A review of the foot function index and the foot function index – revised

Reliability was estimated with Cronbach’s Alpha and Rasch person reliability statistics. Both indices reflect the proportion of variance of the person scores or measures to total variance (i.e., including measurement error). Unlike Cronbach’s Alpha, Rasch person reliability is based on the estimated locations of persons along the measurement continuum, excluding those with measures reflecting extreme (zero or perfect) scores and including cases with missing data. For both indices, our criterion for acceptability was .80.

One requirement of the Rasch model is monotonicity: the requirement that, as person ability increases, the item step response function increases monotonically [20]. This means that choosing one categorical response over the prior—for example, moving from selecting “2 = A little of the time,” to selecting, “3 = Most of the time,”—increases with person ability. The proper functioning of the rating scale is examined using fit statistics, where: (i) outfit mean squares should be less than 2.0, (ii) average measures advance monotonically with each category, and (iii) step calibrations increase monotonically [21, 22].

Table 12 describes the utility of the FFI and FFI-R in studies of foot function measures and includes 17 articles. Category A New Instruments. Includes four articles in which foot health measures are described including the original FFI [7], the FFI-R [11]. The FFI Side to Side was derived from pain and disability subscales of the FFI [23]. The Ankle Osteoarthritis Scale (AOS) [24]; measured foot problems related to foot and ankle osteoarthritis. Agel et al. [25] modified the rating scale of the FFI pain and function subscales from the visual analog rating scale (VAS) to the Likert categorical scale; this modification was tested in a sample of individuals with non-traumatic foot complaints, and the metric of the Likert scale was valid. Category B FFI as Criterion Validity. Articles in this category describe several health measures and use the FFI full scale or subscales to validate these measures. Bal et al. [26] found a strong correlation of FFI scores and scores of RA functional measures: the Health Assessment Questionnaire (HAQ) and Steinbrocker Functional Class (SFC). SooHoo et al. [27] found that the Rand 36-Item Short Form Health Survey (SF-36) scores of a sample of individuals with foot and ankle disorders were moderately correlated with FFI scores and concluded that FFI scores can be used to monitor the quality of life of these patients. Shrader et al. [28] measured the stability of navicular joint alignment and found that this measure correlated well with the FFI scores of the sample. Helliwell et al. [29] developed a new measure, the Foot Impact Scale (FIS), to measure the impact of foot problems on foot health in a sample of individuals with RA; the metric of FIS was validated with the FFI and HAQ. In an RA study, van der Leeden et al. [30] reported that Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) and Disease Activity Scores in 44 joints (DAS 44) were correlated with FFI scores; furthermore, this author discerns the correlations that the FFI pain subscale scores correlated with forefoot pain while the FFI function subscale scores correlated with hindfoot problems. The FFI scores were also used as validation measures of the American Orthopedic Foot and Ankle Society (AOFAS) clinical rating scales, an instrument that was widely used by foot and ankle surgeons [31]. These validation studies were reported by Baumhauer et al. [32] for the AOFAS hallux clinical rating scale and by Ibrahim et al. [33] for the AOFAS clinical rating scale, which was well to moderately correlated with FFI scores. The latter finding was based on his study with a 41% response rate in a sample consisting of 45 individuals. Category C Cultural Adaptation or Translation. The first translation of the FFI was the Dutch-language instrument known as Dutch FFI-5pts [3]. The German-language translation of the instrument is the FFI-G [34]; the FFI was also translated into Brazilian Portuguese [35], Taiwan Chinese [36], Turkish [26], and Czech [37]. There was also a Spanish translation conducted by the MAPI Institute in Lyon, France [38]. These translations complied with rigorous language translation procedures; occasionally, some item adjustments of the scales were needed. In summary, the FFI was developed with good reliability and validity; it also inspired and served as criterion validity for newer foot health measures and attracted the attention of researchers around the world, who conducted translations and adaptations of the tool into their native languages and cultures.

Table 6 is a supplement to Table 5 and displays the clinimetrics of the instruments listed in Table 5; measures were metrically good, with reliability and validity values greater than 0.7 with one exception where the pain subscale had a reliability of 0.64 [3].

The FFI is one of the outcome measures most frequently used by AOFAS members [31]. It was first used to measure surgical outcomes. The surgical interventions and outcomes are summarized in Table 7. There are 30 articles, categorized generally according to type and location of surgical procedure. Five distinct procedural categories were identified as follows: (a) arthrodeses within the foot or ankle [39‐47], (b) arthroplasty within the foot or ankle [48‐51], (c) fracture care of the foot or ankle [52‐55], (d) deformity reconstruction surgery of the foot or ankle [56‐60], and (e) various surgical interventions for chronic conditions [61‐64]. The FFI was also used to assess outcomes of less invasive procedures, such as calcaneal spur treatment by arthroscopy [37], distal tibia repair using fixation with cannulation osteosyntheses [65], arthroscopic chondrocyte implant of the tibia and fibula [66], and surgical interventions for complex ankle injuries [67]. In summary, the FFI and the Dutch FFI-5pts appeared to be useful in measuring outcomes of various surgical procedures in children, adults, and individuals with acute, chronic, and congenital foot and ankle problems.

Table 8 lists studies using foot function outcome measures in orthotic interventions in the foot and ankle. The studies assessed the impact of orthotic treatment on forefoot, midfoot, and hindfoot/ankle pathology. Orthotic treatment on the forefoot in patients with RA improved the scores for pain, disability and activities [68, 69], however the scores were unchanged in the study by Conrad et al. [70]. Other studies using special shoes and shoe inserts showed symptoms of relief in hallux valgus pain [71] hindfoot and forefoot problems [72, 73]; and slowing the progression of hallux valgus in early RA [74]. Midfoot studies assessing the treatment of full length orthoses on pain relief [75], and mobility were performed using the FFI-R as an outcome measures [76]. For hindfoot conditions treatment with orthoses included studies of heel pain [77], plantar fasciitis [35, 78, 79], stabilizing hindfoot valgus [80], correction of posterior tibialis tendon dysfunction [81], destructive hemophilic arthropathy of the foot and ankle [82] and juvenile idiopathic arthritis of the foot and ankle [83]. Shoes/shoe inserts have also been found to relieve foot and ankle pain from arthritides [84, 85]. In summary, the FFI and FFI-R clearly provided useful outcome measures for orthotic management of a wide range of foot and ankle disorders.

The FFI also was used to measure foot health outcomes associated with medical interventions (Table 9), such as cortisone injection of the ankle adhesive capsulitis [86]; the injection resulted in improved FFI pain and disability subscale scores. Di Giovanni et al. [87] measured the outcome of stretching exercises for plantar fasciitis versus Achilles tendonitis; both groups showed improvement in FFI pain subscale scores. Kulig et al. [88] used the FFI pain and disability subscales to measure the outcomes of exercise intervention in posterior tibial tendon dysfunction. Rompe et al. [89] reported the FFI pain score improved in the stretching treatment group of a randomized clinical trial using stretching and shockwave therapy to treat patients with plantar fasciopathy. Overall, the FFI was useful in measuring the outcomes of conservative interventions in chronic foot and ankle conditions.

Investigators had chosen the FFI scores or the subscale scores to determine the prevalence and disease burden of foot and ankle conditions in the general population (Table 10). Novak et al.[4] used FFI scores to evaluate type 2 diabetes with and without neuropathy and identified that group with neuropathy had worse FFI scores. Williams and Bowden [90] correlated high FFI scores to foot morbidity in rheumatic diseases, and estimated cost of care/staffing concerns for that patient subset. Williams [91] also used the FFI scores in patients with Paget’s disease and noted the impacts on plantar foot pressures, gaits, and ambulation abilities. Kamanli et al. [92] correlated the scores of the FFI and foot bone mineral density, then extrapolated these scores to that individual’s skeletal bone density. Kavlak and Demitras [93] reported a strong correlation of FFI scores with the scores of VAS pain scale, foot pain scale (FPS), and hindfoot function scale (HFS) in patients with foot problems. Goldstein et al. [94] noted that FFI scores of individuals with previous foot injuries had a high correlation with 6 other foot function instruments. Rosenbaum et al. [95] found that plantar sensory impairment of the foot in patients with RA was correlated with poor FFI scores. Schmiegel et al. [96] found that pedobarograph scores of patients with RA with foot pain were correlated with poor FFI and HAQ scores. In summary, FFI scores were useful in detecting the prevalence of foot and ankle problems and as a measure of concurrent validity for other foot health measures in various chronic foot conditions.

In all, we found the FFI instrument was frequently chosen as an outcome measure of surgical, orthotic, and medical treatments, but its application was wider than we originally imagined. It was not limited to outcome measures; FFI scores were also applied in the promotion of foot health as a common public health issue and in increasing the awareness of health system administrators. The FFI was also used in the validation of newly developed foot health measures.

For the FFI-R L (68 items) [11], person reliability was high: 0.96, respectively. In the PCA, 56.8% of the variance was explained by the measure, with only 10.6% of the variance explained by the first factor of residuals. These findings support that the full FFI-R meets the unidimensionality requirement of the Rasch model. Further, the criterion for unidimensionality was a ratio of the raw variance in the first contrast of residuals that was 5.4 (i.e., greater than 3). For the FFI-R S (34 items) [11], person reliability was 0.95, similar to the reliability estimates of the FFI-R L. The PCA of the FFI-R S revealed that unidimensionality criteria were also satisfied. This supports the use of a short form of the measure, because the item response burden on patients is lower, at 34 questions. Because this measure is as reliable as the full measure, its use is supported for clinical settings.

All subscales of the FFI-R had strong person reliability estimates (Table 11), ranging from 0.78 to 0.94 for person reliability. The PCA indicated that unidimensionality held for each subscale, with the exception of the stiffness subscale. Further inspection of the data revealed that the two-factor solution reflected groups of the low-severity and high-severity items and was not the result of a competing factor. Unidimensionality for the limitation subscale was met after dropping item 41 (ASSISTO), an item listed in the FFI-R database. Overall, the subscales of the FFI-R satisfied unidimensionality criteria and were reliable measures of the latent traits (Table 11).

The response category analyses for each of the subscales (done after collapsing Categories 5 and 6) revealed that, for the first three subscales (pain, stiffness, and difficulty), the response categories behaved as required by the Rasch model. However, for the subscales of limitation and social issues (both of which are time scales), there was some indication that respondents had difficulty distinguishing between, “2 = A little of the time,” and, “3 = Some of the time.” We considered, then, collapsing these categories and making all FFI-R subscales have four possible response categories. This would ensure uniformity of the measure and decrease the burden on patient response. Therefore, the first three subscales, which measure severity, “3 = Severe pain,” “4 = Very severe pain,” and “5 = Worst pain imaginable,” were collapsed. This was justified because all three captured the notion of severe pain. Overall, the analyses showed that the response to each item functioned well with the four-item response categories.