Predicting Reamputation at a Higher Level After Transmetatarsal Amputation

Abstract

The aim of this study was to develop a model predicting likelihood of progression to a higher level of amputation after transmetatarsal amputation (TMA). All transmetatarsal amputations entered into the national Vascular Quality Initiative database from 2013 to 2019 were investigated. Planned staged amputations were excluded from analysis. Two cohorts were created based on 1-year reamputation at a higher level status. A model for TMA progression was created based on a multivariable logistic regression analysis predicting reamputation at a higher level. The model was validated graphically (depicting predicted vs actual reamputation rates) and statistically utilizing bootstrapping with 1000 repetitions with replacement. During the study period, 781 TMAs were performed, with a 1-year reamputation rate of 27.3% (72.7% below knee amputation, 27.3% above knee amputation). Indications for reamputation included nonhealing wound (58.0%), infection (25.5%), and vascular disease progression (16.6%). The reamputation cohort was less Caucasian (48.8% vs 62.1%; P < .001), less ambulatory preoperatively (81.7% vs 88.5%; P = .01), had higher rates of ankle-brachial index (ABI) <0.7 (43.6% vs 35.6. P = .002), lower rates of ABI >1.4 (24.5% vs 42.96%; P = .002), and were less likely ambulatory at discharge (62.4% vs 72.0%; P = .01). Neuropathic ulcer indication (compared to infection and ischemic disease) and preoperative ambulatory status were protective against reamputation (Table). Active smoking, ABI <0.7 or >1.4, non-Caucasian race, and Hispanic ethnicity were associated with increased reamputation (Table). The TMA progression model was created from the aforementioned variables, with five risk group categories created based on quintile distribution of summed amputation progression scores (1 [lowest risk; score <0], 2 [score 0-1], 3 [score 2], 4 [score 3-4], 5 [highest risk; score >4]). The model displayed good fit with bootstrap validated area under the curve of 0.67 (95% confidence interval, 0.61-0.72) and appropriate Hosmer-Lemeshow goodness of fit testing with a P value of .88. As expected, Kaplan-Meier analysis for reamputation showed corresponding increasing 1 -year reamputation rates based on risk quintile (1 = 9.3%, 2 = 16.0%, 3 = 20.1%, 4 = 36.4%, 5 = 42.7%; P < .001) (Figure). Using six readily obtained preoperative variables (indication for amputation, ambulatory status, smoking status, ABI, race, ethnicity), we created the TMA progression model which accurately predicts 1-year reamputation rates at a higher level. This model can be utilized by physicians in determining the appropriate level of initial amputation in patients presenting with foot and toe wounds without options for revascularization.TableMultivariable logistic regression predicting variables associated with reamputation at a higher levelaVariables associated with reamputation at a higher levelBeta coefficient95% CIP valueScore valueNeuropathic ulcer as indication–1.04(–2.15 to 0.07).07–3Ambulatory preoperatively–0.73(–1.43 to –0.04).04–2Active smoker0.62(0.03 to 1.21).042ABI (compared to reference group 0.7-1.4)0 <0.70.8(0.19 to 1.41).013 >1.40.91(0.25 to 1.57).013Non-Caucasian race0.9(0.38 to 1.41)<.0013Hispanic ethnicity0.93(–0.07 to 1.93).073ABI, Ankle-brachial index.aScore values are attributed to each variable based on beta coefficient value. Each patient’s summed score value was utilized to create an overall transmetatarsal amputation progression score. Open table in a new tab