Increased Incidence and Severity of Postoperative Radiographic Hallux Valgus Interphalangeus With Surgical Correction of Hallux Valgus

Methods

All bunion surgeries in adults performed at a single center by 2 attending surgeons from July 1, 2009, to September 30, 2012, were evaluated for inclusion. The surgeries were identified by a Current Procedural Terminology (CPT) code search for all bunion correction surgeries (28290-9). Exclusion criteria included prior bony surgery to the first ray, inadequate films, nonadult bunion, or treatment with a surgery other than the above-mentioned bunion correction surgeries. There were a total of 298 procedures in 295 feet (265 patients) that underwent a bunion surgery in the time period. Those with an Akin osteotomy (28298) were excluded initially by the CPT code but were confirmed by review of postoperative radiographs. Sixty-nine feet were excluded because of performance of Akin osteotomy at the time of surgery. Seventeen were excluded because the surgery performed was not a bony bunion correction (eg, first metatarsophalangeal fusion was performed, not one of the above-mentioned HV surgeries). Twenty-four were excluded because these were not the index procedure on the first ray, which would affect the preoperative angular measurements. Two procedures were outside the date range, despite the CPT search bringing up the subjects. One subject was excluded for being an adolescent. Ninety-four were excluded for inadequate pre- or postoperative films. This includes those who had their preoperative radiographs taken at outside facilities that were not uploaded into our system, those who did not have weight-bearing pre- or postoperative films, and those who lacked either pre- or postoperative films. Ninety-two feet in 82 patients were eligible for the study.

Films were then reviewed by 2 senior resident orthopedic surgeons matched into foot and ankle fellowships who were not the operating surgeons. The films selected for review were the latest possible preoperative weight-bearing radiograph and the latest possible postoperative weight-bearing radiograph. This radiograph was selected for several reasons. First, the later follow-ups are more likely to be weight-bearing and are taken out of plaster. Second, the previously referenced study by Park et al ⁷ used intraoperative radiographs to demonstrate development of HVIP and we desired to show longer-term results that demonstrated lasting HVIP even after initiation of weight-bearing and, ultimately, healing. Each pre- and postoperative radiograph was reviewed for 2 HV angular measurements and 5 HVIP measurements. Each measurement was done 3 times by each of the 2 orthopedic surgeons. The angles were hallux valgus angle (HVA), first intermetatarsal angle (IMA), hallux interphalangeus angle (HIA), distal metatarsal articular angle (DMAA), proximal phalangeal articular angle (PPAA; also known as distal articular set angle), proximal to distal phalangeal articular angle (PDPAA), and total distal deformity angle (TDD).

The HVA was measured as the angle between a line down the shaft of the hallux metatarsal and a line down the shaft of the proximal phalanx, using the midpoint of the shaft at each metaphysis to draw each line (Figure 1A). The IMA was measured as the angle between a line down the shaft of the first and second metatarsals, using the midpoints at each metaphysis as a guide (Figure 1B). The HIA was measured as the angle between the shaft of the distal phalanx and the proximal phalanx of the hallux (Figure 2A). The DMAA was measured as the angle between a line down the shaft of the hallux metatarsal and the distal articular surface of the hallux, excluding osteophytes (Figure 2B). The PPAA was measured as the angle between the proximal articular surface of the proximal phalanx of the hallux, excluding osteophytes, and another line up the shaft of the proximal phalanx, using the midpoints of each metaphysis as a guide (Figure 2C). The PDPAA was measured between the proximal and distal articular surfaces of the proximal phalanx (Figure 2D). The TDD was measured as described by Elliot and Saxby ⁴ (Figure 2E). Here, a line was made at the borders of the articular surface of the hallux distal phalanx. A line was made at the midpoint of this line toward the medial edge of the proximal articular surface of the proximal phalanx. The angle between that line and perpendicular to the proximal articular surface was measured.

OPEN IN VIEWER

Figure 1.

Hallux valgus angular measurements. All preoperative and postoperative weight-bearing radiographs were assessed for (A) hallux valgus angle and (B) intermetatarsal angle.

OPEN IN VIEWER

Figure 2.

Hallux valgus interphalangeus angular measurements. All preoperative and postoperative weight-bearing radiographs were assessed for (A) hallux interphalangeus angle, (B) distal metatarsal articular angle, (C) proximal phalangeal articular angle, (D) proximal to distal phalangeal articular angle, and (E) total distal deformity angle.

Results

The average HVA improved from preoperative 27 ± 9.7 degrees to postoperative 11 ± 8.1 degrees, for a mean improvement of 16 ± 9.5 degrees (P < .001). Preoperative IMA was 13.6 ± 3.4 degrees and postoperative IMA was 6.1 ± 3.7 degrees, for a mean improvement of 7.7 ± 4.9 degrees (P < .001).

The average HIA worsened after HV surgery without an Akin osteotomy (Figures 3 and 4). Preoperative HIA was 7.2 ± 6.6 degrees and postoperative HIA was 13.2 ± 6.0 degrees, for a mean worsening of 6.0 ± 7.7 degrees (P < .001). DMAA also worsened from 7.3 ± 5.63 degrees preoperatively to 9.2 ± 5.5 degrees postoperatively, for a mean worsening of 1.9 ± 6.1 degrees (P = .001). This mean angle is still less than the defined abnormal angle of 10 degrees. PPAA worsened from 3.2 ± 2.1 degrees to 6.2 ± 3.6 degrees for a mean worsening of 3.0 ± 3.7 degrees (P < .001). In addition, with an upper limit of normal of 6 degrees, the average postoperative measurement was diagnostic of HVIP. PDPAA worsened from 6.7 ± 3.8 degrees preoperatively to 8.2 ± 4.1 degrees postoperatively. The TDD increased from 14.6 ± 3.7 degrees to 17.9 ± 3.1 degrees, for a mean worsening of 3.3 ± 3.3 degrees (P < .001).

OPEN IN VIEWER

Figure 3.

Change in angular measurements after hallux valgus corrective surgery. The hallux valgus angle (HVA) and intermetatarsal angle (IMA), measures of hallux valgus, were both improved postoperatively. The hallux interphalangeus angle (HIA), distal metatarsal articular angle (DMAA), proximal phalangeal articular angle (PPAA), proximal to distal phalangeal articular angle (PDPAA), and total distal deformity angle (TDD) were all worsened with surgery that did not address hallux valgus interphalangeus. Error bars demonstrate standard error.

OPEN IN VIEWER

Figure 4.

Pre- and postoperative hallux valgus interphalangeus (HVIP) measurements. Each of the quantitative measurements of HVIP increased on postoperative weight-bearing radiographs. Error bars represent standard error. DMAA, distal metatarsal articular angle; HIA, hallux interphalangeus angle; PPAA, proximal phalangeal articular angle; PDPAA, proximal to distal phalangeal articular angle; TDD, total distal deformity angle.

Correction of the HVA was strongly correlated to worsening of the HIA (R = −0.64) (Figure 5). Correction of the HVA was moderately correlated to worsening of the TDD (R = −0.40) and the PPAA (R = −0.32). Correction of the HVA was not strongly correlated with change in the DMAA (R = 0.036) or PDPAA (R = 0.030). Correction of the IMA was moderately correlated to worsening of HIA (R = −0.55), PPAA (R = −0.47), and TDD (R = −0.49). It was weakly correlated with improvement of the DMAA (R = 0.22) and worsening of the PDPAA (R = −0.23).

OPEN IN VIEWER

Figure 5.

Correlation of improvement of hallux valgus angle (HVA) to change in hallux interphalangeus angle (HIA). Improvement in HVA is negatively correlated (R = 0.64) with change in HIA. As correction of HVA increases, the apparent HIA on postoperative weight-bearing radiographs worsens.

The prevalence of HVIP pre- and postoperatively as defined by Gentili et al, ⁵ using measurements of HIA, DMAA, and PPAA, was compared in pre- and postoperative groups; the results are shown in Table 1. Preoperatively, 26% (24/92) had HVIP as defined by HIA greater than 10 degrees; this number increased to 79% postoperatively (73/92) (P < .001). By measurement of DMAA, where normal is less than 10 degrees, ¹⁰ 22% (21/92) had HVIP preoperatively, demonstrating a significant increase in HVIP given that 43% (40/92) had HVIP postoperatively (P = .0029). HVIP as defined by PPAA greater than 6 degrees was found in 12% (11/92) preoperatively but was found in 46% (42/92) postoperatively (P < .001).

OPEN IN VIEWER

Table 1.

Prevalence of Hallux Valgus Interphalangeus (HVIP) Pre- and Postoperatively ^a .

Angle Measured	Prevalence of Preoperative HVIP, %	Prevalence of Postoperative HVIP, %	P Value
HIA	26	79	<10−8
DMAA	22	43	.0029
PPAA	12	46	<10−6

a

There was a significantly increased prevalence of HVIP postoperatively by measurements of hallux interphalangeus angle (HIA), distal metatarsal articular angle (DMAA), and proximal phalangeal articular angle (PPAA).

There were 20 chevron osteotomies performed, 2 Lapidus procedures, 57 proximal closing wedge osteotomies, and 13 Scarf osteotomies. As a standard part of each procedure, a modified McBride and medial eminence resection was also performed. ANOVA demonstrated that HIA was significantly different between each of these groups, so a Student t test was performed to determine significance between groups. Table 2 summarizes the results of the Student t test. This demonstrated that the Scarf osteotomy actually improved HIA by 0.32 degrees but the difference was only significant between the effects of Scarf osteotomy and chevron osteotomy (P = .0053), Scarf osteotomy and Lapidus procedure (P = .036), and Scarf osteotomy and proximal closing wedge osteotomy (P < .001), while the remainder of the osteotomy types worsened the HIA. There were no differences between Lapidus procedure, proximal metatarsal closing wedge osteotomy, and Scarf osteotomy in regard to HIA. These worsened the HIA by 7.4, 8.5, and 8.5 degrees, respectively. The DMAA was significantly improved by 2.7 degrees by the chevron osteotomy. This difference was significant between chevron osteotomy and each of the other procedures: Lapidus procedure (P = .023), proximal closing wedge osteotomy (P < .001), and Scarf osteotomy (P < .001). These 3 all worsened DMAA by 3.2, 3.3, and 2.8 degrees, respectively; there were no significant between-group differences among the 3. TDD was worsened by the Scarf osteotomy by 2.3 degrees, which was significantly different from the Lapidus procedure, which worsened it by 4.9 degrees (P = .027), and the proximal closing wedge osteotomy, which worsened it by 4.2 degrees (P < .001). The chevron osteotomy worsened the TDD by 1.2 degrees, but the difference was not significant between the Scarf osteotomy and the chevron osteotomy (P = .90). There were no other differences between the groups. There were no significant differences between the surgeries for PPAA or for PDPAA.

OPEN IN VIEWER

Table 2.

Change in Measurement of Hallux Valgus Interphalangeus (HVIP) Angles Stratified by Type of Procedure ^a .

HVIP Angle	Procedure	Change, degrees	Procedure	Change, degrees	P Value
HIA	Chevron	−2.6	Lapidus	−7.4	.94
	Chevron	−2.6	Prox	−8.5	.58
	Chevron	−2.6	Scarf	0.32	.0053 b
	Lapidus	−7.4	Prox	−8.5	.98
	Lapidus	−7.4	Scarf	0.32	.037 b
	Prox	−8.5	Scarf	0.32	<.001 b
DMAA	Chevron	2.7	Lapidus	−3.2	.023 b
	Chevron	2.7	Prox	−3.3	<.001 b
	Chevron	2.7	Scarf	−2.8	<.001 b
	Lapidus	−3.2	Prox	−3.3	.95
	Lapidus	−3.2	Scarf	−2.8	1.0
	Prox	−3.3	Scarf	−2.8	.98
TDD	Chevron	−1.2	Lapidus	−4.9	.14
	Chevron	−1.2	Prox	−4.2	1.0
	Chevron	−1.2	Scarf	−2.3	.90
	Lapidus	−4.9	Prox	−4.2	.46
	Lapidus	−4.9	Scarf	−2.3	.027 b
	Prox	−4.2	Scarf	−2.3	<.001 b

a

Analysis of variance demonstrated that there were differences in the changes in hallux interphalangeus angle (HIA), distal metatarsal articular angle (DMAA), and total distal deformity (TDD) when different surgeries were performed. For simplicity, negative numbers denote worsening, to decrease the confusion caused by some angles being the supplement or complement of a measured angle. There was a significant difference between improvement in HIA by the Scarf osteotomy relative to chevron, Lapidus, and proximal metatarsal closing wedge osteotomy (Prox), which all worsened the HIA. Similarly, the chevron osteotomy improved the DMAA significantly relative to the worsening caused by Lapidus, Prox, and scarf. The TDD was worsened by all procedures; significantly less so by the scarf than by the Lapidus or Prox.

b

Significant difference by P value.

Discussion

Pronation of the first ray is often a part of HV deformity. Scranton and McDermott ¹² highlighted the importance of recognizing pronation and addressing it surgically by showing that pronation is associated with revision surgery. However, during the correction of HV, the rotationally neutral first ray betrays HVIP that was not previously evident. ⁷ This provides a possible explanation for the prior findings that the incidence of HVIP is lower in those with HV than those without HV. ⁶ That is, the preoperative AP weight-bearing radiograph has low sensitivity for the detection of HVIP in a foot with a pronated first ray.

Our study demonstrated that HVIP is underestimated in preoperative radiographs, which is in accordance with prior research. ⁷ Among those who did not have Akin osteotomies, postoperative weight-bearing AP radiographs demonstrated a worsening of a variety of HVIP measures. The worsening was severe enough to meet defining criteria for HVIP by PPAA. Given that our study excluded any patients who had obvious HVIP or residual deformity intraoperatively and therefore had an Akin osteotomy, our results are an underestimation of the amount of HVIP after correction. Correction of HVA was strongly correlated with worsening of the HIA and moderately correlated with worsening of the PPAA and TDD. The correction of the IMA was moderately correlated with worsening of the HIA, the TDD, and the PPAA. It is possible that larger angular correction was accompanied by a greater rotational correction, although this cannot be accurately measured by 2-dimensional imaging. This would not necessarily mean that the pronation is greater in the greater deformities, but it may mean that the HVIP that is present is more greatly elucidated with the larger HV corrections with larger rotational correction. A prospective study that accurately quantifies rotational correction pre- and postoperatively might help explain the reason behind the worsening appearance of the HVIP with greater corrections. This could be achieved via short axis cuts on a computed tomography scan.

We also showed that the prevalence of HVIP as defined by radiographs was significantly increased postoperatively via measurement of the HIA, DMAA, and PPAA. As mentioned previously, these were the cases that did not undergo an Akin osteotomy to address the HVIP. The clinical importance of the residual HVIP is not known, as clinical outcomes were not measured in this radiographic study. We recommend that surgeons be aware of the higher prevalence of HVIP in HV than can be appreciated preoperatively and be prepared to perform an Akin osteotomy.

When stratified by operative intervention, 3 of the measurements of HVIP were significantly different according to procedure. HIA was positively affected by the scarf osteotomy but not the Lapidus procedure, chevron osteotomy, or proximal closing wedge osteotomy. DMAA was corrected by the chevron osteotomy but worsened by Lapidus procedure, proximal metatarsal osteotomy, and scarf osteotomy. The TDD was worsened by all of the procedures, but less so by the scarf osteotomy. While the differences were statistically significant, the actual differences in angular measurements are small and are likely of minimal clinical value. In addition, there were only 2 Lapidus procedures in the study, and while some results were statistically significant, they may not bear out to be widely applicable.

Strengths of this study include the exhaustive analysis of the data in order to gain more reliable numbers. Much attention has been drawn to the reliability of angular measurements in HV, regarding both intra- and interobserver reliability.^3,8,9 Condon et al ² demonstrated a margin of error of ±3.60 degrees with 95% confidence in the measurement of IMA. ² Resch et al ⁹ evaluated the intraobserver accuracy of measurements of HV with 95% confidence, finding a confidence limit of 4.2 degrees for HVA and 3.1 degrees for IMA; interobserver error was 6.4 degrees for HVA and 5.4 degrees for IMA. ⁹ An earlier study by Saltzman et al ¹¹ evaluated the reliability of measurements on plain radiographs, rather than the digital radiographs that are the current norm. The investigators found that the 95% limit for HVA was 6 degrees and IMA was 4 degrees. While Condon et al ² demonstrated notable improvement in the reliability of measurement of the IMA with the average of only 2 measurements, whether by the same observer or different observers, we averaged 3 measurements by each of 2 separate observers for a total of 6 measurements in the final average of each pre- and postoperative measurement. In addition, the operating surgeons were not involved in the measurements so as to minimize performance bias.

A limitation of this observational, radiographic study is lack of clinical applicability, as we did not investigate clinical outcomes of these patients. Whether the residual HVIP deformity led to revision surgery or future Akins was not investigated. As a tertiary referral center, we often are referred patients with radiographs from outside facilities, and these are often not uploaded to our electronic system, making it impossible to compare pre- and postoperative weight-bearing radiographs, leading to a high number of subjects who were excluded from the study. A true measure of first ray pronation would facilitate this study, but cost and radiation required by advanced imaging would be impractical for most cases of HV. Providing a means of estimating preoperative HVIP via plain films is out of the scope of this paper, but ultimately, this would be the goal of future research in this area. Knowing the true HVIP preoperatively would allow a surgeon to plan for an Akin osteotomy if necessary. This paper is only able to recommend vigilance to surgeons and to be aware of the possibility of residual deformity if the HVIP is not addressed.

The use of the DMAA to estimate pre- and postoperative HVIP was included for completeness, but reliance upon this measurement is inherently flawed. The DMAA has been shown to be less reliable than other measurements of HVIP, ³ and we also found less consistent results with this measurement. Our own study visually demonstrated (Figure 6F) that the medial eminence resection can have an effect on the DMAA due to removal of a portion of the articular surface. The quantification and characterization of the effect are outside the scope of this study and further support the idea that DMAA is a less reliable measurement of HVIP postoperatively.

OPEN IN VIEWER

Figure 6.

Postoperative development of hallux valgus interphalangeus (HVIP). This example of pre- and postoperative radiographs in a single patient demonstrates the radiographic development of HVIP in postoperative follow-up. (A) shows a preoperative weight-bearing AP foot and (B) shows the postoperative foot; the following tiles are magnified versions of the same radiographs demonstrating the angular measurements performed. (C) is magnified to demonstrate hallux interphalangeus angle (HIA) of 2 degrees, which worsens to 12 degrees postoperatively, as seen in (D), which is now consistent with HVIP. The distal metatarsal articular angle (DMAA) shown in (E) is 9 degrees preoperatively and worsens to 20 degrees postoperatively in (F) and now meets criteria for HVIP. The proximal phalangeal articular angle (PPAA) in (G) is 4 degrees preoperatively and worsens to 10 degrees postoperatively in (H), which now is in HVIP.

Conclusions

HVIP was underestimated in preoperative radiographs, and postoperative weight-bearing radiographs demonstrated worsening of standard measurements of HVIP. The average worsening was strong enough to meet criteria for HVIP by PPAA in our study. Correction of HVA was strongly correlated with worsening of the HIA and moderately correlated with worsening of the PPAA and the TDD. The correction of the IMA was moderately correlated with worsening of the HIA, the TDD, and the PPAA. The prevalence of HVIP as defined by radiographs was notably increased postoperatively via measurement of HIA, DMAA, and PPAA. Surgeons should be aware of the higher prevalence of HVIP in bunions than can be appreciated preoperatively and be prepared to perform an Akin osteotomy.