Author + information
- Received February 14, 2018
- Revision received March 12, 2018
- Accepted March 19, 2018
- Published online June 11, 2018.
- Imre Janszky, MD, PhDa,b,∗ (, )@NTNU@karolinskainst,
- Katalin Gémes, PhDc,
- Staffan Ahnve, MD, PhDc,
- Hilmir Asgeirsson, MD, PhDd,e and
- Jette Möller, PhDc
- aDepartment of Public Health and Nursing, Norwegian University of Science and Technology, Trondheim, Norway
- bRegional Center for Health Care Improvement, St. Olavs Hospital, Trondheim, Norway
- cDepartment of Public Health Sciences, Karolinska Institutet, Stockholm, Sweden
- dDepartment of Infectious Diseases, Karolinska University Hospital, Stockholm, Sweden
- eUnit of Infectious Diseases, Department of Medicine Huddinge, Karolinska Institutet, Stockholm, Sweden
- ↵∗Address for correspondence:
Dr. Imre Janszky, Department of Public Health and Nursing, Faculty of Medicine, Norwegian University of Science and Technology, Håkon Jarls gate 11, NO-7489 Trondheim, Norway.
Background Various invasive medical procedures might induce bacteremia and, hence, act as triggers for infective endocarditis. However, empirical data in humans on the potential dangers of invasive medical procedures in this regard are very sparse. Due to lack of sufficient data, it is currently debated whether the risk for endocarditis with medical procedures is substantial or rather negligible.
Objectives The purpose of this nationwide case-crossover study was to quantify the excess risk for infective endocarditis in association with invasive medical and surgical procedures.
Methods The authors identified all adult patients treated for endocarditis in hospitals in Sweden between January 1, 1998, and December 31, 2011. The authors applied a case-crossover design and compared the occurrence of invasive medical procedures 12 weeks before endocarditis with a corresponding 12-week time period exactly 1 year earlier. The authors considered all invasive nondental medical procedures except for those that are likely to be undertaken due to endocarditis or sepsis or due to infections that could possibly lead to endocarditis.
Results The authors identified 7,013 cases of infective endocarditis during the study period. Among others, several cardiovascular procedures, especially coronary artery bypass grafting; procedures of the skin and management of wounds; transfusion; dialysis; bone marrow puncture; and some endoscopies, particularly bronchoscopy, were strongly associated with an increased risk for infective endocarditis.
Conclusions This study suggests that several invasive nondental medical procedures are associated with a markedly increased risk for infective endocarditis.
Infective endocarditis is a relatively rare but life-threatening condition with high mortality and a high incidence of devastating complications among survivors (1,2). Prevention of infective endocarditis is clearly imperative, and therefore, there is a need for identification of possible risk factors.
Various invasive medical procedures might induce bacteremia and, hence, act as triggers for this condition. In animal studies, it is well documented that experimentally induced bacteremia can cause infective endocarditis and that it can be prevented by antibiotic prophylaxis (3,4). However, empirical data in humans on the potential dangers of medical procedures regarding infective endocarditis are very sparse. Evidence is mainly limited to a large number of poorly documented case reports showing a temporal relationship between procedures and infective endocarditis and to a few small case-control studies, which were largely confined to dental procedures (5). Due to lack of sufficient data, it is currently debated whether the risk for endocarditis with medical procedures is substantial or rather negligible (5,6). Consequently, the once widespread use of antibiotic prophylaxis before dental and other invasive medical procedures to prevent infective endocarditis has been questioned by recent guidelines (7–11). However, neither these recent nor previous guidelines were based on strong evidence (1).
As randomized controlled trials are not likely to be performed (10), there is a clear need for large population-based studies to quantify the risk of endocarditis after invasive medical procedures, which could lead to a more evidence-based recommendation on preventive measures such as antibiotic prophylaxis. For example, the American Heart Association called for large case-control studies to better assess the possible risk of infective endocarditis after invasive procedures (9). However, because many conditions could predispose for both medical procedures and endocarditis, and these factors might not be all captured and controlled, the results from such a case-control study are likely to be substantially confounded (9).
In the present study, we analyzed all infective endocarditis cases diagnosed in Sweden over 14 years and assessed the risk of endocarditis in relation to nondental invasive medical procedures. By using a case-crossover design (12), where each patient served as his or her own control, we minimized the risk for unmeasured confounding. Furthermore, by utilizing mandatory registration of hospital admissions and invasive procedures, we also eliminated the possibility of bias due to self-selection or biased recall.
We applied a case-crossover design (12,13), which includes only cases of a disease and applies self-matching by comparing exposure before the disease onset with a sample of disease free time in the past as control information to assess relative risks. The case-crossover design is a special case of the traditional matched case-control design, and it also shares features with crossover trials. Today, it is largely considered a standard method to study acute or so-called triggering effects of transient exposures. For example, much of our knowledge on triggering factors for acute myocardial infarction has come from case-crossover studies (14–16).
The main advantage of the design is that due to the within-person comparisons, stable characteristics cannot confound the observed associations in case-crossover studies. For example, several chronic cardiac conditions or lifestyle related factors such as diabetes might predispose individuals for infective endocarditis, and these conditions are also associated with the probability of undergoing an invasive medical procedure (17,18). However, in the case-crossover approach, due to self-matching, such factors are unlikely to confound the association between procedures and risk for infective endocarditis.
We used the Swedish National Patient Register to identify patients treated at hospital for infective endocarditis. All patients >20 years of age with a primary discharge diagnosis with International Classification of Diseases-10th Revision codes I33, I38, or I39 occurring between January 1, 1998, and December 31, 2011, in Sweden were included. Generally, in Sweden, modified Duke criteria are used for the diagnosis of endocarditis (19,20). For each individual, only the first episode of infective endocarditis was included in the analyses during this period. Exact date for the onset of the disease was not available. We used admission date as a proxy.
Invasive medical procedures
Both inpatient and outpatient medical procedures performed at hospitals and outpatient clinics are registered in Sweden in the National Patient Register, and linkage is possible to the same individual’s previous and subsequent patient records using the Swedish personal identification number.
The National Patient Register has covered the whole country since 1987, but it adopted the NOMESCO (Nordic Medico-Statistical Committee) classification for coding of medical procedures (21), which was used in the present study, on January 1, 1997. For some nonsurgical procedures, the coding system of the Swedish National Board of Health and Welfare was used. Outpatient procedures were available from January 1, 2001.
To avoid reverse causation, we have not investigated procedures, which are more likely to be undertaken due to endocarditis or sepsis than without these conditions. Similarly, to avoid interpretation problems, we have not considered procedures, which are likely to be performed due to infections potentially leading to endocarditis. Procedures not investigated in this study are listed in the Online Appendix. Because only a small fraction of dental procedures in Sweden are performed in hospitals and outpatient clinics and hence recorded in the National Patient Register, these were not investigated. Otherwise, all invasive procedures were the focus of interest in the present study and were included in our analyses.
Identification of individuals at high risk for infective endocarditis
Those with a prior episode of infective endocarditis occurring before January 1, 1998, those with prosthetic cardiac valve or prosthetic material used for cardiac valve repair, those with unrepaired cyanotic congenital heart disease, and those with palliative shunts or conduits and cardiac transplantation recipients were regarded as being at high risk for infective endocarditis (7,9).
In our stratified analyses, we looked at those having high risk and not having high risk for infective endocarditis separately. We also looked separately at those who had no known previous infective endocarditis.
For each patient, we compared the occurrence of invasive medical procedures 12 weeks before endocarditis (case period) with a 12-week time period exactly 1 year earlier (control period). Most previous studies investigating invasive procedures and infective endocarditis used a similar time window, 3 months or 12 weeks prior to disease onset (22–25). Inpatient procedures were not available before 1997. Thus, for this year, there was no corresponding information on the exposure in the preceding year. Consequently, we used 1997 only in the control period, and only endocarditis cases occurring from January 1, 1998, were included in our analyses on inpatient procedures. Similarly, 2001 (i.e., the first year with information on outpatient procedures) was used only in the control period when analyzing outpatient procedures, and only endocarditis cases occurring from January 1, 2002, were included in these analyses.
In case-control studies, the odds of exposure among cases is divided by the odds of exposure among controls, and the resulting odds ratio is an unbiased estimate of the relative risk of the disease. Similarly, in case-crossover studies, the odds ratio is an estimate of the relative risk (12). We calculated odds ratios together with 95% confidence intervals using conditional logistic regression, treating each case as 1 stratum.
Apart from relative risks, we also estimated absolute risk differences in relation to each procedure. For these calculations, we used the number of individuals at risk in Sweden in 2005 (i.e., approximately in the middle of the follow-up) provided by the Swedish National Board of Health and Welfare.
To assess effect modification, we stratified our analyses according to underlying risk for endocarditis (high risk vs. not high risk), age (≤65 years vs. >65 years), sex, and, in case of inpatient procedures, time period (January 1, 1998, to December 31, 2004 vs. January 1, 2005, to December 31, 2011).
Dates for outpatient procedures were always available. In contrast, the dates of inpatient procedures were often not recorded separately, only the date of admission and discharge for the hospitalization during which they were performed. Missing dates for inpatient procedures were set to the admission date for the given hospitalization. In our main analyses, to avoid reverse causation, we excluded all inpatient procedures with missing dates performed during the hospitalization with infective endocarditis. However, because this approach misses cases of endocarditis that develop in hospital after invasive procedures and might therefore lead to underestimation of the true effects for some procedures, we also presented relative risks without such exclusions. Statistical analyses were performed using SAS version 9.3 for Windows (SAS Institute, Cary, North Carolina).
Table 1 presents the general characteristics of 7,013 cases of infective endocarditis we identified during the study period. The majority of patients were men, and patients more frequently underwent an invasive procedure 12 weeks prior to endocarditis than during the same time span a year before.
Table 2 presents the relative risk and risk difference for infective endocarditis after invasive outpatient procedures. The relative risk for endocarditis was markedly increased after several procedures, like transfusion, coronary angiography, dialysis, bone marrow puncture, and some endoscopies (Central Illustration). Phacoemulsification was not associated with any elevated risk.
Transfusion, coronary angiography, dialysis, bone marrow puncture, and some endoscopies, particularly bronchoscopy, also appeared to be strongly associated with an increased risk for infective endocarditis as inpatient procedures. In addition, cardiovascular procedures, especially coronary artery bypass grafting; procedures of the skin; and management of wounds appeared to be associated with the highest risk (Table 3, Central Illustration).
In our secondary analyses, when we did not exclude procedures with a missing date performed during the hospitalization with infective endocarditis, the relative risks and risk differences were even more pronounced (Online Table 1).
In our stratified analyses (Online Table 2), relative risks, regarding both outpatient and inpatient procedures, tended to be somewhat higher among men and among younger patients. However, risk differences were considerably higher for older patients. High-risk patients had lower relative risk for outpatient procedures but higher relative risk for inpatient procedures than patients who were not at high risk. Risk differences for both outpatient and inpatient procedures were much higher for the high-risk group than for those who did not have high risk for infective endocarditis. Inpatient procedures appeared to be somewhat more dangerous when performed in 2005 and onwards than before. Patients without a history of infective endocarditis had similar relative risks and risk differences to those at low risk for infective endocarditis.
Invasive medical procedures appeared to be strongly associated with risk for infective endocarditis among both heroin users and those with an immune-compromised condition (data not shown). However, these analyses, especially the first one, had limited statistical power.
Some contemporary studies suggest that around 30% of all endocarditis patients have a recent exposure to health care (26,27). However, it is not clear to what extent infective endocarditis is really attributable to health care contact, because a comparison with health care exposure without endocarditis was not reported in these previous studies.
In the present study, several invasive medical procedures appeared to be potent triggering factors for infective endocarditis. Among therapeutic procedures, cardiovascular operations, especially coronary revascularizations; procedures on skin and wounds; chronic hemodialysis; blood transfusion; and various noncardiovascular operations carried the highest risk. Among diagnostic procedures, bone marrow puncture, coronary angiography, and some transluminal endoscopies, especially bronchoscopy, were the most important.
Several of these procedures, including hemodialysis (28), transluminal endoscopies (10,26), genitourinary procedures (26), cardiac catheterization (26,29), revascularization procedures (30), and other surgeries (23), were indicated as possible triggering factors for endocarditis in earlier studies. However, so far, our knowledge on the possible dangers of these procedures came from a temporal link established in case reports, simple case series without a comparison group, or from small case-control studies.
The once widespread use of antibiotic prophylaxis before dental and some other invasive medical procedures to prevent endocarditis in high-risk patients has recently been discouraged (7–10). Compared with earlier guidelines, newer guidelines severely restricted the group of patients for whom such prophylaxis is recommended. Also, a number of invasive medical procedures that were strongly associated with endocarditis risk in the present study are not considered high-risk procedures and no longer have indications for prophylaxis in these modern guidelines (7,9,10). The British NICE (National Institute for Health and Clinical Excellence) guideline went even further and suggested a complete cessation of antibiotic prophylaxis for infective endocarditis (8). However, a study from 2015 found a marked decrease in antibiotic use for prophylaxis and, at the same time, an increase in incidence of infective endocarditis after the NICE guideline was implemented in the United Kingdom; this raises the possibility that, contrary to previous beliefs, the danger associated with invasive medical procedures is not negligible (31). The corresponding U.S. guideline might have had a similar effect (32). These findings are in line with our study, showing a substantially high relative risk for infective endocarditis 12 weeks after several common invasive medical procedures.
According to our analyses of absolute risks, 476 (i.e., 1/0.21%) high-risk patients would need to receive prophylactic antibiotics in conjunction with an invasive inpatient procedure to prevent 1 case of infective endocarditis in the 12-week period following the procedure. The corresponding number was 278 for high-risk individuals in our secondary analyses, when we did not exclude procedures with a missing date performed during the hospitalization with infective endocarditis. The numbers refer to an average inpatient procedure, and thus, the number needed to treat is considerably lower for some particularly high-risk procedures. For example, these numbers were 83 and 40 for bronchoscopy in our main and secondary analyses, respectively. Because these calculations assume that antibiotic prophylaxis provides a perfect protection, which is unlikely, the real number of patients needed to be treated is somewhat higher.
Apart from a potential reconsideration of antibiotic prophylaxis before certain particularly high-risk procedures, our findings also suggest that further improvement of aseptic measures before and during invasive procedures to minimize the excess risk for infective endocarditis is of crucial importance. Our results might also be used in cost-benefit considerations of elective procedures for patients at high risk. Furthermore, we believe that both health care providers and patients at high risk should be aware of the markedly increased risk of endocarditis following some procedures. This could lead to an earlier identification of infective endocarditis, which, due to the often nonspecific symptoms, could go undetected for a long time (28). Early diagnosis is a major determinant of successful therapy for this condition (11,28).
Study strengths and limitations
This is by far the largest study on invasive procedures as risk factors for infective endocarditis. Previous studies were statistically underpowered to study relatively rare exposures like medical procedures with temporary effects. Also, to the best of our knowledge, the present study is the first to investigate nondental medical procedures as triggers for endocarditis with the case-crossover method, a research design specifically developed to study effects of short-term exposures (12). Due to this approach, our results, in contrast to previous studies, are unlikely to be confounded by stable patient characteristics like chronic underlying cardiac conditions, chronic immune compromised conditions, or stable lifestyle-related factors predisposing to infective endocarditis (9,17,18).
Other than case reports or simple case series, previous knowledge largely came from small case-control studies. In traditional case-control studies, typically only a fraction of potential control subjects agree to serve as a control, which is a major methodological concern. In such settings, the importance of medical procedures might considerably be overestimated or underestimated. Differential memory of cases and control participants on the history of procedures could also limit the interpretation of the results in these types of studies (33). Mandatory registration of hospital admissions and invasive procedures both at hospitals and outpatient clinics in Sweden eliminated the possibility of bias due to self-selection or biased recall in the current study.
Apart from clear strengths, important limitations of our study have to be considered. Our most important limitation is the uncertainty of the date of real onset of endocarditis and the date of some procedures. However, it is important to recognize that such uncertainties do not lead to overestimation of effects in a case-crossover study (12,13), and therefore, the relative risks in our main analyses are conservative estimates.
We excluded a number of procedures that were possibly performed due to conditions leading to endocarditis. Nevertheless, it is an important limitation that the indication for invasive medical procedures and the effect of these procedures were not separable in our study. Consequently, the interpretation of some of our results is not straightforward. For example, colon cancer, a reason for performing colonoscopy, might increase the risk for infective endocarditis (28). However, given the available knowledge on causes and risk factors for infective endocarditis, we do not expect a major confounding by indication, which could explain the markedly increased risk for endocarditis following the procedures investigated in the present study.
The Swedish Patient Register has a virtually complete coverage of somatic care at hospitals and generally a high accuracy of diagnoses (34). Nearly all endocarditis patients in Sweden are referred to hospital (35), and transesophageal echocardiography was very extensively used during the study period (36). Therefore, we can suppose that both the specificity and sensitivity of the endocarditis diagnosis was high. To further increase specificity, which is generally a greater concern than sensitivity in similar settings (33), we restricted to cases who received infective endocarditis as a primary diagnosis. In the Swedish National Patient Register, the primary diagnosis is given to the main reason for the patient’s need for treatment or investigation at a given hospitalization, and it generally has a higher specificity than secondary diagnoses (37,38). In our previous small validation study, we found that the hospital discharge diagnoses were highly reliable, with <2% of those with infective endocarditis diagnoses failed to meet the modified Duke criteria (39). Furthermore, it is important to emphasize that imperfect accuracy of the diagnosis typically leads to an underestimation of effects; thus, it cannot explain the markedly increased risk we observed after several procedures (33).
We had no information about the use of antibiotic prophylaxis to prevent infective endocarditis. It was likely to be infrequent for most nondental procedures, and antibiotic prophylaxis could have only masked the effects. Finally, an important limitation was the lack of microbiological data for organisms causing endocarditis.
Our study suggests that several invasive nondental medical procedures are associated with a markedly increased risk for infective endocarditis. Health care professionals performing particularly risk-prone procedures should consider every possible preventive measure to decrease the excess risk. Furthermore, increased awareness of the heightened risk in the vulnerable period after these procedures might lead to earlier diagnosis with a better chance for successful therapy and for avoiding the feared complications of the disease.
COMPETENCY IN MEDICAL KNOWLEDGE: Cardiovascular procedures, especially coronary artery bypass grafting; procedures involving the skin and management of wounds; transfusion; dialysis; bone marrow puncture; and certain endoscopic procedures, particularly bronchoscopy, are associated with an increased risk of developing infective endocarditis.
TRANSLATIONAL OUTLOOK: Future studies are needed to assess the efficacy of prophylactic strategies to reduce the risk of infective endocarditis in patients undergoing these particular procedures and to identify patient characteristics that increase the risk to target these measures more precisely.
Grants were provided from the Liaison Committee between the Central Norway Regional Health Authority and the Norwegian University of Science and Technology and from the Swedish Research Council. The funding sources had no role in the study design; collection, analysis, and interpretation of data; writing of the report, or decision to submit the paper for publication. All authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- International Classification of Diseases
- Nordic Medico-Statistical Committee
- Received February 14, 2018.
- Revision received March 12, 2018.
- Accepted March 19, 2018.
- 2018 American College of Cardiology Foundation
- Wilson W.,
- Taubert K.A.,
- Gewitz M.,
- et al.
- Habib G.,
- Hoen B.,
- Tornos P.,
- et al.
- Richey R.,
- Wray D.,
- Stokes T.,
- for the Guideline Development Group
- Wilson W.,
- Taubert K.A.,
- Gewitz M.,
- et al.
- Thornhill M.H.,
- Dayer M.,
- Lockhart P.B.,
- et al.
- Mittleman M.A.,
- Maclure M.,
- Tofler G.H.,
- Sherwood J.B.,
- Goldberg R.J.,
- Muller J.E.,
- for the Determinants of Myocardial Infarction Onset Study Investigators
- Strom B.L.,
- Abrutyn E.,
- Berlin J.A.,
- et al.
- Durack D.T.,
- Lukes A.S.,
- Bright D.K.
- Nordic Medico-Statistical Committee
- Chen P.C.,
- Tung Y.C.,
- Wu P.W.,
- et al.
- Jayasuriya S.,
- Movahed M.R.
- Pant S.,
- Patel N.J.,
- Deshmukh A.,
- et al.
- Rothman K.J.,
- Greenland S.,
- Lash T.L.
- Asgeirsson H.,
- Thalme A.,
- Kristjansson M.,
- Weiland O.