Radial head fractures are the most common elbow fracture, but controversy exists about how to treat them. There is good evidence that undisplaced and minimally displaced radial head fractures without a mechanical block to forearm rotation can be managed nonoperatively with excellent results . It is also known that patients with complete articular radial head fractures with more than three displaced fragments have more favorable outcomes with early radial head replacement than with open reduction and internal fixation (ORIF) [6, 15].
However, the optimal treatment of isolated displaced partial articular fractures of the radial head is not known; good results have been reported with both nonoperative management [1, 4] and ORIF . A recent systematic review found insufficient evidence to draw firm conclusions on the optimal treatment of displaced isolated Mason Type II radial head fractures . Two recent reviews on partial articular radial head fractures have concluded that, while a randomized controlled trial comparing ORIF with nonoperative management in isolated displaced partial articular radial head fractures would be ideal in determining optimal management, important information could be gained initially from a retrospective comparative study focusing on patient-rated and functional outcomes [2, 10].
We therefore (1) compared isolated displaced partial articular radial head fractures managed nonoperatively with those managed with ORIF in terms of validated outcomes scores, ROM, and strength; (2) assessed whether there were any predictor variables (such as age, displacement, or energy) on clinical outcomes scores; and (3) compared the complications between groups.
Patients and Methods
Patients with displaced partial articular radial head fractures were identified from an existing institutional database of prospectively followed radial head fractures from 2004 to 2011. Inclusion criteria were a partial articular radial head fracture that was displaced greater than 2 mm but less than 5 mm. Displacement was defined as the greatest distance of separation of fracture fragments as seen radiographically on any view. Exclusion criteria included any other fracture around the elbow or any other fracture or clinically evident ligament injury elsewhere in the ipsilateral upper limb, a documented elbow dislocation, a complete articular radial head fracture, or skeletal immaturity. In two cases, fracture displacement was not able to be determined due to unavailability of radiographs, and these patients were excluded. Thus, in this initial search, we identified 41 patients who fulfilled our eligibility criteria, 36 treated nonoperatively and five treated with ORIF. Of the nonoperative group, 30 were followed for a minimum of 2 years with clinical assessment, five patients dropped out of the study before 2 years, and one patient died of unrelated causes. Four of the five patients who underwent ORIF fulfilled minimum 2-year followup, while one was not contactable at 2 years.
Because our purpose was to compare operative and nonoperative treatment, we then undertook a more extensive review of our operative records to identify all radial head fractures treated with ORIF at our institution from 1997 to 2011. Seventy-three patients were identified, and 34 patients met the above inclusion criteria. These patients were invited back for a clinical and radiographic assessment by a surgeon not involved in their initial care at a minimum of 2 years and a chart review; 26 consented, seven declined, and one was not contactable. These 26 patients plus the four from the first search comprised the 30 patients who underwent ORIF in our study. The groups were different with respect to age (mean ± SD: 51 ± 17 years in the nonoperative group versus 39 ± 10 years in the operative group; p = 0.001) and fracture displacement (mean: 2.3 ± 0.3 mm in the nonoperative group versus 2.8 ± 0.6 mm in the operative group; p = 0.001) but were not different with respect to sex, number of displaced fragments, dominant hand being affected, occupation, workers compensation, smoking status, or other comorbidities (Table 1).
The patients were reviewed at a mean of 3 ± 1.7 years for the nonoperative group and at a mean of 4.5 ± 1.8 years for the ORIF group. The mean followup for both groups was 3 years 9 months. The study was approved by our institutional review board, and informed consent was obtained from all patients before participation in this study.
As there were no definitive guidelines on whether ORIF or nonoperative treatment represented best management, the type of treatment was decided by the attending surgeon in conjunction with the patient on a case-by-case basis. CT scan was utilized at the discretion of the treating surgeon. The nonoperative group (Fig. 1) was encouraged to begin ROM exercises within 1 week of injury. Patients who underwent ORIF were treated with a minimum of two countersunk screws (Fig. 2), and in all cases, active ROM was encouraged within 1 week.
The primary outcome measure was the Patient-rated Elbow Evaluation (PREE) [13, 16]. Secondary outcome measures included the QuickDASH , the SF-12 , and the Mayo Elbow Performance Score (MEPS) , as well as a clinical measurement of elbow ROM using a goniometer, elbow strength evaluation using a Biodex dynamometer (Biodex Medical Systems, Shirley, NY, USA), hand grip calibrated strength testing, and a sensory examination of the hand. A 10% difference in strength in favor of the dominant hand was corrected for when comparing grip strength measurements only. Fourteen of 30 patients treated nonoperatively agreed to return for a radiographic assessment of their elbow at a minimum of 2 years, while 28 of the 30 patients who received ORIF agreed to radiographic assessment at the time of their review.
Possible predictive variables for outcomes including age, displacement, and energy of injury were recorded from a combination of interview and chart and radiographic review by three fellowship-trained surgeons (AY, JP, RG). For age, we compared patients younger than 60 years (n = 51) with those 60 years or older (n = 9). For fracture displacement, we compared patients with large (3.0-4.9 mm) fracture displacements (n = 12) and small (2.0-2.9 mm) fracture displacements (n = 48). For energy of injury, we compared patients with high-energy injuries (such as a motor vehicle accident or fall from more than 1 m) (n = 11), medium-energy injuries (such as sporting injury or fall from less than 1 m) (n = 21), and low energy of injury (simple fall on outstretched hand) (n = 28).
Complications such as nonunion, heterotopic ossification, and hardware failure were looked for in all patients on their most recent available radiographic evaluation by two fellowship-trained upper-extremity surgeons (AY, JP). Clinical complications were documented during chart review and clinical examination by the same two surgeons.
All continuous variables were analyzed using a two-tailed Student's t-test, while categorical values were analyzed with the chi-square test. Linear correlation of two variables was performed with the Pearson correlation coefficient. A sample size calculation indicated that 30 patients were needed in each group to ensure 80% power based on a variance of 20, alpha of 0.05, and a 15-point minimal clinically important difference between groups for the primary outcome measure (PREE). We performed all statistical analyses using IBM® SPSS® software (Version 21.0; IBM Corp, Armonk, NY, USA).
At followup, there was no difference between groups in terms of our primary outcome variable, the PREE score, but there was a difference in the MEPS favoring the nonoperative group (93 [95% CI, 89-97] versus 86 [95% CI, 80-90]; p = 0.012) (Table 2). Although statistically significant, this difference is not clinically significant. In the nonoperative group, 19 patients had an excellent outcome according to MEPS (score ≥ 90), nine had a good outcome (score = 75-89), and two had a fair outcome (score = 60-74). In the operative group, there were 11 excellent outcomes according to MEPS, 10 good outcomes, and nine fair outcomes. No patient in the study had a poor outcome according to the MEPS. The QuickDASH and SF-12 physical and mental components were similar between groups. With regard to other secondary outcome measures, the nonoperative group was found to have better pronation (86° [95% CI, 84°-88°] versus 83° [95% CI, 80°-85°], p = 0.04), but this was clinically insignificant. There were no other differences between the two groups in terms of ROM (Table 3) or strength (Table 4).
Age was found to be an independent predictor of outcome. Patients in both groups younger than 60 years had worse outcomes than those 60 years or older as measured by PREE (9 [95% CI, 5-13] versus 2 [95% CI, 0-6]; p = 0.01) and MEPS (88 [95% CI, 85-92] versus 97 [95% CI, 92-102]; p = 0.007) (Table 5). Fracture displacement amount did not correlate with the PREE (Pearson correlation coefficient, 0.20; p = 0.14), but there was weak correlation between greater fragment displacement and lower MEPS (Pearson correlation coefficient, −0.29; p = 0.03). Patients in the low-energy injury group were significantly older (mean age, 51 years) than the patients in the medium-energy injury group (p = 0.01) and high-energy injury group (p = 0.02) (mean age, 40 and 37 years, respectively). However, the energy of injury was not found to be an independent predictor of outcome, based on PREE and MEPS.
There were more complications in the operative group (Table 6). In the nonoperative group, there was one case of complex regional pain syndrome, diagnosed by a tertiary referral chronic pain service at our institution, and one case of heterotopic ossification. No patient in the nonoperative group had secondary surgery. In the ORIF group, there were eight cases of heterotopic ossification (Fig. 3) and two cases of early screw failure (Fig. 4). In both cases of screw failure, the initial reduction of the fracture was not maintained at the first postoperative outpatient review radiograph, but the fracture went on to unite, and further surgery was not required. One of these patients had an excellent outcome and the other a good outcome as per the MEPS. In all other ORIF cases, the initial anatomic reduction obtained at surgery was maintained at final followup, and no screws were seen to lose position. One patient who developed heterotopic ossification in the operative group was classified as Hastings and Graham  Class 2c (flexion 30°-124°, pronation to 79°, supination to 50°), but all other cases of heterotopic ossification were Class 1 (mean elbow flexion range of 135° compared to 138° on the unaffected limb; mean combined pronation and supination of 160° compared to 164° on the unaffected limb). There were no cases of nonunion or infection in either group.
Although there are several case series describing the results of either operative or nonoperative management of partial articular radial head fractures, a literature search revealed only two papers directly comparing the two [4, 11]. In our study, we compared outcome scores and clinical assessments of nonoperative and operative fixation of isolated displaced partial articular radial head fractures, assessed the association of age, fracture displacement, and energy of mechanism to outcome scores, and compared complications between nonoperative and operative treatments.
Weaknesses of the study include the fact that it is a nonrandomized review with inherent selection bias and that the two groups did have significant differences in terms of age, fracture displacement, and length of followup. It was not surprising that a lower patient age or a larger displacement of fracture would influence management toward operative intervention, and we accept that these variables could affect outcomes. However, displacement was not significantly related to PREE score and was only weakly related to MEPS, while lower age (younger than 60 years) was independently related to higher pain and disability (MEPS/PREE). We accept that the longer-term followup in the ORIF group is also a confounder, as time may reveal problems that are not obvious early on. Additionally, there were insufficient numbers of long-term followup radiographs for a meaningful comparative analysis of radial head morphology and presence of arthritis between the two groups, and the more significant limitation of long-term radiographic followup in the nonoperative group compared to the ORIF group means the rate of heterotopic ossification in the nonoperative group may be relatively underreported.
Regarding outcome scores, we found no significant difference between the two groups in our primary outcome measure (PREE) despite adequate study power. We did however find a significant difference in MEPS favoring nonoperative treatment. When comparing our results to the current literature, the mean MEPS of 86 in the ORIF group is similar to the mean MEPS of 89 in the long-term (mean 22-year) followup study of 16 patients undergoing ORIF for Mason Type II radial head fractures by Lindenhovius et al. . The mean MEPS of 93 and the proportion of patients scoring excellent or good (93%) in the nonoperative group were comparable to a similar group of patients reported in a large review by Duckworth et al.  (mean MEPS, 93; proportion scoring excellent or good, 96%).
Khalfayan et al.  compared 16 nonoperatively treated patients with 10 patients who had ORIF. Based on MEPS outcome, nine of 10 patients who underwent ORIF had a good or excellent result, compared with seven of 16 nonoperatively managed patients. Our study is adequately powered, comparing a very specific patient group with well-defined and well-matched fracture types treated by four surgeons at one institution, using validated outcome measures, and with an equivalent number of patients in the nonoperative and ORIF groups. There were two patients identified who met inclusion criteria except for having displacement of more than 5 mm of their fracture, and both of these patients received ORIF. The upper limit of 5 mm for this paper was set in an attempt to have well-matched fracture types between the two groups. Other strengths include the mean followup of 3 years 9 months and the relatively high proportion of patients who agreed to return for review. It is interesting to note that, despite our institution being a specialist referral center with a large number of upper-limb surgeons, relatively small numbers of isolated displaced partial articular radial head fractures were seen, as the incidence of associated injuries is high.
Regarding potential predictors of outcome, a patient age of 60 years or more was found to be an independent predictor of better outcome. Fracture displacement and the amount of energy involved in the injury were not found to be independent predictors. Duckworth et al.  recognized in their review of the epidemiology of radial head fractures that the mean age of female patients was significantly higher than that of male patients and that female patients were more likely to sustain their fracture with a lower-energy mechanism than male patients. Our study found similar patterns, and it may be important for any future trial to consider this bimodal distribution of populations.
More complications were seen in the operative group. Most of these complications were mild heterotopic ossification, and there were no cases of nonunion or infection.
In conclusion, there was no clinically significant difference in outcomes, ROM, or strength between patients treated nonoperatively and those treated with ORIF for isolated displaced partial articular radial head fractures. Younger patients fared worse and the ORIF group had more complications. Although no benefit to ORIF over nonoperative treatment could be found in this study, it should be well noted that the two groups were different with respect to age, and this was found to be an independent predictor of outcome. A well-designed randomized controlled trial with long-term followup will reduce the bias encountered in this study and may provide further guidelines as to the optimal management of these fractures. As this was a retrospective comparison, the presence or absence of a mechanical block to rotation influencing the choice of treatment was not considered in this study, and it should be noted that, in treating an isolated displaced partial articular radial head fracture, the presence of a mechanical block to rotation remains an absolute indication for surgical intervention.
We thank Kristie Millman BSc for her invaluable contribution in following up patients and data processing, performed with both cheerful enthusiasm and expertise. We also thank Jeff Pike MD for his contribution in the identification, chart review, examination, and radiographic review of patients involved in this study.
1. Akesson T, Herbertsson P, Josefsson PO, Hasserius R, Besjakov J, Karlsson MK. Primary non-operative treatment of moderately displaced two-part fractures of the radial head. J Bone Joint Surg Am.
2. Athwal GS, King GJ. Partial articular fracture of the radial head. J Hand Surg Am.
3. Duckworth AD, Clement ND, Jenkins PJ, Aitken SA, Court-Brown CM, McQueen MM. The epidemiology of radial head and neck fractures. J Hand Surg Am.
4. Duckworth AD, Watson BS, Will EM, Petrisor BA, Walmsley PJ, Court-Brown CM, McQueen MM. Radial head and neck fractures: functional results and predictors of outcome. J Trauma.
5. Fan ZJ, Smith CK, Silverstein BA. Assessing validity of the QuickDASH and SF-12 as surveillance tools among workers with neck or upper extremity musculoskeletal disorders. J Hand Ther.
6. Grewal R, MacDermid JC, Faber KJ, Drosdowech DS, King GJ. Comminuted radial head fractures treated with a modular metallic radial head arthroplasty: study of outcomes. J Bone Joint Surg Am.
7. Gummesson C, Ward MM, Atroshi I. The shortened Disabilities of the Arm, Shoulder and Hand questionnaire (QuickDASH): validity and reliability based on responses within the full-length DASH. BMC Musculoskelet Disord.
8. Hastings H 2nd, Graham TJ. The classification and treatment of heterotopic ossification about the elbow and forearm. Hand Clin.
9. Herbertsson P, Josefsson PO, Hasserius R, Karlsson C, Besjakov J, Karlsson MK. Displaced Mason type I fractures of the radial head and neck in adults: a fifteen to thirty-three year follow-up study. J Shoulder Elbow Surg.
10. Kaas L, Struijs PA, Ring D, Dijk CN, Eygendaal D. Treatment of Mason type II radial head fractures without associated fractures or elbow dislocation: a systematic review. J Hand Surg Am.
11. Khalfayan EE, Culp RW, Alexander AH. Mason type II radial head fractures: operative versus nonoperative treatment. J Orthop Trauma.
12. Lindenhovius AL, Felsch Q, Ring D, Kloen P. The long-term outcome of open reduction and internal fixation of stable displaced isolated partial articular fractures of the radial head. J Trauma.
13. MacDermid JC. Outcome evaluation in patients with elbow pathology: issues in instrument development and evaluation. J Hand Ther.
14. Morrey BF, An KN, Chao EYS. Morrey BF. Functional evaluation of the elbow. The Elbow and its Disorders
1993; 2, Philadelphia, PAWB Saunders86-97.
15. Ring D, Quintero J, Jupiter JB. Open reduction and internal fixation of fractures of the radial head. J Bone Joint Surg Am.
16. Vincent JI, MacDermid JC, King GJ, Grewal R. Validity and sensitivity to change of patient-reported pain and disability measures for elbow pathologies. J Orthop Sports Phys Ther.