Session IX - Foot & Ankle

Comparison of SF-36 and SMFA in Recovery from Fixation of Unstable Ankle Fractures

William T. Obremskey, MD; Ouida Brown, MS; Douglas R. Dirschl, MD; Robin Driver, RN, WakeMed, Raleigh, NC

Purpose: The purpose of this study was to assess whether the Short Musculoskeletal Functional Assessment (SMFA) questionnaire had fewer floor or ceiling effects than the SF-36 in evaluation of the functional recovery of operatively stabilized isolated ankle fractures. We also hypothesized that the SMFA would be more effective than the SF-36 in discriminating the functional recovery of different groups of patients after an ankle fracture.

Methods: 126 English-competent patients with isolated unstable ankle fractures that had surgical stabilization were identified from a large private community orthopaedic practice and an academic trauma practice. Demographic and radiographic data were gathered from charts and review of radiographs. Patients were contacted and were asked to complete SF-36 and SMFA questionnaires during a phone interview an average of 27 months ± 17 (SD) months (range 6-64 months) after injury. The mean age of the patients was 48 ± 18 (SD) years (range 17-85 years); 54 patients (42.5%) were male and 73 patients (57.5%) were female.

Subscale scores of the SMFA and SF-36 were compared based on sex and OTA fracture classification with an ANOVA analysis of P <0.05 considered significant. The SMFA and SF-36 subscale scores were also compared using Pearson correlation with r = 0.228, P <0.01 considered significant.

Results: The SF-36 demonstrated floor effects (> 5% of scores at poorest possible score) in 2 of 8 categories (role limits-physical function and role limits-emotional function). It also showed ceiling effects of (> 5% of scores at best possible score) in 6 of 8 categories (physical functioning, role limits-physical, emotional, social function, mental health, and bodily pain). The SMFA did not demonstrate floor effect in any category. Ceiling effects were seen in 4 of 6 categories (daily activities, arm/hand function, mobility, and bother index). Both the SF-36 and SMFA subcategory scores showed significant skewness (P >0.05) using the Shapiro-Wilke W test for normalcy, indicating a non-parametric score distribution. Nearly all Pearson correlations of the SMFA and SF-36 subcategory scores were significant (P >0.01).

SMFA/hand function subscale scores were at the maximum in 75% of patients, and all scores were in the top quartile of possible scores. This would be expected in an ankle fracture population with an isolated lower extremity fracture. Women had lower SF-36 physical function scores compared to men (P = 0.004). They also had poorer SMFA daily activity scores (P = 0.03). Women also had subscale scores that tended to reflect poorer outcomes in the SMFA mobility and dysfunction index (P = 0.052) and (P = 0.096) respectively. Significantly poorer scores were seen in patients with 44C type fractures than among those with 44 A/B fractures in the SF-36 physical function and role limits-physical subscales (P = 0005 and P = 0.0089 respectively). Subscale scores in the SMFA daily activities, dysfunction index and bother index were also significantly different from Type 44 A/B fractures (P = 0.05, P = 0.05 and P = 0.065) respectively.

Discussion: The floor and ceiling effects of the SF-36 among patients with musculoskeletal conditions has limited its usefulness in orthopedic outcome research. In a spectrum of musculoskeletal conditions of overuse, injuries or arthritis, the SMFA has been shown to have fewer floor and ceiling effects. This study demonstrated that with an isolated injury of an ankle fracture requiring operative stabilization that the SMFA has no floor effects in assessing functional recovery and approximately equal ceiling effects. The ceiling effects are mostly due to the patient's return to baseline function at 27 months after injury. This is seen in the skewness data showing a preponderance of scores in the high range of function.

In convergent construct validity, scores correlate, as one would expect them to in relation to standards. The high Pearson correlation of the SMFA and SF-36 scores indicate that the SMFA adequately assesses general health, as the SF-36 is the most widely accepted measure of general health and supports the convergent construct validity of the instrument.

The SMFA and SF-36 also correlate well in discriminant construct validity in relation to historical norms of ankle fracture outcome. Historically, women and patients with 44 Type C fractures have had poorer clinical outcomes. This was also shown in this ankle fracture population. It was somewhat surprising to find that the SF-36 subscale scores were sensitive enough to identify groups of patients that may have poorer function.

Conclusions: The SMFA had fewer floor and ceiling effects in long-term follow-up of patients with an isolated unstable ankle fracture. The SMFA and SF-36 both appear to be able to assess the patient's general health, and both instruments identify patients who had significant problems with functional recovery. The SMFA might be a more effective single instrument to track a patient's functional recovery than the SF-36 if it is able to assess general health and identify patients that are having poorer functional activity without significant floor or ceiling effects.