Should we routinely measure patient quality of life after acute coronary syndrome?

This editorial refers to the article ‛Association of baseline and changes in health-related quality of life with mortality following myocardial infarction: multicentre longitudinal linked cohort study’ published in the European Heart Journal—Quality of Care and Clinical Outcomes by T. Dondo et al. https://doi.org/10.1093/ehjqcco/qcae036. The relative value of patient reported outcome (PRO) measures for predicting health outcomes and shared decision-making has been a topic of debate, although evidence in favour of PROs has been growing across a range of conditions.1,2 Nearly 30 years ago, the Studies of Left Ventricular Dysfunction (SOLVD) Investigators reported that quality of life (QoL) at baseline in 5025 patients with heart failure had a strong independent association with mortality and hospitalization over a 3-year follow-up period.3 The large sample size allowed sufficient power to reliably detect associations between QoL and clinical outcomes. The SOLVD Investigators concluded that QoL measurement provided additional prognostic information beyond traditional demographic and clinical covariates (e.g. age, gender, and prior medical history) and could be useful in clinical practice. In the EMMACE-3 and 4 observational studies, Dondo et al. reported results from 9474 patients with acute coronary syndromes (ACS) from 77 UK centres from 2011 to 2015 with a median of 7 years of follow-up.4 Conventional patient characteristics and outcomes were recorded as well as QoL using the EuroQol (EQ5D) and associated visual analogue scale (VAS) at baseline, 30 days, 6 months, and 12 months. The mean age was 64 years, 25% women, and 41% had ST segment elevation myocardial infarction (STEMI). Guideline-directed medical therapy was taken up by >90% of patients. Patients with baseline QoL in the lowest tertile were more frequently women, and had less STEMI and more cardiovascular comorbidities and asthma/chronic obstructive pulmonary disease. Approximately half of the patients at baseline did not contribute QoL data at 12 months, although about 5% of missing QoL will be related to deaths at that time point. Multilevel survival modelling adjusting for common covariates known to influence outcomes was used to explore associations between baseline QoL and survival, and change in QoL, from baseline to 12 months and longer-term mortality. Missing variables were imputed. At baseline, QoL (median EQ-5D = 0.81 and VAS = 0.70) was reduced compared with national age and sex match cohorts and increased steadily at a mean of 0.003 per month up to 12 months but remained reduced compared with the reference values. After about 7 years of follow-up, overall mortality rates were 22.5%, with highest vs. lowest tertile EQ-5D groups having unadjusted mortality rates of 28% and 15.3% (P < 0.001), respectively. Increased EQ-5D values were associated with reduced mortality, but this was not significant for the VAS. The functional and pain/anxiety domains of EQ-5D showed similar results to the overall values, but the anxiety/depression component did not appear to have a strong association with survival. Improvement in EQ-5D from baseline to 12 months was associated with better long-term survival, with about a one-third reduction in risk compared with no improvement, but again this was not demonstrated for the VAS.4 So how does the analysis by Dondo et al. add to our knowledge? First, they have used a well-constructed prospectively collected observational database from the high-quality EMMACE studies describing the management and outcomes of ACS with reliable data on survival up to 7 years. Second, the analysis has sufficient power to provide reliable evidence of the association between baseline QoL and survival in patients with ACS, including the specific domains of the EQ-5D. The VAS component of the QoL assessment appeared to be less reliable in predicting survival than the EQ-5D. Third, this is probably the most reliable study so far to document changes in QoL in ACS patients and confirm that overall improvements over a 12-month period, as well as improvements in the specific domains, are associated with improved survival up to 7 years. Additional strengths of the study are the clear presentation of results and the focus on mortality rather than other less reliable outcome measures. There are some areas that could be clarified for readers. The proportion of patients contributing to 12-month QoL data is about half of the population at baseline and reasons for this do not appear to be provided. It would be helpful to know whether the change from baseline analysis included just the paired available results from baseline and 12 months or whether there was extensive imputation of QoL at 12 months to include more patients, which could have influenced the findings. The association between changes in QoL from baseline to 12 months and longer-term mortality can only be applied to those who survived up to 12 months after ACS, which is a limitation. The multilevel survival modelling mixed baseline and non-baseline factors such as discharge medications, which could introduce bias into the results, and some discussion of this methodology would be helpful. The QoL data are presented as both means and medians, without a clear rationale as to why one or the other was used in the analysis, which was confusing. QoL has been associated with outcomes across several cardiovascular conditions, including heart failure,3 percutaneous coronary intervention,5 coronary artery disease,6 and coronary artery bypass graft (CABG).7 The present analysis by Dondo et al. has extended these observations by showing that changes in QoL after ACS are associated with longer term survival. In general, sample sizes of a few thousand observations are needed to help ensure reliability and validity of QoL scores, which is much larger than seen in most studies. Based on these findings, and the extensive literature showing strong associations of QoL with outcomes in patients with cardiovascular disease, we believe that it is now time to routinely institute QoL assessments, such as the EQ-5D, into clinical practice to support shared decision-making. There are barriers to the adoption of PROs into clinical practice, including licensing requirements, support for patients to complete PROs, use of different languages, training of health professionals to interpret results, and the need to place PROs on existing electronic health care platforms all of which require additional resources.8 Ideally trained facilitators would be available to address these barriers and help ensure quality until PRO measurement and interpretation became routine. Use of PROs is likely to provide key patient input into clinical decision-making and may avoid the need for follow-up if QoL scores are high or improving. Clearly, QoL scores cannot be used in isolation but can provide patients and clinicians with extra confidence about progress and management. At the present time, there is little evidence that PROs are being used to support routine clinical decision-making in cardiovascular care.8 The full impact of measuring and improving QoL is unknown, but present evidence indicates that this is a huge untapped resource for patient benefit and participation in their own health care, which will accrue over time. What is clear is that PROs now need to be introduced into routine care.