A donor heart scoring model to predict transplant outcomes

Abstract

Selection of heart donors is the most important stage on which the success of heart transplantation depends.

Objective: to create a donor heart scoring model based on a number of donor characteristics.

Materials and methods. The study used data from 650 brain-dead donors who underwent organ explantations between January 1, 2012 and December 31, 2017. In binomial logistic regression, non-selection of heart donor was used as a dependent variable, while donor characteristics were used as factor features. In regression model, the odds ratio was determined for each donor factor, which was transformed into points. The sum of the points of each of the donor factors included in the model was taken as the score of the donor heart. The proposed model was validated on a sample of donors  for the period from January 1, 2019 to December 31, 2019; n = 218.

Results. The model includes donor characteristics, such as age, cause of death (traumatic brain injury (TBI)/stroke), history of hypertension and diabetes, cardiac arrest with subsequent recovery, own pathology and traumatic heart disease, as well as heart rate, systolic blood pressure, arterial lactate, and need for norepinephrine immediately before organ harvesting. Based on the average value of the sum of points, low-risk donors (LRD ≤17 points) and high-risk donors (HRD ≥18 points) were identified. In the validation pool of donors, the proportion of heart failure among LRD and HRD was 4.1% and 78.6%, respectively, p < 0.0001, Pearson’s χ2 – 130.9.

Conclusion. The presented donor heart scoring model accurately reflects the probability of using a donor’s heart for transplantation and creates conditions for optimal distribution of heart transplants, especially from high-risk donors.