Fate of Two-Stage Reimplantation After Failed Irrigation and Debridement for Periprosthetic Hip Infection.

BACKGROUND: Irrigation and debridement is an attractive treatment option for acute perioperative and acute hematogenous periprosthetic hip infections. We ask the following questions: (1) What are the results of a two-stage reimplantation if preceded by debridement, antibiotics, and implant retention (DAIR) compared with two-stage reimplantation without an antecedent DAIR? and (2) Do McPherson Musculoskeletal Infection Society (MSIS) host criteria influence results?
METHODS: A total of 114 patients were treated with two-stage exchange for periprosthetic hip infection. Sixty-five patients were treated initially with a two-stage exchange, whereas 49 patients underwent an antecedent DAIR before a two-stage exchange. Patients were classified based on MSIS host criteria. Demographics demonstrated homogeneity between cohorts. Failure was defined as return to the operating room for infection, draining sinus, or systemic infection.
RESULTS: Treatment failure occurred in 42.9% (21 of 49) of patients treated with an antecedent DAIR. In contrast, treatment failure occurred in only 12.3% (8 of 65) of two-stage only procedures (P < .001). Relative risk of return to the operating room after a two-stage reimplantation with an antecedent DAIR compared with initial resection was 4.52 (95% confidence interval: 1.71, 11.9). MSIS host grading was similar between groups and did not influence the rate of failure in a regression model. The DAIR cohort was also found to consume more resources in terms of hospitalization length and operative procedures (P < .001).
CONCLUSIONS: Two-stage exchange procedures for prosthetic hip infections have a higher failure rate and consume more health-care resources when preceded by a failed DAIR. Surgeons and patients should be aware that a failed DAIR may compromise the results of future two-stage procedures.

Introduction

Infection after total hip arthroplasty (THA) is a devastating complication that occurs in 1%-2% of patients, accounting for approximately 15% of all revision THA procedures [1,2]. With the projected exponential increase in THAs performed in the next decade, there is an expectation that there will be a corresponding increase in the number of prosthetic hip infections [3]. This may have a significant economic burden on the health-care system [4]. In an era of cost containment, a thorough understanding of the success of various treatment alternatives is imperative.

Debridement, antibiotics, and implant retention (DAIR) is known to be associated with lower infection control rates, with the percentage of failures reported to be as high as 60% [[5], [6], [7]]. Nevertheless, owing to the low cost and low morbidity of this treatment alternative for acute infections, it remains a common practice. The consequences of a failed DAIR and its effects on subsequent two-stage results in patients with total knee replacements have been previously reported [8,9]. However, to date, there are no reports on how a failed DAIR influences subsequent two-stage results in infected THAs. Because DAIR for the treatment of prosthetic hip infection is fairly common and failure rates can be high, we hypothesized that if DAIR failed to control periprosthetic hip infection, patients may experience inferior results with subsequent two-stage procedures as was seen in prosthetic knee infections [9]. We ask the following questions: (1) What are the results of a two-stage reimplantation if preceded by a DAIR compared with two-stage reimplantation without an antecedent DAIR? and (2) Do patient factors influence the results?

Material and methods

We conducted a multicenter retrospective cohort study of prosthetic hip infections treated at 2 institutions between 2005 and 2015. Institutional review board approval was obtained at each institution. Cases were identified and selected by reviewing the electronic medical records. Prosthetic hip infections treated with two-stage exchange arthroplasty were identified. Patients treated with a two-stage exchange hip arthroplasty for infection were then divided into 2 distinct groups. The first group consisted of those patients who underwent an explant and spacer procedure as their initial operation for infection. The second group included those patients treated with DAIR, defined as debridement with or without modular component (the head and liner) exchange, before a two-stage exchange. Owing to the retrospective nature of this study, there was no standardization of the surgical indications or procedures that were performed.

Patients’ clinical courses were reviewed from the time of first positive hip culture to final hospitalization for infection and subsequent follow-up. Only those patients with at least 2 years of follow-up from their last surgery for infection were included. This allowed us to use a modified version of the Delphi criteria for infection control based on a healed wound and no return to the operating room (OR) for infection [10]. The analysis and reporting of this retrospective cohort study were conducted in accordance with Strengthening the Reporting of Observational Studies in Epidemiology statement [11].

Patient demographic data were collected. The types of organisms and chronicity of the infection were recorded. The authors were able to distinguish between acute postoperative (stage I) periprosthetic infections and hematogenous or late chronic (stage II/III) infections. The comorbidity profile of the patient was classified based on the McPherson Musculoskeletal Infection Society (MSIS) classification system [12]. Patients were categorized based on this host grading system as “uncompromised,” “compromised,” or “significantly compromised” [12]. Owing to the retrospective nature of this investigation and the variance of documentation and tests obtained, laboratory values and assessment of chronic active infections at a distant site were not included in the staging analysis. The McPherson MSIS classification includes a local extremity grade; however, this was excluded from the study as it could not be reliably ascertained. The type and number of procedures each patient underwent for infection were recorded, as well as the total number of days the patient was hospitalized for infection. Whether the infection was “controlled” was also recorded. “Control” was defined as having components in place without need for further surgery or intravenous antibiotics. This definition of control is in accordance with the Delphi-based international multidisciplinary consensus statement defining treatment success in prosthetic infections [10].

A total of 114 prosthetic hip infections were treated with a planned two-stage exchange arthroplasty with adequate follow-up to meet inclusion criteria. The indication for proceeding to a two-stage revision after recovery from the initial DAIR included a positive hip aspirate after DAIR, persistent drainage from the wound, or systemic signs of ongoing infection. The two-stage revision protocol was largely consistent at each center. A two-stage revision protocol consisted of removal of the femoral and acetabular prosthesis, thorough irrigation and debridement, and placement of an antibiotic-loaded polymethylmethacrylate spacer (stage I) in most instances (n = 61 of 65) followed by a minimum of 4 weeks of culture-specific parenteral intravenous antibiotics managed by infectious disease colleagues using data on culture sensitivities and patient allergies. Specifics on the antibiotic dosing of the spacer, spacer composition, static vs articulating, and additive lavages before spacer placement varied by the surgeon. Patients were then reimplanted after a combination of physical examination, serology trending, joint fluid aspiration, and intraoperative pathology findings demonstrating no evidence of persistent infection (stage II). There was no standardized approach to postoperative antibiotic suppression.

The primary outcome variable was success or failure of the two-stage revision procedure in each group. Failure was defined as the need for any additional surgery due to infection or failure to progress to reimplantation, whereas success was defined as no additional surgery required due to infection after reimplantation. Secondary outcome variables were the total number of procedures and days of hospitalization for infection.

JMP Pro, version 15.0.0, from Statistical Analysis Software (Cary, NC) was used to conduct statistical analyses. Descriptive statistics including means and standard deviations for continuous variables and proportions and percentages for categorical variables were calculated. The 2 groups (initial resection vs antecedent DAIR) were compared using Student's t-tests for continuous variables and Pearson's chi-squared analysis for categorical variables to assess equivalence between groups. Multivariable linear and logistic regression main-effect models were constructed to evaluate the impact of the initial resection vs antecedent DAIR groups on study outcomes of return to the OR, the total number of surgeries, and the total length of hospitalization with adjustment for baseline characteristics.

Results

As shown in Table 1, patients undergoing initial resection were significantly older and more likely to have a chronic infection. Otherwise, the distributions of the sex, race, and McPherson host grade were similar between cohorts.

Table 1

Baseline characteristics for patients with and without debridement and implant retention (DAIR) before resection.

Baseline characteristics	Initial resection (n = 65)	Antecedent DAIR (n = 49)	P-value
Age (years)	63.2 (13.2)	57.6 (13.8)	.029
Female sex	30/65 (46.2%)	26/49 (53.1%)	.47
Caucasian race	46/65 (70.8%)	36/49 (73.5%)	.75
McPherson A	12/65 (18.5%)	10/49 (20.4%)	.79
McPherson B	42/65 (64.6%)	25/49 (51%)	.144
McPherson C	11/65 (16.9%)	14/49 (28.6%)	.137
Chronic infection	65/65 (100%)	36/49 (73.5%)	<.001

Bold denotes statistical significance.

As shown in Table 2, of the 49 hips that underwent an antecedent DAIR before two-stage treatment, 21 (42.9%) failed a subsequent 2-stage revision. Therefore, only 28 of 49 (57.1%) were deemed successful using the modified Delphi criteria of a healed wound and no return to the OR. In contrast, of the 65 hips that underwent an initial resection without DAIR, 57 (87.7%) were successful using the modified Delphi criteria of a healed wound and no return to the OR. Only 8 of 65 (12.3%) in this group failed a subsequent two-stage revision surgery requiring a reoperation for persistent infection (Table 2). The initial resection group was notable for a much lower length of total hospitalization, in days, as well a fewer average number of procedures for infection compared with the initial DAIR group (Table 2). Patients who underwent initial resection were hospitalized for an average of 12.9 days, whereas those who underwent an antecedent DAIR were hospitalized for an average of 26.6 days (P < .001). Moreover, the antecedent DAIR group typically underwent 5 surgical procedures, whereas the initial resection group averaged 2.7 surgical procedures to control their hip infection (P < .001).

Table 2

Unadjusted differences in outcomes for patients with and without DAIR before resection.

Outcomes	Initial resection (n = 65)	Antecedent DAIR (n = 49)	P-value
Return to the OR after two-stage revision	8/65 (12.3%)	21/49 (42.9%)	<.001
Total number of surgeries	2.7 (2.1)	5 (2.3)	<.001
Total length of hospitalization	12.9 (10.4)	26.6 (15.5)	<.001

Bold denotes statistical significance.

Adjusted analyses demonstrated that antecedent DAIR was the main risk factor for return to the OR after two-stage revision, increased total number of surgeries, and increased length of hospitalization. McPherson A patients had a significantly shorter duration of hospitalization, whereas McPherson B patients had a significantly lower number of surgeries than the McPherson C patients.

Appendix Table 1 shows additional details regarding the patient’s initial operative treatment between cohorts. In patients both with and without exchange of the head and liner during DAIR, rates of return to the OR after subsequent 2-stage exchange were similar, at 40.9% (13/22) in the group with head and liner exchange vs 44.4% (15/27) in patients without head and liner exchange during DAIR. Outcomes between patients with and without an antibiotic spacer placement at the time of their initial resection also did not differ significantly (11.5 vs 25% rate of return to the OR), although the sample size for patients without an antibiotic spacer placement was quite small (n = 4). Chronic suppression had no significant impact on any of the study outcomes in regression analysis; however, the study was not powered to address this concern. Appendix Table 1 demonstrates the distribution of chronic antibiotic suppression and organism profile between cohorts.

Appendix Table 1

Rates of chronic suppression and organism characteristics between cohorts.

Organisms and treatments	Initial resection (n = 65)	Antecedent DAIR (n = 49)
Chronic suppression	9/65 (13.8%)	13/49 (26.5%)
Any gram positive	50/65 (76.9%)	36/49 (73.5%)
Any gram negative	3/65 (4.6%)	8/49 (16.3%)
Any MRSA	8/65 (12.3%)	9/49 (18.4%)
Any Proteus	0/65 (0%)	2/49 (4.1%)
Any Pseudomonas	0/65 (0%)	1/49 (2%)
Any fungi	2/65 (3.1%)	0/49 (0%)
Acinetobacter baumannii	0/65 (0%)	1/49 (2%)
Candida parapsilosis	1/65 (1.5%)	0/49 (0%)
Coagulase-negative Staphylococcus	14/65 (21.5%)	12/49 (24.5%)
Corynebacterium	0/65 (0%)	1/49 (2%)
Culture negative	11/65 (16.9%)	3/49 (6.1%)
Diphtheroids	0/65 (0%)	2/49 (4.1%)
Enterobacter aerogenes	1/65 (1.5%)	0/49 (0%)
Enterobacter species	0/65 (0%)	2/49 (4.1%)
Enterococcus	6/65 (9.2%)	1/49 (2%)
Enterococcus faecalis	3/65 (4.6%)	1/49 (2%)
Enterococcus species	2/65 (3.1%)	0/49 (0%)
Escherichia coli	0/65 (0%)	1/49 (2%)
Gemella haemolysans	1/65 (1.5%)	0/49 (0%)
Gram-positive cocci	1/65 (1.5%)	0/49 (0%)
Group B beta-hemolytic Streptococcus	4/65 (6.2%)	4/49 (8.2%)
Klebsiella species	0/65 (0%)	1/49 (2%)
Methicillin-resistant Staphylococcus aureus	8/65 (12.3%)	9/49 (18.4%)
Methicillin-sensitive Staphylococcus aureus	11/65 (16.9%)	9/49 (18.4%)
Micrococcus	1/65 (1.5%)	0/49 (0%)
Morganella morganii	0/65 (0%)	1/49 (2%)
Peptostreptococcus	0/65 (0%)	1/49 (2%)
Polymicrobial	0/65 (0%)	1/49 (2%)
Propionibacterium acnes	2/65 (3.1%)	0/49 (0%)
Proteus mirabilis	0/65 (0%)	1/49 (2%)
Proteus species	0/65 (0%)	1/49 (2%)
Pseudomonas aeruginosa	0/65 (0%)	1/49 (2%)
Rhinocladiella species	1/65 (1.5%)	0/49 (0%)
Staphylococcus epidermidis	1/65 (1.5%)	0/49 (0%)
Staphylococcus haemolyticus	1/65 (1.5%)	0/49 (0%)
Staphylococcus species	1/65 (1.5%)	0/49 (0%)
Streptococcus species	1/65 (1.5%)	0/49 (0%)
Streptococcus viridans	1/65 (1.5%)	0/49 (0%)
Unclear	1/65 (1.5%)	4/49 (8.2%)
Viridans streptococcus	3/65 (4.6%)	0/49 (0%)

Discussion

Irrigation and debridement and implant retention (DAIR) for the treatment of prosthetic hip infection is an attractive treatment option, given its technical ease, relative low morbidity, and relative low cost. However, the success rate of irrigation and debridement is limited with failures reported to be as high as 60% [5,7,8,13]. Therefore, it is important to understand the consequences of a failed DAIR for prosthetic hip infections.

This investigation showed that when comparing the outcome of two-stage procedures for periprosthetic hip infection, those patients who had an antecedent DAIR had inferior results than those patients initially resected. There is nothing inherently harmful about the irrigation and debridement procedure itself that should lead to inferior results after subsequent two-stage reimplantation. However, we suspect that the delay between the initial irrigation and debridement and the eventual explantation may have allowed biofilm colonies to mature and bacteria to become more entrenched in the periprosthetic bone. This may have led to more difficult eradication once the two-stage procedure was performed.

The primary limitations of this investigation are related to the retrospective and observational nature of this study. This study collated data across 2 institutions treated by a variety of surgeons with the implied nuanced variety of specific surgical techniques. There is certainly the risk of selection bias such that sicker or more compromised patients may have been preferentially initially treated with DAIR because of a desire to avoid the morbidity of a resection. However, our adjusted analyses including systemic host grading attempt to address this limitation. It is still possible that such selection bias may be responsible for the high failure rate of two-stage reimplantation after a failed DAIR noted in this series. The study design only evaluated patients who underwent a two-stage procedure, some who underwent a prior DAIR (n = 49) and some who did not (n = 65). Those patients who may have had successful results after a DAIR were not considered in this study design because they never needed a two-stage procedure. In fact, at one of our institutions, 84 of 111 (76%) patients who underwent DAIR with the same follow-up duration within the same time frame did not require resection afterward and therefore were considered to have treatment success. Thus, those patients who were good hosts, had a sensitive organism, and received an adequate DAIR in a timely fashion were excluded based on successful eradication of infection. This could possibly have left remaining only those who failed their DAIR as a group of poor hosts, with more virulent organisms, and who had a delayed or inadequate debridement as our study group. Because there is no way to take this into account with a comparative study of this nature, we felt it was important not only to describe this limitation but also to perform a multivariate regression analysis to account for baseline patient characteristics that could have influenced the rate of return to the OR for periprosthetic joint infection. Indeed, patients with better host grades (A and B) had significantly shorter hospitalization and a smaller number of surgeries. However, there was no significant effect of the host grade on return to the OR after two-stage revision although the odds of requiring return to the OR after two-stage revision were slightly lower in patients with better host grades.

Brimmo et al. reviewed 750 patients who underwent a two-stage knee procedure and concluded that failure of two-stage reimplantation was not increased by a prior failed DAIR. However, of the 750 patients who underwent a two-stage procedure, only 57 patients had undergone a prior DAIR. The authors also decided to exclude from the study anyone who had a DAIR greater than 2 years before their two-stage procedure. In addition, anyone who failed a two-stage procedure, greater than 4 years after reimplantation, was also excluded. The limited number of patients who underwent DAIR and the 2 exclusions raise some concerns about the study design [8].

In contrast, the first report on this subject was from one of our centers and described a 34% failure rate in 83 two-stage knee reimplantation procedures after a failed DAIR [9]. Subsequent to this, Cochran et al. identified 16,622 periprosthetic knee infections from a Medicare database. They identified 2806 patients treated with a DAIR without liner exchange before two-stage reimplantation and noted a 28% failure rate at 1 year, increasing to 43% at 6 years. In a group of 2150 patients treated with a DAIR with a liner exchange before two-stage knee reimplantation, they noted a 25.7% failure rate at 1 year, increasing to 40% at 6 years [14]. Two-stage reimplantation, without a prior DAIR had better results—a 19% recurrence rate at 1 year, increasing to 29% at 6 years [14]. Similarly, Rajgopal et al. reviewed 184 two-stage reimplantation procedures. Eighty-eight patients who had a DAIR before two-stage reimplantation had a 24% failure rate. The 96 patients who underwent a two-stage procedure alone failed 16% of the time. Using a Cox regression analysis to calculate the hazard ratio, they noted 2 times the failure rate of two-stage reimplantation after a failed DAIR compared with a two-stage procedure alone [15].

In these times of economic scrutiny, the cost of treatment is an important factor in medical decision-making. However, it is difficult to perform a cost or morbidity analysis for irrigation and debridement when the results in the literature are so varied ranging from 37% success to 83% [16,17]. We have shown that if an irrigation and debridement fails, the increased failure rate of a two-stage procedure with an antecedent DAIR is accompanied by more trips to the OR and longer hospitalizations. In fact, in the antecedent DAIR group, the average number of infection-related procedures (5) was nearly twice that of those initially resected (2.7). This by nature implies a significantly greater burden to the patient and cost to the health-care system. It is also prudent to comment on the potential for success with longer term antibiotic regimens as advocated by some or even suppression regimens in extreme cases [18]. This was not a focus of our investigation but may impact any chosen treatment pathway.

Irrigation and debridement certainly can be appropriate in acute postoperative or acute hematogenous infections. However, the literature is inconsistent in defining success rates. Bryan et al describe the use of DAIR with contemporary debridement and antibiotic regimens reporting an 83% success rate of DAIR [16]. However, Fehring et al reported a success rate of only 37% when performing a DAIR for acute perioperative infections [17]. Predicting results based on a specific organism type has also recently been challenged. Odum et al found no difference in the outcome of DAIR for Streptococcus vs other organisms, while Bradbury et al noted an 84% failure rate in those patients infected with Methicillin Resistant Staph Aureus [6,19].

Although our results show that a failed irrigation and debridement can lead to worse results after a two-stage reimplantation, we want to be careful not to recommend complete abandonment of irrigation and debridement procedures. We still perform DAIR procedures for acute perioperative infections or acute hematogenous infections with symptoms of less than 4 weeks. However, based on the results of this study, we are cognizant of the ramifications of debridement failure and thus the importance of early failure recognition. To that end, consideration should be given to strict surveillance after irrigation and debridement of acute perioperative or acute hematogenous infection.

One must avoid the “out of sight, out of mind” mentality, which is natural in dealing with this patient cohort. Routine clinical, radiographic, and serologic follow-up of DAIR procedures is essential. Deterioration in any of these parameters warrants careful consideration to proceed with a two-stage strategy before bacteria become more deeply entrenched and more difficult to eradicate. Procrastination in dealing with the inevitable may lead to the suboptimal results of two-stage reconstruction reported here.

In light of the data presented here, strategies to improve the results of DAIR such as serial debridement procedures as advocated by Chung et al. [20] who reported an 83% success rate in 83 patients at 3.5-year follow-up should be considered. In addition, reprepping and draping before insertion of new modular parts should also be considered, not only in DAIR procedures but also during the first stage of a two-stage procedure before placement of a spacer. Finally, if an early postoperative infection with cementless implants is encountered before bony ingrowth, consideration should be given to implant removal at that time to prevent later extraction-related bone loss.

Conclusions

Patients with an antecedent DAIR subsequently failed a two-stage procedure 42.9% of the time as opposed to only 12.3% of the time if the initial intervention was a resection. This discrepancy demands caution when weighing treatment options for patients with periprosthetic hip infections. Patients and surgeons must be prudent when weighing the decision to attempt an irrigation and debridement when treating a periprosthetic hip infection. Both the surgeon and patient should be aware that a failed DAIR may compromise the results of future two-stage procedures. Close clinical, serologic, and radiographic surveillance after DAIR may help identify those patients requiring future intervention with the hope of improving these results.

Conflict of interest

The authors declare there are no conflicts of interest.