Multiply imputing missing values arising by design in transplant survival data.

Area of Science:

• Medical Statistics
• Transplantation Research
• Data Science

Background:

• Missing data is a common challenge in transplant survival studies due to varying data collection.
• Explanatory variables, crucial for survival prediction, are often missing for a significant number of recipients.
• Complete case analysis is the conventional method, assuming missing data is not related to survival time.

Purpose of the Study:

• To investigate the effectiveness of multiple imputation in addressing missing-by-design data in transplant survival analysis.
• To compare multiple imputation with complete case analysis using a kidney transplantation survival study.
• To demonstrate the advantages of multiple imputation in handling substantial missing data in a medical context.

Main Methods:

• Utilized multiple imputation techniques to estimate missing explanatory variables.
• Performed a comparative analysis between multiple imputation and complete case analysis.
• Applied methods to a real-world study on survival after kidney transplantation.

Main Results:

• Multiple imputation comprehensively outperformed complete case analysis across various measures.
• The study demonstrated the efficacy of imputation even with large amounts of missing data.
• Results challenge the skepticism often associated with imputing extensive missing data in medical research.

Conclusions:

• Multiple imputation is a superior method for handling missing data in transplant survival studies compared to complete case analysis.
• This finding is particularly relevant for improving the accuracy of survival predictions in kidney transplantation.
• The study advocates for the adoption of multiple imputation in medical research despite potential reservations.