(1007) - A New Prediction Model for Quantifying Mortality Risk in Congenital Heart Disease(CHD) Patients after Heart Transplant (HTx)

https://doi.org/10.1016/j.healun.2018.01.1010Get rights and content

Section snippets

Purpose

To develop a new predictive model for estimating mortality risk in CHD HTx recipients.

Methods

CHD HTx recipients (age > 10) from the ISHLT registry (2005-2013) were randomly divided into training (n=966) and validation (n=415) samples. In the derivation cohort, survival random forest models identified the five most important variables for post HTx survival: gender mismatch, bilirubin, creatinine clearance, recipient age and recipient/donor age ratio. A CHD risk score (CRS) was created based upon their association with mortality. Survival among low (score <10%), medium and high risk

Results

Of 1,381 CHD recipients, 377 (27%) died after HTx during 2.8 median follow-up years. Whereas IMPACT classified all patients as medium and high risk (Figure 1A) the CRS classified a small proportion (4%) of CHD recipients as low risk. In the validation cohort, CRS showed better discrimination than the IMPACT score [Figure 1B, c-statistic 0.65 (0.59-0.70) vs 0.62 (CI 0.56-0.68) for IMPACT]. The CRS showed adequate calibration while IMPACT’s calibration was poor (Fig 1C). The CRS better classified

Conclusion

A new prediction model that incorporates 5 simple variables with adequate discrimination, excellent calibration and better identification of low risk CHD HTx recipients is presented.

References (0)

Cited by (0)

View full text