There is insufficient evidence to support a strong or moderate strength recommendation regarding risks of infection due to retained bone or metal fragments in pTBI.

Routine surgical removal of residual retained bone and metal fragments is not necessary following pTBI.

In the absence of direct scientific evidence, EXPERT CONSENSUS concluded that:

• A foreign body (FB) that is protruding from the skull, such as a knife or an arrow, that penetrates the brain should be removed in a controlled environment, such as the operating room, after appropriate studies have been performed to determine the position of the object, the degree of brain injury and the adjacent structures that need to be evaluated or monitored upon removal. (100% consensus)
• Removal of symptomatic fragments migrating within the skull should be considered. (100% consensus)
• If a foreign body is present at the site of an intracranial infection it should be removed, if possible, to reduce the risk of recurrent infection. (100% consensus)

No evidence or expert opinion supported distinct recommendations based on patient gender, age, wounding mechanism, or military vs. civilian context.

The risk posed by retained bone or metal fragments after a pTBI has guided the degree of initial surgical debridement and the perceived need for reoperation.1-30 Independent of the extent of debridement, here we seek to inform the relative risks and benefits of removing metallic and bony foreign bodies versus those of leaving them in situ. Even if an extensive debridement of a pTBI was performed, fragments may be left behind after surgery. Seizure and infection have been the most common outcome measures used to judge the risk of retained intracranial fragments.2-4,6,9,11,13,16,22,24,25,28 We are considering here, in part, whether re-operation to remove fragments left behind should be considered. Here we also consider the removal of protruding foreign bodies and how this is best conducted. The purpose of this clinical question is to determine benefits and harms of removing retained bone or metal fragments versus those of leaving them, independent of the extent of a debridement that is performed.

The prior guidelines did not specifically address the question of benefits and harms of retained fragments. The discussion of retained fragments and their potential complications was framed around the degree of debridement. The prior guidelines recommended debridement of penetrating cranial wounds and in the setting of mass effect, the removal of safely accessible fragments. The routine removal of fragments not easily accessible or re-operation with the sole purpose of removing retained fragments were not recommended. These recommendations were Options based on Class III evidence.31

This key question (KQ) considers the balance of benefits and harms when inadvertently retained, inaccessible or deep fragments of bone or metal from a penetrating brain injury are removed versus when they are not. The response to this question is related to decisions about: whether primary surgery or debridement should aggressively attempt to remove all fragments during initial treatment (addressed in KQ 37); and when only local debridement may be appropriate (addressed in KQ 35). In addition to the decisions made during initial treatment, this question is also relevant to whether later or additional surgery to remove fragments identified on imaging should be performed in the absence of conditions or symptoms such as infection.

Several studies reported whether patients had retained metal or bone fragments after initial treatment, but in many cases the presence of retained fragments was a descriptive variable and not all studies include data on patient outcomes related to retained fragments. When outcomes were included, the most frequently reported outcome was infection, but seizures, mortality, function, and toxicity were also reported in at least one identified study. In many cases, the impact of retained fragments was not the primary focus of the study. When the overall study sample size was small, the subgroup with retained fragments was often limited to very few or no patients with the reported outcomes. Given that we identified over 90 studies that mentioned retained fragments, we included only studies with an initial sample size of at least 45 people with penetrating brain injuries. At least 17 papers in the prior pTBI Guidelines discuss retained metal or bone fragments. Ten of these papers have been carried forwards to contribute to the evidence base for this KQ.3,4,6,7,11,14,23-25,28

Additionally, we reviewed and present the studies by the year the patients who were injured and initially treated, because the standard of practice shifted following the US experience in the Vietnam War, from removing all fragments to reduce the likelihood of infection to a less aggressive approach that suggested it was better to not remove fragments if they were deep or difficult to reach. Removal of deep fragments might risk damage to more brain tissue, which could outweigh the potential for infection and other negative outcomes.

In total we included 30 articles. These reported results from 25 studies and patient cohorts (three studies are based on the same registry from the Vietnam War,20,24,25 three reports are of different outcomes using data of overlapping cases from the Iran-Iraq War),2-4 and two studies included the same military personnel in Croatia.29,30 Several included studies were not designed to assess the impact of retained fragments and, while studies had to report a patient outcome to be included, authors may not have reported sufficient data to allow a comparison of outcomes for patients with penetrating brain injuries with versus without retained fragments. There is a subset of 11 studies that either reported a comparison (outcomes with retained versus no retained fragments) or reported data that allowed us to calculate a rate for the patients with and without retained fragments and test the difference in the rates.1-4,6,9,11,13,16,18,22,24,25,28

The results of these 11 studies are reported in Table 1 and are used as the basis for our strength of evidence (SOE) assessment. Following this, Table 2 is the evidence table for this question and contains more detail on all 25 studies.1-16,18-20,22-30,32

Treatment since 2000s: 5 studies for strength of evidence (6 total studies)9,10,13,16,18,22. The first five studies in Table 1 and one additional study in Table 2 (for a total of six studies) report results for patients treated since 2000.

Three studies reported that outcomes were not statistically significantly different for patients with and without retained fragments. The largest study by Harmon et al followed 763 civilian patients from 17 trauma centers in the United States who survived at least 72 hours and reported 7% of those with retained fragments and 5% without developed central nervous system infections (not statistically significant).13 A second study by Darwazeh et al of 520 Palestinian civilians reported a lower, but not statistically significantly different rate of local and systemic infections, for patients with retained fragments (6.8% vs. 9.4%).9 In a third study by Larkin et al of 80 U.S. military personnel injured in Afghanistan, the proportion of patients with and without retained fragments with a Glasgow Outcome Scale (GOS) score of 4 or higher was not significantly different at any of the three follow-up times (6 months, 1 year, and 2 years after injury).18

Two studies from this period reported worse outcomes for patients with retained fragments. A study by Pilipenko et al of military injuries in Ukraine that employed comparatively radical debridement as part of treatment for patients who were mostly injured by shrapnel (84.5%) found patients with retained bone fragments were significantly more likely to have an infection during their hospitalization and significantly more likely to have post-traumatic epilepsy when followed for a year after discharge.22 While the odds ratios are significant for both outcomes, the 95% CIs are wide, (e.g., 1.4 to 28.8) indicating that the estimates are not precise due to the relatively small sample size (N=81). A study by Jimenez et al of civilian patients in Colombia with gunshot wounds (GSWs) reported a high rate of intracranial infections overall (25%) and a significantly higher rate for patients with retained fragments (45.8% vs. 12.9%) compared with patients without retained fragments.16 The final study by Fathalla et al covering patients in this time period reported that of 102 civilian and military patients treated at various locations in Egypt, among 52 survivors with 1-year follow-up data, none with retained fragments suffered from lead toxicity.10 However, this article did not report the number of survivors with and without retained fragments.

Treatment in 1980s and 1990s: 5 studies for strength of evidence in 7 articles (11 total studies in 14 articles)1-6,8,11,15,21,26,29,30. All of the included studies from this time period are about military experience or experience in war zones and did not involve U.S. forces.

Five studies of treatment provided from 1980 to 1999 included comparisons of data from patients with and without retained fragments and are included in the SOE assessment. Four studies compared rates of infection,2,6,11,28 three rates of seizures,1,4,6 and one the proportion of patients with good outcomes.3 Three studies found relatively low rates of infection that were not significantly different for patients with and without retained fragments: 4% versus 7% in a study of 148 patients with bone or metal fragments treated in Turkey;11 0% versus 5.9% in a study of 43 patients treated with conservative debridement in Israel;6 and 7.2% versus 2.7% in a study of patients injured during the Iran-Iraq war and treated in Iran.2 One study by Taha et al conducted in Lebanon reported a much higher and statistically significantly different rate of infection for patients with retained bone fragments (20.6%) versus patients without retained bone fragments (1.8%); however further inquiry revealed that having a cerebral spinal fluid (CSF) fistula was an important driver of infection and only 4% of people with fragments and no CSF fistula developed infections, while 84.6% of patients with CSF fistula and bone fragments developed infections.28 The three studies that compared seizure rates all reported that fragments were not associated with higher rates: a study by Brandvold et al in Israel treating military patients evacuated from the military hostilities with Lebanon that also reported infections, evaluated seizures and found a lower, but not significantly different rate for patients who survived hospitalization who had bone or metal and bone fragments versus those without retained fragments (19.2% vs. 29.4%);6 univariable and multivariate analyses of data from the Iran-Iraq war cohort treated in Iraq by Elzayat et al concluded that fragments were not associated with seizures;1 and a study of cases treated in an Iraq hospital during the same conflict reported no difference in the rate of seizures in the first week after injury in patients with and without retained metal fragments.4 Another analysis by Fathalla et al of the Iran data evaluated patient outcomes, found that good outcomes were as likely in patients with retained bone and metal fragments as in patients without retained fragments (81.3% vs. 79.9%).10 The remaining six studies are included only in Table 2 as they did not provide data for patients with and without retained fragments. One study by Singh et al from the Indian military followed 60 patients, 40% with retained bone and 33% with retained metal on follow-up computed tomography (CT). In the 50 surviving patients none with or without fragments had any brain abscess and no other outcome reported by fragment status.26 A series by Cosar et al of 400 military patients treated in Turkey included 32.5% with bone and metal fragments remaining deep in the brain after surgery. The overall infection rate during long term follow-up (mean 2 years) was 5.5% and the authors stated that fragments did not seem to increase infection, but they did not provide the data for infections by fragment status.8 Another analysis by Erdogan et al of data from 374 military and civilian patients treated in Turkey specified that 31.8% (119 of 374) of patients had retained deep bone or metal fragments after surgery and reported 18 infections, including five in cases with fragments and without CSF fistulas, but it is not clear if the other 13 patients with infections had retained fragments or not.21 A study by Jankovic et al of 176 predominately military (86.4%) patients in Croatia did not provide the number of patients with retained fragments or the number with infections; but stated there was a statistically significant association between remaining bone fragments and infection, provided the p-value for a chi square, but added the caveat that all patients with infections and retained fragments also had at least two other risk factors for infection.15 Another study by Tudor et al in Croatia of 176 injured military personnel confirmed that 41% of patients had retained fragments based on CT after surgery and 28 patients (15.9%) developed infections.29,30 The authors stated that infections were more common in the presence of fragments, but data were not provided in a way that allowed comparisons of patients with and without fragments. One additional study by Amirjamshidi et al followed 99 military patients or civilians in battlefield areas in Iran who were assessed and treated with minimal debridement. All the patients had entrance wounds and no exit wounds; all had some retained bone and metal fragments. The majority (91%) achieved good outcomes in long term follow-up and none of these patients developed brain abscesses or epilepsy.5

Treatment before 1980: 1 study for strength of evidence in 2 articles (8 total studies in 10 articles)7,12,14,19,20,23-25,27,32

Data were compiled in a registry from soldiers injured during the Vietnam War (1967 to 1970) who were initially treated in Vietnam and Japan and whose care continued when they returned to the United States. These data were analyzed and reported in several articles, two of which provide evidence for this question. These data contained mixed results for different outcome-fragment combinations. Rish et al. reported that 37 of 1,121 patients developed brain abscesses and that the infection rate was significantly higher (6.4%) in the patients with retained bone fragments than in patients without (2.5%).24 The same article reported that the mortality rate for patients with brain abscesses was 82% in patients with retained fragments versus 42.3% in patients without fragments. The authors added that all patients with abscess and retained fragments who died had other risk factors as well, but they did not report this information for the people without retained fragments, nor did they report the association between mortality and retained fragments in the 1,084 patients who did not have a brain abscess.

Another analysis by Salazar et al of the same registry examined a subgroup of 421 patients who were the first to complete a comprehensive follow-up. The rate of seizures was significantly higher for patients with retained metal fragments (59% vs. 42%), but not significantly different for patients with and without retained bone fragments.25

The remaining eight articles are included in Table 2 as they did not provide comparisons or data that allowed comparisons of outcomes for patients with and without retained fragments.7,12,14,19,20,23,27,32

Five of these studies provided information about small numbers of civilians; three about patients treated in the United States, one in Canada, and one in Lebanon that included civilians and military personnel. In a study by Hubschmann et al that examined the records of 45 patients who had surgery for a GSW to the head at a trauma center in the United States, out of 82 treated patients, 29.9% (N=13) had retained missile/metal fragments but there was no evidence found of infection or other ill effects.14 A second study by Lillard et al of GSW to the head in civilians in the United States found that 68.7% of patients (57 of 83) had retained missile/metal fragments and authors stated there was no evidence of any complication or death related to fragments.25 The third study by Raimondi et al of 150 patients treated for GSW evaluated the 16.7% (25 of 150) who survived with retained missile fragments.19 None of these patients developed brain abscesses; of the 14 patients who had surgery, one had seizures and one developed meningitis. One study by Suddaby et al of 49 patients with GSWs treated in Canada reported that all 19 surviving patients had retained metal fragments after surgery and their complications included one case each of brain abscess, meningitis, and late epilepsy.27 A study by Haddad et al of 164 patients who had surgery, out of 219 civilians and military personnel treated in Lebanon, reported that 8.5% had retained bone fragments after surgery but that none of these patients developed brain abscesses during 15 months of follow-up.12

Additional results in Table 2 from this period include three articles based on military experiences. Meriowsky et al.20 uses the same patient registry from the Vietnam War that was used by Rish et al.24 and Salazar et al.25 This analysis reported a different outcome, that 10.2% of patients had reoperations to remove retained fragments and described the outcomes of those surgeries. Mortality was 10.3% for patients with retained fragments who had another surgery versus 6.2% for those who did not have another surgery.20 Carey et al. provided data on a different group of 103 patients from the Vietnam War and reported the results for 15 of the 16 patients with retained bone fragments who had one or more reoperations: one died (6.3%), but there was no brain abscess in this patient, and two abscesses in patients without retained bone fragments.7 The final study by Webster et al evaluated 206 patients treated by a U.S. Military Hospital Unity in Europe at the end of World War II (~1944). While the number of patients with retained fragments was not reported, 33 patients developed brain abscesses and of these patients 57.6% had retained bone fragments, 36.4% had retained metal fragments, and 6% had no retained fragments. The overall number of patients with retained fragments was not reported.32

Summary and Strength of Evidence

Table 1 summaries the comparisons from 11 studies reported in 14 articles including data from 3,799 patients.1-16,18-20,22-30,32 The studies are all non-randomized, consisting of retrospective cohorts or case series that included data on patient outcomes for patients with and without retained fragments.

These studies provide low strength of evidence that penetrating brain trauma patients with retained fragments have similar rates of infection and other outcomes compared with patients without retained fragments. All but one of these studies was rated high risk of bias and the study limitations for this body of evidence are considered high due to concerns about selection bias, loss to follow-up, and lack of accounting for potential cofounding in analyses. While the majority of the studies found that retained fragments did not increase poor outcomes, there are a minority of studies that report worse outcomes from different time periods, locations, and populations. These inconsistencies in the results create doubt and the need for more studies to either solidify the conclusion of similar outcomes or support another conclusion.

This comprehensive review of the literature showed inconsistent data elements and outcome measures related to the retention of a foreign body after a penetrating brain injury, with a majority of the literature suggesting no change in outcome or risk of infection or seizure whether the retained objects are removed or not. When evaluating this type of literature, it must be acknowledged that each penetrating head injury is unique and with many elements that cannot be controlled. For example, the type of penetrating fragment (bullet versus a fragment of various material from a blast), contamination of the wound at the time of injury, the velocity of the fragment and hence the dissipation of kinetic energy to the surrounding tissues and its effect on outcome and seizures. Variation in prophylactic antibiotic administration is another variable. Furthermore, the mixing of military and civilian data needs to be interpreted with caution as this can introduce a wide variance in the type of weaponry utilized which can directly influence outcome. Despite these limitations, this is still an important question to answer and continue to study.

Some studies within the reviewed literature suggest that there is a higher risk of infection associated with retained bone fragments and a higher rate of seizure associated with retained metallic fragments.15,16,22,24,25,28 However, the overall strength of evidence (SoE) does not warrant treating retained fragments differently based on their composition.

Along with KQ 37, this recommendation is part of a continuum of care for the patient with a penetrating brain injury. During the initial surgery, there is a Level III recommendation for "Conservative debridement of devitalized brain, bone fragments, and foreign bodies including lesions directly visible or encountered during the removal of operative hematomas or during hemostasis should be performed." (Reference KQ37) This approach at the initial surgery sets up the increased possibility of retained fragments post-operatively. The next logical step is to evaluate if these retained fragments should be removed to improve outcome, reduce the risk of infection, or reduce the risk of seizure. This is the impetus for the current question. Although the strength of evidence is low, at this time it does not appear that the retention of foreign bodies alone significantly affects outcome, infection, or seizure risk.

Given that the removal of intracranial fragments can involve iatrogenic trauma to the brain - neurosurgeons know that no study is required to demonstrate that. Surgical removal of foreign bodies would thus need to demonstrate a benefit to be justified. Such benefit was not noted in the literature we found.

There are other caveats here that are important to consider. When foreign bodies are present within a site of intracranial infection it may be difficult or impossible to treat that infection without removal of the foreign body. In this situation the foreign body must be removed if possible. A more extensive effort to remove retained foreign bodies is also appropriate in the context of a heavily contaminated wound.

The topic of retained foreign bodies after penetrating brain injury raised the issue of two closely related topics. The first was the consideration of managing a visibly protruding foreign body resulting in a penetrating brain injury.33 The second was for the management of a migrating intracranial fragment.34-36 A review of the literature on each topic found case reports and case series on these topics. However, the totality of the data did not rise to the level of evidence that would allow a recommendation to be made based on the literature.

The addition of a recommendation based on expert opinion regarding retained fragments that are protruding from the skull reflects a practical application in the management of penetrating brain injury. While commonly dealing with gunshot wounds or blast injury, lower velocity projectiles such as an arrow or a knife may result in visible retained foreign body. There was uniform consensus within the guidelines subject matter expert working group on this matter. When a foreign body that is protruding from skull, such as a knife or an arrow, that penetrates the brain should be removed in a controlled environment, such as the operating room or perhaps an endovascular suite, after appropriate studies have been performed to determine the anatomy and degree of brain injury as well as adjacent structures that need to be evaluated or monitored upon removal. Here potential vascular injury is particularly important to recognize and plan for.

The recommendation based on expert opinion to consider the removal of a migratory intracranial fragment also represents a practical, although less commonly encountered clinical scenario. There was uniform consensus within the guidelines subject matter expert working group that removal of symptomatic migratory fragments should be considered. Neuro-endoscopic approaches could be considered to remove intraventricular migratory fragments.34

In most circumstances limited debridement of pTBI and removal only of easily accessible foreign bodies is appropriate. Heavily contaminated wounds might prompt more aggressive removal. It is important to follow patients with retained fragments over time and to monitor for intracranial infection. Should intracranial infection occur then removal of the relevant foreign body is indicated provided that this is felt to be feasible and with acceptable risk. MRI safety is an important consideration in patients with intracranial metallic FBs.(See KQ7) An further, albeit uncommon indication for removal of a retained metallic fragment, is chronic lead toxicity.17

With respect to protruding or embedded foreign bodies such as knives our group identified a number of key principles for patient care. Protruding foreign bodies should be stabilized to ensure they cannot move and further injure the brain. Initially patients should be imaged with relevant modalities to enable an understanding of the intracranial position and trajectory of the object. It is also important to identify whether a large vessel injury is likely to exist within the cranium. Angiography should be considered when there is involvement of the skull base or the region of the pterion. Risk of vascular injury mandates exposure to facilitate proximal vascular control and an ability to surgically access the site of the injury. Cervical exposure of internal carotid artery may be necessary for proximal control when there is suspected intracranial carotid arterial injury. In many circumstances endovascular treatment can be considered. The ability to temporarily or permanently occlude a major vessel by endovascular means can be a critical intervention and means of proximal vascular control. Generally suitable controlled environments for foreign body removal include the operating room or the endovascular suite. In most cases wide surgical exposure is needed enabling access to known and unanticipated sites of injury. Extended skull base exposures may be required for FB penetration involving the orbits or skull base. Such foreign bodies should be removed by pulling them out opposite the trajectory they went in. Ideally these objects are not removed until needed exposure has been achieved and the plan for removal has been discussed with the operative team. Preparations may be needed for massive blood loss and perhaps for preservation of forensic data. A breadth of difficulties can be encountered during surgical removal of these objects and this can force the operative team to be thoughtful, adaptable and resourceful. The hospital must be aware that such cases can be demanding of human and non-human resources, and communication facilitating preparedness is key. We refer to the algorithms for vascular injury and skull base penetration accompanying these papers.

There are several areas within this question that future research can help strengthen the level of evidence behind these recommendations. Clearly documenting the type of fragment (bone vs metal) and the relationship to the outcome measure, whether it is infection, seizure, or general outcome, provides an important distinction between types of retained fragments and may influence surgical decision making or post-operative management. Further attempts to reduce bias by reporting on all patients with a penetrating brain injury in a prospective fashion would clarify the risk and outcome associated with these injuries. And finally, identification of risk factors for fragments that may migrate and become symptomatic may influence surgical decision making and follow-up care.

Prospective research on protruding or embedded objects is a challenge given the rarity and diversity of injuries encountered. Research comparing endovascular assistance with open vascular approaches would be beneficial. Here case reports sharing new and important lessons learned continue to be valuable.