Efficacy of antibiotic prophylaxis for prevention of native-valve endocarditis
Whether antibiotic prophylaxis can prevent bacterial endocarditis is hotly debated. In an attempt to settle this issue, we have assessed the efficacy of prophylaxis for bacterial endocarditis on native valves in a nationwide, case-control study in the Netherlands. Cases were patients with known cardiac disease in whom endocarditis developed within 180 days of a medical or dental procedure for which prophylaxis was indicated. Of a total of 438 patients with endocarditis diagnosed during 2 years, 48 were eligible for the study. Controls were patients with the same cardiac status in whom endocarditis did not develop within 180 days of a similar procedure; of a total of 889 controls from five hospitals, 200 were eligible. Overall, about 1 in 6 patients in both groups had received prophylaxis. The best estimate of protective efficacy was 49% for first-ever endocarditis occurring within 30 days of a procedure. Endocarditis developed within 30 days of a procedure in only 13% of patients with a previously diagnosed heart lesion which predisposed to the disease. The findings suggest that strict adherence to for recommendations accepted generally do little to decrease the total prophylaxis might number of patients with endocarditis in the
assess the efficacy of antibiotic prophylaxis in man has not yet been done, mainly because the disease is rare and so a large series of patient would be needed.12 We have therefore used an alternative approach-ie, a case-control study-to assess the protective effect of antibiotic prophylaxis in subjects with native-valve and cardiovascular anomalies.
Subjects were eligible if they had native-valve and cardiovascular anomalies that increased their risk of getting endocarditis and that warranted antibiotic prophylaxis before certain medical or dental procedures.13-14 These lesions included congenital heart disease, coarctation of the aorta, rheumatic and other valvular dysfunction, and mitral valve prolapse with mitral regurgitation. Patients with prosthetic heart valves were excluded for two reasons. Firstly, these patients probably have a much higher risk of endocarditis and therefore are a different risk-group. Secondly, there were too few patients with a prosthetic heart valve for a case-control study. Cases had to have undergone, within 180 days of onset of symptoms, a medical or dental procedure that required prophylaxis, and controls had to have had such a procedure within 180 days of the day of the follow-up interview. We chose 180 days to allow subsequent stratification by 30 day intervals in the analysis. Indications for prophylaxis were derived from the guidelines of the Netherlands Heart Foundation, and are similar to recommendations in the UK and USA.13-17 Procedures, indications for prophylaxis, and antibiotics Subjects Information
Introduction There is considerable controversy about the efficacy of antibiotic prophylaxis for prevention of bacterial endocarditis.l-6In the 1940s, sulphonamides and penicillin were successfully used to reduce the incidence and severity of bacteraemia in patients who had undergone dental or surgical procedures, but numerous reports of prophylaxis failure have since appeared.7 11 A controlled clinical trial to
interviewed with a structured questionnaire. obtained on procedures for which antibiotic is indicated that had been done within 180 days of the were was
Departments of Clinical Microbiology, (J T. M. der Meer, MD, Prof M F. Michel, MD), and of Epidemiology and Biostatistics (Prof H A. Valkenburg, MD), Erasmus University, Rotterdam; and Departments of Infectious Diseases (J. T. M. van der Meer, MD, W van Wijk, MSc, J. Thompson, MD) and Clinical Epidemiology (Prof J. P Vandenbroucke, MD), Leiden University Hospital, Leiden, Netherlands. Correspondence to Dr Jan T. M van der Meer, Department of Infectious Diseases, Building 1, C5-P, University Hospital, PO 9600, 2300 RC Leiden, the Netherlands. ADDRESSES: van
onset of disease (cases) or interview (controls). Procedures were named one at a time and the patient was requested to answer "yes" or "no". Questions were reworded only when a subject did not understand the question. When a subject had had more than one procedure, that closest to the onset of symptoms or interview was entered. Proxy responders-ie, spouses and general practitionerswere used for cases who were too ill to be interviewed or had died before they could be visited. All information about recent procedures and the use of prophylaxis was checked with medical and dental specialists, including pharmacists.
TABLEI-PROCEDURES AND MICROORGANISMS IN CASES OF ENDOCARDITIS
Cases From Nov 1, 1986, to Nov 1, 1988, all patients who were consecutively admitted to hospital in the Netherlands (14 491 138 inhabitants) and who were suspected of having bacterial endocarditis were reported to one of us (J. v. d. M.) by the microbiologists who were aware of these patients through submitted blood cultures. While the patient was in hospital, we asked the treating physician to explain the purpose of the study, and sought permission to visit the hospital for a review of the medical records and an interview with the patient. Data about previous diagnoses of heart disease, physical examination, and laboratory results were obtained. The patient was interviewed about recent medical or dental procedures and the use of prophylaxis. We defined day of onset as the first day on which a temperature of 38°C or more was recorded, accompanied by symptoms that finally led to admission to hospital, or the date of the first visit to a doctor for the same complaints. During 2 years, 559 patients suspected of having endocarditis were reported. 210 patients were excluded, 121 because they did not meet the diagnostic criteria for endocarditis as described by Von Reyn et al,18 and 89 because they had endocarditis on a prosthetic valve. Of the 349 patients who had native-valve endocarditis, 197 (56%) had previously had heart disease. Proxy responders were interviewed for 10 of these 197 patients. 54 patients (27%) had undergone, in the 180 days before onset of symptoms, a medical or dental procedure with an indication for prophylaxis. For 6 of these 54, a causal relation between the procedure and endocarditis was ruled out because it was unlikely that the agent isolated from the blood originated from the area of the procedure (Streptococcus bovis
isolated after a dental procedure [2 patients]; a-haemolytic streptococci after cystoscopy without tracheal intubation , after removal of a sebaceous cyst , and after stitching of a head wound, ; and Candida parapsilosis after dental scaling ). The remaining 48 patients with endocarditis formed the case group: for 39 of these patients this episode of endocarditis was their first, and for 9 it was a recurrent episode. None of the proxy responders fell in the case group.
Controls selected from outpatients of the cardiology departments of the Leiden University Hospital and four regional hospitals in Arnhem. All patients with a cardiac lesion and increased risk of endocarditis were selected. These patients were subsequently given a number and a first sample was drawn with the help of random tables. An introductory letter was sent to the control subjects accompanied by a list of of procedures for which antibiotic prophylaxis is indicated. The purpose of this list was to refresh their memory and imitate the recall stimulus experienced by the patients with endocarditis. 192 to 5 days later, the controls were contacted by telephone and the structured questionnaire was completed. Telephone calls were made on weekdays and Saturdays between 0900 h and 2200 h to avoid selection according to job or disease Controls
*Nos in parentheses no of patients from whIch organism isolated tNo further identification possible tNo organism isolated diagnosis based on histology from surgery
the eligibility criteria (53 patients had died, 29 had a heart valve, 62 had moved house and could not be located, 102 could not be contacted by telephone, 418 had not undergone a procedure with an indication for prophylaxis in the previous 180 days, and 25 refused to take part). Overall, of 200 controls included in the analysis, none got endocarditis within 180 days of the procedure.
Analysis The incidence of bacterial endocarditis is low, and endocarditis as the result of a procedure develops in an even smaller number of patients. Therefore, sample size and calculations of power were based on the assumption that a clinically important reduction in risk due to prophylaxis would have to be at least 75%. We also assumed that 40% of the population at risk for endocaditis would be given prophylaxis. On the basis of these assumptions and a one-sided significance level of 5 %, we calculated that a group of 31 cases and 4 controls per case would be needed. The power of the study to detect a protective effect of prophylaxis of at least 75% would then be
90%. To obtain
an estimate of the protective effect of antibiotic odds ratios (ORs) were calculated by the MantelHaenszel method from data stratified for age and indication for prophylaxis. Data were analysed separately for cases with first-time episodes of endocarditis and those with recurrent episodes. For a rare disease the OR derived from a case-control study closely approximates the relative risk (RR) that would have been derived from a cohort study. The protective effect of a prophylactic
TABLE li-CHARACTERISTICS OF CASES AND CONTROLS
Since endocarditis is age related and sex related, whereas cases and controls age related only /-23 were matched for age but not sex. Controls were eligible if they were in the same 5-year-age category as a case and had undergone a medical or dental procedure with an indication for prophylaxis within 180 days of the interview. If there were more than 4 controls per case, all controls were used. However, for some age strata there were fewer than 4. For these age strata, a second random sample was drawn. Letters were sent to 889 patients, 689 (77-5%) of whom did
prophylaxis compliance is
*Interval between procedure and onset of symptoms procedure and interview (controls)
(for cases) and
TABLE III-PROCEDURES AND INDICATION FOR AN EXPOSURE TO ANTIBIOTIC PROPHYLAXIS
Data are no of patients exposed per no of patients not exposed to antibiotics *Indication according to guidelines of Netherlands Heart Foundation tlndication for prophylaxis established with certainty, procedures include denta extractions, dental root work, digestive tract surgery. tlndication for prophylaxis not established with certainty, all patients in this group underwent dental scaling 91nterval between the procedure and onset of symptoms (cases) or mterview
is then expressed as 1- RR. Test-based confidence intervals (CI) for ORs were calculated with the chi-square method derived by Mantel-Haenszel. Ratios and medians were compared by means of the chi-square test and the non-parametric MannWhitney U test, respectively.
Results Cases The procedures and microorganisms associated with the of endocarditis are shown in table 1. Most of these patients had undergone a dental procedure (table n), and prophylaxis was definitely indicated in 28 of these 48 procedures. For the other 20 procedures, the indication for prophylaxis could not be determined with certainty since all concerned removal of tartar. According to the recommendation of the Netherlands Heart Foundation, prophylaxis is indicated for scaling when it includes the subgingival removal of tartar or when it is combined with polishing of the teeth. However, most dentists do not keep records of the extent of scaling, and it proved to be impossible to find out whether scaling included removal of subgingival calculus or polishing. The indication for prophylaxis for these 20 patients was therefore entered as "possible". Median interval between procedure and onset of symptoms was 72 5 days (range 3-170) for procedures with a possible indication for prophylaxis compared with 10 days (range 0-175) for other procedures (p < 0’001). Antibiotics were given before the procedure to 8 of the 48 cases (table in). Prophylaxis was given to 6 of these 8 cases in accordance with the recommendations of the Netherlands Heart Foundation. In the other 2 cases, the antibiotics did not accord with the recommendations but could be considered equivalent. Prophylaxis was given more often to patients who had had previous episodes of endocarditis than to those who had not (3/9 vs 5/39). cases
Controls 181 (91 %) of the 200 procedures in control subjects were dental. Prophylaxis was definitely indicated in 96 of them. In 104 procedures, the indication for prophylaxis was entered as "possible" because dental scaling was done and we could not determine whether subgingival calculus had been removed. The median interval between the procedure ind the day of the interview was 79 days for procedures with
Cumulative percentage of interval between procedure and onset of symptoms or interview.
Top, cases (.) and controls (0) differed significantly (p < 0-001) in median interval between procedure and onset of symptoms (cases) or interview (controls), 30 and 75 days, respectively Bottom, when analysed separately according to certainty of indication for prophylaxis, this difference disappeared for procedures with a "possible" indication and increased for those with a (cases = 0, controls = L) (p