Application of Life Table Method for Measuring Mortality Risk of Non-communicable Diseases in Hong Kong Andrew Kai-ming KWANa,∗, Lee Yuen-hoa, Dr Eddy NGa, Dr Regina CHINGa a Centre for Health Protection, Department of Health, Hong Kong, China ∗ Corresponding author: Tel.: +852 2961 8569; Fax: +852 2893 9425; E-mail: stat_hs1@dh.gov.hk Abstracts In this paper, we apply the actuarial, or life table method to study the mortality risk between ages 30 and 70 from four non-communicable diseases namely, cardiovascular diseases, cancers, chronic respiratory diseases and diabetes (4 NCDs) from 1980 to 2012 in Hong Kong based on the methodology laid down in ‘A Comprehensive Global Monitoring Framework and Voluntary Global Targets for the Prevention and Control of NCDs’ of the World Health Organization. An indicator, namely ‘unconditional probability of dying between ages 30 and 70 from 4 NCDs’, is compiled by the life table method, which allows calculation of the risk of death in the absence of other causes of death. This indicator is useful for monitoring the morality trend in Hong Kong and comparing the mortality risk of 4 NCDs internationally across 1 countries/territories on the same basis. In Hong Kong, the unconditional probability of death from 4 NCDs from age 30 to age 70 decreased from 0.215 in 1980 to 0.091 in 2012. The annual average rate of decline of this probability over the period 1980-2010 is 2.7% per year. It reflects that Hong Kong is a well-performing territory during the past three decades. The annual rate of decline was 2.9% and 5.9% in 2011 and 2012 respectively. Keywords: age-specific mortality rate, unconditional probability of death 2 1. Introduction 1.1. World Health Organization’s global monitoring framework for non-communicable diseases Non-communicable diseases (NCDs) are the leading global cause of death worldwide [1]. Of the 57 million global deaths in 2008, 36 million (63%) of these were due to NCDs, comprising mainly cardiovascular diseases, cancers, chronic respiratory diseases and diabetes (4 NCDs). The combined burden of these diseases is rising rapidly among lower-income countries, populations and communities. About one-fourth of global NCD-related deaths take place before the age of 60 [2]. The 2008–2013 Action Plan was developed by the World Health Organization (WHO) and Member States to translate the Global Strategy for the Prevention and Control of Non-communicable Diseases [3] into concrete action. The Action Plan provides Member States of WHO and the international community with a roadmap to establish and strengthen initiatives for the surveillance, prevention and management of NCDs. The importance of surveillance and monitoring of progress made in the prevention and control of NCDs, was emphasized at the United Nations High-level Meeting of the General Assembly on the Prevention and Control of Non-Communicable Diseases held in September 2011 and the Political Declaration of the High-level Meeting of the General Assembly on the Prevention and Control of Non-Communicable Diseases [4] was adopted 3 by the General Assembly. The Political Declaration urges Member States to consider the development of national targets and indicators, based on national situations, building on guidance provided by WHO. Following the Political Declaration on NCDs, WHO developed a global monitoring framework to enable global tracking of progress in preventing and controlling the 4 NCDs and their key risk factors. The framework comprises 9 voluntary global targets and 25 indicators. Member States are encouraged to consider the development of national NCD targets and indicators building on the global framework [5]. Among these 25 indicators, ‘unconditional probability of dying between ages 30 and 70 from 4 NCDs’ is the indicator recommended by WHO for global monitoring of progress towards reducing premature mortality from NCDs. 1.2. Measuring mortality risk of NCDs in Hong Kong Of the 43 672 deaths in Hong Kong in 2012, 61% were due to the 4 NCDs. In this paper, we apply the actuarial, or life table method to study the mortality risk between ages 30 and 70 from 4 NCDs from 1980 to 2012 in Hong Kong based on the methodology laid down in ‘A Comprehensive Global Monitoring Framework and Voluntary Global Targets for the Prevention and Control of NCDs’ of WHO [6,7,8]. The indicator ‘unconditional probability of dying between ages 30 and 70 from 4 NCDs’ is compiled from age-specific death rates for the combined four causes (in terms of 5-year age groups 30-34, …, 65-69). The life table 4 method allows calculation of the risk of death between ages 30 and 70 from any of these causes, in the absence of other causes of death. This indicator is useful for monitoring the morality trend in Hong Kong and comparing the mortality risk of 4 NCDs internationally across countries/territories on the same basis. 2. Methodology 2.1. WHO recommended indicator for NCD surveillance within the global monitoring framework [6,7,8] According to WHO’s discussion papers in 2011 and 2012, WHO recommended ‘unconditional probability of dying between ages 30 and 70 from 4 NCDs ( 40 q 30 )’ as an indicator for global monitoring of progress toward reducing NCDs. This indicator is calculated from age-specific death rates and the probability of dying from 4 NCDs between ages 30 and 70 is derived by the life table method. This form of indicator was chosen to exclude confounding across countries or over time due to differences or changes in mortality rates for other competing causes and to control for differences in population age structure. The lower age limit for the indicator of 30 years represents the point in the life cycle where the mortality risk for the four selected chronic diseases starts to rise in most populations from very low levels at younger ages. The upper limit of 70 years was chosen for two reasons: 5 (a) to identify an age range in which these chronic disease deaths can truly be considered premature deaths in almost all regions of the world. It shows that estimated regional life expectancies at age 30 for the year 2009; in all regions except the African region, the average expected age at death for 30 year olds already exceeds 70 years; and (b) estimation of cause-specific death rates becomes increasingly uncertain at older ages because of increasing proportions of deaths coded to ill-defined causes, increasing levels of co-morbidity, and increasing rates of age misstatement in mortality and population data sources. 2.2. WHO recommended voluntary target for NCD surveillance within the global monitoring framework WHO recommended the voluntary global mortality target to be set at 25% relative reduction in overall mortality from 4 NCDs over the 15 year period 2010 - 2025. The recommended target was based on an analysis of the historically achieved trends in the indicator in recent decades. To set this target, WHO analysed data from 81 Member States with at least 15 years of vital registration data between 1980-2010 that passed quality criteria on completeness and cause-of-death assignment. WHO calculated the average annual rate of decline in the 40 q 30 for each country from the available data within this 30 year time period and computed the top 24th percentile for the 81 countries. On the conservative assumption that the other Member States without such high-quality historical 6 data would all fall below this level, this percentile corresponds to the 10th percentile for all 193 Member States, and corresponds to an annual average decline of the order of 2% per year [7,8]. The target for the 15-year period 2010-2025 was thus set at a 25% relative reduction (2% annual reduction compounded for 15 years). The historic experience of best performing countries over the period 1980-2010 has shown that very substantial declines in NCD death rates can be achieved, and that the proposed target is achievable. 2.3. Calculation of 40 q 30 2.3.1. Notations The following notations are used for abridged life table: x exact age in years. n length of interval in years. This symbol is omitted when n 1 . n Dx number of deaths occurring to persons aged x to x n of the period under consideration. n Px number of persons aged x to x n alive at the mid-point of the period under consideration. nMx age-specific death rates calculated from information on deaths among persons aged x to x n during a given year and the population aged x to x n at the mid-point of the same year. 7 n qx Probability of dying between exact ages x and x n . 2.3.2. Age-specific mortality rate from 4 NCDs First, we calculate age-specific mortality rate from 4 NCDs for each five-year age group, for each 5-year age range between 30 and 70: 5Mx 5 Dx from 4 NCDs 5 Px 2.3.3. Probability of death in each 5-year age range We then translate the 5-year death rate to the probability of death in each 5-year age range by the actuarial method [9]: 5 qx 5 5M x 1 2.5 5 M x The above formula is derived on the assumption that deaths are linearly distributed throughout the year. For the derivation of this formula, see Appendix. 2.3.4. Unconditional probability of death from age 30 to age 70 We then calculate 40 q 30 : 40 q30 1 1 5 q30 1 5 q35 1 5 q65 2.3.5. Average annual rate of decline At this point, we have 40 q 30 for each year from 1980 to 2010. We next calculate the average annual rate of decline by regressing log probability of death on year using ordinary least squares regression: ln 40 q30 year This generates a coefficient ( ) for each country/territory, from which we calculate the 8 equivalent 30-year relative decline as follows: Annual average change in We can use this annual average change in 40 q30 40q30 1 e for international comparison of mortality risk of 4 NCDs across countries/territories on the same basis. 2. The Data Data of registered death records in Hong Kong for each year from 1980 to 2012 is used in this study. A database of registered death records is maintained by the Department of Health. Registration of death is a legal requirement in Hong Kong and all deaths must be registered at the Deaths Registries of the Immigration Department. For deaths that are not classified as ‘reportable deaths’ under the Coroners Ordinance, doctors are required by law to provide information related to cause of death for deceased persons who were under their care by completing the Medical Certificate of the Cause of Death. The International Classification of Diseases (ICD), developed by WHO, has been adopted for disease coding in Hong Kong and the reason of adopting lCD is to provide good quality of mortality and morbidity statistics for formulating health policy. In this study, the classification of diseases and causes of death is based on the Ninth Revision of the ICD (ICD-9) for each year from 1980 to 2000, while the classification of diseases and causes of death is based on the Tenth Revision of the ICD (ICD-10) for each 9 year from 2001 to 2012. Table 1 exhibits the coverage of the four selected NCDs. Table 1 Coverage of four selected NCDs Four selected NCDs ICD-9 codes and descriptions ICD-10 codes and descriptions (data: 1980-2000) (data: 2001-2012) Diseases of the circulatory system Diseases of the circulatory system (390-459) (I00-I99) Cancer Malignant neoplasms (140-208) Malignant neoplasms (C00-C97) Chronic respiratory Other diseases of upper respiratory Other diseases of upper respiratory disease tract, chronic obstructive pulmonary tract, chronic lower respiratory disease and allied conditions, diseases, lung diseases due to pneumoconioses and other lung external agents, other respiratory diseases due to external agents and diseases principally affecting the other diseases of the respiratory interstitium, suppurative and necrotic system (470-478, 490-519) conditions of lower respiratory tract, Cardiovascular disease other diseases of pleura and other diseases of the respiratory system (J30-J98) Diabetes Diabetes mellitus (250) Diabetes mellitus (E10-E14) Figures for 1980-2000 based on ICD-9 are comparable with those figures for 2001-2012 based on ICD-10. 3. Results 3.1 Age-specific mortality rate from the combined four NCD causes of Hong Kong Figure 1 exhibits the age-specific mortality rates from the combined four causes for each 5-year age group (30-34, 35-39, …, 65-69) over the period 1980-2012. 10 0.02 Age-specific mortality rate 0.015 ₅M₃₀ ₅M₃₅ 0.01 ₅M₄₀ ₅M₄₅ ₅M₅₀ ₅M₅₅ ₅M₆₀ 0.005 ₅M₆₅ 0 1980 1985 1990 2000 1995 2005 2010 Year Fig. 1. Age-specific mortality rate for each 5-year age group, 1980-2012 Figure 1 shows that the age-specific mortality rates from the combined four causes for all 5-year age groups declined over the past three decades. 3.2 Probability of death from the combined four NCD causes of Hong Kong Figure 2 exhibits the probability of dying from the combined four causes for each 5-year age group (30-34, 35-39, …, 65-69) over the period 1980-2012. 11 0.1 0.09 Probabbility of dying between exact age x and x+5 0.08 0.07 0.06 ₅q₃₀ ₅q₃₅ 0.05 ₅q₄₀ ₅q₄₅ 0.04 ₅q₅₀ ₅q₅₅ 0.03 ₅q₆₀ ₅q₆₅ 0.02 0.01 0 1980 1985 1990 1995 2000 2005 2010 Year Fig. 2. Probability of dying between exact age x and x+5, 1980-2012 Figure 2 shows that the probability of dying from the combined four causes for all 5-year age groups declined over the past three decades. 3.3 40 q 30 of Hong Kong Figure 3 exhibits 40 q 30 of Hong Kong over the period 1980-2012. 12 0.25 Probability of dying 0.2 0.15 0.1 0.05 0 1980 1985 1990 1995 2000 2005 2010 Year Fig. 3. Probability of dying from age 30 to age 70 (₄₀q₃₀), 1980-2012 Figure 3 shows that 3.4 40 q 30 decreased from 0.215 in 1980 to 0.091 in 2012. Average annual rate of decline of 40 q 30 We next calculate the average annual rate of decline of 40 q 30 over the period 1980-2010 by regressing ln 40 q 30 on year using ordinary least squares regression: ln 40 q30 year This generates a coefficient ( ), from which we calculate the equivalent 30-year relative decline as follows: Annul average change in 40q30 1 e Figure 4 exhibits the regression results of ln 40 q 30 on year by fitting ordinary least squares regression over the period 1980-2010. 13 Year -1.5 1980 1985 1990 1995 2000 2005 2010 Log probability of dying -1.75 -2 -2.25 ln(₄₀q₃₀) Ordinary least squares regression line -2.5 Fig. 4. Log probability of dying from age 30 to age 70 (ln(₄₀q₃₀)), 1980-2012 The fitted regression line over the period 1980-2010: ln 40 q30 53.1686 0.0276 year F Ratio 2037 .8 with p 0.001 , and R 2 98.6% Regression results show that the annual average rate of decline of 40 q 30 of Hong Kong over the period 1980-2010 is 2.7% per year, which is comparable to the declining rate of the order of 2% per year of the 10th percentile of all Member States of WHO over the same period [7,8]. It reflects that Hong Kong is a well-performing territory during the past three decades. The annual rate of decline persisted in 2011 and 2012 at 2.9% and 5.9% respectively. 4. Discussion The mortality risk of the four major NCDs is highly associated with a wide spectrum of 14 lifestyle or behavioral risk factors. For example, (a) WHO estimated that eight risk factors – alcohol use, tobacco use, high blood pressure, high body mass index, high cholesterol, high blood glucose, low fruit and vegetable intake, and physical inactivity – account for 61% of cardiovascular deaths. Reducing exposure to these eight risk factors would increase global life expectancy by almost 5 years [10]. Additional mortality reductions are feasible through targeted health service interventions for people with high cardiovascular risk and for patients who have experienced an acute cardiovascular disease event [7]. (b) WHO estimated that the five leading behavioral and dietary risks – high body mass index, low fruit and vegetable intake, physical inactivity, tobacco use and alcohol use – are responsible for 30% of cancer deaths. Cancers with the largest proportions attributable to preventable risk factors are cervical cancer (100%) and lung cancer (71%). Seventy percent of liver cancers and 60% of stomach cancers are due to infectious agents [11]. (c) Around 60% of chronic respiratory disease deaths are attributable to tobacco smoking or exposure to indoor smoke from solid fuels [7,10]. (d) Around 60% of the risk of Type 2 diabetes mellitus can be prevented through life style modifications [13] and additional mortality reductions can be achieved through diagnosis and treatment [7]. 15 Strong NCD surveillance system, along with proper public health policies such as health promotion, risk communication, disease prevention and treatment interventions, can result in large reduction in death rates for 4 NCDs in Hong Kong. 5. Conclusions In this paper, we apply the actuarial, or life table method to study the mortality risk between ages 30 and 70 from 4 NCDs from 1980 to 2012 in Hong Kong based on the methodology laid down in ‘A Comprehensive Global Monitoring Framework and Voluntary Global Targets for the Prevention and Control of NCDs’ of WHO. An indicator ‘unconditional probability of dying between ages 30 and 70 from 4 NCDs’ is compiled from age-specific death rates of the combined four cause categories. The life table method is employed to calculate the unconditional probability of dying. The method allows calculation of the risk of death between ages 30 and 70 from any of these causes, in the absence of other causes of death. In Hong Kong, the age-specific mortality rates from the combined four causes for all 5-year age groups declined over the past three decades. The unconditional probability of death from 4 NCDs from age 30 to age 70 decreased from 0.215 in 1980 to 0.091 in 2012. The annual average rate of decline of this probability over the period 1980-2010 is 2.7% per year, which is comparable to the declining rate of the order of 2% per year of the 10th percentile of all Member States of WHO over the same period. It reflects that Hong Kong is 16 a well-performing territory during the past three decades. The annual rate of decline was 2.9% and 5.9% in 2011 and 2012 respectively. This indicator is useful for monitoring the morality trend in Hong Kong and for comparing the mortality risk of 4 NCDs internationally across countries/territories on the same basis. Acknowledgements The authors wish to express their sincere thanks to Louis FAN, Iris CHOW, and Carol TSANG for their valuable support in preparing the mortality dataset and statistical computing for this study. References [1] A. Alwan, D.R. MacLean, L.M. Riley, E.T. d’Espaignet, C.D. Mathers, G.A. Stevens and D. Bettcher, Monitoring and surveillance of chronic non-communicable diseases: progress and capacity in high-burden countries. The Lancet 376 (2010), 1861-1868. [2] Global status report on noncommunicable diseases 2010. Geneva, World Health Organization, 2011. [3] 2008 - 2013 Action plan for the global strategy for the prevention and control of noncommunicable diseases. Geneva, World Health Organization, 2008. [4] Political declaration of the high-level meeting of the general assembly on the prevention and control of non-communicable diseases. New York, United States of America, United Nations, 2011. 17 [5] Resolutions and decisions of the sixth-fifth world health assembly. Geneva, World Health Organization, 21-26 May 2012. [6] WHO discussion paper, A comprehensive global monitoring framework and voluntary global targets for the prevention and control of NCDs. World Health Organization, 2011. [7] Second WHO discussion paper, A comprehensive global monitoring framework including indicators and a set of voluntary global targets for the prevention and control of noncommunicable diseases World Health Organization, 2012. [8] Revised WHO discussion paper, A comprehensive global monitoring framework including indicators, and a set of voluntary global targets for the prevention and control of noncommunicable diseases, World Health Organization, 2012. [9] L.J. Reed and M. Merrell, A short method for constructing an abridged life table. The American Journal of Hygiene 30 (1939), 993-1022. [10] Global health risks: mortality and burden of disease attributable to selected major risks. Geneva, World Health Organization, 2009. [11] J.J. Ott, A. Ullrich, M. Mascarenhas and G.A. Stevens, Global cancer incidence and mortality caused by behavior and infection. Journal of Public Health 32(2011), 223-33. [12] J. Tuomilehto, J. Lindström, J.G Eriksson , T.T. Valle, H. Hämäläinen, 18 P. Ilanne-Parikka, S. Keinänen-Kiukaanniemi, M Laakso, A. Louheranta, M. Rastas , V. Salminen and M. Uusitupa M, Prevention of type 2 diabetes mellitus by changes in lifestyle among subjects with impaired glucose tolerance. The New England Journal of Medicine (2001), 1343-50. 19 Appendix Derivation of the Formula Relating Observed Death Rate and Probability of Death Notation and derivations (a) Notation: x exact age in years. n length of interval in years. This symbol is omitted when n 1. For the observed vital statistics: deaths in the age group x to x n . n Dx n Px population in the age group x to x n . nMx n Dx n Px death rate in the age group x to x n . For the life table population: l x survivors to age x out of a given initial number. If the initial number is unity, l x is the probability of surviving from the initial age to age x (b) Derivation of the formula n qx n nM x : n 1 nM x 2 Assume that over the interval n, l s a bs 20 By definition Mx n l x l xn xn x n { qx (1) ls ds l x l xn lx (2) Then, equations (1) and (2) becomes nMx { n qx a bx a b( x n ) xn x ( a bs)ds b bn a bx 2 a bx a b( x n) bn a bx a bx (3) (4) Eliminating a bx between equations (3) and (4) gives n qx n nM x n 1 nM x 2 For n 5 , 5 qx 5 5M x 1 2.5 5 M x 21