2019 Volume 83 Issue 5 Pages 1006-1010
Background: In Japan, the burden associated with myocardial infarction (MI) and ischemic stroke as well as atherosclerotic cardiovascular disease (ASCVD) generally is high. One key element in measuring disease burden is years of life lost (YLL). The aim of this study was to understand the burden of these diseases by estimating YLL at an overall disease level and average person-YLL (PYLL), a measure of disease burden also used in prior studies.
Methods and Results: Because calculation of YLL and PYLL involves inputs such as disease prevalence, disease-related mortality by age, and general population mortality by age and sex, we searched public databases of disease surveillance to identify comprehensive Japanese-specific inputs. For our reference analysis, disease-specific prevalence and mortality were taken from the Institute for Health Metrics and Evaluation Global Disease Burden study, and background mortality data were from the Japanese life tables published by the Ministry of Health, Labour and Welfare. The average age of patients with MI or stroke was 74 and 70 years, respectively. On average, men comprised 59% and 54% of the MI and stroke population, respectively. The disease-level burden of ASCVD (inclusive of MI, stroke, and peripheral artery disease) was 2,703,711 YLL in 2017. The patient-level burden was 11.99 PYLL for MI and 9.39 PYLL for stroke.
Conclusions: The burden of ASCVD, MI, and stroke in terms of premature mortality is substantial in Japan, both on a population disease level and an individual patient level.
Atherosclerotic cardiovascular disease (ASCVD), which includes myocardial infarction (MI), ischemic stroke, and peripheral artery disease (PAD), is among the most significant health problems affecting the Japanese population. According to the 2016 Institute for Health Metrics and Evaluation (IHME) Global Burden of Disease (GBD) data, cerebrovascular disease (which includes ischemic stroke) and ischemic heart disease were the second and third leading causes of mortality in Japan, respectively.1 An estimated 370,857 deaths were attributable to cardiovascular disease in 2016, which represents 28% of all deaths in Japan. Further, ischemic heart disease was the third leading cause of disability-adjusted life-years in Japan in 2010, and several modifiable lifestyle risk factors were among the top 10 factors contributing to disability-adjusted life-years.2 One study of an insured Japanese population <65 years of age estimated the prevalence of ASCVD to be 2,016 per 100,000 men and 1,718 per 100,000 women, which was associated with annual mean health-care costs (in 2012 US dollars) of $7,031±$14,359 per patient with ASCVD.3
Editorial p 965
Total burden of disease can be defined as the overall impact on individual patients and society in terms of the clinical, economic, and humanistic aspects. Although difficult to quantify, the emotional aspects of caregiver burden, including depression and grief, are other aspects that contribute to the overall burden of non-communicable diseases with high premature mortality rates such as ASCVD and cancer.4,5 Burden of illness is an important indicator of the unmet need associated with a disease, and can be an important factor for public health officials and clinicians responsible for patient care. Estimating the years of life lost (YLL) has been highlighted as a key measure of total population burden for cancer.6 Components of YLL estimation include disease prevalence, disease mortality by age or age group, and general population mortality by age or age group.
Despite the availability of information on some aspects of the burden of ASCVD in Japan, there is an unmet need to characterize this burden in terms of average per-patient and total YLL at a disease level using a clear set of Japanese-specific metrics. Although YLL and person-years of life lost (PYLL) are generally accepted metrics of disease burden, the use of well-defined inputs and reproducible calculations for YLL and PYLL is important, because variability in these inputs could limit comparisons of different estimates within a particular disease as well as between diseases.
The aim of the project was therefore to understand the disease burden of MI and stroke, as well as ASCVD reflecting MI, stroke, and PAD, in Japan by establishing robust estimates of YLL at the societal disease level and PYLL at an individual patient level. YLL, PYLL, and YLL rate were also estimated for cancer.
Analysis objectives involved secondary research methods to compile Japanese-specific data for disease prevalence and mortality as well as age- and sex-specific remaining life expectancy in the general Japanese population. Then YLL were calculated as a total for all people with the disease (i.e., societal burden), burden per person with the disease (i.e., PYLL), and as a rate per 100,000 population in Japan. PYLL calculations are weighted for age and sex; therefore, the total PYLL may not necessarily be between the total values for male and female patients. Details of the calculations are provided in Supplementary Appendix 1. Age- and sex-adjusted total PYLL, YLL, and YLL rate were estimated separately for ASCVD, MI, and stroke and were also projected over 5 years using population growth estimates to better understand the dynamics of the disease burden for each condition. For comparison, YLL, PYLL, and YLL rate were also estimated for cancer.
Disease-specific prevalence and mortality for ASCVD-related conditions were taken from the GBD study; at the time our analysis was conducted, the most recent GBD data were for 2015.7 The GBD study is a program across multiple institutions aiming to measure and quantify the magnitude of health loss due to disease, injury, and risk factors by age, sex, and geography for specific points in time. Previously published methods from the GBD study were used to produce estimates fitting the diseases of interest within this study.8 Specifically, the ischemic heart disease, ischemic stroke, and PAD subcomponents of the cardiovascular diseases category were combined as a proxy for ASCVD; MI due to ischemic heart disease and due to ischemic stroke were also analyzed separately to estimate the burden of these individual components of ASCVD. The neoplasms code was used for cancer calculations. For each of the specific disease groups, data were extracted from the IHME’s online platform (2015) for the following fields: location (Japan), year (2015, most recent year at the time of study), age, sex (male and female), and measurement (deaths, prevalence, YLL, and YLL rate).7
Background mortality data came primarily from the 22nd Japan life table published by the Ministry of Health, Labour and Welfare (Table 1), which was the base case for our calculations, and estimates the remaining life expectancy for a person of a given age and sex specifically for the Japanese population.9
| Age (years) | Life expectancy (years) | |
|---|---|---|
| Men | Women | |
| 1–4 | 79 | 85 |
| 5–9 | 74 | 80 |
| 10–14 | 69 | 75 |
| 15–19 | 64 | 70 |
| 20–24 | 59 | 65 |
| 25–29 | 54 | 60 |
| 30–34 | 50 | 56 |
| 35–39 | 45 | 51 |
| 40–44 | 40 | 46 |
| 45–49 | 35 | 41 |
| 50–54 | 31 | 36 |
| 55–59 | 26 | 32 |
| 60–64 | 22 | 27 |
| 65–69 | 18 | 23 |
| 70–74 | 14 | 18 |
| 75–79 | 11 | 14 |
| ≥80 | 6 | 8 |
Population projections were used to extrapolate future general population size in Japan,10 for which there was negative average population growth of 0.1% per year. Assuming, for simplicity, no dynamic changes in disease incidence, prevalence, or mortality, these projections were also used to extrapolate the respective components of the YLL calculations through the year 2021.
Sensitivity AnalysesVariations in the reference life table, per the method used in the GBD study11,12 and cause-specific decomposition methods used by Beltrán-Sánchez et al,13 were also used in sensitivity analyses (Supplementary Appendix 2).
The average age of Japanese patients with MI or stroke was 74 and 70 years, respectively. The ratio of men to women with MI and stroke varied by age, but on average, men comprised 59% of patients with MI and 54% of those with stroke.
ASCVDWith an estimated ASCVD population of 8,128,004 in Japan, total YLL for ASCVD were estimated to be 2,703,711 in 2017 (Table 2). Using the Japan life table as a reference, PYLL were 10.85 for an individual with ASCVD. Note that PYLL do not change over time, because the same negative growth is attributed to population and deaths. In reality, we would expect a dynamic mortality rate over time, driven by a range of potentially changing factors such as earlier diagnosis and intervention, improved standards of care, improved background mortality, improved health and lifestyle, and changes in other risk factors.
| 2017 | 2018 | 2019 | 2020 | 2021 | |
|---|---|---|---|---|---|
| ASCVD population | 8,128,004 | 8,106,376 | 8,084,806 | 8,063,293 | 8,041,837 |
| PYLL | 10.85 | 10.85 | 10.85 | 10.85 | 10.85 |
| YLL rate† | 2,145.03 | 2,150.76 | 2,156.49 | 2,162.25 | 2,168.02 |
| Total YLL | 2,703,711 | 2,696,516 | 2,689,341 | 2,682,185 | 2,675,048 |
†YLL rate=rate per 100,000 population [total YLL÷(total population÷100,000)]. ASCVD, atherosclerotic cardiovascular disease; PYLL, person-years of life lost; YLL, years of life lost.
The total YLL due to MI in Japan were estimated to be 1,785,800 in 2017, and a patient with MI was expected to lose an average of 11.99 years of life (Table 3). The PYLL for MI were greater than that for ASCVD; however total YLL for MI were lower due to lower disease prevalence.
| 2017 | 2018 | 2019 | 2020 | 2021 | |
|---|---|---|---|---|---|
| MI population | 2,094,348 | 2,088,775 | 2,083,217 | 2,077,674 | 2,072,145 |
| PYLL | 11.99 | 11.99 | 11.99 | 11.99 | 11.99 |
| YLL rate† | 141.68 | 142.06 | 142.44 | 142.82 | 143.20 |
| Total YLL | 1,785,800 | 1,781,048 | 1,776,308 | 1,771,582 | 1,766,868 |
†YLL rate=rate per 100,000 population [total YLL÷(total population÷100,000)]. MI, myocardial infarction; PYLL, person-years of life lost; YLL, years of life lost.
For stroke, the total YLL in 2017 were estimated to be 907,443, and a patient with stroke was expected to lose an average of 9.39 years of life (Table 4).
| 2017 | 2018 | 2019 | 2020 | 2021 | |
|---|---|---|---|---|---|
| Stroke population | 853,133 | 850,863 | 848,598 | 846,340 | 844,088 |
| PYLL | 9.39 | 9.39 | 9.39 | 9.39 | 9.39 |
| YLL rate† | 71.99 | 72.19 | 72.38 | 72.57 | 72.76 |
| Total YLL | 907,443 | 905,028 | 902,620 | 900,218 | 897,823 |
†YLL rate=rate per 100,000 population [total YLL÷(total population÷100,000)]. PYLL, person-years of life lost; YLL, years of life lost.
For comparative purposes, Table 5 presents the estimated PYLL, YLL rate, and total YLL for the overall cancer population in Japan from 2017 to 2021. Although the total YLL were greater than that for ASCVD because of the higher prevalence of cancer, the average PYLL for a patient with cancer (14.88) were comparable to that for the cardiovascular conditions studied.
| 2017 | 2018 | 2019 | 2020 | 2021 | |
|---|---|---|---|---|---|
| All cancer population | 5,894,339 | 5,878,654 | 5,863,012 | 5,847,411 | 5,831,851 |
| PYLL | 14.88 | 14.88 | 14.88 | 14.88 | 14.88 |
| YLL rate† | 4,660.27 | 4,672.71 | 4,685.17 | 4,697.67 | 4,710.21 |
| Total YLL | 5,874,052 | 5,858,421 | 5,842,833 | 5,827,285 | 5,811,779 |
†YLL rate=rate per 100,000 population [total YLL÷(total population÷100,000)]. PYLL, person-years of life lost; YLL, years of life lost.
The per-patient burden of premature death due to ASCVD and MI was numerically higher for men than women, whereas PYLL for cancer were higher in women than men (Table 6). For stroke, PYLL were similar between men and women.
| Disease | PYLL | |
|---|---|---|
| Men | Women | |
| ASCVD | 10.85 | 9.51 |
| MI | 12.04 | 10.02 |
| Stroke | 8.85 | 8.75 |
| All cancer | 13.43 | 15.26 |
ASCVD, atherosclerotic cardiovascular disease; MI, myocardial infarction; PYLL, person-years of life lost.
Premature mortality is the most direct humanistic burden experienced by patients with a specific disease, and premature mortality can also be estimated on a societal level using population-specific disease prevalence and mortality inputs. In the present study, we sought to provide robust estimates of YLL and PYLL for ASCVD, MI, and stroke in Japan. The calculations were based on a method that used published GBD data for prevalence and disease-specific mortality, similar to prior studies; but unlike other studies, the present calculations used the most recent Japanese-specific age- and sex-adjusted background life expectancy data.9 We found the patient-level burden of ASCVD to be greatest for individuals with MI (PYLL, 11.99), whereas the population-level burden was greatest for ASCVD overall because of the higher prevalence (total YLL, 2,703,711 for ASCVD vs. 1,785,800 for MI). Although the total YLL and PYLL for stroke were slightly lower (total YLL, 907,443; PYLL, 9.39), the burden was still substantial.
Providing clear calculations of ASCVD burden data in terms of YLL and PYLL potentially allows further context for comparison vs. other chronic high-burden diseases such as cancer. Our results for cancer burden are consistent with data from a real-world observational cohort study of YLL due to cancer in Japan, which also used Japanese-specific life table data. Pham et al reported average PYLL for all cancers of 13.6 for men and 17.5 for women,14 whereas we calculated PYLL for cancer of 13.43 and 15.26, respectively. Although the YLL for ASCVD in Japan were lower than those for cancer, the YLL for MI were ≥11 in the present population, suggesting a substantial disease burden.
Because YLL can be used as a common measure across countries, populations, and diseases and can accommodate different life expectancies between countries, ages, and sexes,6 they are more sensitive and informative than crude or even age- or sex-adjusted mortality rates. According to the present sensitivity analyses, however, some reference data, mainly the background life-expectancy tables, can significantly affect YLL estimations (Supplementary Appendix 2). Moreover, variation in YLL calculations exists in the published literature, and lack of transparency in other studies makes it challenging to understand exactly the alternative data being used.15,16 These factors affect the ability to make comparisons between studies; therefore, a strength of our analysis was the use of Japanese-specific age- and sex-adjusted background mortality rates, which provided robust, transparent, and reproducible estimates of disease burden for ASCVD relative to the general population.
Although our primary interest in this study was MI and stroke, ASCVD includes PAD in its general definition and analysis of PAD specifically may also be of value when considering the burden of cardiovascular disease in all component conditions. Additionally, although we used YLL because it is a commonly used measure of disease burden, the IHME GBD database also contains data on disability-adjusted life-years, another measure of disease burden, albeit one not as commonly used. Modeling and presenting PAD and disability-adjusted life years are future research directions that could be of value.
Study LimitationsLimitations to the present analysis revolve around 2 assumptions required for YLL calculations. One assumption is that all disease epidemiology, including incidence and mortality, is static over time and that the results are affected only by population growth. This assumption is necessary due to (1) relatively weak data to use for predictive modeling; and (2) the complexity and increased uncertainty in any predictive methods. A second assumption is that life expectancy uses general life tables, as opposed to a disease-specific one that would show changing life expectancy, over time, due to ASCVD specifically rather than all-cause mortality. Both of these limitations are considered to be minor, and the assumptions remain consistent with general methodology for YLL calculations. Another limitation relates to the points of comparison. Other studies of YLL in ASCVD in Japan and the US have used varying data sources and methodologies; some of these methods had to be determined by retro-engineering the results, whereas others were clear but did not match the analyses and methods here. Therefore, it is difficult to establish a common comparator for the results.
When calculated in terms of total YLL, the societal burden of ASCVD-related premature mortality in Japan is significant. For a patient with ASCVD, particularly MI, the burden is comparable to cancer. These robust and reproducible estimates of YLL and PYLL enable a better understanding of the burden of ASCVD in Japan, including the burden specifically attributable to MI and ischemic stroke.
The authors thank Cathryn M. Carter, MS (Amgen Inc.), for medical writing support. She received compensation as an employee of Amgen Inc. This study was sponsored by Amgen Inc. The sponsor contributed to the study design, interpretation of data, and decision to submit the article for publication.
Qualified researchers may request data from Amgen clinical studies. Complete details are available at the following:
http://www.amgen.com/datasharing
Dr. Arai has received remuneration from Astellas, Astellas Amgen Biopharma, Sanofi, Daiichi-Sankyo, Kowa, MSD, and Chugai; Dr. Mortaki is an employee of PRMA Consulting, which was under contract with Amgen to conduct this study; Dr. Quinn is an employee of PRMA Consulting, which was under contract with Amgen to conduct this study, and receives consultancy income from Health Technology and Policy Evaluation, LLC; Dr. Rane, Dr. Zhao, and Dr. Qian are employees of Amgen Inc. and hold Amgen stock.
This study was funded by Amgen Inc., Thousand Oaks, CA, USA.
Please find supplementary file(s);
http://dx.doi.org/10.1253/circj.CJ-18-1216
