Is Asymptomatic Bacteriuria a Risk Factor for Prosthetic Joint Infection ?

Ricardo Sousa, Ernesto Muñoz-Mahamud, Jonathan Quayle, Luis Dias da Costa, Cristina Casals, Phylip Scott, Pedro Leite, Paz Vilanova, Sebastian Garcia, Maria Helena Ramos, Joana Dias, Alex Soriano, and Andrea Guyot Departments of Orthopaedics, and Microbiology, Centro Hospitalar do Porto–Hospital de Santo António, and Department of Biostatistics, Administração Regional de Saúde do Norte, Porto, Portugal; Department of Orthopaedics, Bone and Joint Infection Unit, and Department of Infectious Diseases, Hospital Clínic of Barcelona, Spain; and Department of Orthopaedics and Department of Microbiology, Frimley Park Hospital, Frimley, United Kingdom

Prosthetic joint infection (PJI) is one of the most challenging and frequent complications after joint arthroplasty [1,2].As the demand for total hip and knee joint arthroplasty is expected to increase substantially during the coming decades, so too will the economic burden of prosthetic infections [3,4].Because the incidence of this complication seems to be on the rise worldwide despite antiseptic skin preparation and antibiotic prophylaxis, identifying potentially modifiable preoperative risk factors is of great interest [5,6].
The concern with the genitourinary tract as a possible source of hematogenous seeding has been present as far back as the 1970s, when a few case reports [7][8][9][10] and a retrospective study [11] found a relation between patients with deep joint infection and perioperative urinary tract infection (UTI).Although there seems to be enough evidence supporting a relation between postoperative UTI and PJI [11][12][13][14][15], literature studying the correlation between asymptomatic bacteriuria (ASB) and surgical site infection after joint arthroplasty is scarce [15][16][17][18][19].As a consequence, this finding is not currently considered a criterion for delaying total joint replacement surgery [13].The aims of our study were to describe the prevalence of ASB among candidates for elective total hip and knee arthroplasty, to determine whether ASB is associated with an increased risk of PJI, and, finally, to learn whether an appropriate course of preoperative antibiotics is effective in reducing the risk of PJI.

MATERIAL AND METHODS
From January 2010 to December 2011, in 3 institutions from the United Kingdom, Portugal, and Spain, a preoperative urine culture was collected from all patients undergoing total hip or knee arthroplasty.Relevant information about demographics, body mass index (BMI), diabetes mellitus, American Society of Anesthesiologists (ASA) physical status classification system, and duration of surgery was collected retrospectively, but unfortunately information regarding the duration of surgery and diabetes mellitus was not possible to gather in the British participating institution.The duration of surgery was categorized as less or more than the 75th percentile to account for differences in mean surgical times between centers [20].
In all patients a urine sample was obtained (regardless of dipstick test results), placed in a sterile container, and cultured using conventional methods in the microbiology laboratory.All isolated microorganisms were identified with standard biochemical procedures.ASB was defined as the isolation of ≥10 5 colony-forming units/mL in the absence of symptoms or signs of UTI.
Preoperative treatment of ASB was decided by the treating physician and was not mandatory or randomized.For treatment, an 8-day course of oral antibiotics (according to in vitro susceptibility test) was given the week before hospital admission.Control urine cultures after treatment were not mandatory, and only 26 of the 154 treated patients with ASB had repeat urine cultures (all of them negative) before surgery.In the untreated candidates with ASB no further antibiotics were given preoperatively or perioperatively, nor were any other additional prophylactic measures taken other than normal prophylaxis regimen for each institution (a single 2-g dose of cefazolin in the Portuguese institution, 1.5 g of cefuroxime in the Spanish institution or 600 mg of teicoplanin plus 120 mg of gentamicin in the UK institution during the induction of anesthesia).
Postoperative UTI in the early postoperative period was diagnosed when urinary symptoms of infection were present and urine culture showed bacterial growth (≥10 5 colonyforming units/mL).After hospital discharge, patients were followed up for ≥12 months.The main outcome of the study was the diagnosis of PJI in the first year after surgery, accordingly to the CDC definition of implant-related surgical site infection [21].

Statistical Analysis
Categorical variables were presented as number of patients and percentages.Continuous variables were compared using the nonparametric Mann-Whitney test because the study population did not meet the normality assumption.Proportions were compared using χ 2 and Fisher exact tests when necessary, with statistical significance defined as a 2-tailed P value <.05.
To test the association between study variables and outcome (PJI), logistic regression models were fitted, accounting for spatial clustering because data came from 3 centers (center effect).A multivariable logistic regression model was developed, including variables with P values ≤.20 from the univariable analysis.The role of the variables as potential modifier effect was also studied.The model fit was assessed with the Hosmer-Lemeshow test.Statistical analysis was done using the SPSS program (version 19.0; SPSS).

RESULTS
A total of 2497 patients were included during the study period, with similar proportions undergoing total hip arthroplasty (n = 1248) and total knee arthroplasty (n = 1247).The distribution among participating institutions is shown in Table 1; there were no significant differences between them for ASB or PJI.Approximately two-thirds of patients (63.0%) were women, and the mean age was 68.0 years.ASB was diagnosed in 12.1% of the cohort, in 16.3% of women and 5.0% of men (odds ratio [OR], 3.67; 95% confidence interval [CI], 2.65-5.09;P < .001).
Table 2 shows the microorganisms isolated in these patients.ASB was significantly more common in obese women >71 years  old with an ASA score of ≥3 (Table 3).There was no significant increase in postoperative UTI prevalence in the ASB group.
Postoperative UTI was diagnosed in 26 patients, only 5 of whom had preoperative ASB.All 5 cases occurred in the untreated ASB group, and the same organism was present in urine samples obtained before and after surgery.No resistant strains to the prophylaxis regimen were present in these cases.Despite the higher risk of PJI in the early postoperative UTI group, microorganisms isolated in UTI were always different from those in PJI.
A total of 51 microorganisms were isolated in 43 cases of PJI (Table 5).The proportions of PJI cases involving gram-negative bacteria were 2.0% in the ASB group (6 of 303 patients) and 0.2% in the non-ASB group (5 of 2194 patients; OR, 8.84; 95% CI, 2.68-29.16;P = .001).In 32 of 43 patients (74%), infection was diagnosed within the first 6 weeks after surgery.In the other 11, infection was diagnosed after the first 3 months.The proportions of cases with gram-negative bacteria were identical in the early and late infection groups (8 of 32 and 3 of 11, respectively).Further details on isolated sicroorganisms and time interval   between index surgery and diagnosis of infection can be found on supplementary materials.Microorganisms isolated in PJI were not the same as those in preoperative urine cultures in any patient with ASB (Table 6).Although pulsed-field gel electrophoresis was not performed, Escherichia coli isolates from the urine and the joint of patient 4 (who had untreated ASB) presented different antibiotic resistance profiles, suggesting that they were unrelated.No other variables were significantly associated with gram-negative infections.
A subanalysis was performed on the effect of preoperative treatment of ASB.Because there was no randomization to treatment, there may be biases in the selection of patients for treatment.To address possible unrecognized selection biases, we performed a propensity analysis, and the results did not differ from those obtained with logistic regression analysis.The main risk factors were similar in both groups, except the proportion of female patients was significantly higher in the treated group (Table 7).Infection rates were 4.7% (7 of 149 patients) in the untreated and 3.9% (6 of 154 patients) in the treated ASB group.There was no significant difference between groups (OR, 0.82; 95% CI, .27-2.51;P = .78),and both groups had a significantly higher rate of PJI than the non-ASB group (OR for untreated vs no ASB, 3.56 [95% CI, 1.54-8.24;P = .007];OR for treated vs no ASB, 2.85 [1.20-6.74;P = .03]).

DISCUSSION
PJI is a serious complication of arthroplasty surgery, and its prevention is a priority for US Department of Health and Human Services [22].With contemporary aseptic conditions and the use of antibiotic prophylaxis, the infection rate after joint arthroplasty has significantly decreased [23].Nevertheless, there seems to be a worldwide trend toward an increased incidence of this complication; a recent retrospective study performed in California demonstrated surgical site infection rates of 2.3% after total hip and 2% after total knee arthroplasty [24], rates even higher than in the present study (1.7%).Although many orthopedic surgeons worry about an undiagnosed UTI as a possible source of bacterial contamination, the real effect of ASB as a preoperative marker or risk factor for PJI has not been well established, and, to our knowledge, our study is the largest case series that addresses this matter.
The prevalence of ASB in our cohort was 12.1% (16.3% in women and 5.0% in men), comparable to previously reported results in total joint replacement candidates, ranging from 4% to 19% [16,17,19,25].It is also in agreement with previous reports of the prevalence of ASB in similar age groups in the general population [26,27].In addition, older age, female sex, BMI >30 kg/m 2 , and a higher ASA score were significantly more prevalent in the ASB population, consistent with findings of previous studies [26][27][28][29].
Our data clearly show that patients with preoperative ASB have a significantly higher risk of PJI than patients without ASB (4.3% vs 1.4%).Two classic studies are often cited to illustrate the lack of association between ASB and PJI.Ritter and Fechtman [16] studied 364 total joint replacements and found infections rate of 2.9% (1 in 35) in the ASB group and 0.6% (2 in 329) in the non-ASB group.Glynn and Sheehan [17] reported data from 299 patients who underwent total joint replacement, and found the infection rates of 3.5% (2 of 57) in patients with bacteriuria and 0% (0 of 242) in those without bacteriuria.The results of the latter study should be interpreted cautiously because it included not only patients with ASB but also patients with symptomatic UTIs.Furthermore, different antibiotic treatment regimens (before, during and even exclusively after surgery) were used in patients with bacteriuria.Although both studies found a nonsignificantly higher infection rate in the bacteriuria group, neither set of authors assumed a potential relationship because the microorganisms isolated from surgical site infections and urine cultures were not the same.Nevertheless, our cohort is larger, which made it possible to show not only an increased risk for PJI but also a significant higher rate of gram-negative infections.It is significant that, in accordance with previous findings, microorganisms found in PJI isolates did not direct correspond directly to the species found in urine cultures.
Because part of our cohort of patients with ASB was treated with a course of preoperative antibiotics, we analyzed the potential benefit of this strategy.There was difference in infection rates between untreated and treated groups (4.7% vs 3.9%, respectively).We know of only a single study in which patients with ASB have undergone arthroplasty after randomization to antibiotic therapy [19].The authors identified no case of PJI of urinary origin in patients with ASB, regardless of treatment with specific antibiotics [19].
Although it is extremely difficult to know the exact pathogenesis of infection (hematogenous or acquired during surgery), the majority of infections in our series occurred within 6 weeks after surgery, suggesting that most were caused by wound contamination during surgery [30].The lack of correspondence between ASB and PJI microorganisms could be explained by the fact that patients with ASB are at risk for recurrence with a different organism.However, the short interval between preoperative antibiotic treatment and surgery makes recurrent ASB/UTI with a different organism and subsequent hematogenous seeding of the new organism unlikely to be responsible for most gram-negative infections found.
An alternative explanation for the increased risk of infection could be a relationship between ASB and other known risk factors admittedly more common in patients with ASB.However, the multivariable model showed ASB to be an independent predictor of PJI after adjustment for the main known risk factors (sex, age, location, duration of surgery, BMI, and comorbid conditions), suggesting that it may actually be a surrogate marker for some other feature not yet known.One plausible explanation could be that skin flora differs between patients with or without ASB.The findings of Ollivere et al [31] in a cohort of 558 patients undergoing arthroplasty support the fact that patients with ASB are at increased risk for wound contamination.Fifteen of the 39 patients (38.5%) with preoperative positive urine culture showed some form of postoperative delayed wound healing or confirmed superficial infection, compared with 83 out of 511 (16.2%) in the other subgroup.
Our study has limitations.The first is that the definition of ASB relies on a single urine sample.This is not entirely in accordance with Infectious Diseases Society of America guidelines for the diagnosis of ASB in adults, which require 2 consecutive urine specimens with isolation of the same bacterial strain in women [32].The second limitation is the lack of routine control urine cultures to confirm eradication of ASB before surgery.These are not ideal conditions for assessing the precise value of urine sterilization before arthroplasty.Nonetheless, the prescribed treatment is the usual clinical practice and has been shown to be highly effective in treating UTI [33].Data were not available regarding other possible confounding risk factors for ASB and PJI, such as urinary incontinence, immobility, residence in a nursing home or long-term care facility, or the presence and duration of urinary catheters before and after surgery.Furthermore, preoperative antibiotic prophylaxis regimens were not the same in all 3 institutions and respective compliance rates are not available; however, this concern was addressed by accounting for spatial clustering in our statistical analysis.Finally, antibiotic treatment for ASB was not randomized, which may lead to selection biases (ie, physicians may have chosen to treat the patients they considered at higher risk).However, a propensity analysis was performed and comparing showed no significant differences between treated and untreated ASB populations, apart from a higher proportion of female patients in the treated group.Because female sex was not an independent risk factor for PJI, that difference does not seem be clinically relevant.Possible negative consequences of preoperative antibiotic treatment of ASB (eg, Clostridium difficile infection) were not recorded.On the other hand, the analysis of the influence of treatment helped us understand better the role of ASB.
In conclusion, ASB is a common finding among total joint arthroplasty candidates and it emerges as an independent risk factor for PJI.Our results indicate that there is no direct seeding of urine microorganisms onto the surgical site but rather that ASB is a surrogate marker for some condition that increases the risk of bacterial colonization or infection, especially due to gramnegative microorganisms.Preoperative antibiotic treatment did not show benefit, so postponing surgery or even treating patients with known ASB before surgery cannot be recommended.

Table 1 .
Differences Between the 3 Participating Institutions in Asymptomatic Bacteriuria and Prosthetic Joint Infection Findings

Table 2 .
Microorganisms Isolated From Preoperative Urine Cultures

Table 3 .
Main Population Characteristics in Patients With or Without Asymptomatic Bacteriuria Abbreviations: ASA, American Society of Anesthesiologists; ASB, asymptomatic bacteriuria; BMI, body mass index; CI, confidence interval; OR, odds ratio; UTI, urinary tract infection.a Unless otherwise specified, data represent No. (%) of patients.b Data available for 985 patients.c Data available for 2278 patients.d Data available for 993 patients.e Data available for 1960 patients.

Table 4 .
Risk Factors for Prosthetic Joint Infection a Unless otherwise specified, data represent No. (%) of patients.b Data available for 985 patients.c Data available for 2278 patients.d Data available for 993 patients.e Data available for 1960 patients.

Table 5 .
Microorganisms Isolated in 43 Patients with Prosthetic Joint Infection Polymicrobial refers to number of polymicrobial PJI cases: specific microorganisms involved are reflected under their respective categories.Abbreviation: ASB, asymptomatic bacteriuria.

Table 7 .
Comparison Between Treated and Untreated Patients With Asymptomatic Bacteriuria a Unless otherwise specified, data represent No. (%) of patients.b Data available for 270 patients.c Data available for 118 patients.d Data available for 242 patients.