Routine real-time cost-effectiveness monitoring of a web-based depression intervention: a risk-sharing proposal

Abstract

Background: A new health care technology must be cost-effective in order to be adopted. If evidence regarding cost-effectiveness is uncertain, then the decision maker faces two choices: (1) adopt the technology and run the risk that it is less effective in actual practice, or (2) reject the technology and risk that potential health is forgone. A new depression eHealth service was found to be cost-effective in a previously published study. The results, however, were unreliable because it was based on a pilot clinical trial. A conservative decision maker would normally require stronger evidence for the intervention to be implemented.

Objective: Our objective was to evaluate how to facilitate service implementation by shifting the burden of risk due to uncertainty to the service provider and ensure that the intervention remains cost-effective during routine use.

Methods: We propose a risk-sharing scheme, where the service cost depends on the actual effectiveness of the service in real-life setting. Routine efficacy data can be used as the input to the cost-effectiveness model, which employs a mapping function to translate a depression specific score into quality-adjusted life-years. The latter is the denominator in the cost-effectiveness ratio calculation, required by the health care decision maker. The output of the model is a "value graph", showing intervention value as a function of its observed (future) efficacy, using the €30,000 per quality-adjusted life-year (QALY) threshold.

Results: We found that the eHealth service should improve the patient's outcome by at least 11.9 points on the Beck Depression Inventory scale in order for the cost-effectiveness ratio to remain below the €30,000/QALY threshold. The value of a single point improvement was found to be between €200 and €700, depending on depression severity at treatment start. Value of the eHealth service, based on the current efficacy estimates, is €1900, which is significantly above its estimated cost (€200).

Conclusions: The eHealth depression service is particularly suited to routine monitoring, since data can be gathered through the Internet within the service communication channels. This enables real-time cost-effectiveness evaluation and allows a value-based price to be established. We propose a novel pricing scheme where the price is set to a point in the interval between cost and value, which provides an economic surplus to both the payer and the provider. Such a business model will assure that a portion of the surplus is retained by the payer and not completely appropriated by the private provider. If the eHealth service were to turn out less effective than originally anticipated, then the price would be lowered in order to achieve the cost-effectiveness threshold and this risk of financial loss would be borne by the provider.

Keywords: depression; medical economics; value-based purchasing.

Figure 1

Box-plots of BDI scores for patients, entering the pilot trial (initial BDI) and treatment/control groups at the end of the pilot trial (piecewise linear line shows how BDI score [horizontal axis] is mapped onto the QALY scale [vertical axis]).

Figure 2

Relation between intervention value, price, and cost, depending on the actual efficacy (BDI improvement) during routine use (intervention value calculated using €30,000/QALY threshold; shaded regions depict provider loss and social surplus, shared between the provider and payer at an example ratio of 1:1 [50% private appropriation level, intervention cost at €200 per patient]).

Figure 3

Value of a BDI point improvement (treatment minus control) as a function of the patient's initial BDI score (calculation based on a treatment effect of 10 BDI points).