Faktori koji utiču na nivo osposobljenosti za hod uz pomoć proteze nakon amputacije donjih ekstremiteta

Abstract

Lower limb amputation represents one of the classical rehabilitation problems amenable to intervention by a physiatrist. Because of the aging of the population and the increased incidence of diabetes the number of amputation is expected to increase in the future. Identification of predictive factors of ability to walk with prosthesis is very important in order to define patient's potentials and realistic goals. Goals: To determine a number of comorbidities in persons with lower limb amputation and their influence on the level of ability to walk with prosthesis. To determine the influence of the different factors such as: functional status of the residual limb and time period from amputation to beginning of prosthetic rehabilitation process, on the level of ability to walk with prosthesis. Patients and methods: The study was etrospective-prospective and encompassed patients with major lower limb amputation who underwent prosthetic rehabilitation at Clinic for medical rehabilitation in Clinic center of Vojvodina for the period 2000-2013. Beside classical statistics, we formed regression models using algorithms based on Support Vector Machines (SVM). Results: Vascular disease accounted for the vast majority of amputations: in retrospective study 88,5 percents, and in prospective study for 90 percents of all cases. Majority of amputations were transfemoral: 65 percents in retrospective and 62 percents in prospective study. Average Functional Co morbidity Index (FCI) score was 2,19 in retrospective, and 2,91in prospective study. Average time period from amputation to prosthetic fitting was 186,22 days in retrospective, and 172 days in prospective study. Multivariate analysis in retrospective study identified age, level of amputation, and functional status at the beginning of the rehabilitation treatment as significant predictors for level of ability to walk with prosthesis. Our model identified level correctly in 58,1 percents of all cases. Multivariate analysis in prospective study identified Beck Depression Inventory (BDI form I), level of amputation, muscle strength of the residual and intact limb hip extensors, balance at the beginning of the rehabilitation treatment as significant predictors for level of ability to walk with prosthesis. Our model identified level correctly in 81,4 percents cases. Linear regression model with 2-minute walk test (TMWT) as output variable identified level of amputation, muscle strength of the residual and intact limb hip extensors, age, and balance at the beginning of the rehabilitation treatment as significant predictors (adjusted R2=0,712, p<0,001). Linear regression model with Timed Up and Go test (TUG) as output variable identified level of amputation, age, muscle strength of the residual limb hip extensors, arthritis, Body Mass Index (BMI) and balance at the beginning of the rehabilitation treatment as significant predictors (adjusted R2=0,687, p<0,001). Using parameters measured at the beginning of the rehabilitation treatment, it is possible to predict, with high level of determination: level of ability to walk with prosthesis, results of TUG and TMWT at the end of the treatment. Conclusion: Level of amputation, muscle strength of the residual limb hip extensors, and balance on the intact limb were significant predictors for all three functional parameters tested. Machine learning methods were very efficient in predicting level of ability to walk, results of TUG, and TMWT.