Outcomes, Complications, and Eradication Rates of Two-Stage Revision Surgery for Periprosthetic Shoulder, Elbow, Hip, and Knee Infections: A Systematic Review

Abstract

Background/Objectives: Periprosthetic joint infection (PJI) is one of the most common complications after joint replacement. Two-stage revision remains the standard of care in chronic infections. The aim of this systematic review was to investigate the outcomes, complications, and eradication rates of two-stage revision surgery to treat PJI of the shoulder, elbow, hip, and knee. Methods: A total of 36 studies were included. Patient demographics, follow-up, the visual analog scale (VAS) for pain, the Constant–Murley score (CMS) for shoulder, the Harris Hip Score (HHS) for hip, the Knee Society Score (KSS) for knee, the range of motion (ROM), number and types of complications, and eradication rate were recorded. Results: A total of 2484 patients were identified, of whom 145, 29, 1269, and 1041 underwent two-stage revision surgery for shoulder, elbow, hip, and knee infections, respectively. The overall mean follow-up was 5.7 ± 4.5 years. The overall mean time of re-implantation was 20.8 ± 21.3 weeks. The most common causative bacteria were Cutibacterium acnes (32.7%) for shoulder, Methicillin-Sensitive Staphylococcus aureus (44.4%) for elbow, and Staphylococcus coagulase negative (CNS) (31.3% and 23%) for hip and knee infection, respectively. The mean overall preoperative VAS score was 6.7 ± 2.3, while, postoperatively, the mean score was 4.5 ± 2.7 (p < 0.001). A total of 2059 out of 2484 patients (82.9%) experienced eradication. Conclusions: Two-stage revision is an effective procedure to treat PJI with an overall eradication rate of 83%. A significant recovery of functionality and a decrease in residual pain can be achieved after surgery. Aseptic loosening and re-infection were the most common complications in shoulder and hip infections. Death rate was high in knee infections.

1. Introduction

Periprosthetic joint infection (PJI) is one of the most common complications after joint replacement, occurring in approximately 1–4% of primary total joint arthroplasties [1]. Previous studies showed that incidence ranges from 1% to 4% for periprosthetic shoulder infection (PSI) [2], from 1.5% to 12.5% for periprosthetic elbow infection (PEI) [3], from 1% to 2% for periprosthetic hip infection (PHI), and from 1% to 3% for periprosthetic knee infection (PKI) [4]. PJI is defined by the presence of a sinus tract, intra-articular pus, or two positive cultures with phenotypically identical organisms. Probable and possible infections are referenced based on a scoring rubric of minor criteria [2]. PJI does not only damage the patients’ physical and mental health but is also an economic burden for families and the healthcare system [5]. The managing of PJI is still a challenge for orthopedic surgeons; the options to treat PJI range from irrigation and debridement to amputation depending on multiple factors such as the acuity of infection, the joint involved, the virulence of the organism, and patient-related risks [6,7]. The three major options used to maintain joint functionality are irrigation and debridement, and one-stage and two-stage exchange [8]. A clear superiority of one procedure over the other in terms of infection control and functional outcome has not been identified to date, but two-stage revision remains the standard of care in chronic infections especially when the bacterium involved is unknown or resistant [9,10]. This procedure consists of a first stage involving the removal of implants and foreign material and their replacement with a temporary cement spacer loaded with antibiotics. After a period of intravenous antibiotic therapy and the evidence of clinical and laboratory infection resolution, there is a second stage which involves the implantation of a new prosthesis [11]. However, two-stage revision is associated with the morbidity of a second surgery with high perioperative risk and long hospitalization and healing times. Moreover, PJI is also challenging for surgeons skilled in prosthetic surgery. Different factors can influence the results, including comorbidities, previous surgery, bone stock, soft tissue damage, microbiological features, and antibiotic resistance. In order to maximize clinical outcomes and guide patient care with the best chance of eradication it is essential to identify specific risk factors and evaluate the results of the culture, the duration of the antibiotic treatment, and the proper timing of reimplantation [9,12,13].

The aim of this systematic review was to investigate the outcomes and complications, and the eradication rates of two-stage revision surgery to treat PJI of shoulder, elbow, hip, and knee.

2. Materials and Methods

2.1. Search Strategy

A systematic review of the published literature was conducted and reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [14]. The study protocol was registered in PROSPERO (CRD42024587098). The PubMed, MedLine, Scopus, and Cochrane Central databases were searched in March 2024. The search terms used to retrieve relevant articles were “periprosthetic” AND “joint”, AND “infection”, AND “two stage”, AND “revision”, AND “outcome”, OR “results”, OR “eradication”, OR “complications”. Two authors (EC and SC) independently screened the titles and abstracts to identify articles for inclusion, contacting a third senior author (MM) in cases of major discrepancies. We searched for additional articles by examining the reference lists of every article included and the gray literature available at our institution.

2.2. Inclusion Criteria and Study Selection

The articles were selected based on the following PICO model [15]: (1) Population: patients with PJI; (2) Intervention: patients who underwent two-stage revision surgery; (3) Comparator: all studies were included irrespective of the presence or absence of comparator or control groups; (4) Outcome: patients assessed for functional outcomes, complications, and eradication. Inclusion criteria were applied during the title, abstract, and full-text screenings; these were defined as follows: (1) observational studies including case-control, cohort studies, and randomized controlled trials (RCTs); (2) reporting outcomes or complications of two-stage revision surgery for PJI; (3) reporting of >10 surgically treated cases; (4) a minimum mean follow-up of >24 months; (5) articles written in English; and (6) articles published over the last 5 years. Other reviews, case reports, articles without outcomes or results, cadaveric or biomechanical studies, technical notes, editorials, letters to the editor, and expert opinions were excluded from the analysis but considered for the Section 4.

2.3. Data Extraction and Quality Assessment

Two authors with 4 and 2 years of experience, respectively, (EC and SC), performed comprehensive data extraction from the included articles. The first author, journal name, year of publication, patient demographics, and follow-up period were recorded for each article. Data extracted for quantitative analysis included different patient-reported outcome measures (PROMs), that is, the visual analog scale (VAS) for pain, the Constant–Murley score (CMS) for shoulder, The Harris Hip Score (HHS) for hip, the Knee Society Score (KSS) for knee, the range of motion (ROM), as well as the number and types of complications and eradication rate. A complication was defined as any intraoperative or postoperative event that was likely to have a negative influence on the patient’s final outcome, including fractures, infections, dislocations, nerve palsies, aseptic loosening, modular stem, or polyethylene disassociations.

A methodological quality assessment was conducted independently by three authors (MM, EC, and SC); cohort studies were assessed using the Modified Newcastle–Ottawa Quality Assessment Scale [16]. Randomized controlled trials were assessed with version 2 of the risk of bias tool (RoB2), as recommended by the Cochrane Collaboration. The discrepancies were resolved by consulting a senior reviewer with over 30 years of experience in joint surgery (GG). Details of the quality assessment are shown in

Table 1. Newcastle–Ottawa scale.

Study Author (Year)	Criteria								Total	Quality
	1	2	3	4	5	6	7	8
Stauffer T.P. et al. (2023) [17]	1	1	1	1	2	1	1	1	9	High
Klingebiel S. et al. (2021) [9]	1	1	1	1	2	1	1	1	9	High
Pellegrini A. et al. (2019) [18]	1	1	1	1	2	1	1	1	9	High
Hornung S. et al. (2023) [19]	1	0	1	1	2	1	1	1	8	High
Siegert P. et al. (2023) [20]	1	0	1	1	2	1	1	1	8	High
Saccomanno M.F. et al. (2024) [7]	1	1	1	1	2	1	1	1	9	High
Schwolow F. et al. (2022) [21]	1	0	1	1	2	1	1	1	8	High
Kerbel Y.E. et al. (2021) [22]	1	1	1	1	2	1	1	1	9	High
Białecki J. et al. (2020) [23]	1	0	1	1	2	1	1	1	8	High
Zhang W. et al. (2020) [24]	1	0	1	1	2	1	1	1	8	High
Theil C. et al. (2023) [25]	1	0	1	1	2	1	1	1	8	High
Hipfl C. et al. (2023) [26]	1	1	1	1	2	1	1	1	9	High
Cacciola G. et al. (2023) [27]	1	1	1	1	2	1	1	1	9	High
Veltman E.S. et al. (2019) [28]	1	1	1	1	2	1	1	1	9	High
Gramlich Y. et al. (2019) [29]	1	0	1	1	2	1	1	1	8	High
Santoso A. et al. (2020) [30]	1	1	1	1	2	1	0	1	8	High
Sigmund I.K. et al. (2019) [31]	1	1	1	1	2	1	1	1	9	High
Burastero G. et al. (2020) [32]	1	0	1	1	2	1	1	1	8	High
Bourgonjen Y.P. et al. (2021) [4]	1	1	1	1	2	1	1	1	9	High
Veltman E.S. et al. (2020) [33]	1	1	1	1	2	1	1	1	9	High
Yang F.S. et al. (2019) [34]	1	1	1	1	2	1	1	1	9	High
Hipfl C. et al. (2021) [26]	1	0	1	1	2	1	1	1	8	High
Joo M.-S. et al. (2021) [3]	1	1	1	1	2	1	1	1	9	High
White C.C. et al. (2021) [35]	1	0	1	1	2	1	1	1	8	High
Tuecking L.-R. et al. (2021) [36]	1	0	1	1	2	1	1	1	8	High
Kim Y.-H. et al. (2021) [37]	1	1	1	1	2	1	1	1	9	High
Zamora T. et al. (2020) [38]	1	1	1	1	2	1	1	1	9	High
Cavagnaro L. et al. (2022) [39]	1	0	1	1	2	1	1	1	8	High
Kim C.W. et al. (2021) [40]	1	1	1	1	2	1	1	1	9	High
Kildow B.J. et al. (2022) [41]	1	0	1	1	2	1	1	1	8	High
Fei Z. et al. (2022) [42]	1	0	1	1	2	1	1	1	8	High
Russo A. et al. (2022) [13]	1	1	1	1	2	1	1	1	9	High
Petis S.M. et al. (2019) [43]	1	1	1	1	2	1	1	1	9	High
Şenel A. et al. (2023) [44]	1	0	1	1	2	1	1	1	8	High
Budin M. et al. (2022) [12]	1	0	1	1	2	1	1	1	8	High
Rodriguez J. et al. (2023) [45]	1	1	1	1	2	1	0	1	8	High
Higgins E. et al. (2023) [46]	1	1	1	1	2	1	1	1	9	High
Gabrielli A.S. et al. (2022) [47]	1	0	1	1	2	1	1	1	8	High
Chung A.S. et al. (2019) [1]	1	1	1	1	2	1	1	1	9	High
Lu H. et al. (2024) [48]	1	1	1	1	2	1	1	1	9	High
Winkler T. et al. (2019) [49]	1	0	1	1	2	1	1	1	8	High
Corró S. et al. (2020) [50]	1	1	1	1	2	1	1	1	9	High
Van den Kieboom J. et al. (2022) [51]	1	1	1	1	2	1	1	1	8	High

Based on the total score, quality was classified as “low” (0–3), “moderate” (4–6), and “high” (7–9). Criterion number (in bold): 1, representativeness of the exposed cohort; 2, selection of the nonexposed cohort; 3, ascertainment of exposure; 4, demonstration that outcome of interest was not present at start of study; 5, comparability of cohorts on the basis of the design or analysis; 6, assessment of outcome; 7, was follow-up long enough for outcomes to occur?; 8, adequacy of follow up of cohorts. Each study was awarded a maximum of one or two points for each numbered item within the categories, based on the modified Newcastle-Ottawa scale rules.

. A substantial interobserver agreement (Cohen kappa coefficients ranging between 0.61 and 0.73) was reported.

2.4. Statistical Analysis

The quantitative data were organized for statistical analysis; all data were collected, measured, and reported with one-decimal accuracy. Weighted means and standard deviations (SD) were calculated for data concerning demographic characteristics and outcomes. When SDs were not directly provided, they were calculated with the equation [max range–min range/4] to allow for statistical aggregation. The weighted mean and SD comparisons between pre- and postoperative scores were performed using unpaired t-tests; the comparisons have been performed only when the data were reported in at least two studies. All tests were performed with IBM SPSS Statistics software (version 25.0, IBM Corp., Armonk, NY, USA) and GraphPad Prism (version 7.0, GraphPad Software Inc., San Diego, CA, USA). Confidence intervals (CIs) were set at 95%, and a p-value less than 0.05 was considered significant.

3. Results

A total of 1007 relevant articles were identified through the initial search, 710 abstracts were screened, and 253 full-text articles were assessed for eligibility based on our inclusion criteria, resulting in 36 studies that were eligible for the systematic review (Figure 1).

Articles originated from 11 countries, with Germany having the most contributions (n = 11), followed by Italy (n = 6), the Netherlands (n = 5), United States of America (n = 4), Korea (n = 3), China (n = 2), and Poland, Canada, Indonesia, Taiwan, and Turkey with 1 each. Characteristics of included patients are reported in

Table 2. Characteristics of included studies.

Authors	Journal	Year of Publication	Patient Demographics
			Number of Patients (N)	Sex (N)		Age (Years)			DM	BMI > 30 Kg/m2	HT	Smoke	FU (Years)			Time to Reimplantation (Weeks)
				M	F	Mean	SD	Range					Mean	SD	Range	Mean	SD	Range
SHOULDER
Stauffer et al. [17]	Seminars in Arthroplasty JSES	2023	32	20	22	67	NA	NA	NA	NA	NA	NA	1.3	1.3	0.3–5.3	14.3	28.2	1.7–114.4
Klingebiel et al. [9]	Journal of Clinical Medicine	2021	16	6	10	65	9	41–77	NA	NA	NA	3	2.6	1.5	1–7	18.7	NA	NA
Pellegrini et al. [18]	Orthopaedics and Traumatology: Surgery and Research	2019	30	18	12	68.8	4	64–80	NA	NA	NA	NA	8	2	2–10	14.4	2.3	9–19
Hornung et al. [19]	Arch Orthop Trauma Surg	2023	26	12	14	68.2	13.5	33–87	4	11	10	NA	1.1	0.6	0.1–2.5	NA	NA	NA
Siegert et al. [20]	Archives of Orthopaedic and Trauma Surgery	2023	25	19	6	67.8	11.5	42–88	NA	NA	NA	NA	2.5	2.2	NA	30.9	16	6–70.1
Saccomanno et al. [7]	Journal of Clinical Medicine	2024	16	13	3	69.1	5.4	NA	NA	3	1	NA	3	2.2	NA	20.7	5.4	NA
ELBOW
Joo et al. [3]	Journal of Shoulder and Elbow Surgery	2021	18	8	10	66.9	9.8	43–82	8	NA	NA	NA	2.8	0.7	2–5	NA	NA	NA
White et al. [35]	Seminars in Arthroplasty JSES	2021	11	7	4	49	13	25–77	2	NA	4	5	1.8	0.4	0.6–3.3	65	46.2	15–200
HIP
Schwolow et al. [21]	International Orthopaedics	2022	119	60	59	73	10.6	NA	28	NA	83	15	8	2.3	4–13.1	18.5	NA	12.4
Kerbel et al. [22]	Journal of Arthroplasty	2021	107	53	54	61.8	14	29–85	16	41	81	17	2.9	1.8	0.1–7.4	NA	NA	NA
Białecki et al. [23]	Orthopedic Reviews	2020	184	89	95	61	17.2	18–87	NA	NA	NA	NA	5.9	1.4	3.9–9.3	46	31.5	4–130
Zhang et al. [24]	Bone and Joint Research	2020	36	20	16	62.2	NA	NA	NA	NA	NA	NA	2.2	1	1.0–5.0	44.5	NA	NA
Theil et al. [25]	Journal of Arthroplasty	2023	32	14	18	67	19.5	15–93	NA	NA	NA	NA	3.8	3	0.08–12.4	NA	NA	NA
Hipfl et al. [26]	Archives of Orthopaedic and Trauma Surgery	2023	108	48	60	70.6	12.8	32.2–83.3	26	NA	NA	NA	5.4	1.1	3–7.5	9.2	4	2.7–23
Cacciola et al. [27]	Journal of Clinical Orthopaedics and Trauma	2023	48	18	30	72.7	11.5	NA	NA	NA	NA	NA	2.9	1	1–5	NA	NA	NA
Veltman et al. [28]	World Journal of Orthopedics	2019	29	14	15	66	NA	NA	5	8	NA	7	3.5	2.2	1–9.9	8	3.2	2–15
Gramlich et al. [29]	International Orthopaedics	2019	57	27	30	66.7	1.2	47–83	NA	NA	NA	NA	4.5	2.1	NA	21	20.8	4.7–88
Santoso et al. [30]	Malaysian Orthopaedic Journal	2020	57	32	25	67.38	12	36–84	9	NA	23	NA	2.6	1.1	1–5.6	NA	NA	NA
Sigmund et al. [31]	Journal of Clinical Orthopaedics and Trauma	2019	92	44	48	76	4	50–90	NA	NA	NA	NA	3.5	1.9	1.1–8.7	8.6	2	1–10
Burastero et al. [32]	PLoS ONE	2020	148	68	80	62.2	3.3	53.7–67	21	NA	13	24	4.6	1.9	2–9.7	12.7	4.8	NA
Bourgonjen et al. [4]	Journal of Bone and Joint Infection	2021	31	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA
Veltma et al. [33]	World Journal of Orthopedics	2020	55	30	25	68	8.7	33–88	8	13	NA	NA	4.2	2.4	1–10.7	NA	NA	NA
Yang et al. [34]	BMC Musculoskeletal Disorder	2019	31	19	12	56	5.2	50–71	9	NA	3	NA	2.4	0.26	1.9–2.9	14.5	5.2	11–32
Hipfl et al. [26]	Journal of Arthroplasty	2021	135	60	75	70.6	8.9	32–88	35	NA	NA	NA	5	1	3–7	8.9	4.3	NA
KNEE
Tuecking et al. [36]	Antibiotics	2021	48	28	20	69	10.5	51–93	NA	NA	NA	NA	3.4	1.5	1.5–7.6	NA	NA	NA
Kim et al. [37]	Arthroplasty	2021	63	25	38	67	9.5	40–78	NA	NA	NA	NA	15.1	2.2	10–19	NA	NA	NA
Zamora et al. [38]	Bone and Joint Journal	2020	47	23	24	69	8.8	NA	9	10	18	NA	3.7	1.5	2–6.5	NA	NA	NA
Cavagnaro et al. [39]	Archives of Orthopaedic and Trauma Surgery	2022	16	9	7	68.5	4.5	59.4–77.6	9	NA	NA	7	2.6	0.7	2–4.7	15.3	10.5	NA
Veltman et al. [28]	World Journal of Orthopedics	2019	15	4	11	64	NA	NA	3	9	NA	4	2.3	1	1–4.9	8	5.7	4–27
Kim et al. [40]	Knee Surgery and Related Research	2021	126	20	106	71.7	8	56–88	NA	NA	NA	NA	2.7	2.7	0.1–11	23.4	18.9	9–99
Kildow et al. [41]	Journal of Arthroplasty	2022	178	87	91	66.5	NA	NA	NA	NA	NA	NA	6.6	5.1	2–22.4	NA	NA	NA
Petis et al. [43]	International Journal of General Medicine	2022	46	19	28	67.5	9.5	37–81	NA	NA	NA	NA	4.1	2.5	1.2–11.2	16.2	9.1	NA
Fei et al. [42]	Antibiotics	2021	48	28	20	69.3	11.1	51–93	NA	NA	NA	NA	3.4	1.54	1.5–7.6	53.9	24.7	NA
Russo et al. [13]	The Journal of Bone and Joint Surgery	2019	245	122	123	68	14.5	33–91	NA	NA	NA	NA	14	5.7	2–25	NA	NA	NA
Şenel et al. [44]	Cureus	2023	44	12	32	70	11.2	46–91	14	NA	10	22	4	1.2	2.5–7.4	17.1	18.7	4–120
Budin et al. [12]	Antibiotics	2022	16	8	8	67.8	9.7	NA	NA	NA	NA	NA	4.9	0.6	NA	7	0.5	6–8
Bourgonjen et al. [4]	Journal of Bone and Joint Infection	2021	16	7	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA
Rodriguez et al. [45]	Arthroplast Today	2023	73	26	47	70	3	62–74	26	NA	NA	NA	7.9	0.8	7–10.3	18.5	1.6	13.5–20

FU, follow-up; SD, standard deviation; DM, Diabetes Mellitus; BMI, Body Mass Index; HT, hypertension; NA, not available.

.

A total of 2484 patients were identified, of whom 145, 29, 1269, and 1041 underwent two-stage revision surgery for shoulder, elbow, hip, and knee infections, respectively. Overall, male patients represented 46% of the total. The overall frequency-weighted mean age at the time of the operation was 67.7 ± 11.6 years. The overall prevalence of diabetes, bone mass index (BMI) superior to 30 kg/m², hypertension, and smoking habits were 9.2%, 4%, 9.8%, and 5%, respectively. The overall mean follow-up was 5.7 ± 4.5 years. The overall mean time of re-implantation was 20.8 ± 21.3 weeks.

The causative pathogens are reported in

Table 3. Causative organisms in periprosthetic joint infections undergoing two-stage revision surgery.

PHATOGENS	Shoulder		Elbow		Hip		Knee
	N.	%	N.	%	N.	%	N.	%
Candida albicans	2	1.9			8	0.60	9	1.1
Cutibacterium acnes	33	32.7	2	7.4	60	4.8	17	2
Enterococcus faecalis	2	1.9	3	11.1	47	3.8	35	4.6
Escherichia coli	1	1			18	1.4	12	1.5
Gram-negative					26	2.1	3	0.4
Methicillin-Resistant Staphylococcus aureus	13	12.9	7	25.9	70	5.6	70	8.7
Methicillin-Resistant Staphylococcus epidermidis	13	12.9			1	0.08	27	3.3
Methicillin-Sensitive Staphylococcus aureus	15	14.8	12	44.4	285	23	185	23
Others					31	2.5	44	5.5
Polymicrobial					132	10.6	59	7.3
Pseudomonas aeuruginosa	1	1			8	0.6	13	1.6
Sstaphylococcus agalactie					1	0.08	13	1.6
Sstaphylococcus anginosus							6	0.7
Staphylococcus coagulase negative	9	8.9			388	31.3	183	22.8
Staphylococcus epidermidis	10	9.9	3	11.1	142	11.5	69	8.6
Staphylococcus mutans	1	1
Staphylococcus pyogenes
Streptococcus	1	1			21	1.7	59	7.3
Total	101	100%	27	100%	1238	100%	804	100%

.

Data for perioperative cultured pathogens were reported for 2170 cultures. The most common causative bacteria were Cutibacterium acnes (32.7%), Methicillin-Sensitive Staphylococcus aureus (MSSA) (14.9%), and Methicillin-Resistant Staphylococcus aureus (MRSA) and Methicillin-Resistant Staphylococcus epidermidis MRSE (13%) for shoulder infections. MSSA (44.4%), MRSA (25.9%), and Enterococcus Faecalis and Staphylococcus epidermis (11.1%) were the most common for elbow infections. Staphylococcus coagulase negative (CNS) (31.3%) and Staphylococcus Epidermidis (11.5%) were the most common for hip infections. Finally, CNS (22.8%), MRSA (8.7%) and Staphylococcus Epidermidis (8.6%) were the most common for knee infections. A total of 191 (8.8%) patients had multiple causative pathogens.

The overall pre- and postoperative clinical outcomes are reported in Supplementary Material. The mean overall preoperative VAS score was reported in six studies [3,18,20,24,35,38], and it was 6.7 ± 2.3. Postoperatively, the mean overall VAS score was reported in eight studies [3,18,20,21,24,33,35,38] and it was 4.5 ± 2.7 (p < 0.001). A total of 2092 out of 2532 patients (82.6%) experienced eradication. Complication and eradication rates are reported in

Table 4. Complications in periprosthetic joint infections undergoing two-stage revision surgery.

SHOULDER
Aseptic Loosening		Overall Complications		Periprosthetic Fracture		Re-Infection		Death
N	%	N	%	N	%	N	%	N	%
12	8.3	4	2.7	6	4.1	12	8.3	NA	NA
ELBOW
Aseptic loosening		Overall complications		Periprosthetic fracture		Re-infection		Death
N	%	N	%	N	%	N	%	N	%
NA	NA	4	13.8	NA	NA	0	0	NA	NA
HIP
Aseptic loosening		Overall complications		Periprosthetic fracture		Re-infection		Death
N	%	N	%	N	%	N	%	N	%
71	5.6	68	5.4	37	2.9	86	6.8	16	1.3
KNEE
Aseptic loosening		Overall complications		Periprosthetic fracture		Re-infection		Death
N	%	N	%	N	%	N	%	N	%
60	0.6	27	2.5	18	1.6	79	7.3	67	6.1

and

Table 5. Eradication rate after two-stage revision surgery.

SHOULDER
Authors	Cases	Eradication	%
Stauffer et al. [17]	32	22	68.8%
Klingebiel et al. [9]	16	13	80.0%
Pellegrini et al. [18]	30	11	36.7%
Hornung et al. [19]	26	24	96.0%
Siegert et al. [20]	25	19	77.3%
Saccomanno et al. [7]	16	15	93.7%
Total	145	104	71.70%
ELBOW
Authors	Cases	Eradication	%
Joo et al. [3]	18	18	100.0%
White et al. [35]	11	11	100%
Total	29	29	100%
HIP
Authors	Cases	Eradication	%
Schwolow et al. [21]	119	119	100.0%
Kerbel et al. [22]	107	97	90.60%
Białecki et al. [23]	184	144	78.60%
Zhang et al. [24]	36	34	94.4%%
Theil et al. [25]	32	29	91%
Hipfl et al. [26]	108	61	56.5%
Cacciola et al. [27]	48	45	93.7%
Veltman et al. [28]	29	22	75.8%%
Gramlich et al. [29]	57	52	91.2%
Santoso et al. [30]	57	50	87.7%%
Sigmund et al. [31]	92	83	90.2%
Burastero et al. [32]	148	133	89.9%
Bourgonjen et al. [4]	31	22	70.9%
Veltman et al. [33]	55	51	93.0%
Yang et al. [34]	31	27	86.20%
Hipfl et al. [26]	135	119	88.10%
Total	1269	1088	85.73%
KNEE
Authors	Cases	Eradication	%
Tuecking et al. [36]	48	33	68.7%
Kim et al. [37]	63	49	78.0%
Zamora et al. [38]	47	40	85.1%
Cavagnaro et al. [39]	16	16	100.0%
Veltman et al. [28]	15	11	73%
Kim et al. [40]	126	76	60.3%
Kildow et al. [41]	178	152	85.4%
Petis et al. [43]	245	204	83.2%
Fei et al. [42]	46	45	97.8%
Russo et al. [13]	108	84	77.8%
Şenel et al. [44]	44	39	88.6%
Budin et al. [12]	16	9	56.0%
Bourgonjen et al. [4]	16	14	87.5%
Rodriguez et al. [45]	73	66	90.5%
Total	1041	838	80.5%

.

3.1. Shoulder

Male patients represented 60% of the included patients with a mean age of 68 ± 9.5 years. The mean follow-up was 3.2 ± 3 years and, the mean time of re-implantation was 19.3 ± 18.9 weeks.

The prevalence of diabetes, bone mass index (BMI) superior to 30 kg/m², hypertension, and smoking habits were 2.7%, 9.7%, 1%, and 2%, respectively.

The mean pre- and postoperative VAS score was reported in two studies [18,20], and it was 5.4 ± 2.6 and 1.7 ± 2, respectively (p < 0.001).

The mean pre- and postoperative CMS was measured in two studies [7,18], and it was 35.9 ± 18.8 and 55.6 ± 20.5, respectively (p < 0.001).

The mean preoperative flexion ROM was measured in two studies [17,18], with a mean score of 50.3 ± 26.7, while the postoperative flexion ROM was measured in three studies [17,18,19] with a mean score of 81.4 ± 39.3 (p < 0.001).

The mean preoperative abduction ROM was measured in two studies [17,18], with a mean score of 46.1 ± 26.1, while the postoperative abduction ROM was measured in three studies [17,18,19], with a mean score of 75.4 ± 37.7 (p < 0.001).

The overall complication rate was 2.7%. Aseptic loosening and re-infection were the most common ones (8.3% and 8.3%, respectively).

The eradication rate was 71.7%, ranging from 36.7% to 96%.

3.2. Elbow

Male patients represented 51.7% of the included patients with a mean age of 60.1 ± 14 years. The mean follow-up was 2.4 ± 0.8 years.

The prevalence of diabetes, hypertension, and smoking habits were 34.5%, 14%, and 17%, respectively.

The mean pre- and postoperative VAS score was reported in two studies [3,35], and it was 7.5 ± 1.3 and 3.1 ± 2.1, respectively (p < 0.001).

The mean pre- and postoperative flexion ROM was measured in two studies [3,35], with values of 69.8 ± 30.2 and 98.6 ± 47.1, respectively (p = 0.008). The mean pre- and postoperative extension ROM was measured in two studies [3,35], with values of 21.9 ± 7 and 9 ± 10.3, respectively (p < 0.001).

The overall complication rate was 13.8%.

The eradication rate was 100%.

3.3. Hip

Male patients represented 47% of the included patients, with a mean age of 67 ± 12.5 years. The mean follow-up was 4.7 ± 2.3 years, and the mean time of re-implantation was 19.3 ± 21.3 weeks.

The prevalence of diabetes, bone mass index (BMI) superior to 30 kg/m², hypertension, and smoking habits were 12%, 5%, 16%, and 5%, respectively.

The mean postoperative VAS score was reported in three studies [21,24,33], and it was 6 ± 1.8.

The mean preoperative HHS was measured in four studies [24,26,27,32], with a mean score of 37.6 ± 13.4, while the postoperative HHS was measured in six studies [23,24,26,27,32] with a mean score of 71.7 ± 15 (p < 0.001).

The overall complication rate was 5.4%. Aseptic loosening and re-infection were the most common ones (5.6% and 6.8%, respectively).

The eradication rate was 85.7%, ranging from 56.5% to 100%.

3.4. Knee

Male patients represented 46% of the included patients with a mean age of 68.7 ± 10.5 years. The mean follow-up was 7.4 ± 5.9 years and, the mean time of re-implantation was 22.8 ± 19.3 weeks.

The prevalence of diabetes, bone mass index (BMI) superior to 30 kg/m², hypertension, and smoking habits were 6%, 2%, 3%, and 5%, respectively.

The mean preoperative KSS was measured in five studies [13,37,38,44,45], with a mean score of 56 ± 16.1, while the postoperative KSS was measured in six studies [13,37,38,41,44,45] with a mean score of 67 ± 21.2 (p < 0.001).

The mean preoperative flexion ROM was measured in three studies [37,38,44], with a mean score of 82 ± 27.8, while the postoperative flexion ROM was measured in four studies [37,38,41,44], with a mean score of 88.2 ± 16.6 (p = 0.002).

The overall complication rate was 2.5%. Re-infection was the most common with a rate of 7.3%. Death rate was 6.1%.

The eradication rate was 80.5%, ranging from 56% to 100%.

4. Discussion

The most important finding was that two-stage revision is an effective procedure for treating PJI with an overall eradication rate of 83%. The most causative pathogens causing PJI were Cutibacterium acnes for PSI, Methicillin-Sensitive Staphylococcus aureus for PEI, and CNS for PHI and PKI. The mean time of re-implantation was 21 weeks. A significant recovery of functionality and a decrease in residual pain were found. Aseptic loosening and re-infection were the most common complications in PSI and PHI. The death rate was high in PKI.

This review updates the current evidence from previous systematic reviews by including the largest number of new trials published on this topic in the last five years, including a total of 37 trials that assessed a greater number of outcomes.

PJI remains one of the most challenging and devastating complications in arthroplasty surgery. Different treatment approaches have been used, including conservative and surgical treatments depending on the time of clinical presentation, the general health status of the patients, and the joint involved [52]. Which procedure is most effective at eradicating the infection while providing better functional outcomes and fewer complications is still under debate, and two-stage revision is the most widely used treatment for patients with chronic infection [26].

The success of revision surgery is dependent on the correlation between epidemiological data and comorbidities, and their relationship with PJI [53]. Different studies investigated social demographic factors, such as age over 65 years, BMI, and medical histories, finding that male patients, smokers, and those with diabetes and BMI over 30 kg/m² are associated with a higher risk of PJI [54] and persistent PJI [55,56]. In the current study, 46% of patients were male with a percentage exceeding 50% in shoulder and elbow groups; while the percentage of patients with a BMI exceeding 30 kg/m² and smokers were 4% and 5%, respectively. Watts et al. showed that patients with a BMI over 30 kg/m² had a 5.5-fold increased risk of failed two-stage revision compared with a normal weight cohort of patients [57]. We also reported a prevalence of diabetes of 9.2%. Interestingly, the prevalence of diabetes in the general population is 5.9%, and our data support its pathogenetic role [58]. Other patient-related risk factors are associated with PJI, such as alcohol abuse, drug, chronic renal failure, chronic hepatitis, and immunosuppression. Also, intraoperative factors can increase the risk of PJI, such as the duration of surgical intervention higher than 180 min, the formation of a hematoma from a surgical wound, the inadequacy of the operating room, sterilization, disinfection of the operators’ hands, and the operating field [59]. The use of antibiotic-loaded bone cement was an argument in favor of primary cemented TKA assuming a lower rate of infection, but its clinical benefit remains controversial. A recent meta-analysis that focused on the method of fixation of primary TKA on 5222 patients showed no difference between cemented and cementless TKA groups in terms of PJI rate (1.2% vs. 1.3%, respectively) [60]. In the postoperative period, factors such as respiratory, urinary, and dental infections can increase the risk of infection [59]. Personalization of treatment based on patient characteristics, bone quality, and general health status should be considered. This approach could help reduce the incidence of complications and improve long-term outcomes.

The most common bacteria responsible for PJIs are Staphylococcus aureus, Cutibacterium acnes, and CNS [61], as also reported by the reports from the European facilities list [18,23]. These results concur with ours. Indeed, in the current study, Cutibacterium acnes was the most frequent causative pathogen for PSI (33%); it has been reported that this pathogen is a shoulder skin commensal with a higher bacterial load in male patients, and it is rarely responsible for infections of other joints [62]. CNS was responsible for PEI, PHI, and PKI in 11%, 31%, and 24% of the cases, respectively. MSSA accounted for 44%, 23%, and 23% of the PEI, PHI, and PKI, respectively. A previous systematic review on PEI showed similar results with S. aureus and CNS responsible in 40% and 17% of the cases [53]. Moreover, Bialecki et al. [23] showed that MSSA was the most common causative pathogen for PHI and PKI. It should also be considered that, especially in chronic PJI, pathogens have a longer time to penetrate deep into the tissue and form mature biofilms. Recent research has revealed that S. aureus can penetrate deep into the bone using the osteocyte lacuno-canalicular network, which confirms the high percentage reported among the causative pathogens [63]

In the current study, a total of 197 (10%) patients had multiple infections. Other studies confirmed this result reporting the same percentage (10–11%) [2,23].

The time of reimplantation is still disputable. We reported a mean time ranging from 19 weeks, for PSI and PHI, to 23 weeks for PKI. A study by Vielgut et al. [11] reported a range from 1 week to 91 weeks for the two-stage procedure for PHI. Current evidence suggests that a shorter time to reimplantation might be associated with similar or even better infection control compared to long intervals [64]. Indeed, it has been also reported that antibiotic-loaded acrylic cement samples showed a burst release of antibiotics in the first hour, progressively decreasing over time [65].

Antibiotic therapy is still a challenge in two-stage treatment. There is no consensus on the exact length of antibiotic therapy; 4 to 6 weeks after resection is supported by studies and infectious disease societies. A limited duration of intravenous therapy may be indicated alone, in conjunction with oral antibiotics, or followed by oral antibiotics for continued therapy after discussion by a multidisciplinary team [66]. In this review the mean period was 2 weeks of intravenous therapy followed by 4 weeks of oral therapy, and up to 26 weeks or long-life therapy for persistent infection [28].

In the current study, a significant recovery of functionality and a decrease in residual pain were found. Alternatively, another review on PSI showed that the functional outcomes improved after two-stage revision, but the gain was not statistically significant [67]. These results are variable due to the possible concomitant comorbidities and large bone defects [18]. Similarly to the results of the current study, it has been reported that postoperative functionality improved after two-stage revision surgery for PEI, PHI, and PKI. Satisfactory patient intervention results were also found [53,68].

A total of 83% of the patients achieved eradication after two-stage revision with results ranging from 72% for PSI and 86% for PHI. The eradication rate was 100% for PEI but was calculated considering only the two studies. Overall, these results are encouraging and demonstrate that two-stage revision is a valuable solution for PJI. Indeed, other systematic reviews also reported high eradication rates of 90% [2], 82% [69], 89% [70], and 81% [6] for PSI, PEI, PHI, and PKI, respectively. However, we also reported a high death rate (6%) in PKI. It has been reported that the high mortality rate can be explained by poorer hosts, morbidity of infection, surgical management, and medical complications [56]. A review specifically conducted to assess the mortality after two-stage revision for PKI reported a one-year rate of 4% with an increase of 3% per year thereafter, and an overall rate of 14% [71]. An editorial specified the 10-year mortality risk of PJI after PHI, reporting a rate of 11.4% if the patients had a PJI in the first year [72].

The findings of this study should be interpreted while considering some limitations. First, we only considered English studies, which could potentially contribute to publication bias. Furthermore, although four major literature databases were used for the search, we cannot exclude the possibility of discovering additional articles using other databases. Second, according to the joint, there was heterogeneity in sample size and the number of included studies. Third, due to the lack of relative data or adequate statistical aggregation, it was not possible to compare outcomes based on the implant used. Fourth, the absence of relative data made it impossible to perform an analysis of antibiotic therapy time, and inaccurate statistical aggregation would have caused errors. Finally, we included studies with different evaluation times; it is likely that both clinical scores, complications, and eradication rates are affected by the length of patient follow-up, and these outcomes could also be potentially different if a specific and longer follow-up time was applied. The desire to optimize surgical treatment for PJI should not encourage clinicians to ignore that differences in patients’ characteristics in favor of a certain treatment option. Moreover, appropriate patient selection is critically important to maximize outcomes. The authors are responsible for discussing the results and how they can be interpreted in terms of previous studies and working hypotheses. It is important to discuss the findings and their implications in the broadest context possible. Future research directions may also be highlighted.

5. Conclusions

Two-stage revision is an effective procedure to treat PJI with an overall eradication rate of 83%. The most causative pathogens causing PJI were Cutibacterium acnes for PSI, Methicillin-Sensitive Staphylococcus aureus for PEI, and CNS for PHI and PKI. The mean time of re-implantation was 21 weeks. A significant recovery of functionality and a decrease in residual pain can be achieved after surgery. Aseptic loosening and re-infection were the most common complications in PSI and PHI. The death rate was high in PKI.