Using data safe havens in nephrology

The reality that chronic kidney disease patients are often excluded from randomised clinical trials can result in gaps in evidence-based medicine and clinical questions left unanswered. The School of Medicine has a wealth of experience in addressing some of these questions through the use of real-world health data and health informatics. That’s because the School hosts the original Data Safe Haven (also known as a ‘Trusted Research Environment’), home to the most longitudinal health datasets in the country.  

Here are a few case study examples of how our quality data and health informatics expertise has enhanced randomised controlled trials for the benefit of clinical practice.

Predicting risk of acute kidney injury

Linking anonymised health record data provided by the Health Informatics Centre (HIC) allowed for the development of a risk prediction score to help prevent acute kidney injury in patients undergoing surgery. This score identifies patients at high-risk for developing acute kidney injury preoperatively to try and prevent them from getting postoperative acute kidney injury.

Seven predictors of acute kidney injury, identified from the HIC’s vast collection of datasets, were used to develop this prediction model which, ultimately, can impact patient survival. Access to this data and risk prediction score could not only allow clinicians to identify or be alerted to patients at high risk during preoperative assessment for intervention but could also allow for identifying and selecting those who may benefit from targeted treatment or research in clinical trials.

Impact of vaccination on dialysis

Given that the immune system of patients on dialysis or with a kidney transplant is weakened, these patients are at significant risk if infected with Covid-19 compared to the general population - specifically, they are at a considerably increased risk of death after infection with Covid-19.

Using real-world healthcare data covering Scottish patients on dialysis and with a kidney transplant, the Scottish Renal Registry (SRR) discovered that even after receiving two doses of a Covid-19 vaccine, patients are still at a 9% risk of death following Covid-19 infection. This research conducted by the SRR - Chaired by Dr Samira Bell at the University of Dundee - highlights the critical requirement for a third dose of vaccine in this patient group, and a fourth dose in people with a kidney transplant.

Evaluating risk of adverse events

Currently, there is great interest in using the linkage of routinely collected data and datasets to see if individuals are developing acute kidney injury as a serious adverse event from using certain classes of medicines.

As an example, using the HIC’s prescribing data covering the population of NHS Tayside, it was found that a significant increased risk in acute kidney injury was observed alongside the increased prescription of antibiotics, particularly sulphonamides and trimethoprim. This is a useful illustration of how data can be used in precision medicine to predict and prevent adverse outcomes on renal function.

Evaluating efficacy in dialysis patients

From a regulatory perspective, RCTs are, of course, the required standard for treatment evaluation.  However, many pharmaceutical companies are considering efficient, cost-effective ways of addressing questions left unanswered by RCTs - including where there might be unmet needs in the market.  With a spirit of making the most of new methods available, companies have the opportunity to not only address unanswered clinical questions but also to access and make use of outcome data. 

Trusted research environments for the future of data-driven research

Precision medicine can be taken to the next level by looking at genetics and metabolism to see if there are factors that affect the metabolism of certain medicines - something that will add great value to predicting adverse events in advance of, or in the absence of, observations from human trials.

Furthermore, this level of analysis can help identify certain groups of patients that may benefit from non first-line pharmacological agents - an essential element to accurate health economics and efficient use of health system resources.