Relation of Antibiotic Use to Risk of Myocardial Infarction in the General Population José A. Luchsinger, MD, MPH, Ariel Pablos-Méndez, MD, MPH, Charles Knirsch, Daniel Rabinowitz, PhD, and Steven Shea, MD MD, MPH, There are conflicting reports of an association between Chlamydia pneumoniae (C. pneumoniae) infection and coronary artery disease (CAD); randomized trials of antibiotics for the secondary prevention of CAD are currently underway. Physicians may be tempted to believe that their choice of antibiotic class in treating any infection may alter the risk of CAD. Our objective was to determine if the use of antibiotics with antichlamydial activity in the general population reduces the risk of myocardial infarction. A healthcare claims database with 354,258 patients with continuous health and pharmacy coverage for at least 2 years between January 1, 1991 and December 31, 1997 was used for the analyses. Hazard ratios were derived from proportional hazards models with time-dependent covariates, relating antibiotic prescription to first claim related to incident first myocardial infarction during the observation pe- riod, adjusting for previous CAD, age, sex, diabetes, hypertension, hyperlipidemia, and chronic obstructive pulmonary disease. There were a total of 1,684,091 person-years of observation and 16,139 incident myocardial infarctions. The adjusted hazard ratios were 1.10 (95% confidence intervals [CI] 1.04 to 1.16) for macrolides, 1.20 (95% CI 1.13 to 1.26) for quinolones, 1.10 (95% CI 0.96 to 1.21) for cephalosporins, 1.00 (95% CI 0.96 to 1.06) for tetracyclines, 1.01 (95% CI 0.96 to 1.06) for penicillins, and 1.13 (95% CI 0.98 to 1.30) for trimetroprim-sulfamethoxazole. The hazard ratios for individual antibiotics with activity against C. pneumoniae within each group were similar. Use of antibiotics with activity against C. pneumoniae does not reduce the risk of myocardial infarction in the general population. 䊚2002 by Excerpta Medica, Inc. (Am J Cardiol 2002;89:18 –21) nfection may be causally related to coronary artery Icontribute disease (CAD), and increased antibiotic use may to the ongoing decline in cardiovascular is associated with a decreased risk of myocardial infarction in the general population. mortality.1 There are conflicting data from observational studies linking Chlamydia pneumoniae (C. pneumoniae) infection and CAD.2–9 Three main classes of antibiotics have activity against C. pneumoniae: macrolides, tetracyclines, and quinolones.10,11 Several studies have suggested a protective effect of macrolides in the secondary prevention of CAD,12–14 and clinicians are already prescribing antibiotics with activity against C. pneumoniae for the treatment of CAD.15 It is also possible that macrolides could affect CAD by treating other organisms, or through an anti-inflammatory effect.15 Two case-control studies that examined the relation between incidental use of antibiotics in the general population and risk of myocardial infarction had conflicting results.16,17 The objective of our study was to determine if the use of antibiotics active against C. pneumoniae From the Divisions of General Medicine and Infectious Diseases, Department of Medicine, Columbia University College of Physicians and Surgeons; Division of Epidemiology, Joseph P. Mailman School of Public Health, Columbia University; Department of Statitistics, Columbia University; and Pfizer Pharmaceuticals, New York, New York. This study was supported in part by an unrestricted grant from the Pfizer Corporation, New York, New York. Manuscript received May 24, 2001; revised manuscript received and accepted August 30, 2001. Address for reprints: José Luchsinger, MD, MPH, Division of General Medicine, PH9E-105, 622 West 168th St., New York, New York 10032. E-mail: jal94@columbia.edu. 18 ©2002 by Excerpta Medica, Inc. All rights reserved. The American Journal of Cardiology Vol. 89 January 1, 2002 METHODS A claims database comprising 377,398 subjects of both sexes, with full insurance coverage for pharmacy and health services, was obtained from Protocare Sciences (Herndon, Virginia). Subjects were included in the database if they were ⬎45 years old, had ⱖ2 continuous years of coverage between January 1, 1991 and September 30, 1997, and made ⱖ1 claim of any type in each year of coverage. There were a total of 48,542,468 claims of any type (for any service) for the 354,258 subjects included in our study, but only 4,925,347 claims (10%) were pertinent to the variables in our analyses. The database contained information on demographics, diagnoses, and pharmacy use, all linked by a unique patient identifier. The initiation of observation time for each individual in the database was the date of the beginning of insurance coverage, which occurred between January 1, 1991 and September 30, 1997. For each included individual, all consecutive calendar years of observed persontime were included in the analysis if ⱖ1 claims for any health care service were made in the year of observation time. Coverage was considered continuous if the end date and beginning date of consecutive enrollment periods were separated by no more than 1 day. If a subject had a gap of ⬎1 day between coverage periods, the observation time was terminated at the last date before the gap. This criterion excluded 0002-9149/02/$–see front matter PII S0002-9149(01)02156-7 TABLE 1 Characteristics of Study Subjects Classified According to No Antibiotic Exposure or Exposure to Each of Six Antibiotic Groups Age ⬎65 yrs Male sex Diabetes mellitus Hyperlipidemia Systemic hypertension Previous CAD COPD No Antibiotic Claims (n ⫽ 152,475) (%) Macrolide Claims (n ⫽ 70,801) (%) Quinolone Claims (n ⫽ 61,031) (%) 81,422 66,479 21,194 45,590 81,879 27,141 5,947 38,870 28,887 14,019 27,683 44,746 19,258 8,213 39,792 29,478 15,197 23,924 42,722 20,140 7,385 (53.4) (43.6) (13.9) (29.9) (53.7) (17.8) (3.9) (54.9) (40.8) (19.8) (39.1) (63.2) (27.2) (11.6) Tetracycline Claims (n ⫽ 20,150) (%) (65.2) 11,002 (54.6) (48.3) 8,544 (42.4) (24.9) 4,030 (20.0) (39.2) 7,939 (39.4) (70.0) 13,057 (64.8) (33.0) 5,904 (29.3) (12.1) 2,942 (14.6) Penicillin Cephalosporin Claims Claims (n ⫽ 107,662) (n ⫽ 93,301) (%) (%) TMP-SMZ Claims (n ⫽ 8,257) (%) 61,367 46,295 20,779 41,235 66,212 27,346 9,259 4,162 3,080 1,602 2,906 4,855 1,990 892 (57.0) (43.0) (19.3) (38.3) (61.5) (25.4) (8.6) 55,141 41,332 20,246 36,481 60,366 26,964 10,543 (59.1) (44.3) (21.7) (39.1) (64.7) (28.9) (11.3) (50.4) (37.3) (19.4) (35.2) (58.8) (24.1) (10.8) Percentages in the table refer to proportion of subjects in each antibiotic group with a particular characteristic. The characteristic of the antibiotic groups were compared with those of individuals without antibiotic claims, and all differences were statistically significant with p ⬍0.001. TMP-SMZ ⫽ trimethoprim-sulfamethoxazole. 23,140 subjects with ⬍2 years of continuous observation time (6.1% of the total sample). Those excluded were younger (mean age 65 years); there was a lower proportion of male subjects (42.4%), and lower prevalences of diabetes (6.2%), hyperlipidemia (14.7%), hypertension (24%), previous CAD (5.9%), and chronic obstructive pulmonary disease (COPD) (1.5%) than in the final sample. The incidence of myocardial infarction in the excluded group was 0.61% (140 myocardial infarctions). This study was approved by the Institutional Review Board of Columbia-Presbyterian Medical Center. Ascertainment of exposure to antibiotics: Exposure to antibiotics was based on the first pharmacy claim for antibiotic prescription during the time of observation. These claims were categorized into 7 exposure groups according to common antibiotic classes used by clinicians: (1) macrolides, (2) quinolones, (3) tetracyclines, (4) penicillins, (5) cephalosporins, (6) trimetroprim-sulfamethoxazole, and (7) nonusers of any of these 6 classes of antibiotics. Of the 6 antibiotic classes, macrolides, quinolones, and tetracyclines have significant activity against C. pneumoniae.10,11 Each pharmacy claim included a national drug code and date of fulfillment. Ascertainment of outcome: The primary outcome was first acute myocardial infarction during the observation period, defined by a claim using the International Classification of Diseases-9th edition (ICD-9) code for acute myocardial infarction (ICD-9 410). Previous myocardial infarctions (ICD-9 code 412) were not included in the outcome variable. Covariates: Relevant covariates included in the database were age, sex, and diagnosis of diabetes mellitus, hypertension, hyperlipidemia, history of CAD, and COPD. Age was defined as age at the date of the beginning of observation. Sex was retrieved from demographic data. The other covariates were defined based on pharmacy and/or service claims during the observation period. Diabetes was defined by ICD-9 code 250 and/or claims for insulin or oral hypoglycemics. Hypertension was defined by ICD-9 codes 401, 402, 403, 404, and 405. Hyperlipidemia was defined by ICD-9 code 272 and/or use of medications for hyperlipidemia (statins and fibrates). CAD was de- fined by ICD-9 codes 412, 413, and 414. COPD was defined by ICD-9 codes 491 and 492. Statistical analysis: Bivariate and multivariate analyses were performed using Cox proportional hazards regression with time-dependent covariates18,19 for first antibiotic exposure and time-constant covariates for the other variables. Observation time before antibiotic use was considered unexposed and time after antibiotic use was considered exposed. Observation time subsequent to a first claim for myocardial infarction was censored. The analytic procedures compared the hazard rates for myocardial infarction for each antibiotic group over observation times subsequent to the time of antibiotic exposure with that of observation times before antibiotic exposure and with observation times of subjects not exposed to the antibiotic group. The final model included all antibiotic groups and was stratified by age (22 strata) and calendar year at the beginning of observation (7 strata) to account for age and period effects. All analyses were performed using SAS version 6.12 for Windows (SAS Institute Inc., Cary, North Carolina). RESULTS The 354,258 study subjects contributed a total of 1,684,091 person-years to the analysis. There were 16,139 claims for first myocardial infarctions, which represented an overall cumulative incidence during the time of observation of 4.6% and an average incidence rate of 0.9 events per 100 person-years. The mean times from the beginning of observation to first myocardial infarction and first antibiotic use were 985 ⫾ 617 and 1,122 ⫾ 652 days, respectively. For subjects who had a claim for an antibiotic and a subsequent claim for myocardial infarction, the mean time of this interval was 516 ⫾ 445 days (median 393, interquartile range 157 to 766). The mean prescribed antibiotic supply was 8 days. Subjects without any antibiotic claim had a lower prevalence of cardiovascular risk factors, including diabetes, hypertension, and hyperlipidemia, history of CAD, and COPD (Table 1). The fully adjusted hazard ratios for myocardial infarction were slightly increased for those exposed to macrolides (1.10; 95% confidence intervals [CI] 1.04 to 1.16) and quinolones (1.20; 95% CI 1.13 to 1.26) CORONARY ARTERY DISEASE/ANTIBIOTICS AND RISK OF MYOCARDIAL INFARCTION 19 TABLE 2 Rates and Adjusted Hazard Ratios of Myocardial Infarction for Each Antibiotic Group Rates of Myocardial Infarction per 100 person-yrs Macrolides Quinolones Tetracycline Penicillins Cephalosporins TMP-SMZ 1.05 1.19 1.07 1.14 0.98 1.02 Hazard Ratio (95% CI) Adjusted for Age, Sex and Calendar Year p Value 1.20 1.34 1.17 1.05 1.11 1.15 ⬍0.001 ⬍0.001 0.001 0.003 ⬍0.001 0.047 (1.13–1.26) (1.27–1.41) (1.06–1.29) (1.00–1.11) (1.06–1.17) (1.00–1.33) Hazard Ratio (95% CI) (Full Model) 1.10 1.20 1.10 1.00 1.01 1.13 (1.04–1.16) (1.31–1.26) (0.96–1.21) (0.96–1.06) (0.96–1.06) (0.98–1.30) p Value ⬍0.001 ⬍0.001 0.059 0.712 0.575 0.090 The hazard ratios compare individuals with claims for each antibiotic vs. individuals without claims for other antibiotics. Abbreviation as in Table 1. TABLE 3 Hazard Ratios and 95% Confidence Intervals (CI) of Myocardial Infarction for Each Antibiotic Group from a Cox Regression Analyses Stratified by Gender, Age, Year of Beginning of Observation, and Presence of Diabetes, Hypertension, Previous Coronary Artery Disease, Hyperlipidemia, and Chronic Obstructive Pulmonary Disease Rates of Myocardial Infarction per 100 person-yrs Macrolides Quinolones Tetracycline Penicillins Cephalosporins TMP-SMZ 1.05 1.19 1.07 1.14 0.98 1.02 Hazard Ratio (95% CI) 1.08 1.19 1.06 1.01 1.01 1.14 (1.02–1.15) (1.13–1.27) (0.96–1.18) (0.96–1.06) (0.96–1.06) (0.99–1.33) p Value 0.006 ⬍0.001 0.247 0.718 0.699 0.072 cin (95% CI 1.03 to 1.16), 1.06 for chlarithromycin (95% CI 0.99 to 1.13), and 1.20 for azithromycin (95% CI 0.17 to 8.58 . Ciprofloxacin accounted for 88% of the claims for quinolones, and 12% were for ofloxacin. The hazard ratios for each of these antibiotics also differed little from those of the quinolones as a whole: 1.19 for ciprofloxacin (95% CI 1.12 to 1.26) and 1.28 for ofloxacin (95% CI 1.10 to 1.48). All claims for tetracyclines were for doxycycline (hazard ratio 1.10; 95% CI 0.96 to 1.21). DISCUSSION compared with those not exposed to those antibiotics. The hazard ratios of myocardial infarction for tetracyclines, cephalosporins, penicillins, and trimetroprim-sulfamethoxazole were not significantly different from 1.0 (Table 2). The hazard ratios for male sex (1.75; 95% CI 1.70 to 1.85), diabetes (1.68; 95% CI 1.62 to 1.73), hypertension (2.18; 95% CI 2.09 to 2.28), hyperlipidemia (1.43; 95% CI 1.38 to 1.47), previous CAD (1.37; 95% CI 1.32 to 1.41), and COPD (1.85; 95% CI 1.77 to 1.93) were in the expected directions and statistically significant. We also performed Cox regression analyses stratified by cardiovascular risk factors to address possible residual confounding; this is akin to comparing individuals matched by all covariates. The results were similar to those in the main analyses (Table 3). We performed additional analyses in 88,437 subjects without cardiovascular risk factors, and the results were also similar to the main analysis: the hazard ratios of myocardial infarction were 1.18 for macrolides (95% CI 0.87 to 1.59), 1.43 for quinolones (95% CI 1.06 to 1.94), 0.73 for tetracyclines (95% CI 0.36 to 1.48), 1.01 for penicillins (0.78 to 1.30), 1.24 for cephalosporins (95% CI 0.97 to 1.58), and 0.87 for trimetroprim-sulfamethoxazole (95% CI 0.39 to 1.96). Erythromycin accounted for 71% of macrolide claims, chlarithromycin had 28% of claims, and azithromycin had 1% of claims. The hazard ratios for exposure to these antibiotics were 1.09 for erythromy- These analyses of the longitudinal experience of 354,258 subjects contributing ⬎1,600,000 personyears of observation with ⬎16,000 incident myocardial infarctions indicate that exposure to short courses of single antibiotics with activity against C. pneumoniae is not associated with a decreased risk of myocardial infarction. Atherosclerosis has been induced in rabbits infected with C. pneumoniae, and rabbits given azithromycin have been shown to have less severe atheromata than nontreated infected rabbits.20 Specific mechanisms by which C. pneumoniae might promote atherogenesis include replication in human endothelial and vascular smooth muscle cells,21–23 the antigenic mimicry of heart muscle,24 induction of thrombosis,21 and low-density lipoprotein oxidation and macrophage induction.25,26 Observational studies of the association between C. pneumoniae seropositivity and CAD have had conflicting results.2–9 Macrolides, quinolones, and tetracyclines are considered the main antichlamydial antibiotics,10,11 and macrolides have the best minimal inhibitory concentration against C. pneumoniae.10,11 Two small randomized trials found a protective effect of macrolide antibiotics in the secondary prevention of coronary events,12,13 and the Azithromycin in Coronary Artery Disease: Elimination of Myocardial Infarction with Chlamydia (ACADEMIC) study found a significant decrease in an inflammatory marker score at 6 months,14 but no effect on cardiovascular events at 2 years.27 The ACADEMIC study was only powered to detect a 50% reduction in events.27 Thus, the question of whether macrolides could have a more modest effect in the secondary prevention of CAD remains unsettled.15 To our knowledge, there are no 20 THE AMERICAN JOURNAL OF CARDIOLOGY姞 JANUARY 1, 2002 Abbreviation as in Table 1. VOL. 89 ongoing trials of tetracyclines or quinolones for the primary or secondary prevention of myocardial infarction. Meier et al16 compared 3,315 cases of myocardial infarction with 13,315 controls and found a protective effect in tetracycline and quinolone antibiotic use, which are considered antichlamydial agents, but a protective effect was not found for macrolides. Jackson et al17 compared 1,796 cases of myocardial infarction with 4,882 controls and found no association with the use of erythromycin, tetracycline, or doxycycline, and the risk for myocardial infarction. Our findings are consistent with the report by Jackson et al,17 and those of the ACADEMIC trial, and are inconsistent with the findings of a protective effect for quinolones and tetracyclines found by Meier et al.16 Several explanations for our findings may be considered. It is possible that a protective effect of antibiotics may occur with longer periods of use than the courses represented in our database. It is also possible that some subjects in our study filled their prescriptions but did not actually take the antibiotics. It is possible that our population was not infected with C. pneumoniae. This seems unlikely based on the epidemiology of C. pneumoniae.28,29 We found small detrimental associations between use of macrolides and quinolones and myocardial infarction. One possible explanation may be residual confounding. We repeated our analyses in a subgroup after exclusion of subjects with any of these risk factors for myocardial infarction, matching individuals by all risk factors, and the findings were essentially the same. The most important covariate not measured in our data was tobacco use, and we cannot rule out uncontrolled confounding by smoking. We used claims related to COPD as a proxy for smoking. Another limitation arises from the use of claims data to determine incidence of myocardial infarction. The hazard ratios for the cardiovascular risk factors included as covariates— diabetes, hypertension, hyperlipidemia, history of CAD, and COPD—which showed an increased risk of myocardial infarction as expected, indirectly supporting the validity of the claims database. In one study the overall accuracy of ICD-9 code 410 in identifying definite myocardial infarction was 92%.30 The main strengths of our study are the sample size, the large number of outcome events, and the long periods of follow-up per subject. It is unlikely that any true protective effect was missed due to lack of statistical power or regression dilution bias. All the subjects in the database had full insurance coverage, and it can be assumed that the sample was homogenous in terms of socioeconomic class. In conclusion, our findings do not support the hypothesis that use of antibiotics with activity against C. pneumoniae in the general population decreases the risk of myocardial infarction. Antibiotics should not be chosen based on a possible protective effect on CAD. 1. Nieto FJ. Infections and atherosclerosis. New clues from an old hypothesis? Am J Epidemiol 1998;148:937–948. 2. Saikku P, Leinonen M, Tenkanen L, Linnanmäki E, Ekman M-R, Manninen V, Manttari M, Frick MH, Kuttunen JK. Chronic Chlamydia pneumoniae infection as a risk factor for coronary artery disease in the Helsinki heart study. Ann Intern Med 1992;116:273–278. 3. Thom DH, Wang SP, Grayston JT, Siscovick DS, Stewart DK, Kronmal RA, Weiss NS. Chlamydia pneumoniae strain TWAR antibody and angiographically demonstrated coronary artery disease. Arterioscler Thromb 1991;11:547–551. 4. Thom DH, Grayston JT, Siscovick DS, Wang SP, Weiss NS, Daling JR. Association of prior infection with Chlamydia pneumoniae and angiographically demonstrated coronary artery disease. JAMA 1992;268:68 –72. 5. Campbell LA, O’Brien ER, Capuccio AL, Kuo CC, Wang SP, Stewart D, Patton DL, Cummings PK, Grayston JT. Detection of Chlamydia pneumoniae TWAR in human coronary atherectomy tissues. J Inf Dis 1995;172:585–588. 6. Saikku P, Mattila K, Nieminen MS, Huttunen JK, Leinomen M, Ekman M-R, Makela PH, Valtonen V. Serological evidence of an association of a novel Chlamydia, TWAR, with chronic coronary heart disease and acute myocardial infarction. Lancet 1988;2:983–986. 7. Ridker PM, Kundsin RB, Stampfer MJ, Poulin S, Hennekens CH. Prospective study of Chlamydia pneumoniae IgG seropositivity and risk of future myocardial infarction. Circulation 1999;99:1161–1164. 8. Nieto FJ, Folsom AR, Sorlie PD, Grayston JT, Wang S, Chambless LE. Chlamydia pneumoniae infection and incident coronary heart disease. Am J Epidemiol 1999;150:149 –156. 9. Ridker PM, Hennekens CH, Buring JE, Kundsin R, Shih J. Baseline IgG antibody titers to Chlamydia pneumoniae, Helicobacter pylori, herpes simplex virus, and cytomegalovirus and the risk of cardiovascular disease in women. Ann Intern Med 1999;131:573–577. 10. Hammerschlag MR, Qumei KK, Roblin PM. In vitro activities of azithromycin, clarithromycin, L-ofloxacin, and other antibiotics against Chlamydia pneumoniae. Antimicrob Agents Chemother 1993;36:1573–1574. 11. Hammerschlag MR. Antimicrobial susceptibility and therapy of infections caused by Chlamydia pneumoniae. Antimicrob Agents Chemother 1994;38:1873– 1878. 12. Gupta S, Leatham EW, Carrington D, Mendall MA, Kaski JC, Camm AJ. Elevated Chlamydia pneumoniae antibodies, cardiovascular events, and azithromycin in male survivors of myocardial infarction. Circulation 1997;95:404 –407. 13. Gurfinkel E, Bozovich G, Daroca A, Beck E, Mautner B. Randomized trial of roxithromycin in non-Q wave coronary syndromes: Roxis pilot study. Lancet 1997;350:404 –407. 14. Anderson JL, Muhlestein JB, Carlquist J, Allen A, Trehan S, Nielson C, Hall S, Brady J, Egger M, Horne B, Lim T. Randomized secondary prevention trial of azithromycin in patients with coronary artery disease and serological evidence for Chlamydia pneumoniae infection: The Azithromycin in coronary artery disease: Elimination of Myocardial infection (ACADEMIC) Study. Circulation 1999;99: 1540 –1547. 15. Grayston JT. Secondary prevention antibiotic treatment trials for coronary artery disease. Circulation 2000;102:1742–1743. 16. Meier CR, Derby LE, Jick SS, Vasilakis C, Jick J. Antibiotics and risk of subsequent first-time acute myocardial infarction. JAMA 1999;281:427–431. 17. Jackson LA, Smith NL, Heckbert SR, Grayston JT, Siscovick DS, Psaty BM. Lack of association between first myocardial infarction and past use of erythromycin, tetracycline, or doxycycline. Emerg Infect Dis 1999;5:281–284. 18. Rothman KJ, Greenland S. Modern Epidemiology, 2nd Ed. Philadelphia, PA: Lippincott-Raven, 1998:376 –377. 19. Kahn HA, Sempos CT. Statistical Methods in Epidemiology. New York, NY: Oxford University Press, 1989:193–198. 20. Muhlestein JB, Anderson JL, Hammond EH, Zhao L, Trehan S, Schwobe EP, Carlquist JF. Infection with Chlamydia pneumoniae accelerates the development of atherosclerosis and treatment with azithromycin prevents it in a rabbit model. Circulation 1998;97:633–636. 21. Fryer RH, Schwobe EP, Woods ML, Rodgers GM. Chlamydia species infect human vascular endothelial cells and induce procoagulant activity. J Invest Med 1997;45:168 –174. 22. Halme S, Syrjälä H, Bloigu A, Saikku P, Leinonen M, Airaksinen J, Surcel HM. Lymphocyte responses to Chlamydia antigens in patients with coronary heart disease. Eur Heart J 1997;18:1095–1101. 23. Gaydos CA, Summersgill JT, Sahney NN, Ramirez JA, Quinn TC. Replication of Chlamydia pneumoniae in vitro in human macrophages, endothelial cells, and aortic artery smooth muscle cells. Infect Immun 1996;64:1614 –1620. 24. Bachmaier K Neu, N,, de la Maza L, Sukumar P, Hessel A, Penninger JM. Chlamydia infections and heart disease linked through antigenic mimicry. Science 1999;283:1335–1339. 25. Kol A, Sukhova GK, Lichtman AH, Libby P. Chlamydial heat shock protein 60 localizes in human atheroma and regulates macrophage tumor necrosis factor-␣ and matrix metalloproteinase expression. Circulation 1998:300 –307. 26. Kalayoglu MV, Hoerneman B, LaVerda D. Morrison SG, Morrison RP, Byrne GI. Cellular oxidation of low-density lipoprotein by Chlamydia pneumoniae. J Infect Dis 1999;180:780 –90. 27. Muhlestein JB, Anderson JL, Carlquist JF. Randomized secondary prevention trial of azithromycin in patients with coronary artery disease: primary clinical results of the academic study. Circulation 2000;102:1755–1760. 28. Kauppinen M, Saikku P. Pneumonia due to chlamydia pneumoniae: prevalence, clinical features, diagnosis, and treatment. Clin Infect Dis 1995;21:244 –252. 29. Saikku P. The epidemiology and significance of Chlamydia pneumoniae. J Infect 1992;25(suppl 1):27–34. 30. Pladevall M, Goff DC, Nichaman MZ, Chan F, Ramsey D, Ortiz C, Labarthe DR. An assessment of the validity of ICD-9 code 410 to identify hospital admissions for myocardial infarction: The Corpus Christi Heart Project. Int J Epidemiol 1996;25:948 –952. CORONARY ARTERY DISEASE/ANTIBIOTICS AND RISK OF MYOCARDIAL INFARCTION 21