Non-Invasive Clinical Assessment of Medial Longitudinal Arch Morphology and Its Interrelationships in Foot and Ankle Musculoskeletal Pain: a Cross-Sectional Analytical Study

Abstract

Objectives

Accurate non-radiological assessment of the medial longitudinal arch (MLA) is essential for diagnosing foot deformities and guiding musculoskeletal pain management, yet no consensus exists on which method most reliably reflects MLA morphology in populations with ankle-foot pain, and comparative data across methods within the same clinical sample remain limited. This study evaluated and compared four non-radiological MLA measures and examined their interrelationships with ankle-foot musculoskeletal pain, physical activity and ankle-foot range of motion.

Materials and methods

A cross-sectional analytical study was conducted in 158 participants (66 males, 92 females; mean age 46.7±12.4 years) attending the Orthopaedic Outpatient Department of AIIMS Rajkot, India. Four measures were assessed bilaterally: medial longitudinal arch height (MLAH), arch height index (AHI), longitudinal arch angle (LAA) and Feiss line deviation; physical activity was classified with the IPAQ-SF (High PA ≥600 MET-min/week; Low PA <600 MET-min/week). Analyses included Shapiro-Wilk testing, Mann-Whitney U and t-tests, Kruskal-Wallis H with Dunn's post-hoc, Spearman correlations and hierarchical regression.

Results

Pes cavus was the predominant arch type (71.5%), followed by normal arch (22.8%) and pes planus (5.7%). All four measures showed good-to-excellent intra-rater reliability (ICC 0.931-1.000) and MLAH and AHI correlated very strongly (ρ = 0.846, p<0.001); Feiss-line agreement with the final foot type reference was definitional, since that reference incorporates Feiss-line data. Plantarflexion, subtalar inversion and eversion differed significantly across arch types (all p<0.001). MLAH, AHI and Feiss line correlated significantly with both the numeric rating scale (NRS) and American Orthopaedic Foot and Ankle Society (AOFAS) scores (all p<0.05). Pes cavus was associated with higher pain (NRS 7.0 vs 6.0, p_adj = 0.006) and lower function (AOFAS 70.0 vs 74.0, p_adj = 0.005) than normal arch. In regression, sex (OR = 2.869, p = 0.012) and body mass index (BMI) (OR = 1.129, p = 0.002) were the strongest independent predictors of high pain.

Conclusions

Non-radiological MLA assessment is reliable and clinically feasible. Pes cavus predominated in this ankle-foot pain population and was associated with restricted plantarflexion and increased subtalar inversion. Physical activity level was independently associated with pain intensity beyond arch morphology, but with higher, not lower, pain among more active participants; this cross-sectional association does not show that activity affects pain in either direction. A multimethod classification approach is recommended when arch type directly informs treatment.

INTRODUCTION

Longitudinal and transverse foot arches become apparent in late foetal life, though subcutaneous fat initially makes the newborn foot appear flat (1). Of the three arches, the medial longitudinal arch (MLA) is the highest, most mobile and clinically most important (2); it is visibly absent in toddlers and develops by about age five as ossification proceeds and the protective fat pad resorbs (3-6). The MLA abnormalities are among the most prevalent musculoskeletal conditions: pes planus affects roughly 15-50% of children and adults depending on age and criteria (7) and pes cavus an estimated 10-20% of the general population (8).

The MLA comprises a long relatively weak anterior pillar (first three metatarsals, three cuneiforms, navicular and talus) and a short stronger posterior pillar (medial calcaneus) (9), contacting the ground only at the first-to-third metatarsal heads and medial calcaneal tubercle, with its summit at the talocalcaneonavicular joint (10). Altered arch configuration disturbs load transmission, shock absorption, weight distribution and postural stability, and may cause chronic foot pain (15-17). Clinically, MLA dysfunction has been linked to plantar fasciitis (11), tibialis posterior tendinopathy (12), medial tibial stress syndrome (13) and patellofemoral pain syndrome (14). Flat foot causes excessive pronation and medial knee stress (18), while pes cavus causes poor shock absorption and instability, increasing fall risk and accelerating knee osteoarthritis via varus deformity (19, 20).

Radiological techniques, including weight-bearing radiographs, computed tomography (CT), magnetic resonance imaging (MRI) and 3D imaging, remain the gold standard but are limited by cost, radiation and accessibility; recent additions include deep-learning plantar-pressure mapping (21), AI/smartphone-based arch detection (22) and wearable inertial sensors (23). Non-radiological anthropometric methods (footprint indices, navicular height and drop tests, arch height index, and functional posture assessments) are inexpensive and reproducible but lack imaging precision and their correlation with specific pain syndromes remains unclear.

Despite this evidence, no consensus exists on which non-radiological method most reliably reflects MLA morphology in ankle-foot pain and comparative data across methods within the same population remain limited (24, 25). A structured comparison of modalities is therefore needed (Table 1). The aim of this study was to evaluate and compare non-radiological anthropometric methods of MLA assessment and to examine their interrelationships with ankle-foot musculoskeletal pain outcomes.

TABLE 1.

Assessment modalities of the medial longitudinal arch (MLA)

MATERIALS AND METHODS

Study design, setting and participants

This cross-sectional analytical study was a non-funded collaboration between the Departments of Anatomy and Orthopaedics at an Institute of National Importance. The unit of recruitment was the patient and the unit of measurement was the limb: each participant contributed one symptomatic (index) limb and one contralateral asymptomatic limb, both assessed for MLA morphology (316 limbs from 158 participants; Figure 1) – a within-participant design, not a case-control study of separate individuals. This is the phase-3 (final) assessment of a three-phase project in which the same 158 participants were recruited and followed from 1 August 2022 to 31 July 2025; phase-1 and phase-2 findings from this cohort were published separately (32, 33) and the measures, analyses and findings reported here are previously unpublished. Of 240 patients assessed for eligibility, 82 were excluded or declined, leaving 158 enrolled and retained across all three phases (Figure 1). Reporting followed the STROBE statement (34, 35); the checklist is in Supplement S1.

Participants were recruited by consecutive sampling of outpatients meeting eligibility criteria, assessed separately for each limb. Inclusion required age 14-70 years, plus musculoskeletal foot/ankle pain for the index limb or a current pain-free state for the contralateral limb. Exclusion criteria for both limbs were extreme obesity or emaciation; prior joint transplant or (asymptomatic limb) walking-aid use; congenital/acquired deformity, metabolic disorder, malignancy, pregnancy, or other debilitating disease; and any history of foot/ankle fracture or surgery. Only participants meeting all criteria for both limbs, with written informed consent, were enrolled. Three enrolled participants exceeded the 70-year upper age limit, a protocol deviation reported in the Results section.

FIGURE 1.

STROBE participant flow diagram. Participant recruitment, eligibility assessment, enrolment and bilateral non-radiological assessment of the symptomatic (index) limb and contralateral (asymptomatic) limb of each participant; pain (NRS) and function (AOFAS) were assessed on the index limb only. Reliability subsample (n = 30) randomly selected from enrolled participants. MLA=medial longitudinal arch; MLAH=medial longitudinal arch height; AHI=arch height index; LAA=longitudinal arch angle; ROM=range of motion; NRS=numeric rating scale; AOFAS=American Orthopaedic Foot and Ankle Society score; IPAQ-SF=International Physical Activity Questionnaire – Short Form; ICC=intraclass correlation coefficient; MDC₉₅=minimal detectable change at 95% confidence.

Sample size

Sample size was estimated from Keenan et al. (36) (18.4% prevalence of self-reported foot/ankle pain in adults over 55): with P = 18.4, Q = 81.6, Z = 1.96, L = 5%, n = Z²·PQ/L² ≈ 230.7, rounded to 231; this is a precision-based prevalence estimate, not a power calculation, and none was performed for the between-group comparisons. The recruitment target was raised to 240 to allow for anticipated exclusions; the 158 enrolled (65.8% of target) reflect screening attrition rather than loss of enrolled participants (Figure 1). The study is therefore estimation- rather than power-based and subgroup analyses, particularly the small pes planus (n = 9) and High PA (n = 17) groups, are exploratory and likely underpowered.

Ethics and registration

Ethical approval was obtained from the Institutional Ethics Committee of AIIMS Rajkot (Protocol ID: NF/15/2022; O.W. No.: AIIMS Rajkot/IEC/13/2022, dated 9 July 2022; Supplement S2). Procedures followed the Declaration of Helsinki (2013). Written informed consent was obtained from all participants; for minors (14-17 years), written assent plus parental consent were obtained. The study was prospectively registered with CTRI (CTRI/2023/10/058206).

Anthropometric measurements

Foot length, foot breadth and truncated foot length (TFL) were measured in weight-bearing stance with a MarCal 16 ER digital vernier calliper (±0.01 mm) (37). Foot breadth was the distance between the most medial point of the first metatarsal head and the most lateral point of the fifth metatarsal head. TFL was measured from the most posterior calcaneal load point to the first metatarsal head. Two independent measurements were averaged.

Physical activity assessment

Physical activity was assessed using the International Physical Activity Questionnaire – Short Form (IPAQ-SF). Participants scoring ≥600 MET-min/week (≥150 min/week of moderate-intensity activity, or equivalent) were classified as High PA, consistent with the WHO 2020 guidelines (38); those below 600 MET-min/week were classified as Low PA (38).

Non-radiological assessment of the MLA

All MLA measurements were performed by a single investigator blinded to pain-group allocation, bilaterally, in a standardised sequence (foot length, foot breadth, TFL, MLAH, LAA, AHI, Feiss line); bony landmarks were palpated and marked with a skin-safe dermographic pencil, and three measurements were averaged per continuous variable. Intra-rater reliability was assessed in a randomly selected subsample of 30 participants, with both limbs of each participant re-measured at a second session ≥48 hours later. Inter-rater reliability was not assessed.

The four MLA measures (MLAH, LAA, AHI, Feiss-line deviation) were obtained in weight-bearing stance as defined in Table 2 (definitions, instruments, thresholds, references (37, 39, 40)); the Feiss line joined the medial malleolus centre and first metatarsal head. MLAH was analysed as a continuous variable in mm: published adult norms (Table 2) were not used to assign arch type because the sample median (65.0 mm) is an order of magnitude above them and would classify virtually every limb as pes cavus. LAA followed the interior-angle convention (Option A; higher LAA = higher arch). The reference classification, final foot type (FFT), was assigned by the orthopaedic co-investigator on weight-bearing clinical examination (arch contour, navicular position relative to the Feiss line, forefoot/rearfoot compensation) – an integrated judgement rather than any single measurement. Because the Feiss line contributes to FFT, their agreement is expected by construction and does not validate the Feiss line; only the AHI and LAA concordances with FFT are informative, since these do not contribute to FFT.

TABLE 2.

Non-radiological medial longitudinal arch (MLA) measures and classification thresholds

Ankle-foot musculoskeletal pain assessment

Pain and function were assessed on the symptomatic (index) limb of all 158 participants (Figure 1) by the orthopaedic co-investigator, blinded to MLA results; the contralateral limb was pain-free by definition and not scored. Pain intensity used the 11-point numeric rating scale (NRS) (ICC = 0.96; MCID 2.0) (41, 42), categorised mild (1-3), moderate (4-6), severe (7-10); function used the AOFAS Ankle-Hindfoot scale (0-100, higher = better) (43). Pain location was mapped on a standardised foot diagram, and duration recorded as acute (≤four weeks), subacute (5-12 weeks), or chronic (≥13 weeks) and entered as a regression covariate.

Statistical analysis

Analyses used IBM SPSS v26.0, with significance at p ≤ 0.05 (two-tailed) and Bonferroni correction for multiple comparisons. Normality was assessed by Shapiro-Wilk; continuous data are mean ± SD or median (IQR). Group differences used t-test or Mann-Whitney U; categorical data used chi-square or Fisher's exact test. Reliability used the two-way mixed-effects ICC with absolute agreement (ICC3,1), interpreted per Koo & Li (44); SEM and MDC95 were computed. Correlations used Pearson r or Spearman ρ interpreted per Cohen (45). Arch-type differences used Kruskal-Wallis H with Dunn's post-hoc (adjusted α = 0.017). Concordance used percentage exact agreement and Cohen's/Fleiss' kappa per Landis & Koch (46). Hierarchical linear regression modelled NRS across Model 1 (demographics), Model 2 (+MLA measures) and Model 3 (+PA group), with VIF checks; secondary binary logistic regression modelled high pain (NRS ≥7).

RESULTS

Participant characteristics

Of 158 participants analysed (66 males, 41.8%; 92 females, 58.2%; mean age 46.7±12.4 years, range 16-72; mean BMI 26.79±5.26 kg/m²; Table 3), three (aged 71, 71, 72 years) exceeded the 70-year upper limit and were retained as a disclosed protocol deviation. Most were Low PA (n = 141, 89.2%) versus High PA (n = 17, 10.8%); groups were comparable in age (U = 1324.5, p = 0.481), BMI (U = 997.5, p = 0.261), and sex (χ² = 0.098, p = 0.754). Diagnoses included ankle-and-foot sprain (39.9%), ankle sprain (24.1%), foot sprain (7.0%), plantar fasciitis (5.7%), foot pain (5.1%), flat foot (4.4%), ankle pain (3.8%) and other conditions (10.1%). The High PA group had significantly shorter right foot length and TFL (Table 3).

Descriptive statistics of MLA measures

Shapiro-Wilk indicated non-normal distributions for MLAH, LAA, and Feiss line; AHI was near-normal on the right (W = 0.993, p = 0.503) but not the left. Right-side medians (Table 3) reflect the predominance of pes cavus; left-side medians were similar (MLAH 65.0 mm [58.0-67.0]; LAA 155° [150-163°]; AHI and Feiss deviation identical to the right), and bilateral symmetry was confirmed for all MLA measures (Wilcoxon, all p>0.05). Mean ROM was: dorsiflexion 19.9 ± 6.0°; plantarflexion 40.4 ± 12.2°; subtalar inversion 17.4 ± 5.5°; eversion 8.2 ± 2.9°.

TABLE 3.

Participant characteristics, medial longitudinal arch (MLA) descriptives and arch-type distribution (n = 158)

Arch-type distribution

Based on the FFT classification, pes cavus predominated (71.5%), followed by normal arch (22.8%) and pes planus (5.7%), with no significant difference in distribution between PA groups (χ² = 4.834, df = 2, p = 0.089) (Table 3; Figure 2).

FIGURE 2.

MLA measures by arch type group. Bland–Altman plots showing agreement between test and retest measurements for MLAH (mm), AHI and LAA (°). Solid line=mean difference; dashed lines=95% limits of agreement (±1.96 SD). Diamond markers represent individual participants. ICC=intraclass correlation coefficient (two-way mixed-effects, absolute agreement, ICC(3,1)); MDC₉₅=minimal detectable change at 95% confidence. Feiss line is not shown as perfect test–retest agreement (ICC=1.000) precluded Bland–Altman analysis.

Reliability

Intra-rater reliability (n = 30) was good-to-excellent for all measures (ICC 0.931-1.000; Table 4, Panel A; Figure 3), with perfect agreement for the Feiss line reflecting its categorical nature.

FIGURE 3.

Bland-Altman reliability plots (test-retest, n=30)

Correlations between MLA methods

MLAH and AHI correlated very strongly (right ρ = 0.846; left ρ = 0.840; both p<0.001); both correlated weakly-to-moderately with LAA and Feiss deviation correlated strongly and negatively with MLAH and AHI (Table 4, Panel B). Feiss-line classification agreed with FFT in 100% of cases (definitional, not an independent validation; agreement was lower for LAA (68.4%), AHI (58.2%) and MLAH (43.0%), showing the continuous measures do not always classify a given patient identically.

Correlations of MLA measures with pain outcomes

MLAH, AHI, and Feiss deviation each correlated weakly-to-moderately with both NRS and AOFAS (all p<0.05), in the direction of higher arch with more pain and poorer function, whereas LAA correlated with neither (Table 4, Panel C). Among ROM measures, only subtalar inversion correlated with pain (NRS ρ = +0.231, p = 0.004); dorsiflexion, plantarflexion and eversion did not (all p>0.05), though they differed by arch type.

TABLE 4.

Reliability of measures and correlations with methods and pain outcomes.

Pain and ROM across arch types

Plantarflexion, subtalar inversion, and eversion differed significantly across arch types (Kruskal-Wallis, all p<0.001), but dorsiflexion did not (Table 5, Panel A). NRS (p = 0.007) and AOFAS (p = 0.005) also differed; post-hoc, pes cavus had higher pain and lower function than normal arch (NRS p_adj = 0.006; AOFAS p_adj = 0.005), whereas comparisons involving pes planus were non-significant given the small subgroup (n = 9).

Regression analysis

In the final NRS model (R² = 0.184, adjusted R² = 0.140, F(8,149) = 4.203, p<0.001), male sex, BMI and High PA were significant, but no MLA measure was independently significant (Table 5, Panel B). Adding MLA measures did not improve the model (ΔR² = 0.045, p = 0.096), whereas adding PA group did (ΔR² = 0.024, p = 0.038). For AOFAS (R² = 0.183, p<0.001), sex and BMI were significant and PA group approached significance (p = 0.055). In binary logistic regression (NRS ≥7), male sex (OR = 2.869, 95% CI 1.262-6.521) and BMI (OR = 1.129, 95% CI 1.044-1.221) were the only significant predictors.

TABLE 5.

Range of motion (ROM) and pain outcomes across arch types and regression predictors

DISCUSSION

In this study of 158 orthopaedic outpatients, pes cavus predominated (71.5%), MLAH and AHI were very strongly concordant, arch type was associated with ankle ROM (particularly plantarflexion and subtalar inversion), and sex and BMI were the strongest independent predictors of pain, with PA group an additional NRS predictor.

The pes cavus prevalence (71.5%) markedly exceeds population estimates of 10-20% (8), while pes planus (5.7%) was below the reported 15-50% (7); this skew likely reflects the orthopaedic outpatient sampling frame and the MLAH thresholds applied, and should be interpreted as clinic-specific rather than community prevalence. High-arched feet show reduced shock absorption and subtalar mobility and greater susceptibility to plantar fasciitis, ankle sprains, and peroneal pathology (19), consistent with ankle/foot sprains accounting for ~71% of diagnoses in this cohort (19, 20). The non-significant cavus excess in the High PA group (94.1% vs 68.8%, p = 0.089; n = 17) warrants prospective confirmation.

The good-to-excellent intra-rater reliability (ICC 0.931-1.000) is consistent with, and in places exceeds, prior estimates (Williams & McClay (39): ICC 0.91-0.97 for AHI; Zuil-Escobar et al. (29): ICC 0.82-0.96 for navicular-height measures). LAA showed the lowest ICC (0.931), reflecting goniometric variability on a curved surface (17). SEM/MDC95 values (Table 4) are intra-rater only; for AHI, the intra-rater MDC95 (0.022) is small relative to the between-arch-type thresholds, supporting adequate discriminative sensitivity.

The strong MLAH-AHI correlation reflects their shared basis: both quantify arch height, differing in raw distance (MLAH) versus size-normalised ratio (AHI). AHI's normalisation may be preferable where foot dimensions vary, supported by its negative correlations with foot length (ρ = −0.421) and TFL (ρ = −0.377); MLAH showed no significant anthropometric correlation, consistent with Persiane et al. (15). The moderate MLAH/AHI-LAA correlations suggest LAA captures angular shape rather than absolute height (5), making the constructs complementary; setting aside the definitional Feiss-FFT agreement, the lower AHI and LAA concordance reflects differing thresholds and the continuous-versus-categorical nature of the measures (46).

Higher arch height was associated with greater functional impairment (MLAH-AOFAS ρ = −0.231; AHI-AOFAS ρ = −0.207), consistent with the rigid cavus foot generating increased plantar pressure and impaired shock absorption (19, 20); Burns et al. (19) showed orthotic pressure reduction improved function. Reduced plantarflexion in cavus reflects increased plantar-fascia and intrinsic tension (19) and greater subtalar inversion reflects rearfoot varus (8). Sex and BMI were the primary independent predictors of pain (male sex: ~2.87-fold odds of elevated pain); no MLA measure remained independently associated with pain or function after adjustment, suggesting their unadjusted correlations (Section 3.6) are largely accounted for by demographic factors. The independent association of High PA with higher, not lower, NRS is the opposite of a simple analgesic-effect prediction (38); since it was not reproduced for AOFAS or high pain (OR = 1.960, p = 0.253), added little explained variance, and rested on only 17 High PA participants, this is better regarded as a weak, unreplicated, cross-sectional signal than as evidence that activity affects pain in either direction. Plausible non-causal explanations include reverse causation (more tolerable symptoms permitting greater activity) and unmeasured confounding by occupation or activity type.

Clinically, AHI offers size normalisation, while MLAH is simpler where foot-length measurement is impractical. The Low PA predominance (89.2%) is of public-health relevance (38), but because higher activity was associated with slightly higher, not lower, pain here, these cross-sectional data do not support activity promotion for pain reduction without prospective evaluation. The intra-rater ICCs for MLAH and AHI approach AI-assisted radiographic values (0.88-0.99) reported by Lassalle et al. (26); however, radiographs uniquely provide bony angular parameters (calcaneal pitch, Meary's angle) relevant to surgical planning (27) and dynamic imaging captures arch deformation during gait (28). Emerging plantar-pressure deep-learning, smartphone and wearable-sensor tools (21-23) are promising but require regulatory validation before routine use (Table 1).

Limitations of the study

Firstly, the cross-sectional design precludes causal inference among arch morphology, physical activity and pain. Secondly, single-centre orthopaedic outpatient recruitment in Gujarat limits generalisability and the pes planus and High PA subgroups were small (Section 2.2). Thirdly, the narrow NRS (3-8) and AOFAS (60-85) ranges reflect a moderate-severity outpatient cohort. Fourthly, IPAQ-SF is self-reported and subject to recall/social-desirability bias; objective accelerometry is preferable in future work. Fifthly, inter-rater reliability was not assessed and the LAA interior-angle convention (Option A) differs from the exterior-angle convention used in some published series, so LAA values should not be compared directly with published norms without adjustment (40). Sixthly, three participants exceeded the 70-year upper age limit by up to two years (Section 3.1).

CONCLUSIONS

Four non-radiological MLA measures (MLAH, AHI, LAA, Feiss line) showed good-to-excellent intra-rater reliability and were clinically feasible without radiological equipment. MLAH and AHI were very strongly correlated (ρ = 0.846) but should be regarded as convergent, not interchangeable: they are on different scales, carry different measurement error, and classified the same participants differently under categorical thresholds (Section 3.5); AHI additionally offers foot-size normalisation.

Pes cavus predominated in this clinic-based cohort and was associated with restricted plantarflexion, increased subtalar inversion, higher pain and lower function. Sex and BMI were the strongest independent predictors of pain; physical activity group was also independently associated with higher, not lower, pain, an association not replicated across outcomes and not interpretable causally. A multimethod classification approach is recommended when arch type directly informs treatment and prospective, multi-rater studies with objective activity measurement and larger subgroups are warranted.

Conflicts of interest

none declared.

Financial support

This work was supported by All India Institute of Medical Sciences (AIIMS) Rajkot under the project [PROTOCOL ID: NF/15/2022]; and the publication charges were funded by the Learning Resource Allowance of Dr. Simmi Mehra, corresponding author [LRA/A/RT/F007 to SM].

Author Contributions

Dr Sundip Charmode initiated the idea, collected and compiled the data, performed the analysis and drafted the article. Dr Abhishek Kumar Mishra contributed to data acquisition, analysis and interpretation. Dr Simmi Mehra reviewed the manuscript and provided valuable suggestions.

Use of artificial intelligence

During preparation, the authors used Claude (Anthropic), Copilot and Perplexity.ai to assist with language refinement, clarity, and table/figure formatting. AI tools were not used to generate research data, analyses, or scientific conclusions; all outputs were verified and edited by the authors, who take full responsibility for the content.

Supplementary materials

The STROBE reporting checklist.

SUPPLEMENTARY FILE S3.

Examiner Calibration Protocol

SUPPLEMENTARY FILE S4.

Data Collection Form