896 LETTERS

were done during the first month: clinical (pain, pericardial rub], echocardiographic (M-mode and Z-dimensional echocardioera- phy] and biologic (sedimentation, fib&- emia and C reactive-protein). The initial evaluation was done before the 36th hour af- ter the onset of the pain, and the others on days 5,10 and 30. Pericarditis wasdefined by the presence of echocardiographic PE associ- atedwithorwithoutclinicalsymptoms:group A [n = 151, P- (without pericarditisl and group B (n = 211 P% (with pericarditis]. Then, group B was randomly allocated to nonste- roidtreatment(indomethacin,15Omg,3times daily for 10 days) or no nonsteroid treatment: Bl (n = lo], nonsteroid treatment +, B2 (n = 111, nonsteriod treatment -. Our results showed:

Pericarditiswasfrequentlynotedin58% of cases(2lof36)andPEwasobservedin19of24 (P+]. PE was detected at the first echocardio- graphic examination [l2 of 191, was usually small or moderate (16 of 191. but never showed bad hemodynamic tolerance.

Pericarditis was associated with higher peak creatinine kinase level, yet not signifi- cant (P+: 1,709 f 1,051; P-: 1,126 f 878 LB; difference not significant] and, conversely, with a significantly higher biologic syndrome (sedimentation:42f17~~76&35mm/hour,p <0.05;fibrinemia:6.3 f 22~~7.2 f 1.6g/liter, p <0.02; C reactive protein: 0.043 f 0.030 vs 0.105 f 0.068 UI, p <O.OOl).

The natural history of pericarditis (Avs Bl) is characterized by spontaneous and parallel regression of the biological svndrome. but PE wasstillpresentin7ofiOpatientsattheendof the first month after AMI.

When treated with nonsteroid antiinflam- matory drugs, pericarditis showed a dissoci- ated evolution, significant efficacy on peri- cardial pain, pericardial rub, fever and sedimentation, but was still present in 6 of 11 of the B2 patients at the day. Thus, nonsteroid treatment demonstratesnonsignificant supe- riority on the regression of PE after AM1 [Bl vs BZ).

Despite the limitations of both M-mode and 2-dimensional echocardiography, this study confirms the large frequency of PE af- ter AM1 and its early occurrence seen on repeated echocardiographic examinations. These results are in agreement with the clini- cal correlations of Roeske et al2 who noticed that pericarditis was present in 40% of cases after AMI.

H. lardoux, MD V. Dormagen, MD

E. Francois, MD 1. DeBache, MD

H. Baufine-Ducrocq, MD V. Louvard, MD F. Chouty, MD

Corbeil-Essonnes, France 28June 1985

l.KaplanK, DavisonR,ParkerM,PrzybylekJ,Light A, Bresnahan D, kibner H, Talano J. Frequency of pericardial effusion as determined by m-mode echocardiography in acute myocardial infarction. Am J Cardiol 1985;55:335-337. 2. Roeske WR, Savage RM, O’Rourke RA, Bloor CM. Chnicopathologic correlations in patients af- ter myocardial infarction. Circulation 1981;63:36- 45.

LATE THROMBOLYSIS: CONTROLLED CLINICAL TRIAL OVERDUE

Shapiro et al1 demonstrate the apparent safe- ty and possible efficacy of coronary thrombo- lysis relatively late after the onset of acute myocardialinfarctioninpatientswithpostin- farction angina. They noted that this was not an appropriately designed, randomized con- trolled studv.2 However. as Chalmers3 and Spodick4 point out, “pilot studies” are often misleading, waste time and preempt organi- zation of controlled trials by prematurely popularizing a new treatment. Moreover, when one first tries new treatments or new applications of old treatments, one has high hopes, but cannot know whether it will bring net help or harm to the patient. Therefore, controlled clinical trials should beein with thefirstpatient, bothtoaccomplishtheobjec- tive these investigators propose for future work and to give the patients a 50:50 chance nottogettherisksorbenefitsofthenewthera- py. On scientific grounds, one can randomize from the first patient as described in the AJC.4 Moreover, this also should be done on ethi- ca13,5 and behaviora15v6 grounds.

An example of the diminishing likelihood that a controlled trial will be accomplished despite the undisputed good intentions of Shapiro et al is the case of emergency bypass surgery for acute myocardial infarction. The first report in 1979 was qualified by its work- ers as inconclusive in the absence of appro- priately designed controlled clinical trials.7 Four years later, a report of the same, now burgeoning, series was again accompanied by a call for a controlled clinical trial.8 That the absence of a trial was ascribed to adminis- trative resistance rather than the investiga- tors’ good intentions.9 is just another argu- ment for starting out right-randomizing the first patient-before scientific, ethical and behavioral standards can be circumvented.

David If. Spodick, MD, DSe Worcester, MA

2 July 1985

1. Shapiro EP, Brinker JA, Gottlieb SO, Guzman PA, Bulkley BH. Introcoronary thrombolysis 3 to 13 days after acute myocardial infarction for postin- farction angina pectoris. Am J Cardiol 1985; 55:1453-1458. 2. Spodick DH. Revascularization ofthe heart-nu- merators in search of denominators [editorial]. Am Heart J 1971;81:149-157, 3.ChalmersTC.EthicalaspectsofcJinicaItrials.Am J OphthalmoI1975;79:753-758. 4. Spodick DH. Randomize the first patient: scien- tific, ethical and behavioral bases. Am J Cardiol 1983;51:916-917. 5. SpodickDH. The randomized controlled clinical trial: scientific and ethical bases. Am J Med 1982; 73~420-425. 6.SpodickDH. Barrierstoacceptanceofcontrolled phase III clinical trials: behavioral factors. Biomed Pharmacother 1983;37:60-61. 7. DeWood MA, Spores J, Notske RN, Lang HT, Shields JP, Simpson CS, Rudy LW. Grunwald R. Medical and surgical management of myocardial infarction. Am J Cardiol 1979;44:1356-1364. 8. DeWood MA, Heit J, Spores J. Berg R Jr, Selin- ger SL, Rudy LW. Hensley GR, Shields JP. Anterior transmural myocardial infarction: effects of sur- gical coronary reperfusion on global and region- al left ventricular function. JACC 1983;1:1223- 1234.

9. DeWood M. Reply to Spodick DH [letter]. JACC 1983;2:1249-1242.

SAFETY OF IBUPROFEN FOR ACUTE MYOCARDIAL INFARCTION PERICARDITIS

Boden and Sadaniantzl present observations on 3 patients with ventricular septal rupture during ibuprofen therapy for acute myocar- dial infarction (AMI]-associated pericarditis. They hedge their message by pointing to its anecdotal nature, admitting “no conclusive proof that a causal relationship exists”1 They also appropriately cite Brown et al, 2 who ob- served scar thinning associated with ibupro- fen during experimental AMI. In the same issueofthe AJC, Cannonetalsreporttheben- eficial effect of ibuprofen duringexperimen- tal AMI, including promotion of collagen deuosition with no deleterious effect on sub- sequent scar formation. They point out that Brown et al2 used experimental con- ditions which may have promoted scar thinning.

The Editor had offered me the paper by Boden and Sadaniantz for review. I declined to review it because of involvement in a clini- cal trial of ibuprofen, which included mainly patients with AMI-associated acute pericar- ditis. Since 1978, we have used ibuprofen reg- ularly on a service with approximately 350 “rule ins” for AM1 each year, 6 to 7% with clinically recognizable pericarditis.4 Be- tween 1983 and 1985 we entered a formal study of ibuprofen in acute pericarditis against placebo. Thus, there were obviously many AM1 patients treated informally with ibuprofen, while 15 probably received the drug on protocol. In no patient did we diagnose ventricular septal rupture. More- over, we used larger doses than did Boden andSandaniantz:600mgevery6hoursusual- ly; 800 mg every 6 hours for protocol patients.

Because AM1 pericarditis is almost always unassociated with diagnostic electrocardio- graphic changes, the pericardial rub is the sole objective diagnostic sign of AM1 pericar- ditis4 (Pain alone is not reliable.) Thus, one ‘must assume that Boden and Sadaniantz identified their patients by detecting rubs. Auscultation and phonocardiography show thatpericardialrubsarenotrarelymonopha- sic and are best heard and recorded at the left mid- to low sternal border.5 It is in precisely the same area that precordial murmurs typi- cal of ventricular seotal defect with left-to- right shunt are best heard. An early “leak” of aseptaldefectmayhavebeeninterpretedasa pericardial rub. Yet, Boden is experienced in coronary artery diseases so we can assume thatthepatientsdidhaverubs.Therefore,the occurrence in a relatively short period of sev- eral patients with ventricular septal rupture wascertainlynoteworthy(i.e., publishable] if anecdotal. Although this obliged me to report here our entirely benign experience with large numbers of patients taking ibuprofen, we will remain alert to the possibility. How- ever, the report by Cannon et al3 makes ibu- profen an unlikely culprit.

David H. Spodick, MD Worcester, Massachusetts

3 July 1985