Risk Factors for Surgical Site Infection after Operative Fixation of Acetabular Fractures: Is Psoas Density a Useful Metric?

Abstract

Background

Surgical site infection (SSI) occurs in 5% to 7% of patients undergoing operative fixation of acetabular fractures, with reported risk factors including longer operative duration, increased blood loss, pelvic artery embolization, and concurrent abdominal organ injury, among others. Psoas muscle density is a measure of muscle quality and, as a metric for sarcopenia and/or nutrition status, has been associated with poor outcomes such as mortality across multiple surgical specialties. To date, psoas muscle density has not been explored for its associations with SSI in acetabular fracture patients.

Questions/purposes (1) Is decreased psoas muscle density, as measured by Hounsfield units, associated with an increased SSI risk after acetabular fracture fixation? (2) What patient, operative, and hospital variables are associated with an increased SSI risk after acetabular fracture fixation?

Methods

Between 2012 to 2017, surgeons performed 684 acetabular ORIF procedures at one level I trauma center. Of those, 8% (56 of 684) did not meet inclusion criteria, leaving 92% (n = 628) for analysis in this study. The median (range) follow-up duration was 12 months (0.5-77). Patient demographics, comorbidities, operative and in-hospital variables, and psoas muscle density measured using preoperative pelvic CT images—acquired for all operative acetabular fracture patients—were analyzed. SSI was defined by positive culture results obtained during irrigation and débridement. Overall, 7% (42 of 628) of patients had an SSI. A multivariable regression analysis was performed to identify independent risk factors. Sensitivity analysis was performed with minimum follow-up set at 3 months and 6 months.

Results

There was no difference in the mean psoas muscle density between patients with SSI (50.9 ± 10.2 Hounsfield units [HUs]) and those who did not have an SSI within 1 year of open reduction and internal fixation (51.4 ± 8.1 HUs) (mean difference: 0.5 [95% confidence interval -2.34 to 3.32]; p = 0.69). Four variables were independently associated with an increased risk of SSI: increased operative time (1.04 [95% CI 1.00 to 1.07]; p = 0.03), estimated blood loss (1.08 [95% CI 1.02 to 1.14]; p = 0.01), female sex (2.34 [95% CI 1.19 to 4.60]; p = 0.01), and intravenous drug use (3.95 [95% CI 1.51 to 10.33]; p = 0.01). Sensitivity analysis showed no change in results using either 3-month or 6-month minimum follow-up.

Conclusions

Risk factors for SSI after acetabular fixation include female sex, intravenous drug use, prolonged operative times, and increased intraoperative blood loss. Although the density of the psoas muscle may be a surrogate for nutritional markers, it was not associated with SSI in our patients with acetabular fractures. Thus, it is not useful for risk assessment of SSI in the general population with acetabular fracture; however, future studies with larger sample sizes of patients older than 60 years may re-investigate this marker for SSI risk. Contrary to the results of previous studies, pelvic artery embolization, intraoperative blood transfusion, and intensive care unit stay did not increase the risk of SSI; however, we may have been underpowered to detect differences in these secondary endpoints. Future large, multisite studies may be needed to address these conflicting results more definitively.

Level of Evidence

Level III, therapeutic study.

Introduction

Surgical site infection (SSI) rates remain at 5% to 7% after acetabular open reduction and internal fixation (ORIF) [5, 9, 11, 22]. Various CT measurements of the psoas muscle have recently gained attention because of their association with postoperative mortality when used as a metric of frailty or sarcopenia in patients undergoing various surgical operations, including acetabular fracture repair [4, 16, 21]. Psoas muscle measurements are also associated with postoperative complications in patients undergoing surgery for trauma and elective total joint arthroplasty [1, 24]. The density of the psoas muscle, a measure of the Hounsfield unit (HU) attenuation, provides insight into fat infiltration of the muscle tissue, muscle quality, and muscle strength, acting as a surrogate for overall muscle quality [7]. Psoas density has also been shown to directly correlate with albumin levels in patients undergoing general surgery [14] and is also linked to malnourishment in patients in the intensive care unit (ICU), both at baseline and during hospital admission [23]. Additionally, psoas muscle density may play a role in identifying at-risk patients, as a recent study found an association between decreased psoas muscle density and increased rates of poor outcomes for trauma patients [24]. Thus, while it is typically considered a metric of sarcopenia/frailty, psoas density may be able to help identify patients in which nutritional supplementation may be beneficial and allow providers to track the effectiveness of this supplementation [23].

To date, the association between psoas muscle density and the risk of SSI in patients undergoing surgery for acetabular fractures has not been explored. Preoperative pelvic CT scanning is routinely performed for patients with acetabular fractures, making psoas measurements readily available in this population. The psoas density may therefore provide an efficient, noninvasive metric that offers insight into the nutritional status of the patient and may act as a predictive measure of postoperative complications. Additionally, few studies have reported risk factors for SSI after acetabular fracture fixation [5, 9, 13, 15]. Although several risk factors such as increased operative duration and blood loss appear consistently in these studies, others such as pelvic artery embolization and associated abdominal or genitourinary factors have shown conflicting results with regards to increasing SSI risk in these cases [5, 6, 15, 18].

Therefore, we asked: (1) Is decreased psoas muscle density, as measured by Hounsfield units, associated with an increased SSI risk after acetabular fracture fixation? (2) What patient, operative, and hospital variables are associated with an increased SSI risk after acetabular fracture fixation?

Patients and Methods

Institutional review board approval was obtained for this retrospective study. All patients treated operatively for acetabular fractures at our Level I trauma center from January 2012 through December 2017 were identified using Current Procedural Terminology codes 27226, 27227, and 27228. Patients were included if they were skeletally mature and underwent operative fixation of acetabular fractures. Patients were excluded if they were skeletally immature, nonambulatory at baseline, underwent nonoperative treatment, or their condition was treated acutely with both THA and acetabular ORIF. Of the patients identified by Current Procedural Terminology codes, 3% (17 of 684) of patients were excluded because of skeletal immaturity, 1% (nine of 684) were excluded because they underwent acute primary THA in conjunction with acetabular ORIF, and the remaining 4% (30 of 684) of patients were excluded because they had a nonambulatory status at baseline, life-threatening brain injury, or paralysis at the time of acetabular fracture. The final cohort consisted of 628 acetabular fractures in 620 patients who were eligible for analysis (Fig. 1). Each of the eight patients (1%) with multiple acetabular fractures had the injuries occur to their opposite hip in a separate injury from their initial fracture side. No included patients had bilateral operative acetabular fractures occurring in the same injury.

Fig. 1.

The STROBE flow chart of patients for this study is shown here.

Patient Characteristics

Patient-related characteristics and comorbidities were obtained by medical record review (see Table 1, Supplemental Digital Content 1, http://links.lww.com/CORR/A309). Fractures were classified by the operating surgeon according to the Letournel and Judet classification system [10], and injury-related variables were obtained by further review of the medical records (see Table 2, Supplemental Digital Content 2, http://links.lww.com/CORR/A310). A low-energy injury was defined as a fall from standing or fall from less than 5 feet high, while all other injury mechanisms were considered high-energy. The iNtuition viewer software (v4.4; TeraRecon, Foster City, CA, USA) was used to measure the psoas muscle density in Hounsfield units, as previously described [21]. We did this manually, outlining the borders of both psoas muscles on a single cross-sectional CT image at the mid-L3 vertebral body level, based on a preoperative contrast-enhanced CT image (Fig. 2 A-B). Among the study group, 440 patients (including all patients with infection) who had preoperative contrast-enhanced CT images taken at our institution were included in this portion of the analysis. All 628 patients included in the study cohort had preoperative CT images obtained, however, 30% (188 of 628) of the patients were excluded from this analysis due to a lack of contrast-enhanced CT images, and/or imaging according to another institution’s protocol, which may have caused substantial differences in the measurements.

Fig. 2 A-B.

Representative images demonstrating psoas muscle density as described in this study from a patient with very high psoas density and one with very low psoas density, matched closely for age, gender, and BMI. The red highlighted portion represents how the psoas muscle would be outlined (though we simply outline the outer edges of the muscle for the actual program). The opposite psoas muscle is left untouched in these images to demonstrate how the muscle density actually appears. In this study we outline the borders of both psoas muscles on a single cross-sectional CT image at the mid-L3 vertebral body level, of a preoperative contrast-enhanced CT image. Then the density of the right and left psoas muscles in Hounsfield units (HUs) are averaged for the overall psoas density. (A) Represents the psoas muscle of a 35-year-old female with a BMI of 26, and the mean psoas density for this patient is 75.5 HUs. (B) Represents the psoas muscle of a 37-year-old female with a BMI of 27, and the mean psoas density for this patient is 33.2 HUs.

Surgical Technique and Perioperative Protocols

Eleven orthopaedic trauma fellowship-trained surgeons performed acetabular ORIF during the study period. The surgical approach was dictated by the individual fracture classification and associated pelvic-ring injuries, if present, at the discretion of the operating surgeon. Anterior approaches included lateral window, modified Stoppa/anterior intrapelvic, and ilioinguinal approaches, and posterior approaches included the Kocher-Langenbeck and Gibson approaches. Most procedures involved ORIF with plate and screw constructs using a single anterior or posterior approach (see Table 3, Supplemental Digital Content 3, http://links.lww.com/CORR/A311). Six percent (40 of 628) of patients underwent percutaneous screw fixation only, 4% (25 of 628) underwent combined anterior and posterior approaches, and 0.5% (three of 628) had extended iliofemoral approaches.

Perioperative and postoperative institutional protocols were adhered to, with minor variations depending on the surgeon’s preference. Perioperative details were obtained by medical record review and compared between the two cohorts (see Table 3, Supplemental Digital Content 3, http://links.lww.com/CORR/A311). Between the cohorts, there were no differences in wound lavage or prophylaxis before closure and the use and type of heterotopic ossification prophylaxis (see Table 3, Supplemental Digital Content 3, http://links.lww.com/CORR/A311). All patients received perioperative antibiotics within 1 hour of incision. All patients were made non-weightbearing or toe-touch weightbearing with posterior hip precautions, when appropriate, postoperatively. Patients in whom SSI developed had an increased mean length of stay of 17 ± 9 days compared with 12 ± 8 days for those without infection (p = 0.001) (see Table 4, Supplemental Digital Content 4, http://links.lww.com/CORR/A312). The median (range) follow-up duration was 12 months (0.5-77).

Definitions of SSI and Infection Characteristics

SSI was defined as positive results of cultures taken during an irrigation and débridement performed for suspected infection. Deep SSI was defined as any positive cultures taken from tissues deep in the fascia or from tissue or a swab of the implants or surrounding bone. Superficial SSI was defined as any positive results of a tissue culture or a fluid sample superficial to the deep fascia, without communication with the implants.

Seven percent (42 of 628) of patients had an SSI based on the criteria listed above. Seventy-nine percent (33 of 42) of the infections were down to the fascia and involved the bone and fixation hardware, while the remaining 21% (nine of 42) of infections were superficial to the fascia and did not track deep in the bone or hardware. The mean (range) time from acetabular surgery to diagnosis by irrigation and débridement was 30 days (5-204), and 93% (39 of 42) of the infections occurred within 90 days of fixation (see Table 5, Supplemental Digital Content 5, http://links.lww.com/CORR/A313). Twenty-nine percent (12 of 42) of infections were polymicrobial. The most common organism was methicillin-resistant Staphylococcus aureus, isolated in 38% (16 of 42) of infections. All patients with SSI were initially treated with irrigation and débridement and intravenous antibiotics for 6 weeks, with 64% (27 of 42) of patients underwent no additional surgery. Thirty-three percent (14 of 42) of patients underwent additional irrigation and débridement with antibiotic bead placement and 2% (one of 42) of patient underwent eventual resection arthroplasty to control infection. Thirty-one percent (13 of 42) of patients underwent three or more procedures to control the infection.

Statistical Analysis

We performed univariate analysis using Fisher’s exact and chi-square tests for categorical variables, and a linear regression analysis was performed for continuous variables to compare the SSI and non-SSI cohorts. Variables with p values less than 0.05 on univariate analysis were then included into a multivariable logistic regression analysis performed to identify independent risk factors for SSI by estimating odds ratios and 95% confidence intervals. P values of less than 0.05 were considered statistically significant. The amount of blood transfused and ICU length of stay were calculated based only on patients who had a blood transfusion or ICU stay, respectively.

Power analysis for psoas muscle density was performed post-hoc given that this is the first study to explore this metric for associations with SSI. With the current sample size (628 total patients) and an alpha set at < 0.05, our study had 80% power to detect a statistically significant difference in psoas muscle density of 5.3 HUs (delta) between patients with and without SSI.

Sensitivity Analysis

Due to 18% (113 of 628) of patients not having a true follow-up duration of 6-months, sensitivity analyses were performed by limiting the analysis to include only patients with minimum follow-up set at two different time points: 3 months and 6 months postoperative. These were chosen because the practice of some of our trauma surgeons includes discharging patients from clinic who are doing well and have achieved fracture union, typically around the 6-month follow-up visit, and allowing them to follow-up as needed. The time points were also chosen because most SSIs after acetabular fractures occurred within 90-days in prior publications [5], just as they did in our study.

Results

There was no difference in the mean psoas density between patients with SSI (50.9 ± 10.2 HUs) and those in whom SSI did not develop within 1 year of ORIF (51.4 ± 8.1 HUs) (mean difference: 0.5 [95% CI -2.34 to 3.32]; p = 0.69). When stratified by age, there was no difference in the psoas density between those with SSI and those without among patients at least 60 years old (40.8 versus 44.1 HUs respectively; p = 0.64) or among patients younger than 60 years (52.9 versus 52.3 HUs respectively; p = 0.68) (see Table 1, Supplemental Digital Content 1, http://links.lww.com/CORR/A309). An additional analysis by quartile revealed no difference in the risk of SSI between the lowest quartile of psoas density (< 46.0 HUs) and the remaining quartiles (8% versus 7% respectively, p = 0.27). Five of 21 (24%) patients had a psoas density ≤ 35.0 HUs and had SSI, compared with 37 of 607 (6%) patients with a psoas density > 35.0 HUs (p = 0.12).

Four variables were independently associated with an increased odds of SSI: increased operative time (1.04 [95% CI 1.00 to 1.07]; p = 0.03), estimated blood loss (1.08 [95% CI 1.02 to 1.14]; p = 0.01), female sex (2.34 [95% CI 1.19 to 4.60]; p = 0.01), and intravenous drug use (3.95 [95% CI 1.51 to 10.33]; p = 0.01). Other variables included in the multivariate regression analysis were BMI category and intraoperative blood transfusion, but these were associated with SSI risk (Table 1).

Table 1.

Multivariable regression analysis of independent risk factors for SSI after acetabular fracture fixation for select variables (p < 0.05) from the univariate analysis

Sensitivity analysis was performed with minimum follow-up at 3 months and 6 months. Results of this analysis showed no difference in the overall outcomes of the study, including no change in the psoas density differences between the SSI and no-SSI cohorts. Additionally, there was no change to any of the variables in the univariate analysis and, as a result, there was also no change in the multivariable analysis results (see Tables 6-10, Supplemental Digital Content 6, http://links.lww.com/CORR/A314, Supplemental Digital Content 7, http://links.lww.com/CORR/A315, Supplemental Digital Content 8, http://links.lww.com/CORR/A316, Supplemental Digital Content 9, http://links.lww.com/CORR/A317, and Supplemental Digital Content 10, http://links.lww.com/CORR/A318).

Discussion

Surgical site infections after acetabular ORIF cause patient morbidity and increase expenses to the healthcare system. Although many variables associated with an increased risk of SSI are not modifiable in these patients, it is important to identify those that may be modifiable to help trauma surgeons with preoperative planning, patient counseling, and postoperative monitoring to mitigate the risks when possible. The psoas density has been shown to be correlated directly with the albumin level [14] and is a potential marker of malnutrition or poor overall health. The link between malnutrition and postoperative infection was well-established in previous studies [2, 3]. However, the utility of pre-albumin and albumin may be limited in patients with severe trauma because they are negative, acute-phase reactants and rapidly become injured or ill [12]. Thus, a simple measurement of the psoas density using a preoperative pelvic CT image is a potentially valuable, efficient tool to identify patients in whom SSI may develop. However, we found that a decrease in psoas muscle density was not associated with SSI, while female sex, increased intraoperative blood loss, prolonged operative time, and intravenous drug use were associated with increased SSI risk.

Limitations

This study is not without limitations. First, there is the potential for selection bias. At our institute, adult patients are managed operatively for displaced acetabular fractures (> 2 mm) or hip fracture/dislocations with instability. At times, geriatric acetabular fractures with osteoporotic bone are managed nonoperatively with a plan for early conversion arthroplasty. We do not believe this affected our results given that the aims of this study both involved only those patients undergoing operative fixation of their acetabular fracture. Second, follow-up was variable in our study population, introducing the potential for transfer bias. We required only a short minimum follow-up duration, defined in our inclusion criteria as patients having to at least return for their initial post-surgery follow-up visit, as we felt extending the minimum follow-up may exclude patients with SSI who did not meet the minimum follow-up criteria. To address this potential bias, we performed sensitivity analyses in which the minimum follow-up was set to 3 months and then 6 months. These were chosen because 93% (39 of 42) of the SSI occurred within 3 months and 98% (41 of 42) occurred within 6 months. The results of these analyses showed no change in any of the results of interest, including psoas density measurements, or any of the preliminary analyses, indicating that the results were not influenced by variable loss-to-follow-up.

There is also the possibility of assessment bias resulting from our definition of infection as requiring positive intraoperative cultures during irrigation and débridement. This definition relies on the consistent obtaining of cultures during irrigation and débridement. Five patients who developed postoperative wound complications such as hematoma, seroma, or increased drainage and who returned to the operating room for irrigation and débridement did not have undergo tissue culture because the operating surgeon had a low suspicion of infection. None of these patients returned to the operating room for eventual SSI. By our definition, these patients were not included in the SSI cohort, but indicate that our definition may have resulted in underestimating the true SSI rate. We felt this definition was the most objective definition of SSI and the definition most consistently used in the current evidence [5, 9, 15]. We acknowledge that the definition and diagnostic algorithm for SSI after fracture surgery or fracture-related infection is still a topic of some debate, but we feel with the current available evidence our definition is the least subjective given that it is the most definitive criteria of the recent consensus statement attempting to define criteria for diagnosing fracture-related infection [8]. Finally, in determining risk factors for SSI, our study may have been underpowered to detect differences in the variables that are low-frequency variables. Thus, while our study demonstrated no difference for a number of variables, we cannot definitively say whether there in fact was no difference for these variables or if there was a difference and we were unable to detect one due to a lack of statistical power. Future, multisite studies may be required to determine the true significance of these low-frequency variables.

No Association Between Psoas Density and SSI

Our results showed no difference in the psoas density between patients with SSI and those without, between patients younger than 60 years and those at least 60 years old, and between the lowest quartile of psoas density and the remaining quartiles. Thus, a decreased psoas density does not predispose patients with acetabular fractures undergoing operative fixation to SSI. Although no other studies have examined the utility of the psoas density to predict infection, a recent study reported an increased risk of mortality in elderly patients with acetabular fractures who had sarcopenia, as measured by the skeletal muscle index. The study showed no correlation between the psoas density and SSI [4].

Factors Independently Associated with SSI

Our study identified four independent risk factors for SSI after acetabular ORIF: prolonged operative time, increased intraoperative blood loss, female sex, and intravenous drug use. Two of these variables, prolonged operative time and increased intraoperative blood loss, have been shown to be risk factors for SSI after acetabular ORIF [5, 9, 13]. In our cohort, for every 10 minutes of additional operative time, the odds of SSI increased by 4%, which is consistent with the result of a study by Ding et al. [5], who reported an increased odds of SSI of 22% for every additional hour of operative time. Additionally, based on this multivariate model, for every 100 mL increase in intraoperative blood loss, the odds of SSI increased by 8%. Both aforementioned risk factors likely reflect the complexity of surgery but may also be modifiable if the procedure includes ORIF of multiple fractures during the same visit to the operating room. Intravenous drug use was a major SSI risk factor in this cohort of patients with acetabular fractures. Although this variable was not included in previous studies investigating acetabular fractures, this result is consistent with the result of studies on ORIF of clavicle and ankle fractures as well as studies of elective THA [17, 19, 20]. In contrast to some previous studies, with the numbers available, pelvic artery embolization, associated abdominal and genitourinary injuries, associated Morel-Lavallee lesions, time from admission to surgery, and length of ICU stay were not associated with an increased risk of SSI in our cohort. Pelvic artery embolization has been reported by several studies to be a major risk factor for SSI after acetabular ORIF [5, 15, 18]. However, our results are in line with the conclusions of Firoozabadi et al. [6], who demonstrated no increased risk of SSI with pelvic artery embolization. Conflicting results between studies with regard to pelvic artery embolization may be a result of differences in embolization technique and indications or may differ based on which artery is being embolized (that is, the internal iliac artery compared with the superior gluteal artery, etc). At our facility, selective pelvic artery embolization, when indicated, continues to be an effective part of the acute resuscitation of patients with multi-trauma, and it is a way to control hemorrhage and decrease transfusion rates.

Conclusions

In this study, psoas density was not associated with SSI risk. Thus, despite its reported potential as a surrogate for nutritional status and sarcopenia, it is not useful for SSI risk assessment in the general population with acetabular fracture. However, future studies with larger sample sizes of patients older than 60 years may re-investigate this marker for SSI risk given that these patients appear to have larger variations in muscle density than younger patients. We identified four independent risk factors associated with SSI: prolonged operative time, increased intraoperative blood loss, female sex, and intravenous drug use. With the numbers available, our study found no associated SSI risk for variables including pelvic artery embolization, time from admission to surgery, and length of ICU stay. Future large, multi-site studies may be required to address these conflicting results more definitively.