Introduction

Osteomyelitis is characterized by an infectious and destructive inflammatory process affecting the bone that stems from microorganisms' invasion. The infection's etiology varies, originating either from local spread linked to trauma and surgery or from hematogenous dissemination, particularly in the elderly and children [1, 2]. The disease is often compounded by immune, vascular, and soft tissue problems [2]. The manifestations of chronic osteomyelitis are diverse, often remaining indolent for months before symptoms become apparent. The distinction between acute and chronic osteomyelitis, however, is contentious [3]. While some define chronicity based on histopathological examination and sequestrum formation, others consider it chronic when the infection persists for months to years, an arbitrary but commonly used timeframe [1,2,3,4]. Nevertheless, chronic osteomyelitis evolves over an extended period, potentially leading to sequestrum, bone destruction, marrow infection, soft tissue involvement, and fistulous tracts [4]. The severity can vary widely, from simple, manageable infections to severe cases with extensive bone destruction, significant functional deficit and even limb loss.

The management of osteomyelitis requires a multifaceted and aggressive approach to eradicate the infection and optimize outcomes [5,6,7]. Treatment modalities vary, and decision-making remains challenging as it encompasses various surgical techniques, antibiotic delivery methods, duration of antibiotic treatment, and surgical staging [7]. The problems the patient may encounter are multifaceted, and has been highlighted in the classification by Cierny and Mader et al., which categorizes osteomyelitis based on anatomical location, physiological status, and high-risk factors [8]. The complexity of treating osteomyelitis depends on its location, often involving long-term and debilitating treatment regimens. Success is typically indicated by a prolonged remission period, but conclusively declaring the disease cured is often problematic due to late recurrence.

Traditionally, treatment has relied on prolonged antibiotic use and multiple surgical debridements. In two-stage procedures, the primary focus is on eliminating the infection through bone and soft tissue resection, followed by stabilization of the bone, often externally, using fixators or frames. A second stage is planned approximately 4–8 weeks later, though this period can vary. This stage occurs after a course of antibiotics and once the infection has resolved both clinically and biochemically. The second stage concentrates on restoring function, utilizing techniques like fibular grafts, the Masquelet technique, autologous cancellous bone grafts, or bone transport [9,10,11]. Conversely, single-stage techniques aim to eradicate the infection with appropriate debridement and both local and systemic antibiotics, managing the bone defect in the same stage using techniques similar to those used in the second stage of two-stage management [12,13,14].

Current approaches emphasize a single thorough debridement, effective management of dead space, both local and systemic antibiotic administration, and a multidisciplinary strategy [15,16,17].

Comparisons between these techniques in the literature are scarce. This systematic review and meta-analysis aim to thoroughly review the literature on the treatment of chronic osteomyelitis in long bones and assess cure rates in single versus two-stage surgery for the condition.

Methods

The search and selection process followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines and was prospectively registered with PROSPERO (International Prospective Register of Systematic Reviews) (ID: CRD42021231237).

Search strategy

A systematic search of Medline, Embase, Web of Science, CINAHL (Cumulative Index to Nursing and Allied Health Literature), Healthcare Management Information Consortium (HMIC) and the Allied and Complementary Medicine (AMED) databases was performed using the following search strategy: (("osteomyelitis"[Title/Abstract] OR "bone infection"[Title/Abstract]) AND ("humerus"[Title/Abstract] OR "ulna"[Title/Abstract] OR "radius"[Title/Abstract] OR "fibula"[Title/Abstract] OR "single stage"[Title/Abstract] OR "one stage"[Title/Abstract] OR "two stage"[Title/Abstract] OR "two-stage"[Title/Abstract] OR "long bone"[Title/Abstract] OR "femur"[Title/Abstract] OR "tibia"[Title/Abstract])). Finally, reference lists of relevant articles were reviewed to identify additional articles that were potentially missed during the initial search.

Eligibility criteria

Studies that met the following criteria were included: (1) Management of chronic osteomyelitis in long bones using a planned one-stage or planned two-stage, 2) Patients aged 18 or older, 3) Follow up of at least 12 months, and 4) Clinical outcomes were reported.

Studies were excluded if they met any of the following criteria: (1) Review articles; (2) full text not available; (3) cadaveric studies; (4) Patients less than 18 years; (5) treatment of septic or infected non-unions, (6), non-bacterial osteomyelitis, (7) case series with fewer than 10 patients, (8) joint infections, (9) articles including non-long-bone osteomyelitis were excluded if they did not report outcome data separately for long bones. In addition, articles reporting outcomes on septic non-unions and osteomyelitis collectively were excluded.

Study screening

Titles and abstracts were independently screened for relevance by three authors using Covidence (AL, AE and MP) (Covidence systematic review software, Veritas Health Innovation, Melbourne, Australia. www.covidence.org). Potentially relevant articles underwent full-text screening, with any conflicts between the authors being resolved by discussion and consensus with the senior authors (HS).

Quality assessment and risk of bias

Study quality assessment was conducted using the methodological index for non-randomized studies (MINORS) tool. Methodological quality was categorized prior as follows: a score of 0–8 or 0–12 was considered poor quality, 9–12 or 13–18 was considered fair quality, and 13–16 or 19–24 was considered excellent quality, for non-comparative and comparative studies, respectively. For randomized controlled trials, the Cochrane risk of bias-2 (RoB-2) tool was used to assess study quality. Quality assessment measurements are denoted after each study in Table 1.

Table 1 Baseline demographics and patient characteristics
Full size table

Data extraction

Three authors independently extracted relevant data from the included studies to a previously piloted Microsoft Excel spreadsheet (Microsoft, Redmond, Washington, USA). These data included general article information, patient demographic and surgical procedure details, and relevant outcome measures.

Outcomes

Outcomes included the cure rate (%). A meta-analysis of proportions using the random-effects model was used to pool the cure rate (%) estimates from different studies. Without appropriate data transformation, the accompanying meta-analyses experience threats to statistical conclusion validity [18], such as confidence limits falling outside of the established zero-to-one range and variance instability [19]. While the logit transformation solves the problem of confidence interval estimates falling outside the zero to one range, it does not necessarily resolve the issues regarding variance from extreme proportional datasets. As the double arcsine transformation (Freeman-Tukey transformation) addresses both problems listed above, it is the preferred transformation method and was implemented in the current analysis. Once the meta-analysis had been performed on the transformed proportions, a back-transformation was performed. There is still no consensus about the back-transformation method that should be used with the Freeman-Tukey double arcsine method, although the harmonic mean was suggested for back-transformation [20]. Secondary outcomes included types of treatments used, complications, dead space management techniques, length of hospital stay, data on cost and the need for secondary interventions.

Meta-analysis

Statistical analysis was performed using R v 3.6.3 (R Core Team, Vienna, Austria). The random-effects model (using the maximum likelihood estimator for tau) was used to pool the effect sizes from the included studies. The underlying hypothesis for adopting the random-effects model is that heterogeneity or observed variance of effect is a sum of sampling error and variation in true-effect sizes stemming from inter-population variability. The generic inverse variance method was used to weigh each trial’s per-protocol population. Subgroup analysis was performed based on the stage. The overall proportion was calculated as well as the proportion within each subgroup. Forest plots were used to visualize the results. P values < 0.05 were considered statistically significant.

Prediction interval

The prediction interval was used to assess the treatment effect that may be predicted in future analyses, considering the different settings across different studies. It captures the variability in the true treatment effect across different settings. With substantial heterogeneity, prediction intervals will be broader than confidence intervals and might be considered a more conservative technique to integrate uncertainty in the analysis [21].

Sensitivity analysis

Sensitivity analysis was performed using the leave-one-out method to assess the effect of the different studies on the estimate and heterogeneity. Sensitivity analysis was performed to assess whether the pooled estimate and between-study heterogeneity were significantly affected by the exclusion of certain studies.

Publication bias and heterogeneity between studies

Funnel plots were used to assess publication bias. Egger’s test was used to test the asymmetry of funnel plots [22]. The trim-and-fill method was also used to detect and adjust for publication bias [23]. The I2 statistic was used to explore the percentage of heterogeneity attributed to variation in true-effect sizes secondary to inter-population variation. Estimates from subgroups within the same study were pooled using a fixed-effects model and used in the meta-analysis. The 95% confidence interval (CI) and Z-statistic were calculated and used for hypothesis testing. Heterogeneity between studies was quantified using the I2 statistic. In the case of high heterogeneity, the cause was investigated, the outlier was removed, and a new result was presented.

Results

After the removal of duplicates from the initial search, a total of 3398 references were retrieved for title and abstract screening (Fig. 1). A total of 3237 articles were excluded after the initial title/abstract screening. Next, 161 studies underwent full-text review. A total of 42 studies were included in the final analysis.

Fig. 1
figure 1

PRISMA flowchart illustrating inclusion of studies into the review

Full size image

Sample data

The pooled patient demographics are outlined in Table 1. Among the included studies, all but one were retrospective, encompassing both cohort and retrospective comparative studies. A total of 1605 patients were analyzed, predominantly male (77%), with an average age of 42.7 ± 8.5 years. The mean follow-up duration was 36 ± 18 months. The studies predominantly focused on one-stage management (28 studies), while twelve opted for planned two-stage management, and two studies offered comparisons between single and two-stage management.

Characteristics of osteomyelitis

The infection characteristics, host status, anatomical regions involved, and organisms are detailed in Table 2. The Cierny-Mader (CM) classification, reported in most studies, identified CM type III as the most common (60%) with host type B prevailing (51%). The etiology was primarily post-traumatic (64%, n = 812), followed by hematogenous origins (23%, n = 295). The tibia was the most affected site (57%, n = 819), with the femur (27%, n = 392) and humerus (6.8%, n = 98) following. Methicillin-susceptible Staphylococcus aureus (MSSA) was the predominant organism (28.5%, n = 357), with Methicillin-resistant Staphylococcus aureus (MRSA) found in 8.7% (109) of cases. Notably, 24.8% (n = 310) of cases showed no growth.

Table 2 Characteristics of the osteomyelitis according to classification, location and organisms
Full size table

Management strategies

The surgical treatment strategies are categorized in Table 3, including debridement, dead space management, soft tissue coverage, bone graft, and osseous stabilization. Dead space management techniques varied, with antibiotic-loaded calcium sulphate (CaSO4) beads (e.g., Stimulan, Osteoset T) used in 30.4% (n = 469) of cases. Polymethyl methacrylate (PMMA) cement was utilized in 15% (n = 236) of cases, employed as beads, spacers, and in Masquelet techniques. Other treatments included Cerament G (CaSO4 + hydroxyapatite), S53P4 bioactive glass, and others as described in Table 3. Flaps were required in 21.6% (n = 332) of cases, and bone grafts were used in 17% (n = 274), incorporating autologous, allograft, and reamer aspirate autograft.

Table 3 Management strategies and cure rates within the included studies
Full size table

Complications

Complications reported across studies exhibited considerable heterogeneity, detailed in Table 4. Recurrence of infection was treated as a failure, not a complication, and is thus analyzed separately under cure rates. The overall complication rates were similar for both single-stage and two-stage treatments (26.6% and 27.6%, respectively). The most frequent complication in single-stage procedures was prolonged wound drainage (13%), with stiffness and reduced range of motion also commonly reported.

Table 4 Aggregated complications in both groups
Full size table

Meta-analysis of cure rates

The analysis included 1636 patients. Single stage method was used in 1339 patients and the two-stage method was used in 297 patients. The pooled cure rate was 91% (CI 87%; 93%). Stratifying the analysis by stage did not reveal a statistically significant difference (X2 = 0.76, P > 0.05) with similar cure rate across stages (Fig. 2). The funnel plot was symmetric indicating the absence of publication bias. Egger’s test was not statistically significant (P = 0.64).

Fig. 2
figure 2

Meta-analysis of cure rates in single and two-stage groups

Full size image

Discussion

In our systematic review and meta-analysis on the treatment of osteomyelitis, we examined the evolving therapeutic strategies for this complex condition. Our findings reveal that, in terms of cure rates, or more appropriately termed, non-recurrence rates, there appears to be no significant difference when comparing single versus two-stage management of chronic osteomyelitis. This analysis is the first to collectively assess the success rates of single versus two stage management.

The decision between single-stage and multi-stage procedures is important, particularly considering the implications of lengthier hospital stays, increased costs, and operational complexities associated with two-stage management. Zhou et al. highlighted the notably higher costs and extended hospital stays associated with two-stage procedures compared to single-stage management [12]. Their findings indicate an average hospital stay of 28 days for the two-stage group, versus 18 days for those undergoing single-stage procedures. Similar trends are noted in studies by McNally, Ueng, and Qiu, reporting hospital stays of 27, 22, and 24 days respectively in two-stage treatment [10, 24, 25]. However, variability in hospital stay lengths is influenced by different institutional protocols and the possibility of outpatient management. Across studies, a comprehensive report on the costs and durations of hospital stays is generally deficient.

The surgical aspect of treatment is intricate, and our data indicates that debridement alone is associated with lower cure rates. Quantifying the extent of debridement in various studies presents another challenge, as the terminology used to describe it, such as "radical" or "adequate," is open to diverse interpretations. Consequently, the current data does not allow for distinct categorization of debridement methods.

Dead space management has become increasingly significant in recent years. Techniques such as antibiotic-coated beads and cement, muscle flaps, and bone grafts for addressing compromised soft tissue and bone loss have shown favorable outcomes based studies included in this review.. The induced membrane or Masquelet technique, though requiring a two-stage approach, has shown reliable results in our review [10, 11]. Additionally, bone defect management techniques, such as circular frames and bone segment transfers, offer stability, enabling early range of motion and weight-bearing. Jagadeesh et al.'s study reported a higher success rate with the use of calcium sulfate compared to debridement alone [13]. The current evidence suggests that the effectiveness of various local antibiotic delivery systems is comparable [14, 26].

Complication reporting varied across studies, with a notable incidence of prolonged wound drainage in single-stage procedures, often associated with calcium sulfate beads. While concerning, this drainage is not necessarily a harbinger of infection. Ferguson et al. reported high rates of wound leakage using calcium sulphate beads, but highlighted the low risk of infection associated with it [14]. Jagadeesh et al. also reported 18 out of 50 patients had ongoing serous discharge with the use of calcium sulphate that lasted up to 4 weeks resolving without treatment other than dressing changes [13]. This is in line with reports of 4–30% serous discharge while calcium sulphate is undergoing resorption. It is perhaps mitigated by adequate soft tissue coverage and judicious use of calcium sulphate [12, 14, 27, 28]

Commonly reported complications included wound issues, stiffness, and neuropathic symptoms, which could potentially be alleviated by early rehabilitation following extensive surgeries. Moreover, data on postoperative range of motion is scarce; improved reporting could reveal differences in single-stage groups, potentially allowing for earlier postoperative rehabilitation. More pin site issues were observed in the two-stage group, potentially due to longer durations of fixator use [10, 12, 29]. However, drawing definitive conclusions in this regard is difficult due to the more frequent use of fixators in two-stage management, as well as variations in the definition of pin-site infections. The occurrence of fractures in both treatment approaches necessitates cautious management, particularly regarding the introduction of implants. Systemic complications such as deep vein thrombosis, pulmonary embolism, and acute kidney injuries were also noted, albeit less frequently.

Perhaps it is key to highlight the literature's deficiencies, including the heterogeneity in antibiotic administration, inclusion criteria covering various bones and etiologies, and diverse causative organisms. These variations make it challenging to conclusively determine the superiority of specific treatments. In addition, the analysis represents single-arm comparisons, which are potential sources of bias. Although studies utilized similar techniques for both single and two-stage procedures, certain factors, such as the degree of osteomyelitis and the patient's physiological status, may indicate the use of one technique over the other. Another point of interest would be an analysis of the potential complications in both treatment groups. However, the variation in definitions of complications and the lack of clear reporting of complications arising due to disease and treatment did not allow for an accurate analysis in this regard.

Future research should focus on prospective studies, examining variables like causative organisms, patient demographics, Cierny-Mader classification, and specific treatment modalities. In addition, it will be helpful to know whether major differences exist between different preparations of antibiotic coated beads.

Conclusion

Chronic osteomyelitis is a complex condition with various treatments and interventions described. The data from our analysis suggests that single and two-stage treatment of chronic osteomyelitis yields comparably effective results. The current treatment strategies included a combination of debridement, dead space management, local and systematic antibiotics along with bone stabilization and soft tissue reconstruction if necessary. However, the indications for using either technique may play a role in predicting success rates. Higher-level studies should be conducted to provide more generalizable conclusions.