Early Bayesian modeling of a potassium lab-on-a-chip for monitoring of heart failure patients at increased risk of hyperkalaemia

ObjectivesInnovative point-of-care (POC) diagnostics are likely to have a strong impact on health care. The aim of this study is to conduct an early assessment of a point-of-care chip for the detection of a pathological deviation of the potassium levels in patients at increased risk, specifically in patients with heart failure (HF) requiring diuretics and ACE inhibitors that can both interfere with potassium levels, which may lead to serious clinical complications. This study also identifies the key factors that determine the success of the application under study.MethodsA Markov health state transition model was developed representing the disease process. Model parameters were obtained from various literature sources and estimated using interviews and related data. Simulation was carried out for 60 cycles of 1 month each. A discount rate of 3.5% was used, both for costs and utilities. In order to assess uncertainty a probabilistic sensitivity analysis was carried out from which a cost-effectiveness acceptability curve was derived.ResultsFor an anticipated number of 121 measurements per year with a cost of €16.60 per chip, an incremental cost-effectiveness ratio (ICER) of 34,856 €/QALY was found. Sensitivity analysis revealed that the threshold for the costs per chip was €19.30 in order to maintain a positive net monetary benefit. Also, model results are very sensitive to the utility of hyperkalaemia and to the probability to develop severe hyperkalaemia.ConclusionsThe question whether or not a POC chip to measure potassium concentrations in order to avoid a severe potassium imbalance is likely to be cost-effective cannot be definitively answered with the information at our disposal. Further research should focus on heart failure patients at particularly high risk of severe life-threatening hyperkalaemia, for instance in the presence of significant renal dysfunction. It may be expected that the use of the chip in such patient populations may render this point-of-care application very cost-effective.

► Early HTA is useful in the development stages of a new medical technology.
► Markov modeling is suitable to assess the economic viability of an innovation.
► PSA and VOI analysis are useful to handle uncertainty in CE outcomes.
► Greatest limitation for early HTA is the unavailability of reliable data.