Future investigations on the association between insulin resistance in people without diabetes and chronic kidney disease (CKD) should consider the following aspects to facilitate causal inference and provide more robust findings. The study design should have an adequate follow-up period to rule out reverse causation and pre-existing diabetes, as well as to confirm the diagnosis of CKD. Known causes of CKD and relevant covariates should be identified where possible. Homeostasis model assessment of insulin resistance (HOMA-IR), being an indirect measure of insulin resistance, has limited sensitivity and specificity compared to direct methods like the hyperinsulinaemic-euglycaemic clamp. Regression modelling with HOMA-IR quartiles instead of continuous form may have masked more nuanced relationships. Sensitivity analyses, such as spline regression, could provide more insights about the association and mechanism. Propensity score methods could help address the inadequate overlap in covariate distributions, if present, by ensuring covariate balance. When investigating the CKD diagnostic performance of HOMA-IR, its cut-off for clinically meaningful insulin resistance should be well justified or comprehensively explored to improve the reliability of the results.