Surgical Site Infections after Open Reduction Internal Fixation for Trauma in Low and Middle Human Development Index Countries: A Systematic Review

Patients and Methods

We performed a systematic review of existing literature describing the epidemiology and management of SSIs after ORIF in LMHDI countries. This review was prospectively registered in the Prospero database (project number 42016036658) in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. PubMed, Ovid, and Web of Science databases were searched using the terms “internal fixation” or “open reduction” or “fracture fixation” or “open fractures” or “closed fractures” or “comminuted fractures” or “bone fractures” both alone and in combination by country listed by name (see Supplementary Table S1 online at http://www.liebertpub.com/sur). Countries included in the search were those recognized as LMHDI by the United Nations Development Programme (UNDP) [9]. Low and middle human development index country designation is a composite index focusing on three dimensions of human development: life expectancy at birth, mean years of and expected years of schooling, and gross national income per capita [9]. Inclusion criteria included any randomized, non-randomized, case control, cohort, or demographic studies published between January 2000 and March 2016 so as to obtain the most recent and relevant information. Translation of studies not written in English was performed using Google Translate.

After the initial search, titles and abstracts were screened for eligibility. Eligible articles were assessed independently by blinded reviewers who evaluated each study by pre-established criteria. These criteria included: study duration, location and setting, type of study, study size, age, gender, indication for ORIF, fracture type, implant type, frequency of SSI, grade of infection, need for hardware removal, microbiologic profile, associated morbidity, mortality, and cost of infection. Any disagreement between reviewers was resolved through discussion with a third reviewer. Exclusion criteria included case reports, abstracts without available full-text articles, articles describing fracture repair among only pediatric patients, those written by foreign military medical services, or those not describing patients who experienced their fracture as a result of trauma. A pediatric patient was defined as a patient 18 years of age or younger. Missing data were requested from study authors and incorporated when possible. Additional studies were sought by examining the bibliographies of all studies identified during the search process.

We used established superficial, deep and organ/space SSI definitions as the gold standard definition of infection [10,11]. Deep and organ/space infections, including associated osteomyelitis and joint infections, were combined for analysis. For the purpose of this study, procedures such as intra-medullary nailing of long bone fractures and cancellous screw fixation femoral neck fractures were classified as ORIF although the reduction itself may not have been visualized directly through an open incision [12]. An intra-medullary device was defined as a nail, rod, or implantable Kirschner wire placed in the intra-medullary space for fixation. For the qualitative synthesis, we set no minimum number of patients per study, because we did not expect to identify many articles. For the quantitative pooled infection rate analysis we excluded studies with fewer than 30 patients in concordance with the rule of three sample size and an assumed infection rate of 10% in these settings [13,14]. For studies in which sub-populations were described, sub-groups were divided and considered independently. Pooled infection rates were sample-weighted. A priori a study was required to report both the odds ratio and the 95% confidence interval to be included in the risk factor analysis; χ² analysis was used where appropriate. Statistical analysis was performed using STATA^® (version 14.1, StataCorp, College Station, TX). This was an Institutional Review Board-exempt study because all articles were available publically.

Results

The initial database search identified 657 abstracts (Fig. 1). Five hundred forty-three abstracts remained after 116 duplicates were removed. Four hundred six were excluded after abstract review because of non-relevance. Two additional articles were identified through other sources. Complete articles were obtained for the remaining 139 articles (34% of total). Of these, 97 were excluded for the following reasons: insufficient data on SSIs (n = 53), no ORIFs in the sample (n = 24), pediatric patients (n = 8), not in LMHDICs (n = 4), or other (n = 8). Forty-two studies met inclusion and exclusion criteria and were included in the qualitative synthesis (Table 1) [15–56].

OPEN IN VIEWER

FIG. 1.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flowsheet of manuscript selection process

Of the studies included in the qualitative analysis, 16 (38%) were prospective and 26 (62%) were retrospective. Thirty-five (83%) were observational studies, 4 (10%) were randomized interventional, 2 (5%) were case control, and 1 (2%) was a case series. Twenty-three studies (55%) were conducted in the Southeast Asian region, 15 studies (36%) were conducted in African countries, 3 (7%) were in Eastern Mediterranean countries, and 1 in Western Pacific countries (2%). The most commonly represented African countries were Nigeria (n = 6; 40%) and Malawi (n = 2; 13%), whereas India accounted for the majority of the Southeast Asian region (n = 18; 78%). Thirty-six (86%) studies reported data from a single institution.

Twelve (29%) studies described patients who underwent ORIF exclusively with plates or screws, 15 (36%) exclusively by intra-medullary device, and 15 (36%) studies described a patient population undergoing either of these, or other fixation techniques. Eleven (26%) studies examined only patients with femur fractures; 5 (12%) studies only humerus fractures; 5 (12%) studies only tibia fractures; 4 (10%) studies only forearm fractures; and 16 (38%) fractures in multiple locations. Thirty (71%) studies described patients who had closed fractures; 5 (12%) studies with exclusively open fractures; 5 (12%) studies with both closed and open fractures (1 study did not specify).

Thirty-two studies met criteria for quantitative analysis comprising 3,093 operations. There were 197 SSIs reported, which correlate to a weighted pooled infection rate of 6.4 infections per 100 operations (95% confidence interval [CI] 4.6–8.2 infections) (Table 2). Among the 24 studies with gender information available, 1,684 (74%) operations were performed on males. Twenty-six studies delineated between superficial and deep or organ/space infections; among the 1,983 operations in these studies, superficial infections accounted for 56% of all infections. In the 21 studies that provided data on need for hardware removal, hardware removal was required after 26 operations (1.3 removals per 100 index operations [95% CI 0.4–2.2 hardware removal per 100 operations]). Seven studies specifically commented on presence and causality of peri-operative mortality; there were 2 fatalities after 1,404 operations in this group. Twenty-three of the studies identified populations with either open and closed fractures suitable for sub-group analysis. Among the 1,774 ORIFs performed for closed fractures, an infection rate of 5.9 infections per 100 operations (95% CI 4.0–-7.7 infections per 100 operations) was observed, in contrast to the infection rate of 18.4 infections per 100 operations (95% CI 13.9–23.0 infections per 100 operations) seen after ORIF of 201 open fractures (p < 0.001). Twenty-six studies reported the specific type of fixation in which the infections occurred; among the 1,027 operations were intra-medullary devices were used for fixation, there was a weighted, pooled rate of 7.4 infections per 100 intra-medullary ORIF operations (95% CI 3.6–11.1 infections per 100 operations). In contrast, among the 890 operations in which plate, screw, or other devices were utilized for internal fixation, a weighted, pooled infection rate of 4.8 infections per 100 operations (95% CI 2.6–7.0 infections per 100 operations) was observed (p = 0.02).

The majority of studies (n = 21; 66%) specified pre-operative antibiotic agent use; in this sub-group there was a weighted pooled infection rate of 6.8 infections per 100 operations (95% CI 4.6–8.9 infections per 100 operations). In the 11 studies that did not specify whether the patient received pre-operative antibiotic agents, there was a weighted, pooled infection rate of 4.8 infections per 100 operations (95% CI 1.5–8.2 infections per 100 operations; p = 0.04). Twenty-two studies examined site-specific infection rates on either the upper or lower extremities; analysis of 402 upper extremity cases demonstrated a weighted, pooled infection rate of 5.4 infections per 100 operations (95% CI 0.1–10.9 infections per 100 operations). In the sub-group of 1,009 lower extremity-specific cases, there was a reported weighted, pooled infection rate of 5.9 infections per 100 operations (95% CI 1.9–10.0 infections per 100 operations; p = 0.8).

Discussion

The morbidity associated with musculoskeletal trauma underscores the need for safe and effective orthopedic care in LMHDICs. Surgical site infections may be useful quality indicators for orthopedic surgical services, although orthopedic trauma patients may be at a higher risk for complications, including SSIs, than general orthopedic patients [57]. Surgical site infection rates may be influenced by a wide variety of parameters, from patient delays and mechanism of injury to the healthcare delivery system and the human and material resources available in the operating room [58]. Surgical site infections after ORIF are of particular consequence because infection may result in prolonged courses of oral or intravenous antibiotic agents, wound irrigation and debridement, failure of fixation, malunion or non-union, and potential need for hardware removal [59].

Our comprehensive review identified an SSI rate of 4.6–8.2 infections per 100 ORIF procedures, slightly greater than studies from HHDICs. Prior SSI surveillance from Canada (an HHDIC) reported an infection rate of approximately 6% after various procedures [60], and similar epidemiologic data from the United States Centers for Disease Control and Prevention showed rates of 1%–5% for ORIF [61]. Another U.S.-based study of skeletal trauma revealed an infection rate of 4.2%, with higher rates of infection associated with wound drains, tibia and elbow injuries, and a prior diagnosis of diabetes [58].

In LMHDICs, rates of infection after closed fractures were nearly three times that of comparable closed repair in HHDICs; the same was true with open fractures. Increased SSI rates after open fracture are well-documented in HHDICs. The Study to Prospectively Evaluate Reamed Intramedullary Nails in Patients with Tibial Fractures (SPRINT) trial of tibia fractures in the United States indicated a rate of approximately 2% for closed fractures and 8% for open fractures [62]. Other studies in HHDICs demonstrate infection rates that vary greatly with the extent of soft tissue compromise, ranging from 0%–6% in type I open fractures to 5%–50% in type III [63–65]. Yet in LMHDICs, rates of infection after open fracture are nearly three times those in HHDICs. This finding suggests that patient or healthcare characteristics other than fracture pattern may contribute to the higher observed frequency of SSI. Interventions targeting identification of fractures at risk of infection, more prompt presentation to definitive surgical care, prompt administration of antibiotic agents after open fractures, appropriate timing of pre-operative antibiotic therapy, standardized surgical techniques, and assurance of sterility and infection prevention and control standards remain critical steps in management of these injuries in LMHDICs.

Long bone and fracture pattern heterogeneity among the evaluated studies complicates conclusions about how type of fixation impacts SSI in LMHDICs. Recent studies from HHDICs demonstrate rates among plate and screw constructs may approach 8% in tibial plateau fractures [66], whereas plate and screw constructs in the radius and ulna, even in open fractures, may only approach 5% [67]. This degree of specificity in the reviewed articles was unfortunately reported rarely. Similarly, outcomes after intra-medullary fixation in LMHDICs were variable and likely technique-dependent; lack of fluoroscopy often necessitates multiple incisions to achieve direct anatomic reduction and visualization, which increases tissue damage, operating time, and the chance for a potential wound complication [68]. Implant specific registries, such as data collected by the Surgical Implant Generation Network (SIGN) International, offer broad opportunities to monitor infection and research patient-reported outcomes in these populations [69]. These results support the continued responsible involvement of implant manufacturers in high-quality, rigorous monitoring of implants in LMHDICs.

Unexpectedly, studies not specifying pre-operative antibiotic use reported lower rates of infection than studies that specified pre-operative antibiotic use. This paradoxical finding may be a result of studies not reporting antibiotic use despite actual administration, or antibiotic agents being given only to patients with open fractures. The importance of appropriately timed pre-operative antibiotic therapy is well documented, whereas post-operative duration of antibiotic therapy after hardware implantation remains more nebulous. A recent multi-center study from India suggests that longer duration of antibiotic therapy may be beneficial in LMHDIC settings, given that delayed presentation after open fracture may prevent prompt antibiotic administration [70].

There are multiple limitations to this study. First, as reported for the other articles in our series [71–73], there is inherent selection bias because only 13 LMHDICs were captured. This may limit generalizability to other countries. Second, inclusion of all long bone fractures and all fracture types (both intra- and extra-articular) introduces considerable heterogeneity into the overall rate of SSI. Future prospective studies should evaluate infection with respect to extent of soft-tissue injury using the Gustilo-Andersen classification or other metrics, as this may influence the rate of infection and further characterize research cohorts to aid in risk stratification [63]. Third, publication bias is likely present as surgeons may choose not to publish series or studies with higher rates of SSI. Fourth, studies did not commonly delineate between superficial and deep infection and it is possible that superficial infections that did not require intervention were not captured, lowering the overall reported rate of SSI. Fifth, long-term follow-up after implantation was infrequent, which would lead to under-estimation of long-term complications such as chronic osteomyelitis. Finally, pooled odds ratios were not able to be calculated, because studies with odds ratios did not report enough data on their respective patient populations to enable pooling.

Conclusion

Orthopedic trauma represents a significant burden of surgical disease in LMHDICs, and unaddressed fractures may result in considerable morbidity and mortality. Yet when increasing availability of operative intervention, SSI represents a common complication of ORIF, which appears to occur at higher rates in LMHDIC settings. Open fractures have the greatest SSI rate, exceeding expected infection rates from HHDIC settings. Rigorous post-operative outcomes monitoring is essential to ensuring quality orthopedic care. Standardized, usable definitions for fractures and fracture repair in LMHDIC would greatly improve comparison of SSI rates among LMHDIC settings.