
==== Front
J Am Acad Orthop Surg Glob Res Rev
J Am Acad Orthop Surg Glob Res Rev
JAAOS Glob Res Rev
JAAOS Glob Res Rev
JAAOS Global Research & Reviews
2474-7661
Wolters Kluwer Philadelphia, PA

39236262
JAAOSGlobal-D-24-00214
10.5435/JAAOSGlobal-D-24-00214
00001
3
009
Research Article
A Scoring System for Predicting Nonunion After Intramedullary Nailing of Femoral Shaft Fractures
Kraus Kent R. MD kent.r.kraus@gmail.com

Flores Joshua W. MD joshuawallentinflores@gmail.com

Slaven James E. MS jslaven@iu.edu

Sharma Ishani MD ishani007@gmail.com

Arnold Payton K. MD pkarnoldmd@gmail.com

Mullis Brian H. MD bmullis@iupui.edu

https://orcid.org/0000-0002-4182-3244
Natoli Roman M. MD, PhD
From the Department of Orthopaedic Surgery (Dr. Kraus, Dr. Flores, Dr. Mullis, and Dr. Natoli), the Department of Biostatistics and Health Data Science (Dr. Slaven), Indiana University School of Medicine, Indianapolis, IN; the Indiana University School of Medicine, Indianapolis, IN (Dr. Sharma and Dr. Arnold); and the Indiana University Health Physicians, Indianapolis, IN (Dr. Mullis and Dr. Natoli).
Correspondence to Dr. Natoli: rnatoli@IUHealth.org
9 2024
04 9 2024
8 9 e24.0021413 6 2024
25 6 2024
Copyright © 2024 The Authors. Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Orthopaedic Surgeons.
2024
American Academy of Orthopaedic Surgeons
https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an open access article distributed under the terms of the Creative Commons Attribution-Non Commercial-No Derivatives License 4.0 (CCBY-NC-ND), where it is permissible to download and share the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Introduction:

Femoral shaft nonunion negatively affects patient quality of life. Although multiple risk factors have been identified for femoral shaft nonunion after intramedullary nail (IMN) fixation, there is no quantitative model for predicting nonunion.

Study description:

The study is a retrospective cohort study of patients with femoral shaft fractures treated at two level one trauma centers who were followed to fracture union or nonunion. Patient, injury, and surgical characteristics were analyzed to create a quantitative model for nonunion risk after intramedullary nailing.

Methods:

Eight hundred one patients aged 18 years and older with femoral shaft fractures treated with reamed, locked IMNs were identified. Risk factors including demographics, comorbidities, surgical variables, and injury-related characteristics were evaluated. Multivariate analysis was conducted, and several variables were included in a scoring system to predict nonunion risk.

Results:

The overall nonunion rate was 7.62% (61/801). Multivariate analysis showed significant association among pulmonary injury (odds ratio [OR] = 2.19, P = 0.022), open fracture (OR=2.36, P = 0.02), current smoking (OR=3.05, P < 0.001), postoperative infection (OR=12.1, P = 0.007), AO/OTA fracture pattern type A or B (OR=0.43, P = 0.014), and percent cortical contact obtained intraoperatively ≥25% (OR=0.41, P = 0.021) and nonunion. The scoring system created to quantitatively stratify nonunion risk showed that a score of 3 or more yielded an OR of 6.38 for nonunion (c-statistic = 0.693, P < 0.0001).

Conclusions:

Femoral shaft nonunion risk is quantifiable based on several independent injury, patient, and surgical factors. This scoring system is an additional tool for clinical decision making when caring for patients with femoral shaft fractures treated with IMNs.

OPEN-ACCESSTRUE
SDCT
Read-alongYES
==== Body
pmcFemoral shaft nonunion has a debilitating effect on patient quality of life.1 Multiple risk factors have been identified for femoral shaft fracture nonunion, but adequate prediction of which patients will develop femoral shaft fracture nonunion remains elusive.1234 The most common mode of treatment of femoral shaft fracture fixation is reamed, statically locked intramedullary nailing (IMN). The rate of femoral shaft nonunion after treatment with reamed, statically locked IMN ranges from 0% to 15% but is generally accepted to be in the range of 3% to 5%.3,56789

The AO/OTA system, commonly used to classify femoral shaft fractures, correlates with nonunion risk.3,5,10-12 Another injury characteristic that affects nonunion rates is the presence and type of open fracture.3,13 Reaming, postoperative weight-bearing status, direction of nail insertion, and initial treatment with external fixation are surgical factors that were previously investigated to determine their effect on union rates.3,5-7,9,14-16 Patient factors are increasingly recognized as possible contributors to nonunion, but there are little data on how these may affect femoral shaft union rates.4 Although some scoring systems have been proposed for the tibia17181920 and long bones,21 the only scoring system for the femoral shaft includes pediatric patients, plates, and bone defects in closed fractures.22 While certain risk factors have been shown to be prevalent in patients with femoral shaft fracture nonunion, no scoring system has been developed specifically for femoral shaft fractures treated with IMN that reliably predicts nonunion. This gap represents an area of research that could lead to patient-specific treatments and risk stratification protocols for patients at high risk of nonunion.

We hypothesized that a collection of patient comorbidities, injury characteristics, and surgical factors would be associated with femoral shaft fracture nonunion after IMN fixation. The goal of this study was to identify notable variables for femoral shaft fracture nonunion and create a quantitative model for predicting nonunion risk. We call the scoring system created the “FeNNR score”, short for Femur Nail Nonunion Risk score.

Methods

After institutional review board approval was granted, all patients with femoral shaft fractures at two urban level one trauma centers from 2014 to 2020 were identified using common procedural terminology (CPT) codes 27245, 27511, and 27506. To be inclusive in capturing femoral shaft fractures (ie, AO/OTA type 32 fractures23), we queried the 27245 and 27511 billing codes, in addition to 27506, because some femoral shaft fractures in the subtrochanteric region or closer to the knee may be coded as 27245 or 27511. Radiographs were reviewed for all cases to verify they were true AO/OTA type 32 fractures. Patients then underwent electronic medical record (EMR) review. Inclusions were patients with AO/OTA fracture classification of 32A, B, or C treated with an intramedullary nail, aged at least 18 years, and who completed at least 3 months of follow-up. Exclusion criteria were patients lost to follow-up, plate fixation, presence of nonunion at index procedure, pathologic fractures, cases with segmental bone defects treated with Masquelet technique, and presence of total hip arthroplasty (Figure 1).

Figure 1 Flowchart demonstrating inclusion of patients. 2469 patients with CPT codes 27245, 27506, and 27511 were initially identified. 1137 patients with femoral shaft fracture remained after radiographic screening to confirm AO/OTA type 32 fractures, with 801 meeting eligibility criteria. 61 patients with nonunion were identified and compared with 244 randomly selected patients with union in a 1:4 ratio.

Patients were assigned to the union cohort if the surgeon documented the fracture as healed based on clinical examination and postoperative radiographs and if no additional intervention was conducted to promote healing with a minimum of 3 months of follow-up after surgery (follow-up median 7 months, interquartile range [IQR] 4 to 14 months). Several studies support that femoral shaft union after intramedullary nailing occurs around 3 to 4 months.24-26 Patients were assigned to the nonunion cohort if their attending surgeon documented a diagnosis of nonunion or they underwent additional treatment or procedures to promote bone healing (follow-up median 17 months, IQR 11.5 to 24.5 months). This definition of nonunion is consistent with contemporary understanding,20 whereby a nonunion is defined as a fracture not expected to heal without additional intervention to promote bone healing. Furthermore, all surgeons included in this study have substantial trauma experience, including both acute femur fracture treatment with IMNs and treatment of femur nonunions. 59 of 61 patients with nonunion underwent an additional procedure for femoral shaft fracture nonunion. These nonunion procedures occurred at a median of 7 months (IQR 5 to 9.5 months) after the index procedure.

Fracture characteristics, surgical variables, laboratory values, trauma severity, and patient comorbidities were retrospectively collected through EMR review (Supplemental Tables S1 and S2, http://links.lww.com/JG9/A356 and http://links.lww.com/JG9/A357). Radiographic measurements were calibrated to the diameter of the IMN. Cortical fragment displacement was measured as the shortest distance from the cortex of the fragment to its anatomic position on the intact cortex. Fracture distraction was measured on AP and lateral radiographs in millimeters. The distance between the intact proximal segment and the intact distal segment on anterior, posterior, medial, and lateral cortices was measured for AO/OTA type A and B fractures. For AO/OTA type C fractures, distraction was measured as the distance from the intact proximal segment to the area of comminution added to the distance of the intact distal segment to the area of comminution to approximate the methodology for types A and B. The smallest measure (ie, the closest approximation of bone segments) of the four cortices was reported as the distraction. Cortical contact was determined as previously described.20 In brief, final fluoroscopic views or immediate postoperative radiographs, if taken, were used to determine cortical contact with 25% assigned to each cortex (anterior, posterior, medial, and lateral), with residual fracture gaps of >1 mm scored as 0% (Figure 2). A cortex was assigned to be in contact for AO/OTA type C fractures if the segmental piece achieved a fracture distraction of <1 mm with both the intact proximal and distal segments. Pulmonary injury was defined as pulmonary contusion, acute respiratory distress syndrome, hemopneumothorax or pneumothorax, or any other pulmonary pathology requiring intervention documented in the General Surgery Trauma service history and physical. Finally, postoperative infection was determined by the fracture-related infection (FRI) consensus criteria.27 All patients with infection met confirmatory FRI criteria and completed treatment for the FRI before determination of union status.

Figure 2 Immediate postoperative AP and lateral fluoroscopic images showing an AO type B femoral shaft fracture after intramedullary nailing. As an example of determining cortical contact, this case was assigned 50% cortical contact, 25% for the anterior cortex and 25% for the lateral cortex. The medial and posterior cortices had residual fracture gaps of >1 mm and were assigned 0% each.

Statistical Analysis

A control group of patients with femoral shaft fracture union was randomly selected from the healed fracture cohort to create a 4:1 union-to-nonunion ratio for statistical analysis.28 Bivariate analysis was conducted on all variables (Table 1 and Supplemental Tables 1 and 2, http://links.lww.com/JG9/A356 and http://links.lww.com/JG9/A357). The chi-squared analysis or Fisher exact test was used to analyze categorical variables, and the Wilcoxon rank-sum test was used for analysis of continuous variables. We selected 6 variables identified as statistically significant in bivariate analysis (P < 0.05) for multivariate logistic regression modeling to identify independent risk factors of nonunion (Table 2). The 6 variables were chosen (a) based on the number of nonunion events available to model, using the “rule-of-thumb” of 10 events per predictor,29 and (b) to reduce redundancy in predictors (eg, AO/OTA classification effectively accounts for medial comminution and Winquist classifications, and percent cortical contact effectively accounts for both cortical fragment displacement and fracture distraction). Owing to the small number of patients in the former smoker cohort and several Gustilo open fracture types, both were recategorized as binary for the multivariate analysis to allow for model convergence. Variance inflation factors for all variables were determined to assess multicollinearity, with a conservative value of <2 being considered appropriate. Given the totality of evidence from previous literature and all P-values in the multivariate analysis being <0.05, all 6 variables from the multivariate analysis were used to develop a scoring system (ie, FeNNR score) to quantitatively evaluate femur nonunion risk after IMN (Tables 3 and 4). Weights for each variable from the multivariate logistic regression model were obtained and scaled to integers to create the scoring system. Patients were then assigned the integer value for the variable if present, and the values were summed to create the final score for each case. The FeNNR score was then assessed using receiver operating characteristic (ROC) analysis to characterize its performance (Table 5).

Table 1 Select Risk Factors From the Bivariate Analysesa

	Nonunion (n = 61)	Union (n = 244)	P	
Age	43 (31, 53)	36 (26, 57)	0.1365	
Sex	
 Female	19 (31.2)	87 (35.7)	0.5084	
 Male	42 (68.9)	157 (64.3)		
ASA score	3 (2, 3)	3 (2, 3)	0.7316	
Body mass index (BMI)	28.9 (25.3, 33.1)	29.6 (25.0, 35.3)	0.7147	
Current smoker	36 (59.0)	71 (29.2)	<0.0001	
Current alcohol use	28 (45.9)	95 (39.1)	0.5552	
Diabetes	9 (14.8)	25 (10.3)	0.3170	
Fracture location	
 Proximal third	8 (13.1)	51 (20.9)	0.0875	
 Middle third	30 (49.2)	133 (54.5)		
 Distal third	23 (37.7)	60 (24.6)		
AO/OTA classification	
 A	18 (29.5)	100 (41.0)	0.0173	
 B	22 (36.1)	100 (41.0)		
 C	21 (34.4)	44 (18.0)		
Open fracture	21 (34.4)	37 (15.2)	0.0006	
Gustilo-Anderson classification	
 Closed fracture	40 (65.5)	207 (84.8)	0.0132	
 II	053	1 (0.4)		
 IIIA	19 (31.2)	32 (13.1)		
 IIIB	1 (1.6)	2 (0.8)		
 IIIC	1 (1.6)	2 (0.8)		
 Concomitant vascular injury	3 (4.9)	6 (2.5)	0.3903	
Pulmonary injury	23 (37.7)	56 (23.0)	0.0186	
Percent cortical contact	0% (0-0%)	0% (0-50%)	0.0011	
Postoperative infection	5 (8.2)	2 (0.8)	0.0042	
a Values are frequencies (percentages) for categorical variables and medians (interquartile range) for continuous variables. P-values in bold are variables that were significant (P < 0.05) on bivariate analysis and carried forward to multivariate analysis. All variables assessed by bivariate analysis are available in Supplemental Tables S1 and S2 (http://links.lww.com/JG9/A356 and http://links.lww.com/JG9/A357).

Table 2 Multivariate Logistic Regression Results of Variables From Bivariate Analysis Comparing Nonunion and Union Cohorts

Variable	OR (95% CI)	P a	
Pulmonary injury (yes)	2.19 (1.12, 4.28)	0.0221	
AO/OTA classification	
 A or B	0.43 (0.22, 0.84)		
 C	Reference	0.0140	
Open fracture (yes)	2.36 (1.14, 4.88)	0.0203	
Current smoker (yes)	3.05 (1.63, 5.74)	0.0005	
Percent cortical contact	
 0%	Reference		
 ≥ 25%	0.41 (0.19, 0.87)	0.0208	
Postoperative infection (yes)	12.05 (1.98, 73.48)	0.0070	
a Values are odds ratios (95% confidence intervals) with P-values from multivariate logistic regression models. Variance inflation factors for all variables are <2.

Table 3 Weights of Variables From Multivariate Analysis Based on the Logistic Regression Model

Variable	Weight From Logistic Regression	Value for Scoring System	
Postoperative infection	1.2446	3	
Current smoker	0.5581	1.5	
Pulmonary injury	0.3916	1	
AO/OTA type C fracture	0.4273	1	
Open fracture	0.4295	1	
0% cortical contact	0.4465	1	
Scoring system values were scaled to integers based on weight for ease of calculation. Patients are assigned the value for the variable if present.

Table 4 Femoral Shaft Nonunion Risk Score Frequency and the Number of Nonunions for Each Score

Scorea	Frequency (%)	Number of Nonunions (Cumulative %)	
0	46 (15.1)	0 (0)	
1	96 (31.5)	11 (18.0)	
2	92 (30.2)	17 (45.9)	
3	49 (16.1)	19 (77.1)	
4	16 (5.3)	10 (93.4)	
5	3 (1.0)	1 (95.1)	
6	3 (1.0)	3 (100)	
a As an example, there were 49 of 305 patients who scored a 3 (49/305 = 16.1%). Of those 49 patients, 19 developed a nonunion (19/49 = 38.8%). In other words, if a patient scored a 3, there is over a one-in-three chance they developed a nonunion. When considering all patients with scores ≥3, 33 of 71 patients (46.5%) developed a nonunion.

Table 5 Receiver Operating Characteristic36 Analysis of the Femur Nail Nonunion Risk (“FeNNR”) Score With Various Cut Points Based on the Integer Scoring System

	Sensitivity	Specificity	c-statistica	OR for Nonunion (95% CI)	
Cut point 1 (0 vs. 1+)	100	18.9	0.594	>999 (<0.01, >9990); P = 0.954	
Cut point 2 (0-1 vs. 2+)	82.0	53.7	0.678	5.27 (2.62, 10.61); P < 0.0001	
Cut point 3 (0-2 vs. 3+)	54.1	84.4	0.693	6.39 (3.47, 11.77); P < 0.0001	
Cut point 4 (0-3 vs. 4+)	23.0	96.7	0.598	8.79 (3.49, 22.12); P < 0.0001	
Cut point 5 (0-4 vs. 5+)	6.6	99.2	0.529	8.49 (1.52, 47.50); P = 0.015	
Cut point 6 (0-5 vs. 6+)	4.9	100	0.529	>999 (<0.01, >9990); P = 0.985	
Continuous	n/a	n/a	0.769	2.02 (1.63, 2.52); P < 0.0001	
a c-statistic values that round to or are >0.70 are given in bold. 0.70 is a standard level to identify an acceptable scoring system when analyzing outcomes dichotomously using a cut point.

Results

A total of 2469 patients were identified on CPT code search, and 801 patients met eligibility criteria. 61 patients with nonunion were identified, resulting in an overall nonunion rate of 7.62%. Patients with nonunion were then compared with a random selection of patients with union in a 1:4 ratio, yielding 244 patients in the union cohort and 61 patients in the nonunion cohort (Figure 1).

On bivariate analysis (Table 1 and Supplemental Tables 1 and 2, http://links.lww.com/JG9/A356 and http://links.lww.com/JG9/A357), statistically significant fracture/injury characteristics were as follows: presence of medial comminution (P = 0.0156), AO/OTA classification (P = 0.0173), Winquist classification (P = 0.0033), open fracture (P = 0.0006), Gustilo-Anderson classification (P = 0.0132), and concurrent pulmonary injury (P = 0.0186). Statistically significant surgical variables included percent cortical contact (P = 0.0011) and presence of fracture distraction after IMN (P = 0.0098). Current smoking (P < 0.0001) was found to be statistically significant among patient comorbidities. Finally, postoperative infection was significantly different between the union and nonunion cohorts (P = 0.0042).4,7,11,30 Hospital length of stay (LOS, P = 0.0414) and ICU LOS (P = 0.0289) were not included in the multivariate analysis, given the multitude of potential contributors unrelated to the femur fracture.

As described in Methods, 6 variables were included in the multivariate analysis. Gustilo-Anderson classification was not included because of the small number of patients in several categories and colinearity with open fractures. Pulmonary injury (P = 0.0221), AO/OTA classification (P = 0.0140), open fracture (P = 0.0203), current smoking (P = 0.0005), percent cortical contact (P = 0.0208), and postoperative infection (P = 0.0070) were found to remain independently statistically significant in the multivariate analysis (Table 2). All variance inflation factors for these 6 variables were <2.

These 6 variables were carried forward to create a predictive scoring system using weights from the multivariate logistic regression model to reflect their respective effect on nonunion risk. To ease clinical use of the FeNNR scoring system, the weights were scaled to integer values (Table 3). This scaling resulted in AO/OTA type C fractures, 0% cortical contact, open fractures, and pulmonary injury each receiving 1 point; current smoking receiving 1.5 points; and postoperative infection receiving 3 points. The final scoring system ranged from 0 to 6 across the 305 included patients, with nonunion rates increasing with higher scores (Table 4). ROC analysis optimizing for both sensitivity and specificity identified a sensitivity of 54.1%, specificity of 84.4%, and c-statistic of 0.693 when comparing scores ≥3 with 0 to 2, based on the Youden J-index (Table 5). At this cut point, the odds ratio (OR) for nonunion was 6.39 (95% CI 3.47 to 11.77, P < 0.0001). As a continuous measure, the scoring system had an OR of 2.02 (95% CI 1.63 to 2.52, P < 0.0001), with a c-statistic of 0.769.

Discussion

Interest has grown in predicting nonunion in long bone fractures because of the debilitating effects they have on patients' lives.1,2,21,31,32 Factors at the index procedure, radiographic measurements, and postoperative physical examination findings have been investigated in other long bone fractures to predict which patients will develop nonunion.17,19,20,33-35 The goal of predicting fracture nonunion is to help inform clinicians and patients about their prognosis and guide treatment decisions. To the authors' knowledge, although nonunion prediction systems have been established for tibial shaft fractures, subtrochanteric femur fractures, and lower extremity long bone fractures combining femurs and tibias with both plates and nails, no scoring system specifically for femoral shaft fractures after IMN has been developed.17,19-21,33-35

We created a scoring system to quantify femoral shaft fracture nonunion risk after IMN fixation. The scoring system we created had an optimal c-statistic of 0.693 at a cutoff point of 3 or more (Table 5). Although this is slightly lower than the standard accepted level of 0.70 to identify an acceptable scoring system when analyzed dichotomously using a cut point, it did achieve a c-statistic of >0.70 as a continuous measure.36 Therefore, we suggest that the model still holds clinical value for identifying patients at high risk of nonunion. The continuous ROC analysis should be interpreted as there is an approximately 2-fold increase in nonunion risk for each point increase in the scoring system. Of note, during the creation of the FeNNR scoring system, we considered the totality of the evidence from previous literature in addition to statistically significant factors identified in our data set. Ultimately, we also pared down the scoring system to binary choices, and values were scaled to integer scores for pragmatic use in clinical care. For example, before IMN fixation of a patient's femoral shaft fracture, several components of the scoring system are available. If a patient has an open femoral shaft fracture and a pulmonary injury, then the patient's preoperative score is 2. Our results show that patients with a score of 3 had a nonunion rate of 38.8% compared with an 18.5% rate for those with a score of 2 (Table 4). Therefore, given a score of 2 preoperatively, the surgeon should endeavor to obtain cortical contact during IMN fixation to enhance the patient's probability of fracture healing. Of note, the scoring system is nescient to a patient's temporal course up until nonunion is diagnosed. The scoring system simply assesses whether a factor(s) is(are) present at a given time. Therefore, it could be used at any point along the continuum of care whenever the information regarding a variable in the scoring system becomes available from the preinjury phase, to postoperatively in the hospital, to clinic follow-up.

Pulmonary injury, AO/OTA type C fracture, open fracture, current smoking, no cortical contact, and postoperative infection were shown to associate with femoral shaft fracture nonunion. Many studies have shown open fracture to be a risk factor of nonunion, and our results corroborate that finding.5,12,13,37 While open fracture is a nonmodifiable risk factor, orthopaedic surgeons must be prepared to care for them with early administration of prophylactic antibiotics and urgent, meticulous débridement and irrigation of the fracture site.38,39 Smoking is another nonmodifiable risk factor that our results found to markedly contribute to nonunion risk. Smoking is known to negatively affect orthopaedic surgery patients by decreasing blood flow and oxygenation of tissues, inhibiting immune responses, and altering bone physiology.40 Previous studies have shown that these physiologic changes can contribute to risk of fracture site infection and nonunion.3,5,37,41424344 Current evidence suggests that patients have improved surgical outcomes if a smoking cessation program is completed before surgery for elective operations or postoperatively for acute fracture surgery.45 Orthopaedic surgeons should emphasize the role of smoking cessation postoperatively.

Our results suggest that no cortical contact after IMN fixation plays a role in femoral shaft fracture union rates, reinforcing the need for cortical contact to optimize fracture healing and agreeing with other long bone nonunion prediction models.20,21,24,35 Multiple studies show that AO/OTA fracture classification is a notable risk factor of femoral shaft fracture nonunion.3,11,12,21 In AO/OTA type C fractures, direct cortical contact of the proximal and distal fracture fragments is not possible due to the fracture comminution, but, while maintaining overall length, our data suggest that it may be beneficial to have the intact proximal and distal segments close to the intercalary segment or adjacent comminuted fragments. While there are conflicting data on whether opening the fracture site increases the risk of nonunion,21,37,46 the benefits of obtaining cortical contact need to be considered against the risk of additional morbidity and infection, especially if the patient has multiple risk factors of nonunion at the time of IMN fixation. The most notable risk factor of nonunion was the development of postoperative infection. Postoperative infection is a commonly recognized cause for nonunion, and our results reinforce this.4,21,35,42 Patients who develop a FRI should be promptly cared for and followed closely for nonunion because infection scores 3 points and places patients at high risk of nonunion based on our scoring system.

Some previously recognized risk factors of femoral nonunion, such as nail diameter,16,47 weight bearing,5 and trauma,37 were not identified as statistically significant associations in our data. Nail diameters were comparable in both our union and nonunion cohorts, with similarly low variation in nail diameter to canal ratios that were greater than the 0.7 threshold shown to associate with nonunion,47 which likely mitigated this risk factor of healing in the patients studied. Regarding weight bearing, limitations were only prescribed if there was an ipsilateral lower extremity articular injury while previous literature showing an association with weight bearing limited weight bearing for 6 weeks postoperatively regardless of ipsilateral lower extremity injury.5 Indeed, more recent literature suggests that limiting weight bearing delays radiographic healing but does not increase nonunion rates.25 Finally, although trauma as the indication for nailing has been reported as a risk factor of nonunion compared with pathologic fracture (which we excluded),37 many studies report that ISS (eg, a measure of trauma) is not markedly associated with nonunion risk for IMN of femoral shaft fractures,3,5,48 which is consistent with our data.

This study has limitations. First, it is a retrospective review, so the data are limited to the information available in the EMR. Confounding variables could be present that were not collected or accounted for, although a multitude of variables spanning patient comorbidities, injury characteristics, and surgical factors were included in initial bivariate analysis (Table 1 and Supplemental Tables S1 and S2, http://links.lww.com/JG9/A356 and http://links.lww.com/JG9/A357). Multiple raters did not complete multiple measurements of the radiographs; therefore, we cannot comment on the interobserver and intraobserver reliability of the measurements. 298 patients also did not complete adequate follow-up for inclusion in the study. These patients accounted for approximately 26% of the original cohort, and their exclusion could have altered those variables found to associate with femoral shaft nonunion risk. Even with the large number of exclusions, we were still able to compare a random selection of healed patients in a 4:1 ratio with those with nonunions. Our final sample size of 801 patients is over three times larger than the average sample size of other femoral nonunion studies,4 and our number of nonunions is also comparatively large.3,5,14,16,22,49-51 However, despite this number of nonunions, we were still limited in the ability to stratify certain variables in the multivariate model (eg, open fracture type), and clearly, not all open fractures are the same. In addition, owing to the retrospective nature of our study, some potential factors that affect nonunion risk were not assessable, such as quantification of smoking or other forms of tobacco use and NSAID use. For smoking, we were only able to model current smoking versus not, due to statistical limitations. In addition, during the period investigated, it was predominantly local practice to recommend against NSAIDs for fracture cases. Other limitations due to the retrospective nature of the study include not being able to determine whether nonviable bone was discarded from some cases and whether reduction methods were closed, percutaneous, or open. Finally, when studying nonunion accurate determination of fracture healing is imperative. While there is no universally accepted definition of nonunion,52 our definition of nonunion was based on contemporary literature,20 and healing was determined by review of clinical notes and latest radiographs. However, despite this rigor, it remains possible that some patients in the union cohort later presented to an outside institution with nonunion.

Conclusions

Pulmonary injury, AO/OTA type C fracture, open fracture, current smoking, no cortical contact, and postoperative infection contribute to nonunion after IMN fixation of femoral shaft fractures. Although many of these factors are nonmodifiable, orthopaedic surgeons should endeavor to obtain a reduction with cortical contact, follow protocols to prevent surgical site infection, and counsel patients postoperatively on smoking cessation. Future studies should externally validate the FeNNR score in larger populations and could combine the score with biomarkers to create a stronger scoring system.

Supplementary Material

None of the following authors or any immediate family member has received anything of value from or has stock or stock options held in a commercial company or institution related directly or indirectly to the subject of this article: Dr. Kraus, Dr. Flores, Dr. Slaven, Dr. Sharma, Dr. Mullis, and Dr. Natoli.
==== Refs
References

1. Brinker MR Trivedi A O'Connor DP : Debilitating effects of femoral nonunion on health-related quality of life. J Orthop Trauma 2017;31 :e37-e42.27755332
2. Tay WH de Steiger R Richardson M Gruen R Balogh ZJ : Health outcomes of delayed union and nonunion of femoral and tibial shaft fractures. Injury 2014;45 :1653-1658.25062602
3. Metsemakers WJ Roels N Belmans A Reynders P Nijs S : Risk factors for nonunion after intramedullary nailing of femoral shaft fractures: Remaining controversies. Injury 2015;46 :1601-1607.26026201
4. Zura R Mehta S Della Rocca GJ Steen RG : Biological risk factors for nonunion of bone fracture. JBJS Rev 2016;4 :e5.
5. Taitsman LA Lynch JR Agel J Barei DP Nork SE : Risk factors for femoral nonunion after femoral shaft fracture. J Trauma 2009;67 :1389-1392.19704386
6. Canadian Orthopaedic Trauma Society: Nonunion following intramedullary nailing of the femur with and without reaming. Results of a multicenter randomized clinical trial. J Bone Joint Surg Am. 2003;85 :2093-2096.14630836
7. Ma YG Hu GL Hu W Liang F : Surgical factors contributing to nonunion in femoral shaft fracture following intramedullary nailing. Chin J Traumatol 2016;19 :109-112.27140219
8. Wolinsky PR McCarty E Shyr Y Johnson K : Reamed intramedullary nailing of the femur: 551 cases. J Trauma 1999;46 :392-399.10088839
9. Kuhn KM Ali A Boudreau JA Cannada LK Watson JT : Antegrade versus retrograde intramedullary nailing of proximal third femur fractures. J Surg Orthop Adv 2013;22 :263-269.24393183
10. Marsh JL Slongo TF Agel J : Fracture and dislocation classification compendium - 2007: Orthopaedic Trauma Association classification, database and outcomes committee. J Orthop Trauma 2007;21 :S1-S133.18277234
11. Noumi T Yokoyama K Ohtsuka H Nakamura K Itoman M : Intramedullary nailing for open fractures of the femoral shaft: Evaluation of contributing factors on deep infection and nonunion using multivariate analysis. Injury 2005;36 :1085-1093.16054148
12. Yoon YC Song HK Han JS Lee KC : Antegrade nailing in femoral shaft fracture patients - comparison of outcomes of isolated fractures, multiple fractures and severely injured patients. Injury 2021;52 :3068-3074.33563415
13. Watanabe Y Takenaka N Kobayashi M Matsushita T : Infra-isthmal fracture is a risk factor for nonunion after femoral nailing: A case-control study. J Orthop Sci 2013;18 :76-80.23053587
14. Giannoudis PV MacDonald DA Matthews SJ Smith RM Furlong AJ De Boer P : Nonunion of the femoral diaphysis. The influence of reaming and non-steroidal anti-inflammatory drugs. J Bone Joint Surg Br 2000;82 :655-658.10963160
15. Ostrum RF Agarwal A Lakatos R Poka A : Prospective comparison of retrograde and antegrade femoral intramedullary nailing. J Orthop Trauma 2000;14 :496-501.11083612
16. Serrano R Mir HR Gorman RA 2nd : Effect of nail size, insertion, and delta canal-nail on the development of a nonunion after intramedullary nailing of femoral shaft fractures. J Orthop Trauma 2019;33 :559-563.31464856
17. O'Hara NN Slobogean GP O'Halloran K : Predicting tibia shaft nonunions at initial fixation: An external validation of the Nonunion Risk Determination (NURD) score in the SPRINT trial data. Injury 2020;51 :2302-2308.32622626
18. Ross KA O'Halloran K Castillo RC : Prediction of tibial nonunion at the 6-week time point. Injury 2018;49 :2075-2082.30172349
19. Jang Y Gaski G Natoli R Virkus W McKinley T : Tibial fracture healing score: A novel tool to predict tibial nonunion. Orthopedics 2020;43 :e323-e328.32501520
20. O'Halloran K Coale M Costales T : Will my tibial fracture heal? Predicting nonunion at the time of definitive fixation based on commonly available variables. Clin Orthop Relat Res 2016;474 :1385-1395.27125823
21. Santolini E West RM Giannoudis PV : Leeds-genoa non-union index: A clinical tool for asessing the need for early intervention after long bone fracture fixation. Int Orthop 2020;44 :161-172.31440889
22. Li W Wang Y Zhou S : Development and validation of predictive nomogram for postoperative non-union of closed femoral shaft fracture. Sci Rep 2024;14 :3543.38347044
23. Meinberg EG Agel J Roberts CS Karam MD Kellam JF : Fracture and dislocation classification compendium-2018. J Orthop Trauma 2018;32 suppl 1 :S1-S170.
24. Chang PS Bechtold D Kazarian GS : Small residual fracture gaps delay time to union in length stable femur fractures treated with intrameduallary fixation. Injury 2023;54 :687-693.36402583
25. Flanagan CD Joseph NM Copp J Romeo N Alfonso N Hirschfeld A : Weight-bearing status may influence rates of radiographic healing following reamed, intramedullary fixation of diaphyseal femur fractures. OTA Int 2021;4 :e154.34765904
26. Tornetta P 3rd Tiburzi D : The treatment of femoral shaft fractures using intramedullary interlocked nails with and without intramedullary reaming: A preliminary report. J Orthop Trauma 1997;11 :89-92.9057141
27. Metsemakers WJ Morgenstern M McNally MA : Fracture-related infection: A consensus on definition from an international expert group. Injury 2018;49 :505-510.28867644
28. Linden A Samuels SJ : Using balance statistics to determine the optimal number of controls in matching studies. J Eval Clin Pract 2013;19 :968-975.23910956
29. van Smeden M de Groot JA Moons KG : No rationale for 1 variable per 10 events criterion for binary logistic regression analysis. BMC Med Res Methodol 2016;16 :163.27881078
30. Wu KJ Li SH Yeh KT : The risk factors of nonunion after intramedullary nailing fixation of femur shaft fracture in middle age patients. Medicine 2019;98 :e16559.31335740
31. Brinker MR Hanus BD Sen M O'Connor DP : The devastating effects of tibial nonunion on health-related quality of life. J Bone Joint Surg Am 2013;95 :2170-2176.24352770
32. Brinker MR Loftis CM Khoriaty JD Dunn WR : The devastating effects of humeral nonunion on health-related quality of life. J Shoulder Elbow Surg 2022;31 :2578-2585.35718254
33. Oliver WM Carter TH Graham C : A prospective randomised controlled trial of operative versus non-operative management of fractures of the humeral diaphysis: The HUmeral shaft fracture FIXation (HU-fix) study protocol. Trials 2019;20 :475.31383027
34. Driesman AS Fisher N Karia R Konda S Egol KA : Fracture site mobility at 6 weeks after humeral shaft fracture predicts nonunion without surgery. J Orthop Trauma 2017;31 :657-662.28708781
35. Panteli M Vun JSH West RM Howard AJ Pountos I Giannoudis PV : Development and validation of a post-operative non-union risk score for subtrochanteric femur fractures. J Clin Med 2021;10 :5632.34884334
36. Unal I : Defining an optimal cut-point value in ROC analysis: An alternative approach. Comput Math Methods Med 2017;2017 :3762651.28642804
37. Malik MH Harwood P Diggle P Khan SA : Factors affecting rates of infection and nonunion in intramedullary nailing. J Bone Joint Surg Br 2004;86 :556-560.15174553
38. Foote CJ Tornetta P 3rd Reito A , GOLIATH Investigators: A reevaluation of the risk of infection based on time to debridement in open fractures: Results of the GOLIATH meta-analysis of observational studies and limited trial data. J Bone Joint Surg Am 2021;103 :265-273.33298796
39. Atwan Y Miclau T Schemitsch EH Teague D : Antibiotic utilization in open fractures. OTA Int 2020;3 :e071.33937690
40. Kainiemi K Malmivaara A Sillman-Tetri S : Smoking cessation intervention prior to orthopedic surgery: A study protocol to determine patient outcomes and feasibility. Tob Prev Cessat 2022;8 :33.36118614
41. Panteli M Vun JSH West RM Howard A Pountos I Giannoudis PV : Surgical site infection following intramedullary nailing of subtrochanteric femoral fractures. J Clin Med 2021;10 :3331.34362123
42. Thomas JD Kehoe JL : Bone Nonunion. Treasure Island (FL), StatPearls, 2022.
43. Xu B Anderson DB Park ES Chen L Lee JH : The influence of smoking and alcohol on bone healing: Systematic review and meta-analysis of non-pathological fractures. EClinicalMedicine 2021;42 :101179.34765953
44. Smolle MA Leitner L Bohler N Seibert FJ Glehr M Leithner A : Fracture, nonunion and postoperative infection risk in the smoking orthopaedic patient: A systematic review and meta-analysis. EFORT Open Rev 2021;6 :1006-1019.34909221
45. Nasell H Adami J Samnegard E Tonnesen H Ponzer S : Effect of smoking cessation intervention on results of acute fracture surgery: A randomized controlled trial. J Bone Joint Surg Am 2010;92 :1335-1342.20516308
46. Choi JY Sung YB Yoo JH Chung SJ : Factors affecting time to bony union of femoral subtrochanteric fractures treated with intramedullary devices. Hip Pelvis 2014;26 :107-114.27536567
47. Millar MJ Wilkinson A Navarre P : Nail fit: Does nail diameter to canal ratio predict the need for exchange nailing in the setting of aseptic, hypertrophic femoral nonunions?. J Orthop Trauma 2018;32 :245-250.29401087
48. Eliezer EN Haonga BT Morshed S Shearer DW : Predictors of reoperation for adult femoral shaft fractures managed operatively in a sub-saharan country. J Bone Joint Surg Am 2017;99 :388-395.28244909
49. Yang S Yang Y Huo Y : Effect of the degree of displacement of the third fragment on healing of femoral shaft fracture treated by intramedullary nailing. J Orthop Surg Res 2022;17 :380.35962386
50. Deliberato D Myers DM Schnee B : Weight bearing as tolerated after intramedullary nailing of the femur: A retrospective analysis of clinical and radiographic outcomes. Orthopedics 2022;45 :e148-e153.35021030
51. Lin CY Chen CY Lin KC Chou YJ Chang WN : “Fragment width ratio” as a predictor of nonunion for femoral shaft fracture with third fragments. Orthopedics 2023;46 :169-174.37018623
52. Wittauer M Burch MA McNally M : Definition of long-bone nonunion: A scoping review of prospective clinical trials to evaluate current practice. Injury 2021;52 :3200-3205.34531088
53. Oliver GDWJ Friesen KB Anz AW Dugas JR Andrews JR Washington JK et al. : The effects of a pre-throwing program on collegiate NCAA division I softball pitchers' biomechanical measures of hip and shoulder range of motion. J Orthop Ther 2018:181.
