Dental procedures may result in bacteremia that can cause infection at remote sites, particularly at prosthetic implants. Antibiotic prophylaxis before an invasive dental intervention is frequently believed crucial for preventing such infections. In patients who had total joint arthroplasty (TJA), prosthetic joint infection (PJI) is a rare but recognized complication associated with significant morbidity and mortality.Reference Zimmerli, Trampuz and Ochsner 1 , Reference Berbari, Hanssen and Duffy 2 In 1999, LaPorte reported that transient bacteremia following a dental procedure might lead to PJI in patients with total hip arthroplasty.Reference LaPorte, Waldman, Mont and Hungerford 3 The use of antibiotics for preventing PJI after dental procedures, nevertheless, has remained controversial for decades, with inconclusive advisory statements from different medical associations.Reference LaPorte, Waldman, Mont and Hungerford 3 – Reference Colterjohn, de Beer, Petruccelli, Zabtia and Winemaker 20
The American Academy of Orthopedic Surgeons (AAOS) officially recommended in 1997 that antibiotic prophylaxis was indicated when TJA patients underwent invasive procedures such as dental extractions and periodontal treatment during the first 2 postoperative years, 21 as this period was most critical for developing PJI. Later in 2009, the AAOS suggested that clinicians should consider antibiotic prophylaxis for TJA patients prior to any invasive procedure that may cause bacteremia, for preventing adverse outcomes and obviating the cost of treating PJI.Reference Napenas, Lockhart Pb Fau - Epstein and Epstein 22 In view of the paucity of relevant data, the AAOS in conjunction with the American Dental Association (ADA) stated conservatively in 2012 that practitioners might consider discontinuing routine prescription of prophylactic antibiotics for patients with prosthetic joint implants undergoing dental procedures, and they acknowledged that they were unable to recommend for or against the use of topical oral antimicrobials for dental intervention in patients with prosthetic joints or other orthopedic implants.Reference Hamedani 23
Because the guideline recommendations are not grounded on sufficient evidence, many clinicians tend to maintain the practice of antibiotic prophylaxis, neglecting the potential untoward consequences of misusing antibiotics. To bridge this knowledge gap, this nationwide, population-based study sought to evaluate effectiveness of prophylactic antibiotics for dental procedures in TJA patients. We set out to determine the association between invasive dental treatment and incidence of PJI during the first 2 postoperative years; we then examined whether antibiotic prophylaxis was associated with an attenuated risk of PJI following dental procedures.
MATERIALS AND METHODS
Study Design and Data Source
The data for this retrospective, population-based cohort study were obtained from the Taiwan National Health Insurance Research Database (NHIRD), which was constructed and maintained for academic purposes by the Taiwan National Health Research Institutes (NHRI). The NHIRD contains comprehensive registration and claims data of all beneficiaries enrolled in the National Health Insurance (NHI) program. Because the NHI program is a mandatory and universal insurance policy in Taiwan, the entire population of Taiwan is covered by the NHI and thus is included in the NHIRD.Reference Hsu 24 Disease status in the NHIRD is coded using the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM). The coding accuracy for major diseases has been validated in previous studies.Reference Cheng, Kao, Lin, Lee and Lai 25 , Reference Cheng, Lee, Chen, Li, Lin and Yang 26 Details regarding the ICD-9-CM and procedure codes used in this study are presented in the appendix (Supplementary Tables 1 and 2). The acquisition and analysis of data for this study were approved by the institutional review board of E-Da Hospital (EMRP-103-011; EMRP-103-012) and the research ethics committee of the Taiwan NHRI (NHIRD-103-116).
Definition of the Study Cohorts
We identified from the NHIRD all patients who underwent total hip arthroplasty (THA) or total knee arthroplasty (TKA) for the first time between January 1, 1997, and November 30, 2009, and categorized them into dental and nondental cohorts. The dental cohort comprised patients receiving dental procedures, which included periodontal treatment and surgical or nonsurgical tooth extraction, within the first 2 years after arthroplasty. The nondental cohort comprised patients who did not undergo any dental treatment during this period. The day when the patients underwent dental procedures was assigned as the index date of study baseline. The nondental cohort was matched 1:1 with the dental cohort by index date, age, sex, and propensity score for undergoing a dental procedure.
Because the database contains details of each and every prescription of antibiotics, which can be identified by specific codes, we were able to extract pertinent information regarding when, where, how much, how frequently, and for how long an antibiotic was prescribed. Accordingly, the patients in the dental cohort were further subgrouped on the basis of whether they received prophylactic antibiotics in the week preceding the dental procedure. Given that the indication for antibiotics could not be directly confirmed, the definition of prophylactic use required the following criteria: (1) use of antibiotics within 1 week preceding the dental procedure, (2) medications typical for prophylactic purpose, including the first- or second-generation cephalosporins, penicillins (eg, oxacillin, ampicillin, and amoxicillin), or clindamycin, and, (3) otherwise no indication of an infectious disease. The antibiotic subcohort was matched 1:1 with the nonantibiotic subcohort by index date, age, sex, and propensity score of antibiotic prophylaxis.
Definition of the Study Outcome and Comorbidities
The study outcome was PJI, which was defined as the occurrence of surgically treated osteomyelitis or septic arthritis at the side of joint implants. The surgical intervention could be debridement, prosthesis removal, or resection arthroplasty. We used the Charlson comorbidity index to assess baseline comorbidities that had existed prior to the index date.Reference Charlson, Pompei, Ales and MacKenzie 27 All participants were followed up until occurrence of the study outcome, withdrawal from the NHI, or 1 year after the dental procedure.
Data Analysis and Statistical Tests
Continuous variables are presented as means and standard deviations and categorical variables as numbers and proportions. Propensity score using the greedy approach was applied to match the study cohorts. The propensity scores for dental procedures and those for antibiotic prophylaxis were generated using logistic regression built on baseline characteristics, including specific diseases listed in the Charlson comorbidity index. Cumulative incidences of the study outcomes were calculated and compared between the cohorts. In the unadjusted analysis, the between-cohort difference in the cumulative incidence was analyzed using the 2-tailed log-rank test. A Cox proportional hazard model adjusted potential confounding factors was used to evaluate the multivariate-adjusted association between dental procedures and PJI and to determine whether antibiotic prophylaxis was associated with a decreased risk. Crude and adjusted hazard ratios (HRs) were reported along with their 95% confidence intervals (CIs). All data management and calculations were performed using SPSS version 10.0 software (SPSS, Chicago, IL). All statistical results were considered significant at a P value<.05.
RESULTS
Baseline Characteristics of the Study Population
We identified 255,568 patients from the NHIRD who underwent TKA or THA. Of these patients, 61,917 patients underwent dental procedures, and 193,651 patients did not undergo any dental treatment (Supplementary Figure 1). With 1:1 individual matching by age, sex, propensity score, and the index date, the dental and nondental cohorts each consisted of 57,066 patients. The baseline characteristics, including the comorbidity index, were similar between the 2 cohorts (Table 1).
TABLE 1 Baseline Characteristics of the Study Subjects

NOTE. Continuous variables are expressed with mean±standard deviation, and categorical variables with actual number and proportion.
Among those who underwent dental procedures, 54,977 patients received prophylactic antibiotics and 6,596 did not (Fig. 1). Based on individual matching again, 6,513 pairs of patients were grouped into the antibiotic and nonantibiotic subcohorts, respectively. Both subcohorts were generally comparable in baseline characteristics, although the antibiotic subcohort had more severe in the underlying comorbidities (Table 2).

FIGURE 1 Cumulative incidences of the prosthetic joint infection did not differ between either the dental and nondental subcohorts (panel A) or between the antibiotic and nonantibiotic subcohorts (panel B).
TABLE 2 Baseline Characteristics of the Subcohorts According to Antibiotic Prophylaxis

NOTE. Continuous variables are expressed with mean±standard deviation, and categorical variables with actual number and proportion.
Incidence of the PJI
At 1 year after the index date, 328 patients (0.57%) in the dental subcohort and 348 patients (0.61%) in the nondental subcohort developed PJI. The 1-year cumulative incidences of PJI were 0.6% (95% CI, 0.6%–0.7%) in the dental subcohort and 0.6% (95% CI, 0.5%–0.6%) in the nondental cohorts (P=.3) (Fig. 1A).
During follow-up, infection of the prosthetic joint only occurred in 13 patients (0.2%) in the antibiotic subcohort and 12 patients (0.18%) in the nonantibiotic subcohort (Fig. 1B), and the corresponding 1-year incidence of PJI was 0.2% (95% CI, 0.1%–0.3%) in both of these subcohorts (P=.83). We tried to analyze the infection rates according to the type of dental procedures, and we found that the rates of PJI did not differ among the 3 types of dental intervention (Supplementary Table 3). In light of the low event incidence in each type of dental treatment, it was appropriate to group them together for further analysis.
Association Between Dental Procedure and PJI in TJA Patients
The unadjusted analysis showed that PJI was associated with congestive heart failure (CHF; crude hazard ratio [HR], 1.88; 95% CI, 1.05–3.39), chronic lung disease (crude HR, 1.50; 95% CI, 1.09–2.07), chronic liver disease (crude HR, 1.60; 95% CI, 1.10–2.35), and malignancy (crude HR, 1.89; 95% CI, 1.20–2.29). In the multivariate-adjusted analysis, risk factors for PJI were chronic lung disease (adjusted HR, 2.02; 95% CI, 1.13–3.63), chronic liver disease (adjusted HR, 2.09; 95% CI, 1.12–3.90), and malignant disease (adjusted HR, 2.47; 95% CI, 1.45–4.22). Dental procedure was not linked to an occurrence of PJI (Table 3). As a sensitivity test, the stratified analyses consistently revealed among various subgroups that there was no association between dental treatment and PJI (Supplementary Figure 2).
TABLE 3 Cox Proportional Hazard Model for Prosthetic Joint Infection in Total Joint Arthroplasty Patients

NOTE. HR, hazard ratio; CI, confidence interval; NA, not applicable due to few events.
Association Between Prophylactic Antibiotics and PJI After Dental Procedures
In the unadjusted analysis, the following comorbidities were associated with PJI: CHF (crude HR, 4.72; 95% CI, 1.77–12.58), chronic lung disease (crude HR, 2.69; 95% CI, 1.22–5.92), diabetes with end-organ damage (crude HR, 5.05; 95% CI, 1.73–14.71), hemiplegia (crude HR, 6.36; 95% CI, 1.5–26.95), and metastatic malignancy (crude HR, 7.58; 95% CI, 1.03–55.97).
After adjusting for multiple factors, we identified CHF (adjusted HR, 3.20; 95% CI, 1.03–9.87) and diabetes with end-organ damage (adjusted HR, 5.99; 95% CI, 1.16–31.00) as independent risk factors for PJI following a dental procedure. There was no association between antibiotic prophylaxis and the risk of subsequent PJI (TABLE 4).
Table 4 Cox Proportional Hazard Model for Prosthetic Joint Infection After Dental Procedures

NOTE. HR, hazard ratio; CI, confidence interval; NA, not applicable due to few events.
DISCUSSION
From our analysis of a national healthcare database, this population-based cohort study demonstrated that dental intervention was not associated with the occurrence of PJI within 2 years after TJA and revealed that the risk of PJI following dental procedures was unaffected by administration of prophylactic antibiotics. The risk was determined predominantly by underlying comorbidities such as chronic pulmonary, hepatic, or cardiovascular illness. These findings provide ample evidence to argue against the practice of routinely giving antibiotics to TJA patients for receiving dental treatment.
The oral cavity contains a complex microbiome composed of more than 42 known genera of bacteria.Reference Bagg, MacFarlane, Poxton, Smith and Bagg 28 – Reference Chris, Miller and Palenik. Charles 30 It has been estimated that 10 million to 100 million bacteria per milliliter occur in the saliva and that 200 billion bacteria occur in a dental plaque. Some of these flora are potentially pathogenic, such as Streptococcus spp., Lactobacillus spp., Actinomyces spp., and Bacteroides spp. 31 , Reference Featherstone 32 and may lead to dental caries, periodontitis, or deep neck infection. Some may even have the opportunity to invade tissue and result in bacteremia that in turns causes infection at a remote site like prostheses.
Although the AAOS and ADA have jointly recommended in their current practice guidelines that clinicians should think twice before continuing to prescribe antibiotic prophylaxis for dental procedures in TJA patients, the committee admitted that the evidence level was very limited to support this recommendation. Because data that can elucidate cost and effectiveness of this popular practice are scarce, the AAOS actually encourages practitioners to share decision making with their patients and suggests that patient preference should have a substantial influencing role in deciding whether to forsake this practice.Reference Hamedani 23
Without sufficient evidence as the guidance, healthcare professionals may arbitrarily maintain contrasting perspectives in assessing the risks and benefits of invasive dental procedures and the use of prophylactic antibiotics. Whereas orthopedists must manage the infected prosthetic joint, dentists who perform the procedure will encounter adverse reactions to antibiotic prophylaxis when they occur. Compared with dentists, more orthopedists believed that PJI could result from dental treatment and that antibiotics should be necessary for routine or lengthy dental procedures.Reference Kingston, Kiely and McElwain 33 In a survey of Canadian orthopedic and dental surgeons, more orthopedic specialists than their dental colleagues were prone to use prophylactic antibiotics for a broader range of indications. After all, a clinician may feel safer to practice defensive medicine when the decision cannot be based on convincing evidence. Unfortunately, ineffective use of antibiotics is far from safe. It not only wastes valuable resources but may also precipitate the emergence of drug-resistant pathogens and potential side effects from the antibiotics such as Clostridium difficile infection and allergic reactions.
Previous studies have shown that PJI is a serious but uncommon complication that occurs in 0.3%–1.0% of patients after total hip replacement and in 1.0%–2.0% of patients after TKA.Reference Enzler, Berbari and Osmon 34 Thus, exploring different PJI incidences among these patients will require a large number of study subjects to mitigate the probability of type 2 error. For instance, 32,324 patients in each group are essential to detect an absolute difference of 0.2% (0.5% vs 0.7%). In view of the difficulty to conduct quality research on this issue, relevant literature is limited despite its important implication for public health. With the sample size of more than 100,000 patients, the statistical power of our study should be sufficient.
One may argue that a blinded randomized controlled trial is ideally the best way to confirm or refute the effectiveness of antibiotic prophylaxis, but such research is unlikely to be conducted in the foreseeable future. Although this study is indeed retrospective, meticulous efforts were taken in the study design to avoid bias and adjust for confounding. Matched ages, sex, and propensity scores were selected to optimize comparability between cohorts. Furthermore, the day of dental treatment was deliberately defined as the index date for matching and entry of observation. We did not match the study cohorts at the time of TJA or any outset prior to dental procedures to preclude the immortal time bias. Moreover, the sensitivity tests further affirmed our conclusion by showing a consistent result across subgroups of patients.
Previous studies have shown an important role of underlying comorbidity in the risk determination of PJI following joint replacement. Zhu et alReference Zhu, Zhang, Chen, Liu, Zhang and Zhang 35 reported in a systematic review that increased body mass index, corticosteroid therapy, hypoalbuminemia, wound dehiscence, superficial infection of the surgical site, and prolonged operation were detrimental. Consistent with our finding, CHF has also been identified as an independent risk factor. For instance, Rodriguez-Merchan et alReference Rodriguez-Merchan 36 summarized 22 studies and found CHF, rheumatology disease, psychoses, renal diseases, and chronic urinary tract infection were risk factors of infection following TKA. Bozic et alReference Bozic, Lau, Kurtz, Ong and Berry 37 analyzed the Medicare database to demonstrate that CHF was a risk factor for both postoperative mortality and infection following TKA. The mechanism of this association remains to be explored, but probably relates to the impaired immunity of the patients.
Our study had some limitations. First, laboratory and pathological data are unavailable in the NHIRD database, and we could not analyze patient-level data on microbiology of the infected prosthesis. Similarly, we had to use admission diagnoses and treatment procedures to define study participants and outcomes. Nonetheless, the study cohorts adopted identical definitions and thus should be comparable. Also, we could not determine the severity of the dental diseases, which might affect the chance of bacteremia. Nevertheless, the low incidence of PJI suggests that this factor should not have confounded the result. Moreover, confounding from unmeasured factors remains possible. We could not examine the potential influence of body weight, cigarette smoking, alcohol drinking, or dietary habits because this information is not included in the NHIRD. To overcome this limitation, the analysis was adjusted for a number of representative comorbidities such as chronic lung disease, diabetes mellitus, cardiovascular illness, and so on.
In conclusion, dental treatment in patients with TKA or THA is not a risk factor for PJI, which uncommonly occurs in <1% of patients 1 year after TJA. Moreover, there is no association between the prescription of antibiotics and the prevention of PJI following a dental procedure. After balancing risks and benefits as well as cost and effectiveness, our study reminds patients, clinicians, policy makers, and all stakeholders in this issue that less is more.
ACKNOWLEDGMENTS
This study is based on data from the National Health Insurance Research Database provided by the Bureau of National Health Insurance, Taiwan Department of Health, and managed by the National Health Research Institutes (NHRI). The interpretation and conclusions contained herein do not represent those of the Bureau of National Health Insurance or the NHRI. The authors are grateful to Ms. Chen Tzu-Shan for her efficient assistance.
Financial support: This study was supported by the Center for Database Research, E-DA Healthcare Group (grant no. CFDR-B-103-1-4).
Potential conflicts of interest: None of the authors have any conflict of interest to declare.
SUPPLEMENTARY MATERIAL
To view supplementary material for this article, please visit https://doi.org/10.1017/ice.2016.248