Estimates of the incidence and impact of bacterial infective endocarditis (IE) have been limited by the infrequency of the disease. Administrative data analyses can provide important information across a broad range of hospitals and regions. We used a recent nationally representative sample1 to estimate the incidence of hospitalizations for bacterial IE in the United States.
We conducted a retrospective cohort study using the 1999 through 2008 Nationwide Inpatient Sample (NIS), which is produced by the Agency for Healthcare Research and Quality.1 The NIS is the largest all-payer inpatient database in the United States (approximately 8 million records per year).1 Admissions related to bacterial IE were identified by the presence of International Classification of Diseases, Ninth Revision (ICD-9) codes 421.0, 421.1, 421.9, or 996.61, combining 2 previous strategies.2,3 The etiologic agent of IE was determined by the presence of organism-specific infection (eg, 041.x) and bacteremia codes (038.x). Incidence was estimated using the rate of IE-related discharges per 100 000 US population-years. Rates were calculated quarterly based on discharge date; the denominator was adjusted annually based on the US population.
Trends in admission rate were evaluated using joinpoint methods, allowing the trend to change over time.4 We measured the effect of organism on in-hospital mortality using logistic regression, adjusting for age, sex, payer, and comorbidities. The data set was constructed in SAS software, version 9.22 (SAS Institute Inc); analyses were performed in Stata/IC, version 11.2 (StataCorp LP) and the Joinpoint Regression Program, version 3.4.3 (National Cancer Institute), using the stratification and weighting the data provided in the NIS to create nationally representative estimates. Additional detail on study methods is available in the eAppendix.
Of the 78.2 million records in the 1999 through 2008 NIS database, 93 511 met our inclusion criteria. Using weights, we determined that these records corresponded to 457 690 discharges nationwide. After exclusion of 9538 admissions that ended in inpatient transfer and 273 (0.3%) with unknown disposition, the main study sample consisted of 83 700 discharges (409 665 weighted). Most episodes involved patients who were male (59.3%), white (71.4%), and insured by Medicare (57.2%). Of those discharges for which an organism was identified, staphylococci were the most common (57.5%), followed by streptococci and/or enterococci (33.3%).
Between the first quarter of 1999 and the first quarter of 2006, the rate of bacterial IE–related hospitalizations increased from 11.4 per 100 000 population-years to 16.6 per 100 000 population-years (test of trend, P < .001). This trend corresponds to an average percentage change (APC) of 1.1% per quarter (95% CI, 0.9%-1.3%). After the first quarter of 2006, the rate stabilized, with an APC of 0.1% (95% CI, −0.6% to 0.8%). Substantial differences were evident in the rate of IE-related admissions caused by different organisms over the study period (Figure, A). Admissions associated with staphylococcal IE grew at a rate of 1.1% per quarter (95% CI, 0.9%-1.3%), rising from 3.3 to 5.4 cases per 100 000 population-years from first quarter of 1999 to the fourth quarter of 2008 (test of linear trend, P < .001). Most of the increase in staphylococcal IE admissions was due to IE caused by Staphylococcus aureus, which increased at a rate of 1.7% per quarter between the first quarter of 1999 and the first quarter of 2006 (95% CI, 1.3%-2.0%). Interestingly, rates of S aureus –associated IE stabilized between 2006 and 2008 (APC, 0.1%; 95% CI, −1.1% to 1.2%).
We limited the cohort to 33 956 admissions (165 563 weighted) that occurred in 2002 or later for which an etiologic organism was identified (excluding unknown organisms and unspeciated staphylococci) and that had complete covariate data. Admissions for S aureus –related IE were associated with a higher probability of in-hospital mortality than streptococcal and/or enterococcal IE (17.5% vs 8.9%) (P < .001). After adjustment, IE caused by S aureus was associated with a 57.1% greater risk of in-hospital mortality (risk difference, 5.9%) (P < .001) compared with streptococcal and/or enterococcal IE (Figure, B).
This report estimates the current incidence and trends in hospitalizations for bacterial IE in the United States at the beginning of the 21st century. We found that the rate of IE-related hospitalizations grew markedly and that this growth was driven primarily by increases in S aureus IE. Patients with S aureus IE were also more likely to die during the hospitalization. Since these findings were drawn from a representative sample of over 78 million hospitalizations, they are generalizable to contemporary medical practice.
Our finding that S aureus is the predominant cause of IE in the 21st century is consistent with results from the International Collaboration on Endocarditis.3 Our results contrast with those of Tleyjeh and colleagues,5 who found no significant change in IE rates or cause during the 1970-2000 period in Olmstead County, Minnesota. The generalizability of that important study was limited by its small size (107 IE cases during the 30-year period), small geographic scope, and lack of racial and ethnic diversity. By contrast, the current investigation involved a nationally representative sample of contemporary US admissions.
This investigation has important strengths. Our study used a large, contemporary, and nationally representative data set. Such a data set avoids referral bias,6 ensures that our results are broadly generalizable, and allows for sufficient sample size for statistical inference. The study is limited by its use of ICD-9 diagnosis codes, which have been associated with both false-positive and false-negative findings.7 However, others have reported good agreement between IE diagnosis codes and clinical criteria obtained from medical records.8 Our inability to access laboratory results resulted in a relatively high proportion of IE cases without organism identification (43.8%). However, the number of unidentified cases grew at a rate similar to that of identified cases (APC, 1.1% vs 1.2%), suggesting that our findings are not owing to improvements in coding.
Despite these limitations, our study is uniquely able to make a number of observations. We estimate the incidence of bacterial IE in the United States and document its increasing prevalence in the early 21st century. This growth is primarily attributable to increased rates of IE related to S aureus, which was associated with worse outcomes and higher costs than IE caused by other organisms.
Correspondence: Dr Fowler, Box 102359 Medical Center, 153 Hanes House, Trent Drive, Durham, NC 27710 (email@example.com).
Author Contributions: Mr Federspiel had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Federspiel, Stearns, Peppercorn, and Fowler. Acquisition of data: Federspiel, Stearns, and Fowler. Analysis and interpretation of data: Federspiel, Stearns, Peppercorn, Chu, and Fowler. Drafting of the manuscript: Federspiel, Peppercorn, Chu, and Fowler. Critical revision of the manuscript for important intellectual content: Federspiel, Stearns, Peppercorn, and Fowler. Statistical analysis: Federspiel and Stearns. Study supervision: Peppercorn.
Financial Disclosure: Mr Federspiel and Dr Stearns are employed by the University of North Carolina at Chapel Hill and have received grant support from the National Institutes of Heath (NIH). Dr Peppercorn is employed by GlaxoSmithKline. Dr Chu has received research support from Merck and Pfizer and is employed by Duke University. Dr Fowler has served as a consultant for Astellas, Cubist, Inhibitex, Merck, Johnson & Johnson, Shire, Leo Pharmaceuticals, NovaDigm, The Medicines Company, Baxter Pharmaceuticals, and Biosynexus; he has received grant or research support from Astellas, Cubist, Inhibitex, Merck, Theravance, Cerexa, Pfizer, Novartis, Advanced Liquid Logic, and NIH; he has received honoraria from Arpida, Astellas, Cubist, Inhibitex, Merck, Pfizer, Targanta, Theravance, Wyeth, Ortho-McNeil, Novartis, and Vertex Pharmaceuticals; he has served on an advisory committee and a speakers' bureau for Cubist; he is employed by Duke University.
Funding/Support: This study was supported by grant T32 GM008719 from the National Institute of General Medical Sciences (Mr Federspiel); grant R01 AG025801 from the National Institute on Aging (Dr Stearns); grants R01-AI068804 and K24 AI093969 from the National Institute of Allergy and Infectious Diseases (Dr Fowler); grant F30 HL110483 from the National Heart, Lung, and Blood Institute (Mr Federspiel); and grant 0675027N from the American Heart Association (Dr Chu).
Role of the Sponsors: None of the funding agencies had any role in the design and conduct of the study, the collection, management, analysis, and interpretation of the data, or the preparation, review or approval of the manuscript.
Previous Presentation: Preliminary data from this study was presented at the Interscience Conference on Antimicrobial Agents and Chemotherapy; September 17, 2011; Chicago, Illinois.
Additional Contributions: Tiffany Jewell, RHIA, assisted in developing the coding criteria. Peter H. Gilligan, PhD assisted in formulating the research objectives and design. Any errors are solely the responsibility of the authors.
Noskin GA, Rubin RJ, Schentag JJ,
et al. The burden of Staphylococcus aureus
infections on hospitals in the United States: an analysis of the 2000 and 2001 Nationwide Inpatient Sample database. Arch Intern Med
. 2005;165(15):1756-176116087824PubMedGoogle ScholarCrossref
Cabell CH, Heidenreich PA, Chu VH,
et al. Increasing rates of cardiac device infections among Medicare beneficiaries: 1990-1999. Am Heart J
. 2004;147(4):582-58615077071PubMedGoogle ScholarCrossref
Kim HJ, Fay MP, Feuer EJ, Midthune DN. Permutation tests for joinpoint regression with applications to cancer rates. Stat Med
. 2000;19(3):335-35110649300PubMedGoogle ScholarCrossref
Tleyjeh IM, Steckelberg JM, Murad HS,
et al. Temporal trends in infective endocarditis: a population-based study in Olmsted County, Minnesota. JAMA
. 2005;293(24):3022-302815972564PubMedGoogle ScholarCrossref
Steckelberg JM, Melton LJ III, Ilstrup DM, Rouse MS, Wilson WR. Influence of referral bias on the apparent clinical spectrum of infective endocarditis. Am J Med
. 1990;88(6):582-5882346159PubMedGoogle ScholarCrossref
Austin PC, Daly PA, Tu JV. A multicenter study of the coding accuracy of hospital discharge administrative data for patients admitted to cardiac care units in Ontario. Am Heart J
. 2002;144(2):290-29612177647PubMedGoogle ScholarCrossref
Schneeweiss S, Robicsek A, Scranton R, Zuckerman D, Solomon DH. Veteran's affairs hospital discharge databases coded serious bacterial infections accurately. J Clin Epidemiol
. 2007;60(4):397-40917346615PubMedGoogle ScholarCrossref