Abstract
Introduction
Geriatric femur fractures occur in complex hosts that pose significant challenges in perioperative management, with venous thromboembolism (VTE) as a well described complication. Hyperglycemia, prevalent in the perioperative period, has been implicated in increasing thrombotic risk. However, data on its impact on outcomes following femur fracture fixation in the elderly population remains limited.
Methods
A retrospective cohort analysis was conducted on adults >60 years old undergoing operative fixation for femur fractures between January 2017 and December 2019. Demographic information, comorbidities, and outcomes were collected from electronic medical records.
Results
Six hundred and fourteen patients met inclusion criteria. The average age was 78.7 years (30.3% male; 25.0% with diabetes mellitus (DM) diagnosis). Patients with postoperative glucose ≥180 mg/dL had significantly higher rates of pulmonary embolism (PE) and VTE within 90 days, and glucose ≥180 mg/dL was an independent predictor for these events. Those with a DM diagnosis had higher rates of sepsis but not PE, VTE, or other complications compared to those without DM.
Conclusion
Acute perioperative hyperglycemia, regardless of DM diagnosis, is an independent risk factor for VTE and PE following geriatric femur fracture fixation. Strict glucose control in the postoperative period may improve outcomes in this patient population.
Keywords
Introduction
As the population continues to age, the number of geriatric patients with femoral fractures rises. 1 High rates of perioperative medical complications in the acute to subacute postoperative period are well described and include venous thromboembolism (VTE), pulmonary embolism (PE), superficial or deep infection, and acute blood loss anemia.2,3 Comorbidities such as cardiovascular disease, chronic obstructive pulmonary disease (COPD), hyperglycemia and diabetes mellitus (DM), as well as tobacco use have been commonly cited as risk factors which increase the likelihood of these complications. 3
Hyperglycemia is common in the orthopedic trauma perioperative period, even in patients without DM, likely due to systemic stress which stimulates the release of glucocorticoids and reduces insulin sensitivity. 4 Furthermore, hyperglycemia has been associated with increased production of coagulation factors and impaired fibrinolysis, 5 resulting in elevated risk for venous clot formation. Previous studies have linked preoperative hyperglycemia with higher rates of postoperative surgical site infection in orthopedic trauma patients, 6 and postoperative DVT/PE following total joint arthroplasty.7-10 However, little data exists on the effect of acute perioperative blood glucose control and its effect on acute to subacute outcomes following femur fracture fixation in elderly patients.
As postoperative hyperglycemia is a modifiable risk factor, the purpose of this study was to examine the effect of postoperative blood glucose control on outcomes following operative fixation of femoral fractures in geriatric patients.
Methods
This study is an Institutional Review Board (IRB) approved (STUDY22120129) retrospective cohort analysis following the STROBE guidelines that was performed on all adults undergoing operative fixation for a femur fracture at three affiliated academic institutions between January 2017 and December 2019. Inclusion criteria were age ≥60 years, availability of postoperative glucose values, operative fixation of an isolated femur fracture, including intertrochanteric, subtrochanteric, femoral shaft, femoral neck, supracondylar, or periprosthetic femur fractures (AO/OTA 31, 32, 33). Exclusion criteria were polytrauma patients, those with impending pathologic or pathologic fractures, atypical fractures secondary to medications, or any patient without at least two postoperative glucose values recorded in the electronic medical record following operative fixation of the femur fracture. The first three postoperative blood glucose values for each patient were collected and averaged. Patients with a known DM diagnosis received glucose measurements every 8 hours, while those without a diagnosis received daily glucose measurements.
The electronic medical record was utilized to collect demographic information including age, sex, body mass index (BMI), presence of diabetes mellitus (DM) diagnosis, fracture type, hospital length of stay (LOS), Charlson Comorbidity Index (CCI), age adjusted CCI, American Society of Anaesthesiologists (ASA) score, and glucose values at admission, morning of operation, and post-operatively for each patient. Complications following surgery included PE and VTE (defined as deep vein thrombosis (DVT) or PE) within 90 days postoperatively, surgical site infection (superficial infection or deep infection), hardware infection, other infections, sepsis, reoperation, nonunion, and 30-day mortality were also collected. Clinical symptoms of DVT and PE were tested using lower extremity dopplers and computed tomography (CT) pulmonary angiography to confirm or deny diagnosis. Superficial infection was defined as any infection at the surgical site requiring antibiotics without operative treatment, while deep infection was any infection requiring operative irrigation and debridement and antibiotics. Hardware infection was defined as any infection that required hardware removal at time of irrigation and debridement. Infections including urinary tract infections, pneumonia, clostridium difficile, etc. were evaluated as a collective that the authors defined as other infections. Reoperation was defined as return to the operating room for any reason related to the original surgery (i.e., irrigation and debridement, hematoma evacuation, revision, etc.). As consistent with prior literature examining the effect of preoperative or perioperative hyperglycemia on surgical outcomes, patients were stratified based on average postoperative blood glucose levels above or below 180 mg/dL, irrespective of DM diagnosis.11,12 Those with an average postoperative glucose value of 180 mg/dL or greater were compared to those with values less than 180 mg/dL. These groups were analyzed for differences in demographics and outcomes/complications. A secondary analysis of differences in demographics and outcomes/complications was also completed on patients with vs without a DM diagnosis.
Statistical Testing
Statistical analysis was performed using SPSS (IBM Corp. IBM SPSS Statistics for Mac, Version 27.0. Armonk, NY: IBM Corp). All patients were initially analyzed via multiple independent t-tests with average postoperative glucose as a continuous variable to determine associations with hyperglycemia and outcomes or complications. Then, the primary analysis evaluated patients with average postoperative glucoses of 180 mg/dL or greater against those with average postoperative glucoses under 180 mg/dL. These groups were compared based on demographics and outcomes/complications. Continuous variables were analyzed via multiple independent two-sided t-tests, while categorical variables were analyzed with two-sided Chi-square or Fischer exact tests. The continuous variables defined in this study were age, BMI, LOS, hemoglobin A1C, and CCI. The categorical variables were sex, fracture type, DM diagnosis, ASA score, PE, VTE, SSI30 (surgical site infection within 30 days), SSI90 (surgical site infection within 90 days), superficial infection, deep infection, hardware infection, additional infection, sepsis, reoperation, nonunion, and 30-day mortality. Patients who had less than two postoperative glucoses recorded were excluded from the comparison analysis. Logistic regression analysis was performed to evaluate whether the postoperative glucose cutoff value was an independent predictor for outcomes, regardless of baseline group demographics. A secondary analysis compared patients with a DM diagnosis to those without a diagnosis to evaluate baseline demographics and outcomes/complications between these groups. Continuous variables were analyzed via multiple independent two-sided t-tests, while categorical variables were analyzed with two-sided Chi-square or Fischer exact tests as was performed for the primary analysis. A P-value of <0.05 was considered statistically significant. An a priori power analysis performed using G* Power Version 3.1 (Heinrich-Heine-Universität Düsseldorf, Düsseldorf, Germany) determined a necessary sample size of 558 patients in one group (<180 mg/dL) and 58 patients in the second group (≥180 mg/dL) was required to detect a medium effect size using an allocation ratio of 8.6 based on our group sizes. To detect a large effect size, 197 patients in the <180 mg/dL group and 23 patients in the ≥180 mg/dL group was required. Post-hoc power analyses performed for continuous and independent variables confirmed a power of greater than 0.95 using this study’s sample size.
Results
Demographics
Six hundred and fourteen patients met the criteria for study inclusion (Figure 1). Amongst these 614 patients, the mean age was 78.7 ± 10.0 years. One hundred eighty-six (30.3%) patients were male, and 152 patients (25.0%) had a DM diagnosis. The mean BMI was 27.0 ± 7.7, the mean age adjusted CCI was 9.8 ± 3.1, and the mean ASA score was 3.0 ± 0.6 (Table 1). Inclusion, Exclusion, and Final Patient Cohort. Entire Patient Cohort Demographics. Continuous Variables Described as Mean ± Standard Deviation, While Categorical Variables Defined as Number (Percentage).
Analyzing average postoperative glucose levels as a continuous variable, the authors found that hyperglycemia was associated with increased rates of PE (P = .024), VTE (P = .016), DM diagnosis (P < .001), and BMI (P < .001), but did not differ in terms of age (P = 50), sex (P = .18), SSI30 (P = .32), SSI90 (P = .38), superficial infection (P = .48), deep infection (P = .40), hardware infection (P = .63), additional infection (P = .58), sepsis (P = .19), nonunion (P = .97), or reoperation (P = .62).
Primary Outcome: Postoperative Average Glucose Cutoff of 180 mg/dL
Cohort Demographics and Outcome Measures Separated Based on Average Post-operative Glucose Values. Continuous Variables are Displayed as Mean ± Standard Deviation, While Categorical Variables are Displayed as Number of Patients (Percentage).
Patients with mean postoperative glucose values ≥ 180 mg/dL were more likely to have a PE (7.8% vs 1.8%, P = .014) and VTE (10.9% vs 3.1%, P = .003) within 90 days (Table 2). When controlling for age, sex, BMI, DM diagnosis, CCI, and ASA score via logistic regression analysis, a mean postoperative glucose value ≥ 180 mg/dL was an independent predictor for PE (P = .047) and VTE (P = .012). The outcomes of LOS (P = .27), SSI 30d (P = 1.00), SSI 90d (P = 1.00), superficial infection (P = 1.00), deep infection (P = .73), hardware infection (P = 1.00), additional infection (P = 1.00), sepsis (P = .20), reoperation (P = .64), nonunion (P = .68), or 30 day mortality (P = 1.00) did not differ between groups.
Secondary Outcome: DM Diagnosis
Cohort Demographics and Outcome Measures Separated Based on Diabetes Mellitus Diagnosis. Continuous Variables are Displayed as Mean ± Standard Deviation, While Categorical Variables are Displayed as Number of Patients (Percentage).
Those with a DM diagnosis had higher admission glucoses, higher morning of operative intervention glucoses, higher average postoperative glucoses, and higher hemoglobin A1C values compared to those without diagnosed DM (all P < .001). Additionally, those with a DM had higher rates of sepsis (6.0% vs 1.3%, P = .003), but there were no differences in other outcome or complication measures (all P > .05, Table 3). The rate of nonunion was higher in patients with a diagnosis of DM (4.6% vs 1.8%, P = .06) which approached but did not reach statistical significance. No differences in outcomes or complications were seen when comparing glucose cutoff groups only in patients without diabetes (all P > .21) or only in patients with diabetes (all P > .052), however there was a trend towards significance for PE (P = .052) and VTE (P = .055) for those with a diabetes diagnosis.
Discussion
A mean postoperative glucose value ≥ 180 mg/dL was an independent predictor for PE and VTE in geriatric patients following femur fracture fixation. Little data exists describing the effect of postoperative blood glucose control on outcomes following femoral fracture fixation in this population. In a study of 217 femoral neck fractures, Yao et al, 8 found an association between preoperative blood glucose values and increased rates of preoperative VTE/PE. Liu and colleagues observed an association between preoperative hyperglycemia and rates of preoperative DVT identified on screening ultrasound in patients presenting with lower extremity fracture requiring fixation. 13 Perioperative and postoperative hyperglycemia have been linked to the development of embolic events in other orthopedic surgical subspecialities, as several studies have shown an association between hyperglycemia and postoperative DVT/PE following joint arthroplasty,7,9,14 yet minimal evidence is available to support this in the orthopaedic trauma setting. Still, it is accepted that hyperglycemia contributes to an elevation in coagulation factors, impaired fibrinolysis, and increased likelihood of thrombosis.15,16 This, in conjunction with the prolonged immobilization and pro-inflammatory state following lower extremity fracture fixation, may provide explanation for the main finding of the present study.
Demographically, patients with average postoperative blood glucose values ≥ 180 mg/dL were more likely to have a diagnosis of DM, elevated hemoglobin A1c at presentation, elevated BMI, CCI, and age adjusted CCI scores. Many of these factors have been linked to adverse events following operative intervention in orthopaedic trauma including higher rates of SSI, reoperation, and mortality.17-19 Although this current study’s analysis found that average postoperative blood glucose values ≥ 180 mg/dL were independently associated with increased incidence of VTE and PE, it is important to understand the medical complexity of patients and the cumulation of risk factors that are present and play a role in outcomes in this demographic of patients.
Perioperative hyperglycemia has also been cited as a risk factor for postoperative infection following operative intervention of orthopaedic trauma. Interestingly, the current study did not find elevated mean postoperative blood glucose of ≥180 mg/dL to be an independent risk factor for any infectious parameter. This is likely explained by the small overall percentage of infections in both groups. Additionally, glycemic control was only evaluated in the immediate perioperative period, so the effect of more prolonged hyperglycemia is not evaluated here. Perhaps longer follow up, stratifying patients after more prolonged uncontrolled blood glucose would better evaluate the effect of hyperglycemia on surgical site infection.20-26
DM has been widely accepted as having a negative impact on outcomes of orthopaedic surgery, regardless of the subspecialty.27-29 In orthopaedic trauma, diagnoses of diabetes have been linked to greater risk of mortality, nonunion, and deep postoperative infections.28,30 When stratifying by preexisting diagnosis of DM instead of average postoperative glucose measurements, patients with preexisting diagnoses of DM did have higher admission and postoperative blood glucose values in the current study. However, the current study did not find any significant differences in infectious or thromboembolic outcomes in patients with a preexisting diagnosis of DM. Similarly, differences in outcomes or complications were not seen when further breaking down the glucose cutoff groups into only patients without diabetes or only in patients with diabetes. Those with diabetes had a trend towards significantly more PE and VTE if their blood glucose was ≥180 mg/dL, but this was not statistically significant given the lack of power from additional breakdown of groups. Nonetheless, the regression analysis demonstrated that a glucose of ≥180 mg/dL was an independent risk factor for increased rate of PE and VTE in this cohort. While in the acute trauma setting there is not an opportunity to optimize A1c or DM management preoperatively, the current study demonstrates that blood glucose control in the acute perioperative setting, irrespective of a diagnosis of DM, plays an important role in preventing thromboembolic events.
The current study provides support for the importance of controlling blood glucose in the acute postoperative period following orthopaedic surgery for fixation of femoral fractures in geriatric patients to lower risk of adverse events such as VTE and PE. However, it is not without limitations. First, the study is a retrospective in nature and is limited to the data present in the medical record. Factors such as chronicity of hyperglycemia may be important to outcomes and were not formally assessed in this pilot study. Additionally, evaluation of outcomes such as nonunion which require follow up past 90 days may be limited by loss to follow up in a population notorious for limited follow up. However, the study was conducted as part of a large (>40 hospital) system with a unified electronic medical record which allows capture of reoperation and complications that occur in the system but outside of the hospital where the index operation was conducted. Other confounders related to this study include no record of preoperative anticoagulation, varied time to ambulation postoperatively, and lack of preoperative imaging evaluations for VTE or PE. However, given the large sample size and lack of clinical VTE or PE symptoms at this time of surgery, there was no clinical need for preoperative imaging. Lastly, more extensive research is needed to elucidate the negative effects of poor blood glucose control on long term outcomes (including functional outcomes) in this patient population. Yet, orthopaedic surgeons can use the results of this study to encourage strict control of acute postoperative blood glucose, understanding the possible deleterious effects of poor blood glucose control in elderly patients following operative fixation of femoral fractures.
Conclusions
Acute perioperative hyperglycemia, regardless of DM diagnosis, is an independent risk factor for DVT and PE following geriatric femur fracture fixation and represents a modifiable risk factor that has potential to improve outcomes.
Footnotes
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.