ECONOMIC EVALUATION OF COMBINATION TREATMENTS – DOES USE OF NETWORK META-ANALYSIS MAKE A DIFFERENCE?

Purpose:

The impact of using network meta-analysis (NMA) instead of conventional meta-analysis (CMA) for estimating comparative effectiveness is not yet fully explored in health economic evaluations. We aimed to compare the cost-effectiveness of antihypertensive drugs, both monotherapy and combination therapy, using CMA and NMA.

Method:

We used NorCaD (Norwegian Cardiovascular Disease model) to compare five different classes of antihypertensive drugs (calcium channel blocker, thiazide, ACE inhibitor, angiotensin receptor blocker (ARB) and beta blocker). We compared no therapy with monotherapies and various combination therapies of two or three drugs (in total 26 different alternatives). Analyses were performed for both women and men.

In the CMA analysis we made an assumption of multiplicativity between relative risks when estimating the effectiveness of combination therapy. This approach creates problems when using several direct comparison CMAs, for instance; (1) for each comparison where no direct evidence is available, there are different possible alternatives of combinations; and (2) when combining two or three treatments, the effect estimate can be different if different direct CMAs are combined. Both these problems are easily and intuitively solved using NMA.

We ran the NorCaD model using CMA and NMA estimates for effectiveness, but otherwise used identical parameter values. The results were expressed as net health benefit (NHB) assuming a national “reference value” for cost-effectiveness threshold of $89,000 per life year gained.

Result:

For monotherapy, NHB was consistently lower with analyses based on NMA compared to CMA, but the differences were small, ranging from -0.3% to -1.5%. The ranking of drug classes with respect to cost-effectiveness did not change, neither for men, nor women. 

For combination therapy, the differences in NHB were also small (range: from -0.2% to -1.9%), but the ranking of the 20 different combinations varied slightly. 

Conclusion:

Although NMA has obvious advantages over CMA when modelling effectiveness in economic evaluations, our analyses indicated only small differences in cost-effectiveness.