.12-13.47) 1.08(0.65-2.03) 0.97(0.72-1.36) showing no benefit are not. If publica¬
Gurfinkel et al, 1995" 1.20(0.82-1.78) 5.21(0.34-106.67) 0.81(0.42-2.06)_NA tion bias exists, small studies with nega¬
Summary 0.68(0.40-1.17) 1.89(0.66-5.38) 1.03(0.84-1.43) 0.82(0.56-1.20) tive findings are unlikely to be published,
while small studies with positive find¬
*RR Indicates relative risk; 95% CI, confidence interval for patients treated with aspirin plus heparin compared
with those treated with aspirin alone; Ml, myocardial infarction; CABG, coronary artery bypass grafting; PTCA, ings are likely to be published. This could
percutaneous translumlnal coronary angioplasty; NA, not available; and NB, no major bleeding in the aspirin group result in a correlation between study
or the aspirin plus heparin group.
tResults from this trial were excluded from the meta-analysis of recurrent ischemie pain because of heterogeneity size and RR estimate. We found no such
(see text). correlation among the 6 randomized tri¬
als in our meta-analysis (r=0.25; P= .64).
with heparin plus aspirin compared with moved from the meta-analysis, the sum¬ The validity of results from a meta-
those treated with aspirin alone. There¬ mary RR was 0.68 (95% CI, 0.40-1.17), analysis depends on the quality of the tri¬
fore, 29 to 34 patients would need to be suggesting a 32% risk reduction which als included.38 Formal weighting of the
treated with heparin for every 1 MI or is similar to the risk reduction found for quality of individual studies in this meta-
death prevented. Despite the risks of MI or death. This finding is consistent analysis was not done because all 6 were
intravenous heparin therapy including with the results of 2 other studies that well-performed randomized controlled tri¬
bleeding, thrombocytopenia, skin necro¬ reported increased infarction rate and als. A strength of these trials is that the
sis, hypersensitivity reactions, and cath¬ reactivation of unstable angina follow¬ primary outcomes (MI and death) were
eter-related infections,34 we believe that ing discontinuation of heparin.36·37 objective, and MI was diagnosed using
this low number needed to treat justi¬ Because the anticoagulant effects of standard criteria. In addition, follow-up
fies treating patients with unstable an¬ heparin are brief, any benefit of therapy among all 6 studies was excellent. Only 1
gina with 2 to 7 days of heparin added is unlikely to last beyond the duration of study11 reported incomplete follow-up: 2
to aspirin. However, it should be noted treatment. Consistent with this theory, patients were lost to follow-up at 12 weeks.
that the majority of primary outcome we found no reduction in risk of MI or Bias may result from inadequate blind¬
events in the 4 studies that reported MI death between 2 and 12 weeks following ing in randomized controlled trials. Table
and death separately were nonfatal MI. randomization in patients with unstable 1 shows that only 2 trials were double-
The number of deaths in these studies angina who received heparin and aspirin blinded,10·14 2 blinded only the partici¬
was too low to calculate a meaningful compared with those who received aspi¬ pants,12·13and 2 were unblinded.7·11 The
difference between therapies for death rin alone. This result underscores that trial by Cohen et al12 was stopped early
alone. The enhanced risk of bleeding on heparin is a short-acting, temporizing due to high withdrawal rates, particu¬
heparin might not be worth the benefit therapy, and not an intervention that al¬ larly in the aspirin plus heparin arm. Bi¬
of reduced nonfatal MI for some patients. ters underlying atherosclerotic disease. ased findings from this trial are unlikely,
Any benefit of adding heparin to aspi¬ A high percentage of patients who are however, because most of the difference
rin most likely occurs because heparin pre¬ admitted to the hospital with unstable in withdrawal between the 2 groups was
vents propagation of established throm¬ angina go on to have coronary angiogra- due to medication intolerance or patient
bus and allows time for endogenous phy. The decision to revascularize is based request and not to an increased rate of
fibrinolysis to occur. In theory, preven¬ on coronary anatomy and the degree of events in the aspirin plus heparin group.
tion offurther thrombus formation should underlying atherosclerotic disease. Be¬ The trial by Gurfinkel et al14 was stopped
act synergistically with the antiplatelet cause heparin does not change underly¬ prematurely because of much greater
effects of aspirin to reduce intracoronary ing atherosclerotic disease, we did not efficacy of a third arm, low-molecular-
obstruction and reduce myocardial ische¬ expect heparin to reduce the incidence of weight heparin plus aspirin. While ter¬
mia in patients with unstable angina. coronary artery bypass grafting or per¬ minating the trial early might have re¬
Heparin alone has been demonstrated cutaneous transluminal coronary angio- duced the power of the study (and our
to reduce the incidence of recurrent is¬ plasty following treatment for unstable meta-analysis) to detect a difference be¬
chemie pain in patients with unstable angina. The results of our meta-analysis tween unfractionated heparin and aspi¬
angina.32·35 There was heterogeneity in support this theory. rin compared with aspirin alone, it should
the findings of the 6 clinical trials re¬ We found a RR of 1.89 (95% CI, 0.66- not have biased the outcomes.
garding the benefit of adding heparin to 5.38) for major bleeding in patients with The Agency for Health Care Policy and
aspirin for the prevention of recurrent unstable angina treated with aspirin plus Research Clinical Practice Guideline39
ischemie pain. Much of this heterogene¬ heparin compared with those treated has recommended that, unless contrain-
ity was due to the findings of 1 trial11 with aspirin alone. This result is not dicated, patients with unstable angina be
that used 80 mg of aspirin per day in the statistically significant but is consistent treated with heparin for 2 to 5 days. This
aspirin plus heparin group, but 325 mg with the results of other studies of the recommendation was based on a panel
per day in the aspirin only group. While adverse effects associated with heparin consensus in the absence of directly
each dose should have provided adequate therapy. If adding heparin to aspirin applicable clinical studies (strength of
antiplatelet therapy, it is possible the reduces the risk of MI or death by one evidence C). This meta-analysis pro¬
=
higher dose also provided analgesia, re¬ third, we believe that the 1.5% incidence vides statistical evidence to support the
ducing the observed benefit of heparin of major bleeding reported in patients recommendation stated in this guideline.
on recurrent pain. With this study re- treated with aspirin plus heparin com- Our conclusion is further supported by
Downloaded from www.jama.com by guest on April 16, 2010

5. 2 randomized controlled trials comparing dence suggests that unless heparin is 21. Bishop YMM, Feinberg SE, Holland PW. Dis-
crete Multivariate Analysis: Theory and Practice.
low-molecular-weight heparin plus aspi¬ contraindicated, most patients with un¬ Cambridge, Mass: MIT Press; 1975.
rin with aspirin alone in patients with stable angina should be treated with both 22. Cohen M, Parry G, Adams PC, et al. Prospec-
unstable angina.14·40 The first study ran¬ aspirin and heparin. tive evaluation of a prostacyclin-sparing aspirin for-
domized patients to 3 arms (low-molecu¬ mulation and heparin/warfarin in aspirin users with
unstable angina or non-Q wave myocardial infarc-
lar-weight heparin plus aspirin, unfrac¬ References tion at rest. Eur Heart J. 1994;15:1196-1203.
tionated heparin plus aspirin, or aspirin 1. Fuster V, Badimon L, Cohen M, Ambrose JA, 23. Fuchs J, Cannon CP. Hirulog in the treatment
alone) and found a RR of 0.07 (95% CI, Badimon JJ, Chesebro J. Insights into the patho- of unstable angina: results of the Thrombin Inhi-
0.0-1.28) for MI and death in 68 patients genesis of acute ischemic syndromes. Circulation. bition in Myocardial Ischemia (TIMI) 7 trial. Cir-
treated with low-molecular-weight hep¬ 1988;77:1213-1220. culation. 1995;92:727-733.
2. Gazes PC, Mobley E Jr, Faris H Jr, Duncan RC, 24. Serneri GGN, Modesti PA, Gensini GF, et al.
arin compared with 73 patients treated Humphries GB. Preinfarctional (unstable) anginap=m-a Randomised comparison of subcutaneous heparin,
with aspirin alone.14 This trial also com¬ prospective studyp=m-tenyear follow-up: prognostic intravenous heparin, and aspirin in unstable an-
pared low-molecular-weight heparin to significance of electrocardiographic changes. Cir- gina: Studio Epoorine Sottocutanea nell'Angina In-
culation. 1973;48:331-337. stobile (SESAIR) Refrattorie Group. Lancet. 1995;
unfractionated heparin and found a RR
3. Duncan B, Fulton M, Morrison SL, et al. Prog- 345:1201-1204.
of 0.11 (95% CI, 0.01-2.17) for MI and nosis of new and worsening angina pectoris. BMJ. 25. Moise A, Roos M, Roos M. Aspirin versus hep-
death in patients treated with low-mo¬ 1976;1:981-985. arin in the acute phase of unstable angina. Circu-
lecular-weight heparin plus aspirin com¬ 4. Rahimtoola SH. Coronary bypass surgery for lation. 1994;90:1107.
unstable angina. Circulation. 1984;69:842-848. 26. Theroux P, Waters D, Qiu S, et al. Aspirin
pared with unfractionated heparin plus 5. Cairns JA, Gent M, Singer J, et al. Aspirin, sul- versus heparin to prevent myocardial infarction
aspirin. The second study randomized pa¬ finpyrazone, or both in unstable angina: results of during the acute phase of unstable angina. Circu-
tients to 2 arms (low-molecular-weight a Canadian multicenter trial. N Engl J Med. 1985; lation. 1993;88:2045-2048.
heparin plus aspirin or aspirin alone) and 313:1369-1375. 27. Cohen M Xiong J, Parry G, et al. Prospective
found a RR of 0.37 (95% CI, 0.20-0.68) for 6. Lewis H Jr, Davis JW, Archibald DG, et al. comparison of unstable angina versus non-Q wave
MI and death in 741 patients treated with
Protective effects of aspirin against acute myocar- myocardial infarction during antithrombotic ther-
dial infarction and death in men with unstable an- apy: Antithrombotic Therapy in Acute Coronary
low-molecular-weight heparin plus aspi¬ gina: results of a Veterans Administration Coop- Syndromes Research Group. J Am Coll Cardiol.
rin compared with 757 patients treated erative Study. N Engl J Med. 1983;309:396-403. 1993;22:1338-1343.
with aspirin alone.40 In this large trial, 7. The RISC Group. Risk of myocardial infarction 28. Averkov OV, Zateishchikov DA, Gratsianskii
and death during treatment with low dose aspirin NA, Dobrovol'skii AV. Unstable angina: tissue-type
major bleeding occurred in 0.5% of the and intravenous heparin in men with unstable coro- plasminogen activator inhibitor and other hemo-
aspirin group and 0.8% of the low-mo¬ nary artery disease: the RISC Group. Lancet. 1990; static factors in aspirin and intravenous heparin
administration. Kardiologiia. 1993;33:16-22.
lecular-weight heparin group (RR, 1.53, 336:827-830.
8. Rosenberg RD. Actions and interactions of an- 29. Averkov OV, Zateishchikov DA, Gratsianskii
95% CI, 0.43-5.44). tithrombin and heparin. N Engl J Med. 1975;292: NA, Logutov I, Iavelov IS, Ianus V. Unstable an-
If therapeutically equivalent, low-mo¬ 146-151. gina: effect of aspirin and heparin on treatment
lecular-weight heparin might be preferred 9. Wallis DE, Boden WE, Califf R, et al. Failure of outcome in hospital patients (a double-blind, placebo-
over unfractionated heparin for several adjuvant heparin to reduce myocardial ischemia in controlled study). Kardiologiia. 1993;33:4-9.
the early treatment of patients with unstable an- 30. Zwerner PL, Gore JM, Corrao JM, et al. Heparinnnnn`
reasons. Unlike unfractionated heparin,
gina. Am Heart J. 1991;122:949-954. in the treatment of unstable angina: a randomized
which has a bioavailability of only 30%, all 10. Theroux P, Ouimet H, McCans J, et al. Aspirin, prospective trial. Circulation. 1987;76(IV):180.
low-molecular-weight heparins have a bio¬ heparin, or both to treat acute unstable angina. 31. Charvat J, Fiserova J. Anticoagulation and an-
availability that approaches 100%.41 Low- N Engl J Med. 1988;319:1105-1111.
11. Cohen M, Adams PC, Hawkins L, Bach M, Fus-
tiaggregation therapy in patients with unstable an-
gina pectoris. Vnitr Lek. 1990;36:537-543.
molecular-weight heparin is administered ter V. Usefulness of antithrombotic therapy in rest- 32. Serneri GGN, Gensini GF, Poggesi L, et al.
subcutaneously rather than intrave¬ ing angina pectoris or non-Q-wave myocardial in- Effect of heparin, aspirin, or alteplase in reduction
nously, its half-life is longer than that of farction in preventing death and myocardial of myocardial ischaemia in refractory unstable an-
unfractionated heparin, and therapeutic infarction (a pilot study from the Antithrombotic gina. Lancet. 1990;335:615-618. Correction: Lancet.
monitoring is not needed. Therapy in Acute Coronary Syndromes Study 1990;335:868.
Group). Am J Cardiol. 1990;66:1287-1292. 33. OASIS. Comparison of hirudin with heparin
While low-molecular-weight heparin 12. Cohen M, Adams PC, Parry G, et al. Combina- and warfarin with control for unstable angina and
and unfractionated heparin are differ¬ tion antithrombotic therapy in unstable rest angina non q wave MI in a randomized controlled trial.
ent preparations, they have similar and non-Q-wave infarction in nonprior aspirin users: Circulation. 1995;92:416.
mechanisms of action.41 Including the 2 primary end points analysis from the ATACS trial: 34. Hirsh J, Raschke R, Warkentin TE, Dalen JE,
Antithrombotic Therapy in Acute Coronary Syn- Deykin D, Poller L. Heparin: mechanism of action,
studies of low-molecular-weight hepa¬ dromes Research Group. Circulation. 1994;89:81-88. pharmacokinetics, dosing considerations, monitor-
rin in our meta-analysis would have 13. Holdright D, Patel D, Cunningham D, et al. ing, efficacy, and safety. Chest. 1995;108:258S-275S.
yielded a summary RR of MI or death Comparison of the effect of heparin and aspirin 35. Williams DO, Kirby MG, McPherson K, Phear
versus aspirin alone on transient myocardial ische- DN. Anticoagulant treatment of unstable angina.
during randomized treatment of 0.56 mia and in-hospital prognosis in patients with un- Br J Clin Pract. 1986;40:114-116.
(95% CI, 0.40-0.80; for test of heteroge¬ stable angina. J Am Coll Cardiol. 1994;24:39-45. 36. Theroux P, Waters D, Lam J, Juneau M, McCans
neity, P=.52) in patients with unstable 14. Gurfinkel EP, Manos EJ, Mejail RI, et al. Low J. Reactivation of unstable angina after the discon-
angina treated with aspirin plus heparin molecular weight heparin versus regular heparin or tinuation of heparin. N Engl J Med. 1992;327:141-145.
aspirin in the treatment of unstable angina and silent 37. Granger CB, Armstrong PW. Reinfarction fol-
compared with those treated with as¬ ischemia. J Am Coll Cardiol. 1995;26:313-318. lowing discontinuation of intravenous heparin or
pirin alone. We believe the possibility 15. Walter S, Cook R. A comparison of several hirudin for unstable angina and acute myocardial
that low-molecular-weight heparin is su¬ point estimators of the odds ratios in a single 2m=x infarction. Circulation. 1995;92(81):460.
perior to unfractionated heparin in pa¬ contingency table. Biometrics. 1991;47:795-811. 38. Moher D, Olkin I. Meta-analysis of randomized
tients with unstable angina should be 16. Fleiss JL. Statistical Methods for Rates and controlled trials. JAMA. 1995;274:1962-1964.
Proportions. 2nd ed. New York, NY: John Wiley & 39. Unstable Angina Panel. Unstable Angina: Di-
explored in a randomized controlled trial. Sons Inc; 1981. agnosis and Management. Clinical Practice Guide-
17. Fisher LD, van Belle G. Biostatistics: A Meth- line. Washington, DC: US Dept of Health and Hu-
CONCLUSION odology for the Health Sciences. New York, NY: man Services, Public Health Service, Agency for
John Wiley & Sons Inc; 1993. Health Care Policy and Research; 1994. AHCPR
This meta-analysis of 6 randomized 18. Berlin JA, Laird NM, Sacks HS, Chalmers TC. 94-0602.
controlled trials demonstrated a strong A comparison of statistical methods for combining 40. The FRISC Study Group. Low-molecular-
trend toward reduction in risk of MI or event rates from clinical trials. Stat Med. 1989;8: weight heparin during instability in coronary ar-
death during randomized therapy in pa¬ 141-151. tery disease: Fragmin During Instability in Coro-
tients with unstable angina treated with 19. Fleiss JL. The statistical basis of meta-analy- nary Artery Disease (FRISC) Study Group. Lancet.
sis. Stat Methods Med Res. 1993;2:121-145. 1996;347:561-568.
aspirin plus heparin compared with those 20. BMDP Statistical Software, Inc. PC90 ed. Los 41. Barrowcliffe TW. Low molecular weight hepa-
treated with aspirin alone. Current evi- Angeles: University of California Press; 1990. rin(s). Br J Haematol. 1995;90:1-7.
Downloaded from www.jama.com by guest on April 16, 2010