Logistic Regression Analysis of Risk Factors for Femoral head Osteonecrosis after Healed Intertrochanteric Fractures

Abstract

Purpose

To evaluate the potential risk factors of the development of femoral head osteonecrosis after healed intertrochanteric fractures.

Methods

We retrospectively reviewed all patients who were operated upon with closed reduction and internal fixation for intertrochanteric fractures by our medical group from December 1993 to December 2012. Patients with healed fractures were identified. Age, gender, comorbidities favouring osteonecrosis, causes of injuries, fracture patterns, the location of the primary fracture line, time from injury to surgery, fixation methods, and the development of femur head osteonecrosis of these patients were summarised. Univariate and multivariate logistic regression analysis were performed to evaluate the correlation between potential risk factors and the development of femoral head osteonecrosis.

Results

A total of 916 patients with healed intertrochanteric fractures were identified. Femoral head osteonecrosis was found in 8 cases (0.87%). According to the results of univariate logistic regression, a more proximal fracture line, fixation with dynamic hip screws and age were found to be statistically significant factors. The results of multivariate logistic regression analysis indicated that the statistically significant predictors of femoral head osteonecrosis were younger age (odds ratio [OR] = 17.103; 95% confidence interval [CI], 1.988-147.111), a more proximal fracture line (OR = 31.439; 95% CI, 3.700-267.119) and applying dynamic hip screw as the internal fixation (OR = 11.114; 95% CI, 2.064-59.854).

Conclusions

Regular follow-up is commended in young patients with a proximal fracture line who underwent closed reduction and internal fixation with dynamic hip screw, even though the bone had healed.

Keywords

Close reduction and internal fixation Femoral head osteonecrosis Intertrochanteric fractures Logistic models Risk factors

Introduction

Intertrochanteric fractures (ITFs) are most common among the elderly, particularly those with osteoporosis (1–2–3). The incidence of ITFs has increased significantly during recent decades, and is likely to rise substantially due to the continued ageing of the population (4). Although surgical management has been regarded as golden standard for most ITFs because of the poor outcome following conservative management (5), challenges are still frequently encountered as the risk of complications and mortality following these fractures is inevitable (5, 6).

Femoral head ostoenecrosis (FHO) is a pathologic process that results from interruption of blood supply to the femoral head (7). The primary source of blood supply to the femoral head is branches of the medial circumflex femoral arteries, which emerge on the superior aspect of the femoral neck and supply the capital epiphysis and head-neck region (8). These branches could be disrupted directly in case of femoral neck fractures, especially sub-capital cases. As such, traumatic FHO is common among the patients with femoral neck fractures, even if there is eventual healing of the bone (9). The abundant blood supply to the area between the greater and lesser trochanter, which is involved in ITFs, makes ITFs much less susceptible to FHO and nonunion than are femoral neck fractures. In the literature, the estimated incidence of FHO after ITFs has ranged between 0.13% and 2.46% (10). However, of these patients, many require operative intervention for pain relief and better function, which would increase the socioeconomic burden of these fractures (3, 11). Although several potential risk factors have been proposed including increasing age (12, 13), high energy trauma (13–14–15–16), a more proximal fracture line (13–14–15–16–17) and fixation methods (13–14–15), most of these risk factors remain controversial due to the low incidence of FHO after ITFs. Furthermore, there is few study that focused on evaluating the potential risk factors of FHO after healed ITFs. The purpose of current study is to evaluate the potential risk factors influencing the development of FHO after healed ITFs.

Methods

Institutional review board approval was granted for this investigation before medical record and radiographic review. We retrospectively reviewed all patients who were operated upon with closed reduction and internal fixation for ITFs by our medical group from December 1993 to December 2012. Inclusion criterion was ITFs healed within 9 months after the surgery. Exclusion criteria included open fractures, pathological fractures, previous surgery on the ipsilateral hip or femur, failure to complete 2 years of follow-up, suffering from complications associated with fractures (including infection, fixation failure, nonunion), and patients with incomplete medical record.

Age, gender, comorbidities favouring osteonecrosis, causes of injuries, time from injury to surgery and fixation methods were obtained from the medical records. Fracture patterns and fracture line location were evaluated based on plain x-ray and/or CT obtained before surgery. The fracture line passed higher than the classical fracture line of ITFs and located near the femoral neck or at the base of the femoral neck was defined as a proximal fracture line (13–14–15). Comorbidities favouring osteonecrosis included venous thromboembolism, diabetes, hypercholesterolaemia, hypertriglyceridaemia, steroid management, alcohol abuse, and smoking. The diagnosis of venous thromboembolism was made by general surgeons according to the result of Doppler ultrasound of the vein of the lower limbs. The diagnosis of diabetes, hypercholesterolaemia and hypertriglyceridaemia were made by endocrinologists according to the results of clinical laboratory tests and the diagnostic criteria devised by Chinese Medical Association. Slipping and falls from a normal height were classified as low energy trauma. Falls from a height greater than 2 metres, motor vehicle accidents, motorcycle accidents, automobile pedestrian encounters, and industrial crush injuries were classified as high energy trauma. After discharge, patients were followed up monthly to union of the fractures and annually thereafter. Patients with persistent hip pain were followed up monthly until pain relief was achieved or definite diagnosis was made. Radiological evaluation was done every visit. Diagnosis of FNO was based on the medical history, clinical evaluation, and plain x-ray. CT and MRI were employed to confirm the diagnosis of FNO when it was necessary.

Data was analysed using the Statistical Package for Social Sciences version 19.0 (SPSS, Inc., Chicago, IL). Kolmogorov-Smirnov test was performed to evaluate the distribution of continuous data. Mean ± standard deviation, median and range, and number of patients or percentage were employed to describe variables as appropriate. Univariate and multivariate binary logistic regression analysis were performed to evaluate the correlation between potential risk factors and FHO developing. A P value less than 0.05 (two-tailed) was considered statistically significant.

Results

General clinical characteristics

Records of 1,662 patients who were operated upon for ITFs from December 1993 to December 2012 were reviewed retrospectively. Open fractures were seen in 24 patients, pathological fractures were seen in 6 patients, previous surgery on the ipsilateral hip or femur was seen in 14 patients, and incomplete medical records were seen in 132 patients. During the follow-up, 495 patients were lost to follow-up or passed away within 2 years after surgery, and 75 patients suffered complications including infection, implant migration, fixation failure, nonunion, and delayed union. Therefore, the above-mentioned 746 patients were excluded (Fig. 1), and the remaining 916 patients were evaluated finally.

Fig. 1

Flow chart of patients selecting intertrochanteric fractures (ITFs).

The median of follow-up period was 6 years (range 2-20 years). The basic data of patients are shown in Table I. 8 patients (0.87%) were found to have developed FHO during the follow-up. The basic data of these patients was shown in Table I.

TABLE I

Basic data of 916 patients included

	All patients		FHO
	All patients		No	Yes
No. of patients		916	908	8
Follow-up period (years)
	Median	6	6	7
	Range	2-20	2-20	2-9
Age (years)
	Median	72	72	51
	Range	14-100	18-100	14-60
Gender
	Female	430	429	1
	Male	486	479	7
Comorbidities
	None	770	765	5
	One or more	146	143	3
Causes of injuries
	Low-energy trauma	713	707	6
	High-energy trauma	203	201	2
Fracture patterns
Stable fractures		344	340	4
	31A1	344	340	4
Unstable fractures		572	568	4
	31A2	552	548	4
	31A3	20	20	0
Fracture line location
Proximal	Proximal	165	158	7
Others		751	750	1
	Typical	731	730	1
	Distal	20	20	0
Time from injury to surgery (days)
	Median	4	4	3
	Range	0-10	0-10	2-7
Fixation methods
Dynamic hip screw	Dynamic hip screw	207	201	6
Others		709	707	2
	PFN	113	113	0
	Gamma nail	304	303	1
	InterTAN	202	202	0
	Ender nail	1	1	0
	Cannulated screws	1	0	1
	Femoral LISS	42	42	0
	Polyax	46	46	0

FHO = femoral head osteonecrosis; PFN = proximal femoral nail (Synthes); LISS = less invasive stabilization system (Synthes).

Univariate binary logistic regression analysis

All the potential risk factors, including age, gender, comorbidities, causes of injuries, fracture patterns, fracture line location, time from injury to surgery and fixation methods, were considered as covariates, respectively. The development of FHO was considered as dependent variable. The results of univariate binary logistic regression analysis were shown in Table II.

TABLE II

Results of univariate binary logistic regression analysis

Variable	p value	OR	95% CI
Variable	p value	OR	Lower	Upper
Age (reference: ≥60 years)	0.012*	14.917	1.827	121.808
Gender (reference: female)	0.087	6.269	0.768	51.162
Comorbidities (reference: without any comorbidities)	0.113	3.210	0.759	13.581
Causes of injuries (reference: low-energy trauma)	0.846	1.172	0.235	5.854
Fracture patterns (reference: stable fractures)	0.470	0.599	0.149	2.409
Fracture line location (reference: typical and distal)	0.001*	33.228	4.060	271.970
Time from injury to surgery (reference: >4 days)	0.146	4.749	0.582	38.758
Fixation methods (reference: methods besides DHS)	0.004*	10.552	2.114	52.683

DHS = dynamic hip screws; OR = odds ratio; CI = confidence interval.

p<0.05.

According to age, patients were divided into 2 groups: 297 cases younger than 60 years and 619 cases not younger than 60 years. There were 7 FHO cases in the former group (7/297, 2.36%) and 1 in the latter group (1/619, 0.16%). The result of univariate binary logistic regression demonstrated that the risk of developing FHO increased steadily with the decreasing age (odds ratio [OR] = 14.917; 95% confidence interval [CI],1.827-121.808), and the difference was statistically significant (p = 0.012). The result suggested that the younger patients had a higher incidence of FHO developing after healed ITFs. Younger age could increase the risk of FHO developing after healed ITFs.

According to location of the primary fracture line, patients were divided into 2 groups: 165 cases with a more proximal fracture line and the other 751 cases. There were 7 FHO cases in the former group (7/165, 4.24%) and 1 case in the latter group (1/751, 0.13%). The result of univariate binary logistic regression demonstrated that the risk of developing FHO in the former group was 33.228 times more than the latter group (OR = 33.228; 95% CI, 4.060-271.970), and the difference was statistically significant (p = 0.001). The result suggested that the fractures with a more proximal fracture line had a higher incidence of FHO developing after healed ITFs. The proximal fracture line could increase the risk of FHO developing after healed ITFs.

According to fixation methods, patients were divided into 2 groups: 207 cases whose internal fixation after reduction was achieved with dynamic hip screw and the other 709 cases. There were 6 FHO cases in the former group (6/207, 2.90%) and 2 cases in the latter group (2/709, 0.28%). The result of univariate binary logistic regression demonstrated that the risk of developing FHO in the former group was 10.552 times more than the latter group (OR = 10.552, 95% CI, 2.114-52.683), and the difference was statistically significant (p = 0.004). The result suggested that the fractures with a fixation with dynamic hip screw had a higher incidence of FHO developing after healed ITFs. The dynamic hip screw applying could increase the risk of FHO developing after healed ITFs.

There were no statistically significant differences for FHO developing in ITFs divided by gender (p = 0.087), comorbidities (p = 0.113), causes of injuries (p = 0.846), fractures patterns (p = 0.470) and time from injury to surgery (p = 0.146).

Multivariate binary logistic regression analysis

The potential risk factors that had a statistically significant correlation with FHO developing according to the results of univariate binary logistic regression analysis, including age, fracture line location and fixation methods, were considered as covariates. The development of FHO was considered as dependent variable. Multivariate binary logistic regression analysis was performed to evaluated the correlation between covariates and dependent variable.

The results of multivariate binary logistic regression analysis were shown in Table III. The results indicated that the significant predictors of FHO developing were younger age (p = 0.010; OR = 17.103; 95% CI, 1.988-147.111), a more proximal fracture line (p = 0.002; OR = 31.439; 95% CI, 3.700-267.119) and applying dynamic hip screw as the internal fixation (p = 0.005, OR = 11.114, 95% CI, 2.064-59.854).

TABLE III

Results of multivariate binary logistic regression analysis

Variable	p value	OR	95% CI
Variable	p value	OR	Lower	Upper
Age (reference: ≥ 60 years)	0.010*	17.103	1.988	147.111
Fracture line location (reference: typical and distal)	0.002*	31.439	3.700	267.119
Fixation method (reference: methods besides DHS)	0.005*	11.114	2.064	59.854

DHS = dynamic hip screws; OR = odds ratio; CI = confidence interval.

p<0.05.

Discussion

This study evaluated the correlation between age, gender, causes of injuries, comorbidities favouring osteonecrosis, fracture patterns, fracture line location, time from injury to surgery, fixation methods and the development of FHO after healed ITFs. We found that patients with younger age, proximal fractures line location and dynamic hip screw implanting were more likely to develop FHO after healed ITFs.

Compared with femoral neck fractures, FHO after ITFs is quite rare. The first case was reported by Taylor et al (18) in 1955 and the largest group of 12 cases was reported by Baixauli et al (17) in 1999. The estimated incidence of FHO after ITFs has ranged between 0.13% and 2.46% (10). Due to the low incidence, the risk factors of FHO after ITFs are controversial.

A more proximal fracture line, which locates near the femoral neck or at the base of the femoral neck, has been advocated as one of the potential risk factors for FHO after ITFs in the literature (13–14–15–16–17). In accordance with the literature, our results suggested that a more proximal fracture line was one of the significant predictor of the development of FHO after healed ITFs. The primary aetiology may be the disruption of blood supply to the femoral head. The typical ITFs are totally extracapsular fractures with a fracture line between the greater and lesser trochanters. The primary source for blood supply of femoral head is branches of the medial circumflex femoral arteries (8), which are usually spared in ITFs, as the typical fracture line is distal to them. However, anatomical variations in the medial circumflex femoral arteries are not rare (8). ITFs with a more proximal fracture line might disrupt the blood supply to the femoral head via these vessels. Furthermore, ITFs with a more proximal fracture line could be partially intracapsular fractures and lead to haemarthrosis. An increase in intracapsular pressure, which could diminish the blood supply to the femoral head through a tamponade effect, caused by haemathrosis, has been well accepted as a risk factor of FHO after femoral neck fractures (7). Therefore, ITFs with a more proximal fracture line might disrupt the blood supply to the femoral head directly and indirectly, and increase the risk of FHO.

Increasing age is another risk factor of FHO that has been extensively debated in the literature (12, 13). Various vessel diseases often affect the elasticity of vessels of elder patients. Therefore, they are less resistant to overstretching during fracture reduction manoeuvres, which may jeopardise the already precarious blood supply of femoral head (12, 13). Besides that, bone quality, which decreases with the advancing age, may be a risk factor of FHO. In a study on animals, Naito et al (19) reported that the vulnerable bone matrix plays a crucial role in the occurrence of FHO in stroke-prone spontaneously hypertensive rats (SHRSPs), whether based on generalised or localised osteopenia. However, a study, which evaluated the potential risk factors of FHO after femoral neck fractures, failed to find a correlation between age and FHO developing (20). Interestingly, instead of negative correlation, our results demonstrated that decreasing age had a positive correlation with FHO developing. The positive correlation might not reflect the exact correlation between age and FHO developing. All patients included in our study was followed up for at least 2 years, and 75.8% FHO was diagnosed within 2 years after fracture (10). Nevertheless, most elder patients with ITFs do not seek for further medical attention and are often lost to follow-up after 2 years, especially after fracture union. Therefore, the shortness of mid- and long-term evaluation may account for the young age of patients with FHO in our study.

A lot of attention has been drawn to the influence of fixation methods on the vascularity of femoral head as well (13–14–15). Our results suggested that fixation with dynamic hip screw was one of the significant predictor of the development of FHO after healed ITFs. The intraoperative disruption of blood supply to the femoral head appears to be a possible aetiology. In 1983, an intravital study by Strömqvist et al (21) showed that postoperative blood supply of femoral head might be influenced by the operative procedure. The potential adverse effects of reaming, which are involved before dynamic hip screws implanting, include greater bone loss, heating of cancellous bone, and an application of rotational forces to the femoral head (14, 17). Each of these factors may influence the vascularity of the femoral head (14).

Some researchers have related FHO after ITFs to high-energy trauma and fracture patterns (13–14–15–16). They postulated that high-energy trauma could make fractures more comminute and more unstable, and increase the incidence of displacement of fractures, thereby compromising the blood supply at the time of fracture. However, we could not find any correlation between them and the development of FHO. Time from injury to surgery has been related to FHO after femoral neck fractures (22). Nevertheless, we could not find any correlation between it and the development of FHO. Several other risk factors have been related to non-traumatic FHO, including include venous thromboembolism, diabetes, hypercholesterolaemia, hypertriglyceridaemia, steroid management, alcohol abuse, smoking (7). Although some of these comorbidities were present in some FHO cases in our study, we could not find any correlation between them and the development of FHO.

The major limitation of the current study was that the sample size of the current study could not provide adequate power to detect a significant difference of gender and time from injury to surgery (63% and 64%, respectively). Therefore, it may be inappropriate to exclude these 2 factors from risk factors. Further study with a larger sample size is still needed to evaluate the correlation between the development of FNH after healed ITFs and gender and time from injury to surgery.

Generally, our results indicated that the significant predictors of FHO developing were a more proximal fracture line and applying dynamic hip screw as internal fixation. Elder age was a statistically significant protector of FHO. Regular follow-up is commended in young patients with a proximal fracture line who underwent closed reduction and internal fixation with dynamic hip screw, even though the bone had healed.

Footnotes

Financial support: The study was supported by the National Natural Science Foundation of China (Grant no. 81272004).

Conflict of interest: None.

References

Rosengren

Ahlborg

Gärdsell

et al.

Bone mineral density and incidence of hip fracture in Swedish urban and rural women 1987-2002.

Acta Orthop. 2010;81(4):453–459.

van Zeeland

Genovesi

Mulder

Strating

Glas

Engel

POSSUM predicts hospital mortality and long-term survival in patients with hip fractures.

J Trauma. 2011;70(4):E67–E72.

Johnell

Kanis

An estimate of the worldwide prevalence and disability associated with osteoporotic fractures.

Osteoporos Int. 2006;17(12):1726–1733.

Verettas

Ifantidis

Chatzipapas

et al.

Systematic effects of surgical treatment of hip fractures: gliding screw-plating vs intramedullary nailing.

Injury. 2010;41(3):279–284.

Kaplan

Miyamoto

Levine

Egol

Zuckerman

Surgical management of hip fractures: an evidence-based review of the literature.

II: intertrochanteric fractures. J Am Acad Orthop Surg. 2008;16(11):665–673.

Burgers

Van Lieshout

Verhelst

Dawson

de Rijcke

Implementing a clinical pathway for hip fractures; effects on hospital length of stay and complication rates in five hundred and twenty six patients.

Int Orthop. 2014;38(5):1045–1050.

Babis

Sakellariou

Parvizi

Soucacos

Osteonecrosis of the femoral head.

Orthopedics. 2011;34(1):39–48.

Gautier

Ganz

Krügel

Gill

Ganz

Anatomy of the medial femoral circumflex artery and its surgical implications.

J Bone Joint Surg Br. 2000;82(5):679–683.

Davidovitch

Jordan

Egol

Vrahas

Challenges in the treatment of femoral neck fractures in the nonelderly adult.

J Trauma. 2010;68(1):236–242.

10.

Barquet

Mayora

Guimaraes

Suárez

Giannoudis

Avascular necrosis of the femoral head following trochanteric fractures in adults: a systematic review.

Injury. 2014;45(12):1848–1858.

11.

Johnell

The socioeconomic burden of fractures: today and in the 21^st century.

Am J Med. 1997;103(2A):20S–25S.

12.

Mallina

Dinah

Avascular necrosis of femoral head: a rare complication of a common fracture in an octogenarian.

Geriatr Orthop Surg Rehabil. 2013;4(3):74–77.

13.

Bartonícek

Fric

Skála-Rosenbaum

Dousa

Avascular necrosis of the femoral head in pertrochanteric fractures: a report of 8 cases and a review of the literature.

J Orthop Trauma. 2007;21(4):229–236.

14.

Shih

Chen

Avascular necrosis of the femoral head—an unusual complication of an intertrochanteric fracture.

J Orthop Trauma. 1992;6(3):382–385.

15.

Chen

Chiu

Avascular necrosis of femoral head after gamma-nailing for unstable intertrochanteric fractures.

Arch Orthop Trauma Surg. 2001;121(9):505–507.

16.

Mattan

Dimant

Mosheiff

Peyser

Mendelson

Liebergall

Avascular necrosis and related complications following healed osteoporotic intertrochanteric fractures.

Isr Med Assoc J. 2002;4(6):434–437.

17.

Baixauli

Jr Baixauli

Lozano

Avascular necrosis of the femoral head after intertrochanteric fractures.

J Orthop Trauma. 1999;13(2):134–137.

18.

Taylor

Neufeld

Nickel

Complications and failures in the operative treatment of intertrochanteric fractures of the femur.

J Bone Joint Surg Am. 1955;37(2):306–316.

19.

Naito

Ito

Sekine

et al.

Femoral head necrosis and osteopenia in stroke-prone spontaneously hypertensive rats (SHRSPs).

Bone. 1993;14(5):745–753.

20.

Min

Kim

Avascular necrosis of the femoral head after osteosynthesis of femoral neck fracture.

Orthopedics. 2011;34(5):349.

21.

Strömqvist

Femoral head vitality after intracapsular hip fracture.

490 cases studied by intravital tetracycline labeling and Tc-MDP radionuclide imaging. Acta Orthop Scand Suppl. 1983;200:1–71.

22.

Razik

Alexopoulos

El-Osta

et al.

Time to internal fixation of femoral neck fractures in patients under sixty years—does this matter in the development of osteonecrosis of femoral head?

Int Orthop. 2012;36(10):2127–2132.