There are limits in conducting clinical trials… the biggest of which boils down to ethics and feasibility. Sometimes, an RCT is not going to happen. So, we are stuck with observational studies which have the inherent trouble in dealing with confounding and attributing causality.
In an attempt to overcome these limitations, these researchers performed a regression discontinuity design (RDD).
The RDD relies upon the existence of a continuous variable (i.e. D-dimer) for which there is a cut-off that determines treatment or investigation. Patients on either side of the cut off have much different probabilities of investigation but probably are reasonably matched for confounding factors.
These researchers included adult patients who had a D-dimer test during an ED visit over a 5-year period in the province of Alberta, Canada. They compared the group that had elevated D-dimers vs. those that were normal. The primary outcome was long term renal function (up to six months from ED visit).
150,028 patients met inclusion criteria. Mean baseline eGFR was 86.
There was no evidence of renal damage with a change of eGFR of only 0.4 (95%CI -4.9 to 4.0) up to 6 six months later.
The authors conclude;
“To our knowledge, this study provides the strongest evidence to date that intravenous contrast is not associated with significant kidney injury…”
This overall conclusion is correct. But... we are not worried about patients with near normal renal function to start with. We are much more interested in the cohort that has borderline function.
They tried a subgroup analysis by looking at patients with eGFR <45 but state that the analysis was hindered by missing data and small numbers. This is unfortunate.
What should we conclude?
Overall, the modern IV contrast agents probably don’t do much to kidneys. But we are not entirely sure in the cohort with more severe baseline renal impairment.
Goulden R, Rowe B, Abrahamowicz M, et al. Association of Intravenous Radiocontrast with Kidney Function. A Regression Discontinuity Analysis. JAMA Intern Med. 2021;181:767-774. [Link to article]