Participants
The human research committee at our institution approved this prospective study. All adult patients presenting, to one of two upper-extremity orthopaedic surgeons (D.R. and J.B.J.) at a tertiary care institution, with a loss of flexion or extension of at least 30°, measured with a handheld goniometer at least four months after elbow trauma, who had not responded to supervised exercises and splinting (i.e., had had no measurable gains in the active range of flexion and extension over a thirty-day period) were eligible for enrollment. Exclusion criteria included severe burns, severe injury to the central nervous system with residual motor or cognitive deficits, nontraumatic arthritis, active infection, and severe articular injury requiring interposition arthroplasty or a total joint arthroplasty. The protocol was designed to evaluate all patients prior to the operative elbow contracture release and at least one year after the release and to record motion, disability, and general health status measures at both time points.
Evaluation Prior to and at Least One Year After Operative Contracture Release
Both prior to and at least one year after the contracture release, elbow flexion and extension and forearm rotation were measured with a handheld goniometer, and anteroposterior and lateral elbow radiographs were obtained. Patients completed the DASH questionnaire3 and the Short Form-36 (SF-36)4. In addition, we recorded patient age, sex, and occupation; the dominant and affected limb; the type of injury; associated injuries; the initial treatment; ulnar neuropathy; the number of surgical procedures prior to the release; the time between the injury and release; the time between the release and the final follow-up evaluation; the number of additional surgical procedures between the release and the time of final follow-up; and the arthritis grade at the time of final follow-up. All evaluations were performed by an independent observer (A.L.C.L.) not involved with the care of the patients.
The DASH questionnaire3 is an upper-extremity-specific disability measure used to evaluate difficulty with performing specific tasks as well as symptoms, social function, work function, sleep, and confidence. The score is scaled between 0 and 100 points, with higher scores indicating worse upper-extremity function. The SF-364 is a widely used general health status questionnaire that consists of a physical component and a mental component. The Physical Component Summary (PCS) score and the Mental Component Summary (MCS) score are norm-based scores that range from 0 to 100 points, with 50 points being the average score in the United States and 10 points representing one standard deviation. Norm-based scoring equates all scores, so scores of >50 points are better than the general population average for both summary measures, while scores of <50 points are worse. Norm-based scoring facilitates comparison and interpretation of summary scores and their subscales5.
To measure pain quantitatively, we recorded the scores for the pain subscale of the American Shoulder and Elbow Surgeons (ASES) Elbow Evaluation instrument6. Patients rated their pain from 0 points, indicating no pain, to 10 points, indicating the worst imaginable pain, on five scales: (1) pain when it is at its worst, (2) pain at rest, (3) pain when lifting a heavy object, (4) pain when doing a task with repeated elbow movements, and (5) pain at night. These five ratings were recalculated as a summary pain score that ranged from 0 to 25 points, with 25 points indicating no pain.
Ulnar neuropathy was graded according to the McGowan scale7, with Grade 1 indicating paresthesias in the ulnar nerve distribution with no detectable motor weakness of the hand; Grade 2, intermediate lesions with weak interossei and muscle wasting; and Grade 3, severe lesions with paralysis of the interossei and marked muscle weakness.
Radiographic evidence of arthritis was rated by an independent observer according to the system of Broberg and Morrey8. With this system, Grade 0 indicates a normal joint; Grade 1, slight joint-space narrowing with minimum osteophyte formation; Grade 2, moderate joint-space narrowing with moderate osteophyte formation; and Grade 3, severe degenerative change with gross destruction of the joint.
Statistical Analysis
Primary Study Question
Continuous data are presented in terms of the mean and range. Improvements in the arc of flexion and extension and in forearm rotation, and differences between preoperative and postoperative PCS, MCS, and DASH scores, were evaluated with use of paired t tests. To answer our primary study question, we used the Pearson correlation to assess the correlation between improvement in the arc of flexion and extension and improvement in disability (DASH) and general health status (SF-36 PCS and MCS) scores. A power analysis indicated that a total sample size of twenty-four patients would provide an 80% power to detect a significant correlation, with rho = 0.4 (ß = 0.20, a = 0.10), between improvement in the flexion-extension arc and improvement in disability and health status scores. A two-tailed p value of <0.05 was considered significant. To account for a possible loss to follow-up of 20%, we anticipated enrolling twenty-nine patients.
In order to account for confounding with other variables, we also performed bivariate and multivariate analyses. The number of explanatory variables that can be included in a multivariate model is limited by the overall sample size of the study. Therefore, instead of entering all potential explanatory variables into multivariate models, we ran a bivariate analysis first. Only those variables that were either significant (p < 0.05) or nearly significant (p < 0.10) in the bivariate analysis were entered into the multivariate analysis.
Bivariate analysis:
Pearson correlations (r) were used to assess the association of continuous variables (age, number of surgical procedures prior to the release, time between the injury and release, number of additional surgical procedures between the release and the final evaluation, time between the last surgery and the final evaluation, arc of flexion and extension, forearm rotation, improvement in the arc of flexion and extension, improvement in forearm rotation, and pain) with improvement in the PCS, MCS, and DASH scores and with the final PCS, MCS, and DASH scores. Correlation was categorized as small (r = 0.10 to 0.29), medium (r = 0.30 to 0.49), or large (r = 0.50 to 1.00)9. Unpaired Student t tests were used to evaluate differences with regard to improvement in PCS, MCS, and DASH scores and in final PCS, MCS, and DASH scores between dichotomous variables (sex, laborer [any person employed to do physical or heavy manual work] versus non-laborer, limb dominance, distal humeral fracture, associated ipsilateral injuries, arthritis, and ulnar neuropathy).
Multivariate analysis:
Backward stepwise multiple linear regression analysis was performed to determine the best predictors of improvement in the PCS, MCS, and DASH scores and in the arc of flexion and extension as well as the best predictors of the final PCS, MCS, and DASH scores and the final arc of flexion and extension, thereby accounting for confounding between explanatory variables. Thus, we ran multivariate models for each of the six outcome variables. A backward stepwise multiple regression model initially includes all of the entered variables and then iteratively removes variables from the model until the best-fit model is achieved. Multiple linear regression models produce a statistic called the adjusted R2, which reflects the percentage of the overall variability in the dependent (outcome) variable that can be explained or accounted for by the predictor (explanatory) variables included in the multiple linear regression model. If pain was among the predictors, an additional model with only the variable pain entered was run. Comparison of the variability accounted for by each model (the adjusted R2) provides a measure of the relative influence of each explanatory variable on the overall variation in the response variable.
Multivariate analysis of variance was performed to assess the significance of the models, with significance indicating a linear relationship between at least one of the predictor variables in the model and the dependent variable.
Secondary Study Question
Pearson correlation was used to analyze the association between the final arc of flexion and extension and the final disability and health status scores (DASH, PCS, and MCS).
Source of Funding
No outside funding was received in direct support of this study.
Patients
Between January 2004 and May 2006, twenty-eight of twenty-nine eligible patients were enrolled in this study. Five patients did not return for follow-up (tthree patients declined, one patient was terminally ill, and one patient could not be located), leaving twenty-three patients included in the current study. One patient who had massive heterotopic bone and a proximal radioulnar synostosis at the index contracture release underwent multiple additional procedures to treat recurrent heterotopic bone and stiffness and finally underwent a total elbow arthroplasty. This patient was considered to have had a failure of the release and was excluded from further analysis.
Preoperative Evaluation
Patient and injury characteristics as well as information on the initial treatment of the traumatic injury and additional procedures prior to the index contracture release are presented in Table I. The preoperative and postoperative evaluations (elbow function, DASH, SF-36, and pain scores as well as radiographic data) are shown in Table II.
Operative Treatment
The time between the initial injury and the operative contracture release averaged twenty-one months (range, four to 210 months). Combined medial and lateral intervals through a posterior incision were used in eleven patients, lateral and medial intervals through separate incisions were employed in two patients, a lateral interval through a lateral or previous posterior incision was utilized in eight patients10-12, and a medial interval through a medial incision13 was used in one patient. The ulnar nerve was addressed in thirteen patients: it was transposed anteriorly in five patients, and it was released in eight patients in whom it had already been transposed during a previous surgical procedure. Heterotopic bone was excised in sixteen patients, including two who underwent resection of a proximal radioulnar synostosis. Implants were removed from seven patients. In the three patients who had a nonunion of the distal part of the humerus, release of the capsules and the ulnar nerve was followed by debridement of the fracture site, removal of loose implants, and fixation of the nonunion; two of the patients had autogenous bone-grafting as well.
Four patients received preoperative irradiation (a single dose of 7 Gy) on the morning of the surgery as prophylaxis against formation or recurrence of heterotopic bone. Nonsteroidal anti-inflammatory drugs were not prescribed for any patient after the surgery. All patients began passive and active-assisted exercises on the first postoperative day, and eleven patients used a continuous-passive-motion device. Twelve patients began using a splint between three and six weeks after the surgery to help regain motion; ten used a static progressive splint, and two used a dynamic splint.
Additional Operations
Four patients had a total of eight additional operations after the index contracture release and prior to the one-year evaluation. One patient had three procedures to treat a forearm compartment syndrome. The other three patients had a subsequent elbow release, with excision of recurrent heterotopic bone (preceded by preoperative irradiation with a single dose of 7 Gy) in two of them. The third patient also had an interposition arthroplasty and temporary hinged external fixation.
Final Evaluation
The final evaluation was performed at an average of twenty-three months (median, sixteen months; range, twelve to sixty-eight months) after the index contracture release and an average of twenty-two months (median, sixteen months; range, twelve to sixty-two months) after the most recent surgery. The patients obtained significant improvements in the flexion-extension arc and in the arc of forearm rotation. In addition, there were significant improvements in the DASH and SF-36 PCS and MCS scores (Table II).
Correspondence of Motion with Disability and Health Status
There was no significant correlation between the improvement in the arc of flexion and extension and the improvement in the PCS (p = 0.73), MCS (p = 0.41), or DASH score (p = 0.53). Neither was there a significant correlation between the final arc of flexion and extension and the final PCS (p = 0.52), MCS (p = 0.42), or DASH score (p = 0.39).
Bivariate and Multivariate Analysis
All explanatory variables were entered into a bivariate analysis to find associations between explanatory variables and each of the six outcome variables: improvements in the PCS, MCS, and DASH scores, and the final PCS, MCS, and DASH scores. Explanatory variables that had a significant or nearly significant association with the outcome variables in the bivariate analysis (listed in Table III) were entered into a multivariate analysis to identify the variables that explained the variation in each of the outcome variables best, while accounting for confounding between the variables. This multivariate analysis provides information about the degree to which the outcome variables are explained by the variables in the model. Overall, the models explained the final PCS, MCS, and DASH scores better than they explained the improvement in the PCS, MCS, and DASH scores. We performed additional analyses with pain as the only variable entered for the outcome variables that were associated with pain in the bivariate analysis: pain explained 21% of the improvement in the PCS, 9% of the improvement in the DASH score, 15% of the final MCS score, and 17% of the final DASH score. The results of the multivariate analysis are summarized in Table IV.
Although additional surgery was often performed, operative contracture release resulted in substantial improvements in elbow motion (average, 55° of improvement in the arc of flexion and extension and 29° of improvement in forearm rotation). Health status improved as well: the SF-36 PCS and MCS scores increased an average 10 and 5 points, respectively. The improvement in the disability (DASH) score averaged 20 points. However, our hypothesis was not confirmed: despite the significant improvements in all outcome measures, there was no significant association between improvement in motion and improvement in disability scores and health status. Although open elbow contracture release resulted in less disability and better health status, the degree of improvement in objectively measured motion was not associated with the degree of improvement in perceived health. Thus, none of the disability and health status measures were sensitive to changes in motion. However, the outcome measures were sensitive to other factors. In particular, pain turned out to be an important predictor of disability and health status.
When possible, predictors of final disability and health status scores and of improvement in disability and health status scores were analyzed in a multivariate regression analysis; ulnar nerve dysfunction and the number of surgical procedures prior to the release were the most prevalent objective predictors. A previous study showed the results in patients without arthritis to be better than those in patients with arthritis14, a finding that makes sense but that was not observed in our study. A previous meta-analysis of articles reporting associations between impairment and patient-rated disability and health status revealed that only 36% of disability scores and 13% of health status scores were explained by impairment15. In the current study, the relationship of the flexion-extension arc (the primary measure of impairment) with disability and health status was limited as well. Pain and ulnar neuropathy were the only predictors of improvement in health status in our study. Furthermore, models that included pain alone explained a considerable part of the variability in disability and (mental) health status: pain was a predictor of PCS (21%), MCS (9%), and DASH scores (15%). In spite of the clear association of pain with disability and health status, our numbers were somewhat less pronounced than those reported in a previous study16, in which 36% of the variation in DASH scores after treatment of elbow fractures was found to be explained by pain. The perception of pain is highly variable among individuals and strongly psychosocially mediated17-21. Patients with low depression scores have less disability22, and good coping skills seem especially important in patients with pain resulting from unclear and vague causes23.
Most relatively simple posttraumatic elbow contractures respond to exercises (eventually including a splinting program24) and patience. The majority of posttraumatic contractures for which surgery is elected in our practice are very complex: the cohort that was studied in this investigation had a severely stiff elbow, with an average 51° arc of flexion and extension prior to the release, most patients had sustained a distal humeral fracture or elbow fracture-dislocation, and in many patients the stiffness was complicated by fracture nonunion or heterotopic bone. Heterotopic bone has traditionally been considered to be a poor prognostic factor, although a recent investigation showed the results of operative contracture release in patients who had had excision of heterotopic bone blocking motion to be better than those in patients with a capsular contracture alone25.
In addition to the excluded patient who underwent a total elbow arthroplasty, three of the twenty-two analyzed patients had subsequent surgery for elbow stiffness prior to the time of final follow-up. Including one of these three patients, there were five patients who had an arc of flexion and extension of =80° at the follow-up evaluation performed at least one year postoperatively. The improvement in the arc of flexion and extension that was found in this study compared well with that in prior reports (an average improvement of 55° in our study compared with improvements ranging from 21° to 66° in previous studies26,27).
Ten (45%) of our twenty-two patients had ulnar nerve dysfunction at the final evaluation. In six patients it was present prior to the initial contracture release, but the four remaining patients had no prior dysfunction of the nerve. Three of those four patients had a transposition or release during the index contracture release. Three patients with preoperative ulnar nerve dysfunction had improvement after release or transposition of the nerve during the index release, and two of them had no symptoms of ulnar neuropathy at the final evaluation. In two of the patients who had additional surgery to address stiffness, the ulnar nerve was transposed: at the final evaluation, one of these patients had altered sensibility in the distribution of the ulnar nerve. With increased flexion the ulnar nerve flattens against the medial epicondyle28, and with full flexion the nerve elongates approximately 5 mm29,30. A contracture release may thus increase traction on the ulnar nerve and thereby put it at risk. The benefit of an ulnar nerve release or transposition in patients without symptoms of ulnar nerve dysfunction seems questionable on the basis of our data. However, when interpreting our results, one should keep in mind that most patients had a transposition or release of the ulnar nerve because of placement of implants in the case of nonunion, because of heterotopic bone entrapping the nerve, or because of preexisting ulnar neuropathy. There is no conclusive evidence regarding routine decompression or transposition of the ulnar nerve during elbow contracture release reported in the literature31. The limited sample size of our study does not allow definitive conclusions on this topic.
The data in this report should be interpreted in light of the shortcomings of the study, such as the diversity of the population, which included patients with complex contractures. (Complex contractures fit our inclusion criteria as they reflect the true nature of posttraumatic elbow stiffness.) Furthermore, we had difficulty inducing patients to return for follow-up at a standard interval after the surgery. While we did not analyze the influence of follow-up time on motion, there was no correlation between the time since the surgery and any of the health status or disability outcome measures. Another limitation is the fact that, as a result of the loss to follow-up, the late exclusion of a patient who underwent an interposition arthroplasty but was erroneously kept in the study, and the exclusion of a patient because of a total elbow arthroplasty, we eventually had only twenty-two patients instead of the twenty-four needed according to our power analysis. Nonetheless, the lack of correlation between improvements in motion and improvements in disability and health status scores was fairly convincing. Given the p values ranging from 0.10 to 0.98, it seems unlikely that a single additional patient would have affected our findings. This lack of correlation seems counterintuitive and brings into question the traditional surgeon and patient focus on motion as the most important measure of success. On the other hand, it might be argued that disability and health status instruments measure a much more complex construct than does a simple measure of impairment such as range of motion. The failure to demonstrate a correlation may not have as much to do with the relative importance of improving motion as with the fact that motion is a very simple construct and what is measured by the DASH and SF-36 is much more complex. Additional study is merited to better determine how impairments in motion affect disability and why decreases in disability are not correlated with increases in motion. This will better define the role of operative release of posttraumatic elbow contracture.