Studies have implicated foot posture and foot function as risk factors for lower extremity pain. Empirical population-based evidence for this assertion is lacking; therefore, the purpose of this study was to evaluate cross-sectional associations of foot posture and foot function to lower extremity joint pain in a population-based study of adults.
Participants were members of the Framingham Foot Study. Lower extremity joint pain was determined by the response to the National Health and Nutrition Examination Survey–type question, “On most days do you have pain, aching or stiffness in your (hips, knees, ankles, or feet)?” The Modified Arch Index classified participants as having planus, rectus (referent), or cavus foot posture. The Center of Pressure Excursion Index classified participants as having overpronated, normal (referent), or oversupinated foot function. Crude and adjusted (age, sex, and body mass index) logistic regression determined associations of foot posture and function to lower extremity pain.
Participants with planus structure had higher odds of knee (odds ratio [OR] 1.57, 95% confidence interval [95% CI] 1.24–1.99) or ankle (OR 1.47, 95% CI 1.05–2.06) pain, whereas those with a cavus foot structure had increased odds of ankle pain only (OR 7.56, 95% CI 1.99–28.8) and pain at 1 lower extremity site (OR 1.37, 95% CI 1.04–1.80). Associations between foot function and lower extremity joint pain were not statistically significant except for a reduced risk of hip pain in those with an oversupinated foot function (OR 0.69, 95% CI 0.51–0.93).
These findings offer a link between foot posture and lower extremity pain, highlighting the need for longitudinal or intervention studies.
Lower extremity (LE) joint pain is highly prevalent. Estimates suggest that up to 40% of women and 30% of men have experienced nonminor musculoskeletal pain in the past month ([1, 2]), with older adults, women, and obese individuals at greater risk of LE joint pain ([3-6]). Suspected, but lesser known, risk factors for LE joint pain may include foot posture (i.e., planus, cavus, and rectus) and foot function (i.e., overpronated, oversupinated, and normal) as individuals with LE joint pain are more likely to have planus (low-arched) ([7-13]), cavus (high-arched) ([8-11, 14-17]), overpronated ([7, 8, 18-23]), or oversupinated ([8, 20, 24, 25]) feet. However, associations with foot posture or foot function to risk of LE joint pain are not universal ([10, 26, 27]).
Evidence regarding foot posture or foot function as an LE joint pain risk factor is sparse, conflicting, and typically evaluates specialized populations (e.g., athletes [[10, 11, 18, 20, 21, 24-27]], military trainees [[8, 10, 22]], and patients [[12, 14-16]]). A retrospective study of triathletes reported that risk of an overuse injury was increased 4-fold in those with an oversupinated foot function (), while a similar study of professional baseball players showed no association between LE joint pain to foot posture or foot function (). A prospective study of United States Naval Special Warfare trainees noted that planus, cavus, and abnormal foot and ankle function were associated with LE injuries (), while a similar study of Australian Defense Force Academy military trainees showed no association with foot posture and LE pain (). Part of the discrepancy in results may reside in the specific populations evaluated in these studies, as studies of foot posture and foot pain present conflicting results depending on the level of physical activity of the study participants ([14, 15]).
Biomechanical evidence also supports the notion that feet with extremes of foot posture (i.e., planus or cavus) or foot function (i.e., overpronated or oversupinated) are associated with altered LE functional alignment ([11, 21, 28]) and muscle activation patterns ([29, 30]). These studies suggest overpronated foot function yields excessive internal rotation of the shank with excessive anterior pelvic tilt ([23, 31]) as well as altered knee movements and excessive joint stress (), whereas a cavus foot may add excessive stress at the ankle due to limited ankle eversion range of motion ().
While studies have reported associations of foot posture and foot function to LE joint pain, these studies have restrictive participant inclusion criteria ([8, 10-12, 14-16, 18, 20-22, 25, 28]), failed to adjust for common intrinsic factors of LE pain ([11, 18, 19]) (e.g., sex, body mass index [BMI], and age [[3-6]]), or utilized a convenience sample of healthy participants to test external devices, such as orthotics, to experimentally manipulate foot posture or function ([28, 29]). As a result, the generalizability of these results to the adult population may be compromised. Therefore, the purpose of this study was to evaluate the cross-sectional associations of foot posture and foot function to LE joint pain in a population-based study of adult men and women.
Box 1. Significance & Innovations
Planus foot structure is associated with increased odds of knee and ankle pain.
Cavus foot structure is associated with increased odds of ankle pain or pain at 1 lower extremity (LE) site.
Extremes of foot function are not associated with increased odds of LE pain.
SUBJECTS AND METHODS
Study participants are members of the population-based Framingham Foot Study (), which for this analysis included the Framingham Heart Study and Offspring () and community cohorts. The community cohort consists of Framingham residents, recruited between 2002–2005 through random digit dialing from town lists, who were age >50 years and not members of the Framingham Heart Study.
Hebrew SeniorLife and Boston University Institutional Review Boards approved the Framingham Foot Study and all participants signed an informed consent prior to enrollment. Between 2002 and 2008, participants in the Framingham Foot Study received a plantar pressure and loading assessment during standing and walking, and they were queried regarding musculoskeletal pain.
Inclusion criteria in this analysis included those with complete data on LE joint pain, bilateral measures of foot posture and function, and covariate data. Participants with LE amputations were excluded.
LE joint pain
LE joint pain was determined by the response to the US-based National Health and Nutrition Examination Survey–type question, “On most days do you have pain, aching or stiffness in your (LE joints)?” LE joints queried included the hips, knees, ankles, and feet. Participants' response options were as follows: 1) yes, right side only, 2) yes, left side only, 3) yes, both sides, and 4), no pain on either side. Unilateral and bilateral pain at each site were weighted the same, with responses dichotomized to yes or no. To evaluate widespread LE joint pain, we created variables as follows: 1) pain in 1 LE site, 2) pain in 2 LE sites, 3) pain in ≥3 LE sites, and 4) no LE pain (referent).
Foot posture and function
A Matscan pressure mat (Tekscan) was used to collect data regarding foot posture and function during quiet bipedal standing and during gait, respectively. The Matscan system is a 43.6 × 36.9 cm system with 2,288 sensels for a resolution of 1.4 sensels/cm2 that is placed on the floor and has a height of 0.2 cm. Matscan data were recorded at 40 frames per second. With each participant, the Matscan was calibrated by having the individual stand on the mat to set it to the individual's body weight.
To determine foot posture, participants were asked to stand in their usual bipedal stance and the pressure under the foot was recorded on the Matscan system. From the quiet bipedal pressure data, each foot was individually analyzed to calculate its Modified Arch Index (MAI) value (Figure 1). To determine the MAI value, the foot, not including the toes, is divided into 3 equal parts, and the pressure in the middle third is divided by the total pressure in all 3 foot regions (). The MAI is correlated with other measures of foot posture, notably navicular height ().
To determine foot function, participants were asked to walk barefoot at their usual pace over the Matscan pressure mat using the two-step method (). The two-step method involves participants stepping on the pressure mat on the second step and has a similar reliability, as well as reduces the participant burden as they only have to take a minimum of 4 steps for the assessment, when compared to the midgait approach (). Prior to the gait data collection, participants practiced walking using this protocol to become familiar with the requirements of the assessment in order to ensure that the collected data were representative of the participant's normal gait. Each trial was visually inspected by the examiner and if the participant seemed to deviate from his or her gait (e.g., overreaching) to target the pressure mat, the participant had another trial collected. There were 2 trials collected, 1 for each foot. From the walking data, each foot was analyzed to calculate its center of pressure excursion index (CPEI) value (Figure 1). To determine the CPEI value, a line is drawn from the first and last points of each foot's center of pressure trajectory and the distance of the center of pressure curve at the distal third of the foot from the constructed line is recorded. This value is normalized by foot width and multiplied by 100 to obtain a percentage excursion of the center of pressure (). CPEI has been shown to be sensitive to changes in clinical measures of static foot alignment ().
As each participant had 2 values of CPEI and MAI, the more extreme MAI and CPEI values, relative to median, were used in the analysis, similar to prior studies of bilateral conditions ([38, 39]). To classify foot posture and function, the distributions of MAI and CPEI values used were divided into quintiles. Feet in the top and bottom 20% of the MAI values were defined as planus and cavus, respectively, with the middle 60% as the rectus feet. Feet in the top and bottom 20% of CPEI values were defined as oversupinated and overpronated, respectively, with the middle 60% classified as having normal foot function.
In this analysis, covariates included sex, age, and BMI. Weight was measured with the participant barefoot using a standardized balance beam scale and recorded to the nearest 0.5 pound (0.23 kg). Participant height, without shoes, was measured in inches to the nearest 0.25 inch (6.35 mm) using a calibrated stadiometer. Weight and height were used to determine BMI. Although smoking and alcohol use are available data with the Framingham Foot Study, a similarly aged population-based study of joint pain showed that these covariates were not significant risk factors of LE pain (); therefore, we did not adjust for these in our analysis.
Means and SDs or frequencies of participants' characteristics variables were calculated for included participants. To assess odds of LE pain at the individual joints (e.g., hip pain, knee pain) by foot structure and foot function, logistic regression analyses were used for each joint. To assess odds of LE pain at the specific individual joint (e.g., hip pain only, knee pain only) by foot structure and foot function, a multinomial logistic regression analysis was used. To evaluate if widespread LE pain was associated with foot posture or foot function, multinomial logistic regression using number of sites of LE pain was used, with the referent being individuals with no LE pain. Because LE pain is thought to have common patterns (e.g., knee and foot pain) (), a multinomial logistic regression evaluated the patterns of pain in individuals with 2 sites of LE pain.
In addition to these crude models, adjusted models were evaluated for all the regressions analyses. Adjusted models included covariates of age, BMI, and sex ([3-6]). Further, as prior studies have used either a normal or cavus () foot posture as the referent population, associations between the extremes of foot posture were evaluated, as well as between extremes of foot function. All statistical analyses were conducted using the SAS statistical analysis package, version 9.3, with alpha set to P less than or equal to 0.05.
Of the eligible 2,188 participants in the Framingham Offspring and community cohorts, there were 1,856 participants included in this study, with 944 (50.9%) experiencing LE pain (Table 1). There were 332 participants excluded for missing data (327 were missing CPEI data and 5 were missing covariate data). Of the LE sites assessed, knee pain was the most common at 29.4% (n = 546), followed by foot pain at 26.2% (n = 487), hip pain at 17.5% (n = 325), and ankle pain at 11.2% (n = 207). There were 525 (28.3%), 256 (13.8%), and 163 (8.8%) participants who had pain in 1, 2, or ≥3 LE sites, respectively.
Table 1. Participant characteristics*
Study participants (n = 1,856)
Values are the mean ± SD or the number (percentage). BMI = body mass index; MAI = Modified Arch Index; CPEI = Center of Pressure Excursion Index; LE = lower extremity.
63.8 ± 8.9
28.6 ± 5.6
0.115 ± 0.101
13.9 ± 9.95
Hip pain only
Knee pain only
Ankle pain only
Foot pain only
1 LE pain site
2 LE pain sites
≥3 LE pain sites
No LE pain
Individuals who had a foot with an MAI value >0.164 were considered to have planus feet, whereas a score <0.030 was considered to be cavus, with rectus feet being in between these cut points (Table 2). Individuals who had a CPEI <7.30 were considered to have overpronated feet, whereas a score >21.00 was considered to be oversupinated feet, with normal foot function being in between these cut points.
Table 2. Modified Arch Index (MAI) values to define pes planus, cavus, and rectus foot posture, and Center of Pressure Excursion Index (CPEI) values to define overpronated, oversupinated, and normal foot function
−25.30 to 7.30
Individuals with a planus foot posture were more likely to have knee pain (odds ratio [OR] 1.57, 95% confidence interval [95% CI] 1.24–1.99) and ankle pain (OR 1.47, 95% CI 1.05–2.06) compared to individuals with a rectus foot posture (Table 3), with CIs widening after adjusting for age, sex, and BMI. In the multivariate analysis to evaluate the effects of foot posture on a single site of pain (Table 4), results were similar, with individuals with planus foot posture having higher odds of knee pain only (OR 1.94, 95% CI 1.38–2.74) and ankle pain only (OR 5.04, 95% CI 1.29–19.7). After adjusting for age, sex, and BMI, the odds of knee pain only remained significant (OR 1.52, 95% CI 1.05–2.19), while the odds of ankle pain only were reduced (OR 3.58, 95% CI 0.88–14.6) with wider CIs. Individuals with planus feet showed increased odds of a single LE site, 2 LE sites, and ≥3 LE sites of pain, relative to a rectus foot; after adjustment, ORs were attenuated (Figure 2). When pain patterns were evaluated in individuals with 2 LE sites of pain, there were no significant differences in odds of LE pain by foot posture; however, there was an overall tendency of higher odds of pain for individuals with planus feet and lower odds of pain with cavus feet (see Supplementary Tables 1 and 2, available in the online version of this article at http://onlinelibrary.wiley.com/doi/10.1002/acr.22049/abstract).
Table 3. Crude and adjusted odds ratios (ORs) with 95% confidence intervals (95% CIs) of associations of foot posture and foot function to lower extremity pain at any joint
Individuals with cavus foot posture had a 5.93 (95% CI 1.59–22.1) increased odds of ankle pain and after adjustment the increased odds were maintained (OR 7.56, 95% CI 1.99–28.8). Relative to rectus feet and after adjusting for age, BMI, and sex, cavus feet had higher odds of LE pain at 1 site (OR 1.37, 95% CI 1.04–1.80).
Individuals with an overpronated foot function had lower odds of hip pain (OR 0.69, 95% CI 0.51–0.93) relative to those with a normal foot function, but after adjustment the association was reduced. In general, compared to individuals with normal foot function, individuals with overpronated or with oversupinated foot function had similar odds of pain at the LE sites assessed. Further, odds of a single LE site, 2 LE sites, and ≥3 LE sites of pain showed no significant associations in the crude or adjusted models, regardless of referent population. When pain patterns were evaluated in individuals with 2 sites of LE pain, odds of both hip and knee pain were decreased with an oversupinated foot function (OR 0.49, 95% CI 0.25–0.98); after adjustment these associations were attenuated (OR 0.56, 95% CI 0.28–1.13).
The purpose of this cross-sectional study was to evaluate the associations of foot posture and foot function to LE joint pain in a population-based study of adult men and women. Relative to rectus foot posture, those with a planus foot had greater odds of knee and ankle pain, whereas those with a cavus foot posture had increased odds of ankle pain. Further, individuals with a cavus foot type had greater odds overall of having 1 site of LE joint (i.e., hips, knees, ankle, or foot) pain. Foot function during gait, as measured by CPEI, showed no associations to LE joint pain; however, those with an oversupinated foot were at reduced odds of having hip pain as well as concurrent hip and knee pain. The results suggest that having extremes of foot posture (cavus or planus) predisposes individuals to a greater risk of LE joint pain, whereas extremes of foot function (overpronated or oversupinated) show little effect on odds of LE joint pain. Since foot posture is modifiable using foot orthoses (), and other common risk factors (e.g., increasing age, sex, obesity [[3-6]]) are less modifiable, these findings may have implications for the prevention and/or treatment of knee and ankle pain in adults.
Our results suggest that foot posture (planus and cavus feet) rather than foot function (overpronated and oversupinated feet) may be an independent risk factor for LE joint pain. Similar to our work, prior research suggests that knee pain is associated with a planus foot ([8, 10-12]), while both cavus () and planus () foot types are associated with ankle injuries and pain. While some studies report similar findings ([8, 10-12]), our work is counter to others ([18-22, 24, 25]). Our results may differ from these prior findings; however, many of the differences are mitigated through study design characteristics ([18, 20-22, 24, 25]), such as using restrictive inclusion/exclusion criteria and referent populations ([18, 19]).
The novelty in understanding our results in relation to prior work does not lay in comparisons with prior cross-sectional, retrospective, or even prospective studies, but in shoewear and orthotic intervention studies, especially with regard to knee pain, which is a major contributor to disabilities in adults ().
A common aspect of nonpharmacologic intervention studies of LE pain is the use of orthotic devices and shoewear modifications for these patients (). Lateral wedge orthotics for patients with medial compartment knee osteoarthritis (OA) is a common, conservative treatment strategy. Lateral wedge orthotics shift the center of pressure laterally to reduce knee adductor moments (), which has been linked to medial knee OA progression and knee joint pain ([46, 47]). Correspondingly, individuals with a planus foot posture display a more medial center of pressure when standing, relative to cavus or rectus foot postures (), which may explain the increased risk of knee pain within the planus foot posture noted within our study. These lateral wedge orthotics that shift the center of pressure laterally decrease use of nonsteroidal antiinflammatory drugs and yield better orthotic compliance compared to those with the neutral-wedged insole for the populations with symptomatic knee OA ().
Lateral wedge orthotics can also increase gluteus medius muscle activity (), which is altered in those with knee () or hip pain (). Increases in gluteus medius muscle activity stabilize the hip and control lumbopelvic motion (). Causes of the muscle activation changes with orthotics are unknown; however, orthotic use may influence plantar tactile sense () and foot alignment or posture () to enhance muscle activation and gait (). In light of our current findings, these studies highlight the need to evaluate the mechanism by which orthotics and shoewear interventions affect the kinetic chain, starting with how foot posture and function influence alignment, function, and pain in the hips, knees, ankles, and feet.
Although we found associations of foot posture and function to LE joint pain, these results should be evaluated with consideration to the study's strengths and limitations. First, our analysis was a cross-sectional evaluation of foot biomechanics and regional foot pain, meaning that causal relations cannot be inferred. While LE pain, particularly knee pain with joint space narrowing (), may cause planus feet through a change in knee alignment as compensation for the pain, results of other studies suggest foot posture is a predictor of knee pain (). Further, given that the MAI is noted as the pedobarographic measure to use as a surrogate for foot structure (), it is not a direct measure of arch structure or forefoot–rearfoot alignment, which may also be an important component of foot posture (). Further, neither the MAI nor the CPEI have clinically defined cut points to determine foot posture or foot function. However, our 20% cut points are similar to or more restrictive than the proposed cut points of cavus and planus feet using the MAI, which are noted as lower than 0.047 () or lower than 0.0926 () for cavus feet and greater than 0.158 () or greater than 0.1778 for planus feet (). Given the lack of clinically derived cut points, using locomotor extremes (such as the top and bottom 20% of MAI and CPEI values) may provide evidence of structure–function relationships (). Further, although other studies have suggested that multiple trials may be required for a reliable measure using the Tekscan Matscan (), given the large participant numbers only 1 scan per foot was collected. However, in prior work on the Framingham Foot Study, it was noted that with a reliability of 0.50 (Metscan reliability measures noted as >0.70) (), a sample size of <300 participants was needed to appropriately power the study (). Therefore, due to the higher sample size of our study, we are confident that the findings noted are not an artifact of the single scan of each foot and that the study was adequately powered given the single assessment of foot posture and foot function. These results suggest that foot posture is associated with LE pain, and further work using radiographic and/or kinematic measures to assess foot posture and its relation to LE pain is warranted.
The strengths of this study include its participants, data collection, and analysis. The Framingham Foot Study is a well described, population-based cohort of adults evaluating foot posture, foot function, and LE joint pain. The range of participant demographics (e.g., men and women, ages 36–100 years, and BMI range 15.3–58.5 kg/m2) provides greater generalizability of the study results. Moreover, the MAI is a simple, clinically relevant measure of foot posture that can identify whether individuals have planus feet () and are at an increased risk of knee, ankle, or widespread LE pain. Although confirmatory studies (including longitudinal studies) of foot posture and LE joint pain are needed to substantiate our cross-sectional results, this work suggests that foot posture should be an important consideration in evaluating LE joint pain risk.
In conclusion, planus foot posture, as measured by the MAI, is associated with greater risk of knee and widespread LE pain, while cavus foot posture is associated with ankle joint pain. Foot function as measured by the CPEI shows no association with LE joint pain. These findings offer preliminary support of a mechanistic link between foot posture and LE pain. This study also points to a need for longitudinal studies to evaluate the role of foot posture in LE pain and intervention studies to evaluate if corrective orthotics or shoewear can be used to prevent or treat LE pain.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be submitted for publication. Dr. Hannan had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Riskowski, Hillstrom, Hannan.
Acquisition of data. Hannan.
Analysis and interpretation of data. Riskowski, Dufour, Hagedorn, Hillstrom, Casey, Hannan.