Does Prophylaxis Prevent Postdental Infective Endocarditis? A Controlled Evaluation of Protective Efficacy THOMAS F. IMPERIALE,M.D.C/~V~/ZI~, OMORALPH I. HORWITZ,M.D. NewHaven,Connecticut
PURPOSE: Despite the American Heart Association’s (AHA) recommendations for antibiotic prophylaxis to prevent infective endocarditis, no controlled clinical evidence exists for the effectiveness of this intervention. The purpose of this casecontrol study was to determine whether antibiotic prophylaxis for a dental procedure reduces the risk of infective endocarditis in persons with highrisk cardiac lesions. PATIENTSANDMETHODS: Cases consisted of eight subjects with high-risk lesions (six mitral, one aortic, one uncorrected tetralogy) whose first-time, native-valve infective endocarditis occurred within 12 weeks of a dental procedure and was diagnosed between 1980 and 1986. For each case subject, three control subjects were chosen from patients who underwent echocardiographic evaluation between 1980 and 1986, and who were matched for the specific high-risk lesion and age. Use of antibiotic pronhylaxis. which was determined bv interviews with patients and supplemented by the dentists, was defined as antibiotic taken both before and after the dental procedure. RESULTS: Antibiotic prophylaxis was used by only one of eight (13%) case subjects compared with 15 of 24 (63%) control subjects, for an odds ratio of 0.09, which is clinically impressive (indicating 91% protective efficacy) and statistically significant (p = 0.025). CONCLUSION: Although this report does not specifically assess the value of antibiotic prophylaxis for the current AHA recommendations, the use of antibiotic prophylaxis in persons with high-risk cardiac lesions is supported by the magnitude of protective efficacy observed in this study.
From the Departments of Medrcme and Epidemiology, Yale Un~versrty School of Medrcrne. and the Department of Medicine, Hosprtal of St. Raphael, New Haven, Connectrcut. Dr. lmperrale IS a former Robert Wood Johnson Clrnrcal Scholar. Requests for reprints should be addressed to Thomas F. Imperrale, M.D., Cleveland Metropolitan General Hosprtal, Hamann 311, 3395 Scranton Road, Cleveland, Ohio 44109. Manuscript submitted December 21, 1988, and accepted I” rewed form September 7, 1989.
A
lthough an uncommon disease and despite advances in antimicrobial therapy, postdental infective endocarditis remains a significant source of morbidity and mortality. Therefore, prevention, if possible, would be worthwhile. Between 25% and 40% of patients with endocarditis have a history of recent dental manipulation [l]. To prevent the occurrence of infective endocarditis, antibiotic prophylaxis has been the cornerstone of therapeutic strategy. Antimicrobial prophylaxis for prevention of infective endocarditis is one of the enduring and substantial controversies in clinical medicine. From 1979 to 1982, an American Heart Association (AHA) committee collected and recorded examples of apparent failures of endocarditis prophylaxis. Among the 52 cases, 48 (92%) occurred after a dental procedure; 39 (75%) were caused by viridans streptococci; and only six patients (12%) received regimens of antibiotic prophylaxis recommended by the AHA. These data appeared to indicate that endocarditis prophylaxis failures were more common than was previously believed [2]. Despite uncertainty over the efficacy of prophylaxis, the concept has achieved widespread acceptance and is recommended in all major textbooks of internal medicine, cardiology, infectious disease, and general surgery. Surveys of practitioners confirm this acceptance, and the AHA recommendations are referred to as the currently recognized standard of care for practicing physicians and dentists as well as for the medical-legal apparatus [3]. Since the evidence that endocarditis prophylaxis might be effective is indirect, questions still exist about its value for the individual patient and as health care policy. The purpose of this case-control study was to conduct a controlled evaluation to determine whether antibiotic prophylaxis for a dental procedure reduces the risk of developing infective endocarditis in persons with high-risk cardiac lesions. Among its many advantages, the case-control method is particularly useful for studying uncommon diseases, because cases are gathered unconstrained by the natural frequency of disease, and because far fewer subjects are required for a meaningful comparison. We assembled two groups of subjects: a case group of patients who acquired infective endocarditis after a dental procedure and a control group who did not. We then ascertained the relative frequency of antibiotic prophylaxis used in these two groups to provide a controlled estimate of the risk reduction with prophylaxis.
PATIENTS AND METHODS This study involved subjects from two hospitals: Yale-New Haven Hospital, an 850-bed facility that serves as the main teaching hospital for Yale University School of Medicine; and the Hospital of St. Raphael, a 480-bed Yale University-affiliated hospital locatFebruary
1990
The American
Journal
of Medicine
Volume
88
131
POSTDENTAL INFECTIVE ENDOCARDITIS / IMPERIALE AND HORWITZ
required. (The echocardiogram obtained during hospitalization for endocarditis could have served as the confirmatory test.); and (3) A dental procedure with high likelihood of causing transient bacteremia must have occurred within 12 weeks of hospital admission. We documented this procedure using medical records, patient interview, and dental records. Acceptable dental procedures were those likely to cause gingival bleeding, such as extractions, cleanings, fillings below the dentate line, oral surgery, and periodontal surgery [41*
TABLEI Diagnostic Criteria for Infective Endocarditis” These criteria require either pathologic documentation of infective endocarditis or the fulfillment of certain clinical criteria A. Pathologic documentation-based on histology from surgery or on bacteriology (Gram stain or culture) of valvular vegetation or peripheral embolus 6. Clinical documentation-one or more of the followinn: 1. Persistently positive blood culturest plus one of the following: (a) a changed murmur (b) predisposing heart disease and vascular phenomena* (c) predisposing heart disease, fever without an extracardiac source, and a murmur 2. For viridans streptococcal cases only: at least two positive blood cultures without an extracardiac source; fever 3. Negative or intermittently positive blood cultures§ with all three of the following: (a) fever (b) predisposing heart disease (c) vascular phenomena
Definition and Selection of Control Subjects Potential control subjects were selected from a roster of patients who underwent echocardiographic evaluation during the period covered by the study, Criteria for potential eligibility, which were similar to those in the case subjects, included the following: age at least 15 years at the time of the chosen dental procedure; no history of infective endocarditis; and a native highrisk cardiac lesion documented by echocardiogram. Three eligible controls were chosen for each case and were matched to the case for the type of specific valvular lesion and patient’s age within five years. Since the method of documentation of the high-risk lesion (either physical examination or echocardiogram) that preceded the dental procedure may have influenced the decision to administer prophylaxis, we also matched cases and controls for the method of documentation of the high-risk lesion.
Wodifled from [16]. + At leas1 lwo Dlooo cultures obtarnea With two ot two posltlve, three of three posltlve, or at least 70% cultures Dosibve if four cultures or more obtained. * Includes petechiae, splrnter hemorrhages, conjunctrval hemorrhages, Roth spots, 0s. ler’s nodes, Janeway lesions, splenomegaly, aseptlc meningltls, glomerulonephritls, and DUlmOnarV. central nervous svstem. or DercDheral emboli. b Any rate of blood culture dositivity that does not meet the definition of persistently positive.
ed in New Haven and in close proximity Haven Hospital.
to Yale-New
Definition and Selection of Case Subjects Case eligibility was determined using a two-step process: preliminary determinations was based on an abbreviated review of the medical record; final determination of eligibility was established after a detailed extraction of the medical record and completion of a telephone interview with patients and their dentists. At both hospitals, listings of patients with a discharge diagnosis of endocarditis from 1980 to 1986 were obtained from the medical records department. Charts were reviewed to identify cases of first-time, native-valve infective endocarditis occurring in subjects at least 15 years of age that was either caused by organisms comprising normal mouth flora (viridans streptococci, Haemophilus species other than Haemophilus influenzae, Corynebacterium hominis, Actinobacillus actinomycetemcomitans) or that was culturenegative in the absence of apparent primary sources of bacterial contamination (intravenous drug abuse, recent manipulation of the gastrointestinal or genitourinary tract, or cellulitis). The specific criteria for infective endocarditis are presented in Table I. For subjects fulfilling these preliminary criteria, three additional conditions were required. (1) An echocardiographic evaluation must have been performed either before or during the hospitalization for infective endocarditis to document the site and type of cardiac lesion; (2) A high-risk cardiac lesion or a murmur must have been documented and known by the patient before the dental procedure. The high-risk lesions included ventricular septal defects, rheumatic and other valvular dysfunction, idiopathic hypertrophic subaortic stenosis, and mitral valve prolapse (including both a click and a murmur) [4,5]. For the circumstance in which a murmur was documented by physical examination, subsequent confirmation that the murmur represented a high-risk lesion was 132
February
1990
The American
Journal
of Medicine
Volume
Exclusions Case and control subjects would have been excluded from the study if one or more of the following were present: (1) signs or symptoms consistent with infective endocarditis before or during the dental procedure; (2) use of antibiotics for any condition except rheumatic fever prophylaxis within 48 hours of the dental procedure (The regimen of continuous lowdose penicillin used for rheumatic fever prophylaxis is considered inadequate to prevent endocarditis [4].); (3) subjects (cases or controls) whose dentition would have necessitated administration of antibiotics regardless of their cardiac status; (4) subjects (cases or controls) with allergies to all the antibiotic regimens recommended by the AHA, who therefore would have been unlikely to receive customary antibiotic prophylaxis; and (5) subjects (cases) who had undergone any other procedure for which prophylaxis has been suggested by the AHA subsequent to the dental procedure and before the development of endocarditis. Choice of a Specified Zero Time Each case subject with infective endocarditis and each control subject must have a chronologic reference point at which the pretherapeutic baseline state is characterized, and the principal treatment (antibiotic prophylaxis) is classified. This point, which we refer to as zero time, corresponds best to the time at which randomization occurs in a clinical trial. For this study, zero time was chosen as the day of the dental procedure. Ascertainment of Antibiotic Use and Other Pertinent Data Potentially eligible subjects (cases and controls) received a letter describing the general purpose of the 88
POSTDENTAL
INFECTIVE
ENDOCARDITIS
/ IMPERIALE
AND HORWITZ
TABLE II Description of Case Subjects with Postdental Infective Endocarditis
Case Subject Number
Age and Sex
40 M 41 F 43 M : 53 M 67 F 2 79 M 80 M L 88 F i,D - m,+r?.l/,~l>,,.nrnl^nr^. h”D - -,+rnl r^ni,rn:+^+,^“. :
High-Risk Cardiac Lesion
Type of Dental Procedure
MVP, MR MVP ;;q;
Cleaning Cleaning Cleaning Cleaning Filling Cleaning Cleaning Extraction
ASiMR Ii: MS nc - -^-I:^ r+^^^r:r
Interval between Dental Procedure and Diagnosis (weeks)
Organism Causing Infective Endocarditis Culture-negative Streptococcus Streptococcus Streptococcus Culture-negative Strep tococcus Strep tococcus Streptococcus
i 4 5 :; :
vhdans vidans viridans viridans vidans vhdans
LIC - -,I”.. ?+^^^~,..
research as a health survey of medical and dental care. Accompanying the letter was a release-of-information form so that dental records could be reviewed. Soon afterwards, subjects were interviewed by telephone to determine the source of dental care for the focal time procedure, and pertinent medical and dental history, particularly the frequency of and reasons for prior dental visits and patterns of antibiotic use for the visits. The interviews were conducted so that the specific purpose of the research was not revealed to the interviewed subjects-questions regarding dental history and use of antibiotic prophylaxis were accompanied by questions regarding medical history and the taking of prescribed medications. Next, the dentists of case and control subjects were contacted to provide supplementary and confirmatory information about dental history and use of antibiotic prophylaxis. When dental records did not uniformly record whether prophylaxis was given, we accepted the subject’s recall regarding the use of antibiotic prophylaxis (both customary use and specifically for the zero-time dental procedure) as the basis for classifying exposure to prophylaxis. We considered case and control subjects “exposed” to antibiotic prophylaxis if they reported taking one of the AHA-recommended antibiotics both before and after the dental procedure regardless of dose or duration of use. Strategy of Data Analysis and Sample Size Estimates When a rarely occurring disease such as infective endocarditis is investigated with the case-control method, the odds ratio approximates the risk ratio that would have been derived from a cohort study comparing groups with and without the exposure [6]. (In this case, “exposure” is antibiotic prophylaxis.) To obtain a controlled measure of the (protective) effect of antibiotic prophylaxis in persons with native highrisk lesions, we calculated the odds ratio from a matched analysis of the data according to standard techniques [7]. To test the statistical significance of this odds ratio, we used the chi-square method derived by Mantel and Haenszel[8]. In addition, we calculated the more familiar unmatched odds ratio and its upper 95% confidence limit according to Cornfield’s [9] exact method. Finally, we estimated the protective efficacy of antibiotic prophylaxis that describes the proportion of infective endocarditis that would have been prevented by antibiotic prophylaxis. Protective efficacy is defined as “l-relative risk.” Since the odds ratio in a case-control study approximates the relative risk, the
protective efficacy can be estimated by the quantity: l-odds ratio [lo]. RESULTS By record review, we found 34 cases of infective endocarditis that met the following inclusion criteria: age at least 15 years; first-time, native-valve endocarditis either caused by mouth flora or considered culture-negative without an apparent primary source of bacterial contamination; a high-risk cardiac lesion or murmur whose documentation preceded admission; and echocardiographic evidence of a high-risk lesion. Fourteen cases could not be included because no dental procedure had occurred within 12 weeks of hospitalization. Of the remaining 20 cases, 12 were not included because we could not determine whether a dental procedure had occurred for the following reasons: seven subjects had died; three had changed address and could not be located; and two refused to participate. Thus, eight patients had undergone a dental procedure 12 weeks before diagnosis and became the final case group. There were no further exclusions within this group. The case group of eight subjects, described in Table II, consisted of five men and three women. The average age at the time of hospitalization for endocarditis was 61 years (range, 40 to 88 years). Six case subjects had cardiac lesions involving the mitral valve alone, one case subject had an uncorrected tetralogy of Fallot, and one case subject had both mitral and aortic valve lesions. The dental procedures implicated in causing endocarditis were cleaning and scaling in six cases, and extraction and filling in one case each. The average interval between dental procedure and diagnosis of endocarditis was six weeks (range, three to 12 weeks). Streptococcus viridans was isolated from the blood of six case subjects. The remaining two had culture-negative endocarditis. Both culture-negative patients received a course of antibiotics subsequent to dental treatment for early manifestations of endocarditis. In Table III, the eight case and 24 control subjects are compared for several important clinical features. As would be expected from the matching procedure, the two groups were similar in age and type of valvular lesion. There were no statistically significant differences between the groups for history of rheumatic fever, murmur duration, frequency of dental visits, type of dental procedure for which exposure was ascertained, and use of digitalis or diuretics. None of the
February
1990
The American
Journal
of Medicine
Volume
88
133
POSTDENTAL
INFECTIVE
ENDOCARDITIS
/ IMPERIALE
AND HORWITZ
TABLE III
~
Selected Characteristics of Case and Control Subjects Cases AMge:mean f SD)
(n = 8)
Controls (n = 24)
61 f 19
57 f 16
ll(46)
5 (63)* Cardiac lesion Mitral valve Mitral and aortic valves Tetralogy History of rheumatic fever Murmur duration lO years Frequency of dental visits 2 per year Dental cleaning as zero time visit Use of diuretics Use of digitalis Penicillin allergy
f gg : {q 0 (0) : jtg 6 (75)
COMMENTS Developments in the field of prevention of infective endocarditis have been hampered by a lack of information on which to base recommendations. Reliable data have not been available to answer basic questions regarding the risk of developing endocarditis, those procedures and operations that should be covered by antibiotics, and whether antibiotic prophylaxis works. In this study, we have used the case-control method to determine whether prophylaxis reduces the risk of developing endocarditis after a dental procedure. We found that only one of eight case subjects with postdental endocarditis received prophylaxis (defined as antibiotic taken both before and after the dental procedure), whereas 15 of 24 control subjects received prophylaxis, for an odds ratio of 0.09, which is clinically impressive (indicating 91% protective efficacy) and statistically significant (p = 0.025). These results, which support the use of prophylaxis in persons with high-risk lesions who undergo dental procedures with a propensity to cause bacteremia, agree with those of Horstkotte et al Ill], who conducted a retrospective study and demonstrated a benefit of prophylaxis for infective endocarditis in patients with prosthetic heart valves. Our results may appear to conflict with those from the AHA registry study [2], but we do not believe that this is the case. The AHA registry analysis was a case series of 52 apparent failures of endocarditis prophylaxis that were reported to a national registry during a four-year period. Since the denominator of all persons at risk is unknown, a “rate” of prophylaxis failure cannot be calculated. In addition, there was no comparison group-neither the number nor the “rate” of cases of endocarditis that occurred without the use of antibiotic prophylaxis was recorded. There are several methodologic limitations of the present study. One important limitation is sample size. Our case group is limited in size largely because of our strict inclusion criteria. Although the analysis of eight cases and 24 controls resulted in a clinically important and statistically significant difference, this difference may be fragile. Misclassification of one unexposed case subject or three or more exposed control
-igures ~nparentheses denote percentages.
TABLE IV Association between Antibiotic Prophylaxis and Postdental Infective Endocarditis: Matched Results Number of Controlswith Antibiotic Prophylaxis 0 1 2 3 Antibiotic prophylaxis in cases No antibiotic prophylaxis in cases Total matched sets
0 1
i
1
3
;
2
Total Matched Sets
i
:
2
8
dds ratio (matched) = O/13 = 0. antel-Haenszel chi-square = 6.00, p = 0.025.
TABLE V Association between Antibiotic Prophylaxis and Postdental Infective Endocarditis: Unmatched Results Cases
Controls 15
Antibiotic prophylaxis No antibiotic prophylaxis Total
:
8
2:
jds ratio (and upper 95% confidence Ilmlt) = 0.09 (0.93). otectlve efficacy = (l-odds ratio) = 0.91.
case subjects were allergic to penicillin whereas five (21%) control subjects reported penicillin allergy. Tables IV and V list the overall results of the study in both matched and unmatched format. The matched odds ratio is calculated from only those matched sets that are discordant for use of antibiotic prophylaxis. Since the only exposed case subject that received antibiotic prophylaxis was matched to three control subjects, all of whom received antibiotic prophylaxis, there are no discordant sets in the numerator (Table IV). The matched odds ratio of zero, although statisti-
134
February
1990
cally significant, does not provide a suitable point estimate of the risk reduction of endocarditis with prophylaxis. (If it did, it would imply that antibiotic prophylaxis was 100% protective.) The more familiar unmatched analysis is displayed in Table V. Only one of eight cases (12.5%) received prophylaxis whereas 15 of 24 controls (62.5%) received prophylaxis. To obtain a reasonable point estimate of the risk reduction of infective endocarditis with antibiotic prophylaxis, we computed the more familiar unmatched odds ratio as 0.09 (Table V). To calculate the most conservative confidence limit for this point estimate, we used Cornfield’s [9] exact method, which provided only the upper (and more important) 95% confidence limit of 0.91. From the unmatched odds ratio of 0.09, we determined protective efficacy to be (1 - 0.09) = 0.91 or 91%. The results are essentially unchanged when the culture-negative case subjects and their control subjects are excluded: the unmatched odds ratio is 0.08 (p = 0.028 by Fisher’s exact test, onetailed), and protective efficacy is 92%.
The American
Journal
of Medicine
Volume
88
POSTDENTAL
subjects, or both, would reduce the protective efficacy and eliminate statistical significance of the unmatched analysis. On the other hand, the matched analysis is more resistant to misclassification of subjects. In addition, that we achieved meaningful differences with this small case series indicates the substantial magnitude of efficacy of antibiotic prophylaxis. A more important consequence of the small sample size may be the inability to perform subgroup analyses and to calculate the protective efficacy of antibiotic prophylaxis for different high-risk lesions and for different antibiotic regimens. A second possible limitation of this study may be our acceptance of a subject’s recall regarding use of antibiotic prophylaxis as the basis for ascertaining exposure. In designing this study, we had planned to obtain this information solely from dental records. It became clear, however, that some patients take prophylaxis without notifying their dentists, and, more commonly, that this information is not routinely recorded in the dental record. Information from dental records and from patient memory regarding prophylaxis was in agreement for all eight case subjects and for 19 of 24 control subjects. The remaining five control subjects recalled using antibiotic prophylaxis routinely before “every” dental procedure, whereas their records did not indicate that prophylaxis had been taken. There were no instances in which dental records indicated use of prophylaxis but subjects denied using it. A third limitation of this study may be that our results do not necessarily represent the protective efficacy of the AHA-recommended regimens. We accepted as “prophylaxis” the use of any AHA-recommended antibiotic both before and after the dental procedure regardless of dose or duration of use. We used this definition of prophylaxis because it is impossible to know with certainty whether all subjects taking antibiotic prophylaxis did so according to AHA recommendations. Although several subjects remembered exactly how they used prophylaxis, others were less certain of dose and duration of use, but could readily recall using prophylaxis both before and after dental procedures. Moreover, our findings do not specifically reflect the efficacy of the currently recommended regimens since the study period spans two eras of recommendations for antibiotic prophylaxis. This final limitation, which may reduce the generalizability of the study results, does not affect its validity. We believe that the strength of this analysis lies in the care with which cases and controls were matched for critical clinical features. We matched cases and controls for features related both to the risk of developing infective endocarditis and to the likelihood of receiving prophylaxis. Most important among these variables was the particular cardiac lesion, since clinicians may consider some lesions to be more “high risk” than others and thus selectively administer prophylaxis to patients with such lesions. Since we were concerned that patients who had prior documentation of a high-risk lesion by echocardiography might be more likely to receive prophylaxis than patients who had a high-risk lesion documented by auscultation alone, we matched cases and controls so that a similar mode of documentation of the high-risk lesions existed before the dental procedure in ques-
INFECTIVE
ENDOCARDITIS
/ IMPERIALE
AND HORWITZ
tion. In support of matching for this variable is the fact that several control subjects with high-risk lesions by auscultation never received antibiotic prophylaxis before their echocardiogram, but had received it routinely subsequent to echocardiography. These data suggest that if we had ignored this variable, a higher proportion of control subjects would have received prophylaxis, and protective efficacy would have appeared to be even greater. Our results support the use of antibiotic prophylaxis by persons with high-risk cardiac lesions in the dental setting, and better justify it as a preventive strategy. However, prophylaxis is not without side effects, which range from gastrointestinal upset and skin rash to death from anaphylaxis. In a review of the world literature, Idsoe et al [12] estimated the incidence of skin rashes and urticaria at 0.7% to lo%, and the incidence of anaphylaxis at 0.015% to 0.04%. More recent estimates range from 2.4 mild reactions per thousand courses of oral penicillin (range, 2.0 to 7.0) to 9.0 fatal reactions per million courses of oral penicillin (range, 6.0 to 11) [13]. Although these probabilities are small from the individual’s perspective, they represent a significant risk from a societal perspective because of the frequency with which measures to prevent endocarditis are used. In certain instances, the probability of experiencing a serious reaction to penicillin may exceed that of developing endocarditis without prophylaxis. Current estimates of the risk of developing endocarditis in persons with high-risk lesions without prophylaxis are imprecise, and range from 1 in 500 [14] to 1 in 115,000 [15]. Clearly, the risks and benefits of antibiotic prophylaxis need to be more precisely defined. It is necessary to explore the possibilities that prophylactic efficacy may depend on the particular antibiotic regimen or valvular lesion. It may be that certain regimens prevent endocarditis for specific cardiac lesions or that other regimens offer good overall protection. Such information is required to maximize the benefit and minimize the risk of this preventive strategy.
ACKNOWLEDGMENT We are indebted to Elizabeth T. Pesapane for her special care and expertise preparation of this manuscript.
in
REFERENCES 1. Littneo MM, Kaffe I, Tamse A, Buchner A: New concepts in chemoprophylaxis of bacterial endocarditis resulting from dental treatment. Oral Surg Oral Med Oral Pathol 1986; 61: 338-342. 2. Durack DT, Kaplan EL, Bisno AL: Apparent failures of endocardrtrs prophylaxis. JAMA 1983; 258: 2318-2322. 3. Kaplan EL: Bacterial endocarditis prophylaxis: tradition or necessrty? Am J Cardiol 1986; 57: 478479. 4. American Heart Association Commrttee Report: Prevention of bactenal endocarditis. Circulation 1984; 70: 1123A-1127A. 5. MacMahon SW, Hickey AJ, Wilcken DEL, et al: Rusk of Infective endocarditis in mitral valve prolapse with and without precordial systolic murmurs, Am J Cardiol 1986; 58. 105-108. 6. Horwitz RI: The experimental paradigm and observational studres of causeeffect relattonships in clrnical medicine. J Chronic Dis 1987; 40: 91-99. 7. Schlesselman JJ: Case-control studies. New York: Oxford University Press, 1982. 8. Flerss JL: Stabsbcal methods for rates and proportions. New York: John Wiley and Sons, 1981; 123-126. 9. Cornfield J: A statistrcal problem ansing from retrospective studres. In. Neyman J, ed. Proceedings of the Third Berkeley Symposium on Mathematical Stabs&s and Probabrlrties. Berkeley: University of Cakfornia Press, 1956; 135-148. 10. Clemens JD. Shapiro ED: Resolving the pneumococcal vaccrne controversy:
February
1990
The American
Journal
of Medicine
Volume
88
135
POSTDENTAL gainere
INFECTIVE
alternatives
ENDOCARDITIS
to randomized
/ IMPERIALE
AND HORWITZ
clinical trials? Rev Infect Dis 1984; 6: 589-
prophylaxis for pabents with mitral valve prolapse. J Chronic Dis 1984; 37: 531544. 14. Kelson SR, White PD: Notes on 250 cases of subacute bacterial (streptococcal) endocarditis studied and treated between 1927 and 1939. Ann Intern Med 1945; 22: 40-60. 15. Pogrel MA, Welsby PD: The dentist and prevention of Infective endocarditrs. Br Dent J 1975: 139: 12-16. 16.Von Reyn CF, Levy BS, Arbeit RD, et al: Infective endocarditis: an analysis based on strict case definitions. Ann Intern Med 1981; 94: 505-518.
11. Horstkotte D. Friedrichs W, Pippert H, Bircks W, Loogen F: Benefit of prophylaxis for infectious endocarditis in patients with prosthetic heart valves. 2 Kardiol 1986; 75: 8-l 1. 12. ldsoe 0, Guthe T, Wilcox L, DeWeck AL: Nature and extent of penicillin srdereactions, with particular reference to fatalities from anaphylactic shock. Bull WHO 1968; 38: 159-188. 13. Clemens JD, Ransohoff DF: A quantitative assessment of pre-dental antibiotic
136
February
1990
The American
Journal
of Medicine
Volume
88
PURPOSE: Despite the American Heart Association’s (AHA) recommendations for antibiotic prophylaxis to prevent infective endocarditis, no controlled clinical evidence exists for the effectiveness of this intervention. The purpose of this casecontrol study was to determine whether antibiotic prophylaxis for a dental procedure reduces the risk of infective endocarditis in persons with highrisk cardiac lesions. PATIENTSANDMETHODS: Cases consisted of eight subjects with high-risk lesions (six mitral, one aortic, one uncorrected tetralogy) whose first-time, native-valve infective endocarditis occurred within 12 weeks of a dental procedure and was diagnosed between 1980 and 1986. For each case subject, three control subjects were chosen from patients who underwent echocardiographic evaluation between 1980 and 1986, and who were matched for the specific high-risk lesion and age. Use of antibiotic pronhylaxis. which was determined bv interviews with patients and supplemented by the dentists, was defined as antibiotic taken both before and after the dental procedure. RESULTS: Antibiotic prophylaxis was used by only one of eight (13%) case subjects compared with 15 of 24 (63%) control subjects, for an odds ratio of 0.09, which is clinically impressive (indicating 91% protective efficacy) and statistically significant (p = 0.025). CONCLUSION: Although this report does not specifically assess the value of antibiotic prophylaxis for the current AHA recommendations, the use of antibiotic prophylaxis in persons with high-risk cardiac lesions is supported by the magnitude of protective efficacy observed in this study.
From the Departments of Medrcme and Epidemiology, Yale Un~versrty School of Medrcrne. and the Department of Medicine, Hosprtal of St. Raphael, New Haven, Connectrcut. Dr. lmperrale IS a former Robert Wood Johnson Clrnrcal Scholar. Requests for reprints should be addressed to Thomas F. Imperrale, M.D., Cleveland Metropolitan General Hosprtal, Hamann 311, 3395 Scranton Road, Cleveland, Ohio 44109. Manuscript submitted December 21, 1988, and accepted I” rewed form September 7, 1989.
A
lthough an uncommon disease and despite advances in antimicrobial therapy, postdental infective endocarditis remains a significant source of morbidity and mortality. Therefore, prevention, if possible, would be worthwhile. Between 25% and 40% of patients with endocarditis have a history of recent dental manipulation [l]. To prevent the occurrence of infective endocarditis, antibiotic prophylaxis has been the cornerstone of therapeutic strategy. Antimicrobial prophylaxis for prevention of infective endocarditis is one of the enduring and substantial controversies in clinical medicine. From 1979 to 1982, an American Heart Association (AHA) committee collected and recorded examples of apparent failures of endocarditis prophylaxis. Among the 52 cases, 48 (92%) occurred after a dental procedure; 39 (75%) were caused by viridans streptococci; and only six patients (12%) received regimens of antibiotic prophylaxis recommended by the AHA. These data appeared to indicate that endocarditis prophylaxis failures were more common than was previously believed [2]. Despite uncertainty over the efficacy of prophylaxis, the concept has achieved widespread acceptance and is recommended in all major textbooks of internal medicine, cardiology, infectious disease, and general surgery. Surveys of practitioners confirm this acceptance, and the AHA recommendations are referred to as the currently recognized standard of care for practicing physicians and dentists as well as for the medical-legal apparatus [3]. Since the evidence that endocarditis prophylaxis might be effective is indirect, questions still exist about its value for the individual patient and as health care policy. The purpose of this case-control study was to conduct a controlled evaluation to determine whether antibiotic prophylaxis for a dental procedure reduces the risk of developing infective endocarditis in persons with high-risk cardiac lesions. Among its many advantages, the case-control method is particularly useful for studying uncommon diseases, because cases are gathered unconstrained by the natural frequency of disease, and because far fewer subjects are required for a meaningful comparison. We assembled two groups of subjects: a case group of patients who acquired infective endocarditis after a dental procedure and a control group who did not. We then ascertained the relative frequency of antibiotic prophylaxis used in these two groups to provide a controlled estimate of the risk reduction with prophylaxis.
PATIENTS AND METHODS This study involved subjects from two hospitals: Yale-New Haven Hospital, an 850-bed facility that serves as the main teaching hospital for Yale University School of Medicine; and the Hospital of St. Raphael, a 480-bed Yale University-affiliated hospital locatFebruary
1990
The American
Journal
of Medicine
Volume
88
131
POSTDENTAL INFECTIVE ENDOCARDITIS / IMPERIALE AND HORWITZ
required. (The echocardiogram obtained during hospitalization for endocarditis could have served as the confirmatory test.); and (3) A dental procedure with high likelihood of causing transient bacteremia must have occurred within 12 weeks of hospital admission. We documented this procedure using medical records, patient interview, and dental records. Acceptable dental procedures were those likely to cause gingival bleeding, such as extractions, cleanings, fillings below the dentate line, oral surgery, and periodontal surgery [41*
TABLEI Diagnostic Criteria for Infective Endocarditis” These criteria require either pathologic documentation of infective endocarditis or the fulfillment of certain clinical criteria A. Pathologic documentation-based on histology from surgery or on bacteriology (Gram stain or culture) of valvular vegetation or peripheral embolus 6. Clinical documentation-one or more of the followinn: 1. Persistently positive blood culturest plus one of the following: (a) a changed murmur (b) predisposing heart disease and vascular phenomena* (c) predisposing heart disease, fever without an extracardiac source, and a murmur 2. For viridans streptococcal cases only: at least two positive blood cultures without an extracardiac source; fever 3. Negative or intermittently positive blood cultures§ with all three of the following: (a) fever (b) predisposing heart disease (c) vascular phenomena
Definition and Selection of Control Subjects Potential control subjects were selected from a roster of patients who underwent echocardiographic evaluation during the period covered by the study, Criteria for potential eligibility, which were similar to those in the case subjects, included the following: age at least 15 years at the time of the chosen dental procedure; no history of infective endocarditis; and a native highrisk cardiac lesion documented by echocardiogram. Three eligible controls were chosen for each case and were matched to the case for the type of specific valvular lesion and patient’s age within five years. Since the method of documentation of the high-risk lesion (either physical examination or echocardiogram) that preceded the dental procedure may have influenced the decision to administer prophylaxis, we also matched cases and controls for the method of documentation of the high-risk lesion.
Wodifled from [16]. + At leas1 lwo Dlooo cultures obtarnea With two ot two posltlve, three of three posltlve, or at least 70% cultures Dosibve if four cultures or more obtained. * Includes petechiae, splrnter hemorrhages, conjunctrval hemorrhages, Roth spots, 0s. ler’s nodes, Janeway lesions, splenomegaly, aseptlc meningltls, glomerulonephritls, and DUlmOnarV. central nervous svstem. or DercDheral emboli. b Any rate of blood culture dositivity that does not meet the definition of persistently positive.
ed in New Haven and in close proximity Haven Hospital.
to Yale-New
Definition and Selection of Case Subjects Case eligibility was determined using a two-step process: preliminary determinations was based on an abbreviated review of the medical record; final determination of eligibility was established after a detailed extraction of the medical record and completion of a telephone interview with patients and their dentists. At both hospitals, listings of patients with a discharge diagnosis of endocarditis from 1980 to 1986 were obtained from the medical records department. Charts were reviewed to identify cases of first-time, native-valve infective endocarditis occurring in subjects at least 15 years of age that was either caused by organisms comprising normal mouth flora (viridans streptococci, Haemophilus species other than Haemophilus influenzae, Corynebacterium hominis, Actinobacillus actinomycetemcomitans) or that was culturenegative in the absence of apparent primary sources of bacterial contamination (intravenous drug abuse, recent manipulation of the gastrointestinal or genitourinary tract, or cellulitis). The specific criteria for infective endocarditis are presented in Table I. For subjects fulfilling these preliminary criteria, three additional conditions were required. (1) An echocardiographic evaluation must have been performed either before or during the hospitalization for infective endocarditis to document the site and type of cardiac lesion; (2) A high-risk cardiac lesion or a murmur must have been documented and known by the patient before the dental procedure. The high-risk lesions included ventricular septal defects, rheumatic and other valvular dysfunction, idiopathic hypertrophic subaortic stenosis, and mitral valve prolapse (including both a click and a murmur) [4,5]. For the circumstance in which a murmur was documented by physical examination, subsequent confirmation that the murmur represented a high-risk lesion was 132
February
1990
The American
Journal
of Medicine
Volume
Exclusions Case and control subjects would have been excluded from the study if one or more of the following were present: (1) signs or symptoms consistent with infective endocarditis before or during the dental procedure; (2) use of antibiotics for any condition except rheumatic fever prophylaxis within 48 hours of the dental procedure (The regimen of continuous lowdose penicillin used for rheumatic fever prophylaxis is considered inadequate to prevent endocarditis [4].); (3) subjects (cases or controls) whose dentition would have necessitated administration of antibiotics regardless of their cardiac status; (4) subjects (cases or controls) with allergies to all the antibiotic regimens recommended by the AHA, who therefore would have been unlikely to receive customary antibiotic prophylaxis; and (5) subjects (cases) who had undergone any other procedure for which prophylaxis has been suggested by the AHA subsequent to the dental procedure and before the development of endocarditis. Choice of a Specified Zero Time Each case subject with infective endocarditis and each control subject must have a chronologic reference point at which the pretherapeutic baseline state is characterized, and the principal treatment (antibiotic prophylaxis) is classified. This point, which we refer to as zero time, corresponds best to the time at which randomization occurs in a clinical trial. For this study, zero time was chosen as the day of the dental procedure. Ascertainment of Antibiotic Use and Other Pertinent Data Potentially eligible subjects (cases and controls) received a letter describing the general purpose of the 88
POSTDENTAL
INFECTIVE
ENDOCARDITIS
/ IMPERIALE
AND HORWITZ
TABLE II Description of Case Subjects with Postdental Infective Endocarditis
Case Subject Number
Age and Sex
40 M 41 F 43 M : 53 M 67 F 2 79 M 80 M L 88 F i,D - m,+r?.l/,~l>,,.nrnl^nr^. h”D - -,+rnl r^ni,rn:+^+,^“. :
High-Risk Cardiac Lesion
Type of Dental Procedure
MVP, MR MVP ;;q;
Cleaning Cleaning Cleaning Cleaning Filling Cleaning Cleaning Extraction
ASiMR Ii: MS nc - -^-I:^ r+^^^r:r
Interval between Dental Procedure and Diagnosis (weeks)
Organism Causing Infective Endocarditis Culture-negative Streptococcus Streptococcus Streptococcus Culture-negative Strep tococcus Strep tococcus Streptococcus
i 4 5 :; :
vhdans vidans viridans viridans vidans vhdans
LIC - -,I”.. ?+^^^~,..
research as a health survey of medical and dental care. Accompanying the letter was a release-of-information form so that dental records could be reviewed. Soon afterwards, subjects were interviewed by telephone to determine the source of dental care for the focal time procedure, and pertinent medical and dental history, particularly the frequency of and reasons for prior dental visits and patterns of antibiotic use for the visits. The interviews were conducted so that the specific purpose of the research was not revealed to the interviewed subjects-questions regarding dental history and use of antibiotic prophylaxis were accompanied by questions regarding medical history and the taking of prescribed medications. Next, the dentists of case and control subjects were contacted to provide supplementary and confirmatory information about dental history and use of antibiotic prophylaxis. When dental records did not uniformly record whether prophylaxis was given, we accepted the subject’s recall regarding the use of antibiotic prophylaxis (both customary use and specifically for the zero-time dental procedure) as the basis for classifying exposure to prophylaxis. We considered case and control subjects “exposed” to antibiotic prophylaxis if they reported taking one of the AHA-recommended antibiotics both before and after the dental procedure regardless of dose or duration of use. Strategy of Data Analysis and Sample Size Estimates When a rarely occurring disease such as infective endocarditis is investigated with the case-control method, the odds ratio approximates the risk ratio that would have been derived from a cohort study comparing groups with and without the exposure [6]. (In this case, “exposure” is antibiotic prophylaxis.) To obtain a controlled measure of the (protective) effect of antibiotic prophylaxis in persons with native highrisk lesions, we calculated the odds ratio from a matched analysis of the data according to standard techniques [7]. To test the statistical significance of this odds ratio, we used the chi-square method derived by Mantel and Haenszel[8]. In addition, we calculated the more familiar unmatched odds ratio and its upper 95% confidence limit according to Cornfield’s [9] exact method. Finally, we estimated the protective efficacy of antibiotic prophylaxis that describes the proportion of infective endocarditis that would have been prevented by antibiotic prophylaxis. Protective efficacy is defined as “l-relative risk.” Since the odds ratio in a case-control study approximates the relative risk, the
protective efficacy can be estimated by the quantity: l-odds ratio [lo]. RESULTS By record review, we found 34 cases of infective endocarditis that met the following inclusion criteria: age at least 15 years; first-time, native-valve endocarditis either caused by mouth flora or considered culture-negative without an apparent primary source of bacterial contamination; a high-risk cardiac lesion or murmur whose documentation preceded admission; and echocardiographic evidence of a high-risk lesion. Fourteen cases could not be included because no dental procedure had occurred within 12 weeks of hospitalization. Of the remaining 20 cases, 12 were not included because we could not determine whether a dental procedure had occurred for the following reasons: seven subjects had died; three had changed address and could not be located; and two refused to participate. Thus, eight patients had undergone a dental procedure 12 weeks before diagnosis and became the final case group. There were no further exclusions within this group. The case group of eight subjects, described in Table II, consisted of five men and three women. The average age at the time of hospitalization for endocarditis was 61 years (range, 40 to 88 years). Six case subjects had cardiac lesions involving the mitral valve alone, one case subject had an uncorrected tetralogy of Fallot, and one case subject had both mitral and aortic valve lesions. The dental procedures implicated in causing endocarditis were cleaning and scaling in six cases, and extraction and filling in one case each. The average interval between dental procedure and diagnosis of endocarditis was six weeks (range, three to 12 weeks). Streptococcus viridans was isolated from the blood of six case subjects. The remaining two had culture-negative endocarditis. Both culture-negative patients received a course of antibiotics subsequent to dental treatment for early manifestations of endocarditis. In Table III, the eight case and 24 control subjects are compared for several important clinical features. As would be expected from the matching procedure, the two groups were similar in age and type of valvular lesion. There were no statistically significant differences between the groups for history of rheumatic fever, murmur duration, frequency of dental visits, type of dental procedure for which exposure was ascertained, and use of digitalis or diuretics. None of the
February
1990
The American
Journal
of Medicine
Volume
88
133
POSTDENTAL
INFECTIVE
ENDOCARDITIS
/ IMPERIALE
AND HORWITZ
TABLE III
~
Selected Characteristics of Case and Control Subjects Cases AMge:mean f SD)
(n = 8)
Controls (n = 24)
61 f 19
57 f 16
ll(46)
5 (63)* Cardiac lesion Mitral valve Mitral and aortic valves Tetralogy History of rheumatic fever Murmur duration lO years Frequency of dental visits 2 per year Dental cleaning as zero time visit Use of diuretics Use of digitalis Penicillin allergy
f gg : {q 0 (0) : jtg 6 (75)
COMMENTS Developments in the field of prevention of infective endocarditis have been hampered by a lack of information on which to base recommendations. Reliable data have not been available to answer basic questions regarding the risk of developing endocarditis, those procedures and operations that should be covered by antibiotics, and whether antibiotic prophylaxis works. In this study, we have used the case-control method to determine whether prophylaxis reduces the risk of developing endocarditis after a dental procedure. We found that only one of eight case subjects with postdental endocarditis received prophylaxis (defined as antibiotic taken both before and after the dental procedure), whereas 15 of 24 control subjects received prophylaxis, for an odds ratio of 0.09, which is clinically impressive (indicating 91% protective efficacy) and statistically significant (p = 0.025). These results, which support the use of prophylaxis in persons with high-risk lesions who undergo dental procedures with a propensity to cause bacteremia, agree with those of Horstkotte et al Ill], who conducted a retrospective study and demonstrated a benefit of prophylaxis for infective endocarditis in patients with prosthetic heart valves. Our results may appear to conflict with those from the AHA registry study [2], but we do not believe that this is the case. The AHA registry analysis was a case series of 52 apparent failures of endocarditis prophylaxis that were reported to a national registry during a four-year period. Since the denominator of all persons at risk is unknown, a “rate” of prophylaxis failure cannot be calculated. In addition, there was no comparison group-neither the number nor the “rate” of cases of endocarditis that occurred without the use of antibiotic prophylaxis was recorded. There are several methodologic limitations of the present study. One important limitation is sample size. Our case group is limited in size largely because of our strict inclusion criteria. Although the analysis of eight cases and 24 controls resulted in a clinically important and statistically significant difference, this difference may be fragile. Misclassification of one unexposed case subject or three or more exposed control
-igures ~nparentheses denote percentages.
TABLE IV Association between Antibiotic Prophylaxis and Postdental Infective Endocarditis: Matched Results Number of Controlswith Antibiotic Prophylaxis 0 1 2 3 Antibiotic prophylaxis in cases No antibiotic prophylaxis in cases Total matched sets
0 1
i
1
3
;
2
Total Matched Sets
i
:
2
8
dds ratio (matched) = O/13 = 0. antel-Haenszel chi-square = 6.00, p = 0.025.
TABLE V Association between Antibiotic Prophylaxis and Postdental Infective Endocarditis: Unmatched Results Cases
Controls 15
Antibiotic prophylaxis No antibiotic prophylaxis Total
:
8
2:
jds ratio (and upper 95% confidence Ilmlt) = 0.09 (0.93). otectlve efficacy = (l-odds ratio) = 0.91.
case subjects were allergic to penicillin whereas five (21%) control subjects reported penicillin allergy. Tables IV and V list the overall results of the study in both matched and unmatched format. The matched odds ratio is calculated from only those matched sets that are discordant for use of antibiotic prophylaxis. Since the only exposed case subject that received antibiotic prophylaxis was matched to three control subjects, all of whom received antibiotic prophylaxis, there are no discordant sets in the numerator (Table IV). The matched odds ratio of zero, although statisti-
134
February
1990
cally significant, does not provide a suitable point estimate of the risk reduction of endocarditis with prophylaxis. (If it did, it would imply that antibiotic prophylaxis was 100% protective.) The more familiar unmatched analysis is displayed in Table V. Only one of eight cases (12.5%) received prophylaxis whereas 15 of 24 controls (62.5%) received prophylaxis. To obtain a reasonable point estimate of the risk reduction of infective endocarditis with antibiotic prophylaxis, we computed the more familiar unmatched odds ratio as 0.09 (Table V). To calculate the most conservative confidence limit for this point estimate, we used Cornfield’s [9] exact method, which provided only the upper (and more important) 95% confidence limit of 0.91. From the unmatched odds ratio of 0.09, we determined protective efficacy to be (1 - 0.09) = 0.91 or 91%. The results are essentially unchanged when the culture-negative case subjects and their control subjects are excluded: the unmatched odds ratio is 0.08 (p = 0.028 by Fisher’s exact test, onetailed), and protective efficacy is 92%.
The American
Journal
of Medicine
Volume
88
POSTDENTAL
subjects, or both, would reduce the protective efficacy and eliminate statistical significance of the unmatched analysis. On the other hand, the matched analysis is more resistant to misclassification of subjects. In addition, that we achieved meaningful differences with this small case series indicates the substantial magnitude of efficacy of antibiotic prophylaxis. A more important consequence of the small sample size may be the inability to perform subgroup analyses and to calculate the protective efficacy of antibiotic prophylaxis for different high-risk lesions and for different antibiotic regimens. A second possible limitation of this study may be our acceptance of a subject’s recall regarding use of antibiotic prophylaxis as the basis for ascertaining exposure. In designing this study, we had planned to obtain this information solely from dental records. It became clear, however, that some patients take prophylaxis without notifying their dentists, and, more commonly, that this information is not routinely recorded in the dental record. Information from dental records and from patient memory regarding prophylaxis was in agreement for all eight case subjects and for 19 of 24 control subjects. The remaining five control subjects recalled using antibiotic prophylaxis routinely before “every” dental procedure, whereas their records did not indicate that prophylaxis had been taken. There were no instances in which dental records indicated use of prophylaxis but subjects denied using it. A third limitation of this study may be that our results do not necessarily represent the protective efficacy of the AHA-recommended regimens. We accepted as “prophylaxis” the use of any AHA-recommended antibiotic both before and after the dental procedure regardless of dose or duration of use. We used this definition of prophylaxis because it is impossible to know with certainty whether all subjects taking antibiotic prophylaxis did so according to AHA recommendations. Although several subjects remembered exactly how they used prophylaxis, others were less certain of dose and duration of use, but could readily recall using prophylaxis both before and after dental procedures. Moreover, our findings do not specifically reflect the efficacy of the currently recommended regimens since the study period spans two eras of recommendations for antibiotic prophylaxis. This final limitation, which may reduce the generalizability of the study results, does not affect its validity. We believe that the strength of this analysis lies in the care with which cases and controls were matched for critical clinical features. We matched cases and controls for features related both to the risk of developing infective endocarditis and to the likelihood of receiving prophylaxis. Most important among these variables was the particular cardiac lesion, since clinicians may consider some lesions to be more “high risk” than others and thus selectively administer prophylaxis to patients with such lesions. Since we were concerned that patients who had prior documentation of a high-risk lesion by echocardiography might be more likely to receive prophylaxis than patients who had a high-risk lesion documented by auscultation alone, we matched cases and controls so that a similar mode of documentation of the high-risk lesions existed before the dental procedure in ques-
INFECTIVE
ENDOCARDITIS
/ IMPERIALE
AND HORWITZ
tion. In support of matching for this variable is the fact that several control subjects with high-risk lesions by auscultation never received antibiotic prophylaxis before their echocardiogram, but had received it routinely subsequent to echocardiography. These data suggest that if we had ignored this variable, a higher proportion of control subjects would have received prophylaxis, and protective efficacy would have appeared to be even greater. Our results support the use of antibiotic prophylaxis by persons with high-risk cardiac lesions in the dental setting, and better justify it as a preventive strategy. However, prophylaxis is not without side effects, which range from gastrointestinal upset and skin rash to death from anaphylaxis. In a review of the world literature, Idsoe et al [12] estimated the incidence of skin rashes and urticaria at 0.7% to lo%, and the incidence of anaphylaxis at 0.015% to 0.04%. More recent estimates range from 2.4 mild reactions per thousand courses of oral penicillin (range, 2.0 to 7.0) to 9.0 fatal reactions per million courses of oral penicillin (range, 6.0 to 11) [13]. Although these probabilities are small from the individual’s perspective, they represent a significant risk from a societal perspective because of the frequency with which measures to prevent endocarditis are used. In certain instances, the probability of experiencing a serious reaction to penicillin may exceed that of developing endocarditis without prophylaxis. Current estimates of the risk of developing endocarditis in persons with high-risk lesions without prophylaxis are imprecise, and range from 1 in 500 [14] to 1 in 115,000 [15]. Clearly, the risks and benefits of antibiotic prophylaxis need to be more precisely defined. It is necessary to explore the possibilities that prophylactic efficacy may depend on the particular antibiotic regimen or valvular lesion. It may be that certain regimens prevent endocarditis for specific cardiac lesions or that other regimens offer good overall protection. Such information is required to maximize the benefit and minimize the risk of this preventive strategy.
ACKNOWLEDGMENT We are indebted to Elizabeth T. Pesapane for her special care and expertise preparation of this manuscript.
in
REFERENCES 1. Littneo MM, Kaffe I, Tamse A, Buchner A: New concepts in chemoprophylaxis of bacterial endocarditis resulting from dental treatment. Oral Surg Oral Med Oral Pathol 1986; 61: 338-342. 2. Durack DT, Kaplan EL, Bisno AL: Apparent failures of endocardrtrs prophylaxis. JAMA 1983; 258: 2318-2322. 3. Kaplan EL: Bacterial endocarditis prophylaxis: tradition or necessrty? Am J Cardiol 1986; 57: 478479. 4. American Heart Association Commrttee Report: Prevention of bactenal endocarditis. Circulation 1984; 70: 1123A-1127A. 5. MacMahon SW, Hickey AJ, Wilcken DEL, et al: Rusk of Infective endocarditis in mitral valve prolapse with and without precordial systolic murmurs, Am J Cardiol 1986; 58. 105-108. 6. Horwitz RI: The experimental paradigm and observational studres of causeeffect relattonships in clrnical medicine. J Chronic Dis 1987; 40: 91-99. 7. Schlesselman JJ: Case-control studies. New York: Oxford University Press, 1982. 8. Flerss JL: Stabsbcal methods for rates and proportions. New York: John Wiley and Sons, 1981; 123-126. 9. Cornfield J: A statistrcal problem ansing from retrospective studres. In. Neyman J, ed. Proceedings of the Third Berkeley Symposium on Mathematical Stabs&s and Probabrlrties. Berkeley: University of Cakfornia Press, 1956; 135-148. 10. Clemens JD. Shapiro ED: Resolving the pneumococcal vaccrne controversy:
February
1990
The American
Journal
of Medicine
Volume
88
135
POSTDENTAL gainere
INFECTIVE
alternatives
ENDOCARDITIS
to randomized
/ IMPERIALE
AND HORWITZ
clinical trials? Rev Infect Dis 1984; 6: 589-
prophylaxis for pabents with mitral valve prolapse. J Chronic Dis 1984; 37: 531544. 14. Kelson SR, White PD: Notes on 250 cases of subacute bacterial (streptococcal) endocarditis studied and treated between 1927 and 1939. Ann Intern Med 1945; 22: 40-60. 15. Pogrel MA, Welsby PD: The dentist and prevention of Infective endocarditrs. Br Dent J 1975: 139: 12-16. 16.Von Reyn CF, Levy BS, Arbeit RD, et al: Infective endocarditis: an analysis based on strict case definitions. Ann Intern Med 1981; 94: 505-518.
11. Horstkotte D. Friedrichs W, Pippert H, Bircks W, Loogen F: Benefit of prophylaxis for infectious endocarditis in patients with prosthetic heart valves. 2 Kardiol 1986; 75: 8-l 1. 12. ldsoe 0, Guthe T, Wilcox L, DeWeck AL: Nature and extent of penicillin srdereactions, with particular reference to fatalities from anaphylactic shock. Bull WHO 1968; 38: 159-188. 13. Clemens JD, Ransohoff DF: A quantitative assessment of pre-dental antibiotic
136
February
1990
The American
Journal
of Medicine
Volume
88
