OBJECTIVES:

A Singapore cost-effectiveness analysis (CEA) of diabetes prevention based on the US Diabetes Prevention Program (DPP) concluded that metformin and lifestyle modifications are both cost-effective. We corrected a number of errors in that analysis, concluding that only lifestyle was cost-effective and explored various graphics, indicating the appropriate conclusion, minimizing chance of misinterpretation. We then assessed the impact of adopting metformin as cost-effective in error.

METHODS:

We corrected probabilities, QALY calculations, discounted both costs and QALYs, and corrected Incremental Cost-effectiveness Ratios (ICERs). We present results as appropriate ICERs and in various graphical forms intended to clearly differentiate optimal/non-optimal treatments and indicate costs of errors in interpretation (Loss Curves and Technical and Economic Inefficiency measures).

RESULTS:

The difference between expected values for Costs and QALYs vs. our results ranged from -11.9% to 6.6% (costs), and from -3.3% to -2.6% (QALYs). Placebo and metformin were both strongly dominated by lifestyle. Lifestyle was cost-effective at any conventional WTP. Various visual depictions clearly show this. The cost-effectiveness plane shows metformin and placebo as dominated. The cost-disutility plane shows how placebo and metformin are technically and economically inefficient (and can measure the degrees of inefficiency) and CEACs indicate low decision uncertainty. Loss Curves show losses from implementing metformin over lifestyle Net Loss Curves show losses of choosing metformin over lifestyle (at US$70,000 WTP, approximately US$2,650 per patient).

CONCLUSIONS:

When more than two alternatives are included in CEAs, ICER calculations can easily be made in error. The use of various graphics and calculation of Expected Losses has advantages for communicating CEA results, providing extra information to 1) reduce chances of misinterpreting non-optimal choices as optimal and 2) measuring the impact of such errors. The Singapore DPP CEA miscalculated ICERs and made other errors; decision makers adopting those results could impose significant costs on patients.