A systematic review of ankle fracture treatment modalities in diabetic patients

Abstract

Aim

This systematic review evaluated the surgical outcomes of various ankle fracture treatment modalities in patients with Diabetes Mellitus as well as the methodological quality of the studies.

Methods

For our review, four online databases were searched: PubMed, MEDLINE (Clarivate Analytics), CINAHL (Cumulative Index to Nursing and Allied Health) and Web of Science (Clarivate Analytics). The overall methodological quality of the studies was assessed with the Coleman Methodology Score. Data regarding diabetic ankle fractures were pooled into three outcomes groups for comparison: (1) the standard fixation cohort with management of diabetic ankle fractures using ORIF with small or mini fragment internal fixation techniques following AO principles, (2) the minimally invasive cohort with diabetic ankle fracture management utilizing percutaneous cannulated screws or intramedullary fixation, and (3) the combined construct cohort treated with a combination of ORIF and another construct (transarticular or external fixation).

Results

The search strategy identified 2228 potential studies from the four databases and 11 were included in the final review. Compared to the standard fixation cohort, the minimally invasive cohort had increased risk of hardware breakage or migration and the combined constructs cohort had increased risk of hardware breakage or migration, surgical site infection and nonunion. Limb salvage rates were similar for the standard fixation and minimally invasive cohorts; however, the combined constructs cohort had a significantly lower limb salvage rate compared to that of the standard fixation cohort. The mean Coleman Methodology Score indicated the quality of the studies in the review was poor and consistent with its limitations.

Discussion

The overall quality of published studies on operative treatment of diabetic ankle fractures is low. Treating diabetic ankle fractures operatively results in a high number of complications regardless of fixation method. However, limb salvage rates remain high overall at 97.9% at a mean follow-up of 21.7 months. To achieve improved limb salvage rates and decrease complications, it is critical is to follow basic AO principles, respect the soft tissue envelope or utilize minimally invasive techniques, and be wary that certain combined constructs may be associated with higher complication rates.

Level of evidence

2.

1. Introduction

Ankle fractures are one of the most common orthopaedic injuries, accounting for 9% of fractures. As many fractures are treated operatively and the incidence of these fractures rises, the financial burden on the healthcare system also increases.¹ Diabetes is well documented as a significant risk factor for loss of reduction, impaired wound healing, delayed fracture healing, and Charcot neuroarthropathy.^{2, 3, 4} In 2018, the Centers of Disease Control reported that 26.9 million people (8.2% of the U.S. population) had diabetes and the incidence is rising every year. This will further increase the prevalence of diabetic ankle fractures.^5,6

Over the past couple of decades, the risks and complications of diabetic ankle fractures have been recognized. Most of these patients also have additional comorbidities including peripheral arterial disease and diabetic neuropathy. Therefore, on one hand surgical intervention can pose challenges for wound healing, but non-operative management increases the risk for loss of reduction and therefore stresses the condition of the surrounding soft tissue. Both non-operative and operative management can increase the risk for Charcot neuroarthropathy.^7,8 Diabetic neuropathy affects patients’ balance and they cannot fully appreciate the detriment of early weight bearing, often leading to loss of reduction and malunion. This can further lead to Charcot neuroarthropathy, infections, or ulcerations, which are challenging conditions to treat. Complications arising from diabetic ankle fractures are not only associated with a significant economic burden to the healthcare system,^9,10 but also decrease patient quality of life impacting their ability to work or their well-being.¹¹ Therefore, as our understanding of the pathophysiology improves, techniques to treat these fractures have been evolving.

Operative diabetic ankle fracture treatment has centered around standard open reduction internal fixation (ORIF) with small or mini fragment (locking or nonlocking) plates following principles espoused out by Arbeitsgemeinschaft für Osteosynthesefragen (AO). However, with the increased awareness of the tenuous soft tissue envelope and the lack of adequate biology at times, recent authors have advocated for alternative fixation methods to help increase rates of limb salvage. Minimally invasive approaches including percutaneous cannulated screws^12,13 or Minimally Invasive Plate Osteosynthesis (MIPO)¹⁴ have been described as techniques to avoid large surgical incisions and maintain periosteal blood supply. Likewise, the use of intramedullary fixation including locked fibular nailing or tibiotalocalcaneal (TTC) nailing also utilizes smaller incisions and can preserve the soft tissue envelope.^{15, 16, 17, 18} Another form of alternative fixation involves a combination of the constructs (i.e. ORIF + external fixation, ORIF + transarticular fixation). These techniques have been extensively described as a means to help prevent further instability.^19,20

Many published studies discuss the various roles of surgical management and describe the techniques of fixation to prevent further instability. But few compare these various treatment modalities to each other and have long term follow-up. The purpose of this systematic review was to (1) evaluate the outcomes of standard ankle fracture treatment (ORIF utilizing AO principles) versus alternative fixation modalities (percutaneous cannulated screws, intramedullary fixation, combined constructs) in diabetic ankle fractures and (2) evaluate the methodological quality of the studies.

2. Materials and methods

We performed a systematic review to evaluate the outcomes of standard ankle fracture treatment versus alternative fixation modalities (percutaneous cannulated screws, intramedullary fixation, combined constructs) in diabetic ankle fractures. We defined the standard ankle fracture treatment to include open reduction internal fixation with standard AO principles. We used the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.²¹ To assist in our review, four online databases were searched: PubMed, MEDLINE (Clarivate Analytics), CINAHL (Cumulative Index to Nursing and Allied Health) and Web of Science (Clarivate Analytics). A specific time period in regards to publication date was not chosen; however, the search was done in June 2020. The key Boolean phrases used in the search engines were: Diabetes, Diabetic, Ankle Fractures. For this study, Institutional Review Board (IRB) approval was not required as this was a review of published studies.

2.1. Inclusion criteria

In order to perform a thorough evaluation of the studies for this systematic review, the following inclusion criteria were used:

• •Study of the operative management of acute diabetic ankle fractures.
• •Study must be full length and published in a peer-reviewed journal.
• •Study must have clinical and radiographic outcome assessment.
• •Study must have at least 6 months of follow-up and the duration of follow-up must be documented.
• •Study must be published in English.

2.2. Exclusion criteria

Studies were excluded in the systematic review for the following reasons:

• •Study was an expert opinion, technique article, or case report without clinical or radiographic follow-up.
• •Study was a review article including systematic reviews.
• •Study was an abstract of a conference proceeding and did not have a full-length published article.
• •Study describing fractures or neuropathy of other anatomic sites (midfoot, hindfoot, forefoot, knee, hip, etc).
• •Study did not document at least 6 months of follow-up.
• •Study focused on other high-risk comorbidities (e.g., obesity, PAD, renal disease) as well which could be a confounder.
• •Study was not published in English.

2.3. Study screening

Three of the reviewers helped identify studies by screening titles and abstracts of the manuscripts. If further clarification was needed, then the full text of the manuscript was reviewed. Additionally, references from these studies were assessed to supplement the review. Fig. 1 is the PRISMA Flow Diagram of included studies.

Fig. 1.

Prisma flow diagram.

Included studies for systematic review of ankle fractures in diabetic patients.

The complete full text of the manuscripts was distributed among three academic surgeons (one Board-Certified Orthopaedic Trauma Surgeon, one Board-Certified Orthopaedic Foot and Ankle Surgeon, and one Orthopaedic Foot and Ankle Surgery Fellow) who are very actively involved in treating diabetic ankle fractures. The predetermined variables of interest were populated into a data collection sheet as shown in Table 1.

Table 1.

Predetermined variables of interest.

1. Number of Ankle Fractures in the Study

2. Mean Age (years)

3. Gender (number of males)

4. Number of closed fractures

5. Fracture Classification (Unimalleolar vs Bimalleolar vs Trimalleolar)

6. Surgical Approach (Open versus Percutaneous)

7. Primary Fixation/Implant Type

8. Overall Hardware Complications – Hardware Breakage or Migration (Y/N)

9. Surgical Site Infection (Deep or Superficial)? (Y/N)

10. Charcot Neuroarthropathy? (Y/N)

11. Loss of Reduction? (Y/N)

12. Nonunion? (Y/N)

13. Unplanned Return to OR? (Y/N)

14. Post-operative Wound Healing Issues/Cast Ulcers? (Y/N)

15. Limb Salvage? (Y/N)

16. Mean Follow-up (Months)

17. Coleman Methodology Score

Two surgeons independently reviewed each article to record the variables of interest. Any disagreements with the recorded variables were resolved by consensus. We elected to pool the data into three outcomes groups for comparison (one control and two study groups). For the purpose of our analysis, our control group focused on the standard operative ankle fracture management of diabetic ankle fractures (ORIF), while the study group was subdivided into a “minimally invasive” cohort which had treatment with percutaneous cannulated screws and intramedullary fixation (fibular nails, TTC nails) and a “combined constructs” cohort which were treated with a combination of ORIF and another construct (transarticular or external fixation). Once the outcomes were collected, the data was reviewed by a biostatistician and five academic surgeons (three Board-Certified Orthopaedic Trauma Surgeons, one Board-Certified Orthopaedic Foot and Ankle Surgeon, and one Orthopaedic Foot and Ankle Surgery Fellow) to determine the overall trends.

Coleman Methodology Score (CMS), as depicted in Fig. 2, was used to assess the quality of the studies by each of the three reviewers. The CMS scores each study from 0 to 100 after assessing ten distinct criteria to evaluate study designs and outcomes. Scores between 85 and 100 are deemed excellent, 70–84 are good, 55–69 are fair and less than 55 are poor. As a score approaches 100, the study’s design tends to minimize chance, bias, and confounding.²²

Fig. 2.

Adapted coleman methodology score (CMS).

2.4. Statistical analysis

The weighted incidence rate for each of the nine outcomes of diabetic ankle fractures was calculated for the six standard fixation treatment studies and separately for the two alternative fixation treatment (four minimally invasive and two combined constructs) studies. The incidence rate was calculated as the number of positive cases of the event (outcome) divided by the total number of persons at risk for the event. The weighted odds ratio (OR), with penalized maximum likelihood estimation and Firth’s bias correction, along with the 95% confidence interval was also estimated. An estimated odds ratio >1 indicated greater odds of an outcome of diabetic ankle fractures for the alternative fixation treatment vs. standard fixation treatment. Wald chi-square from the penalized maximum likelihood estimates (categorical variables) and two-independent samples t-test with the Satterthwaite method for unequal variances (continuous variables) were used to identify any differences between the study cohorts (alternative fixation treatment vs. standard fixation treatment) on each of the outcomes.

The I² statistic and Cochran Q test were evaluated to assess observed heterogeneity for each outcome across the study cohorts. I² values range from 0% to 100%. A value of 0% indicates no observed heterogeneity and larger values show increasing heterogeneity; as a rule of thumb: Low (≤25%), Moderate (50%), and High (≥75%). Egger’s test for each outcome across all studies was used to evaluate potential publication bias.

Statistical analyses were carried out using SAS software, version 9.4 (SAS Institute, Inc., Cary, NC) and MedCalc for Windows, version 19.5.2 (MedCalc Software, Ostend, Belgium). The level of significance was set at α = 0.05 (two-tailed) and we implemented the False Discovery Rate (FDR) procedure, where applicable, to control false positives over the multiple tests.²³

3. Results

To begin our systematic review, we identified 2228 potential studies from the four online databases as shown in Fig. 1. The databases were reviewed and 630 studies were excluded by simply screening the titles and abstracts. From the remaining studies, further exclusions were done by removing technique articles, expert opinions, other reviews, and studies not focusing exclusively on ankle fractures or diabetic patients. From this, 63 abstracts were selected which were then further narrowed down to full text review of 27 studies to help identify which would evaluate the outcomes of standard ankle fracture treatment versus alternative fixation modalities (percutaneous cannulated screws, intramedullary fixation, combined constructs) in diabetic ankle fractures. Studies which were technique articles without outcomes, or which did not have adequate radiographic and clinical follow-up of at least six months were excluded. For this review, 11 studies were included as depicted in Table 2. Six studies focused on treatment of diabetic ankle fractures using ORIF with a standard AO technique using small fragment internal fixation,^{24, 25, 26, 27, 28, 29} three studies focused on treatment of diabetic ankle fractures using minimally invasive techniques and implants,^12,15,18 one study focused on treatment of diabetic ankle fractures using combined constructs,¹⁹ and one study highlighted the use of minimally invasive techniques as well as combined constructs.²⁰ One of the included studies was a case report by Facaros et al.¹⁹ which utilized a combined construct and included radiographic and clinical followup of greater than six months. The dates of publication of all included studies ranged from 2003 to 2020.

Table 2.

Included studies in the systematic review.

Study	# of ankles	Mean Age (years)	Gender (Males) (n)	Closed fractures (n)	Fracture Classification∗	Approach	Surgical Fixation Type	Hardware Breakage or Migration (n)	Infection (Deep or Superficial) (n)	Charcot Neuroarthropathy (n)	Loss of Reduction (n)	Nonunion (n)	Unplanned return to OR (n)	Wound Healing/Cast Ulcers (n)	Limb Salvage (n)	Mean Follow-up (mos)	CMS
Standard Fixation Techniques
Bazarov et al.24	48	64	15	48	19U 18B 11T	ORIF	SFIF	0	6	0	2	1	3	9	48	11.7	47
Costigan et al.25	84	49	51	75	+	ORIF	SFIF	0	10	5	0	3	9	3	82	49.2	58
Guo et al.26	36	56	22	36	6U 14B 16T	ORIF	SFIF	0	13	3	0	0	25	12	36	12	48
Jones et al.27	19	57	1	19	^	ORIF	SFIF	^	3	3	0	0	^	^	19	6	44
Lovy et al.28	8	62	1	8	++	ORIF	SFIF	^	0	0	0	0	0	1	8	7	45
Schmidt et al.29	131	56	59	105	35U 55B 41T	ORIF	SFIF	2	17	2	3	6	24	8	127	12	53
Alternative Fixation Techniques (Minimally Invasive)
Ashman et al.15	24	67	6	24	2U 10B 12T	ATS, Perc	FN	0	2	0	0	0	4	6	24	12	55
Ebaugh et al.18	27	66	20	21	+++	Perc	TTC	3	4	0	0	3	6	1	26	29.2	53
Emara et al.12	26	67	8	24	10U 11B 5T	Perc	PCS	0	0	1	2	0	1	0	26	34.2	47
Jani et al.20	3	69	^	3	1U 2B	Perc	RTF	1	0	0	0	0	0	0	3	15.9	51
Alternative Fixation Techniques (Combined Constructs)
Facaros et al.19	1	67	1	0	1U	ORIF + Ex Fix	LPS, CEF	0	1	0	0	0	1	1	1	8	37
Jani et al.20	13	61	^	10	6U 7B	ORIF + Perc	SFIF + RTF	2	4	0	0	2	4	0	11	15.9	51

Legend.

∗ Fractures classified as unimalleolar (U), bimalleolar (B) or trimalleolar (T).

+ Article classified fractures using Weber Classification: 28 Wb C, 52 Wb B, 4 Wb A.

++ Article classified fractures using Weber Classification: 1 Wb C, 7 Wb B.

+++ Article classified fractures using AO Classification: 11 44-B3, 11 44-B2, 1 44-B1, 1 44-A2, 2 43-C1, 1 43-A1.

^ Article did not provide this data.

ORIF = Open Reduction Internal Fixation.

ATS = Arthroscopy.

Perc = Percutaneous.

Ex Fix = External Fixation.

CMS = Coleman Methodology Score.

SFIF = standard fixation technique with small or mini fragment internal fixation (locking/nonlocking) following AO Principles.

FN = fibular nail + syndesmosis screw.

TTC = tibiotalocalcaneal nail (Zimmer Biomet) – no joint preparation.

PCS = Percutaneous Cannulated Screws (Partially-Threaded; 6.5 mm for lateral malleolus, 4 mm for medial malleolus, 3.5 mm for posterior malleolus and syndesmosis).

RTF = Retrograde trans-calcaneal-talar-tibial fixation using 7.3 mm cannulated screw or Steinmann pin (1/8–5/32 inch).

LPS = 3.5 mm LC-DCP small fragment plate with Pro-Syndesmotic Screws.

CEF = Multi-planar circular external fixator.

The overall patient cohort was comprised of 420 operative ankle fractures with a mean patient age of 57.90 ± 6.11 years. Males were 45.41% of the cohort and the overall mean follow-up was 21.65 ± 15.35 months. Overall, for each outcome across all 12 studies, Egger’s test suggested minimal concern over general publication bias (p-values ranged from 0.06 for hardware complications to 0.84 for nonunion). Further demographic and clinical data can be seen in Table 2. The mean overall CMS was 51.95 ± 4.27.

The standard fixation cohort included diabetic patients undergoing operative treatment for their ankle fractures with a standard AO technique using small or mini fragment internal fixation. This cohort was comprised of 326 ankle fractures with a mean patient age of 55.58 ± 4.78 years. Males were 45.71% of the cohort and the overall mean follow-up was 21.06 ± 16.67 months. Six studies were reviewed for this cohort and significant heterogeneity was observed on only one of the nine outcome measures: wound healing problems (I² = 88.86%, Q = 7.61, p = 0.0058). Values of I² ranged from 0.00% to 88.86%, with a mean I² of 29.71 ± 36.28%. The mean CMS for these six studies was 52.13 ± 4.48.

The minimally invasive fixation cohort was comprised of 80 ankle fractures with a mean patient age of 66.73 ± 0.65 years. Males were 44.16% of the cohort and the overall mean follow-up was 25.16 ± 9.40 months. Four studies were reviewed for this cohort and significant heterogeneity was observed on three of the nine outcome measures: closed fractures (I² = 64.98%, Q = 8.56, p = 0.0356), hardware breakage/migration (I² = 63.75%, Q = 8.27, p = 0.0406), and wound healing problems (I² = 71.14%, Q = 10.39, p = 0.0155). Values of I² ranged from 0.00% to 71.14%, with a mean I² of 36.74 ± 28.14%. The mean CMS for these four studies was 51.57 ± 3.34.

The combined constructs fixation cohort was comprised of 14 ankle fractures with a mean patient age of 62.42 ± 1.60 years. The overall mean follow-up was 15.33 ± 2.11 months. Two studies were reviewed for this group and significant heterogeneity was observed on five of the nine outcome measures: closed fractures (I² = 85.91%, Q = 35.48, p = 0.0001), surgical site infections (I² = 59.11%, Q = 12.23, p = 0.0318), Charcot neuroarthropathy (I² = 60.08%, Q = 12.52, p = 0.0283), unplanned return to the operating room (I² = 92.69%, Q = 54.71, p = 0.0001), and wound healing problems (I² = 83.33%, Q = 23.99, p = 0.0001). Values of I² ranged from 0.00% to 92.69%, with a mean I² of 42.35 ± 41.65%. The mean CMS for these two studies was 50.00 ± 3.74.

3.1. Incidence and predicted odds of outcomes of the alternative fixation vs. standard fixation cohorts

The current study evaluated soft tissue integrity of the injury (open versus closed injury) and outcomes involving hardware complications (breakage/migration), surgical site infection, Charcot neuroarthropathy, loss of reduction, nonunion, unplanned return to the operating room, wound healing/cast ulcers, and limb salvage. Overall, there were 47 ankles (11.2%) with open injuries, nine ankles (2.3%) which had hardware complications, 60 ankles (14.3%) with infection, and 41 ankles (10.2%) with wound healing issues or cast ulcers. Additionally, 14 ankles (3.3%) went on to develop Charcot neuroarthropathy, seven ankles (1.7%) lost reduction, and 15 ankles (3.8%) developed nonunion. These factors lead to 67 ankles (16.7%) requiring unplanned return to the operating room and an overall limb salvage rate (avoidance of a major amputation) of 97.9%.

Predicted odds of hardware breakage or migration (OR = 7.00, 95% CI: 1.45 to 33.69, p = 0.0152, FDR = 0.0659) were greater for patients of minimally invasive fixation treatment than for patients of standard fixation treatment. Also, predicted odds of hardware breakage or migration (OR = 36.22, 95% CI: 6.24 to 209.88, p = 0.0001, FDR = 0.0004), nonunion (OR = 6.03, 95% CI: 1.30 to 27.97, p = 0.0217, FDR = 0.0521), infection (OR = 3.24, 95% CI: 1.05 to 10.03, p = 0.0409, FDR = 0.0701), and reoperation (OR = 2.88, 95% CI: 0.93 to 8.90, p = 0.0657, FDR = 0.0986) were greater for patients of combined constructs fixation treatment than for patients of standard fixation treatment. However, combined constructs fixation treatment had lower odds of closed fractures (OR = 0.28, 95% CI: 0.08 to 0.94, p = 0.0390, FDR = 0.0701) and limb salvage (OR = 0.10, 95% CI: 0.02 to 0.05, p = 0.0052, FDR = 0.0156) than standard fixation treatment. Incidence rates and odds ratios for all outcomes of diabetic ankle fractures are reported in Table 3, Table 4.

Table 3.

Incidence rates and odds ratios of reported outcomes of standard and alternative fixation (minimally invasive) methods in diabetic ankle fractures.

Characteristics and Outcomes	Total Ankle Fractures	Standard Fixation	Alternative Fixation (Minimally Invasive)	Odds Ratio	95% CI for Odds Ratio	P-value (FDR)
Mean Age, years (SD)	57.90 (6.11)	55.58 (4.78)	66.73 (0.65)	–	–	0.0001 (0.0013)
Male Gender	183/403 = 45.4%	149/326 = 45.7%	34/77 = 44.2%	0.94	0.57 to 1.55	0.8136 (0.9374)
Closed Fractures	373/420 = 88.8%	291/326 = 89.3%	72/80 = 90%	1.04	0.47 to 2.30	0.9253 (0.9374)
Hardware Breakage/Migration	9/393 = 2.3%	2/299 = 0.7%	4/80 = 5%	7.00	1.45 to 33.69	0.0152 (0.0659)
Infection	60/420 = 14.3%	49/326 = 15.0%	6/80 = 7.5%	0.49	0.20 to 1.15	0.1038 (0.3374)
Charcot Neuroarthropathy	14/420 = 3.3%	13/326 = 4.0%	1/80 = 1.3%	0.43	0.08 to 2.43	0.3455 (0.7486)
Loss of Reduction	7/420 = 1.7%	5/326 = 1.5%	2/80 = 2.5%	1.86	0.40 to 8.53	0.4238 (0.7871)
Nonunion	15/420 = 3.8%	10/326 = 3.1%	3/80 = 3.8%	1.36	0.39 to 4.71	0.6261 (0.8776)
Unplanned return to OR	67/401 = 16.7%	51/307 = 16.6%	11/80 = 13.8%	0.82	0.41 to 1.65	0.5857 (0.8776)
Wound Healing/Cast Ulcers	41/401 = 10.2%	33/307 = 10.7%	7/80 = 8.8%	0.83	0.36 to 1.93	0.6751 (0.8776)
Limb Salvage	411/420 = 97.9%	320/326 = 98.2%	79/80 = 98.8%	1.07	0.17 to 6.51	0.9374 (0.9374)
Mean Follow-up, months (SD)	21.65 (15.35)	21.06 (16.67)	25.16 (9.40)	–	–	0.0038 (0.0247)
Coleman Methodology Score (SD)	51.95 (4.27)	52.13 (4.48)	51.57 (3.34)	–	–	0.2162 (0.5621)

M = Sample Mean; SD = Standard Deviation.

Odds ratio = predicted odds (with Firth’s bias correction) of each outcome for alternative fixation – minimally invasive vs. standard fixation.

P-value (two-tailed) associated with the test of group differences (alternative fixation – minimally invasive vs. standard fixation) on each characteristic. FDR = False Discovery Rate.

Table 4.

Incidence rates and odds ratios of reported outcomes of standard and alternative fixation (combined constructs) m

ethods in diabetic ankle fractures.

Characteristics and Outcomes	Total Ankle Fractures	Standard Fixation	Alternative Fixation (Combined Constructs)	Odds Ratio	95% CI for Odds Ratio	P-value
Mean Age, years (SD)	57.90 (6.11)	55.58 (4.78)	61.42 (1.60)	–	–	0.0001 (0.0004)
Male Gender	183/403 = 45.4%	149/326 = 45.7%	N/Aa	–	–	–
Closed Fractures	373/420 = 88.8%	291/326 = 89.3%	10/14 = 71.4%	0.28	0.08 to 0.94	0.0390 (0.0701)
Hardware Breakage/Migration	9/393 = 2.3%	2/299 = 0.7%	3/14 = 21.4%	36.22	6.24 to 209.88	0.0001 (0.0004)
Infection	60/420 = 14.3%	49/326 = 15.0%	5/14 = 35.7%	3.24	1.05 to 10.03	0.0409 (0.0701)
Charcot Neuroarthropathy	14/420 = 3.3%	13/326 = 4.0%	0/14 = 0%	0.80	0.04 to 15.59	0.8832 (0.9180)
Loss of Reduction	7/420 = 1.7%	5/326 = 1.5%	0/14 = 0%	0.98	0.10 to 38.22	0.6411 (0.7693)
Nonunion	15/420 = 3.8%	10/326 = 3.1%	2/14 = 14.3%	6.03	1.30 to 27.97	0.0217 (0.0521)
Unplanned return to OR	67/401 = 16.7%	51/307 = 16.6%	5/14 = 35.7%	2.88	0.93 to 8.90	0.0657 (0.0986)
Wound Healing/Cast Ulcers	41/401 = 10.2%	33/307 = 10.7%	1/14 = 7.1%	0.91	0.15 to 5.41	0.9180 (0.9180)
Limb Salvage	411/420 = 97.9%	320/326 = 98.2%	12/14 = 85.7%	0.10	0.02 to 0.50	0.0052 (0.0156)
Mean Follow-up, months (SD)	21.65 (15.35)	21.06 (16.67)	15.33 (2.11)	–	–	0.0001 (0.0004)
Coleman Methodology Score (SD)	51.95 (4.27)	52.13 (4.48)	50.00 (3.74)	–	–	0.0806 (0.1075)

M = Sample Mean; SD = Standard Deviation.

Odds ratio = predicted odds (with Firth’s bias correction) of each outcome for alternative fixation – combined constructs vs. standard fixation.

P-value (two-tailed) associated with the test of group differences (alternative fixation – combined constructs vs. standard fixation) on each characteristic. FDR = False Discovery Rate.

^aUnable to obtain gender data from Jani et al.²⁰ study.

4. Discussion

Our systematic review of eleven studies focuses on determining the long-term outcomes of fixation of diabetic ankle fractures. The aim of this review was to evaluate if alternative fixation modalities like minimally invasive or combined constructs have decreased complication rates compared to standard fixation techniques. Overall, the review demonstrates an average limb salvage rate of 97.9% and an average follow-up of 21.7 months. Of the 420 ankle fractures, 88.8% were closed injuries. Surgical site infections were noted in 14.3% of fractures, wound healing problems in 10.2%, and 16.7% of fractures had an unplanned return to the operating room. In the overall cohort, 3.8% of fractures went on to nonunion, 1.7% lost reduction and 3.3% developed Charcot neuroarthropathy. These complication rates are consistent with previous studies and reviews discussing the overall complication rates in diabetic ankle fractures.^30,31 Despite these large complication rates, there was a high rate of successful limb salvage indicating the need to act on these unstable injuries. As medical management of diabetes continues to improve, these patients have longer survival rates and therefore the prevalence of these injuries will continue to increase.

This systematic review compared two study groups (minimally invasive and combined constructs) to a control group (standard fixation). Previously published studies have described various fixation techniques in higher risk patients including diabetics; however, few of these have compared these techniques to each other to determine if outcomes are improved over a long-term period. The analysis showed a significantly higher rate of adverse outcomes in the study groups compared to the control group. Specifically, the minimally invasive cohort had an increased risk of hardware breakage/migration (4/80 = 5%), which could be a testament to the fixation methods studied in this review. The minimally invasive fixation methods reported in this review included fibular nails, tibiotalocalcaneal nails/retrograde transarticular screws and percutaneous cannulated screws. The analysis showed that all of the hardware failures occurred within the tibiotalocalcaneal nail or retrograde transarticular screw patients with 2 of them being proximal locking screw breakage (patients remained asymptomatic), 1 having nail breakage due to persistent nonunion requiring reoperation, and 1 having a retained transarticular pin which proved to be clinically insignificant.^18,20 Overall, three of these four hardware complications were minor. The patient who had nail breakage requiring further surgery was also morbidly obese (BMI of 68) and had diabetic nephropathy. Most of these minimally invasive implants rely on indirect reduction and relative stability, rather than direct, anatomic reduction and fixation with implants imparting absolute stability. Therefore, the implants utilized in this cohort may not provide as rigid of fixation when compared to standard fixation with plate and screws. Bazarov et al. discussed the use of Minimally Invasive Plate Osteosynthesis (MIPO)¹⁴ in higher risk patient populations (including diabetic patients, elderly, patients with compromised soft tissues) and found similar complication rates to standard, open techniques of fixation demonstrating the possible differences in implants with absolute versus relative stability. Our review also demonstrated a trend towards a difference in infection rates with minimally invasive implants (7.5% versus 15%). Although not statistically significant, we feel this clinically relevant as one of the major indications for performing minimally invasive surgery is to preserve the soft tissue envelope in these patients and decrease infection. In all likelihood, this review was underpowered to detect a significant difference. Even though there is a higher risk of hardware breakage or migration in the minimally invasive cohort, most of these situations are clinically insignificant, and the rates of nonunion and limb salvage are similar with lower infection rates.

The combined constructs cohort was a relatively small group comprised of 14 ankle fractures when compared to standard fixation group. One of these fractures was from a case report by Facaros et al.¹⁹ which utilized ORIF techniques in combination with multi-planar circular external fixation. The other 13 fractures utilized ORIF techniques in combination with retrograde transarticular screws.²⁰ Unfortunately, the small numbers can bias the results. Nevertheless, this cohort had an increased risk of hardware breakage/migration, infection and nonunion. These results can be understood in the context of the implants used as well as the patient population. Particularly, the external fixation patient also suffered from peripheral neuropathy and dementia (was not oriented to person, place or time). He presented with deformity, multiple fracture blisters and pressure necrosis over the medial malleolus. This patient ultimately had problems with infection and wound healing which required repeat interventions in the operating room. The hardware problems and nonunions were seen in patients undergoing supplementary fixation with retrograde transarticular screws or pins. All of these patients had evidence of diabetic neuroarthropathy based on clinical examination. The two hardware problems were due to retained or broken implants which proved to be clinically insignificant. These screws are placed across joints and are meant to add more rigid fixation. However, in the long-term, motion at the tibiotalar joint can cause these screws to fail and break or migrate. Although post-operative glycemic control or peripheral vascular status was not reported for these patients, the presence of neuropathy on examination can be reflective of this status and explain the increased incidence of infections. Of note, overall limb salvage rates were better with standard fixation than in the combined constructs.

There are several limitations in the systematic review. All of the studies discussed were retrospective case series or case reports and variables were not consistently reported. Despite this, among the six standard fixation technique studies, heterogeneity was observed among one of the nine outcomes, while in the study groups it was observed in three out of the nine for minimally invasive and five out of nine for combined constructs. The Coleman Methodology Scores ranged from 37 to 58 with the average of 51, demonstrating that the methodological quality overall was poor. Our study focused on the treatment of diabetic ankle fractures, but few studies quantified patients’ diabetic control (either through Hemoglobin A1c or perioperative glucose measurements) or baseline neurovascular status (neuropathy, peripheral arterial disease). Of the eleven studies, only three reported glycemic control or Hemoglobin A1c and eight studies detailed patients’ contributing risk factors and medical comorbidities. As Wukich et al.³² reported, these factors greatly impact rates of surgical site infection, so having more consistent reporting of these findings perhaps could have helped in developing important conclusions. Although fractures were generally classified, there was variability in schemes (Lauge-Hansen versus Weber or AO) along with the general reporting of medical comorbidities and glycemic control. This made it difficult to compare operative indications between studies. Outcome measures were inconsistently reported accounting for variability among studies. During our database search, since we restricted studies to full length, English, peer-reviewed articles, potentially other relevant articles may have been excluded. Therefore, the overall heterogeneity, patient selection and outcome parameters make it difficult to make a direct conclusion. Lastly, the combined constructs cohort was relatively small (14 ankles) and so there may have been a strong element of selection bias in these results arising from the included studies. Alternative fixation methods, in particular the combined constructs, were used in higher risk patients (all had neuropathy). In the future, prospective studies should aim to address these issues.

In conclusion, diabetic ankle fractures are a challenging entity to treat. Nonoperative treatment has been associated with increased loss of reduction and development of Charcot neuroarthropathy.⁴ Therefore, operative treatment has been advocated, but this is associated with higher risks of surgical site infections or wound complications due complication of diabetes (peripheral artery disease, neuropathy and end stage renal disease). Diabetic patients are at risk for delayed bone healing and therefore prolonged immobilization is necessary contributing to the morbidity of the injury. To optimize operative management, techniques have continued to evolve from standard internal fixation. Whether it be using more minimally invasive incisions to respect the soft tissue envelope or increasing fixation to achieve more stability, it is important to adhere to the fundamentals of fracture management. This systematic review echoes previous studies in showing that operative diabetic ankle fractures are associated with higher complication rates, but also high limb salvage rates. Critical to achieving improved limb salvage rates and decreasing complications is to continue following basic AO principles, and to further study the role of minimally invasive techniques. While this study demonstrated increased complication rates with large complex combined constructs, the reader should not interpret this as a condemnation of very robust fixation. These findings were based on very small numbers, and the reader should interpret the results in this context. In addition to stable fixation, judicious glycemic control and prolonged immobilization are necessary to achieve optimal results. While the hallmark of these injuries is bone fracture, soft tissue injury and postoperative wound complications play a major role in determining outcomes.