Introduction
Foot osteoarthritis (OA) is a significant cause of foot pain and disability in older adults [
1]. Midfoot OA is one of the most common forms of foot OA affecting 1 in 8 adults aged 50 years and over [
1], and is associated with increased age, female sex, lower socioeconomic class, obesity, pain in other weight-bearing joints, the presence of non-musculoskeletal comorbidities, and previous foot and ankle injury [
2]. Individuals with midfoot OA report impaired physical [
2‐
4] and mental [
2] function compared to individuals without midfoot OA, with over 80% reporting the condition to be disabling [
2]. The midfoot is a complex region made of multiple articulations, and midfoot OA has been reported to present as three phenotypes based on the radiographic pattern of joint involvement; medial midfoot (talonavicular, navicular-first cuneiform, or cuneiform-first metatarsal joint), the central midfoot (second cuneiform-metatarsal joint), or both the medial and central midfoot joints [
5]. Although midfoot OA is typically described as pain occurring at the dorsal aspect of the midfoot with the presence of radiographic changes of midfoot joints [
5,
6], there is no consensus regarding its definition [
7].
Currently, there are no evidence-based clinical guidelines to inform the management of midfoot OA. However, non-surgical interventions such as anti-inflammatory and analgesic medications, intra-articular corticosteroid injections, physical therapy, foot orthoses (FOs), and footwear modifications are commonly used as a first line approach to manage foot OA [
8,
9]. Non-steroidal anti-inflammatory drugs (NSAIDs) are anti-inflammatory and analgesic agents, primarily exerting their effects by inhibiting prostaglandin synthesis via inhibition of cyclooxygenase (COX) enzymes [
10]. Intra-articular corticosteroids [
11,
12] are potent anti-inflammatory compounds that exert their effects by acting directly on nuclear steroid receptors [
13]. Low-dose radiotherapy has been used for pain relief for foot and ankle osteoarthritis [
14], however the mechanism of action is unknown [
15]. As alterations in foot and lower limb biomechanics are likely to play an important role in the development and progression of midfoot OA [
16], interventions such as FOs and footwear modifications that can alter midfoot joint movement and forces during gait [
17‐
19] are speculated to be effective for midfoot OA [
19‐
22]. Where these non-surgical interventions are unable to improve symptoms, surgery can be considered [
9].
As no systematic review of interventions for midfoot OA currently exists, the aim of this study was to systematically review and summarise the evidence relating to studies that have evaluated the efficacy of non-surgical interventions for this condition.
Methods
Review registration
This review was prospectively registered with PROSPERO (CRD42021273375) and has been reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines [
23].
Study inclusion
All trial designs that included adults with midfoot OA confirmed by radiology or physician diagnosis who underwent any non-surgical intervention were eligible for the review. As there is currently no consensus definition for midfoot OA [
7], we included all trials that described any symptoms and/ or radiographic degenerative joint changes around the midfoot joints (Supplementary file 1). Trials were excluded if they included participants who: (i) were under 18 years of age, (ii) had neuromuscular or inflammatory arthritic conditions, or (iii) had undergone lower limb surgery. Single case reports, expert opinion pieces, protocols, abstracts without full text, or conference proceedings were excluded.
Search strategy
A comprehensive literature search was conducted without date restriction up to 16 September 2021 and updated on 23 February 2023 in the following electronic databases: Medline, CINAHL, Embase, the Cochrane Library, and the WHO International Clinical Trials Registry. Two strings of search terms were developed including MeSH terms, keywords, and synonyms: (i) “midfoot” AND “osteoarthritis” and (ii) “non-operative intervention” or “treatment” or “therapy”. Truncation, proximity, and Boolean operators were used as appropriate, and limiters were applied for human studies (Supplementary file 1). In addition to the electronic database search, reference lists of included trials were hand-searched, and citation tracking was performed. There were no date or language restrictions applied.
Study selection and data extraction
Search results were imported into Endnote 20.1 (Clarivate Analytics, New York, USA) and Covidence (Veritas Health Innovation, Melbourne, Australia), and duplicates were removed. Two investigators (PQXL, SEM) independently screened all titles and abstracts. Full-text articles were obtained if the investigators were not able to determine whether to include the record from the title and abstract. Any disagreements were discussed and resolved by a third investigator (HBM).
The following outcome measures to be extracted were pre-specified before reviewing the articles: (i) pain, (ii) function, (iii) health-related quality of life, (iv) number of participants experiencing any adverse event, and (v) number of participant withdrawals due to adverse events. Outcome measures were obtained for the following time-points: short term (0 to 12 weeks), medium term (> 12 to 52 weeks), long term (> 52 weeks). If two follow-up assessments were completed within one of the defined time-points, the results of the latter of the two assessments were selected. For studies that used multiple measures to evaluate the same outcome (e.g., multiple pain measures), a consensus approach (involving PQXL, SEM, KBL, HBM and MRK) was undertaken to select one outcome measure considered to be the most valid representation of the outcome. This was performed without knowledge of the results for these outcomes to minimise bias. Relevant data were then extracted independently by PQXL and SEM and entered into Microsoft® Excel (Microsoft Corporation, Redmond, Washington, USA) using a standardised extraction form. Attempts were made to contact the authors for missing data, and any disagreements were discussed and resolved by consensus with a third investigator (HBM).
Quality assessment
The methodological quality of included trials was assessed using the National Institutes of Health (NIH) Quality Assessment Tools [
24]. Randomised controlled trials were appraised using the ‘Controlled Intervention’ Tool, which consisted of 14 items. Case series trials were assessed using the ‘Before-After (Pre-Post) Studies With No Control’ Tool. For this tool, one item (‘if the intervention was conducted at a group level’) was excluded as it was considered not applicable to the samples in the trials included in our review, resulting in 11 items of this tool being used. Although the NIH tools [
24] allow an overall rating (‘poor’, ‘fair’ or ‘good’) to be applied to each trial, there were no specific scoring thresholds provided to determine the overall rating. Therefore, prior to conducting the quality assessment, a consensus approach (involving PQXL, SEM, KBL, MRK and HBM) was undertaken to determine key criteria that needed to be satisfied to be considered a poor, fair, or good quality trial. For the ‘Controlled Intervention’ Tool, the following items needed to be satisfied: all items for a good quality trial; items 2–8 and 11–14 for a fair quality trial; and none of items 2–8 and 11–14 for a poor quality trial. For the ‘Before-After (Pre-Post) Studies With No Control’ Tool, the following items needed to be satisfied: all items for a good quality trial; items 2, 3, 5, 6, 7, 9, 10 for a fair quality trial; and none of items 2, 3, 5, 6, 7, 9, 10 for a poor quality trial. Quality assessments were undertaken independently by PQXL and SEM, and any disagreements were resolved by consensus with two other investigators (KBL and HBM).
Data synthesis and analysis
Data were entered into the RevMan software program (V5.4.1; Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2014) to obtain estimates of treatment effect. For trials that included a control group, treatment effects from between-group analyses were used. For trials that did not have a control group (i.e., case series), treatment effects from within-group analyses were used. For continuous scaled outcome measures, estimates were analysed as mean differences (MDs) with 95% confidence intervals (CIs) and standardised mean differences (SMDs) with 95% CIs. Results are presented such that a positive MD and SMD value would indicate an effect favouring the experimental intervention (for between-group analyses) or an improvement in an outcome following treatment (for case series trials). SMD effect sizes and 95% CIs were calculated to obtain a measure of the magnitude of differences. SMD values were classified as very small (0.01), small (0.2), medium (0.5), large (0.8), very large (1.2), and huge (2.0) [
25]. For dichotomous scaled outcome measures, estimates were analysed as risk ratios with 95% CIs (for between-group analyses) or presented using descriptive statistics (proportions for within-group analyses). A quantitative synthesis (meta-analysis) was not performed as the data from included trials were not sufficiently homogenous due to the variability in trial designs, interventions, and the reported outcomes.
Discussion
This systematic review aimed to summarise the evidence for the efficacy of non-surgical interventions for midfoot OA. There are no randomised controlled trials – the existing trials are limited to a feasibility trial or case series trials. However, the current, albeit limited, evidence indicates that arch contouring FOs, shoe stiffening inserts and corticosteroid injections may be effective for improving pain and/or function in midfoot OA. The effects of interventions for midfoot OA on health-related quality of life are unknown. Further, there is limited evidence regarding the harms of interventions for midfoot OA, as few trials reported adverse events.
Two trials investigated the efficacy of arch-contouring FOs [
20,
22]. Arch contouring FOs are used to support the midfoot with a close-fitting orthotic shell and increase contact time and maximum force underneath the midfoot [
18,
20]. This is theorised to reduce bending moments across the midfoot joints, prevent medial longitudinal arch deformation and reduce midfoot joint compression and pain [
18,
26]. Arch contouring FOs were found to have none [
20] to large [
22] effects on pain in the short term (≤ 12 weeks), very large [
22] effects on pain in the medium term (> 12 to 52 weeks), medium [
20] to very large [
22] effects on function in the short term (≤ 12 weeks), and very large [
22] effects on function in the medium term (> 12 to 52 weeks). Although it appears that arch contouring FOs have very large effects in one trial [
22], the authors reported that a proportion of participants (36 out of 57) had additional rigid carbon fibre plates incorporated into the soles of their shoes. This could have potentiated the effects of the intervention. Overall, arch contouring FOs are well tolerated, with only one participant reporting an adverse event in one trial [
20]. However, the methodological quality of these trials was judged to be poor and the effects beyond 52 weeks have not been investigated. Appropriately powered randomised trials are required to determine the effects of arch contouring foot orthoses for midfoot OA.
The efficacy of shoe stiffening inserts was investigated in two trials [
17,
19]. Shoe stiffening inserts have been found to reduce first metatarsal joint dorsiflexion and first metatarsal plantarflexion during gait [
17]. As the proximal aspects of the first, second and third metatarsals form part of the distal articulation of the tarsometatarsal joint, reducing movement of the metatarsals is theorised to limit articular stress within the midfoot and potentially reduce midfoot joint pain [
17]. Shoe stiffening inserts were found to have no effects [
19] in reducing pain in individuals with mild midfoot OA in the immediate timepoint (within a session), huge effects [
19] in reducing pain in individuals with moderate midfoot OA in the immediate timepoint, medium effects [
21] on pain improvement in the short term (≤ 12 weeks), and small effects [
21] on function in the short term (≤ 12 weeks). Shoe stiffening inserts are well tolerated with one trial [
21] reporting no adverse events. However, these trials were judged to be of poor methodological quality and the effects of the intervention beyond 12 weeks are unknown.
The efficacy of image-guided intra-articular corticosteroid injection was investigated in two trials [
11,
12]. Corticosteroid injections are common anti-inflammatory pharmacological agents used for inflammation and pain management in ankle and knee osteoarthritis [
27,
28]. In the included trials, methylprednisolone was the primary agent used and was injected into midfoot joints guided by ultrasound [
11] or x-ray [
12]. Corticosteroid injection was found to provide short term (≤ 12 weeks) improvement in pain and function with heterogenous effects that ranged from small to large [
11,
12]. These intervention effects were not maintained in the medium (> 12 to 52 weeks) and longer term (> 52 weeks). The observed pattern of efficacy in the short term, but reducing in the medium and long term, is consistent with findings reported in trials evaluating corticosteroid injection for other musculoskeletal conditions of the foot such as plantar heel pain and tendon disorders [
29‐
31]. This suggests that corticosteroid injection alone may have limited utility for midfoot OA. Adverse events associated with intra-articular corticosteroid injections were reported in both trials and were uncommon (0 to 13.5%) and minor [
11,
12]. As the included trials were rated as poor quality, rigorous randomised trials are required to improve our understanding of the efficacy of corticosteroid for midfoot OA.
There are several strengths of this review. The review protocol was prospectively registered with PROSPERO. We used a robust search strategy that was comprehensive and not restricted by language or date. We used two independent investigators to screen studies for inclusion, perform data extraction and conduct quality assessment with conflicts resolved by two investigators. The reporting of data and results were cross checked. We also used pre-determined decision rules to identify and extract data for outcome measures at specific time points. Furthermore, we pre-specified important criteria within the NIH Quality Assessment Tools [
24] that had to be satisfied prior to rating the quality for each trial. As such, our assessments regarding the methodological quality of the evidence are transparent and unbiased.
The quality of the evidence in this review is limited by the small number and low level of evidence of the included trials, and their poor methodological quality. Our quality assessment identified several common issues across the included trials, so all were rated as poor quality. First, the inclusion criteria for the diagnosis of midfoot OA were unclear in most trials [
11,
12,
19,
21,
22] which makes it difficult to generalise the findings. Further work is required to develop a consensus definition for midfoot OA [
7]. Second, the outcome measures varied in type and assessment timepoints across the trials, and only three trials [
12,
20,
21] used validated measures for pain and function [
32]. Third, there was a lack of participant and assessor blinding across all trials [
11,
12,
19,
21,
22] apart from the feasibility trial [
20]. There was considerable intervention variability in the provision and types of FOs used, and the intra-articular corticosteroid injections performed were guided by different imaging techniques and operators. As such, comparisons could not be made between the included trials. Finally, only the feasibility trial [
20] provided a prospective sample size calculation, so it is possible that the remaining trials [
11,
12,
19,
21,
22] may have been under-powered leading to an uncertainty in the observed intervention effect [
33]. Although the authors [
20] reported appropriate sample size calculations adequate for a feasibility trial, key characteristics of participants in each group at baseline were notably different, and this may have confounded the reported effect of the intervention.
In conclusion, the aim of this review was to assess the current evidence for non-surgical interventions for midfoot OA. The available evidence indicates that arch contouring FOs may reduce pain and improve function in the short and medium term, and shoe stiffening inserts may improve pain and function in the short term. However, the long-term effects of these interventions are unknown. Although intra-articular corticosteroid injection therapy may reduce pain in the short term, the effects do not persist. Overall, the quality of evidence that these conclusions are drawn from is low. Therefore, rigorous randomised trials are required to evaluate the efficacy of these non-surgical interventions.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.