Variations in foot posture are thought to influence the function of the lower limb and may therefore play a role in predisposition to overuse injury [1–4]. Despite these observations, there is still considerable disagreement regarding the most appropriate method for categorizing foot type [5]. A wide array of techniques have been used, including visual observation [3, 6], various footprint parameters [7, 8], measurement of frontal plane heel position [9, 10] and assessment of the position of the navicular tuberosity [11].

Recently, a six-item criterion reference tool (the Foot Posture Index, or FPI) was developed in response to a requirement for a quick, easy and reliable method for measuring foot position in a variety of clinical settings [12]. The FPI consists of six validated, criterion-based observations of the rearfoot and forefoot of a subject standing in a relaxed position. The rearfoot is assessed via palpation of the head of the talus, observation of the curves above and below the lateral malleoli and the extent of the inversion/eversion of the calcaneus. The observations of the forefoot consist of assessing the bulge in the region of the talo-navicular joint, the congruence of the medial longitudinal arch and the extent of abduction/adduction of the forefoot on the rearfoot [12].

The concurrent validity of the FPI has been investigated fully and reported previously [12]. A more recent study has also demonstrated good internal construct validity and fit of the scoring system to the Rasch model, a useful statistical model of the uni-dimensionality (capacity to measure a single construct) and scale stability (or linearity across a range of values) of a measure [13]. The FPI is suitable for a range of clinical applications and yields high quality linear metric data [13]. The original authors now recommend the use of the six item FPI tool, replacing the eight item version reported previously [14, 15].

The FPI has been used in a variety of clinical and research settings. The applications of the FPI include studies of biomechanical risk factors for neuropathic ulceration in diabetes [16], identifying foot type as a basis for screening subjects as inclusion or exclusion criteria in clinical research [17, 18], investigating the relationship between foot types and risk factors for sports and training injuries [19–21], investigating whether foot posture is associated with falls in older people [22] and as a means of assessing age-related differences in foot structure [23].

One of the limitations of the FPI is that, to date, there have been no normative data available for comparison and reference. The aim of this study therefore, was to establish normative FPI reference values for use in research and to assist in clinical decision making.

The FPI is only one of a number of measures of foot posture currently available. Razeghi and Batt [5] discuss the current measures available based on foot morphology and classify them according to four categories: visual assessment, anthropometric values, footprint measures and radiographic appraisal. To date, there are only two foot posture measures – the arch index [7] and the rearfoot angle [10] – for which valid normative data are available. The FPI is the only approach that captures information about standing foot posture in multiple foot segments without a requirement for complex measurement techniques.

The FPI has now been employed in several studies and median FPI raw scores for normal samples have been reported to lie consistently around +5 [19, 25]. Other studies have confirmed this tendency towards normal feet as being pronated rather than 'neutral' [20, 21]. The current study, employing a large sample indicates that in the normal adult population the mean (back-transformed) FPI score is +4, confirming that a slightly pronated foot posture is the normal position at rest.

Statistically determined reference ranges for postural variations such as standing foot position are inherently wide, so must be used as a general guide only in interpreting FPI scores in a clinical context. It is recognised that clinically, relatively minor variations from the mean may increase risk of mechanically induced pathology, although the strength of these relationships have not been confirmed scientifically and certainly vary for different pathological groups. Except for foot postures falling clearly outside the normal range, the reference ranges alone are probably not adequate for clinical decision making.

There was some evidence of age-related variation in mean foot posture scores and this is in agreement with previous studies. In the recent study by Scott et al [23], a sample of older adults (mean age 80.2 ± S.D. 5.7) had more pronated foot postures than a group of younger adults (mean age 20.9 ± S.D. 2.6). A tendency toward more pronated foot postures in younger children is also well documented. A flatter, more pronated foot has been reported in young children as a consequence of the process of development of the longitudinal arch [8]. The values reported in this study of FPI normative values support the notion of a U-shaped relationship between age and foot posture reported by Staheli et al [8].

While age was found to have an effect on foot posture there was no evidence of any systematic difference between the FPI scores of males (logit mean = 2.3, SD = 2.4) and females (logit mean = 2.5, SD = 2.3). This is again in agreement with the longitudinal arch study by Staheli et al [8] who found minimal differences between male and female foot postures. Although studies have been conducted to analyse foot morphology based on gender [26], studies investigating gender differences in foot posture are limited and our data suggest that gender related differences are small enough to be considered negligible.

The current study found no relationship between BMI and the FPI. Previous studies undertaken using measures such as the footprint angle (FA) and the Chippaux-Smirak index (CSI) have reported lowered longitudinal arches, a broader midfoot area and subsequently flatter feet in people with high BMI values [27]. However, the studies reporting BMI related differences have exclusively used footprint measures, and the postural data may be confounded by the effect of body adiposity on the interpretation of arch height based on these footprint estimates. Indeed, it has been suggested previously that footprint parameters are a measure of "fat feet" rather than "flat feet" [28].

It is known from empirical observation and previous studies that foot posture differences may be encountered in association with underlying disease processes or functional pathology. Comparison of the FPI scores from the normal sample with data from participants known to have identified pathology revealed variations consistent with those predicted by theory. The group with neurogenic pes cavus (mean FPI logit score = -2.78, SD = 2.32) and idiopathic pes cavus (mean = -2.63, SD = 1.25) had FPI scores significantly different from the normal population (mean logit score = +2.4) indicating that the FPI data was sensitive to disease-related postural changes. Data have also been reported elsewhere indicating the sensitivity of the FPI to postural change associated with pathological pes planovalgus (median FPI raw score = +12) [29]. Conversely, the otherwise healthy group with minor musculoskeletal symptoms (mean FPI logit score = 2.23, SD = 2.35) was not systematically different from the normal population (mean FPI logit score = 2.4, SD = 2.3), nor was a sample of patients with diabetes (mean = 2.14, SD = 2.96). There appears therefore to be scope for using FPI scores and associated normative values to help identify groups with structural pathology and to assist in the clinical decision-making process.

There are several limitations to this study that warrant discussion. The most compelling of these is that the data used did not come from a prospectively constructed random sample, such as a general practice or telephone directory derived random sampling frame. Such sampling methods are extremely resource intensive and financially costly whereas the retrospective compilation of a large sample from existing sources covering both normal and pathological subgroups was felt to be a realistic compromise between impact and resource. The dataset was compiled using data from nine centres which raises the possibility of some inconsistency in data collection. One centre had recorded age as a range rather than an integer in years, although sufficient detail was provided to allow classification according to the cut points provided in the analysis. Body mass index was also of less importance to some studies and was not recorded by all centres. However, all incomplete datasets were missing only variables informing the secondary analysis, and for variables of primary importance such as presence or absence of pathology, the dataset was complete.

Observations were derived from either FPI-6 total scores, or through the extraction of the six relevant items from studies using the older eight item version of the FPI. All observers were trained using the official FPI user manual, but it is acknowledged that minor variations in interpretation may have occurred and could not have been controlled for. Conversely, the use of data from multiple centres limits the potential for bias in the total sample and could be considered to enhance the validity of the results.

In summary, this study has provided a set of normative values for FPI scores in a healthy adult population. The data also provides mean and standard deviation values to act as comparators for future studies in a range of potentially pathological groups. Future studies defining FPI ranges of normal and abnormal explicitly according to resulting pathology would supplement this statistical definition and would be helpful to our understanding to the link between foot posture and mechanical 'overuse' type symptoms. The FPI scores did not vary systematically with gender, side of observation or BMI, although did vary at the extremes of age. FPI scores in groups with confirmed structural pathology were systematically different from normal, indicating some sensitivity of the instrument. This now requires further confirmation in specific pathological groups. Further investment in studies to determine definitive reference ranges for children and older adults may help to complete the picture.

Normative data for the FPI obtained from 619 healthy adults has been presented and compared with grouped data from 1,029 further observations. Based on the analyses presented here, it is concluded that foot posture is influenced by age and presence of pathology, but is not influenced by sex or BMI. The use of age-specific reference ranges provided in this paper will assist in classifying foot type for the purpose of research and clinical decision-making.