Introduction

Pelvic ring injuries are potentially life threatening and necessitate multidisciplinary care [1]. Surgical management of these injuries has revolved around the use of open reduction and internal fixation with plates, iliosacral screws and external fixation [2]. Simple injury patterns like diastasis can be managed well with plates, but complex anterior ring fractures were traditionally managed with external fixation as these can be quite challenging. However, external fixators have been associated with complications like pin site infection, limited access to the abdomen, decreased patient mobility, loosening and nerve damage, and are therefore not tolerated well by patients [3].

To overcome these limitations, the anterior subcutaneous internal fixator (ASIF) was developed from spine surgery instrumentation. This modality, termed as the INFIX or the pelvic bridge technique, utilizes pedicle screws that are placed into the ilium on either side, and connected to each other with a subcutaneous rod, with or without fixation into the symphyseal region [4]. The proposed potential benefits include fewer soft tissue infections, better pain control and patient mobilization and hence faster rehabilitation. Moreover, this technique is minimally invasive and therefore thought to be less time-consuming and associated with less blood loss.

This technique has been used for more than a decade, and several studies have reported clinical and functional outcomes with this technique. However, there is a need to critically analyze the available evidence in a comprehensive fashion so as to determine its efficacy and safety in routine surgical practice, as well as to refine its indications. Therefore, the aim of this study was to conduct this systematic review and meta-analysis to determine the outcomes and complications of the INFIX technique for unstable pelvic ring injuries following high-velocity trauma.

Methods

Objectives

  1. 1.

    To determine the clinical and radiological outcomes of the INFIX technique for unstable pelvic ring injuries.

  2. 2.

    To determine the complication rates of the INFIX technique for unstable pelvic ring injuries.

Study design

A systematic review and meta-analysis of literature were planned and in accordance with the PRISMA guidelines [5].

Inclusion and exclusion criteria

Prospective and retrospective studies in all languages, whether comparative or non-comparative, pertaining to the use of INFIX in pelvic fractures, with a minimum of 10 cases were included. Studies which did not evaluate INFIX, case reports, conference abstracts and those with less than 10 cases were excluded. Cadaveric studies and technique papers that did not describe the prespecified outcome measures were also excluded. Since this study involved unstable pelvic ring injuries with high-velocity trauma, INFIX used for fragility fractures of the pelvis were not included.

Search methodology

Using a well-defined search strategy, a primary search was conducted on PubMed, Scopus and Embase on 18 May 2020 (Table 1). A total number of 384 results were obtained. The secondary search was conducted by screening the references from included studies and other relevant publications on the subject.

Table 1 The detailed search methodology
Full size table

Study selection

The selection of studies was done independently by two reviewers (SP and KJ). Titles and abstracts of the search results were screened initially. This was followed by retrieval of the full texts of relevant articles. Articles were included if they fulfilled the pre-specified inclusion criteria. For studies reported by the same author, all studies were included in the systematic review, but the study with maximum number of patients was included for analysis. Discrepancies were resolved by mutual agreement.

Data collection, data items and outcome measures

Data collection was performed by KJ and SS on a pre-piloted data collection sheet; all entries were cross-checked by SP to ensure accuracy. The following baseline data items were extracted: Journal and year of publication, language, gender, age, injury severity score (ISS and fracture classification).

Outcome measures that were extracted included perioperative blood loss and time, functional and radiological outcomes, Majeed scoring and complications.

Risk-of-bias assessment

The Methodological Index for Non-Randomised Studies (MINORS) tool [6] was used to assess the risk-of-bias for all studies by 2 independent observers (SS and VK) (Table 2). This tool is a 12-item questionnaire that can be used for both comparative and non-comparative studies. Grading is done from a score of 16 for non-comparative and 24 for comparative studies.

Table 2 MINORS tool for risk of bias assessment
Full size table

Data synthesis and analysis

Qualitative and quantitative analyses were performed after data extraction. For qualitative analysis, the data were organized in appropriate tables. Quantitative data synthesis consisted of two different arms: a comparative meta-analysis arm and a non-comparative meta-analysis arm.

The comparative meta-analysis arm was conducted to determine the differences in outcomes and complications between INFIX and plating for unstable pelvic ring injuries. A random-effects model was employed for this purpose; measures of treatment effects included mean difference (MD) for continuous variables (such as functional outcome score, perioperative blood loss, operative time) and odds ratio (OR) for dichotomous variables (such as complications). 95% confidence intervals (CI) were calculated for each outcome of interest. The I2 test was used to assess statistical heterogeneity. Forest plots were constructed to provide a visual summary for each outcome of interest. The RevMan 5.4 software was used for these analyses [7].

The non-comparative meta-analysis arm was conducted to determine the pooled effect size for outcomes and complication rates. Outcome measures included event rate for dichotomous variables and mean for continuous variables; 95% CI were calculated for each outcome of interest. The I2 test was used to assess statistical heterogeneity. Forest plots were constructed to provide a visual summary for each outcome of interest. All analyses were performed by the Open MetaAnalyst software [8].

Exploration of heterogeneity

For outcomes which showed high statistical heterogeneity (I2 > 75%), leave-one-out meta-analysis and meta-regression (by a random-effects model) were performed. To facilitate meta-regression, the following covariates were studied:

(a) the percentage of complex cases in the study (defined as the percentage of vertical shear and/or combined mechanism injuries, or Tiles C) (continuous variable).

(b) total cases (continuous variable).

(c) ethnicity—Asian versus non-Asian study population (dichotomous variable).

Subgroup analysis

Subgroup analysis was done by dividing the studies into those with patients of Asian ethnicity versus those of non-Asian ethnicity.

Results

Literature search

The PRISMA flowchart for the study has been presented in Fig. 7. A total of 384 records were identified, and full text retrieved for 49 studies. Twenty-seven studies were excluded as per the exclusion criteria, and a total of 22 studies were included for qualitative analysis [9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30]. The details of these studies are presented in Table 3. Comparative meta-analysis was performed on 3 studies that compared INFIX with plating [12, 25, 30] and pooled analysis was performed on 18 studies.

Table 3 Characteristics of studies included
Full size table

Characteristics of study

A summary of the studies included in the review has been presented in Table 4. A total of 746 patients were included for qualitative analysis; of these, pooled analysis was done for 589 patients. The average follow-up of these studies ranged from a minimum of 5.4 months to a maximum of 54 months. The mean time to radiological union was reported in 3 studies with an average time of 4.1 months [11, 15, 18].

Table 4 Summary of the studies included
Full size table

Comparative meta-analysis—plating versus INFIX (Fig. 1)

Fig. 1
figure 1

Forest plots for comparison of INFIX vs plating. a Functional outcomes. b Number of complications. c Blood loss. d Perioperative time

Full size image

Comparative meta-analysis revealed significantly less blood loss with the INFIX procedure as compared to plating (mean difference = 176.46 mL; 95% CI − 207.54 to − 145.38) and significantly shorter operative time (mean difference = 26.43 min, 95% CI − 31.79 to − 21.07) and no significant difference in the overall complication rates (OR 1.59, 95% CI 0.83–3.05) and functional outcome score (MD − 2.51, 95% CI − 5.73 to 0.71).

Pooled analysis

Perioperative events

The pooled mean operative time estimated from 12 studies (330 patients) was 52.1 min (95% CI 43.69–60.51); the heterogeneity for this estimate was high (I2 = 99.42%, p < 0.01) (Fig. 2). Meta-regression revealed that the high heterogeneity was not explained by the percentage of complex pelvic injury cases, total number of cases or the surgical volume (less than or more than 25 cases).

Fig. 2
figure 2

Forest plots for perioperative events reported in various studies. a Blood loss. b Intra operative time

Full size image

The pooled mean blood loss estimated from 10 studies (266 patients) was 65.425 mL (95% CI 51.022–79.828); the heterogeneity for this estimate was high (I2 = 99.61%, p < 0.001) (Fig. 2). Covariate analysis for the percentage of complex cases (vertical shear or combined type/tiles type C) showed significant increase in blood loss with the increase in complex cases (p = 0.03).

Outcomes (Table 5)

Table 5 Assessment of radiological and functional outcomes
Full size table

Radiological outcomes

The quality of reduction was reported as per Matta’s criteria [31] by 7 studies in 161 patients. Pooled analysis showed an excellent to good reduction in 91.4% of the cases (95% CI 0.860–0.969), with a moderate heterogeneity (I2 = 45.26%, p = 0.09) (Fig. 3).

Fig. 3
figure 3

Forest plots for studies reporting outcomes a radiological, b functional, c Majeed Score

Full size image

Functional outcomes

Majority of the studies reported Majeed scoring [32]. Only 1 study [10] used the Iowa pelvic score as they consider it to have the best correlation with SF-36 and is a good tool for measuring outcomes in severe pelvic injury. Pooled analysis of 8 studies reporting Majeed score showed an average score of 86.48 (95% CI 83.34–89.61); the heterogeneity for this event was high (I2 = 92.37%, p < 0.01) (Fig. 3). Meta-regression analysis showed that better Majeed score was obtained with complex fracture patterns (p = 0.001).

Outcome scoring classified as excellent or good was reported in 82.8% cases in 10 studies with 228 patients (95% CI 73.7–91.9%); the heterogeneity for this event was high (I2 = 82.24%, p < 0.001) (Fig. 3). Meta-regression showed a trend of poorer results with complex patterns but was not statistically significant (p = 0.118). Leave-one-out analysis excluding Kuttner et al. [27] improved overall estimate from 82.8 to 88.2% (95% CI 81.9–94.5%).

Complications

The complications of the INFIX procedure have been presented in Table 6.

Table 6 Complications
Full size table

Non-union

Only 2 cases of non-union have been reported in literature. Muller et al. [22] reported nonunion in a schizophrenic non-compliant patient while Dahill et al. [29] reported 1 asymptomatic nonunion. Pooled analysis showed an overall nonunion rate of 1.2% (95% CI 0.4–2.1%); the heterogeneity for this event was low (I2 = 0%, p = 1.000) (Fig. 4).

Fig. 4
figure 4

Forest plots of comparison of a non-union, b infection, c implant loosening, d heterotopic ossification

Full size image

Infection

The overall rate of infection with INFIX was seen to be 2.3% (95% CI 1.1–3.4%); the heterogeneity for this event was low (I2 = 0%, p = 0.971). Seventeen cases of infections have been reported overall in the studies (Fig. 4).

Loosening

Loosening of implant is a rare complication reported in 7 cases. The overall rate of loosening was found to be 1.3% (95% CI 0.4–2.2%); the heterogeneity for this event was low (I2 = 0%, p = 0.995) (Fig. 4).

HO

HO though usually asymptomatic is a commonly associated complication with an overall rate of 9.4% (95% CI 5.5–13.3%); the heterogeneity for this event was high (I2 = 87%, p < 0.001). High number of cases of HO have been reported by Vaidya et al. [23] (n = 32), Hoskins et al. [21] (n = 14) and Yin et al. [12] (n = 12) in their studies (Fig. 4). Subgroup analysis was done according to ethnicity (Asians and non-Asians), and it was seen that non-Asians were more susceptible (23.6% versus 4.2% in Asians). Meta-regression according to ethnicity showed a strong correlation (p = 0.01).

LFCN injury

LFCN injury, mostly self-limiting is the most commonly seen complication with INFIX. Pooled analysis showed an overall rate of LFCN injury as 28% (95% CI 15.1–41%); the heterogeneity for this event was high (I2 = 95.7%, p < 0.001). Vaidya et al. [23] reported 27 LFCN injuries in a series of 96 patients, while Hoskin et al. [21] reported 20 nerve injuries in 21 cases (42 screws) (Fig. 5). Cumulative meta-analysis shows that from the year 2018, the complication rate of LFCN injury is similar, but despite 9-year experience, the complication rate is stationary around 30%. Subgroup analysis was done according to ethnicity (Asians and non-Asians), and it was seen that non-Asians were more susceptible to LFCN injury (40.9% versus 18.9% in Asians). However, meta-regression according to ethnicity showed a weak correlation (p = 0.053).

Fig. 5
figure 5

Forest plot of comparison of LFCN injury in various studies

Full size image

Deep venous thrombosis (DVT)

Pooled analysis showed that the overall rate of DVT was 1.2% (95% CI 0.3–2.1%); the heterogeneity for this event was low (I2 = 0%, p = 0.926). A total of 8 cases of DVT have been reported in literature (Fig. 6).

Fig. 6
figure 6

Forest plots of comparison of a DVT, b femoral nerve injury, c loss of reduction, d early implant removal

Full size image

Femoral nerve injury

Pooled analysis showed that overall rate of femoral nerve injury was 1.3% (95% CI 0.4–2.2%); the heterogeneity for this event was low (I2 = 0%, p = 0.998) (Figs. 6, 7).

Fig. 7
figure 7

PRISMA flow diagram for the study

Full size image

Loss of reduction

Loss of reduction has been reported in 6 cases with overall rate of 1.8% (95% CI 0.7–2.8%); the heterogeneity for this event was low (I2 = 0%, p = 1.000). This loss of reduction has been reported in the immediate post-operative phase, and no study has reported any loss of reduction post-implant removal (Fig. 6).

Early removal

Early implant removal due to loosening, infection or nerve injury was done in 18 patients with overall rate of 2.1% (95% CI 0.9–3.2%); the heterogeneity for the event was low (I2 = 0%, p = 0.761). However, studies have reported that despite the early removal of implant in these cases, the patients have satisfactory functional outcomes (Fig. 6).

Risk of bias assessment

The risk of bias assessment is depicted in Table 2 using MINORS tool. For non-comparative studies, 7 studies had a score of 12, 5 had a score of 11, 3 had a score of 10 and 1 had a score of 9. For comparative studies, 1 study had a score of 21, 3 had a score of 20, 1 had a score of 19 and 1 had a score of 18. There were no randomized control trials reported.

Discussion

Pelvic fractures are complex injuries with disruption of multiple bony or ligamentous components. The indications for anterior fixation with INFIX are unstable pelvic ring injuries (Tiles B and C or Young and Burgess-LC 2 and 3, APC 2 and 3, Vertical Shear and Combined) in conjunction with posterior fixation. Symphyseal injuries are conventionally treated with plating [19, 21]. However, in Types B and C injuries it is associated with extensive exposure and increased complications, and thus the role of INFIX has been developed as a useful modality. The sequence of fixation is usually posterior fixation followed by anterior fixation. However, Gardner et al. [24] demonstrated that anterior fixation should be considered first for pelvic reduction.

Biomechanical studies done by Vaidya et al. [33], Osterhoff et al. [34], Eagan et al. [35] show that mechanical stiffness of an INFIX construct is similar or superior to that of an external fixator. Monoaxial screws are 26% more rigid than polyaxial screws but are technically more difficult. However, Vigdorchik [36] showed that internal fixation of the anterior ring with plates was still the strongest when compared to INFIX or an external fixator. As such, in patients with pure symphyseal injury, internal fixation is still the preferred treatment.

INFIX being a subcutaneous procedure was associated with relatively low blood loss (mean 65.425 mL). However, no study reported the method of estimation of blood loss which could account for the variability. Moreover, few studies have reported blood loss for anterior and posterior fixation combined which could be a possible cause of high heterogeneity in our analysis.

The duration of INFIX fixation was seen to be an average of 52.1 min. The pooled analysis shows that it a short procedure with excellent results. However, the need for a concomitant surgical procedure such as spino-pelvic fixation by Du et al. [9] increased the operative time in their series. We also observed that being a percutaneous procedure with only indirect reductions being done, the operative time does not increase significantly with complex anterior ring injuries.

Only 2 nonunion cases have been reported in literature which is indicative of a very high union rate, which is also true for pelvic fractures in general. Adequate reduction of fracture fragments is the key for a successful outcome. Outcome scores have mainly reported excellent to good results. These are better in comparison with those studies where anterior external fixators were applied or when injuries were left conservatively to heal on their own [15]. However, many pelvic ring injuries are associated with multiple other injuries that affect the overall outcome and make it impossible to assess treatment [19]. For these, patient-reported outcome scores have been developed which may need to be reported more in the forthcoming studies.

No procedure is without complications and INFIX has been associated with LFCN injury, HO, vascular injury, DVT, femoral nerve palsy and surgical site infection.

LFCN injury is a common complication but is usually self-limiting. Rudin et al. [37] studied in detail the anatomical variations of LFCN in 18 cadavers and saw different pathways. They observed that it crossed the level of the inguinal ligament just superficial to the ASIS in 29%, lateral to ASIS in 11% and medial to ASIS in 69%. Osterhoff et al. [38] observed that the LFCN is found to run 13.49 mm from lateral end of bar. Thus, a very short lateral end of the bar prevents the irritation of LFCN. However, we observed that the incidence of LFCN injury remains as high as up to 30% and thus proper identification of the nerve by careful dissection up to the bone and extending the incision as per comfort is important to minimize its occurrence. We also noticed that non-Asians were more susceptible to LFCN injury (40.9% versus 18.9% in Asians) and this could be because of anatomical variations.

HO though common is usually asymptomatic. Vaidya et al. in their study [19] found older age and higher injury severity score as risk factors for developing HO. However, no significant relation was found with sex, race, Majeed score and Glasgow coma scale in their study. This can be usually removed at the time of implant removal in cases of excessive bone formation. In our pooled analysis we observed a higher incidence of HO in non-Asian population and this could be related to the genetics as incidence of HO is known to be influenced by race [39]. HO was usually Grade 1 or 2 and no active intervention was required.

Femoral nerve palsy has also been reported as a rare complication. Hesse et al. [40] in their study reported 8 femoral nerve palsies in 6 patients. They said that injury is likely due to limited space available for the psoas muscle and femoral nerve and that sinking screws or under contouring of the rod is a possible cause of compression. Thus studies by Scherer et al. [41] recommended bar to bone distance of 20–25 mm to avoid compression. Vaidya et al. [23] recommended the screw head to be kept at the level of the sartorius such that the subcutaneous bar remains above the level of deep fascia.

Smith et al. [42] reported a case of bilateral compression of the external iliac artery after the application of INFIX. They identified on CT angiogram the bulging of the rectus muscle underneath the INFIX and subsequent compression against the bar which resolved by increasing the arc of curvature. Merriman et al. [43] calculated average bar to vessel distance of 2.2 cm and rod to bladder distance of 2.6 cm. Thus, it can be usually avoided by proper implant placement and adequate contouring of the anterior bar.

Surgical site infection was seen in 17 cases (2.3%) with no evidence of osteomyelitis. Pelvic INFIX is seen to be associated with low rates of infection and is a safe procedure.

The rates of DVT have been seen to be low with the application of INFIX (1.2%). This may be due to early mobilization with this implant and the routine use of thromboprophylaxis.

There is an inherent risk of abdominal wall perforation with errant rod placement. It is important to maintain the rod in a subcutaneous position without violation of the abdominal fascia. This can be performed reliably by tunneling a path below the abdominal wall before rod insertion. Jain et al. [44] reported a rare case of bladder incarceration following application of INFIX in a case of pubic diastasis. This occurred at the time of compression of the diastasis after application of INFIX. Thus, an index of suspicion is warranted and timely INFIX removal with open reduction and stabilization is warranted.

Patients with INFIX have “awareness” about their implants due to its subcutaneous position which is a cause of uneasiness as compared to conventional plating [39]. Thus, most of the implants require removal after radiological union. However, as opposed to an external fixator, the subcutaneous nature of this construct requires removal in an operating room.

Some modifications have been applied to INFIX technique with time. Few studies such as Bi et al. [28] and Wu et al. [16] have added a 3 cm Pfannenstiel incision 2 cm above the pubic symphysis and inserted a polyaxial pedicle screw through the ipsilateral or contralateral pubic tubercle to the inferior ramus. They believe that it forms a stable geometric triangle allowing better fit of the rod with the anatomy of the anterior pelvic ring. It also helps in better restriction of relative micro-movement between the sites of the fractured pubis. Wang et al. [30] used the modified INFIX technique for cases of pubic diastasis by inserting 4 screws (one each in pubic tubercle) in a series of 29 cases and reported excellent to good outcomes in 27 patients. Robot-assisted placement of screws has also been reported by Du et al. [9] and Liu et al. [11]. Liu et al. in their study compared robot-assisted INFIX with the conventional technique and showed shorter duration of surgery, need for lesser fluoroscopy, and reduced blood loss with the robotic technique.

There are some limitations in this study. The included studies as mentioned are all case series or comparative studies with no randomized control trial. There were only 3 studies available for comparative meta-analysis. Secondly, a high statistical heterogeneity was noted for some of the outcomes reported in our analysis. We have, however, performed meta-regression and subgroup analysis to try and identify the reasons for heterogeneity.

From our systematic review and meta-analysis, it appears that INFIX is a valuable tool for reduction and fixation in unstable pelvic ring fractures and is well tolerated by the patient, allowing good mobility with acceptable complications and outcomes. It is a minimally invasive technique with less blood loss and surgical time in comparison with conventional plating with comparable complications and similar outcomes.

Conclusion

INFIX is a biomechanically stable and effective minimally invasive surgical technique for management of unstable pelvic ring fractures. However, it is imperative to stabilize the posterior ring first if required with use of plates and screws.