Impact of angina frequency on health utility values of patients with chronic stable angina

It is estimated that nearly 8 million people in the United States (US) suffer from chronic stable angina [1]. Previous studies have demonstrated chronic stable angina has a significant negative impact on health-related quality-of-life (HrQoL) due to the bodily pain, psychological distress, impaired functioning, activity restriction and inability to care for oneself that is common to the disease [2].

The Seattle Angina Questionnaire (SAQ) [3] is a coronary disease-specific patient reported outcome measure (PROM) that has been included as a key endpoint in numerous clinical trials of stable coronary disease [4‐9]. Some of these trials have published analyses evaluating morbidity, mortality and healthcare utilization across different frequencies of angina attacks; often categorizing patients as having no, monthly, weekly or daily angina symptoms based upon the angina frequency domain [8‐11]. These analyses provide valuable insight into the relationship between angina frequency and event rates, healthcare utilization and treatment costs; however, in order to conduct a thorough cost-effectiveness analysis using these data, an assessment of patient health utility [values of preference for a disease state on a scale of 1.0 (perfect health) to 0.0 (death)] using matching angina frequency groupings are required [12].

There is currently a paucity of data describing relative health utility values for persons with chronic stable angina reporting the abovementioned frequencies of angina attacks. In these situations, guidance from the National Institute of Health and Care Excellence (NICE) [13] supports the use of mapping equations to translate responses from a PROM to an appropriate health utility tool (e.g. EuroQoL [EQ]-5D) [14]. In this study, we used two previously published mapping equations to convert individual patient SAQ data from a randomized controlled trial of chronic stable angina [4] into EQ-5D health utility scores [15, 16] to evaluate the relationship between angina frequency and health utility.

Data used in this post-hoc analysis were from the multi-national, double-blind, randomized, placebo-controlled, parallel group Efficacy of Ranolazine in Chronic Angina (ERICA) trial (clinicaltrials.gov registration: NCT00091429) [4]. The ERICA trial was approved by the institutional review board at each participating hospital, and each patient provided written informed consent. ERICA evaluated the effectiveness of ranolazine (1,000 mg twice daily) in adults with a documented history of coronary disease, at least a 3-month history of chronic stable angina, and reporting ≥3 episodes of angina/week during a 2-week qualification period despite treatment with 10 mg/day of amlodipine.

The SAQ was a secondary outcome measure administered at baseline and at the end of the 6-week double-blind full-dose trial phase. The SAQ is a validated, coronary disease-specific PROM comprising 19 items that quantify 5 clinically relevant domains of health status (angina frequency, physical limitation, angina stability, treatment satisfaction and disease perception) each scored on a scale of 0 to 100, with higher scores indicating better health status [3]. For this analysis, we used the SAQ angina frequency (SAQAF) domain scores to classify patients into discrete angina frequency categories; with a score of 100 = no; 61-99 = monthly; 31-60 = weekly; 0-30 = daily angina symptoms [9, 10, 17].

Two unique and independently derived mapping equations were used to translate (or cross-walk) individual SAQ domain scores for patients in the ERICA trial to EQ-5D-based health utility scores [15, 16]. The first equation was derived in nearly 3,000 participants of the Alberta Provincial Project for Outcome Assessment in Coronary Heart Disease (APPROACH) database using linear regression within a Bayesian framework and scoring the EQ-5D using the US scoring algorithm (which ranges from −0.11 to 1.0 on a scale where 0.0 = death and 1.0 = perfect health). The equation [EQ-5D = 0.0010*(angina frequency domain score) – 0.0002*(angina stability domain score) + 0.0023 (disease perception domain score) + 0.0019 * (physical limitation domain score) + 0.0004 * (treatment satisfaction domain score) + 0.4388] uses all 5 domains of the SAQ to estimate EQ-5D health utility scores with a mean absolute error (MAE) of 0.088 (adjusted R² = 0.37). The second equation used patient-level data from 5 studies of cardiac interventions, 3 of the 5 SAQ domains along with patient demographics, ordinary least squares linear regression methods and the United Kingdom (UK) scoring algorithm for the EQ-5D (which ranges from −0.594 to 1.0 on a scale where 0.0 = death and 1.0 = perfect health) to derive the following equation: EQ-5D = 0.002*(age) - 0.009 (if male) + 0.021(if medically managed) + 0.048 (if pre-percutaneous coronary intervention (PCI)) + 0.018 (if post-PCI) + 0.073(if pre-coronary artery bypass grafting) + 0.0036 *(physical limitation domain score) + 0.0021* (disease perception domain score) + 0.0015 * (angina frequency domain score) + 0.147 to estimate EQ-5D index scores with a MAE of 0.123 (adjusted R² = 0.44).

Our analysis included all patients randomized in the ERICA trial that received at least one dose of double-blind investigational drug, had at least one subsequent primary efficacy assessment during the double-blind trial period and had a complete SAQ assessment at randomization and at the end-of-trial.

We estimated median (along with 25%, 75% ranges) EQ-5D health utility scores at randomization and at the end-of-trial for each patient using the above-mentioned mapping equations. Comparison of EQ-5D health utility scores across the four SAQ-based angina frequency classifications were made using end-of-trial values only (as no subjects were classified as having no angina symptoms and few were classified as having monthly symptoms at baseline due to trial inclusion criteria). We calculated change in EQ-5D health utility scores from randomization to end-of-trial and compared these values between subjects improving and not improving ≥1 SAQAF classification. Additionally we compared the change EQ-5D health utility scores for subjects achieving and not achieving a ≥20-point improvement in SAQAF domain score; a threshold previously reported as the minimally important clinical difference on the SAQAF domain score [11].

Categorical data were compared using chi-squared tests. Continuous data were compared using a Mann–Whitney U or Kruskal-Wallis one-way analysis of variance test, where appropriate. A p-value <0.05 was considered statistically significant in all situations. All analyses were conducted using SPSS version 20.0 (SPSS Inc., Chicago, IL, USA).

Characteristics of the 565 subjects randomized in the ERICA trial have been published previously [4]. We briefly summarize their key characteristics in Table 1.

Both SAQAF and derived EQ-5D health utility scores were available in 548 of the patients (97% of all randomized). At baseline, the total trial population reported median SAQAF domain score of 40 (40, 50) with 6%, 71% and 23% reporting monthly, weekly and daily angina symptoms, respectively. Median EQ-5D health utility scores were 0.68 (0.64, 0.73) and 0.60 (0.55, 0.66) at baseline based on Equation 1 and 2.

Median EQ-5D health utility scores for each of the a priori determined SAQAF classifications at end-of-trial are reported in Table 2. A statistically significant association was seen between angina frequency classifications and EQ-5D health utility scores; with more frequent angina resulting in decrements in health utility (p < 0.001 for association for both Equations 1 and 2).

Patients who improved at least one SAQAF category from baseline (n = 254/541, 47%) experienced a median 0.05/0.07 greater improvement in EQ-5D health utility score compared to those not improving by at least one SAQAF classification (p < 0.001 for data based on both mapping equations) (Figure 1). Patients who improved ≥20-points on their SAQAF domain score from baseline (n = 355/541, 66%) experienced a median 0.06/0.07 greater improvement in EQ-5D health utility score compared to those not achieving a ≥20-point improvement (p < 0.001 for both equations) (Figure 2).

Using patient data from a randomized chronic stable angina trial [4] 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 [12].

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 [6]. 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 [4] 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 [13], 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 [12]. 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.

Chronic angina patient health utility values decrease as angina frequency increases. Our health utility estimates for stable angina patients stratified by angina frequency may be useful for conducting cost-effectiveness analyses in the future. Patients reporting at least a clinically important improvement in angina frequency experience a tangible improvement in health utility. Due to the strong relationship between these two outcome measures, future studies of chronic stable angina interventions should include health utility measures to aid in health economic evaluation.