Author + information
- Received June 20, 1996
- Revision received August 28, 1996
- Accepted September 25, 1996
- Published online January 1, 1997.
- Scott L. Woodfield, MDA,
- Conor F. Lundergan, MDA,*,
- Jonathan S. Reiner, MD, FACCA,
- Mark A. Thompson, MD, FACCA,
- Steven C. Rohrbeck, MD, FACCA,
- Yuri Deychak, MD, FACCA,
- James O. Smith, MD, FACCB,
- Jeffrey R. Burton, MD, FACCC,
- William F. McCarthy, PhDA,
- Robert M. Califf, MD, FACCD,
- Harvey D. White, DSc, FACCE,
- W.Douglas Weaver, MD, FACCF,
- Eric J. Topol, MD, FACCG and
- Allan M. Ross, MD, FACCA
- ↵*Dr. Conor Lundergan, Cardiovascular Research Institute, Division of Cardiology, The George Washington University, 2150 Pennsylvania Avenue, NW, Suite 4-414, Washington, D.C. 20037.
Objectives. This study sought to 1) determine the effect of gender on early and late infarct-related artery patency and reocclusion after thrombolytic therapy for acute myocardial infarction; 2) examine the effect of gender on left ventricular function in response to injury/reperfusion; and 3) assess the independent contribution of gender to early (30-day) mortality after acute myocardial infarction.
Background. Women have a higher mortality rate than men after myocardial infarction. However, the effect of gender on infarct-related coronary artery patency and left ventricular response to injury/reperfusion have not been fully defined in the thrombolytic era.
Methods. Patency rates and global and regional left ventricular function were determined in patients at 90 min and 5 to 7 days after thrombolytic therapy for acute myocardial infarction. The effect of gender on infarct-related artery patency and left ventricular function was determined. Thirty-day mortality differences between women and men were compared.
Results. Women were significantly older and had more hypertension, diabetes, hypercholesterolemia, heart failure and shock. They were less likely to have had a previous myocardial infarction, history of smoking or previous bypass surgery. Ninety-minute patency rates (Thrombolysis in Myocardial Infarction [TIMI] flow grade 3) in women and men were 39% and 38%, respectively (p = 0.5). Reocclusion rates were 8.7% in women versus 5.1% in men (p = 0.14). Women had more recurrent ischemia than men (21.4% vs. 17.0%, respectively, p = 0.01). Ninety-minute ejection fraction and regional ventricular function were clinically similar in women and men with TIMI 2 or 3 flow (ejection fraction [mean ± SD]: 63.4 ± 6% vs. 59.4 ± 0.7%, p = 0.02; number of chords: 21.4 ± 0.9 vs. 21.0 ± 1.9, p = 0.7; SD/chord: −2.4 ± 08 vs. −2.4 ± 0.2, p = 0.9, respectively). No clinically significant differences in left ventricular function were noted at 5- to 7-day follow-up. Women had a greater hyperkinetic response than men in the noninfarct zone (SD/chord: 2.4 ± 0.2 vs. 1.7 ± 0.1, p = 0.005). The 30-day mortality rate was 13.1% in women versus 4.8% in men (p ≤ 0.0001). After adjustment for other clinical and angiographic variables, gender remained an independent determinant of 30-day mortality.
Conclusions. Women do not differ significantly from men with regard to either early infarct-related artery patency rates or reocclusion after thrombolytic therapy or ventricular functional response to injury/reperfusion. Gender was an independent determinant of 30-day mortality after acute myocardial infarction.
(J Am Coll Cardiol 1997;29:35–42)>
The incidence of coronary artery disease in women increases after menopause, equaling that of men by the age of 70 (). It is the leading cause of mortality in older women, accounting for one-third of all deaths. The mortality rate of women in the setting of acute myocardial infarction is generally regarded as higher than that in men ([2–6]), and this difference continues to exist in the era of reperfusion therapy despite an equivalent decrease in mortality in both men and women ([7–11]). Most of the excess mortality has been accounted for by increased rates of comorbidities, such as hypertension (), advanced age ([5, 13]) and diabetes mellitus ([4, 12, 14]). There is evidence of decreased rates of catheterization and angioplasty ([14–16]) and underutilization of thrombolytic agents ([17, 18]) in women, possibly due to fears of increased risk of complications or to selection bias ([12, 19–25]), factors that may contribute to the greater mortality noted in women.
Other potential explanations for the worse outcome in women include higher rates of failed thrombolysis, increased incidence of reocclusion and reinfarction and accentuated myocardial injury in response to ischemia and reperfusion. The effect of gender on the rate of successful thrombolysis and ventricular response to injury and reperfusion has not been well studied.
Analysis of the Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary Arteries (GUSTO-I) trial () found gender to be an independent predictor of 30-day mortality, after adjustment for clinical variables (). In a separate analysis, the mortality benefit of thrombolytic therapy was equivalent in men and women, despite a higher stroke rate in women (). The GUSTO-I Angiographic Study () offered a unique opportunity to assess the effect of gender on infarct-related artery patency after thrombolytic therapy and the effect of gender on left ventricular response to ischemia/reperfusion. In addition, the independent effect of gender on short-term (30-day) mortality could be determined after adjustment for both clinical and angiographic variables.
1.1 Study patients.
The GUSTO-I Angiographic Study has been described in detail previously (). Briefly, patients were included in the GUSTO-I trial if they had chest pain <6 h in duration and ST segment elevation in at least two contiguous electrocardiographic leads. We compared global and regional left ventricular function after thrombolysis in men and women who had angiographically documented patency (Thrombolysis in Myocardial Infarction [TIMI] flow grade 2 and 3) in the infarct-related coronary artery 90 min (±45 min) after thrombolytic therapy, sustained patency at 5- to 7-day follow-up angiography and ventriculograms adequate for analysis. We also compared noninfarct zone function at 90 min in all male and female patients randomized to 90-min angiography with adequate ventriculograms for analysis regardless of 90-min TIMI flow grade.
1.2 Thrombolytic therapy and coronary angiography.
As described previously (), patients were randomized to one of four thrombolytic regimens. At the time of drug randomization, patients were also randomized to one of four times for initial coronary angiography after the start of thrombolytic therapy. This report includes angiographic data only from those patients randomized to 90-min postthrombolytic therapy angiography. These patients also had 5- to 7-day follow-up catheterization.
1.3 Core angiographic laboratory procedures and cineangiographic analysis.
Films were interpreted by an experienced angiographer (C.F.L., J.S.R., A.M.R.) who had no knowledge of treatment allocation, angiographic randomization or gender. Infarct-related artery patency as defined by TIMI flow grade was assessed according to standard methods (). Ventriculographic silhouettes were acquired digitally at end-systole and end-diastole, and the borders were defined by the core laboratory angiographer. Ventricular volumes and ejection fraction were calculated by the single plane area-length method (). Regional ventricular function was assessed by the method of Bolson et al. () and included mean excursion of the most abnormal 50% of chords in the infarct region, expressed as the number of standard deviations per chord (SD/chord) and the number of consecutive chords in the infarct region more than two standard deviations below the norm. The noninfarct region was similarly assessed as the SD/chord of the most normal 50% of chords. Patients who underwent coronary angioplasty were excluded from analysis of ventricular function.
1.4 Rescue angioplasty definition.
Patients with TIMI flow grade 0 or 1 in the infarct-related artery at the time of 90-min catheterization or who had reocclusion of an initially patent (TIMI flow grade 2 or 3) infarct-related artery and underwent immediate attempted angioplasty with cine films adequate for analysis were classified as having undergone rescue angioplasty attempts. Successful rescue angioplastywas defined as conversion from TIMI flow grade 0 or 1 to TIMI flow grade 2 or 3 and <50% residual stenosis at the completion of the angioplasty procedure.
1.5 Statistical analysis.
Results are expressed as mean value ± SEM, unless otherwise noted. Chi-square analysis was utilized to compare categoric data, and continuous variables were compared using the unpaired Student ttest (two-tailed). Multivariable linear regression analysis () was utilized to determine the independent effect of clinical and angiographic variables in the prediction of regional and global left ventricular function indexes. The effect of gender on mortality was analyzed by multivariable logistic regression analysis (). Demographic and angiographic variables that were considered possibly related to left ventricular functional outcomes or mortality, or both, were incorporated into the models, and the independent effect of each variable was determined. Clinical variables for testing were chosen on the basis of their previously described independent effect on 30-day mortality () and included gender, age, diabetes, previous history of infarction and initial heart rate and systolic blood pressure. Angiographic variables included 90-min (±45) TIMI flow grade and multivessel coronary artery disease. The TIMI flow grade was entered as a dichotomous variable (TIMI flow grade 0, 1 and 2 vs. TIMI flow grade 3) to reflect its biologic significance (). Continuous variables were entered into the model without modification except where indicated. Adjusted relative odds ratios were also determined with 95% confidence intervals. A p < 0.05 was considered indicative of statistical significance in all cases.
The difference in 30-day survival between women and men was determined by Kaplan-Meier curves. The significance of the difference was determined by the log rank test.
2.1 Patient characteristics.
The GUSTO-I Angiographic Study enrolled 2,431 patients (543 women, [22.3%], 1,887 men [77.7%]). The gender of one patient was not documented. Women differed from men in many baseline variables (Table 1): They were older (66.2 ± 11.2 vs. 59.0 ± 11.5, p ≤ 0.01) and more often had a history of hypertension, diabetes and hypercholesterolemia. Women were also more likely to be treated later than men (mean time from symptom onset to initiation of thrombolytic therapy 3.3 ± 1.4 vs. 3.0 ± 1.5 h, p = 0.003). Although women had a similar incidence of multivessel coronary artery disease at the time of enrollment in the trial, the incidence of previous coronary artery bypass graft surgery (CABG) was more than twofold less than that in men (5.1% vs. 2%, p = 0.002). Also, despite a similar incidence of multivessel coronary artery disease and anterior infarction, women were twice as likely to have congestive heart failure or cardiogenic shock during the course of their hospital stay than men. Although women were significantly more likely to have recurrent ischemia during their hospital stay, there was no significant difference between genders with regard to reinfarction (Table 2).
2.2 Thrombolytic regimen randomization.
A similar proportion of women and men were randomized to each thrombolytic regimen: recombinant tissue-type plasminogen activator plus intravenous heparin 29.1% versus 31.3%; streptokinase plus intravenous heparin 24.2% versus 21.0%; streptokinase plus subcutaneous heparin 15.2% versus 20.5%; and combination therapy 31.5% versus 31.3% (all p > 0.05).
2.3 Effect of gender on 90-min patency, reocclusion and collateral circulation.
Ninety-minutes after thrombolytic therapy, no significant difference was noted with regard to infarct-related artery patency (TIMI flow grade 2 and 3) rates in women versus men (69.0% versus 66.5%, respectively, p = 0.46). This sample size of women and men is sufficient to detect a 10 percentage point decrease in patency (TIMI 2 and 3 flow grades) in women compared with men with a power of 80%, or a 20 percentage point difference with >99% power. Also of note, there was an equal proportion of women and men with fully normal (TIMI flow grade 3) flow 90-min after thrombolytic therapy (Table 3). The power to detect a 10 percentage point difference in 90-min TIMI flow grade 3 between women and men with this sample size is 83%. Reocclusion of initially patent (TIMI flow grade 2 and 3) infarct-related arteries was rare and occurred in 8.7% of female and 5.1% of male patients (p = 0.14). The power to detect a 10 or 15 percentage point difference in reocclusion of initialy patent infarct-related arteries between women and men is 91% and 99.5%, respectively. Angiographically visible collateral vessels were seen in 16.0% of women and 14.1% of men (p = 0.4).
2.4 Follow-up angiography.
Of 721 patients (165 women, 556 men) with a patent (TIMI flow grade 2 or 3) infarct-related artery at 90 min, 1-week follow-up angiography was performed in 87 (52.7%) of 165 women and 394 (70.8%) of 556 men (p ≤ 0.0001). Among the 240 patients who did not have follow-up angiography (78 women, 162 men), 18 (23.1%) of 78 women and 17 (10.5%) of 162 men died before follow-up (odds ratio [OR] 2.6, 95% confidence interval [CI] 1.2 to 5.3), whereas 60 (76.9%) of 78 women and 145 (89.5%) of 162 men failed to have follow-up angiography because of patient or physician refusal, reocclusion, intervening percutaneous transluminal coronary angioplasty (PTCA) or follow-up angiography that was performed before the prespecified 5- to 7-day time period.
2.5 Procedures and outcomes.
Percutaneous transluminal coronary angioplasty was performed in 185 (35.9%) of 515 women versus 654 (35.8%) of 1,826 men (p = 0.9). Data for PTCA was not available for 28 women, 61 men and 1 patient of unknown gender. The overall mortality rate in women undergoing PTCA was 8.1% (15 of 185) versus 5.2% (34 of 654) in men (p = 0.14). Cineangiograms were available for core laboratory analysis from 49 women and 181 men who underwent rescue PTCA. Rescue PTCA was successful in 43 (87.8%) of 49 women versus 157 (88.7%) of 181 men (p = 0.8). The mortality rate in women with successful rescue PTCA was 11.6% (5 of 43) versus 7.6% (12 of 157) in men (OR 1.6, 95% CI 0.5 to 4.7). Coronary artery bypass surgery was performed in 56 (10.3%) of 542 women and 186 (9.9%) of 1,884 men (p = 0.75). The status of five patients (one woman, three men, 1 patient of unknown gender) with regard to CABG is unknown. The mortality rate in women with CABG was 10.7% (6 of 56) versus 3.8% (7 of 186) in men (p = 0.08).
2.6 Effect of gender on left ventricular function.
Indexes of global and regional left ventricular function in men and women 90 min and 5 to 7 days after successful thrombolysis in patients with ventriculograms adequate for analysis are shown in Table 4. There were no clinically significant differences in either global or infarct zone regional left ventricular function either at 90 min or 5- to 7 days after successful thrombolysis in men compared with women. However, women did have a significantly greater degree of hyperkinesia in the noninfarct zone acutely and at 1 week follow-up than men.
The male and female patient groups were of greatly different sizes. The power to detect a difference of eight chords, or 0.6 SD/chord in regional ventricular function, in men compared with women at the 0.05 level of significance with the number of patients shown in Table 3for 90-min ventriculography is 99% and 75%, respectively. The corresponding powers to detect similar differences at 5- to 7-day ventriculography with the number of patients in each group are 97% and 75%, respectively.
2.7 Multivariable analysis: independent determinants of ventricular function.
The independent determinants of left ventricular function at 90 min and 5 to 7 days after myocardial infarction are shown in Table 5. Female gender is a significant determinant of lower end-systole volume index and noninfarct zone ventricular function at 90 min but not of ejection fraction. Female gender is a significant determinant of 5- to 7-day global left ventricular function (i.e., female gender is associated with a higher 5- to 7-day ejection fraction and lower end-systolic volume index than male gender after adjusting for other variables). However, gender has no effect on regional (infarct or noninfarct zone) ventricular function 1-week after occlusion/reperfusion of the infarct-related artery.
The unadjusted 30-day mortality rate in women was 13.1% versus 4.8% in men (p < 0.0001) (Fig. 1). Female patients with a patent infarct-related artery 90-min after thrombolytic therapy had a mortality rate of 11.5% versus 3.8% in men (OR 3.3, 95% CI 1.7 to 6.3). Women with an occluded infarct-related artery at 90-min had a mortality rate of 19.0% versus 5.4% in men (OR 4.1, 95% CI 2.0 to 9.0). The mortality rate in those with stroke was 38% in women versus 33% in men (p = 0.75). After adjustment for age, gender remained a significant independent determinant of 30-day mortality, as shown in Fig. 2(p = 0.002). After adjustment for clinical or angiographic variables, or both, gender remained an independent predictor of 30-day mortality (Fig. 3).
3.1 Infarct-related artery patency and reocclusion.
The early postmyocardial infarction mortality rate for women continues to exceed that of men in the reperfusion era ([7–11]). However, our data demonstrate that successful thrombolysis 90 min after administration of a thrombolytic regimen is achieved in equal proportions of women and men. Complete normalization of coronary flow (TIMI flow grade 3) in the infarct-related artery, a finding previously shown () to be highly correlated with convalescent ventricular function and early mortality, also occurs in equal proportions of women and men. Angiographic reocclusion of an initially patent infarct-related artery, an event associated with high mortality (), occurred more frequently in women than men, although the difference did not reach statistical significance. This may be due to the relatively small sample size of our study or the paucity of reocclusion in general in the present study. Of note, the main GUSTO-I trial did show a slight increase in clinical reinfarction in women compared with men (OR 1.3, 95% CI 1.12 to 1.41) (). Our study also demonstrated a similar trend toward increased reinfarction in women, although it failed to reach statistical significance.
Women were less likely to have a history of smoking than men. Several studies, including an analysis from GUSTO-I ([35, 36]), have suggested that smokers have a higher thrombus burden at the infarct site and are more likely to respond favorably to thrombolysis, with higher patency rates and lower rates of reinfarction and recurrent ischemia. However, it would appear unlikely that a significant proportion of the excess mortality in women noted in the present study can be attributed to failed thrombolysis. Increased rates of reocclusion/reinfarction may contribute to the increased early mortality noted in women, although this remains to be definitely determined.
3.2 Left ventricular function.
We documented that there is no difference in the infarct zone regional response of the myocardium to injury/reperfusion in women compared with men. However, women had a slightly greater ejection fraction and a smaller end-systolic volume index than men 90 min and 5 to 7 days after successful thrombolysis, a finding that did reach statistical significance. Despite this finding, women were twice as likely to develop congestive heart failure and cardiogenic shock than men. Numerous studies have shown ([4–6, 12, 37, 38]) that the incidence of congestive heart failure is higher in women after myocardial infarction, although the Thrombolysis and Angioplasty in Myocardial Infarction (TAMI) investigators () found no difference. There is also conflicting data on the incidence of cardiogenic shock in women in the setting of acute myocardial infarction ([6, 12, 37]). It is possible that the female population, due to a greater prevalence of hypertension and diabetes, has a greater degree of diastolic dysfunction, leading to more congestive heart failure ([40, 41]). Notably, in our cohort, women had significantly better noninfarct zone systolic function than men. Because there was no difference in the prevalence of previous myocardial infarction or multivessel disease, one potential explanation for this finding is that women have a different adrenergic response to infarction, leading to an increased hyperkinesia of the noninfarct zone. Also, the relative lack of previous CABG procedures in women despite equivalent rates of previous infarction and multivessel disease might be expected to adversely affect noninfarct zone function, thus contributing to the excess in heart failure. However, this was not found in our study cohort.
3.3 In-hospital procedures.
Some investigators ([14–16]) have reported that women are less likely to be referred for cardiac catheterization and subsequent revascularization procedures than men after myocardial infarction. However, in the present study, women and men were referred for revascularization in equal proportions. Of note, the success rates of rescue angioplasty were equivalent in each gender. There was a trend toward increased mortality after CABG in women in this study, although not statistically significant in this cohort. However, given the small sample size we are unable to state that differences in outcomes between women and men after CABG do not account for at least a proportion of the excess early mortality noted in women in the present study ().
The increased mortality noted in women after myocardial infarction has been attributed to an excess of adverse comorbid factors ([4, 5, 12–14]). The GUSTO-I trial of >41,000 patients identified a number of predictors of 30-day mortality (). After adjustment for multiple baseline characteristics, female gender was a significant independent prognostic factor for 30-day mortality but only marginally so (). The only previous angiographic study to evaluate this group (TAMI []), did not find gender to be a significant independent determinant of early postinfarction mortality after adjustment for clinical or angiographic variables, or both, although a trend was evident. However, our data suggest that the excess early mortality in women cannot be fully explained by more adverse clinical characteristics noted in the female population by adjustment for angiographic variables. Furthermore, our results suggest that within this trial cohort, there was not a systematic gender bias in referrals for revascularization therapies, which may effect mortality. It should also be noted that the confidence limits surrounding the point estimate of 30-day mortality after adjustment for clinical and angiographic variables are wide, ranging between 1.2 and 3.7. These data are in accord with the results of the GUSTO-I main trial in 41,021 patients ([27, 28]); the wide confidence intervals in our study are due to the relatively smaller sample size.
Our findings suggest, therefore, that the early (30-day) mortality difference between men and women is due to yet to be defined variables or to the small but cumulative effects of numerous variables (e.g., reocclusion differences, post-CABG mortality), which may not be captured within the confines of multivariable analysis. However, our data strongly suggest that neither differences between women and men with regard to thrombolytic-induced infarct-related artery patency rates nor differences in the response of the ventricular myocardium to ischemia/reperfusion contribute significantly to any difference in a 30-day mortality.
3.5 Study limitations.
The present study has several limitations inherent to subgroup analysis. The GUSTO-I angiographic trial data base was designed to evaluate the efficacy of several thrombolytic regimens on early infarct-related artery patency and left ventricular functional outcomes. Because patients were not randomized according to gender, the proportion of female and male patients in this study was unequal. The failure to detect differences in regional left ventricular function between women and men could thus be due to lack of sufficient statistical power. However, we provided power calculations for the determination of differences in regional function that we consider to be clinically significant in relation to the number of female and male patients included in our analysis. Similarly, the failure to detect a difference between women and men with regard to 90-min patency and subsequent infarct-related artery reocclusion may be limited by our sample size. However, as we showed, we do have sufficient power to detect differences that would be considered clinically significant. A ≥10% difference in early patency or reocclusion rates would be detected with a power of >80% in each instance.
We also provided data on mortality differences (univariable analyses) between women and men after CABG and PTCA. These mortality rates should only be considered descriptive of this particular cohort. Our sample size is too limited to draw any specific conclusions regarding postprocedural mortality.
Other limitations include the fact that 1-week follow-up angiography rates differed between women and men. This difference was largely due to the excess number of female patient deaths relative to male patient deaths in the first postinfarction week. The lower rate of angiographic follow-up may have reduced our ability to detect significant differences in the male and female ventricular response to injury/reperfusion. Also, because it was not a requirement of the trial to have all rescue or elective angioplasty films reviewed in the core angiographic laboratory, a potential for bias exists in that particular analysis.
Thrombolytic therapy is equally efficacious in establishing early patency in the infarct-related artery in women and men. Reinfarction and reocclusion of an initially patent infarct-related artery tended to be more common in female patients but were not statistically significant in our sample. There is no gender-specific difference between the female and male myocardium in response to injury/reperfusion. Hyperkinesia of the noninfarct zone is significantly greater in women than men. Despite these findings, the 30-day mortality rate in women remained significantly greater after adjustment for clinical and angiographic variables than that in men. These data support the use of thrombolytic therapy as an effective means of ventricular salvage regardless of gender.
☆ This study was supported by a combined grant from Bayer, New York, New York; CIBA-Corning, Medfield, Massachusetts; Genentech, South San Francisco, California; ICI Pharmaceutical, Wilmington, Delaware; and Sanofi Pharmaceutical, Paris, France.
- coronary artery bypass graft surgery
- confidence interval
- Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary Arteries trial
- odds ratio
- percutaneous transluminal coronary angioplasty
- standard deviations per chord
- Thrombolysis in Myocardial Infarction
- Received June 20, 1996.
- Revision received August 28, 1996.
- Accepted September 25, 1996.
- The American College of Cardiology
- for the MILIS Study Group,
- Tofler GH,
- Stone PH,
- Muller JE,
- et al.
- the Israeli SPRINT Investigators,
- Greenland P,
- Reicher-Reiss H,
- Goldbourt U,
- Behar S
- Wilcox RG,
- Olsson CG,
- Skene AM,
- et al.
- for the Thrombolysis and Angioplasty in Myocardial Infarction Study Group,
- Lincoff AM,
- Califf RM,
- Ellis SE,
- Sigmon KN,
- Lee KL,
- Leimberger JD,
- Topol EJ
- for the Investigators of the International Tissue Plasminogen Activator/Streptokinase Mortality Study,
- White HD,
- Barbash GI,
- Modan M
- for the MIDAS Study Group,
- Kostis JB,
- Wilson AC,
- O’Dowd K,
- et al.
- Krumholz HM,
- Douglas PS,
- Lauer MS,
- Pasternak RC
- Ellis SG,
- Roubin GS,
- King SB,
- et al.
- Bovill EG,
- Terrin ML,
- Stump DC,
- et al.
- Lee KL,
- Woodlief LH,
- Topol EJ,
- et al.
- Chesebro JH,
- Knatterud G,
- Roberts R,
- et al.
- Bolson EL,
- Kliman S,
- Sheehan F,
- Dodge HT
- Freund RJ,
- Littell RC
- the TAMI Study Group,
- Ohman EM,
- Califf RM,
- Topol EJ,
- et al.
- Barbash GI,
- Reiner J,
- White HD,
- et al.
- DeChillou C,
- Riff P,
- Sadoul N,
- et al.
- Lincoff AM,
- Califf RM,
- Ellis SG,
- et al.
- Fein F