What Is the Efficacy of a Nonoperative Program Including a Specific Stretching Protocol for Flexor Hallucis Longus Tendonitis?

Abstract

Background

Diagnosis and treatment of tendonitis/entrapment of the flexor hallucis longus (FHL) has been sporadically described in the evidence, primarily in the context of dancers and other athletes. Although various nonspecific nonoperative treatments have been described, it is not clear how often they achieve a satisfactory amount of symptom improvement.

Questions/purposes

The present study was designed to address the following questions regarding the nonoperative treatment of FHL tendonitis: (1) In a population of patients where the default management option for FHL tendonitis is a comprehensive nonsurgical approach, what proportion of patients thus treated opted not to have surgery? (2) What factors were associated with a patient’s decision to undergo surgery after a period of nonsurgical management?

Methods

The 656 patients included were all those diagnosed with FHL tendonitis who were initially treated nonoperatively in the foot and ankle division between January 2009 and December 2018. Demographics, comorbidities, examination findings, imaging results, pain scores, treatment instituted, and final outcome were obtained from the electronic medical record. The primary outcome was the decision to have surgery due to unsatisfactory symptom improvement. We compared patients who opted for surgery with those who did not after nonoperative treatment with univariable and multivariable statistics using demographics, comorbidities, and clinical findings as potential risk factors, with p < 0.05.

Results

Forty-four percent (180 of 409) of patients decided to forgo surgery after the institution of a specific FHL stretching program. Surgery was more likely in patients with clinical hallux rigidus (OR 2.4 [95% CI 1.16 to 4.97]; p = 0.02) or posteromedial ankle pain (OR 1.78 [95% CI 1.12 to 2.83]; p = 0.01) and less likely in those who completed an FHL stretching program (OR 0.15 [95% CI 0.08 to 0.27]; p < 0.001).

Conclusion

FHL tendonitis is more common than the previous evidence suggests and frequently occurs in nonathletes. Once it was diagnosed by detection of tenderness anywhere along the tendon, most frequently at the fibroosseous tunnel, nonoperative treatment focused on specific FHL stretching and immobilization in more severe cases reduced the symptoms to the extent that 44% of patients decided that surgery was unnecessary. The key to its diagnosis is awareness that this injury is possible because most patients treated in this study had been previously seen by orthopaedic providers who had not appreciated the presence of the condition, leading to a delay in diagnosis and treatment of more than a year in many patients.

Level of Evidence

Level III, therapeutic study.

Introduction

Diagnosis and treatment of tendonitis/entrapment of the flexor hallucis longus (FHL) has been sporadically described in the evidence, primarily in the context of dancers and other athletes [1, 4, 5, 8, 15]. Although various nonspecific nonoperative treatments have been described—such as rest, ice, compression, and elevation (RICE); injections; and strengthening of the deep posterior musculature in dancers [1, 5, 8, 15]—it is not clear how often they are successful. The total number of patients who have been reported on with clinical follow-up after nonoperative treatment in the evidence is 179 [1, 4, 5, 8, 11, 15], of which 61 were patients primarily selected for study because they had undergone surgery. Of the remaining 118 patients, 81 were in a single study, and each of the studies encompassed the practice of a single orthopaedist. Even though most of the studies published have focused on the occurrence of this entity in ballet dancers, soccer players, and other athletes [1, 4, 5, 8, 15], it has been our experience that FHL disorders are commonly seen in the general population, and their diagnosis is frequently missed because of the perception that it only occurs in athletes.

Tendonitis of the FHL can occur from impingement during extreme plantarflexion, such as in ballet [1, 4, 14, 16], irritation from an os trigonum [5, 8, 10, 13, 18, 19], direct trauma [3], or as a consequence of entrapment in the proximal end of the fibroosseous tunnel by a low-lying FHL muscle belly [4, 11]. Although motor imbalance may play a role in the etiology in ballet dancers [1, 4, 19], a common clinical finding for all patients with FHL tendonitis is a limitation of hallux dorsiflexion when the ankle is maximally dorsiflexed [11]. This is a consequence of preventing the FHL muscle belly from further distal excursion by restriction of movement into the fibroosseous tunnel. This formed the basis of the development of an FHL-specific stretch protocol that involves placing the hallux on a small platform to dorsiflex it, followed by maximally dorsiflexing the ankle to place the FHL under maximal stretch (Fig. 1). Only one previous study included a stretching program explicitly designed to address the limited excursion of the FHL [11], but it did not ascertain the efficacy of the stretching program itself. Furthermore, most of the evidence has focused on the presentation of FHL tendonitis in dancers and other running/jumping athletes, without investigating its occurrence in the general, nonathletic population. The current study was conducted to expand the understanding of the spectrum of patients with FHL tendonitis, their clinical signs and symptoms, and the efficacy of nonoperative treatment. The foot and ankle division in our medical center has been treating patients with FHL tendonitis using a standardized protocol that includes FHL-specific stretching since 2007. The institution of an electronic medical record in 2009 made it possible to begin to collect extensive data, including demographics and comorbidities, which enabled us to identify and conduct a retrospective review of a decade of patients with FHL tendonitis who had been treated.

Fig. 1.

A-B Demonstration of the flexor hallucis longus (FHL) stretch used to treat patients with FHL tendonitis. (A) The patient stands with the foot flat on the floor while placing the hallux on top of a relatively thin (about 1 cm) book. This serves to pretension the FHL. (B) The patient then maximally dorsiflexes the ankle while keeping the foot flat on the floor. Since the FHL limits ankle dorsiflexion while the hallux is also dorsiflexed [12], this maximizes the tension in the FHL muscle-tendon complex during the stretch.

This single-center retrospective study was designed to address the following questions regarding the nonoperative treatment of FHL tendonitis: (1) In a population of patients where the default management option for FHL tendonitis is a comprehensive nonsurgical approach, what proportion of patients thus treated opted not to have surgery? (2) What factors were associated with a patient’s decision to undergo surgery after a period of nonsurgical management?

Patients and Methods

Study Design and Setting

This study drew from the integrated practices of two fellowship-trained foot and ankle orthopaedic surgeons, two podiatrists, and two foot and ankle trained nurse practitioners.

Participants

The diagnosis of FHL tendonitis is primarily clinical, with imaging playing an ancillary or confirmatory role. The existence of FHL tendonitis can be demonstrated by the presence of tenderness elicited along the flexor hallucis longus at the proximal entrance of the fibroosseous tunnel in the posterior medial ankle [11, 16, 19], underneath the sustentaculum talus, and at the knot of Henry (Fig. 2) [11, 16, 19]. There may also be pain produced in the dorsal first metatarsophalangeal (MTP) joint with hallux dorsiflexion [5]. Posterior ankle pain with forceful ankle plantarflexion is a sign of impingement [5, 9, 14], but this typically does not occur with pure FHL tendonitis [4]. Assessment of stenosing tenosynovitis in the fibroosseous tunnel was determined by the FHL stretch test [4, 11, 16, 17]. In this maneuver, the maximal amount of hallux metatarsophalangeal joint (MTPJ) dorsiflexion is determined while holding the ankle in full dorsiflexion and stabilizing the first metatarsal head to prevent its plantarflexion (Fig. 3). This normally produces 20° or more of hallux MTPJ dorsiflexion, with less than 10° seen in the presence of FHL entrapment at the fibroosseous tunnel [11, 12]. This maneuver may also elicit pain in the dorsal hallux MTPJ [5, 9, 11, 16]. Radiographs can demonstrate an associated os trigonum [5, 10]. Although an MRI can show FHL tendonitis with excess tenosynovial fluid around the tendon, false-positive results are not unusual [10], and an MRI is not required for making a diagnosis [9]. The clinical diagnosis of FHL tendonitis was determined by the presence of tenderness and/or pain localized to the FHL. The finding of restricted hallux dorsiflexion (less than 10°) with the FHL stretch test was considered confirmatory, but it was not required for the diagnosis. Imaging was also not required for the diagnosis.

Fig. 2.

Photograph demonstrating the areas that are palpated to elicit tenderness of the FHL. The black arrow at the postero-medial ankle is at the superior end of the fibroosseous tunnel. The dotted arrow indicates the area under the sustentaculum talus where the FHL runs within the fibroosseous tunnel. The white arrow is inferior to the navicular, just medial to the plantar fascia, where the knot of Henry can be palpated.

Fig. 3.

Demonstration of the FHL stretch test. The ankle is held in maximal dorsiflexion. The first metatarsal is stabilized in dorsiflexion with the thumb of one hand to prevent its plantarflexion during attempted hallux dorsiflexion. The hallux MTP joint is then maximally dorsiflexed. Normal FHL excursion will permit at least 20° of hallux dorsiflexion at the MTP joint with this maneuver. Restricted FHL excursion results in less than 10° of hallux MTP joint dorsiflexion [11].

Patients were identified through the electronic medical record system, comprising all patients seen in the Foot and Ankle Division of the Department of Orthopedics and Rehabilitation, Larner College of Medicine at the University of Vermont, South Burlington, who had their initial clinic visits between January 1, 2009 and December 31, 2018, and who had the diagnosis of flexor hallucis longus tendonitis (ICD-9 code 727.06, ICD-10 code 77.9). Outpatient follow-up lasted until the end of their treatment or for a minimum of 1 year if still under treatment. The study duration was limited to 2018 to permit at least 1 year of surveillance (through the end of 2019) of the medical records for any indication of continuing treatment for the index condition. Data collected comprised demographics, medical comorbidities including systemic steroid use and smoking and alcohol consumption history, and all clinic notes. Clinic notes were reviewed manually and abstracted using a standardized data collection form. Information that was recorded included the provider, dates of visits, the side affected, previous history of injury, location of pain and tenderness, other foot and ankle complaints, athletic history, imaging results as interpreted by the provider, pain score (VAS), treatment instituted at each visit, and whether surgery was either recommended or carried out.

All patients were initially seen by one of six providers in the foot and ankle division, comprising two nurse practitioners, two podiatrists (SJM, LMS), and two orthopaedic surgeons (MDC, JDM). Evaluation and initial management was similar for all providers. If the patient and the nonorthopaedic provider agreed that there was not satisfactory symptom improvement, the patient was seen by the orthopaedic surgeons for surgical consultation. It was the same two surgeons (MDC, JDM) throughout the study duration, and the criteria for offering surgery was consistent between and throughout the study period.

In all, 656 patients were identified as having flexor hallucis longus (Fig. 4). Of these, 84 did not undergo nonoperative treatment, either because similar treatment had been attempted previously by the referring providers or there were intercurrent conditions that guided the move to surgical treatment. Of the remaining 572 patients treated who began with nonoperative treatment, 12 decided to have surgery, which was scheduled but not performed for unknown reasons. Of the remaining 560 patients who underwent some form of nonoperative treatment, 151 did not return after their initial visit when they were started on the stretching program. All had been scheduled for a 6-week follow-up visit but told they could cancel it if they were satisfied with their progress. This left 409 patients in the analysis who had more than a single clinic visit and complete records until they either had resolution of their symptoms or had surgery. The mean age of the patients in the final study group was 45 years (95% confidence interval 42 to 48), with 74% (302 of 409) female patients, and 75% (308 of 409) not professing any athletic endeavors (Table 1). Twenty-nine percent (120 of 409) of patients reported a history of a relevant prior ankle or foot injury. The most common associated injuries were ankle sprain reported in 11% (46 of 409) of patients, ankle fracture in 7% (29 of 409) of patients, and either hyper plantarflexion or dorsiflexion of the ankle in 2% (8 of 409) of patients. VAS pain scores were routinely obtained at all clinic visits. They were available for the analysis in 99% (406 of 409) of patients at their first visit and 99% (405 of 409) patients at their last visit.

Fig. 4.

Flow diagram of patient selection for the study. Of the initial 656 patients identified with the diagnosis of flexor hallucis longus tendonitis in the electronic medical record, 409 were included in the final study population.

Table 1.

Study demographics

	Total patients (n = 409 patients)	Nonoperative patients (% of row category)	Surgical patients (% of row category)	p valuea
Women	74 (302)	44 (133)	56 (169)	0.99
Diabetes	8 (33)	27 (9)	73 (24)	0.04
Steroid use	28 (113)	42 (47)	58 (66)	0.5
Tobacco use	17	8	9	0.81
Alcohol use	17	8	9	0.81
Athletic activities	25 (101)	52 (53)	48 (48)	0.05
Age in years	45 ± 15	45 ± 15	44 ± 15	0.33
BMI, kg/m2	29 ± 7	29 ± 7	30 ± 8	0.08
Follow-up duration in weeks	29 (29-30.0)	34 (8-117)	64 (30-148)	< 0.001

Data presented as % (n), mean ± SD, or median (interquartile range).

^aStatistical comparison is between patients who did and did not have surgery

The sites of FHL-related pain reported by the patients were at the posteromedial ankle in 55% (225 of 409), the hallux in 28% (116 of 409), and the arch in 19% (78 of 409), with many patients having pain in more than one FHL-related location. Tenderness on examination was highly correlated with the sites of reported pain, but additional FHL-related sites of tenderness were frequently noted (Table 2). Pain was reported in one, two, or three locations by 67% (274 of 409), 14% (56 of 409), and 3% (11 of 409) of patients, respectively. Tenderness was elicited at one, two, or three sites in 56% (231 of 409), 23% (96 of 409), and 14% (58 of 409) of patients, respectively. For the 13 patients who had no pain or tenderness directly related to the FHL, all had restricted FHL excursion with the FHL stretch test, and their symptoms (usually lateral forefoot overload with pain) were felt to be a direct mechanical consequence of the alteration in gait necessitated by the reduced FHL excursion. Of the 215 patients who had radiographs obtained, 11% (45 of 409) had an os trigonum. In the presence of an os trigonum and posteromedial ankle tenderness, it may not be possible to distinguish between isolated FHL tendonitis and a symptomatic os trigonum. However, this is probably a moot question, since the os trigonum forms part of the floor of the fibroosseous tunnel through which the FHL passes and irritation of the former will cause inflammation of the latter [5]. The contribution that an otherwise asymptomatic os trigonum may make to stenosis of the fibroosseous tunnel is not known [18].

Table 2.

Correspondence between location of pain and tenderness

		Posteromedial ankle (n = 225)a	Arch (n = 78)	Hallux (n = 116)	None (n = 68)b
Location of FHL tenderness	Posteromedial ankle	93% (210)	77% (60)	60% (70)	66% (45)
	Arch	26% (59)	73% (57)	27% (31)	31% (21)
	Hallux	26% (58)	44% (34)	76% (88)	32% (22)
	None	2% (4)	3% (2)	5% (6)	19% (13)

Numbers in columns and rows do not add up to 100% since some patients had pain or tenderness in more than one location (as detailed in Methods); location of pain and tenderness are as described in Methods.

^aData represents percent of patients (n) with pain at the site indicated by the column header.

^bPatients whose pain was primarily located other than along the FHL tendon.

Hallux rigidus was noted in 47 patients clinically and was radiographically confirmed in 25. The clinical diagnosis of hallux rigidus was based on the presence of pain or tenderness in the first MTPJ associated with any dorsal bone prominence at the joint. Joint motion did not have to be restricted. The distinction between hallux rigidus and stenosing FHL tenosynovitis was made based on the physical examination. In early hallux rigidus, there may be a small dorsal first MTP osteophyte that mildly limits hallux dorsiflexion (45°). However, with the FHL stretch test, hallux dorsiflexion may be normal (20° or more) or restricted (10° or less). In the latter situation, the patient has both hallux rigidus and stenosing FHL tenosynovitis. In more severe hallux rigidus, first MTPJ dorsiflexion is limited regardless of ankle position, so it would not be possible to make the diagnosis of stenosing tenosynovitis, since it is not possible to detect a change in hallux dorsiflexion with the FHL stretch test. However, if the patient has tenderness along the FHL (usually at the fibroosseous tunnel), then the coexistence of FHL tendonitis can be made in the absence of demonstrable stenosis.

The mean follow-up for the overall population was 54 weeks (95% CI 19 to 132), with longer follow-up for the patients ultimately treated with surgery than those treated purely nonoperatively (64 weeks [95% CI 30 to 148] versus 33.5 weeks [95% CI 8.25 to 117]; p < 0.001). The only significant demographic differences between patients who ultimately had surgery and those who did not was that the former were more likely to have diabetes and had a longer duration of follow-up. There was no right-left predilection for symptoms, with 14% (59 of 409) reporting bilateral symptoms.

Nonsurgical Treatment Protocol

The initial nonoperative treatment was dictated by symptom severity. Most patients were started on an FHL stretching protocol at their first visit. This consisted of a stretch with the hallux passively dorsiflexed on a small book while simultaneously maximally dorsiflexing the ankle with the knee bent with the forefoot and heel kept on the ground [11] (see Fig. 1). Each stretch is held for 30 seconds, then repeated three times. The entire stretching set is repeated three to four times per day. All patients were given a handout describing and illustrating the FHL stretch, which was also demonstrated for them during their clinic visit. Formal physical therapy was not used. Patients with more severe symptoms who could not tolerate the stretching protocol were initially treated by immobilization, generally a walking boot, for 6 weeks. The selection of immobilization type was up to the provider/patient based on considerations such as anticipated compliance, comfort, and convenience. NSAIDs were not consistently used. The patients were told they could take them for pain control if they wished, particularly for those who underwent immobilization. Since this is predominantly a mechanical problem (entrapment of the FHL in the fibroosseous tunnel), it was not felt that NSAIDs would be particularly useful unless the underlying mechanical problem was addressed through the stretching program. If they were sufficiently improved by their 6-week follow-up, the FHL stretching protocol was started. This overall protocol was standardized between providers.

All clinic visits were spaced 6 weeks apart. Patients who felt that the nonoperative treatment was effective in restoring their ability to resume their normal activities were given follow-up appointments that they could cancel if they were continuing to do well. Imaging was usually obtained for patients contemplating surgery, primarily to identify a possible os trigonum. Otherwise, imaging was not routinely required for the diagnosis and nonoperative treatment of FHL tendonitis, as the diagnosis is based on the clinical exam. Many patients referred to us already had imaging obtained before being seen.

Decision to Have Surgery

If the patient did not feel that they were sufficiently improved by the second or subsequent clinic visits, radiographs were obtained and the possibility of surgical treatment was discussed. This consisted of an FHL release at the fibroosseous tunnel combined with any other relevant procedures based on their underlying clinical condition (such as, os trigonum, hallux rigidus, hallux valgus). The criteria for offering surgery (based on the presence of pain or functional impairment that was interfering with normal daily activities) was consistent between the providers, but the patient always made the final decision.

Bias

As a retrospective study at a tertiary referral center, this study has several potential sources of bias, which fall into four general categories. The most obvious is that this study reflects the selection bias of our patient population for those with more recalcitrant symptoms. This is a direct consequence of our referral pattern, with our results reflecting the outcomes in a worst-case patient population. The second is performance bias, specifically in the variation between providers in diagnosing FHL tendonitis and the criteria followed to offer surgical treatment. The third is detection bias, primarily the completeness of the capture of patients who have been diagnosed with FHL tendonitis based on searching the electronic medical record using the appropriate ICD-9/10 codes. The last source of potential bias is transfer bias, in the form of patients pursuing further treatment elsewhere.

Primary and Secondary Study Outcomes

Our primary study outcome was the proportion of patients treated for FHL tendonitis with the nonsurgical approach who opted not to have surgery performed at our center within 1 year of diagnosis. This was determined by the manual review of all clinic notes, where the patient was characterized as being satisfied with their progress and/or canceling tentatively scheduled surgery specifically because they felt sufficiently improved that surgery was unnecessary. All patients undergoing surgery were identified by the presence of an appropriate operative note. Most patients had enough symptomatology that they had seen multiple other providers over the course of a year or more before their being seen in our clinic. As such, they had already self-selected as patients who would not have decided to live with their symptoms if the nonoperative treatment was not helping, and they knew that a relatively simple surgical treatment existed. This makes it unlikely that a sizable number of patients who ultimately did not have surgery had simply given up on treatment.

With univariable analysis, patients with posteromedial ankle pain/tenderness, an os trigonum, diabetes, clinical hallux rigidus, or higher VAS scores were more likely to have surgical intervention. In contrast, patients with bilateral symptoms, a history of athletic participation, arch pain/tenderness, or having used the FHL stretching protocol were less likely to have surgery. Our secondary study outcome was factors associated with a patient’s decision to opt for surgical treatment for FHL tendonitis, after controlling for univariable variables found to be different between the surgical and nonsurgical groups. We ascertained this using multivariable analysis, as described below.

Ethical Approval

Ethical approval for this study was given by the University of Vermont, Burlington, VT, USA (number CHRMS (Medical): STUDY00000805).

Statistical Analysis

Statistical analysis was performed using SPSS (version 26, IBM Corp). Normally distributed continuous data were summarized using means with 95% confidence intervals, with nonnormally distributed continuous data summarized by medians with interquartile ranges. Statistical comparisons for continuous data were performed using ANOVA or the Mann-Whitney U test, respectively. Categorical data were assessed using a chi-square statistic. Variables found to be statistically significant (p < 0.05) with univariable analysis were entered into a backwards stepwise multivariable logistic regression analysis. Given the low percentage of patients with imaging (47% [194]), radiographic findings were not included in the multivariable linear regression. Adjusted odds ratios and CIs were presented for all statistically significant variables. For all comparisons, the significance level was set at p < 0.05.

Results

Proportion of Patients Treated Without Surgery for FHL Tendonitis

Of the 409 patients begun on the nonoperative FHL program, 44% (180 of 409) finished their treatment without having surgery. The remaining 55% (229 of 409) ultimately had surgical intervention, which included performing an FHL release and tenosynovectomy at the fibroosseous tunnel. Overall, 56% (164 of 295) of patients who followed the FHL stretching protocol did not go on to have surgery; 14% (16 of 114) of patients who did not complete the FHL stretching protocol did not opt for surgery. The VAS scores improved in both groups during treatment, with the improvement not being different between the groups (Table 3). The initial VAS scores were higher for the patients who decided to have surgery after nonoperative treatment than for those who did not have surgery (median 5 [IQR 2 to 5] versus median 3 [IQR 1 to 3]; p < 0.001).

Table 3.

Change in VAS scores after nonoperative treatment

		Initial VAS	Final VAS	delta VASa	p value (initial-final)b
Total population (n = 409)	VAS	4 (2-4)	2 (0-2)	1 (0-1)	< 0.001
	Number of patients	n = 406	n = 405	n = 405
Nonoperative treatment (n = 180)	VAS	3 (1-3)	2 (0-2)	1 (-1-1)	< 0.001
	Number of patients	n = 178	n = 177	n = 177
Surgical treatment (n = 229)	VAS	5 (2-5)	2 (0-2)	1 (0-1)	< 0.001
	Number of patients	n = 228	n = 229	n = 228
p value (nonoperative vs operative)c		< 0.001	0.09	0.10

^aChange in each patient in their VAS scores from beginning to end of treatment.

^bp value compares initial VAS score to final score.

^cp value compares differences between treatment groups.

Factors Associated with a Patient’s Decision to Undergo Surgery

After controlling for potential confounding variables, including comorbidities, location or pain/tenderness, and athletic participation, the predictors for having surgery were the diagnosis of clinical hallux rigidus (OR 2.4 [95% CI 1.16 to 4.97]; p = 0.02), presenting with posteromedial ankle pain (OR 1.78 [95% CI 1.12 to 2.83]; p = 0.01), and having a higher initial VAS (OR 1.15 [95% CI 1.06 to 1.25]; p = 0.001) (Table 4). Having performed the FHL stretches was the strongest predictor of not having surgery (OR 0.15 [95% CI 0.08 to 0.27]; p < 0.001).

Table 4.

Multivariable regression for predictive factors for the decision to have surgery^a

Predictive factor	OR (95% CI)	p value
Posteromedial ankle pain	1.78 (1.12-2.83)	0.01
Arch tenderness	0.56 (0.34-0.92)	0.02
max VAS (before treatment)	1.15 (1.06-1.25)	0.001
Clinical hallux rigidusb	2.4 (1.16-4.97)	0.02
FHL stretches	0.15 (0.08-0.27)	< 0.001
Predictive value	67.5%

Logistic regression entry variables included all univariate significant predictors, excluding the presence of an os trigonum on radiograph, since radiographs were obtained for less than 50% of patients.

^aPercent of patients in analysis: 92% (378 of 409).

^bClinical hallux rigidus is as defined in Methods.

Discussion

Although tendonitis of the flexor hallucis longus is a condition that has been recognized for almost 40 years [4], most investigators associate this condition with ballet dancers and other running and jumping athletes [4, 5, 8, 15] without addressing its occurrence in the general population. Further, most reported studies focus on the surgical treatment and findings in such patients, with relatively little attention paid to nonoperative treatment. The current study reporting the 10-year experience of a multiprovider specialist foot and ankle group expands on our understanding of this clinical problem by showing that it is seen commonly in nonathletes (75% [308 of 409]), with a high proportion (44% [180 of 409]) responding to a specific FHL stretching protocol. The presence of hallux pain/tenderness was the best predictor (OR 2.4 (95% CI 1.16 to 4.97); p = 0.02) of those who decided to have surgery after an adequate nonoperative treatment trial.

Limitations

The main limitations of this study are the biases inherent to a single-center retrospective study. Because our institution is a tertiary referral center, there is severe selection bias of our patient population toward patients with more refractory symptoms, many of whom already had undergone prolonged nonoperative treatment by other providers without sufficient symptomatic improvement. Our results, consequently, reflect the outcomes in a worst-case patient population that is not representative of a nonspecialist community orthopaedist. Efforts have been made over the years to standardize the assessment and treatment of patients with FHL tendonitis, thereby lessening the potential impact of performance bias between the providers. Through weekly clinical conferences before and during the study, the clinical examination and diagnostic criteria for FHL tendonitis was discussed and standardized between all providers. Although no formal reliability testing was performed, the experience of seeing one another’s patients indicates that there is consistency between all the providers in the examination and assessment. The possibility of detection bias and incomplete capture of all patients with FHL tendonitis based on the presence of clinical visit diagnostic codes also exists. However, it is not possible to close a clinical encounter without entering a visit diagnosis, so the search based on diagnostic codes should be accurate. Although it is possible that some patients with FHL tendonitis were never coded correctly through any of their clinic visits, given the workflow established to ensure adequate clinical documentation, the number of such patients would be expected to be small. The last major limitation is that of transfer bias, due to patients pursuing further care elsewhere. This is somewhat mitigated by our practice environment, which is a regional tertiary care center that has been the only source of specialized foot and ankle care for the region. Most patients were seen initially as a referral from the community orthopaedists and podiatrists after a nonoperative treatment by the referring provider did not improve their symptoms. To our knowledge, there were no other providers in the region during the period of this study who were performing surgery involving FHL release/tenosynovectomy. This would severely limit the options patients would have for surgical treatment outside of our medical center. Consequently, the likelihood of a patient who did not have surgery at our center (that is, being classified as nonoperative treatment) having surgery elsewhere is low. Conversely, these patients were sufficiently symptomatic that they pursued treatment with us due to persistent symptoms after treatment by other providers, and they were unlikely to simply live with their symptoms if the nonoperative treatment was not helping and they knew that a relatively simple surgical treatment existed. Finally, the role of the os trigonum during treatment is similarly subject to detection bias, since radiographs were generally not obtained unless surgery was being considered (for example, in patients who were not responding satisfactorily to nonoperative treatment). For this reason, the presence of an os trigonum was not included in the multivariable analysis, and this study cannot answer the question of its role in the occurrence of FHL tendonitis.

Proportion of Patients Treated Without Surgery for FHL Tendonitis

The current study demonstrates that 44% of patients treated using an FHL stretching nonoperative protocol will have sufficient symptom improvement that they will decide to not have surgery. This is likely to be an underestimation of the true positive response rate to the stretching protocol, since a further 151 patients were started on the FHL stretching protocol and cancelled their follow-up appointment. Since they were explicitly told that they could do so if they were satisfied with their improvement, and given the medically underserved region in which we practice (there are only two other orthopaedic foot and ankle–trained surgeons in our state), it is unlikely that a sizable number would have sought care elsewhere. Furthermore, the period that we studied was designed to capture at least 1 year of subsequent clinical follow-up by the patient by anyone in our medical system. The concept of the FHL-specific stretch was first introduced by Michelson and Dunn [11], based on the concept that by holding the hallux dorsiflexed pretightens the FHL, which is then maximally tensioned during forced ankle dorsiflexion. This concept has been validated both in the lab [6, 7] and clinically [12]. Most patients in the current study had persistent symptoms despite nonoperative care provided for more than a year by previous providers. That their symptoms abated in a high percentage of cases solely with the institution of the FHL-specific stretching program is consistent with the paradigm that restriction of the FHL motion excursion was the underlying pathomechanical etiology of their condition. This finding extends that of a previous study using a similar regimen that demonstrated successful nonoperative treatment for FHL tenosynovitis in 46% (37 of 81) of patients [11]. Other reports have emphasized conservative treatment such as rest, ice, compression, and elevation (RICE) [1, 5, 15]; occasional immobilization; strengthening of the posterior muscles [8, 13, 19]; and vaguely defined physical therapy. The current study identifies a stretching program physiologically appropriate for the pathophysiology that yields satisfactory patient outcomes.

Factors Associated with a Patient’s Decision to Undergo Surgery

The presence of clinical hallux rigidus symptoms (OR 2.4) is associated with ultimately having surgery. Since entrapment of the FHL at the fibroosseous tunnel can result in decreased distal excursion of the tendon with consequent increased loading of the first MTP joint [6], the coexistence of hallux rigidus with FHL tendonitis is expected. That an FHL stretching program is less likely to achieve a satisfactory improvement in symptoms in these patients is likely due to the long-standing nature of the condition. In such patients, the limited FHL excursion probably predates the onset of clinical symptoms in the hallux [2], and may therefore be less responsive to attempts at increasing tendon motion through stretching. It is also possible that the pain at the hallux limited the ability to perform the FHL stretches effectively. Undergoing the FHL stretching protocol was the largest predictor of not deciding to have surgery (OR 0.15).

Conclusion

This study found that tendonitis of the flexor hallucis longus is a relatively underrecognized condition, with many patients in the current study having been seen by multiple previous providers without an accurate diagnosis having been made. Nor is it predominantly seen in athletes, as 75% of patients in the study did not pursue any athletic activities. The keys to making the diagnosis are having an awareness of the condition and eliciting specific signs confirming its presence, such as tenderness of the tendon at the fibroosseous tunnel and reduced hallux dorsiflexion on the FHL stretch test. Conservative treatment using an FHL stretching protocol resulted in 44% of patients having sufficient improvement to opt out of surgery. Future prospective studies should help in further defining which patients are more likely to benefit from this stretching program.