Economic evaluation in chronic pain: a systematic review and de novo flexible economic model

Abstract

There is unmet need in patients suffering from chronic pain, yet innovation may be impeded by the difficulty of justifying economic value in a field beset by data limitations and methodological variability. A systematic review was conducted to identify and summarise the key areas of variability and limitations in modelling approaches in the economic evaluation of treatments for chronic pain. The results of the literature review were then used to support the development of a fully flexible open-source economic model structure, designed to test structural and data assumptions and act as a reference for future modelling practice. The key model design themes identified from the systematic review included: time horizon; titration and stabilisation; number of treatment lines; choice/ordering of treatment; and the impact of parameter uncertainty (given reliance on expert opinion). Exploratory analyses using the model to compare a hypothetical novel therapy versus morphine as first-line treatments showed cost-effectiveness results to be sensitive to structural and data assumptions. Assumptions about the treatment pathway and choice of time horizon were key model drivers. Our results suggest structural model design and data assumptions may have driven previous cost-effectiveness results and ultimately decisions based on economic value. We therefore conclude that it is vital that future economic models in chronic pain are designed to be fully transparent and hope our open-source code is useful in order to aspire to a common approach to modelling pain that includes robust sensitivity analyses to test structural and parameter uncertainty.

Keywords: Chronic pain; Economic evaluation; Modelling assumptions; Transparency.

Fig. 1

Model structure—one treatment arm

Fig. 2

Scatterplot of base case cost-effectiveness pairs, assuming 50 % standard errors

Fig. 3

Cost-effectiveness acceptability curve using the Fig. 2 results

Fig. 4

Scatterplot of base case cost-effectiveness pairs, base case, assuming 10 % standard errors

Fig. 5

Cost-effectiveness acceptability curve using the Fig. 4 results

Fig. 6

Tornado diagram of the top ten most influential parameters from OWSA, assuming 50 % standard errors

Fig. 7

Tornado diagram of the top ten most influential parameters from OWSA, assuming 10 % standard errors