Article Text

Download PDFPDF
115 Using data science to improve patient outcomes in renal transplantation
  1. William Bryant,
  2. John Booth,
  3. Wesley Hayes
  1. Great Ormond Street Hospital


Introduction Conducting research in a hospital environment offers many opportunities, but also many challenges. Historically, accessing data for research has been a tedious and error-prone process, and advanced analytics tools have not been readily available to researchers. We have applied advanced computational tools to overcome these issues in order to investigate the prevalence of postoperative plasma electrolyte imbalance in paediatric kidney transplant recipients by use of high-frequency measurements for patients in a 72-hour post-operative window. We have shown that these imbalances occur with high prevalence in this cohort.

Methods Electrolyte data were extracted using SQL from the GOSH Data Warehouse. Plasma electrolytes in the first 72-hours post kidney transplant were analysed in 76 children transplanted between January 1, 2015, and January 31, 2018, managed with a standard intravenous fluid strategy used in most UK paediatric transplant centers.

Results 16,582 lab test results were extracted along with weight measurements for each child, which were filtered and transformed for downstream analysis. Of 76 pediatric transplant recipients of median age 9.9 (range 2.2–17.9) years predominantly managed with 0.45% sodium chloride 5% glucose, 45 (59%) developed acute hyponatremia, 23 (30%) hyperkalemia, and 43 (57%) non–anion–gap acidosis in the postoperative period. Hyperglycemia occurred in 74 (97%) patients. Hyperkalemia was more prevalent in deceased than live donor recipients (P=0.003) and was significantly associated with non-anion-gap acidosis (P < 0.001). Recipient weight was not associated with overt electrolyte imbalance.

Conclusion Postoperative plasma electrolyte imbalance is common in paediatric kidney transplant recipients. Current clinical care strategies mitigate the associated risks of neurological sequelae to some degree. Programmatic access to data drastically reduces time and transcription errors in acquiring patient data, resulting in a very fast turnaround and has contributed to a successful application for a multi-site clinical trial within 13 months of pilot project inception.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.