PUK12 Using Real-World DATA to Improve the Accuracy of Economic Models; Experience from the Swedish Renal Register

Abstract

When data from clinical trials is lacking economic models typically obtain their estimates from published literature, which is often derived from non-representative populations, with varying confounders and observation periods. To inform economic models in the setting of chronic kidney disease (CKD), we generated reliable estimates on transition probabilities during the disease progression cycle using real-world data from the nationwide Swedish Renal Register. Observational analysis from all nephrologist-referred patients with CKD stages 3-5 in Sweden during 2010-2016, with complete capture of disease progression and death events. Using Markov processes, we estimated the annual probability of dying or progressing to a more severe CKD stage. The six possible states were CKD stages 3, 4 and 5, chronic dialysis, kidney transplant and death. We also evaluated the transition probability by CKD state in individuals with high (>130 pg/mL) and low (= <130 pg/mL) serum levels of parathyroid hormone (PTH). In total 17,679 patients fulfilled our inclusion and exclusion criteria. Median age was 73 years, and 35% were women. The annual probability of transitioning to more severe CKD stages was 51% in stage 3 and 44% in stage 4 and 5; the probability of dying increased from 4% in stage 3, to 22% in stage 5 and 20% in patients on dialysis. 8,246 patients had high PTH levels, and their annual probability of transitioning to death or the next severity stage of CKD was between 40-60% significantly higher compared to patients with low PTH. Data from rich and representative patient registers can improve the estimation of reliable risk estimates to inform economic models. Our observed transition probabilities of patients with CKD can be of use to any future studies on this disease.