Vol. 6. No. 2
International Journal of Epidemiology O Oxford Unhranity Prsn 1977
Printed In Great Britain
Potential Years of Life Lost Between Ages 1 and 70: An Indicator of Premature Mortality for Health Planning* J-M ROMEDERi
J R McWHINNIE2
INTRODUCTION
This paper addresses itself to the basic question of 'How to compare the importance of major causes of death ?' The indicator of 'Potential Years of Life Lost between ages 1 and 70' is a concept which has been discussed by a number of authors over the past 30 years using various methods of calculation. This indicator is an attempt to incorporate both theoretical and practical aspects already discussed by others in the field of mortality analysis, with the primary objective of helping health planners define priorities, particularly with respect to prevention. Obviously untimely or premature death constitutes only one aspect of all health problems, and other dimensions such as morbidity and disability have to be considered if one requires an overview of major health problems. However, in many jurisdictions routine data on these other dimensions are lacking.! 1 ' Long Range Health Planning Branch, Department of National Health and Welfare, Jeanne Mance Building, Ottawa K1A OK9, Canada. * This work was first presented at the Canadian Public Health Association, June 1974; a more detailed version is available in French and in English. t In Canada, such data will be regularly collected by the Canada Health Survey which has been planned since carry 1974 (1) and will provide the first annual data for the year 1978. Reprint rtqpetts thould be addrtued to Dr. J-M Romeder.
The objectives of this paper are to review existing or proposed mortality indicators and indices, demonstrate how to calculate Potential Years of Life Lost between ages 1 and 70t with associated rates, discuss alternative methodological approaches and underlying assumptions, and finally present some Canadian data and interpretation. 1. REVIEW OF MORTALITY INDICATORS AND INDICES
Most mortality indicators or indices have been proposed in order to compare mortality in different geographic areas, occupational groups or for different years. The concept of potential years of life lost, however, originated with the primary objective of comparing the relative importance of different causes of death for a particular population. An excellent review of mortality indicators was made in 1943 by Woolsey (2) who described direct and indirect methods for standardizing death rates, with the resulting indices being called 'Standardized Mortality Ratios'. The life table death rate, promoted by Brownlee (3) was reviewed and mention t The indicator of 'Potential Yean of life Lost between ages 1 and 70' is referred to subsequently as the PYLL indicator.
143
Downloaded from http://ije.oxfordjournals.org/ at Mount Royal University on April 30, 2015
Romeder, J - M (Long Range Health Planning Branch, Department of National Health and Welfare, Jeanne Ma nee Building, Ottawa, Canada) and McWhinnie, J R. Potential years of life lost between ages 1 and 70: an indicator of premature mortality for health planning. International Journal of Epidemiology 1 9 7 7 , 6 : 1 4 3 - 1 5 1 . The indicator of Potential Years of Life Lost between ages 1 and 70 (PYLL) is proposed with the primary objective of ranking major causes of premature mortality. This proposal is based on a review of existing mortality indicators and indices and of the history of the concept of potential years of life lost The method of calculation along with the corresponding rate and the age-adjusted rate are discussed and presented with applications to Canadian data and interpretation. Several methodological aspects are discussed, particularly the comparison with more sophisticated approaches based on life tables which do not appear to alter the ranking of major causes of premature death. This indicator fits well Into the category of Social Indicators and can help health planners define priorities for the prevention of premature deaths. Epidemiological studies could also make use of this indicator of premature mortality. The simplicity of calculation and ease of comprehension should facilitate its use.
144
INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
the same year Martin (20) proposed the use of life expectancy values restricted to the first 70 years of We. In 1953 Logan and Benjamin (21) reviewed the subject to show changes in mortality patterns from 1848-72 to 1952. They proposed two further variations on the years of life lost concept which was put into perspective more recently in a book by Benjamin in 1970 (22). Another proposal for 'future working years lost' was made by Stocks (23) in 1953. Rates of years of life lost per 10,000 total population for ages 15-64 and the total to age 85 are regularly published for England and Wales by the Office of Population Censuses and Surveys.
1 • 1 Concept ofpotential years of life lost The concept seems to have been introduced for the first time by Dempsey (15) in 1947, with the objective of comparing mortality due to tuberculosis with heart diseases and cancer. For each death, she calculated the years of life remaining up until the current life expectancy. In 1948 a paper entitled 'What is the Leading Cause of Death' by Dickinson and Welker (16) proposed 'life years lost' and 'working years lost', which differed from Dempsey's method by using life expectancy at different ages instead of life expectancy at birth. This answered one of Greville's (17) criticisms of Dempsey's method. In 1950, Haenszel (18) compared five different measures of years of life lost with corresponding standardized rates and showed that the ranking of different causes of death was unaltered whether or not one used life table values. As a result, he recommended the simple method of using the difference between age at death and age 75, chosen as an upper limit. This method was used by Doughty (19) in 1951 with an upper age limit of 70. In
Therefore, PYLL is given by:
2 CALCULATION, RATE AND STANDARDIZATION
2 • 1 Method of calculation ofPYLL* The method of calculating PYLL for a particular cause or group of causes consists of a summation of the number of deaths at each age (between 1 and 70) multiplied by theremainingyears of life up to age 70. Let
dj = number of deaths between ages i and / -+- 1 ai = remaining years to live until age 70 when death occurs between ages / and i + 1 = 70 — ( i + - 5 ) assuming uniform distribution of deaths within age groups, where t represents age at last birthday
69 PYLL =
69 (70-/-
-5)di(2-l)
It can be seen that PYLL is nothing but a function of the mean age at death, for deaths between ages 1 and 70. An example of the calculation of PYLL for ischaemic heart disease is given in table I, using deaths byfive-yearage groups (with the exception of ages one to four) and the corresponding values of aj. This shorter method for the calculation of PYLL is preferred as a good approximation of formula 2 • 1. When both calculations were made for several causes of death they differed by less than two per cent for any one cause. * PYLL refers to Potential Yean of Life Lost between age* 1 and 70, calculated as indicated in this section.
Downloaded from http://ije.oxfordjournals.org/ at Mount Royal University on April 30, 2015
was made of the interesting debate which followed in 1922 in England on "The Value of Life-Tables in Statistical Research' (4). The question of 'Standard Population' was discussed, followed by "The Method of Equivalent Average Death Rates' (5) which takes account of mortality before age 65 only. The last index reviewed by Woolsey, the 'Relative Mortality Index', gives a decreasing importance to older age-groups, being a weighted average of age-specific rate ratios (ratios of given age-specific rates to standard age-specific rates) with weights equal to the population proportions of each age group. In 1951, Yerushalmy (6) reviewed mortality indices and proposed a new 'Mortality Index', similar to the Relative Mortality Index, where the weights are the relative lengths of the age intervals (such as five years for five-year age groups). His major criticism of the use of age-adjusted rates for comparing group mortalities was that 'it puts relatively heavy premiums and penalties on minor proportionate changes in the older ages*. A similar indicator called 'An Objective Mortality Indicator' was proposed the same year by Kohn (7), with weights equal to the reciprocals of the age at which death occurs, which constitute 'objective weights', constant for all countries and for all periods. Most of the previously proposed mortality indices have been reviewed subsequently in several papers (8, 9, 10, 11, 12, 13, 14) which also present new applications and methodological extensions.
145
POTENTIAL YEARS OF UFE LOST TABLE I
Caleulation of Potential Yean of Life Lost between aget 1 and 70 (P YLL), rat* and age-adjusted rate, Ontario, Ischaemie Heart Disease, Males, 1974 Standardized PYLL and rate (according to Canada age-structure)
PYLL and rate A«es
Remaining yean
Number of deaths
PYLL
Correcting factor Pk .Pi tf,' N
670 62-5 57-5 52-5 47-5 42-5 37-5 32-5 27-5 22-5 17-5 12-5 7-5 2-5
0 0 1 1 3 9 26 89 198 489 772 1015 1419 1630 5652
0 0 57-5 52-5 142-5 382-5 975 2892-5 5445 11,002-5 13,510 12,687-5 10,642-5 4075 61,865 0
Pi
108 102 103 105 103 •97 •96 •96 •95 •94 •95 100 100 101
— xN N,
0 0 59-2 551 146-8 371 0 9360 2776-8 5172-8 10,342-4 12,834-5 12,687-5 10.642-5 4115-8 60,140-4
60,140-4
61,865
Rate of PYLL
mx-x
x 1000
3,791,600 -16-3 per 1000
x 1000
3,791,600 - 1 5 - 9 per 1000
N •= 3,791,600 represents the Ontario male population between ages 1 and 70 in 1974. N r •= 10,531,000 represents the
International Journal of Epidemiology O Oxford Unhranity Prsn 1977
Printed In Great Britain
Potential Years of Life Lost Between Ages 1 and 70: An Indicator of Premature Mortality for Health Planning* J-M ROMEDERi
J R McWHINNIE2
INTRODUCTION
This paper addresses itself to the basic question of 'How to compare the importance of major causes of death ?' The indicator of 'Potential Years of Life Lost between ages 1 and 70' is a concept which has been discussed by a number of authors over the past 30 years using various methods of calculation. This indicator is an attempt to incorporate both theoretical and practical aspects already discussed by others in the field of mortality analysis, with the primary objective of helping health planners define priorities, particularly with respect to prevention. Obviously untimely or premature death constitutes only one aspect of all health problems, and other dimensions such as morbidity and disability have to be considered if one requires an overview of major health problems. However, in many jurisdictions routine data on these other dimensions are lacking.! 1 ' Long Range Health Planning Branch, Department of National Health and Welfare, Jeanne Mance Building, Ottawa K1A OK9, Canada. * This work was first presented at the Canadian Public Health Association, June 1974; a more detailed version is available in French and in English. t In Canada, such data will be regularly collected by the Canada Health Survey which has been planned since carry 1974 (1) and will provide the first annual data for the year 1978. Reprint rtqpetts thould be addrtued to Dr. J-M Romeder.
The objectives of this paper are to review existing or proposed mortality indicators and indices, demonstrate how to calculate Potential Years of Life Lost between ages 1 and 70t with associated rates, discuss alternative methodological approaches and underlying assumptions, and finally present some Canadian data and interpretation. 1. REVIEW OF MORTALITY INDICATORS AND INDICES
Most mortality indicators or indices have been proposed in order to compare mortality in different geographic areas, occupational groups or for different years. The concept of potential years of life lost, however, originated with the primary objective of comparing the relative importance of different causes of death for a particular population. An excellent review of mortality indicators was made in 1943 by Woolsey (2) who described direct and indirect methods for standardizing death rates, with the resulting indices being called 'Standardized Mortality Ratios'. The life table death rate, promoted by Brownlee (3) was reviewed and mention t The indicator of 'Potential Yean of life Lost between ages 1 and 70' is referred to subsequently as the PYLL indicator.
143
Downloaded from http://ije.oxfordjournals.org/ at Mount Royal University on April 30, 2015
Romeder, J - M (Long Range Health Planning Branch, Department of National Health and Welfare, Jeanne Ma nee Building, Ottawa, Canada) and McWhinnie, J R. Potential years of life lost between ages 1 and 70: an indicator of premature mortality for health planning. International Journal of Epidemiology 1 9 7 7 , 6 : 1 4 3 - 1 5 1 . The indicator of Potential Years of Life Lost between ages 1 and 70 (PYLL) is proposed with the primary objective of ranking major causes of premature mortality. This proposal is based on a review of existing mortality indicators and indices and of the history of the concept of potential years of life lost The method of calculation along with the corresponding rate and the age-adjusted rate are discussed and presented with applications to Canadian data and interpretation. Several methodological aspects are discussed, particularly the comparison with more sophisticated approaches based on life tables which do not appear to alter the ranking of major causes of premature death. This indicator fits well Into the category of Social Indicators and can help health planners define priorities for the prevention of premature deaths. Epidemiological studies could also make use of this indicator of premature mortality. The simplicity of calculation and ease of comprehension should facilitate its use.
144
INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
the same year Martin (20) proposed the use of life expectancy values restricted to the first 70 years of We. In 1953 Logan and Benjamin (21) reviewed the subject to show changes in mortality patterns from 1848-72 to 1952. They proposed two further variations on the years of life lost concept which was put into perspective more recently in a book by Benjamin in 1970 (22). Another proposal for 'future working years lost' was made by Stocks (23) in 1953. Rates of years of life lost per 10,000 total population for ages 15-64 and the total to age 85 are regularly published for England and Wales by the Office of Population Censuses and Surveys.
1 • 1 Concept ofpotential years of life lost The concept seems to have been introduced for the first time by Dempsey (15) in 1947, with the objective of comparing mortality due to tuberculosis with heart diseases and cancer. For each death, she calculated the years of life remaining up until the current life expectancy. In 1948 a paper entitled 'What is the Leading Cause of Death' by Dickinson and Welker (16) proposed 'life years lost' and 'working years lost', which differed from Dempsey's method by using life expectancy at different ages instead of life expectancy at birth. This answered one of Greville's (17) criticisms of Dempsey's method. In 1950, Haenszel (18) compared five different measures of years of life lost with corresponding standardized rates and showed that the ranking of different causes of death was unaltered whether or not one used life table values. As a result, he recommended the simple method of using the difference between age at death and age 75, chosen as an upper limit. This method was used by Doughty (19) in 1951 with an upper age limit of 70. In
Therefore, PYLL is given by:
2 CALCULATION, RATE AND STANDARDIZATION
2 • 1 Method of calculation ofPYLL* The method of calculating PYLL for a particular cause or group of causes consists of a summation of the number of deaths at each age (between 1 and 70) multiplied by theremainingyears of life up to age 70. Let
dj = number of deaths between ages i and / -+- 1 ai = remaining years to live until age 70 when death occurs between ages / and i + 1 = 70 — ( i + - 5 ) assuming uniform distribution of deaths within age groups, where t represents age at last birthday
69 PYLL =
69 (70-/-
-5)di(2-l)
It can be seen that PYLL is nothing but a function of the mean age at death, for deaths between ages 1 and 70. An example of the calculation of PYLL for ischaemic heart disease is given in table I, using deaths byfive-yearage groups (with the exception of ages one to four) and the corresponding values of aj. This shorter method for the calculation of PYLL is preferred as a good approximation of formula 2 • 1. When both calculations were made for several causes of death they differed by less than two per cent for any one cause. * PYLL refers to Potential Yean of Life Lost between age* 1 and 70, calculated as indicated in this section.
Downloaded from http://ije.oxfordjournals.org/ at Mount Royal University on April 30, 2015
was made of the interesting debate which followed in 1922 in England on "The Value of Life-Tables in Statistical Research' (4). The question of 'Standard Population' was discussed, followed by "The Method of Equivalent Average Death Rates' (5) which takes account of mortality before age 65 only. The last index reviewed by Woolsey, the 'Relative Mortality Index', gives a decreasing importance to older age-groups, being a weighted average of age-specific rate ratios (ratios of given age-specific rates to standard age-specific rates) with weights equal to the population proportions of each age group. In 1951, Yerushalmy (6) reviewed mortality indices and proposed a new 'Mortality Index', similar to the Relative Mortality Index, where the weights are the relative lengths of the age intervals (such as five years for five-year age groups). His major criticism of the use of age-adjusted rates for comparing group mortalities was that 'it puts relatively heavy premiums and penalties on minor proportionate changes in the older ages*. A similar indicator called 'An Objective Mortality Indicator' was proposed the same year by Kohn (7), with weights equal to the reciprocals of the age at which death occurs, which constitute 'objective weights', constant for all countries and for all periods. Most of the previously proposed mortality indices have been reviewed subsequently in several papers (8, 9, 10, 11, 12, 13, 14) which also present new applications and methodological extensions.
145
POTENTIAL YEARS OF UFE LOST TABLE I
Caleulation of Potential Yean of Life Lost between aget 1 and 70 (P YLL), rat* and age-adjusted rate, Ontario, Ischaemie Heart Disease, Males, 1974 Standardized PYLL and rate (according to Canada age-structure)
PYLL and rate A«es
Remaining yean
Number of deaths
PYLL
Correcting factor Pk .Pi tf,' N
670 62-5 57-5 52-5 47-5 42-5 37-5 32-5 27-5 22-5 17-5 12-5 7-5 2-5
0 0 1 1 3 9 26 89 198 489 772 1015 1419 1630 5652
0 0 57-5 52-5 142-5 382-5 975 2892-5 5445 11,002-5 13,510 12,687-5 10,642-5 4075 61,865 0
Pi
108 102 103 105 103 •97 •96 •96 •95 •94 •95 100 100 101
— xN N,
0 0 59-2 551 146-8 371 0 9360 2776-8 5172-8 10,342-4 12,834-5 12,687-5 10.642-5 4115-8 60,140-4
60,140-4
61,865
Rate of PYLL
mx-x
x 1000
3,791,600 -16-3 per 1000
x 1000
3,791,600 - 1 5 - 9 per 1000
N •= 3,791,600 represents the Ontario male population between ages 1 and 70 in 1974. N r •= 10,531,000 represents the
