Using patient data from a randomized chronic stable angina trial  and 2 two different mapping equations [15, 16], we have demonstrated that EQ-5D health utility values decrease significantly with worsening angina frequency categorization. The two independently derived mapping equations used in this analysis provided similar results; and depending on the equation used, patients regularly experiencing angina attacks reported clinically-relevant 0.07 (for monthly) to 0.33 (for daily) unit decrements (11% to 61% relative decrements) in health utility compared to patients reporting no angina. We also demonstrated patients improving by at least one SAQAF classification or reporting at least a 20-point improvement on the SAQAF domain score (previously estimated to signify a minimally important improvement) experienced a statistically significant and clinically-relevant improvement in health utility score. Thus, the above data suggests that appropriate management of stable angina symptoms can result in important improvement in patient HrQoL. In addition, our analysis provides the needed health utility values for stable angina patients with differing frequencies of angina symptoms required to calculate quality-adjusted life-years (QALYs) in cost-effectiveness (utility) analyses .
A previous analysis by the MERLIN –TIMI 36 investigators [16, 18] (published in abstract form only) has also reported health utility values based upon the same SAQAF domain score categories we used. As in ours, this analysis demonstrated a strong and statistically significant (p < 0.001) association between angina frequency and health utility (“no” = 0.96; “monthly” = 0.81; “weekly” = 0.72; and “daily” = 0.65). Of note, unlike our own analysis, the MERLIN trial elicited health utility values by administering the EQ-5D tool to a large number of subjects (n = 5,388) 4-months after randomization. However, since MERLIN only included patients within 48-hours of a non-ST-segment elevation acute coronary syndrome, the reported health utility values may not fully represent those of a stable angina population . Therefore, our analysis adds important information to current body of literature.
The two mapping equations [15, 16] we used in our study to estimate health utility values had some important differences worthy of discussion. While both equations used SAQ domain scores to estimate EQ-5D health utility values; the equation by Wijeysundera and colleagues utilized all 5 SAQ domains, while the equation by Goldsmith and colleagues used only 3 (angina frequency, physical limitation and disease perception). Next, the equation by Goldsmith included demographic variables such as age, gender and use of PCI and CABG along with SAQ domains. By using this additional information, they were able to develop an equation that explained/predicted a greater proportion of the total variation in EQ-5D health utility scores evidenced by its higher adjusted R2 compared to Wijeysundera. However, a potential downside of including this data is that researchers wanting to utilize a mapping equation may not have access to some or all of these demographic variables. Finally, while the equation by Wijeysundera used the US algorithm to score the EQ-5D, Goldsmith used the UK scoring algorithm. Given this is a multi-national trial, neither equation is preferable; however, it is important to note that the scoring algorithms result in different potential ranges of values (−0.11 to 1.0 for the US and −0.594 to 1.0 for the UK equation) [14, 15]. It is likely that each of the above-mentioned differences between the two mapping equations contributed to the differing EQ-5D health utility estimates arrived at by the two equations in our study.
There are some limitations to our analysis worth further discussion. First, since the SAQ domain scores required for mapping came from a single, moderately-sized randomized trial  that initially enrolled patients experiencing a relatively high frequency of angina and then treated these patients with an effective antianginal (ie, ranolazine), few patients finished the trial in the “no” or “daily” angina frequency categories. As a result, our estimates of health utility in these categories are associated with greater variance than the estimates in the ”monthly” and “weekly” groups with larger sample sizes. Next, as highlighted in the NICE guidance document , mapping is “at best, a second-best solution” to the direct collection of EQ-5D health utility values. However, in order to conduct a thorough cost-effectiveness analysis of stable angina interventions, an assessment of patient health utility using matching angina frequency groupings are required . Unfortunately, there is a paucity of health utility data for discrete angina frequency categories in patients with chronic stable angina, and the ERICA trial did not utilize the EQ-5D or similar tool to elicit them directly. Consequently, our data likely represents some of the best estimates currently available. Finally, to address the potential short-coming of using any one mapping equation, we used multiple equations in this analysis to estimate a range of potential values.