Validation of Arterio Venous Access Stage Classification by Using Machine Learning in Prediction of Vascular Access Creation and Uninterrupted Use of the Vascular Access for Dialysis
Validation of Arterio Venous Access Stage Classification by Using Machine Learning in Prediction of Vascular Access Creation and Uninterrupted Use of the Vascular Access for Dialysis
VAVASC trial is an observational multicentre study. The aim of this trial is to validate AVAS (arteriovenous vascular access stage) classification. The classification is used for determining which type of access is the most suitable for the patient on the basis of the patient´s vascular anatomy
The methodology of this trial is to apply AVAS classification on patients who are indicated for creation of vascular access for hemodialysis. Data on these patients (vascular anatomy status, AVAS type, and predicted type of arterio venous access, demographic data etc.) will be than statistically analysed. Patients will then undergo creation of the selected arteriovenous access. They will be observed in terms of the access functionality. The follow up will be 1 to 3 years. The second aim of this study is to evaluate the relationship between AVAS classification and uninterrupted use of the created arterio venous access.
Ultrasound of upper limb vessels as a part of preoperative assessment of patients referred for arteriovenous access placement has been recommended repeatedly in several studies and guidelines. Each patient needing arterio-venous access (AVA) is unique and has different possibilities for fistula creation that depend on many factors as well the anatomical condition of inflow arteries and outflow veins, the so-called 'arteriovenous access map'. In 2020 we created, published and retrospectively evaluated the inter-rater reliability of the AVAS classification, which is a simplified way of sharing the information about suitability for access creation depending on vascular anatomy (doi: 10.1093/ckj/sfaa189).
However, before routine use of the AVAS classification system in clinical practise it requires prospective validation and comparison with other clinical factors. Finally, whether the AVAS classification can be used for the prediction of vascular access function as defined by uninterrupted use of vascular access for dialysis without need for any access intervention/procedures (SONG-HD, doi: 10.1053/j.ajkd.2017.12.003) remains to be tested.
This project has two main parts:
Validation of AVAS classification with and without combination with other factors using a predictive statistical model
.
Validation of AVAS classification in terms of prediction of uninterrupted use of successfully created vascular access for dialysis.
The data will be collected prospectively. Both parts of the project will be taking place simultaneously.
Data will be collected prospectively (Belfast City Hospital, UK, University hospital Královské Vinohrady Prague CZ). Other centres around the world will be asked for cooperation (announcement will be posted via Twitter, ResearchGate, LinkedIn, Facebook and Kidney Academy) in January 2021. The data collected will be the parameters of the vessels on the upper limb, in order to be able to evaluate AVAS (VAVASC FORM). Along with anatomical dispositions of the patients, other parameters will be collected.
Obligatory parameters (parameters A):
Side of the arm (Left / Right) Dominant hand (Left / Right) Allen's test (Positive / Negative) (A negative Allen's test is demonstration of a complete palmar arch and intact collateral blood flow to the hand) Sex(M / F) Age, Height, Weight, BMI, Diabetes (Yes / No) Central venous line previous or current (Yes / No) Side of the central venous line if place (Left / Right)
Non-obligatory parameters (parameters B):
Smoking history (Yes / No) Hypertension (Yes / No) Heart failure (Yes / No) Ischemic heart disease (Yes / No) Cancer (previous or current) (Yes / No)
In order to evaluate uninterrupted use of the vascular access for dialysis (part 2 of the study), we will follow the patients and do a time to event analysis of the arteriovenous accesses for one year after creation or up until the first intervention for maintaining patency has been performed. If the patient dies during follow up the date of death should be recorded. The data for this part will be collected prospectively (Belfast City Hospital, UK, University hospital Královské Vinohrady Prague, CZ and other centres).
Predictive models will be used for validation of AVAS classification. The data will be registered via electronic database and analysed by Waldauf, P. and O'Neill, S.
Inclusion Criteria:
Exclusion Criteria: