Use of CT Vs. MRI for Diagnosis of Hip or Pelvic Fractures in Elderly Patients After Low Energy Trauma

Abstract

Background:

Elderly patients who fall account for more than two million emergency department visits each year. In 4-10 % of these patients, initial plain radiographs do not identify a hip or pelvis fracture later diagnosed with advanced imaging. No consensus exists about what type of imaging, CT or MRI, should be obtained in elderly patients with hip or pelvic pain after a low energy trauma. The purpose of this study was to determine whether CT or MRI is more likely to result in a definitive fracture diagnosis in elderly patients with hip or pelvic pain after low energy trauma.

Methods:

A retrospective chart review was conducted of all patients who presented to the ED at a single level one trauma center over a 4.5 year period. Inclusion criteria were age greater than fifty years old, presentation with hip and/ or pelvis pain due to a low energy trauma, and radiographic imaging including both plain radiographs and at least one pelvis MRI or CT.

Results:

Of the 218 patients who met inclusion criteria and had negative initial plain radiographs, CT or MRI later diagnosed a fracture in 69 patients (32%). Seventy eight patients underwent MRI (24 fractures, 32%), 132 underwent CT imaging (41 fractures, 31%), and eight had both CT and MRI (5 fractures, 63%). Patients who underwent CT spent less time in the ED on average (430 minutes) than those who underwent MRI, or MRI and CT (502 minutes and 620 minutes respectively). Patients who underwent CT were just as likely to be diagnosed with a fracture as those who underwent MRI (p= 0.002). We encountered no cases where CT imaging did not identify a fracture that was later identified on MRI. Fifty six patients (26%) had at least one contraindication to MRI.

Conclusions:

Our study suggests CT may be adequate to rule out hip and pelvic fractures in this patient population. CT may be preferable to MRI based on decreased time spent in the ED and the large percentage of elderly patients with contraindications to MRI.

Level of Evidence: III

Introduction

Hip and pelvic pain after low energy trauma is a common cause for emergency room visits. A hip or pelvic fracture can often be diagnosed with plain radiographs alone, but in 2-10% of patients, fractures cannot be identified on plain radiographs.¹ Physicians may further evaluate patients with advanced imaging such as CT or MRI, but disagreement persists about which imaging modality may be the best choice. In several small series of patients, MRI appears to be better than CT for detection of hip and pelvic fractures in elderly patients with low energy trauma. In a study of 13 patients, Lubovsky et al. reported that MRI more often correctly identified the location of pelvic and hip fractures when compared to pelvic CT.² Cabarrus et al. evaluated 64 patients with suspected hip fractures with both CT and MRI.³ MRI detected fractures in 99% of patients, while CT identified fractures in 53% of patients.³ Likewise, Haubro et al. reported greater sensitivity for detecting proximal femur fractures with MRI (100% vs 87% for CT), although this was not statistically significant.⁴ Verbeeten et al. found MRI specificity and sensitivity was near 100% for hip and pelvic fractures.⁵ MRI also offers the ability to identify other soft tissue problems that may be the source of pain, such as tumors, muscle tears, and hematoma.

Although pelvic MRI may be more sensitive than CT for fracture detection, MRI is also more expensive, more time consuming, and may not be available at all times at all institutions. Many elderly patients also have implants or medical devices that preclude MR imaging. In addition, the fractures identified by MRI may not result in changes in treatment. Of 69 fractures identified by MRI in one study, only 23 (33%) required operative intervention.⁶ Improvements in CT technology may also result in increased sensitivity compared to older studies.³

The primary goal of this study is to determine which secondary study (CT or MRI) is most likely to result in a fracture diagnosis in elderly patients presenting with hip or pelvic pain. Our secondary goal was to determine the ED length of stay based on type of study performed. The ultimate goal is to determine the most efficient manner to evaluate elderly patients with a potential hip or pelvic fracture but negative plain radiographs.

Materials and Methods

Under IRB approval we retrospectively reviewed the charts of all patients over age 50 years who presented to the Emergency Department at our Level 1 trauma center with hip pain after a low energy fall between 2009 and 2013. We identified these patients by searching order codes for radiographic studies performed. We included all patients who had negative plain radiographs (AP pelvis and AP and lateral hip) and a subsequent pelvic MRI or CT. We defined “negative plain radiographs” as the official radiology reading not reporting any fracture. In an effort to identify all patients who may have been diagnosed with a fracture at some point after their ED visit, we included any patient who had negative plain radiographs in the ED and pelvic MRI or CT within 30 days of their initial visit. We excluded any patients who suffered a high energy trauma (anything greater than fall from a standing height). We obtained demographic data such as patient age, sex, mechanism of injury, and premorbid living condition (home, assisted living, nursing home). When available in the records, we recorded if the patient was able to ambulate. We reviewed the radiology reports for any fracture diagnosed on CT or MRI. We also recorded the time spent in the ED for each patient and if the patient had a contradiction to MRI such as an implanted defibrillator.

We divided patients into 3 groups: those who underwent CT, those who underwent MRI, and those who had both studies. We compared the demographic and medical history information between the CT and MRI groups using t-test for all variables except age, which was compared using ANOVA. We calculated the percentage of patients in each group who were diagnosed with fractures, and compared the likelihood of diagnosing a fracture between the different groups. We used ANOVA to compare time spent ER between the three groups (CT, MRI, and CT and MRI).

We performed a second analysis comparing demographic and medical history between those patients with fractures and those without fractures using ANOVA for age and t-test for the remaining medical history variables.

Results

We identified 218 patients over age 50 who presented to the Emergency Department with hip pain after low energy trauma with negative plain radiographs and a subsequent hip or pelvis CT or MRI. Seventy-eight patients underwent MRI only, 132 patients underwent CT only, and eight patients had both MRI and CT. Nine patients had more than one visit for the same problem and underwent advanced imaging at the subsequent visit. These patients were excluded from the ED length of stay calculations, but included in all other analyses.

The CT only group was slightly older than the MRI only group (mean age 79 vs 76 years) (Table 1). No significant difference existed between the two groups with regard to the other demographic and clinical variables.

Table 1.

Baseline Characteristics by Use of Imaging Technology

	MRI only (n=76)	CT Only (n=131)	MRI + CT (n=8)	P-value
Age (mean (std dev))*	76 (12)	79 (12)	68 (10)	0.031
Male	22 (29%)	39 (30%)	1 (13%)	0.71
History of Osteoporosis	25 (33%)	52 (40%)	1 (13%)	0.22
Ambulatory Before	75 (100%)	120 (92%)	8 (100%)	0.019
Ambulatory Now	30 (42%)	56 (46%)	4 (50%)	0.76
History of Dementia	17 (23%)	53 (40%)	2 (25%)	0.026
History of Insufficiency Fx	9 (12%)	14 (11%)	1 (13%)	0.93
Living Situation:				0.33
Assisted Living	17 (22%)	29 (22%)	3 (38%)
Home	49 (64%)	70 (54%)	5 (63%)
Nursing Home	10 (13%)	28 (22%)	0 (0%)
Other	0 (0%)	3 (2%)	0 (0%)
Weight Bearing	34 (49%)	60 (52%)	4 (50%)	0.93

We found no significant difference in the likelihood of fracture diagnosis between the CT only and MRI only groups. The CT identified a fracture in 41 of 132 patients (31%) and MRI identified a fracture in 24 of 78 patients (32%). Of the eight patients who underwent both MRI and CT, five patients had a fracture (63%). No patient who had a negative CT later had a fracture diagnosed on MRI. Rather, in those five cases, surgeons requested a CT after the fracture diagnosis was made on MRI to better characterize the fracture pattern. Fractures identified included femoral neck fractures, intertrochanteric fractures, sacral fractures, greater trochanter fractures, isolated pubic rami fractures, acetabular fractures, and pelvic ring patterns (e.g. sacrum and pubic rami) (Table 2).

Table 2.

Fractures Identified on Advanced Imaging

Type of fracture	CT only	MRI only	CT	and MRI
Pelvic ring	9	4		0
Femoral neck	8	5		1
Greater trochanter	3	3		0
Sacral	9	0		1
Intertrochanteric	1	2		0
Pubic rami	5	4		1
Other proximal femur	2	4		0
Acetabular	2	1		2
Other	2	1		0

Of those patients who made multiple visits for the same problem within 30 days of the index visit, seven underwent A CT only (3 fractures) and one underwent MRI only (no fracture). One patient had a negative CT and MRI at the index visit, but a sacral fracture was identified on a second CT exam obtained 15 days later after a second fall.

Patients who went underwent MR imaging spent significantly more time in the ED than those who underwent CT only (502 minutes vs 430 minutes, p=<0.0001). Patients who underwent both tests spent an average of 630 minutes in the ED. Fifty-six patients (26%) had at least one contraindication to MRI.

We found no significant difference between patients with and without fracture with regard to age, sex, medical history, and premorbid living situation (Table 3). Ability to ambulate after the fall was not predictive of a fracture diagnosis. Patients who could bear weight were less likely to have a fracture (p=0.035), but ability to bear weight did not exclude a fracture diagnosis.

Table 3.

Baseline Characteristics for Patients With and Without Diagnosed Fractures

	No Fracture (n=147)	Fracture (n=70)	P-value
Age (mean (std dev))	77 (13)	78 (11)	0.46
Male	43 (29%)	19 (27%)	0.87
History of Osteoporosis	48 (33%)	31 (45%)	0.10
Ambulatory Before	138 (95%)	67 (96%)	1.00
Ambulatory Now	64 (47%)	26 (39%)	0.30
History of Dementia	50 (34%)	23 (33%)	0.89
History of Insufficiency Fx	16 (11%)	8 (11%)	1.00
Living Situation:			0.81
Assisted Living	34 (23%)	14 (21%)
Home	83 (57%)	43 (61%)
Nursing Home	26 (18%)	12 (17%)
Other	3 (2%)	0 (0%)
Weight Bearing	72 (56%)	26 (39%)	0.035

Discussion

Most previous studies examining these modalities reported MRI to be superior to CT for diagnosis of fracture. In contrast, our research suggests that CT and MRI are equally likely to diagnosis fractures in this patient population. Demonstrating which study (CT or MRI) is most likely to provide a diagnosis can increase the likelihood of providing a timely diagnosis and decrease unnecessary imaging studies.

Although previous studies found MRI to be more sensitive, most of these studies included a small number of patients and utilized older CT scanners.^2-6 Advances in CT scanner technology may improve the ability of modern CT to detect more subtle fractures. For example, Mallee et al. found comparable sensitivity, specificity, and accuracy when comparing MRI with 64 slice multidetector CT (MDCT).⁷ Heikal et al. reviewed 65 patients evaluated by MDCT for occult femoral neck fractures, and found no instances where a patient with a negative CT was later found to have a fracture.⁸

Of studies using current CT technology, our results nearly match those of Gill et al., who reported occult fracture rates of 36% (CT) and 38% (MRI) among 92 patients drawn from a similar patient population.⁹ As in our study, patients underwent either CT or MRI based on the ordering physician, and they could not directly compare sensitivities of the two studies. Our study includes a larger patient group and more clearly defines the patient population of interest (over age 50 years, low energy mechanism) which allows more direct application of this information to other clinical practices.

In a smaller study of 24 patients, Hakkarinen reported 4 cases where MRI detected fractures that were missed on CT.¹⁰ In contrast, our study found no instance where MRI identified a fracture missed on CT. One explanation for this discrepancy could be variation in radiologist experience and variability in what constitutes a fracture diagnosis on MRI or CT. Collin et al. reported variation in the number of hip MRI/CT images read as “equivocal” for fracture based on the experience and specialty of the radiologist.¹¹ Likewise, Haubro et al. found differences in ability to detect fracture by surgeons and radiologists with varying levels of experience.⁴ Several studies describe intertrochanteric fractures identified on MRI that were interpreted as isolated greater trochanter fractures on CT or A plain films.^4,9,12 The clinical significance of this injury pattern is not clear. For example, two of four patients in the Haubro study with greater trochanter fracture on CT and intertrochanteric extension on MRI, underwent surgical fixation while two other patients with identical fractures were managed nonoperatively. The authors could not identify the reason for variation in treatment. Our radiologists may have been less likely to interpret this pattern as a true fracture.

We also analyzed all radiographic studies obtained within 30 days of the index ED visit to ensure that we did not overlook fractures identified on later exam. We found nine patients with multiple visits during that time span, but none of the patients with a negative CT were later diagnosed with a fracture.

In an effort to describe patient factors that may increase the likelihood of occult fracture, we recorded demographic data, medical history, and ability to walk when available in the chart. We were unable to identify any specific patient characteristic that was associated with fracture. Like other studies, we found that patients were less likely to have a fracture if they could bear weight, but neither ability to walk nor ability to bear weight could exclude a fracture diagnosis.

Our study also provides new information on length of stay in the ED based on imaging choice. Our findings confirmed our anecdotal impression that patients who went underwent MRI imaging spent significantly more time in the ED than those who underwent CT (502 minutes vs 430 minutes) at our Level 1 trauma center. The length of stay differences would likely be exaggerated in smaller facilities where MRI is not readily available.

Limitations of our study include retrospective data collection with some information (such as ability to bear weight) being absent from some patient records. Because most patients had only CT or MRI, we can not declare with certainty that MRI would not have detected additional fractures in patients who underwent CT with a negative result. However, given that the two patient populations - those who underwent CT and those who underwent MRI- were demographically very similar, we would expect that the incidence of fracture would also be similar between the two groups. If MRI was truly significantly more sensitive to diagnose a fracture, we would expect to find a higher incidence of fracture diagnosis in the MRI only group, or multiple patients returning with displaced fractures after negative CT. Instead, the incidence of fractures was nearly identical.

Conclusion

Although previous studies suggest MRI has a superior sensitivity for fractures, our study found CT to be adequate to rule out hip and pelvic fractures in this patient population. CT may also be preferable to MRI based on decreased time spent in the ED and the large percentage of elderly patients with contraindications to MRI. Further prospective study is necessary to definitively determine the best and most efficient diagnostic tool in this patient population.