This study included participants from the Framingham Foot Study (FFS), comprised of members of the Framingham Heart Study (Offspring Cohort) who completed the FFS examination during 2002–2008. The FFS was approved by the Institutional Review Boards (IRBs) at Boston University and Hebrew SeniorLife, and all participants provided written informed consent prior to enrollment. Participants in the FFS received a physical examination of the foot to assess structural disorders and were queried on presence and locations of foot pain. Participants provided information on health status, history, and symptoms through a structured questionnaire. Each participant received a single FFS examination and contributed one observation to the present analysis.
We included only FFS participants with available data on CRP, IL-6, BMI, age, sex, and physical activity index (PAI). Because many participants were missing TNF-α, participants were included with or without TNF-α (Supplementary Fig. 1).
Foot pain and structural foot disorders
Foot pain was assessed with the question “On most days do you have pain, aching, or stiffness in either of your feet?” Answers were dichotomized as yes (pain in one or both feet) and no (no pain in either foot). Forefoot and hindfoot pain were identified by asking participants to locate areas of pain on a picture of the foot.
Feet were examined to determine the presence of hallux valgus, hallux rigidus, hammer toes, claw toes, or overlapping toes . For this analysis, toe deformities were considered present if a hammer, claw, or overlapping toes were present on one or both feet. Hallux valgus and hallux rigidus were each considered present if present on one or both feet.
Information on CRP, TNF-α, and IL-6 was obtained from Framingham Offspring Study visits from 1998—2001. Although these visits preceded FFS visits by 1 to 7 years, prior analyses of longitudinal CRP measurements in the Framingham Offspring Cohort have found measurements to be fairly stable over short and long-term periods averaging 4 and 16 years, respectively . Similar stability has been noted for IL-6 and TNF-α in other populations .
Inflammatory markers were measured from fasting blood draws. CRP was measured through particle enhanced immunonephelometry, TNF-α through enzyme immunoassay, and IL-6 through a quantitative enzyme-linked immunosorbent assay. Because the distributions of all inflammatory markers were right skewed, values were log-transformed before modeling.
A normal CRP range has been proposed by the American Board of Internal Medicine as < 8 mg/L , however, internal reference ranges vary by institution. Normal ranges for IL-6 and TNF-\(\alpha\) have yet to be established. Higher values for all three markers indicate a greater degree of systemic inflammation.
Age (in years), current smoking status (yes or no), BMI, and physical activity index (PAI) were measured as potential confounders. The PAI is a validated weighted score of usual metabolic activity in a 24-h period, and ranges from 24 to 120 (lowest to highest activity). Covariates were obtained at the time of the FFS exam.
Distributions of inflammatory markers, foot outcomes, and confounders were calculated for the whole population and separately for male and female participants. Prior literature has found that incidence of and risk factors for foot pain vary by sex . Therefore sex-specific logistic regression models were used to evaluate unadjusted and adjusted (age, smoking status, BMI, and PAI) relationships between each inflammatory marker and each foot outcome, independently. Covariates were selected for inclusion based on prior literature and a priori hypotheses about confounding relationships. These covariates were included in the model regardless of their statistical significance with the outcome. Because some participants were missing data on TNF-α level, TNF-α models were evaluated on a subset of the full study population.
For each pain outcome (foot, forefoot, or hindfoot), we tested interactions between BMI and each inflammatory marker to determine whether associations varied by BMI.