|Home | About | Journals | Submit | Contact Us | Français|
Plantar fasciitis is a common disabling condition and the prognosis of conservative treatment is difficult to predict.
To determine whether initial clinical findings could help predict patient response to conservative treatment primarily consisting of supportive footwear and stretching.
Patients were recruited and seen at two outpatient podiatric clinics in the Chicago, Illinois metropolitan area.
Seventy-seven patients with non-chronic plantar fasciitis were recruited. Patients were excluded if they had a heel injection in the previous six months or were currently utilizing custom foot orthoses at the time of screening. Sixty-nine patients completed the final follow-up visit three months after receiving the footwear intervention.
Treatment failure was considered a <50% reduction in heel pain at three month follow23 Logistic regression models evaluated the possible association between over thirty clinical and physical exam findings prospectively assessed at enrollment, and treatment response.
Inability to dorsiflex the ankle past −5° (OR 27 3.9, p=.024), non-severe (≤ 7 on ordinal scale) first-step pain (OR 3.8, p=.021), and heel valgus in relaxed stance (OR 4.0, p=.014) each predicted treatment failure in multivariable analysis (Receiver operating characteristic area under the curve=.769). Limited ankle dorsiflexion also correlated with higher heel pain severity at initial presentation (r = −.312, p =.006).
Patients with severe ankle equinus were nearly four times more likely to experience a favorable response to treatment centered on home Achilles tendon stretching and supportive therapy. Thus earlier use of more advanced therapies may be most appropriate in those presenting without severe ankle equinus or without severe first step pain. The findings from our study may not be clinically intuitive as patients with less severe equinus and less severe pain at presentation did worse with conservative care.
Plantar fasciitis is the most common cause of plantar heel pain,1,2 and is an ongoing recognized economic burden. It is commonly defined as an overload injury of the proximal plantar fascia at the infracalcaneal insertion. Some studies suggest plantar fasciitis is more common in athletes3 and those that are obese.4 Others report it affects approximately ten percent of both the general and runner populations.1 In the United States, it accounts for nearly one million patient visits each year,5 and direct costs associated with prescription therapy and outpatient visits alone exceed 284 million dollars annually.6 Plantar fasciitis has a significant negative impact on general health- related quality of life which is independent of body mass index.7
The primary etiology of plantar fasciitis is unknown, and there is not a large body of evidence supporting one treatment over another. In the vast majority of cases, symptoms are resolved within a year of onset, regardless of treatment type.8 There are multiple treatment options available including stretching, foot orthoses, corticosteroid injections, night splints, extracorporeal shock wave therapy, and surgery.9 However, it is generally accepted that fewer than 10% of patients will require surgery.10 Considerable research exists on the effectiveness of various treatments,11–15 and on factors associated with the development of plantar fasciitis.16–19 However, evidence for factors predictive of a positive or negative outcome using supportive foot therapy and patient-directed interventions for plantar fasciitis is absent from the current literature.
With the more recent paradigm shift that is starting in the United States healthcare system supporting value based payment dependent on patient outcomes, it is imperative this condition be treated in a more efficient and effective manner. The primary aim of this study was to determine whether clinical findings observed during initial presentation can be predictive of treatment response in patients receiving conservative, supportive therapy for non-chronic plantar fasciitis. A secondary aim was to determine which, if any, baseline observations are associated with degree of functional impairment and/or heel pain severity.
This was a secondary data analysis of seventy seven subjects who participated in a randomized controlled trial (ClinicalTrials.gov Identifier NCT00765843) to evaluate the efficacy of three different foot orthoses conditions for the treatment of proximal plantar fasciitis 20. All participants were ambulatory adult men and women with non-chronic heel pain (i.e., symptoms for less than 12 months). All subjects were recruited from two podiatry specialty clinics located in the greater Chicago metropolitan area. The data analysis received an exempt status determination from the Institutional Review Board.
All included patients had pain at the plantar fascial attachment to calcaneal tubercle and/or pain distal from the tubercle along the plantar fascial band with typical post-static dyskinesia or pain with activity following rest. Patient history, physical exam, plain radiograph assessment, and diagnostic ultrasound were utilized to rule out other etiologies of heel pain including proximal or local nerve entrapment, arthritis, bone cyst or tumor, or stress fracture.
Patients with a previous heel injection within the past six months or who were currently using custom foot orthoses were excluded. Patients with any of the following pathologies were also excluded: proximal musculoskeletal pathology (e.g., knee or hip arthritis, sciatica secondary to back pathology, significant limb length discrepancy), use of gait assistive devices (e.g., crutches, canes, walkers), inability to wear supportive closed toed shoes, or lack of range motion at the first metatarsophalangeal or subtalar joints.
Details regarding the full randomized controlled trial can be found in a previous publication 20. Briefly, seventy-seven patients met the eligibility criteria and agreed to participate. Patients were randomized at their screening visit into one of three treatment groups: custom foot orthoses, prefabricated orthoses, or sham orthoses. Participants and investigators were blinded to group allocation. Blinding was possible by utilizing the same 3mm neoprene top cover on each type of orthoses and providing each subject with standardized shoes (Brooks Dyad, Brooks Sports Inc., Bothell WA) in which the orthoses were to be worn. All participants received a standardized baseline assessment including a comprehensive lower extremity biomechanical exam during the screening/enrollment visit from a single examiner. The same single examiner also measured the shoes, casted the patient, fit, and dispensed the orthoses and shoes. Treatment was also instituted at that time. Participants were evaluated again at one and three months for follow-up.
Those patients reporting less than 50% improvement in their heel pain symptoms (using an average of their ordinal pain scale first step and end-of-day pain scores) at 3 month follow-up were classified as having had an unfavorable treatment response, while those reporting greater than or equal to 50% improvement were classified as having had a favorable treatment response.
Each participant completed an ordinal pain scale (1–10) first step pain, end-of-day pain, and FFI-R (Foot Function Index-Revised) surveys regarding their study foot at enrollment. These scales are recognized as reliable and valid outcomes measures.11,21,22 Patients completed these instruments in a private exam room during the visit. Information regarding other potential covariates including, but not limited to, age, gender, duration of symptoms, height, weight, laterality, foot type, duration of symptoms, and co-existing medical conditions was also recorded (Table 2). Additionally, a standardized lower extremity biomechanical examination was performed by a single podiatric physician with greater than 20 years of experience (BJ). The biomechanical exam measurements were performed via the techniques described (BJ).23 The measurements were performed using a standard goniometer, with all measurements being done off-weightbearing except the stance positions (relaxed and neutral calcaneal stance positions and tibial influence). Limb length was measured off-weightbearing using a standard measuring tape measuring the anterior superior iliac spine to the medial malleolus and the umbilicus to the medial malleolus. In addition, visual gait analysis was performed to facilitate identification of asymmetries consistent with limb length inequality. Ankle joint dorsiflexion, specifically, was measured off-weightbearing both with the knee flexed and the knee extended. The patient was placed in a supine position, and the subtalar joint was slightly supinated. A goniometer was aligned such that one arm of the goniometer bisected the shaft of the fibula on the lateral side of the leg, and the other arm was aligned with the lateral border of the foot. The ankle joint was manually dorsiflexed to first resistance. Ankle joint dorsiflexion with the knee extended was measured first. The physical exam variables assessed are provided in Table 3. In addition, we explored for different foot types previously believed to be important in plantar fasciitis and favorable treatment response. These included hallux limitus 24, functional hallux limitus, hypermobile 1st ray, compensated forefoot varus, and uncompensated rearfoot varus. Participants again completed all questionnaires regarding their study foot at their 1 and 3 month follow-up visits.
Outside of the type of foot orthoses and shoes received, all patients were placed on the same patient-directed heel pain treatment protocol.9 During the screening process participants were casted for orthoses and provided removable longitudinal and metatarsal (L&M) pads fashioned from ¼ inch felt for use until their orthoses were distributed at the baseline visit. One investigator (BJ) measured and dispensed all study shoes. At baseline, only 11% of patients presented with appropriately fitting shoes when measured from heel-to-ball of foot. Sixty-one percent of patients had one size or greater difference when using heel-to-ball of foot measure. Patients were instructed to wear shoes and foot supports (initially L&M pads until orthoses were provided) at all times when ambulating.
They were also instructed to avoid going barefoot. Patients were also instructed to avoid the use of ladders. Patients were asked to perform Achilles tendon stretches using both a stair (Figure 1a.) and a wall (Figure 1b., runner's stretch) three times daily (typically before or after meals) for a total of one minute each. Specifically, participants were instructed to perform an initial passive stretch for 30 seconds followed by a deeper stretch for the remaining 30 seconds. This technique was demonstrated by the examiner, and comprehension of the stretching technique was assessed at each study visit. Patient adherence was assessed at one month and at three month follow-up with a simple questionnaire asking for the number of days they performed stretching, and the average number of times per day. Patients were also asked to moderate their recreational activities until their pain had improved. Additionally, they were instructed to avoid non-steroidal anti-inflammatory drugs (NSAIDs) and instead asked to perform ice massage to the symptomatic heel for 20 minutes per hour, for as long as necessary at the end of the day, as a method of controlling inflammation. Patients were also asked about other ant-inflammatory medication they may have been taking. Written instructions and paper cups for ice massage were provided. A rescue analgesia protocol consisting of supplied acetaminophen was incorporated into the study. Patients were asked to use the acetaminophen as little as possible and were provided a diary to track their use of it. Lastly, patients were notified that if their pain was still severe after three months of orthoses use, they would have the option to receive a steroid injection.
For patient's presenting with bilateral heel pain, the study foot was chosen as the more symptomatic foot or, in the case of equal symptoms, the patient's dominant foot was used. Pearson correlation coefficient (r) was used to examine for correlation between the study’s continuous clinical covariates and pain level and disability scores. Patient demographic and baseline characteristics were stratified by treatment response (favorable versus unfavorable). Univariate logistic regression models were used to identify candidate variables associated with treatment response. Variables with P value thresholds < 0.10 were tested in the multivariable analysis. Receiver operating characteristic (ROC) curves were used to identify cut points for continuous candidate variables and these were also tested for significance in multivariable models. The final multivariable model was determined using stepwise logistic regression with P < .05 as the criterion for model entry and P ≥ .05 for removal. Missing values were imputed with the mean value for the study population during the multivariable analysis. A sensitivity analysis was also performed to evaluate the influence of participants lost to follow-up (n=8), by included them as non-responders. While there were no significant baseline differences between those lost to follow-up and those completing the study, the average baseline AM pain was 5.71 v. 4.78 for those that completed, PM pain 5.71 v. 4.68 for those that completed, and age 44.9 years v. 49.6 for those that completed. All P values less than .05 were considered significant, and all were two tailed. Estimates of risk were given by odds ratios (OR) and ninety five percent confidence intervals (CI). SAS (Statistical Analysis System, SAS Institute Inc., Cary, NC, USA) version 9.2 was used for all data analysis.
Sixty-nine of seventy-seven participants (89.6%) returned for reassessment at three months. Characteristics of the study population with complete follow-up data are provided in Table 1. The study population consisted mostly of white women with mildly obese habitus. The average duration of heel pain at initial presentation was slightly more than five months (5.27 ± 3.2 mos. [range .25 – 12]). Twenty one participants presented with bilateral heel pain (21/69, 30%). Full compliance with stretching (defined as > 80% of the days) was reported in greater than 50% of the study population at both one month and three month follow-ups (38/66, 58% [3 missing] and 38/68, 56% [1 missing], respectively). More than 80% of the study population stretched on more than half of the study days (53/66, 80% at one month and 57/68, 84% at three months). Participants admitted to stretching for a mean 2.6 ± 1.4 times per day (range 0 to 10) at one month follow-up and 2.2 ± 1.3 times per day (range 1 to 10) at three month follow-up. No participant admitted to using NSAIDs during the study period.
Table 2 displays the distribution of select baseline characteristics, including foot support allocation and pain/disability scores, stratified by treatment response. Treatment response was not associated with the type (sham, prefabricated, or custom orthoses) of foot orthoses received (p > .10). Table 3 describes the distribution of baseline biomechanical exam variables stratified by treatment response. Of the variables examined in the univariate analysis, hip internal rotation (with hip extended, p = .008), hip neutral position (with hip extended, p = .016), relaxed calcaneal stance position (p = .035), heel valgus in relaxed stance (p = .013), ankle joint range of motion < negative 5° (with knee extended, p = .022), and first step heel pain ≤ 7 (p = .031) were each associated with treatment response. Other potentially important variables (p < .10) that were tested in the multivariable analysis included baseline first step pain, hip neutral position (with hip flexed), and amount of passive subtalar joint pronation. The only significant predictors of an unfavorable treatment response in the multivariable analysis were having heel valgus in relaxed stance, inability to dorsiflex past −5° (with knee extended), and having first step heel pain ≤ 7 at initial presentation (Table 4). Seventy percent (22/31) of participants with severe first step heel pain (≥7/10) responded to treatment in our study compared to only 45% (17/38) of those without severe heel pain. Seventy one percent (25/35) of participants without heel valgus responded to treatment compared to 41% (14/34) of participants with heel valgus. Sixty seven percent (30/45) of participants with severe ankle equinus (≤ neg. 5 degrees) responded compared to only 38% (9/24) of participants with without severe equinus. We found no differences in foot types, including foot posture index, hallux limitus, functional hallux limitus, hypermobile 1st ray, compensated forefoot varus, and uncompensated rearfoot varus. (Table 5)
Ankle joint dorsiflexion (with the knee flexed) demonstrated a moderate correlation with pain and a modest correlation with disability at initial presentation. Patients with less ankle joint dorsiflexion generally reported higher end-of-day (r = −.357, p = .002) and higher average heel pain levels (r = −.312, p = .006). Patients with less ankle joint dorsiflexion also tended to report higher total scores (e.g., more disability) on FFI-R at presentation (r = .303, p = .008).
BMI demonstrated a modest correlation with both pain and disability scores at initial presentation. Patients with higher BMI tended to report higher levels of end-of-day pain (r = .244, p = .033), higher average heel pain levels (r = .224, p = .049), higher scores on the FFI-R pain subscale (r = .268, p = .019), and higher total disability/dysfunction scores on the FFI-R (r = .303, p = .008). Participants with higher pain levels on FFI-R pain subscale also tended to exhibit a narrower base of gait (r = −.302, p = .008).
Finally, Hispanic patients (n=11) reported higher first step heel pain (mean 8.1 ± 1.9 vs. 6.5 ± 2.2, p = .026) and higher average heel pain (mean 8.0 ± 1.5 vs. 6.0 ± 1.9, p = .002) levels compared to non-Hispanic patients. There were no other correlations observed between demographic, historical and/or physical exam characteristics and magnitude of pain/disability at initial presentation. Furthermore, we explored treatment response thresholds. There were only 9 subjects (out of 69) that had no pain (VAS am + VAS pm pain = 0) at 3 months, and 18 (out of 69) that had 80% improvement or greater at 3 months. Not surprisingly when these outcomes were modeled in the multivariate analysis, there were no significant predictors identified.
In our sensitivity analysis, there were no significant differences in age, baseline AM pain, and baseline PM pain for the lost to follow-up group (n=8) and those completing the study. When we included the lost to follow-up group as non-responders and repeated our logistic model, the effect sizes did not change substantially. (Data not shown)
Studying a compliant patient population with non-chronic plantar fasciitis, we found that patients with severe ankle joint equinus were more likely to respond favorably to a treatment program that centered around conservative care including Achilles tendon stretching. Specifically, patients unable to dorsiflex to at least −5 degrees (with the knee extended) at initial presentation were nearly four times more likely to experience an unfavorable response at three month follow-up. While equinus is certainly a recognized risk factor for development of plantar fasciitis,25–27 we believe our study is the first to actually conclude that more ankle equinus at the start of therapy translates into improved outcome after conservative therapy.9
Our findings are encouraging as they suggest, like prior work, many people with non-chronic plantar fasciitis will benefit from conservative care.2 After all, 65% (45/69) of the participants in our study exhibited a severe ankle joint equinus at their initial presentation and it is well recognized that many individuals suffering with plantar fasciitis will demonstrate restricted ankle joint motion.16,25–27 Despite there being a lack of clear evidence surrounding the benefit of Achilles tendon stretching,28 authorities still maintain that home Achilles tendon stretching is an integral part in the conservative management of plantar fasciitis.29,30 Our work may help explain the mixed results of past work. Our protocol relied on stretching, ice massage, standardized shoes, and altering footwear behaviors. Our work also found that stretching alone was not enough for experiencing relief. More specifically, we found that the amount of stretching (i.e., days or times per day) determined by patient self-report was not associated with a favorable treatment response (p > .05). Therefore it may be that it is the combined effect of regular stretching along with an initial severe posterior muscle group tightness that favors a good response to treatment. Our study was underpowered to fully examine this interaction effect, but future work is probably warranted to further explore this effect.
Subjects with a valgus rearfoot position in relaxed stance were also more likely to experience an unfavorable response to treatment. There is some limited evidence which suggests that patients with pronated feet may be more likely to develop chronic plantar fasciitis symptoms. In a recent case-control study, Irving et al.,4 found that patients with chronic heel pain (defined as > 6 months of symptoms) were nearly four times more likely (OR 3.7, 95% CI 1.6 to 8.7) to have a pronated foot posture determined using FPI. Our findings are in close agreement with these, as our patients with non-chronic plantar fasciitis and heel valgus were four times less likely to respond to patient-directed treatment measures. These findings make some intuitive sense as excessive pronation of the foot increases tensile loads on the plantar fascia.31,32 It is also possible that an everted heel may allow for more displacement of the plantar calcaneal fat pad, and therefore possibly less shock absorption during impact (i.e., heel strike). Indeed, we examined our data for other foot types that were poor shock attenuators. We found uncompensated rearfoot varus was prevalent at 30%; however, its association with treatment response only trended towards significance. (p=.12)
Outside of heel valgus and a tight gastro-soleal complex, we did not find any other biomechanical measurements associated with treatment response. Although hamstring tightness has been implicated in the development of plantar fasciits in previous work164, hip flexion did not seem to correlate with likelihood of achieving a favorable response in our work. This finding was not entirely unexpected; however, as there was no effort made during any of our offered therapies to address a possible contracture of the proximal limb posterior muscles. Further work to address whether therapies aimed at relieving hamstring tightness for plantar fasciitis may be warranted. We were surprised by our findings that hallux limitus or functional hallux limitus were not associated with treatment response especially given our novel findings of poorly fitting shoes at baseline. When measuring from heel-to-ball of foot, only 11% presented with appropriately fitting shoes with 61% having > one size difference when using heel-to-ball of foot measure. As the plantar medial fascial band is the widest band of tissue, we were surprised hallux limitus was not associated with treatment response. In a case control study, Aranda and Munuera reported significantly more 1st metatarsal phalangeal joint dorsiflexion in controls than those with plantar fasciitis (71° v. 49°). 33
Finally, we found that patients with severe (> 7/10) first step heel pain were approximately four times more likely to respond favorably to conservative treatment measures. This means that patients presenting for treatment of mild to moderate symptoms were less responsive to conservative therapies for heel pain. In a largely self- limiting condition with a high placebo treatment effect,1 it is not surprising to see that patients with the greatest pain may be more likely to experience relief with the introduction of relatively straight-forward program of footwear, insole, and gentle stretching recommendations.
This work also identified several variables that are, at least, modestly associated with greater amounts of pain and/or disability in patients presenting with plantar fasciitis. We found that patients with greater amounts of equinus and greater BMI were also more likely to present with higher average heel pain levels and more self-reported foot disability. These findings add further supportive evidence to the notion that ankle equiuns and BMI may be important risk factors in the causal pathway for developing plantar fasciitis.4,27
The primary strength of this study is that it is the first to look at presenting factors that might be predictive of treatment response in patients suffering with non-chronic plantar fasciitis. Because pain response was not associated with the type of foot orthoses received we were able to readily evaluate factors that were predictive of pain response independent of treatment allocation. The complete exam prospectively performed at presentation by a single experienced physician allowed for the consideration of many variables. Since all subjects received the same shoes and the same instructions on stretching and icing, those aspects were standardized.
We recognize limitations exist and should be addressed in future studies. The primary weakness of the study is that it is a secondary analysis. It would be beneficial to reproduce these findings using a prospective design and in another, larger patient population. Additionally, ankle equinus was only measured during the screening session. It would be interesting to follow potential flexibility improvements in all participants after the initiation of the included stretching protocol. It would also be valuable to look at the potential additive benefit that might be achieved with greater stretching in those that exhibit the greatest posterior group tightness. All of the measures were determined by a sole examiner. Intra-rater reliability for measures of the foot and ankle tend to be quite higher than inter-rater reliability.34 However, many of these measures do not have published clinical effect sizes to assist in determining the clinical significance of our findings. Lastly, there were a substantially higher number of female participants and the potential effect of this is unknown.
Patients with severe ankle equinus were nearly four times more likely to experience a favorable response to a plantar fasciitis treatment program that centered around a home Achilles tendon stretching regimen and conservative therapy. Patients presenting with more pronated foot types and less severe presenting pain were also less likely to respond to conservative care consisting of ankle stretching, ice massage, standardized shoes, and altering footwear behaviors.
This study received funding from the American Podiatric Medical Association (APMA). It also received support from the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), Grant 1T35DK074390. NIDDK and APMA had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Funding source: American Podiatric Medical Association and National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), Grant 1T35DK074390.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
James S. Wrobel, University of Michigan Medical School, Domino's Farm, Lobby G, Suite 1500, 24 Frank Lloyd Wright, Dr. Ann Arbor, MI 48105, Office: 734-232-3611.
Adam E. Fleischer, Dr. William M. Scholl College of Podiatric Medicine, Rosalind Franklin University of Medicine and Science, 3471 Green Bay Road, North Chicago, IL 60064, Office: 847-578-8449.
Jonathon Matzkin-Bridger, Dr. William M. Scholl College of Podiatric Medicine, Rosalind Franklin University of Medicine and Science, 3333 Green Bay Road North Chicago, IL 60064, Office: 847-578-8420.
Jeanna Fascione, Advocate Illinois Masonic Medical Center, 836 West Wellington, Chicago, IL 60657, Office: 773-296-8285.
Ryan Crews, Dr. William M. Scholl College of Podiatric Medicine, Rosalind Franklin University of Medicine and Science, 3333 Green Bay Road, North Chicago, IL 60064, Office: 847-578-8437.
Nicholas Bruning, University of Michigan Medical School, Domino's Farm, Lobby G, Suite 1500, 24 Frank Lloyd Wright, Dr. Ann Arbor, MI 48105.
Beth Jarrett, Dr. William M. Scholl College of Podiatric Medicine, Rosalind Franklin University of Medicine and Science, 3333 Green Bay Road, North Chicago, IL 60064, Office: 847-578-8415.