Predictors of Patient-Reported Function and Pain Outcomes in Operative Ankle Fractures

Introduction

Ankle fractures are common musculoskeletal injuries, with a reported incidence of 187 per 100 000 person-years. ⁶ An epidemiologic study of Swedish patients shows that ankle fractures continue to increase at a rate of 0.2% per year, with even more rapid growth in the elderly, with Kannus et al showing a 164% increase in women older than 60 years from 1970 to 2000.^12,18 Open reduction and internal fixation (ORIF) is generally accepted as the treatment for closed, unstable ankle fractures. ORIF is performed to restore the anatomic alignment of the ankle mortise and minimize the risk of joint arthrosis in hopes of improving patient functionality.^3,15,22 Despite operative management, previous literature has reported that many patients continue to complain of subjective discomfort and functional limitations 5 years postoperatively. ¹⁶

Prior studies have identified patient and injury characteristics associated with poor functional outcomes in the short term following operative treatment of ankle fractures. Patient factors associated with poor short-term functional outcome included lower education level, smoking history, female sex, older age, and American Society of Anesthesiologists (ASA) score. Injury characteristics including fracture-dislocation, medial malleolus fractures, and bimalleolar and trimalleolar fractures were also associated with a worse functional outcome.^4,7,21 However, very few studies have evaluated the outcomes of operatively treated ankle fractures outside of the year immediately following surgery. The few published studies do not evaluate preoperative patient characteristics associated with reduced functional outcome.^16,19

The purpose of this study was to determine the predictors of functional outcome in operatively treated ankle fractures. To our knowledge, there are no other studies that have specifically evaluated the effect of preoperative risk factors on functional outcome at follow-up of greater than 2 years. Additionally, no previous studies have been published assessing predictors of functional outcomes in ankle fractures using Patient Reported Outcomes Measurement System (PROMIS), which has been shown to be reliable and has been validated in the lower extremity.^10,11

Methods

After obtaining approval by our institutional review board, we retrospectively reviewed a multicenter cohort of all operatively treated ankle fractures from January 2001 to January 2013. Patients 18 years of age and older with a minimum follow-up of 2 years were included in the study. The exclusion criteria included no valid contact information, open fracture, prior ankle surgery, pilon variants, fractures related to Charcot arthropathy, and Maisonneuve fractures. Pilon variants were excluded, as they are generally associated with more articular destruction and are difficult to compare to ankle fractures. Maisonneuve injuries are primarily a ligamentous injury and are also difficult to compare to ankle fractures. Fractures that underwent operative treatment greater than 3 weeks from initial injury were also excluded to prevent confounding of the analysis with chronic injuries.

A total of 463 patients were eligible for inclusion and were contacted by phone or email. Patients were administered the PROMIS physical function v1.2 (PF) and PROMIS pain interference v1.1 (PI) computerized adaptive tests (CATs) through the PROMIS website or by phone. PROMIS was developed by the NIH and uses CATs based on item response theory, meaning that a patient’s response to a previous question determines the subsequent question that he or she is asked from a larger bank of questions. ⁵ Physical function measures the ability to carry out activities that require physical actions, and includes mobility, dexterity, axial function (head and neck), and activities of daily living. A higher PF score indicates higher physical function. Pain interference measures the degree to which pain limits or interferes with the patients’ physical, mental, and social activities. A higher PI score indicates greater pain interference with activities. Each PROMIS score is standardized with a range of 0-100, a mean of 50, and a standard deviation of 10 based on the general US population.

A total of 142 patients completed the PROMIS instruments and were included in the analysis with 64 women and 78 men (Figure 1). Mean age of the cohort was 52.7 years (range, 23-86). Mean follow-up was 6.3 years (range, 2-14). A chart review was performed on the patients to record preoperative risk factors, including sex, age, diabetes, smoking, ASA class, body mass index (BMI), education level, and energy of injury. Fractures related to the mechanisms of motorcycle crash, motor vehicle crash, pedestrian struck, and falls from greater than 10 feet were considered high energy. All other fractures were considered low energy. The distinction of insulin- and non–insulin-dependent diabetes was not available in our chart review and was not recorded. Information about peripheral neuropathy related to diabetes was also unavailable. Preoperative and immediate postoperative radiographs were evaluated and fractures were classified by the AO/OTA classification system. ¹⁴ Presence of fracture-dislocation was also recorded from review of injury radiographs. Intraoperative radiographs and operative reports were reviewed to determine whether there were posterior malleolus fractures or syndesmosis injuries associated with each ankle fracture. Information on any postoperative complications, defined as any need for reoperation, were obtained and recorded from chart review. Hardware, including syndesmotic screws, was not routinely removed in our cohort. Thus, return to the operating room for removal of symptomatic hardware was considered a complication.

OPEN IN VIEWER

Figure 1.

Study flow chart.

Patients were treated with a similar postoperative protocol, which consisted of placement of a splint for 2 weeks. Patients were then transitioned to a functional brace with range of motion exercises and continued non-weight bearing on the operative extremity until 6 weeks postoperatively. Patients were examined clinically and radiographically at 2 weeks, 6 weeks, 12 weeks, and 1 year postoperatively.

Statistical analysis was performed with IBM SPSS Statistics (version 22). In all cases, a P value of <.05 was deemed statistically significant. Preliminary analysis with Student t tests for continuous variables and Pearson correlation coefficients for categorical variables was used to identify risk factors that had a statistically significant association with PF and PI scores. The independent predictive value of these significant risk factors was then analyzed using multivariable linear regression analysis to control for covariate and confounding factors. The results of our regression models are presented with unstandardized B values and standardized β values indicating the magnitude of effect of each independent variable on the linear regression analysis.

Results

Mean PF score was 51.9 (SD = 10.0) and mean PI score was 47.8 (SD = 8.45). Twenty patients (14%) had complications. Nine patients (6.3%) had removal of symptomatic hardware at 1 year after surgery. One patient (0.7%) had a nonunion of a medial malleolus fracture that required reoperation. Two patients (1.4%) had a deep vein thrombosis, 1 (0.7%) had a pulmonary embolism, 2 (1.4%) had a superficial wound infection, and 1 (0.7%) had deep wound infection requiring an operative debridement. There were 8 (1.7%) known mortalities in the possible cohort, with none being directly related to the operative intervention. However, 256 (55.3%) patients could not be reached for follow-up, and current status is unknown.

There was one AO/OTA 44-A fracture, 117 44-B fractures, and 24 44-C fractures. There were no significant differences in PF or PI score between different fracture classifications (Table 1). The presence of a syndesmosis injury did not result in differences in PF score (P = .41) or PI score (P = .86). The presence of a posterior malleolus fracture also did not result in a worse PF score (P = .38) or PI score (P = .64). Furthermore, the presence of a medial malleolus fracture did not result in differences in PF score (P = .16) or PI score (P = .40).

OPEN IN VIEWER

Table 1.

Ankle Fracture Characteristics.

		Significance (P Value)
	Percentage	Physical Function	Pain Interference
AO/OTA
44-A	1	.18	.43
44-B	82
44-C	17
Syndesmosis	43	.41	.86
Medial malleolus	40	.16	.40
Posterior malleolus	28	.38	.64
Ankle dislocation	35	.95	.98
Energy of injury
Low High	8614	.28	.25

Female sex, older age, diabetes, higher BMI, and higher ASA class were associated with lower PF scores (Table 2). Smoking, education level, energy of injury, and ankle dislocation were not associated with significant differences in PF scores. Diabetes, higher ASA class, and lower BMI were associated with higher PI scores. Sex, age, smoking, education level, energy of injury, and ankle dislocation were not associated with significant differences in PI scores.

OPEN IN VIEWER

Table 2.

Risk Factors.

		Significance (P Value)
Risk Factor	Percentage	Physical Function	Pain Interference
Sex
Male	53	.02	.63
Female	47
Age	58 (19-85) a	<.01	.95
Diabetes	8	.02	.07
Smoking	16	.95	.29
BMI	28 (20-44) a	<.01	<.01
ASA class
1	52	<.01	<.01
2	45
3	3
Education
High school	7	.43	.48
Attended college	16
College degree	30
Graduate degree	47

Abbreviations: ASA, American Society of Anesthesiologists; BMI, body mass index.

a

Reported as mean (range).

After controlling for covariate and confounding variables, multivariable regression showed that only higher ASA class (P = .003), higher BMI (P = .008), and higher age (P = .025) were independently associated with decreased PF scores (Table 3). ASA had the largest magnitude of effect on PF scores with a B of −10.3 (95% CI, –3.56-–17.0) (Figure 2). Only higher ASA class (P = .001) and lower BMI (P = .014) were independently associated with increased PI scores (Table 4). ASA also had the largest effect on PI score with a B of 11.5 (95% CI, 4.67-18.4) (Figure 3).

OPEN IN VIEWER

Table 3.

Independent Risk Factors for Increased PF Score.

Variable	B	β	P Value	95% CI
ASA	−10.301	−0.542	.003	−3.559, −17.044
Age	−0.163	−0.25	.025	−0.304, −0.021
BMI	−0.436	−0.255	.008	−0.753, −0.119

Abbreviations: ASA, American Society of Anesthesiologists; BMI, body mass index; PF, PROMIS physical function v1.2.

OPEN IN VIEWER

Figure 2.

Coefficient of effect (B) for independent predictors of PF score. ASA, American Society of Anesthesiologists; BMI, body mass index.

OPEN IN VIEWER

Table 4.

Independent Risk Factors for Increased PI Score.

Variable	B	β	P Value	95% CI
ASA	11.512	0.605	.001	4.67, 18.4
BMI	−0.410	−0.240	.014	−0.735, −0.085

Abbreviations: ASA, American Society of Anesthesiologists; BMI, body mass index; PI, PROMIS pain interference v1.1

OPEN IN VIEWER

Figure 3.

Coefficient of effect (B) for independent predictors of PI score. ASA, American Society of Anesthesiologists; BMI, body mass index.

Discussion

Open reduction and internal fixation for ankle fractures has been shown to have excellent outcomes, with Stufkens et al reporting a mean AOFAS score of 86.6 at a 13-year follow-up in type 4 supination external rotation injuries. ¹⁷ However, satisfactory outcomes are not universal. Shah et al showed that up to 50% of patients still have significant pain 5 years postoperatively following fixation of ankle fractures, and 39% of patients report that they have not completely recovered functionally. ¹⁶ Our study, using PROMIS PI and PF scores, revealed a mean PROMIS PF score of 51.9 and a mean PI score of 47.8, suggesting that patients have pain and function similar to population-based controls.

Previous studies have shown that short-term function after operative ankle fractures is associated with preoperative patient and injury risk factors. Bhanderi et al identified an association between smoking, lower education level, and the presence of a medial malleolus fracture with lower SF-36 scores at 3 months postoperative, whereas Egol et al found that diabetes and ASA class were associated with decreased AOFAS scores at a 1-year follow-up.^4,7 Fracture-dislocations have also been identified as being predictive of poorer short-term functional outcomes. ²¹ We found that only higher ASA class, older age, and higher BMI were associated with decreased physical function (Table 3) whereas higher ASA class and lower BMI were associated with increased pain interference (Table 4) at a minimum follow-up of 2 years. Although previous studies found an association between diabetes and physical function, our study found no association between diabetes and physical function or pain interference after controlling for BMI and ASA class.

Our study found ASA class to have the strongest magnitude of effect on both PI and PF scores. ASA class has previously been associated with increased hospital readmission rates and increased adverse events postoperatively following operative fixation of ankle fractures and fractures of other areas in the lower extremity.^1,2 Functional outcomes have also been shown to be influenced by ASA score following arthroplasty of both the hip and knee. ⁹ The findings of our study reinforce the importance of ASA as a useful predictor of postoperative outcomes following orthopedic surgery.

Studies looking at intermediate to long-term outcomes have analyzed differences in functional outcomes based on injury characteristics alone and have not included patient characteristics.^17,19 Verhage et al found a significant difference in AOFAS scores between unimalleolar and bimalleolar fracture patterns at an average follow-up of 9.6 years. ¹⁹ Our study did not find fracture pattern or any other injury characteristics to significantly impact PROMIS PI or PF scores. However, we did find that patient factors such as ASA class and obesity were strong independent predictors of poor functional outcomes using a validated outcome measure.

The presence of a posterior malleolar fragment did not negatively impact functional or pain outcomes in our study. However, we did not evaluate the effect of the size of the posterior malleolar fragment on PI and PF scores. Hong et al reported that increasing size of the posterior malleolar fragment is correlated with worse functional outcomes at a 1-year follow-up. ⁸ Further investigation on the impact of posterior malleolar fragment size on functional outcomes is warranted.

Van Vligmen et al found that at an average of 5.1 years’ follow-up, 60% of patients with syndesmotic injury reported pain and 26% had evidence of degenerative changes. ²⁰ However, in our cohort, syndesmosis injury was not associated with a difference in function. Our finding is similar to the findings reported by Litrenta et al in a multicenter study of 242 SE-4 ankle fractures at a 1-year follow-up. ¹³

We believe that the use of patient-reported outcomes (PROMIS) is a strength of our study. PROMIS has previously been validated in the lower extremity and has been proven to be reliable and flexible. ⁸ Additionally, PROMIS results are not limited to the lower extremity like the AOFAS and Olerud-Molander scores, which allows for comparison across varying extremities and injuries. Additionally, PROMIS has been shown to perform better than other previous legacy scales such as Foot and Function Index (FFI) and Foot and Ankle Ability Measure (FAAM). ¹¹

Limitations of our study include the retrospective design and the unstandardized length of follow-up. Future studies following a cohort prospectively from the immediate postoperative period to long-term follow-up would be helpful. Furthermore, the study of a traumatic injury did not allow us to establish preinjury function levels, which would have allowed us to examine which factors were associated with the largest decreases in functional scores after ORIF of the ankle. Another limitation of our study is the low rate of follow-up for our patients at 31%, which could lead to selection bias. Possible reasons for the relatively low rate of follow-up include the transient nature of trauma patients and the lack of cellular phone contact information for some of the patients from early in the cohort. Of the 199 patients we were able to contact, 142 (71.4%) agreed to participate in the study. Finally, we did not perform a physical examination or obtain radiographs at follow-up and therefore can only make conclusions based on patient-reported outcomes. The addition of these to any future studies would be valuable.

Conclusion

In summary, sex of the patient, diabetes, smoking status, education level, AO/OTA fracture classification, syndesmotic injury, presence of a posterior or medial malleolar fractures, presence of a fracture-dislocation, and energy of the injury did not show an association with functional outcomes at a minimum of 2 years’ follow-up. Factors that were associated with decreased physical function included higher ASA class, age, and BMI. Factors that were associated with increased pain included lower BMI and higher ASA class. ASA class had the largest effect on postoperative outcomes. Patients with these characteristics may need to be counseled more extensively on expected outcomes following operative fixation of ankle fractures. Patients with operative ankle fractures had pain interference and physical functions scores similar to population means at follow-up of greater than 2 years postoperatively.