Overview
A retrospective cohort design was used to document the health status and functional outcomes of adults in whom a pilon fracture had been treated with either open reduction and internal fixation or external fixation with or without limited internal fixation at one of two North American trauma centers during 1994 and 1995. These two particular centers were selected because, during the years studied, pilon fractures had been treated primarily with open reduction and internal fixation at one site whereas external fixation with or without limited internal fixation typically had been used at the other site. Two to five years after injury, patients were asked to return to the medical center where they had been treated to undergo a comprehensive health status evaluation. The clinical characteristics of the pilon fracture and its treatment were ascertained by abstraction of the medical records. We examined the variations in health, function, and midterm impairment status according to patient, injury, and treatment characteristics.
Participants and Setting
Patients who were between sixteen and sixty-nine years old at the time of injury and had been admitted to one of two trauma centers (Sites A and B) between January 1994 and December 1995 for treatment of an AO/OTA class-43B or 43C pilon fracture were eligible for the study
18 . The centers were urban-based trauma referral centers from distinct geographic regions of the country. Large numbers of high-energy injuries are treated at both facilities, and both centers are part of university-based academic medical centers.
Patients were excluded if (1) they had undergone primary amputation or had not been treated with either open reduction and internal fixation or external fixation; (2) they had received primary or operative treatment elsewhere; (3) they had sustained a bilateral pilon fracture and had been treated with both methods; (4) they did not speak English; (5) they were younger than sixteen or older than sixty-nine years at the time of injury; (6) they had died; (7) they had had a preexisting medical condition that severely limited physical or mental health before the injury; or (8) they had sustained a concomitant moderate-to-severe traumatic brain injury (i.e., an injury with an Abbreviated Injury Scale
19 score for the head region of =4), a spinal cord injury with a motor deficit, or an ipsilateral fracture of the femur, patella, or tibia or dislocation of the knee.
Procedures
The institutional review boards at both centers approved the study. Informed consent was obtained from all study participants. The trauma registries at the two sites were used to identify all patients with eligible injuries, and 152 patients met the inclusion criteria. However, 32% (forty-nine patients) were ineligible because of one or more of the above exclusion criteria. Of the forty-nine patients who were excluded, twenty-seven had sustained a concomitant ipsilateral injury, nine were ineligible because of their age at the time of injury, six had died, five met multiple exclusion criteria, and two had a bilateral pilon fracture and had received both types of treatment.
Each of the 103 eligible patients was sent an introductory letter describing the study and a copy of the consent form. The letter was followed up with a telephone call from the site coordinator, who answered any questions that the prospective participant had. Patients who agreed to participate were each given an appointment to return to the medical center for an evaluation consisting of a face-to-face interview and a functional assessment. A small number of patients who were unable to return to the medical centers were interviewed by telephone. Each participant who completed the health status evaluation was paid $50.
Patient Characteristics
As part of the health status evaluation, the site coordinator conducted a thirty to forty-five-minute interview with each patient. Participants were queried about sociodemographic information that was hypothesized to influence health outcomes. They were asked their age, marital status, highest grade completed in school, total personal annual income before taxes, and type of health insurance that they had at the time of the interview, if any. They were asked whether they presently smoked cigarettes and, if so, how many. Finally, they were read a list of major medical conditions and asked whether a doctor had ever told them they had any of those conditions.
Injury Characteristics
Data regarding the circumstances of the injury and the severity of all associated injuries were obtained from each center's trauma registry. The nature and severity of each injury was graded according to the Abbreviated Injury Scale
19 , which classifies individual injuries by body region on an ordinal scale of 1 (minor) to 6 (unsurvivable). The side (right or left) of all extremity fractures was also noted. Head injuries were documented according to both the head region score of the Abbreviated Injury Scale and the admission Glasgow Coma Scale score, a measure of impaired consciousness that ranges from a score of 3 (comatose) to a score of 15 (no impaired consciousness)
20 .
Radiographs made at the time of admission and initial postfixation radiographs of all participants who completed the follow-up evaluation were reviewed by the principal investigator at each site. The pilon fracture was categorized according to the AO/OTA classification of long-bone fractures and whether the fracture was open or closed
18 . To ensure consistency of grading between the two sites, the radiographs of the patients treated at one center were then sent to the other center and were blindly graded by the principal investigator at that site. Consistency between the two graders, with regard to fracture type (AO/OTA type B or C) and location (AO/OTA location 1, 2, or 3), was evaluated. Overall, agreement between the two site investigators was high for fracture type but not for both fracture type and location. There was perfect agreement regarding the type of fracture in 86% of the cases. However, when the type was combined with the location of the fracture, perfect agreement dropped to 55%. When the two site investigators disagreed about the type of fracture, they conferred and reached a consensus regarding the final grade.
Treatment Characteristics
Data regarding treatment related to the pilon fractures that was rendered at the respective medical centers (i.e., during the initial hospitalization as well as any rehospitalizations and outpatient follow-up visits) were abstracted by the site coordinators. The following information was collected: (1) date and method of definitive fixation; (2) date and method of revisions of the original fixation; (3) date, number, and type of soft-tissue-coverage procedures; (4) total number of débridements during the initial hospitalization; (5) date of bone-grafting of the tibia; (6) date and type of arthrodesis; (7) date and level of amputation; (8) date and type of stabilization of ipsilateral talar and calcaneal fractures; (9) any other procedures performed on the tibia; and (10) any documented limb complications.
Definitive fixation of the pilon fracture was defined in relation to fracture-healing. If an external fixator was placed and not removed until the fracture had healed, then external fixation was considered to be the definitive treatment modality. If external fixation was used temporarily until open reduction and internal fixation could be safely performed and was then removed immediately thereafter, the definitive treatment was considered to be the open reduction and internal fixation. If the external fixator was left in place after open reduction and internal fixation and removed after the fracture had healed, the open reduction and internal fixation procedure was considered to be limited and the definitive treatment was considered to be external fixation with or without limited internal fixation. All external fixation used in the series was bridging external fixation. No hybrid frames were used.
Outcomes Assessed
The primary outcomes studied included (1) general health status as measured by the eight scales of physical and psychosocial health of the Short Form-36 (SF-36), (2) lower-extremity function as reflected by the ambulation scale of the Sickness Impact Profile, (3) range-of-motion impairment as measured by the American Medical Association range-of-motion impairment rating, (4) pain as indicated on a 100-point visual analog scale, and (5) stair-climbing performance as measured by the ability to ascend and descend a flight of stairs reciprocally. In addition, a secondary outcome of employment status was assessed for those who were employed at the time of injury.
The Medical Outcomes Study thirty-six-item Short Form Health Survey
21 (SF-36) consists of thirty-six items or questions representing eight scales: physical function, role disability due to physical health problems, bodily pain, general health perceptions, vitality, social function, role disability due to emotional problems, and mental health. Scale scores range from 0 (worst) to 100 (best)
21 . Age and gender-specific population norms have been developed for the SF-36 and were used to interpret our findings
21 . The SF-36 has been extensively tested for its reliability and validity and has been used previously to measure the health status of other populations with traumatic injuries
22-26 .
Participants were also questioned about their ability to perform different lower-extremity activities. They completed the ambulation subscale of the Sickness Impact Profile, which consists of twelve statements related to walking and climbing stairs
27,28 . The twelve items describe limitations such as walking more slowly, being able to walk for only short distances, needing the assistance of another person to walk, and not being able to walk at all. The Sickness Impact Profile ambulation subscale ranges from 0 to 100; the higher the score, the greater the degree of ambulatory dysfunction.
Participants were also asked to report whether they had difficulty with more strenuous or challenging lower-extremity activities. All participants were asked how frequently the ankle caused them difficulty when running one block, wearing different types of shoes, climbing a ladder, or participating in recreational activities. Finally, participants were queried about any equipment or devices that they usually used to get around and how frequently they experienced pain, swelling, or stiffness in the ankle.
The lower-extremity impairment and lower-extremity function of each participant were evaluated by a physical therapist or certified athletic trainer during the functional evaluation. The active and passive ranges of motion were measured with use of the start and end positions for each motion of the hip, knee, and ankle recommended by the American Academy of Orthopaedic Surgeons
29 . With use of the American Medical Association's
Guides to the Evaluation of Permanent Impairment, range-of-motion values can be summarized as an impairment rating for the injured extremity
30 . The lower-extremity impairment ratings range from 0% to 100%. The higher the score, the greater the range-of-motion impairment.
Patients were also asked to mark on a visual analog scale the degree of pain that they experienced in the injured leg during a typical day. Pain scores range from 0 (no pain) to 100 (unbearable pain).
Lower-extremity function was evaluated by asking participants to perform several activities, including complete full toe excursion (rising onto the toes such that the weight of the body is borne on and balanced between the metatarsal heads and the phalanges), ascending and descending a flight of stairs, squatting and picking up a light object (a pencil) and a heavy object (a 10-lb [4.5-kg] weight), standing on the injured leg for thirty seconds, rising from a chair without using the arms five times within fifteen seconds, and running in place for thirty seconds. The evaluators observed each participant's performance and graded each activity according to whether it was completed and, if appropriate, whether it was completed within the specified time.
We also asked participants specifically about employment status. Participants who reported that they had been employed at the time of injury but were not employed at the time of the follow-up interview were asked whether the pilon fracture had prevented them from resuming work.
Analysis
Analysis was conducted in four phases. First, the patients who did and those who did not complete the follow-up evaluation were compared on the basis of the available injury and treatment data. Second, the general health status outcomes in our study sample were compared with age and gender-matched norms from a general United States population sample; 95% confidence intervals were calculated around the sample mean for each SF-36 scale. Third, bivariate analysis was conducted to examine the primary outcomes according to different patient, injury, and treatment characteristics. A chi-square statistic was used to test for associations between the independent variables and the dichotomous outcome variable (stair-climbing ability); a Student t test or analysis of variance was used for the continuous outcome variables (SF-36 physical function, Sickness Impact Profile ambulation, American Medical Association range-of-motion impairment, and visual analog scale pain scores). Finally, multivariate regression techniques were used to simultaneously control for the influence of different patient, injury, and treatment characteristics on outcome. Because the ambulation scores, range-of-motion impairment ratings, and pain scores were all positively skewed, log scores of each were used in the linear regression models. Stair-climbing was modeled with use of logistic regression. All outcomes were modeled separately as a function of the type of fracture wound, AO/OTA fracture classification, treatment method, presence of a contralateral injury, mechanism of injury, bilaterality of the pilon fracture, age, gender, marital status, education, presence of comorbidities, personal annual income, health insurance status, duration to follow-up, and site of initial treatment. An interaction term was constructed between treatment method and site to determine whether there were significant differences in the outcomes of patients who were treated with the nondominant treatment method at each site.
The final models included all participants who had sustained either a unilateral or a bilateral pilon fracture because the results were similar regardless of whether the models included only patients with a unilateral fracture or both those with a unilateral fracture and those with a bilateral fracture. Differences were considered significant when p was =0.05. However, because of the small sample size and the limited ability to detect smaller but potentially meaningful differences in outcome, differences were also noted when p was =0.10.
Of the 103 patients who met the study inclusion criteria, eighty (78%) completed a follow-up health status evaluation. Of those eighty, the vast majority (88%) completed both the interview and the functional status assessment. Ten participants completed a telephone interview but not the functional assessment. The mean duration of follow-up was 3.2 years (range, two to five years). Of the twenty-three patients who did not complete the follow-up evaluation, twenty-two could not be located and one refused to participate. The twenty-three patients who were not followed were significantly more likely than the patients who were followed to have sustained a closed fracture that was stabilized with open reduction and internal fixation (53%; p < 0.05). No other differences were noted between those who did and those who did not complete the follow-up evaluation.
The mean age of the participants at the time of the follow-up evaluation was forty-four years (range, nineteen to seventy-two years). Participants were more likely to be male (78%), to be a high-school graduate (64%), and to have health insurance (79%) at the time of the evaluation. Patient characteristics did not vary significantly by treatment method (see Appendix).
Overall, study participants were more likely than not to have sustained a closed pilon fracture (61%) and to have sustained an AO/OTA Type-C pilon fracture (74%). Only six (8%) of the participants sustained a bilateral pilon fracture, and nine (11%) had a major injury of the contralateral lower extremity. Participants treated with external fixation with or without limited internal fixation were significantly more likely to have sustained an open fracture (53%) and an AO Type-C fracture (90%) than were participants treated with open reduction and internal fixation (26% and 60%, respectively) (p < 0.05).
None of the participants underwent delayed amputation (i.e., after initial surgical treatment) during the index hospitalization. However, five underwent late amputation (after hospital discharge). One patient underwent an above-the-knee amputation, and the other four underwent a below-the-knee amputation. All five patients had sustained an AO Type-C fracture, and four of the five fractures were open. All of the late amputations were performed in patients in whom the fracture had been treated with external fixation with or without limited internal fixation. None of the patients had major complications during the initial hospital stay; however, four of the five patients had a severe wound infection that required operative treatment before the late amputation.
As reflected by the SF-36 scales, the poorest domain of the participants' health was role disability due to physical health problems (
Fig. 1 )
21 . The greatest differences in health between the study sample and the United States age and gender-specific norms were in role disability due to physical health problems (mean difference = 35.6) and physical function (mean difference = 21.9). Moreover, for all except three of the SF-36 scales (vitality, role disability due to emotional health, and mental health), the scores for the study sample were significantly worse than the general population norms (p < 0.05).
Figure 2 shows a comparison of the average SF-36 physical function score of the study sample with those of other clinical samples
21,31-36 .
Of the sixty-five participants who were employed at the time of injury, twenty-eight (43%) were not working at the time of follow-up. More than two-thirds (68%) of those not working stated that the pilon fracture prevented them from being able to work. Approximately one-third of the participants reported notable difficulty with ankle stiffness (35%), swelling (29%), or pain (33%). Twenty-five percent of the participants reported that they usually wore an orthotic device, and 13% reported that they usually used a walking aid (e.g., a cane, crutches, or a walker). Participants also demonstrated moderate-to-severe difficulty with the performance of many activities that depend on lower-extremity function. Difficulties included not being able to complete full toe excursion (52%) and not being able to stand with the weight borne by the injured leg for thirty seconds (71%) (see Appendix).
Overall, the results of the bivariate and multivariate analyses were fairly consistent, with two exceptions. First, marital status was not found to have a strong relationship with any of the outcomes in the bivariate analyses. However, in the multivariate analyses, marital status was significantly related to the physical function, Sickness Impact Profile ambulation, range-of-motion impairment, and pain scores. Second, a bilateral pilon fracture was associated with poorer physical function, a worse ambulation score, and greater range-of-motion impairment in the bivariate analyses but was not found to be a significant factor in the multivariate models.
In the multivariate analyses, several patient characteristics demonstrated strong relationships with two or more of the selected outcomes (see Appendix). Married patients reported worse health and demonstrated more range-of-motion impairment than did patients who were not married at the time of the interview. Patients with a lower income level or a lower level of education were significantly more likely to report and/or demonstrate poorer health and function than were patients who had more financial resources and education. All of the primary outcomes that were studied, except for range-of-motion impairment, were significantly poorer in the presence of two or more comorbidities.
The only injury or treatment characteristic that was significantly related to several of the selected outcomes was treatment method. After controlling for other patient and injury characteristics, participants treated with external fixation with or without limited internal fixation had more overall range-of-motion impairment and reported more pain and ambulatory dysfunction than did participants treated with open reduction and internal fixation (p < 0.05). The average range-of-motion impairment rating of patients treated with external fixation with or without limited internal fixation (27%) was more than twice as high as that of patients treated with open reduction and internal fixation (12%). The average pain score of patients treated with external fixation with or without limited internal fixation was 25.1 points higher (worse) than that of patients treated with open reduction and internal fixation. Although the general physical health of the two treatment groups was not significantly different, the average Sickness Impact Profile ambulation subscale score was 19.8 points higher (poorer) for patients treated with external fixation with or without limited internal fixation. We found no significant interaction effect between treatment method and enrollment site, so it was not included in the final models (Table E-6 in Appendix).
This study demonstrated that overall midterm outcomes after pilon fractures are not good. In general, study participants reported relatively poor physical and psychosocial health outcomes as measured by the SF-36. The Sickness Impact Profile ambulation scores reflected substantial dysfunction in basic walking ability at two to five years after injury. When asked to perform various lower-extremity activities, such as ascending and descending stairs reciprocally or running in place for thirty seconds, one-third to one-half of the participants were unable to do so. Moreover, only 57% of the patients who had been employed at the time of injury were working at the time of follow-up.
Previous studies have documented conflicting results regarding clinical and functional outcomes after pilon fracture. Rüedi and Allgöwer
1 reported that 71% of their patients treated with open reduction and internal fixation had a good-to-excellent result four years after injury. Ninety-two percent of their study participants had returned to work at the time of follow-up. However, only 6% of their patients had sustained an open fracture compared with 39% in our sample. Ovadia and Beals
37 documented a 65% rate of good-to-excellent results after the treatment of 145 pilon fractures. Only 20% of the fractures were open. Many other authors have reported poorer overall results. For example, Teeny and Wiss
38 reported that 50% of their fifty-eight patients had a poor clinical result after treatment of a pilon fracture with open reduction and internal fixation. Sands et al.
17 reported results that were remarkably similar to ours, in a sample of patients evaluated with the SF-36 at the time of follow-up after open reduction and internal fixation of a pilon fracture.
In addition to the overall results, the influence of patient, injury, and treatment characteristics on midterm outcomes was examined in our study. We chose outcomes that represented a wide spectrum, including global health, specific lower-extremity functional activities, limitation of motion, and pain. In addition, the selected outcomes were determined by a combination of clinician-rated and patient-reported measures.
Regarding patient characteristics, the presence of two or more comorbidities resulted in significantly poorer scores for all selected outcomes except range-of-motion impairment. This finding is in agreement with those of a previous study of the effects of comorbidities on long-term health outcomes of injured patients
23 . Patients with a lower income level or lower level of education were significantly more likely to report and/or demonstrate poorer health and function than were patients who had more financial resources and education. The relationship of health with socioeconomic status and education has been well documented in the literature and may reflect differences in access to appropriate medical and social services
39-41 . In the present study, patients who were married at the time of the follow-up interview reported worse health than did those who were not married. This finding was unexpected and contrary to findings reported in the literature
42-44 .
Patients treated with external fixation with or without limited internal fixation reported and demonstrated significantly poorer results for three of the five primary outcomes—walking ability, range-of-motion impairment, and pain—than did patients treated with open reduction and internal fixation. These findings are somewhat contrary to the results reported by Wyrsch et al.
45 , who used a randomized-surgeon design to compare open reduction and internal fixation with external fixation with or without limited internal fixation. Although the differences were not significant, Wyrsch et al. found that patients treated with open reduction and internal fixation tended to have worse clinical scores. However, it is difficult to directly compare their study with ours because of the difference in the severity of the pilon fractures between the two treatment groups in their study and because we assessed outcomes differently.
Other studies have demonstrated significant differences in outcomes based on the clinical characteristics of the fracture or its treatment, such as the degree to which anatomic reduction was achieved
37,38 . For example, Teeny and Wiss
38 reported a 37% rate of good-to-excellent results after the treatment of Rüedi Type-I and II fractures compared with a rate of only 13% after the treatment of Type-III fractures. Ovadia and Beals
37 reported that 89% of their patients in whom the reduction was rated as good had a good-to-excellent clinical result. Conversely, all of their patients with poor fracture reduction had a poor clinical result. In the present study, both the fracture wound and the fracture type were found to be significantly associated with two or more of the selected outcomes in the bivariate analyses. However, in the multivariate analyses, neither was significantly associated with at least two outcomes. This may be the result of the small sample size or our inclusion of only severe injuries. These findings emphasize the importance of controlling for other factors when judging differences in outcomes associated with different treatment techniques.
The following limitations must be kept in mind when interpreting the results of this study. First, the study was based on a relatively small number of patients. Although a host of patient, injury, and treatment factors may influence health outcomes, especially more global ones, this study had limited statistical power to detect those factors. For example, the bivariate analyses clearly demonstrated that patients with a bilateral pilon fracture fared significantly worse than did patients with a unilateral pilon fracture. However, this difference did not remain significant in the multivariate analyses. We attribute this to a lack of statistical power.
Second, this was an observational, not a randomized, study. Not all patients received the same treatment at each study site, and we were unable to ascertain the specific criteria for treatment selection for each patient. Although we tried to account for variations in injury severity and patient characteristics when we compared outcomes according to the treatment method in our multivariate regression analyses, the small sample size limited our ability to adequately address potential treatment bias that may have been present in our cohort. Therefore, the differences in range-of-motion impairment, walking ability, and pain that we found between the treatment methods must be interpreted with caution and examined more rigorously in future studies.
A third limitation of the present study is that it did not address postdischarge complications (except for late amputation). Because a large proportion of our patients received care at other facilities during a relatively long follow-up period, we chose not to rely solely on the complications noted in the medical records of the initial treating centers and the participants' recall at the time of the follow-up interview. We judged those sources to be incomplete and inaccurate. It should be noted, however, that previous investigators have reported frequent and serious limb and systemic complications associated with pilon fractures
38,46-48 .
Fourth, we did not evaluate the degree of restoration of articular congruity after operative treatment because it is our opinion that the methodology of reviewing postoperative radiographs after treatment of pilon fractures is inherently flawed because hardware often obscures the details of articular alignment.
Finally, the generalizability of the findings of the present study is limited to intra-articular fractures managed at trauma centers that routinely treat patients who have sustained a high-energy fracture. All surgery was closely supervised or directly performed by an experienced attending orthopaedic traumatologist. The results of this study may therefore underestimate the disability following the treatment of these fractures in other settings.
In summary, we used well-validated outcomes instruments and other parameters to conduct a comprehensive evaluation of midterm outcomes after treatment of pilon fractures at two trauma centers. We found that patients who sustain a pilon fracture continue to experience major physical and psychosocial health problems long after the initial injury. Although certain patient characteristics can mitigate these poor outcomes, individuals who sustain such injuries do not enjoy the same physical or psychosocial health as do adults in the general population. Additional, prospective study is warranted to find better ways to treat these severe injuries so that functional outcomes can be improved.
Tables showing patient and injury characteristics, according to the treatment method, of the eighty patients who completed the follow-up evaluation; symptoms and functional activities at the time of follow-up; outcomes scores according to patient, injury, and treatment characteristics; and parameter estimates for selected outcomes are available with the electronic versions of this article, on our web site at www.jbjs.org (go to the article citation and click on "Supplementary Material") and on our quarterly CD-ROM (call our subscription department, at 781-449-9780, to order the CD-ROM).