Study Selection

Searches of PubMed, MEDLINE, Embase, and the Cochrane Database of Systematic Reviews were performed for articles published from January 1, 1990, to March 22, 2014, using the search terms Charcot, diabetes, arthropathy, neuroarthropathy, internal fixation, external fixation, or Ilizarov. The reference lists of the retrieved articles were reviewed manually for any additional studies that met inclusion criteria.

Eligibility

To meet the criteria for inclusion, a study must have reported on: (1) patients with Charcot arthropathy of the midfoot secondary to diabetes mellitus who underwent Lisfranc or Chopart joint arthrodesis, (2) surgical treatment with either internal or external fixation, and (3) outcomes of interest. Articles were excluded from review when they met any of the following criteria: (1) studies in which surgery was performed for other causes of midfoot deformity, such as posttraumatic arthritis, (2) studies using a surgical technique that combined the use of both internal and external fixation, and (3) studies that reported on Charcot arthropathy and surgery at the ankle joint, where midfoot cases could not be extracted and analyzed separately.

Involvement of the hindfoot was neither an inclusion nor exclusion criterion (ie, as long as the midfoot was involved, patients also could have undergone surgery at the subtalar joint, calcaneocuboid joint, and the talonavicular joint, but not the ankle joint). All study designs were considered, and there were no restrictions based on language or country in which the study was performed. Duplicate data on the same patient population were excluded to avoid counting patients more than one time.

Quality Assessment of the Literature

The quality of each study selected for inclusion was evaluated using MacDermid's Structured Effectiveness Quality Evaluation Scale (SEQES) and by the level of evidence criteria. The SEQES contains 24 items each scored from 0 to 2 based on adherence to specific criteria.⁷ Two independent reviewers (D.J.L., J.S.) scored each study using the specific descriptors provided in MacDermid's original paper to limit subjectivity and variance between reviewers. Consensus was reached using MacDermid's adjucation process. Each study was assigned a cumulative score (out of 48) to rank the study as low (scores 1–16), moderate (scores 17–32), or high quality (scores 33–48).⁷

Data Extraction

The literature search was conducted independently by 2 reviewers (D.J.L., J.S.). Abstracts of eligible studies were reviewed, and relevant manuscripts were read in full. Each reviewer (D.J.L., J.S.) independently extracted the data and entered them into a standardized data-extraction form. Any differences were resolved by discussion until consensus was reached. In studies that included some patients without midfoot involvement but where these patients could be excluded on an individual basis, data were extracted for only those cases that involved the mid-foot, with or without hindfoot involvement. Cases involving the ankle joint were excluded.

Outcome Measures

Outcomes of interest included: return to functional ambulation, postoperative ulcer occurrence, radiographic nonunion, extremity amputation, unplanned further surgery, deep infection, wound healing problems, peri- or intraoperative fractures, and total cases with any complication. Significant heterogeneity in the reporting of outcomes across the studies required the use of standardized definitions for each of the parameters examined.

Return to Functional Ambulation. This was defined as the ability to ambulate after surgery, with or without assistive devices and with or without special diabetic footwear. This excluded amputees or those who died prior to returning to functional ambulation, but included those who returned to functional ambulation then later died for unrelated causes.

Ulcer Occurrence. This was defined as the presence of any ulcer located in either the plantar midfoot or at a site other than the midfoot at any point during the follow-up period.

Radiographic Nonunion. This was defined as occurrence of any failure of solid bony union at any fusion or osteotomy site. A fibrous union was included as a nonunion, even if it was functionally stable.

Extremity Amputation. This was defined as any major amputation of the operative extremity, excluding toe amputations.

Unplanned Further Surgery. This was defined as the occurrence of any unplanned further surgery for any indication, not including amputation. These included incision and drainage for infection, management of wound complications (eg, skin grafting or delayed closure), revision of fixation, hardware removal including in-office procedures, and management of postoperative fractures.

Deep Infection. This was defined as the presence of any postoperative infection (eg, osteomyelitis), not including superficial incision site infection or pin-tract infection.

Wound Healing Problems. This was defined as any wound dehiscence, skin sloughing, or other wound healing issues.

Peri- or Intraoperative Fractures. This was defined as any fracture occurring during the procedure or during the immediate postoperative period.

Total Cases With Any Complication. This was defined as the total number of cases with any complication.

Statistical Analysis

A mixed-effects logistic regression analysis was used to account for the significant heterogeneity across the different studies. This model evaluated each of the outcomes described above and its association with the type of fixation method (internal vs external). The response was defined as the presence or absence of a specific outcome or individual complication event and entered into the model as the number of events (report of number with a given outcome) divided by the number of study participants. This formed the basis for the unweighted analyses. Logistic regression was used to estimate the odds ratio (OR) with 95% confidence intervals (CIs) for individual parameters.

In addition, each study was weighted by its sample size to investigate possible changes in the OR, under the assumption that those studies with a larger number of observations reflected greater experience among practitioners with a procedure and might provide a more appropriate estimate of the outcomes of interest. In fact, the weighted OR, rather than the unweighted incidence rates, likely provide the most accurate description of the differences between the 2 interventions with respect to individual parameters. All analyses were performed using SAS version 9.3 software (SAS Institute Inc, Cary, North Carolina) on a Windows 7 (Microsoft, Redmond, Washington) platform.

Studies and Baseline Patient Characteristics

The initial search yielded 184 references. A review of the bibliographies of the obtained articles produced 12 additional studies. After stringent inclusion and exclusion criteria were applied, 11 studies were determined to meet the criteria (Figure). The characteristics of the included studies are summarized in Table 1. All of the articles were published between 1996 and 2014. The majority of studies were conducted at single centers in the United States and Europe. All 11 studies represented Level IV evidence (case series or case reports), with 10 studies using a retrospective design and 1 study using a prospective design.

Figure: Flowchart showing literature search strategy and results.

Table 1: Case Series or Reports Included in Systematic Review of Internal Versus External Fixation

None of the studies directly compared internal and external fixation techniques. Of the 11 studies, 8 (73%) reported on internal fixation and 3 (27%) reported on external fixation. A total of 88 feet in 85 patients underwent internal fixation and 38 feet in 38 patients underwent external fixation for realignment of midfoot Charcot deformity. The studies on internal fixation used interfragmentary screws (n=1), plate and screws or screws alone (n=2), intramedullary cannulated screws (n=2), and the solid intramedullary fusion bolt (n=3). The studies on external fixation used monoplanar external fixators (n=2) and ring external fixators (n=1). Mean length of follow-up was 37.4 and 23.1 months for internal and external fixation, respectively.

Specific Outcome Measures

Results of statistical analysis and specific outcome measures are reported in Table 2, which includes the unweighted incidence rates in each treatment group, as well as the unweighted and weighted OR, 95% CI, and P values. Although P values are reported for the weighted analyses, it should be understood that significance is an artifact of the weighting scheme and serves to highlight differences between the techniques.

Table 2: Outcomes of Internal Versus External Fixation

Return to Functional Ambulation

Ten studies reported the return to functional ambulation (7 internal fixation studies and 3 external fixation studies).^4,8–16 The rate of return to functional ambulation was 93.8% in the internal fixation group (range, 82.4%–100%) and 92.1% in the external fixation group (range, 90.9%–100%). The odds of returning to functional ambulation were 25% higher for the internal fixation group, although this difference was not statistically significant in the weighted analysis (OR, 1.259; P=.9766).

Ulcer Occurrence

All 11 of the studies reported the total number of cases with any ulcer occurrence.^4,6,8–16 The rate of ulcer occurrence was 8% (range, 0–18.2%) in the internal fixation group and 15.8% (range, 0.0–18.2%) in the external fixation group. The internal fixation group had a 42% reduced rate for ulcer occurrence, which was significant in the weighted analysis (OR, 0.578; P<.0001).

Radiographic Nonunion

Eight studies reported the radiographic nonunion rate (6 internal fixation studies and 2 external fixation studies).^{4,8–10,12–15} The rate of radiographic nonunion was 15.1% (range, 0–27.3%) in the internal fixation group and 58.3% (range, 0–63.6%) in the external fixation group. Taking the inverse of the OR (reversing the group comparison), external fixation was approximately 8 times more likely to result in nonunion than internal fixation (OR, 8.197 for the weighted analysis; OR, 7.874 for the unweighted analysis); this difference was significant in the un-weighted analysis (P=.002).

Extremity Amputation

All 11 of the studies reported the rate of major extremity amputation.^4,6,8–16 The rate of extremity amputation was 5.6% (range, 0–28.6%) in the internal fixation group and 2.6% (range, 0–3.8%) in the external fixation group. The OR suggested an almost 50% greater likelihood of extremity amputation for the internal fixation group, although this difference was not statistically significant in the weighted analysis (OR, 1.488; P=.0972).

Unplanned Further Surgery

All 11 of the studies reported the rate of unplanned further surgery.^4,6,8–16 The rate of unplanned further surgery was 21.6% (range, 0–45.5%) in the internal fixation group and 15.8% (range, 0–23.1%) in the external fixation group. The OR suggested a 20% increase for the internal fixation group with respect to unplanned further surgery, although this difference did not reach statistical significance in the weighted analysis (OR, 1.221; P=.0715).

Deep Infection

Nine studies reported the rate of deep infection (7 internal fixation studies and 2 external fixation studies).^{4,6,8,9,11–14,16} The rate of deep infection was 9.4% (range, 0–42.9%) in the internal fixation group and 3.7% (range, 0–3.8%) in the external fixation group. Those in the internal fixation group were twice as likely to report deep infection as those in the external fixation group; this difference was significant in the weighted analysis (OR, 2.068; P=.0013).

Wound Healing Problems

Six studies reported the rate of wound healing problems (4 internal fixation studies and 2 external fixation studies).^{6,9,11,13,14,16} The rate of wound healing problems was 12.2% (range, 0–28.6%) in the internal fixation group and 3.7% (range, 0–3.8%) in the external fixation group. The OR showed a 3.4-fold increase for the internal fixation group, which was significant in the weighted analysis (OR, 3.405; P<.0001).

Peri- or Intraoperative Fractures

Five studies reported the rate of peri- or intraoperative fractures (3 internal fixation studies and 2 external fixation studies).^6,9,11,12,14 The rate of peri- or intraoperative fractures was 10.9% (range, 5.9%–14.3%) in the internal fixation group and 7.4% (range, 0–7.7%) in the external fixation group. There was a 50% increased odds of fractures for the internal fixation group, which was significant in the weighted analysis (OR, 1.478; P=.0324).

Total Number of Cases With Any Complication

All 11 of the studies reported the total number of cases with any complication.^4,6,8–16 The rate of total cases with any complication was 44.3% (range, 0–63.6%) in the internal fixation group and 42.1% (range, 0–72.7%) in the external fixation group. The internal fixation group reported a 50% increased odds of complications, which was significant in the weighted analysis (OR, 1.525; P<.0001).