During the 90-month period, a total of 24 patients (14 male and 10 female; age range, 16-64 years; mean age, 42 years) who had sustained a knee dislocation were identified (Table 1). Using the assumptions described in Materials and Methods, we estimated the annual incidence of knee dislocations in adults at 2.3 per million. Patients were included in the study if they had dislocation on radiographs (n = 17) or if they had a well-documented description of the dislocation in their patient files (n = 7). Patients younger than 16 years were not included because they are not treated at our hospital. Twenty-two patients underwent surgery. One patient died before the knee surgery, and one patient's knee was not operated on because of extensive muscle necrosis due to vascular injury and compartment syndrome. Nineteen patients had an MRI examination of diagnostic quality before surgery (mean, 5 days; range, 0-26 days). Patient mean BMI was 28 (range, 19-42).

Of the 24 patients, seven had knee dislocation caused by high-energy trauma such as a motor vehicle or motorcycle accident. Two of these high-energy trauma patients were overweight (BMI, 25-29.9) and five had a normal weight (BMI, 18.5-24.9). Seven patients had knee dislocation due to athletic injury. Five of these patients had normal weight, one was overweight, and one patient was obese (BMI, 30-39.9). Ten patients had dislocation due to a simple fall. Of these patients, one had normal weight, three were overweight, four were obese, and two were morbidly obese (BMI > 40) (Table 2).

A BMI of less than 25 is considered normal [16]. Eleven patients (five women, six men; age range, 27-64 years; mean age, 49 years) had a BMI greater than 25 (mean, 33; range, 27-42) and had knee dislocation due to low-energy trauma (nine due to a simple fall, two due to downhill skiing). These patients were of special interest in this study and are referred to as group A. The other 13 patients—patients with dislocation due to high-energy trauma (n = 7) and normal-weight patients with dislocation due to low-energy trauma (n = 6)—are referred to as group B (mean BMI, 22; BMI range, 19-28; five female; eight male; age range, 16-62 years; mean age, 36 years). The annual incidence of patients having a BMI greater than 25 and had knee dislocation due to low-energy trauma is about 1.0 per million.

The subjective quality of all MRI studies was classified as diagnostic by both readers. In the reviewed original reports, there was no mention of nondiagnostic image quality. In every reviewed report, a radiologist was able to answer the questions that the clinician had asked in the request.

Operative findings revealed that all 22 patients had a complete tear of the ACL. All patients in group A had also a complete tear of the PCL. In group B, one patient had an intact PCL and all others had a complete tear of the PCL. Complete tear of the MCL was present in seven and five patients in groups A and B, respectively. A complete tear of the LCL was found in five patients in both groups.

Complete popliteal tendon tears occurred in five patients, all of whom were in group B (p = 0.0262). One patient from group B had a complete rupture of the biceps femoris tendon. There was no complete tear of the quadriceps femoris tendon or patellar tendon in any patient. Complete rupture of the medial retinaculum was identified on MRI examinations in 10 patients.

MRI evaluation of the ligaments and tendons resulted in one false-negative and one false-positive diagnosis of a complete tear in group A (both LCLs). In group B, there were one false-negative (ACL) and two false-positive (one MCL and one LCL) diagnoses of complete tears. Operative findings are shown in Table 3.

Bone, cartilage, and meniscal injuries— Ten avulsion fractures and one tibial plateau fracture were seen in eight patients. Two of these fractures were in group A and nine in group B. Six avulsion fractures of the intercondylar eminence of the tibia (four PCL and two ACL) and four avulsion fractures of the proximal fibula (all LCL) were seen. One patient's lateral tibial plateau was fractured. MRI evaluation of the fractures resulted in one false-positive diagnosis of avulsion fracture of the tibia.

Chondral injuries were classified into four grades according to the evaluation system of the International Cartilage Repair Society [17]. Nine grade I, two grade II, and one grade III injuries were observed at surgery. Of these 12 chondral injuries, eight were found in patients in group A. MR images were evaluated for the presence of an osteochondral fracture or a chondral loose fragment, but none was found.

Patients in group A had eight meniscal tears; group B patients had four meniscal tears. Of the meniscal tears, four were medial and eight were lateral. Three meniscal tears were radial, one was horizontal, and four were vertical. One bucket-handle tear was seen, and three menisci had ruptured with a displaced fragment. MRI evaluation resulted in four false-positive diagnoses of meniscal tears (one vertical, two radial, and one with a displaced fragment). False-positive MRI diagnoses were equally divided between the two groups (Figs. 1A, and 1B).

No statistically significant differences in bone, cartilage, or meniscal injuries between the two groups were found.

Neurovascular injuries—Three patients had signs of complete peroneal nerve injury after knee dislocation. Early nerve reconstruction surgery was performed in one patient. None of these three patients has regained peroneal function after 1 year of follow-up. All three patients belonged to group B.

Four patients had signs of partial peroneal injury: weak ankle dorsiflexion in three and paresthesias in one. After 1 year of follow-up, three of these patients regained function with rehabilitation. Two patients were in group A and two in group B.

On MR images, seven patients had edema or hematoma surrounding the peroneal nerve. Two of these patients had clinical signs of a partial peroneal injury. On the MRI examinations of seven patients, the peroneal nerve was interpreted as normal; none of these patients were symptomatic from nerve injury. Of the three patients with complete irreversible peroneal paresis, only one had an MRI examination before surgery. In this patient, the peroneal nerve was not identified on the MR images.

Two patients in group A and one patient in group B had popliteal artery injuries. In these patients, vascular injury was suspected at the time of admission, and they underwent subsequent conventional angiography and were treated surgically. All three patients had return of blood flow after surgery, and no reoperations or amputations were necessary. None of the patients underwent MR angiography.

In the literature, dislocation of the knee has been reported to occur mainly as a result of high-energy trauma or contact-sport-related trauma [1, 2, 4, 6, 8, 10]. Single case reports of low-energy knee dislocation in obese patients have been published [11, 12]. In our study, however, 46% of patients with knee dislocation were overweight and suffered low-energy trauma. One reason for this difference is the strict inclusion criteria of our study. We included only patients with dislocation on radiographs or patients with clearly documented dislocation. For example, patients with a dislocated knee at surgery but who had no evidence of previous true dislocation were not included. Most likely, some spontaneously reduced dislocations were not included in this study because there was no documented evidence of dislocation or because the patient could not remember the knee dislocation because of a high-energy multitrauma. Other possible explanations for the difference are the increasing prevalence of obesity [13-15] and the evolution of safety mechanisms in motor vehicles. Nevertheless, our study shows that the annual incidence of knee dislocation due to low-energy trauma in obese patients is not insignificant at a level 1 trauma center (Figs. 2A, 2B, 2C, and 2D).

A downhill skiing accident can be either low- or high-energy trauma. In our study, all three downhill skiing trauma patients were nonprofessionals; there were no competition- or jump-related downhill skiing accidents. All cases were considered to be caused by low-energy trauma.

The annual incidence of knee dislocation was calculated by comparing the patients with knee dislocation identified in our trauma center with the general population. This is a reasonable estimate because our hospital is the only one in an 8,929 km² area that takes care of major trauma patients, such as those with knee dislocation. In case a patient with a knee dislocation is initially taken to some other hospital, the patient is eventually transferred to our hospital for final treatment. The entire population is entitled to a general health care system, and there are no private hospitals taking care of trauma patients in this area. No patients younger than 16 years are treated at our hospital. However, a traumatic knee dislocation is rare in the pediatric population. Therefore, the lack of identified pediatric dislocations most likely does not have a significant effect on its incidence. Therefore, we think that the defined approximation of the incidence of dislocated knees is useful and informative.

Prior reports have shown that almost every patient with a knee dislocation has ruptured both cruciate ligaments [1, 6, 8-10]. Some studies have reported cases in which one of the cruciate ligaments has been intact [1, 18, 19]. Usually the intact cruciate ligament is the PCL [18], but rare cases of knee dislocation without ACL disruption have been reported [19]. The PCL prevents the posterior displacement of the knee, and the cases in which the PCL has been spared have been anterior or anteromedial knee dislocations [9, 18]. Our findings are in agreement with those of previous studies. In our series, every patient had a torn ACL, and one patient with anterior dislocation of the knee had an intact PCL (Figs. 3A, and 3B).

In previous studies, the incidence of meniscal tears has ranged between 13% and 53% in medial menisci and between 7% and 71% in lateral menisci [6, 8, 9]. In our study, 18% of patients had a medial meniscal tear and 36% had a lateral meniscal tear. Chondral injuries have been reported to occur in 57% of patients [9]. In our series, 55% of patients had chondral injury.

In previous studies, 20-25% of patients with knee dislocation had severe injury to the peroneal nerve [1, 20]. In our series, three of 24 patients (13%) had complete peroneal palsy. Severe MRI findings of peroneal injury, such as an encasing hematoma or partial disruption of fibers, also display some degree of symptomatic peroneal nerve injury [9]. In our series, we found a good correlation between patients who had normal peroneal nerve findings on MRI and patients with no clinical signs or symptoms of peroneal nerve injury. If a hematoma surrounded an intact peroneal nerve, two of seven patients had symptoms of partial nerve injury; both of these patients had a complete neurologic recovery during follow-up. Of the three patients with complete irreversible peroneal paresis, only one underwent MRI before surgery. MR images in that patient showed severe hematoma along the course of the common peroneal nerve, and the nerve was not identified. However, we cannot make reliable conclusions based on a single case (Figs. 4A, 4B, and 4C).

When we compared injuries between groups A and B, the only statistically significant difference was that in group B there were five complete ruptures of the popliteal tendon, and in group A every patient had an intact popliteal tendon. It has been suggested that the varus type of injury that results in a popliteal tendon rupture also increases the possibility of peroneal nerve injury [8]. Our findings are in agreement with that hypothesis. All three patients who had irreversible peroneal nerve injury were in group B.

The low incidence (13%) of major popliteal artery injury in our series may be explained by the high proportion of low-energy trauma. Note that, because arteriography was performed only in the setting of a clinical indication, the true incidence of popliteal artery injury has not been evaluated because a population of intimal injuries that did not progress to clinical signs would not be detected in this study.

A dedicated knee coil was used in each patient, and the subjective quality of all MR images was assessed as diagnostic. We did not find a significant difference in the number of false-positive or false-negative results between groups A and B.

Eleven of 24 patients (46%) with knee dislocation were overweight and sustained low-energy trauma; a simple fall was a common cause. These patients had no injuries to the popliteal tendon or the peroneal nerve. Only 13% of 24 patients had major popliteal artery injury. Obesity did not interfere with the performance or subjective image quality of knee MRI examinations. As the prevalence of obesity increases, it may alter the injury patterns seen in emergency departments. The radiologist should be aware that, even after a simple fall, these patients may have a knee dislocation.