Does Minimally Invasive Surgery Provide Better Clinical or Radiographic Outcomes Than Open Surgery in the Treatment of Hallux Valgus Deformity? A Systematic Review and Meta-analysis

Abstract

Background

Hallux valgus is the most common foot deformity and affects 23% to 35% of the general population. More than 150 different techniques have been described for surgical correction. Recently, there has been increasing interest in the use of minimally invasive surgery to correct hallux valgus deformities. A variety of studies have been published with differing outcomes regarding minimally invasive surgery. However, most studies lack sufficient power and are small, making it difficult to draw adequate conclusions. A meta-analysis can therefore be helpful to evaluate and compare minimally invasive and open surgery.

Questions/purposes

We performed a systematic review and meta-analysis of randomized controlled trials and prospective controlled studies to answer the following question: Compared with open surgery, does minimally invasive surgery for hallux valgus result in (1) improved American Orthopaedic Foot and Ankle Society (AOFAS) scores and VAS scores for pain, (2) improved radiologic outcomes, (3) fewer complications, or (4) a shorter duration of surgery?

Methods

The systematic review and meta-analysis was conducted according to the guidelines of the Cochrane Handbook for Systematic Reviews of Intervention and the Preferred Reporting Items for Systematic Reviews and Meta-analyses. A search was performed in the PubMed, Embase, Scopus, CINAHL, and CENTRAL databases on May 3, 2022. Studies were eligible if they were randomized controlled or prospective controlled studies that compared minimally invasive surgery and open surgery to treat patients with hallux valgus. We defined minimally invasive surgery as surgery performed through the smallest incision required to perform the procedure accurately, with an incision length of approximately 2 cm at maximum. Open surgery, on the other hand, involves a larger incision and direct visualization of deeper structures. Seven studies (395 feet), consisting of six randomized controlled studies and one prospective comparative study, were included in the qualitative and quantitative data synthesis. There were no differences between the minimally invasive and open surgery groups regarding age, gender, or severity of hallux valgus deformity. Each included study was assessed for the risk of bias using the second version of the Cochrane tool for assessing the risk of bias in randomized trials or by using the Newcastle-Ottawa Scale for comparative studies. Most of the included studies had intermediate quality regarding the risk of bias. We excluded one study from our analysis because of its high risk of bias to avoid serious distortions in the meta-analysis. We performed a sensitivity analysis to confirm that our meta-analysis was robust by including only studies with a low risk of bias. The analyzed endpoints included the AOFAS score (range 0 to 100), where higher scores represent less pain and better function; the minimum clinically important difference on this scale was 29 points. In addition, the VAS score was analyzed, which is based on a pain rating scale (range 0 to 10), with higher scores representing greater pain. Radiologic outcomes included the hallux valgus angle, intermetatarsal angle, and distal metatarsal articular angle. Complications were qualitatively assessed and evaluated for differences. A random-effects model was used if substantial heterogeneity (I² > 50%) was found; otherwise, a fixed-effects model was used.

Results

We found no clinically important difference between minimally invasive and open surgery in terms of the AOFAS score (88 ± 7 versus 85 ± 8, respectively; mean difference 4 points [95% CI 1 to 6]; p < 0.01). There were no differences between the minimally invasive and open surgery groups in terms of VAS scores (0 ± 0 versus 0 ± 1, respectively; standardized mean difference 0 points [95% CI -1 to 0]; p = 0.08). There were no differences between the minimally invasive and open surgery groups in terms of the hallux valgus angle (12° ± 4° versus 12° ± 4°; mean difference 0 points [95% CI -2 to 2]; p = 0.76). Radiographic measurements of the intermetatarsal angle did not differ between the minimally invasive and open surgery groups (7° ± 2° versus 7° ± 2°; mean difference 0 points [95% CI -1 to 1]; p = 0.69). In addition, there were no differences between the minimally invasive and open surgery groups in terms of the distal metatarsal articular angle (7° ± 4° versus 8° ± 4°; mean difference -1 point [95% CI -4 to 2]; p = 0.28). The qualitative analysis revealed no difference in the frequency or severity of complications between the minimally invasive and the open surgery groups. The minimally invasive and open surgery groups did not differ in terms of the duration of surgery (28 ± 8 minutes versus 40 ± 10 minutes; mean difference -12 minutes [95% CI -25 to 1]; p = 0.06).

Conclusion

This meta-analysis found that hallux valgus treated with minimally invasive surgery did not result in improved clinical or radiologic outcomes compared with open surgery. Methodologic shortcomings of the source studies in this meta-analysis likely inflated the apparent benefits of minimally invasive surgery, such that in reality it may be inferior to the traditional approach. Given the associated learning curves—during which patients may be harmed by surgeons who are gaining familiarity with a new technique—we are unable to recommend the minimally invasive approach over traditional approaches, in light of the absence of any clinically important benefits identified in this meta-analysis. Future research should ensure studies are methodologically robust using validated clinical and radiologic parameters, as well as patient-reported outcome measures, to assess the long-term outcomes of minimally invasive surgery.

Introduction

Hallux valgus is the most common foot deformity and affects 23% to 35% of the general population [35]. Symptoms include pain, bursal inflammation, and ulceration, which lead to problems with mobility and wearing shoes [55]. The prevalence of hallux valgus deformity is higher among women than among men and increases with age [35]. More than 150 different surgical procedures have been described to treat hallux valgus [58]. However, consensus on a gold standard operative approach has not been established. Minimally invasive surgery (MIS) has been increasingly used in surgery in general and particularly in orthopaedic foot surgery [44]. MIS techniques for hallux valgus correction are increasing in popularity because of proposed advantages such as decreased stiffness, reduced postoperative morbidity, shorter surgical duration, and smaller scars [15, 44].

There are important differences in the key findings of studies on MIS for hallux valgus, with some supporting the technique and others recommending against it [5, 13, 22, 28-31, 36]. However, these results are mainly derived from small and underpowered studies and do not allow adequate conclusions and recommendations for clinical practice. This leaves surgeons with uncertainty regarding the adoption of these novel surgical approaches. A meta-analysis can therefore be helpful to evaluate and compare MIS and open surgery.

Thus, we asked, compared with open surgery, does MIS for hallux valgus result in (1) improved American Orthopaedic Foot and Ankle Society (AOFAS) scores and VAS scores for pain, (2) improved radiologic outcomes, (3) fewer complications, or (4) a shorter duration of surgery?

Materials and Methods

Search Strategy

The systematic review and meta-analysis was performed according to the guidelines detailed in the Cochrane Handbook for Systematic Reviews of Intervention and the Preferred Reporting Items for Systematic Reviews and Meta-analyses [17, 37]. We registered the review protocol in the PROSPERO database (CRD42022330035) [46]. The last search was performed on May 3, 2022, in the PubMed, Embase, Scopus, CINAHL, and CENTRAL databases. Furthermore, we conducted a gray-literature search for conference proceedings in Scopus and Embase, and the reference lists of the included studies were manually searched to identify studies that might have been missed in the systematic search. The search strategy was composed of three main concepts: minimal invasive, surgery, and hallux valgus. The different terms for each concept were combined with the OR operator, and the three concepts were combined using the AND operator (Supplemental Digital Content 1; http://links.lww.com/CORR/A971).

Inclusion and Exclusion Criteria

The inclusion and exclusion criteria were designated according to the Participants, Intervention, Comparison, Outcome, Study design strategy [7]. Studies were eligible if they were randomized controlled or prospective controlled studies that compared MIS and open surgery to treat patients with hallux valgus. We defined MIS as surgery performed through the smallest incision necessary to perform the procedure properly, with a maximal incision length of approximately 2 cm. We defined open surgery as a surgical approach performed through larger incisions and under direct visualization of deeper structures. The outcomes of interest were the AOFAS score and VAS score for pain at rest. Furthermore, radiologic outcomes (hallux valgus angle, intermetatarsal angle, and distal metatarsal articular angle) and complications were assessed. In addition to a minimum follow-up period of 6 months, the latest available follow-up results were analyzed to evaluate outcomes over a longer period, which allowed us to compare long-term results and consistency among patients. Further, we did not limit or filter the search strategy according to date, language, or patient age.

Data Collection and Abstraction

Each database was searched separately, and the resulting datasets were exported to Endnote^TM (version X9, Clarivate). After removing duplicates, we exported the final dataset to Covidence^TM (Veritas Health Innovation). Each step of the study selection and data extraction process was performed independently by two investigators (ARA and AO). Disagreements were resolved by discussion with the senior author (HW).

Data were extracted using a standardized proforma. For qualitative data synthesis, the following study characteristics were collected: authors, the period in which the study was conducted, year of publication, MIS techniques, open or conventional surgical techniques in the comparison group, demographic data of the study population, the country in which the study was conducted, number of centers involved in the study, duration of follow-up, and duration of surgery in minutes. We extracted AOFAS and VAS scores from the included studies. The radiologic parameters (hallux valgus angle, intermetatarsal angle, and distal metatarsal articular angle) along with complications and duration of surgery were also obtained.

For quantitative data synthesis, means, standard deviations, and sample size were taken at the latest available follow-up interval. In the absence of means and SDs, medians and ranges were converted, as recommended by Wan et al. [56].

Study Characteristics

The search strategy identified 11,768 articles, which yielded 6458 records after we removed duplicates (Fig. 1). The titles and abstracts of the 6458 articles were screened, and irrelevant articles were excluded. Eighty-three studies were assessed for eligibility. Eight studies, consisting of seven randomized controlled studies and one prospective comparative study, were identified and met the inclusion criteria. A study with a high risk of bias and inconsistent data reporting was excluded from our analysis [39]. Seven studies were included in the qualitative and quantitative data synthesis [10-12, 23, 25, 42, 52].

Fig. 1.

This Preferred Reporting Items for Systematic Reviews and Meta-analyses flow diagram shows the studies that were included in our review.

A total of 395 feet were included in the data synthesis. Because patient-level data were unavailable, bilaterally treated hallux valgus was included. There were no differences between the MIS and open surgery groups in terms of age (50 ± 17 years versus 51 ± 15 years) and gender (91% [177 of 195] women versus 92% [183 of 200] women) (Table 1). All included studies used distal metatarsal osteotomies to correct mild-to-moderate hallux valgus and provide appropriate similarity to allow pooling.

Table 1.

Characteristics of the included studies

Author	Feet		Women		Men		Mean age in years		Surgical approach		Mean follow-up in months		Risk of biasa	Country
	MIS	Open	MIS	Open	MIS	Open	MIS	Open	MIS	Open	MIS	Open
Radwan and Mansour 2012 [42]	29	31	25	28	4	3	33 ± 7	36 ± 7	Bösch	Chevron	22± 9	20 ± 9	Intermediate	Egypt
Giannini et al. 2013 [12]	20	20	20	20	0	0	53 ± 11	54 ± 11	SERI	Scarf	84	84	Intermediate	Italy
Lee et al. 2017 [25]	25	25	23	22	2	3	53	53	MICA	Scarf-Akin	6	6	Low	Australia
Frigg et al. 2019 [11]	48	50	41	44	7	6	48 ± 25	48 ± 19	MICA	Scarf-Akin	24	24	Low	Switzerland
Kaufmann et al. 2020 [23]	19	20	16	17	3	3	54 ± 15	47 ± 14	MICA	Chevron	67 ± 6	67 ± 6	Low	Austria
Torrent et al. 2021 [52]	30	28	28	28	2	0	61 ± 28	64 ± 24	Mini-scarf	Scarf	21 ± 19	21 ± 19	Intermediate	Spain
Dragosloveanu et al. 2022 [10]	24	26	24	24	0	2	49 ± 15	55 ± 14	MICA	Chevron	12	12	Intermediate	Romania

^aRisk of bias was assessed using the second version of the Cochrane tool for assessing the risk of bias in randomized trials or by using the Newcastle-Ottawa Scale for cohort studies. MIS = minimally invasive surgery; SERI = simple, effective, rapid, and inexpensive; MICA = minimally invasive chevron-akin osteotomy.

Specifically, four studies use a minimally invasive chevron osteotomy [10, 11, 23, 25]. Two further studies used a combined approach of a minimally invasive chevron-akin osteotomy [11, 25]. The minimally invasive chevron-akin was first described by Redfern and Perrera [44]. The minimally invasive chevron-akin osteotomy is performed with a 2-mm burr under fluoroscopic control. Periosteal elevators are inserted into the diaphysis of the proximal fragment and levered laterally to translate the metatarsal head [44]. As opposed to other minimally invasive techniques, the minimally invasive chevron-akin osteotomy uses cannulated screws to achieve stable internal fixation and adheres to the principles of the Arbeitsgemeinschaft für Osteosynthesefragen group for internal fixation [44]. Kaufmann et al. [23] and Dragosloveanu et al. [10] used a single dorsal screw fixation technique, whereas Lee et al. [25] used two medial-to-lateral screws. Frigg et al. [11] used a combination of these techniques. In contrast, Giannini et al. [12] performed a modified Bösch osteotomy, the simple, effective, rapid, and inexpensive (SERI), on 20 feet. SERI and Bösch osteotomies are characterized by transverse subcapital osteotomies that are fixed by percutaneous medial K-wires inserted into the medullary canal. They differ in terms of the approach because a Bösch osteotomy uses a 2-mm burr, whereas a SERI osteotomy uses an oscillating saw. Furthermore, Torrent et al. [52] used a mini-scarf osteotomy in 30 patients and assessed the outcome compared with that of open scarf osteotomy in 28 patients. The scarf osteotomy is characterized by a z-shaped osteotomy stabilized by screws. During a minimally invasive scarf osteotomy, a 0.5-cm-long incision is made and a 2-mm burr is used to perform the osteotomy [12]. By contrast, open scarf osteotomies are performed through a 3-cm-long incision and are made with 9-mm saws [12].

Assessment of Study Quality

Each study was assessed for the risk of bias using the second version of the Cochrane tool for assessing the risk of bias in randomized trials or by using the Newcastle-Ottawa Scale for comparative studies [49, 57]. Two authors (ARA and AO) reviewed the risk of bias, and any disagreements were resolved through discussion and consensus with the senior author (HW). The Cochrane tool allows for an assessment of the risk of bias by evaluating six domains independently. Whenever all domains were rated as low or up to one domain was rated as intermediate, a study was deemed to have a low risk of bias. The studies classified as having a high risk of bias had a high risk of bias in at least one domain.

The evaluation of the individual domains for each study is visualized by a risk of bias graph and a risk of bias summary figure that was generated using robvis [33]. We found that three studies [11, 23, 25] had low risks of bias, four [10, 12, 42, 52] had an intermediate risk, and one [39] had a high risk (Fig. 2). We therefore excluded this study [39] because of its high risk of bias and inconsistencies in our quantitative data analysis. To further assess the robustness of our findings, we performed an additional sensitivity analysis by only including studies deemed to have a low risk of bias. The sensitivity analysis did not reveal any changes from our original findings and confirmed its robustness.

Fig. 2.

An assessment of the risk of bias is shown. We made judgements regarding each risk of bias domain in (A) randomized controlled studies and (B) comparative studies. Assessment is visualized as (C) percentages across all studies. A color image accompanies the online version of this article.

Primary and Secondary Study Outcomes

Our primary study goal was to determine whether MIS can provide better symptom resolution and perceptible benefits to patients than open surgery. To achieve this, we pooled data from studies reporting the AOFAS and VAS scores. The AOFAS score is a nonvalidated, clinician-based, region-specific outcome measure with a maximum score of 100, representing better function and no symptoms or impairment [3, 8, 19, 47, 48, 50]. It evaluates pain, function, and alignment and incorporates subjective and objective factors [24]. As one of the most widely used outcome measures, AOFAS scores can be used for comparisons between studies [34]; the minimum clinically important difference on this scale was 29 points [6].

The AOFAS score was reported in seven studies [10-12, 23, 25, 42, 52]. To further evaluate pain, we extracted the VAS score from four studies [10, 11, 23, 25]. The VAS is a self-reported measure of pain (range 0 to 10), with higher scores representing more pain [4].

Our secondary study goals were to assess and compare the accuracy of MIS with that of open surgery. For this analysis, we pooled data from studies reporting radiographic measurements, complications, and duration of surgery. Radiographic measurements include the hallux valgus angle (HVA), intermetatarsal angle (IMA), and distal metatarsal articular angle (DMAA), which were obtained from seven [10-12, 23, 25, 42, 52], six [10-12, 25, 42, 52], and four [11, 12, 23, 52] studies, respectively. The HVA represents the angle between the longitudinal axes of the first metatarsal and the proximal phalanx of the hallux, with normal values ranging from 10° to 15° [20]. The IMA is formed by the longitudinal axis of the first and second metatarsals, with values of 7° to 9° considered normal [20]. The DMAA is the angle between the distal articular surface and the longitudinal axis of the fist metatarsal. DMAA angles smaller than 20° are considered normal [41]. We extracted data from the latest available follow-up interval. We evaluated and compared the duration of surgery for MIS and open surgery, which was reported in four studies [11, 12, 42, 52]. Additionally, we evaluated and compared the frequency and severity of complications across all studies qualitatively, because pooling complications with disparate severity levels is inappropriate. We were not able to perform a sex- and gender-specific analysis because patient-level data were not available.

Meta-analysis Methods

Statistical calculations were performed using R. The meta-analysis was performed using the “metafor” package [14, 51, 54]. Heterogeneity between studies was assessed using the I² measure, assuming a value between 0% and 100% [18]. Depending on the I² value, the heterogeneity was categorized into low (< 50%) or substantial (> 50%) degrees of heterogeneity [9].

The AOFAS score (I² = 65%) and the radiographic parameters of HVA (I² = 82%), IMA (I² = 82%), and DMAA (I² = 63%), as well as the duration of surgery (I² = 86%), revealed substantial heterogeneity; thus, a random-effects model was used in the meta-analysis.

Regarding VAS scores (I² = 0%), low heterogeneity was found, and thus a fixed-effects model was used. We used the inverse-variance method for continuous data [9].The summary effect’s confidence interval was calculated using the Hartung-Knapp adjustment for the random-effects model [16]. The methods proposed by Wan et al. [56] were used to calculate means and standard deviations, if median and ranges were reported. We assessed and confirmed our estimates using a second sensitivity analysis that excluded two studies [11, 23]; in this analysis. we converted medians and ranges. There were differences among the included studies regarding measurement methods for VAS scores (continuous bounded scales or ordinal scales [10, 11, 23, 25]). Our method for standardizing the results of the included studies was to use the standardized mean difference. In this way, related data reported using different scales could be compared [17].

Publication Bias

We intended to use funnel plot techniques, the Begg rank correlation test, and the Egger regression test for assessing publication bias. However, neither funnel plots nor more advanced regression analyses were performed because of insufficient numbers of included trials. A meta-analysis with fewer than 10 studies should not test funnel plot asymmetry because fewer studies cannot distinguish chance from asymmetry [17].

Power Analysis

We performed a power calculation for our endpoints to indicate what difference (delta) can be detected, with the numbers available, at an alpha level of 0.05 and with 80% power. Regarding the AOFAS and VAS scores, a difference of 2.12 of 100 points can be detected for the AOFAS and 0.37 of 10 points can be detected for the VAS. In terms of radiographic outcomes, differences of 1.13°, 0.70°, and 1.58° can be detected for the HVA, IMA, and DMAA, respectively. Considering the duration of surgery, we can detect a difference of 3.38 minutes, with the numbers available.

Results

Pain and Function

We found no clinically important difference between MIS and open surgery in terms of the AOFAS score (88 ± 7 versus 85 ± 8, respectively; mean difference 4 points [95% CI 1 to 6]; p < 0.01) (Supplemental Table 1; http://links.lww.com/CORR/A972). There were no differences between the MIS and open surgery groups in terms of VAS scores (0 ± 0 versus 0 ± 1; standardized mean difference 0 points [95% CI -0.1 to 0]; p = 0.08) (Fig. 3).

Fig. 3.

Forest plots of (A) the AOFAS score and (B) VAS score are shown. AOFAS = American Orthopaedic Foot and Ankle Society; MIS = minimally invasive surgery; MD = mean difference; SMD = standardized mean difference.

Radiographic Correction

In terms of radiographic parameters, there were no differences between the MIS and open surgery groups. There were no differences between the MIS and open surgery groups in the HVA (12° ± 4° versus 12° ± 4°; mean difference 0 points [95% CI -2 to 2]; p = 0.76) (Fig. 4). Radiographic measurements of the IMA did not differ between the MIS and open surgery groups (7° ± 2° versus 7° ± 2°; mean difference 0 points [95% CI -1 to 1]; p = 0.69) (Fig. 4). In addition, there were no differences between the MIS and open surgery groups in terms of DMAA (7° ± 4° versus 8° ± 4°; mean difference -1 point [95% CI -4 to 2]; p = 0.28) (Fig. 4).

Fig. 4.

Forest plots of radiologic parameters are shown for (A) the hallux valgus angle (HVA), (B) intermetatarsal angle (IMA), and (C) distal metatarsal articular angle (DMAA). MD = mean difference.

Complications

There was no obvious difference between the MIS (Supplemental Table 2; http://links.lww.com/CORR/A973) and open surgery groups (Supplemental Table 3; http://links.lww.com/CORR/A974) in terms of the frequency and severity of complications. We analyzed complications qualitatively because it is inappropriate to combine complications that have disparate severity levels; however, when we did so, we found no differences between MIS and open surgery for treating hallux valgus.

Duration of Surgery

The MIS and open surgery groups did not differ in terms of the duration of surgery (28 ± 8 minutes versus 40 ± 10 minutes; mean difference -12 minutes [95% CI -25 to 1]; p = 0.06) (Fig. 5).

Fig. 5.

A forest plot of the duration of surgery is shown. MD = mean difference.

Discussion

Hallux valgus is the most common foot deformity; it affects 23% to 35% of the general population [35]. A variety of techniques have been described for surgical correction [58]. However, there is still no consensus on a standard operative approach. MIS has been increasingly used in surgery in general and particularly in orthopaedic foot surgery [44]. The rationale for using MIS techniques for hallux valgus correction includes a number of proposed advantages, including decreased stiffness, faster recovery, reduced postoperative morbidity, shorter surgical duration, and smaller scars [15, 44]. However, studies disagree regarding whether MIS is advantageous or should be adopted [5, 13, 22, 28-31, 36]. These results are mainly derived from small and underpowered studies and do not give recommendations for clinical practice. This leaves surgeons with uncertainty regarding the adoption of these novel surgical approaches. A meta-analysis can therefore be helpful to evaluate and compare MIS and open surgery. In this meta-analysis, we found no difference between the MIS and open surgery groups across most endpoints. This is particularly important because most of the included studies had intermediate quality. Biases that accrue in such studies tend to inflate the benefits and underestimate the harms and risks of MIS; for example, they overlook or exclude learning curves and use nonvalidated outcome measures. Surgeons may be intrigued by emerging techniques but need to carefully evaluate their risks. In particular, the learning curve and its associated risks and complications are important when considering adopting minimally invasive techniques [27]. We are unable to recommend the MIS approach over traditional approaches, in light of the absence of any clinically important benefits to MIS identified in this meta-analysis. Future studies should be methodologically robust and use validated clinical and radiologic parameters as well as patient-reported outcome measures to assess the long-term outcomes of MIS.

Limitations

Our review has some limitations. First, MIS does not consist of one single procedure but rather includes a variety of techniques, such as the Bösch osteotomy and minimally invasive chevron-akin osteotomy. There were even differences between studies using the same technique such as the minimally invasive chevron-akin osteotomy; some surgeons used two screws, while others used only one [11, 23]. By lumping these approaches, heterogeneity is increased and the ability to identify true differences may be diminished. However, a random-effects model was used to further account for heterogeneity. Additionally, because these studies used distal metatarsal osteotomies, particularly because point estimates were similar in all MIS subgroups, it is logical to combine them.

A further limitation of our study was the inclusion of studies that enrolled bilateral feet in the same patients and treated them as independent observations. In general, doing so is not appropriate. If bilateral feet are assumed to be independent in the statistical analysis, the sample size, confidence intervals, and levels of significance may be inaccurately estimated [40, 43]. There might be false perceptions of a therapeutic benefit, and MIS may appear to be more effective than it is. However, because patient-level data were not available, bilateral procedures could not be excluded from this meta-analysis, and caution is advised when interpreting these findings. An additional methodologic limitation is the use of an as-treated analysis in at least one of our source studies; it was specifically mentioned in one of the included studies [11], and may have been used in others. This analytic approach may result in inflated estimates of the treatment benefit of the new (MIS) treatment compared with studies that used an intention-to-treat analysis [32]. Given that we found few differences between the MIS and traditional approaches in this meta-analysis, this limitation could mean the MIS approach is not as beneficial.

One limitation of our study is the pooling of studies with different study durations. The included studies differed regarding length of follow-up, and they assessed outcomes at different timepoints. To establish what patients might expect at full or near-full recovery, we decided to evaluate our endpoints at the latest available follow-up and focus on long-term outcomes. Offsetting this concern, only one study [11] had a follow-up period of less than 1 year (in that study, patients were evaluated at 6 months); all other studies had a 1-year minimum follow-up period. Therefore, we believe our endpoints are similar enough to allow us to pool the results because they represent patients who are likely to be fully recovered or nearly so.

Another limitation of our study is the lack of sex- and gender-specific analyses, which could not be performed because patient-level data were unavailable. Treatment results may differ because of sex- or gender-related differences. Most of the included patients were women, and we did not perform a by-gender analysis. Thus, caution is warranted, because the results, obtained in a population consisting mostly of women, may not apply to men.

Furthermore, medians and ranges were converted into means and SDs using the formula provided by Wan et al [56]. Therefore, converting non-normally distributed data might have reduced the precision of our meta-analysis. To assess the impact of this conversion on our conclusion, we conducted a sensitivity analysis. The sensitivity analysis, however, did not reveal any differences in results when these converted data were excluded. Furthermore, our study is limited because of the different measurement scales used in the included studies. Studies used a continuous bounded interval and ordinal scales to measure VAS scores. Because of the differences in assessment among the studies, these results may be more unstable. However, the standardized mean difference allowed us to standardize the study results in our meta-analysis. Thus, related data reported using disparate scales can be compared, minimizing the risk of bias. Another limitation of our study is the use of the AOFAS score, which is a nonvalidated outcome measure. There is controversy over the AOFAS score because it may result in biased assessments [3, 8, 19, 47, 48, 50]. Nonetheless, the AOFAS score remains one of the most widely used outcome measures and allows for a comparison between studies [34]. The AOFAS score, however, may overstate effectiveness, be misleading, and tend to inflate the apparent benefits of MIS. Therefore, its results should be interpreted cautiously.

AOFAS Scores, VAS Scores, and Radiologic Correction

We found no clinically important differences in AOFAS scores and no differences in VAS pain scores between minimally invasive and open surgery. We caution readers not to overvalue the small statistical difference we observed in the AOFAS score; first, as noted, this outcomes tool is not validated. More importantly, prior research suggests that the minimum clinically important difference for the AOFAS score is 29 points [6], which is much greater than our reported mean difference of 3.6 points. This suggests that patients are very unlikely to perceive any clinical advantage to MIS as assessed with that scoring tool. Likewise, we found no differences between the approaches in terms of radiographic correction. Based on these findings, we believe our results do not support the adoption of MIS techniques over open surgery.

Complications

We analyzed complications qualitatively, because pooling complications with disparate levels of severity is not appropriate. When we did so, it did not appear there were differences in either the severity or frequency of serious complications between MIS and open surgery for hallux valgus, but this approach to analysis has obvious shortcomings, as noted earlier. The complications associated with MIS of hallux valgus are similar in many ways to those of open surgery [1, 12, 15, 42, 45]. The most frequent complication was implant removal. This was widely reported in SERI and Bösch osteotomies using K-wires and in minimally invasive chevron-akin osteotomies using cannulated screws [23]. However, screw designs have improved, leading to new screw morphology that may reduce irritation of the soft tissues, potentially decreasing the frequency of removal [25]. In addition, there may be more complications among patients treated with earlier generations of MIS techniques, including SERI and Bösch osteotomies. This may reflect the uncertain nature of new techniques and the unrecognized risks associated with them. In a study by Kadakia et al. [22], SERI osteotomy was performed in 13 patients, five of whom experiences a recurrent hallux valgus deformity. Additionally, nine of the 13 patients experienced dorsal malunion [22]. However, these complications may have been a function of the learning curve associated with MIS for hallux valgus deformities, because the authors had little experience in MIS [27, 53]. This should not be dismissed—any surgeon adopting a new technique might find him or herself in this same situation, and the risk of novelty and potential harms arising in patients when the surgeon is surmounting a learning curve must be considered [26]. Together with a lack of benefit of MIS seen in our meta-analysis, we urge caution when adopting such new approaches. Owing to burrs, burning of the skin may be present in up to 13% of MIS-treated patients [25]. However, none of our included studies reported wound complications such as burning of the skin, perhaps because most of the included studies initiated their analyses after the participating surgeons were beyond their learning curves [11, 12, 23, 25, 39, 42]. Burning of the skin may occur when less-experienced surgeons adopt MIS techniques [25].

Several factors contribute to the challenging learning curve associated with MIS for hallux valgus. Surgeons must practice using MIS-specific instruments and gain experience regarding tactile sensations, which is time- and resource-consuming [2, 21, 38]. Training in cadavers may help decrease the risk of complications and shorten the learning curve [44]. However, complications during a surgeon’s learning curve should not be dismissed as merely a byproduct of that learning curve; they are real and may have lasting consequences for the patients who experience them.

Duration of Surgery

We found no difference between MIS and open surgery in terms of the duration of surgery. Notably, Palmanovich et al. [38] reported a learning curve in which it took 2 years for a surgeon to achieve comparable operative times to those of open surgery. Most included studies did not comment on the learning curve associated with MIS and rather began the study after participating surgeons had surmounted the learning curve. Thus, the duration of MIS may be longer than reported when a surgeon is early in the learning curve, and caution should be exercised when interpreting our results regarding surgical duration.

Conclusion

In our meta-analysis of the best-available evidence, MIS for hallux valgus deformity did not provide improved clinical or radiologic outcomes compared with open surgery. The current evidence to support MIS is limited because of the biases noted earlier (which likely inflated the apparent benefits of MIS; in reality, it may be inferior to the traditional approach), small sample sizes, and underpowering. Given the associated learning curves—during which patients may be harmed by surgeons who are gaining familiarity with a new technique—we are unable to recommend the MIS approach over traditional approaches, in light of the absence of any clinically important benefits to MIS surgery identified in this meta-analysis. Future studies should be methodologically robust and use validated clinical and radiologic parameters, as well as patient-reported outcome measures, to assess the long-term outcomes of MIS.