Abstract
Objectives: The effects of acute coronary syndrome (ACS) events on health-related quality-of-life (HRQoL) and the time dependency of these effects are unknown. This study aimed to characterize health utilities in ACS patients to aid development of future economic models estimating the cost per quality-adjusted life-year impact of ACS events and potential treatments.
Methods: Multi-center, non-interventional, longitudinal evaluation of health utility in patients experiencing ACS or stroke events. EuroQol-5 dimension 3 level (EQ-5D-3L) surveys were sent to patients (≥18 years) from three UK centers, 1 month after hospital discharge for myocardial infarction (MI), unstable angina (UA), or stroke. Patient demographics, lifestyle, and baseline utility score were collected in the first survey. Follow-up surveys were sent at 6, 12, 18, and 24 months to prospectively capture utility and subsequent health events. Two methods of patient identification were adopted—prospective, where the patient’s qualifying events occurred after the study index date, and retrospective, where the patient’s qualifying event occurred prior to the study index date. General healthy population utility values were assumed for pre-event HRQoL.
Results: Between January 2011 and March 2014, 2,103 prospectively (n = 1,350)/retrospectively (n = 753) identified patients (mean age = 68.3 years; 67.9% male) responded: MI = 55.9% (n = 1,176), UA = 42.7% (n = 898), stroke = 1.4% (n = 29); 24% had type 2 diabetes. Post-event utility values were lower than general healthy population values, although significant differences in utility between subsequent 6 (n = 1,031, change = –0.002), 12 (n = 1,096, change = –0.008), 18 (n = 1,246, change = –0.007), and 24 (n = 1,277, change = –0.004) month timepoints were not detected. Through multivariate regression analyses, wheelchair use, current smoking, and secondary mental and joint health events were associated with the greatest statistically significant utility decrements.
Conclusions: This study indicates that health utility decreases following a cardiovascular event and, although some improvement occurs over the subsequent 24 months, general healthy population utility is not necessarily attained.
Introduction
Cardiovascular (CV) disease causes substantial morbidity and mortality worldwide, being the primary cause of non-communicable disease-related death in the US and EuropeCitation1–3. Acute coronary syndrome (ACS) is an umbrella term for clinical conditions characterized by signs and symptoms of myocardial ischemia. ACS includes unstable angina (UA), ST segment elevation myocardial infarction (MI), and non-ST elevation MI, differentiated by ECG and biomarker (troponin) changes. Following an initial ACS event, the likelihood of a subsequent event, with its associated morbidity and mortality, increasesCitation4,Citation5. Furthermore, such events are associated with more chronic conditions such as heart failure and strokes (REACH registry), long-term disability, and resulting costs to individual patients and society as a wholeCitation6,Citation7.
Patient-reported health status measures are often used as end-points in clinical trials, although they are rarely used in the real-world settingCitation8. As such, the published data often reflects a healthier clinical trial eligible patient cohort, and does not reflect the real-world cohort fully. Furthermore, most published utilities represent post-event health states without distinguishing between the acute impact of the CV event and the chronic post-event effectsCitation9. Consequently, the effects of initial and subsequent ACS events on health-related quality-of-life (HRQoL) and the variables that may contribute to the magnitude of these effects are not well characterized or referenced. For health technology appraisal bodies in the UK, characterization of health utilities in patients with ACS helps when estimating the economic impact of both ACS events and potential treatments to mitigate these events.
The objectives of this study were to quantify: (1) The utility associated with a CV event; and (2) The change (decrement or increment) in utility associated with a CV event over a time frame of 24 months.
Materials and methods
Study design
The study was a multi-center, non-interventional, longitudinal evaluation of health utility in patients who had experienced an ACS or stroke event. Patients were surveyed for a maximum of 2 years from their discharge date following their qualifying ACS or stroke event. The main study end-point was health utility index value determined from the EuroQoL 5-Dimensions 3 level Questionnaire (EQ-5D-3L).
Setting
Three National Health Service (NHS) trust hospitals (Barnet Hospital, Royal Free London NHS Foundation Trust; University Hospital of Wales, Cardiff and Vale University Health Board; and Peterborough City Hospital, Peterborough and Stamford Hospitals NHS Foundation Trust) with well-characterized CV patient populations participated in the study. In the national MINAP data-set of management of ACS patients, these centers were all normal or above normal in performance with evidence-based therapies.
Patients
Patients experiencing an ACS event or a stroke following an ACS event were eligible for inclusion. Potential study participants were identified from participating site health records and, 1 month post-discharge following a qualifying CV event (MI, UA, or stroke), were mailed an informed consent form and an initial study survey.
Patients had to meet each of the following criteria to be considered eligible for inclusion in the study: age ≥18 years; experiencing a qualifying event of ACS (MI or UA) or stroke (defined as: stroke with a prior history of CV disease and with their most recent previous event prior to their qualifying stroke being an ACS event). Stroke and ACS patients were sought to understand the impact of stroke deliberately, particularly as ACS patients often do not survive a subsequent stroke event and, therefore, are usually unavailable to contribute.
Patients were excluded if they met any of the following criteria: stroke but no history of CV disease; stroke and prior history of CV disease, but their most recent CV event was not an ACS event (i.e. stroke, stable angina, myocarditis, etc…); prior diagnosis of type 1 diabetes mellitus; prior elective or emergency coronary revascularization or coronary artery bypass graft surgery in the 6 weeks before their subsequent CV event. Patients who had undergone revascularization more than 6 weeks before their CV event were assumed to have stabilized, and were considered eligible for the study.
Patients were identified by two methods; (1) prospectively, when the patient’s qualifying event occurred after the start of the study; or (2) retrospectively (Cardiff only) when the patient’s qualifying event had occurred before the start of the study. The inclusion of both prospectively and retrospectively identified patients enabled the collection of both early-term and late-term time-dependent utility data from the start of the study. Inclusion of the retrospectively identified cohort also ensured adequate sampling of later timepoints, when patients in the prospectively identified group may have been lost to follow-up.
Evaluations
HRQoL was measured using the EuroQol five dimension–three level (EQ-5D-3L) outcome measureCitation10, a standardized measure of health status that has been validated in a wide range of health conditions and as a general population health measure in the UK and many other countriesCitation11, and utilizing the UK value setCitation12. The visual analog scale (VAS) part of the EQ-5D was not utilized, as the study was designed to generate utility scores to inform economic models; as VAS scores are not used in this process they were deemed an unnecessary burden to the patients.
The initial study survey included a demographic questionnaire; health status, as measured by the EQ-5D (mobility, self-care, usual activities, pain/discomfort, and anxiety/depression), and a question regarding current wheelchair use. Patients who provided signed informed consent and completed the initial survey were considered study responders. Study responders were mailed up to five surveys (one initial survey and four follow-up surveys at 6, 12, 18, and 24 months after their qualifying event). The follow-up surveys consisted of the EQ-5D, the current wheelchair-use query, and two additional questions regarding the occurrence and type of any injury or illness that may have affected their health since they had completed the preceding study survey.
The retrospectively identified patients were sent informed consent forms and surveys (initial and follow-up) consistent with the process established for the prospectively identified patients, except that that they could only receive a maximum of four surveys (at 6-, 12-, 18-, and 24-month follow-ups), depending on the time elapsed since their qualifying CV event had occurred.
Accrual period and sample size
The study accrual period (identification of potential study subjects, informed consent form, and initial survey mailings) began in January 2011, following activation of the first study site on January 5, 2011, and continued until May 2012. Data collection continued until March 2014, when the last follow-up patient questionnaire was returned.
Data sources and measurement
The majority of study variables were assessed by patient-completed questionnaires. The data captured included baseline characteristics, HRQoL, and the occurrence and type of secondary illness or injury affecting health between surveys. Patient questionnaires were anonymized: all included a unique identifier on each page that linked the questionnaire to each patient’s unique participant code. Type of qualifying event (MI, UA, or stroke) and subsequent in-hospital death during the study period were recorded by the site investigator based on institutional health records (it was not possible to capture deaths in the community, prior to study entry).
At the Barnet and Peterborough study sites, questionnaire data and investigator-reported variables were entered into a Microsoft Excel spreadsheet by investigator staff. To allow for interim data reviews and provide data back-ups, copies of the spreadsheets were regularly transferred from the study sites during the study period. Spreadsheet data were uploaded into a site-specific Microsoft Access database; surveys were mailed via commercial courier.
Completed questionnaires from patients enrolled at the Cardiff study site were scanned into the site-specific Access database using optical character recognition software. Qualifying CV event type and death during the study period were recorded by investigator staff in a text file provided for uploading into the site-specific Access database linked to the scanned survey data. A separate study database was maintained for each study site; data from all sites were amalgamated only for study analyses. Routine quality assurance checking was undertaken by two members of the analysis team.
Sample size and statistical considerations
It was determined that a population of 669 patients would be needed to attain a power of 80% at a 5% α level to detect a difference in EQ-5D score of 0.03. A difference of 0.03 in EQ-5D score is consistent with previously reported minimum clinically meaningful differences in health utilityCitation13. Loss to follow-up over the study duration was estimated to be 20% at 6 months (i.e. the second survey) and maintained for the duration of the study.
Baseline characteristics, qualifying event type, death (if noted in health records), and patient-reported causes of change in health during the study period were tabulated and analyzed using simple summary statistics (proportions, means, standard deviations [SDs], and ranges).
HRQoL was assessed using linear regression analysis to examine change in utility between timepoints and by multivariable linear regression analysis stratified by qualifying CV event and controlling for variables (i.e. age, sex, weight, type 2 diabetes mellitus [T2DM] status, smoking status, wheelchair use, and causes of changes in health) and interactions between variables. The calculated β, standard error, 95% confidence intervals, and p-values were tabulated, and the utility decrements associated with baseline characteristics and patient-reported causes of change in health were calculated and presented by qualifying event.
Regression analyses were conducted by patient group (prospectively identified patients, retrospectively identified patients, and combined patient groups) and survey time point. EQ-5D was used as the dependent variable, with all other variables entered as independent variables. Backward elimination stepwise regression models were utilized with non-statistically significant variables removed one at a time; for consistency, qualifying event, gender, age, and weight were forced variables entered into the model, irrespective of their statistical significance.
This study was approved by the National Institute for Social Care and Health Research (NISCHR) Research Ethics Service (RES), Rec reference: 10/WSE03/37, IRAS project ID: 61628.
Results
A total of 2,103 patients returned surveys; 1,350 (64.2%) prospectively identified patients and 753 (35.8%) retrospectively identified patients. Most patients in the prospective group were enrolled from Cardiff (48.4%), with Peterborough and Barnet recruiting 30.3% and 21.3% of patients, respectively. The overall survey response rate at baseline was 62.0%, with follow-up response rates ranging from 66.3–70.6% ().
Table 1. Survey response by time point.
Baseline characteristics
Baseline characteristics of the patient group by qualifying event are shown in . Of the 2,103 patients enrolled, 1,176 (55.9%) had an MI, 898 (42.7%) had UA, and 29 (1.4%) had a stroke. Approximately one-quarter of patients had been diagnosed with type 2 diabetes mellitus at the time of completion of their initial survey ().
Table 2. Baseline patient characteristics.
Patients reported wheelchair use at baseline and at each follow-up survey. Wheelchair use was reported in 6.3% of patients at baseline, and varied from 5.1% at month 6 to 6.2% at month 18 ().
The majority of patients enrolled (86.9%) were not current smokers at the time of completing their initial study survey. Cardiff (12.6%) had the highest proportion of smokers, followed by Peterborough (10.5%) and Barnet (6.3%).
Outcomes
Health utility: EQ-5D index scores
As shown in , mean EQ-5D scores increased from baseline to month 6, then subsequently decreased through month 12 and month 18, before increasing again at month 24.
Table 3. Mean EQ-5D index scores.
Mean EQ-5D score was lowest in the stroke sub-group, ranging from 0.448 (SD = 0.425) at month 18 to 0.527 (SD = 0.403) at month 24. Mean EQ-5D was highest in the MI group, ranging from 0.690 (SD = 0.290) at baseline to 0.708 (SD = 0.322) at month 18.
Utility over time
shows changes in utility over time. Although decreases in utility were observed between each timepoint, none of these decreases were statistically significant (p = .267–.785).
Table 4. Unadjusted change in EQ-5D score over time.
Change in utility with secondary events
The numbers of patients reporting an event that significantly affected their health, as assessed in each follow-up survey, are shown in . Joint and CV disease events were the most common event types influencing self-reported health. The ∼12% of further CV events over the 24 months of follow-up is consistent whatever the follow-up duration available. This rate may be somewhat low for an average follow-up of 24 months, compared to ACS registries such as GRACECitation4.
Table 5. Patients reporting secondary health events.
Changes in utility associated with baseline characteristics, wheelchair use, and secondary health events, as determined by multivariate regression analysis, are shown in . For discussion and presentation of regression model results, the initial survey was estimated to have been completed 1 month after the qualifying event.
Table 6. Adjusted EQ-5D utility scores.
Baseline characteristics (e.g. qualifying event, gender) were forced into the regression analysis model, regardless of their significance to evaluate their effect over time. Changes in health due to secondary events were not included in the regression analysis model if they were not statistically significant. For example, cancer and complications of diabetes did not have a statistically significant effect on utility, most likely due to the small number of occurrences (cancer) or the effect on utility (complications of diabetes) not being significantly greater than already incurred by having the disease (baseline T2DM status).
Adjusted EQ-5D scores over time for study participants whose qualifying event was an MI or UA are shown in ; stroke patients were excluded due to the small number of patients. The utility point estimated at months 6, 12, 18, and 24 are contrasted with mean utilities for individuals aged 68 years without complications derived from the English Health SurveyCitation14.
Figure 1. Adjusted EQ-5D by qualifying event over time. Results of the English Health Survey (2003 publication) for individuals aged 68 years were used as a pre-event baseline. Initial survey completion estimated to have occurred 1 month after the qualifying event. EQ-5D, EuroQoL 5-Dimensions questionnaire.
Male patients who experienced an MI had a mean utility score of 0.692 at month 1 after the event and 0.649 at month 6, before increasing to 0.727 at month 12, 0.782 at month 18, and 0.765 at month 24, all lower than the general population mean of 0.886 (). In patients with UA, mean utility was 0.649 at month 1 after the event, decreasing to 0.614 at month 6, before increasing to 0.688 at month 12 and 0.761 at month 18, before decreasing to 0.65 at month 24; again, all lower than the general population mean.
Discussion
Several utility values are available to represent health states in patients with ACSCitation15. However, many such values do not take into account the time between the CV event and the utility assessment. This limits interpretation of the impact of the event and the resulting quality-of-life implications, which can occur acutely and chronically following the event. In the current study, regression analyses indicated a modest increase utility from 1-month after the initial event (i.e. at initial survey completion) to month 24, although significant differences in utility were not detected between baseline and subsequent month 6, 18, 12, and 24 timepoints. It is noteworthy that health utility remained significantly lower than population norms for age-matched subjects without complications. Wheelchair use, current smoking, and secondary mental and joint health events were associated with the greatest utility decrements (>0.250 decrease). The effect of these baseline characteristics and secondary events were more profound than baseline T2DM status. The number of stroke patients enrolled in the study was inadequate for any conclusions to be made regarding this group.
A potential limitation of the study was that CV history (aside from the qualifying event) was not available, and that secondary or subsequent clinical events were captured via self-reporting rather than via health records. Nevertheless, the high patient response to study surveys allowed for the collection of valuable utility information which could be examined in the context of initial CV events that characterized participants’ ACS. In addition, the limited number of sites could introduce characteristic biases, such as the smoking rate differences seen. We have not looked for or adjusted for social class demographic differences between the sites, but these may be present given the average survival ages for the three regions is known to differ. Study patient death was likely under-reported, as only those deaths documented in hospital records were included. Furthermore, patients whose poor health or undocumented death precluded survey completion would have inappropriately biased results toward better utility. The study did not ask investigators to chase patient status if questionnaires were unreturned.
Conclusions
This real-world, survey based, study indicates that health utility decreases following a CV event, although some improvement occurs over the subsequent 24 months. If we assume a general population quality-of-life baseline from published data, this level of quality-of-life score is not approached by our study group as a whole.
Transparency
Declaration of funding
This study was sponsored by Roche Products Ltd. The non-interventional nature of the study means that it was not registered on any commonly used clinical trial databases. Third-party medical writing assistance was funded by Roche Products Ltd.
Declaration of financial/other relationships
JR and SS are employed and paid by F. Hoffmann-La Roche, Basel, Switzerland. IT is employed and paid by Roche Products Ltd, Welwyn Garden City, UK. All other authors declare no financial/other relationships. JME peer reviewers on this manuscript have no relevant financial or other relationships to disclose.
Acknowledgments
Roche thanks the Swansea Centre for Health Economics (SCHE), formerly the Cardiff Research Consortium (CRC), for contributing to the study, data analysis, and interpretation. The protocol and data analysis plan were developed jointly by the sponsor and CRC personnel with final approval by the sponsor. Support for third-party medical writing assistance was provided by SCHE and Miller Medical Communications Ltd (both funded by Roche Products Ltd).
References
- Go AS, Mozaffarian D, Roger VL, et al. Heart disease and stroke statistics – 2014 update: a report from the American Heart Association. Circulation 2014;129:e28-e292
- Nichols M, Townsend N, Luengo-Fernandez R, et al. European cardiovascular disease statistics, 2012 ed. Brussels, Belgium, Sophia Antipolis, France: European Heart Network, European Society of Cardiology; September 2012. p. 125
- World Health Organization (WHO). Global status report on noncommunicable diseases, 2010. Geneva, Switzerland: WHO Library; 2011. p. 161
- Alnasser SM, Huang W, Gore JM, et al. Late consequences of acute coronary syndromes: global registry of acute coronary events (GRACE) follow-up. Am J Med 2015;128:766-75
- Erdem G, Bakhai A, Taneja AK, et al. Rates and causes of death from non-ST elevation acute coronary syndromes: ten-year follow-up of the PRAIS-UK registry. Int J Cardiol 2013;168:490-4
- Leal J, Luengo-Fernandez R, Gray A, et al. Economic burden of cardiovascular diseases in the enlarged European Union. Eur Heart J 2006;27:1610-19
- Luengo-Fernandez R, Leal J, Gray A, et al. Cost of cardiovascular diseases in the United Kingdom. Heart 2006;92:1384-9
- Spertus JA, Jones P, McDonell M, et al. Health status predicts long-term outcome in outpatients with coronary disease. Circulation 2002;106:43-9
- Matza LS, Stewart KD, Gandra SR, et al. Acute and chronic impact of cardiovascular events on health state utilities. BMC Health Serv Res 2015;15:173
- EuroQol Group. EuroQol – a new facility for the measurement of health-related quality of life. Health Policy 1990;16:199-208
- Szende A, Janssen B, Cabases J, eds. EQ-5D value sets: inventory, comparative review and user guide. Netherlands: Springer; 2014
- Dolan P. Modeling valuations for EuroQol health states. Med Care 1997;35:1095-108
- Coretti S, Ruggeri M, McNamee P. The minimum clinically important difference for EQ-5D index: a critical review. Expert Rev Pharmacoecon Outcomes Res 2014;14:221-33
- Kind P, Dolan P, Gudex C, et al. Variations in population health status: results from a United Kingdom national questionnaire survey. BMJ 1998;316:736-41
- Smith DW, Davies EW, Wissinger E, et al. A systematic literature review of cardiovascular event utilities. Expert Rev Pharmacoecon Outcomes Res 2013;13:767-90