Impact of Podiatry Resident Experience Level in Hallux Valgus Surgery on Postoperative Outcomes

Abstract

Background

Despite modern advancements in transosseous fixation and operative technique, hallux valgus (i.e., bunion) surgery is still associated with a higher than usual amount of patient dissatisfaction, and is generally recognized as a complex and nuanced procedure requiring precise osseous and capsulotendon balancing. It stands to reason then that familiarity and skill level of trainee surgeons might impact surgical outcomes in this surgery. The aim of this study was to determine whether podiatry resident experience level influences mid-term outcomes in hallux valgus surgery.

Methods

Consecutive adults who underwent isolated hallux valgus surgery via distal metatarsal osteotomy at a single US metropolitan teaching hospital from January 2004 to January 2009 were contacted and asked to complete a validated outcome measure of foot health (Manchester-Oxford Foot Questionnaire) regarding their operated foot. Resident experience level was quantified using the surgical logs for the primary resident of record at the time of each case. Associations were assessed using simple, multiple and logistic regression analyses.

Results

A total of 102 adult patients (n=102 feet) agreed to participate with a mean age of 46.8 (SD 13.1 years, range 18-71) and average length of follow-up 6.2 years (SD 1.4, range 3.6-8.6). Level of trainee experience was not associated with postoperative outcomes in either the univariate (odds ratio 0.99 [95% CI 0.98-1.01], p = 0.827) or multivariate analyses (odds ratio 1.00 [95% CI 0.97-1.02], p = 0.907).

Conclusions

We conclude that podiatry resident level of experience in hallux valgus surgery does not contribute appreciably to postoperative clinical outcomes.

Introduction

It is estimated that one third of the world’s population will develop hallux valgus (i.e., bunion) deformity at some point in their lifetime (1,2). Hallux valgus surgery (HVS) is therefore commonplace in podiatric and foot and ankle orthopedic circles around the globe. Despite modern advancements in transosseous fixation and operative technique, HVS is still somewhat tainted with higher than expected rates of suboptimal outcomes (3). In fact, a recent Cochran review by Ferrari et al. (2004) estimated that as many as 25 to 33 percent of patients remain dissatisfied following HVS (4).

The difficulties surrounding HVS may be related to the inherent challenges of trying to re-establish the complex triplanar and reciprocal motion required at the 1^st metatarsophalangeal joint. Proper intra-operative tensioning and balancing of the intrinsic muscles around the great toe represent additional challenges to maintaining long-term correction of the deformity (5). For these reasons, HVS is generally considered a highly nuanced procedure requiring precise osseous and capsulo-tendon manipulation that is only fully appreciated after years of clinical and operative experience (3).

Resident participation in operative procedures is essential to surgical training (6). Although the exact proportion of surgeries performed with a resident or trainee surgeon is uncertain, it is likely to be significant. In the Veterans Affairs (VA) system, teaching hospitals perform approximately 80% of the surgical workload and identify a resident as the primary surgeon over 90% of the time (7). While several studies have previously examined general surgery resident participation and outcomes, each with varying but generally acceptable results (6-11), there are no studies yet that have addressed the interaction between podiatry residents and surgical outcomes. The purpose of this study then was to identify whether podiatry resident experience level influences intermediate and long-term postoperative outcomes using a common, yet technically challenging, foot surgery. The study’s null hypothesis was that resident experience level in hallux valgus surgery is not associated with postoperative outcomes.

Patients and Methods

Study Population and Setting

Consecutive adult patients (n=295) who underwent isolated HVS by the podiatry service at Advocate Illinois Masonic Medical Center (AIMMC) from January 2004 to January 2009 were contacted by mail and asked to participate. AIMMC is a level 1 trauma center and tertiary care hospital located in metropolitan Chicago, IL. It serves as the primary teaching hospital for an established podiatry residency program in which residents complete their 2^nd and 3^rd years of post-graduate surgical training.

Eligible patients were ambulatory men and women who underwent HVS via first metatarsal osteotomy for predominantly medial foot “bump” pain and at least 36 months follow-up. Patients with a chief complaint of joint pain (e.g., hallux rigidus) or those with prior first ray surgery were excluded. Patients requiring osseous surgery other than hammertoe repair (e.g., Weil osteotomy, tailor’s bunionectomy, extra-osseus talo-tarsal stabilization) at the time of their bunion surgery or those with seronegative (e.g., ankalosing spondylitis) or seropositive (e.g., RA) inflammatory arthridities were also excluded.

In instances when patients underwent bilateral foot surgery during the same operative setting (n=13), the study foot was determined by a coin flip. For those who underwent bilateral HVS with first metatarsal osteotomy at different times (n=3), the first operated foot only was used in the analysis. Local ethics committee approval was obtained (Advocate Health Care Institutional Review Board) prior to the start of the study and all subjects provided written consent.

Primary Outcome Measure—The MOXFQ

In response to the recent Cochrane review calling for an increased need to evaluate patient-centered outcomes when assessing the success of HVS (4), we selected the Manchester-Oxford Foot Questionnaire (MOXFQ) as the study’s primary outcome measure. The MOXFQ is a brief 16 question patient-focused survey designed specifically to assess pain and function following foot surgery with three domains (foot pain, walking/standing, and social interaction). The total score and each domain are scored on a scale ranging from 0-100, with larger numbers indicating greater levels of disability or pain. The MOXFQ was validated in the context of hallux valgus surgery in 2006 by Dawson et al. (12), demonstrates high correlation with the Short Form 36 (SF-36) and AOFAS hallux metarsophalangeal-interphalangeal clinical scale (13), and has greater readability and patient comprehension than the Foot Health Status Questionnaire (14).

Eligible patients, identified through chart review, received a survey via standard mail with instructions to complete and return the form. Participants were then stratified in the following way: Patient’s reporting a score greater than 1 SD above previously published averages (12) for any domain on the MOXFQ, and those requiring revision HVS (n=4), were classified as having had a ‘suboptimal’ postoperative outcome. This equated to a walking/standing score greater than 39.4, pain greater than 30.7 and social interaction greater than 30.5. Similarly, patients returning surveys without satisfying either of these two conditions were classified as having had a ‘favorable’ postoperative outcome.

Primary Predictor Variable

Resident experience level with HVS served as the primary predictor variable in this study. The following observations were recorded from the surgical logs for the principle resident surgeon on record for each procedure: 1) post graduate year of training, 2) number of ‘C level’ HVS cases requiring first metatarsal osteotomy, and 3) number of ‘C level’ + ‘B level’ HVS cases requiring first metatarsal osteotomy. During the time frame covered with this study, the Council of Podiatric Medical Education defined a C level case as “the resident participates actively in performing the procedure.” This level of participation could be claimed on surgical logs when the resident performed at least the majority (greater than fifty percent) of the procedure under direct supervision of the attending (17). B level cases, in contrast, could be claimed by the resident when their participation in the case was less than 50% and typically involved greater amounts of observation.

Other Predictor Variables (Independent Variables)

Several potentially important confounders were also examined in the analysis including patient age at the time of surgery, gender, highest level of education attained, length of follow-up, bunionectomy type (e.g., scarf, modified Austin, modified Kalish), whether bilateral HVS was performed, board certification status of the attending at the time of surgery, length of surgical residency training of the attending, and whether the case was performed by a “core” attending faculty member of the residency training program. For the purposes of this paper core faculty were defined as any surgeon who receives a stipend from the teaching program. Individual attending surgeons were also considered as independent variables in the analysis if they served as the surgeon on record in 5 or more cases. Likewise, individual resident physicians were also considered as independent variables if they were involved in 5 or more included cases. Postoperative complications such as infection, return to the operating room, recurrence, and resultant hallux varus were also recorded from outpatient and hospital records.

Data Analysis

Missing data (n=10 subjects, 11 items total) on the MOXFQ was imputed with the patient’s mean value for completed items on the survey. Patient’s answering “n/a” were given a score of “0” for that item. Relationships between continuous variables and MOXFQ scores were graphically explored using scatter plots. Assessment of a linear correlation was determined using Pearson correlation coefficient (r). Dichotomous variables were tested for their association with surgical outcomes using an independent t-test, Wilcoxon rank-sum test, or simple regression.

Multivariable logistic regression was used to predict suboptimal outcomes following HVS based on the patient and provider predictor variables in the study. Significant predictor variables were determined by first conducting a univariate analysis for each predictor. Potentially important independent variables (with p<0.25) in the univariate analysis were then entered into a multivariable, logistic regression model. Statistical interaction (p-values < 0.05) was tested by looking at all pairwise interactions among the predictors using forward selection. Potential confounders were identified using stepwise backward elimination working back from a full model (without the interaction terms). The final regression model consisted of the clinically important predictors, significant confounders and significant interaction terms. The Hosmer-Lemeshow goodness-of-fit test was used to assess how effectively the final model described the outcome of interest. Data was analyzed using SAS software (version 9.2 of the SAS System for Windows; SAS Institute, Cary, North Carolina). All tests were two-tailed with p values less than 0.05 considered statistically significant.

Results

A total of 102 adult patients (45 modified Austin buniononectomies, 10 Kalish-type procedures and 47 scarf bunionectomies) returned usable surveys (102/295, 35% response rate). The mean age of the cohort was 46.8 (SD 13.1 years, range 18-71) and most participants were women (91/102, 89%). The average length of follow-up was 6.2 years (SD 1.4, range 3.6-8.6). There were eighteen attending surgeons and 18 different resident physicians involved in patient care. Three attendings performed over half of the surgeries (59/102, 58%), and the majority of trainees (10/18, 55%) claimed to be the principle resident surgeon on at least 5 of the included cases. The mean postoperative MOXFQ score for our cohort was 20.4 (SD 25.1, range 0-95.3). The mean postoperative MOXFQ scores in each of the domains were: walking/standing 19.6 (SD 27.1), pain 23.5 (SD 26.8), and social interaction 17.7 (SD 25.1). There were three minor complications each attributed to screw removal for painful hardware (3/102, 3%). Four patients required surgical revision (two for hallux varus and two for joint stiffness) at an average 2.3 years. No patient was re-admitted for postoperative infection, DVT, or other major complication.

Patient and provider characteristics stratified by postoperative outcome are provided in Table 1. There were 64 favorable postoperative outcomes and 38 suboptimal outcomes in our cohort. The only variable in the univariate analysis found to be associated with suboptimal HVS outcomes was length of follow-up (p = 0.006), with longer lengths of follow up found in the suboptimal group (mean 5.9 [1.4 yrs] vs. 6.7 [1.4 yrs]). Level of resident experience with HVS showed no associated with postoperative outcome when using both resident C level cases only (mean 41.4 [26] vs. 40.4 [21], p = 0.838) and resident C+B level cases (mean 54.4 [32] vs. 52.8 [24], p =0.7903). This is clearly demonstrated in the scatter plot looking at resident number of prior HVS cases (C level) and postoperative MOXFQ total scores for the cohort (Figure 1, r = 0.002, p = 0.978).

Table 1.

Patient and provider characteristics stratified by postoperative Hallux Valgus Surgery (HVS) outcome

	Favorable Outcome (n=64)	Suboptimal Outcome (n=38)	p-value
Patient Characteristics
Age	45.6 (13.4)	48.9 (12.3)	0.205
Female gender	59/64 (0.92)	32/28 (0.84)	0.212
Follow-up in yrs	5.9 (1.4)	6.7 (1.4)	0.006*
Bilateral foot surgery	9/64 (0.14)	4/38 (0.11)	0.606
Study foot = right	33/64 (0.52)	22/38 (0.58)	0.537
Education (highest level)
High school	15/64 (0.23)	8/38 (0.21)	0.782
Some college	22/64 (0.34)	12/38 (0.32)	0.773
College graduate	14/64 (0.22)	11/38 (0.29)	0.424
Masters or higher	13/64 (0.20)	7/38 (0.18)	0.817
Surgical procedure
Modified Austin	26/64 (0.41)	19/38 (0.50)	0.359
Modified Kalish	6/64 (0.09)	4/38 (0.11)	0.851
Scarf buninonectomy	32/64 (0.50)	15/38 (0.39)	0.305
Provider Characteristics
Board certified attending	51/64 (0.80)	27/38 (0.71)	0.323
Attending yrs of surgical training	2.03 (0.77)	1.73 (0.95)	0.126
Core attending in training prog.	47/64 (0.73)	23/38 (0.60)	0.176
Resident post graduate year	2.47 (0.50)	2.45 (0.50)	0.838
Resident number of HVS cases¥	41.4 (25.6)	40.4 (20.6)	0.830

All values are displayed as frequency (percentage of column total) or as mean (sd) t-test or Wilcoxon’s rank-sum test was used for comparisons of continuous variables χ² or Fisher’s exact test was used for comparisons of categorical variables

^*statistically significant test result

^¥refers to the total number of ‘C level’ HVS cases performed prior to surgery

Figure 1.

Resident number of ‘C Level’ hallux valgus surgical (HVS) cases with 1st metatarsal distal osteotomy versus Manchester-Oxford Foot Questionnaire (MOXFQ) Total Scores at mid and long-term follow-up (avg. 6.2 yrs).

Potential important predictors in the univariate analysis (p < 0.25) that were entered into the multivariable analysis along with level of resident experience (C level cases) were: length of follow-up (in yrs), age, gender, attending years of surgical training, and whether the attending was “core” faculty. Additionally, one attending (‘attending X’, univariate p = 0.075) and one resident (‘resident Y’, univariate p = 0.095) were eligible for entry into the final multivariable analysis.

Table 2 shows the final multivariate model for predicting suboptimal outcomes following HVS. There were no statistically significant interaction terms added to the model. As expected, age and follow-up were retained in the final model, but interestingly ‘attending X’ and ‘resident y’ were also identified as significant independent predictors of suboptimal outcomes. The Hosmer-Lemeshow goodness of fit test was not significant (0.768, p = 0.479), indicating that the model is well fit throughout the spectrum of predicted risk. Finally, the final model was able to discriminate between suboptimal and favorable HVS outcomes more than three quarters of the time (area under the ROC curve [AUC] = 0.768).

Table 2.

Final multivariable model for predicting suboptimal outcomes following HVS

Risk Factor	Regression Coefficient	Odds Ratio	95% CI	p value
Intercept	−6.675	–	–	–
Resident number of HVS Cases	−0.002	1.00	0.97-1.02	0.907
Age	0.049	1.05	1.01-1.10	0.016
Follow-up (yrs)	0.536	1.71	1.19-2.45	0.004
Attending X	2.140	8.50	1.67-43.30	0.010
Resident Y	1.742	5.71	1.28-25.51	0.023

HVS, hallux valgus surgery

The area under the ROC curve (AUC) = 0.768. Hosmer-Lemeshow goodness of fit test = 7.546, p = 0.479.

Discussion

While previous work has examined how general surgery residents impact surgical outcomes (6-11), this is the first study to critically evaluate podiatry residents and postoperative outcomes. We found that podiatry resident level of experience in HVS was not associated with postoperative outcomes using a relatively common, yet technically challenging, foot surgery. Additionally, operating with podiatry residents appears to be safe, as our minor (3/102, 3%) and major (0/102, 0%) complication rates, and rate of revision (4/102, 4%) did not differ notably from previously published rates (4).

That being said, the mean MOXFQ scores in our series were slightly higher (i.e., worse) than those previously reported in the literature. Dawson et al. (12), for example, reported postoperative means of 16, 20, and 12 for the walking/standing, pain, and social interaction domains, respectively. Similarly, Maher and Kilmartin (16) reported mean scores of 12, 15, and 19.1 for the same MOXFQ domains using a cohort of 71 postoperative HVS patients. It is possible that the slightly higher MOXFQ scores encountered in our series may be, at least, partly attributed to resident participation. However, we believe this discrepancy is more likely due to the larger number of attending surgeons which we included in our study and their varying levels of operative training and expertise. When considering only the core attendings involved in the training program—those who also typically attained higher levels of board certification and completed longer programs of post-graduate residency training—the MOXFQ scores, although not statistically significant, more closely resembled those from prior studies (Figure 2). This observation may also suggest that regular and repetitive interaction between trainees and mentors might facilitate improved outcomes in HVS. Unfortunately, our study was underpowered to confirm this type of interaction effect with statistical significance.

Figure 2.

Mean Manchester-Oxford Foot Questionnaire (MOXFQ) scores across the three domains for “core” (n=70 procedures) and other (n=32 procedures) faculty in the residency training program.

Although resident experience did not correlate with postoperative results in this study, certain individuals themselves (e.g., attending X and resident Y) were associated with poorer HVS outcomes. Until now, quality assurance exercises at our hospital have been limited to ad hoc review of postoperative X-rays for surgeries performed by members of the foot and ankle service. Our study however raises the possibility of also using patient-reported outcomes to identify resident and/or attending surgeons in need of independent review of their techniques in order to improve their patient outcomes to within expected levels.

Like many retrospective cohort studies, our results may suffer from residual confounding, inability to account for the baseline severity of the hallux valgus deformity, and selection/respondant bias. Also, since all procedures in this study involved a post-graduate trainee, we were unable to assess the impact which resident involvement (versus no resident involvement) had on postoperative outcomes. Further large scale studies examining both operating conditions would be needed to better address this. Finally, our study was conducted within a large, tertiary care teaching hospital with several dedicated, core faculty members responsible for residency education, so is not clear just how generalizable our findings will be to smaller podiatry residency training programs or larger programs that may have less oversight and/or individual instruction.

In summary, we found that neither podiatry resident seniority (PGY) or surgical experience level influenced mid and long-term postoperative outcomes in HVS. These findings may provide guidance for attending and resident physicians during office consultations and better equip patients with appropriate postoperative expectations when contemplating HVS in teaching environments.