SISA Research Paper
Life expectancy and SMR applied to migrant groups living in Amsterdam, The Netherlands.
Abstract from: Uitenbroek DG. Verhoeff A. Life expectancy and mortality differences between migrant groups living in Amsterdam, The Netherlands. Social Sciences and Medicine 2002 (54/9): 1379-1388. Please also see the lifetable spreadsheet on which this paper is based.
This paper presents a study of mortality in Amsterdam, the Netherlands. The study is informed by previous research, which shows that in Amsterdam life expectancy is lower than life expectancy in the Netherlands as a whole. Several studies in the Netherlands have shown that groups of immigrants have a high relative mortality as expressed in the standardised mortality ratio (SMR) estimates (Uniken Venema, Garretsen & van der Maas, 1995; Reijneveld, 1998; Kocken, Mackenbach, Oers & Uniken Venema, 1994; Hoogenboezem & Israëls, 1990) which might explain the low life expectancy in Amsterdam, a city with a high proportion of migrants in the population.
The present study includes an analysis based on data made available by the Amsterdam city council as to age-specific number of deaths and population size (O&S). The analysis was carried out on five nationality groups. The differences in mortality between these five nationality groups and the populations of Amsterdam and the Netherlands as a whole are examined by means of life table analysis, a method which consists of applying the current age-specific mortality rates to a hypothetical cohort. The life expectancy measure is the most important outcome measure of life table analysis and will be at the centre of interest in this paper. Life expectancy is the mean number of years a cohort of people might expect to live in accordance with the current age-specific mortality rate. Life expectancy is the most important indicator used in demography to compare periods and regions in terms of health and health care (including prevention) (Pressat, 1969; Shryrock. & Siegel, 1973). The life expectancy measure is generally considered to reflect differences in mortality well, but on the other hand to be insensitive to age structures in populations, changes in birth rates, and other demographic phenomena.
In migrant studies the SMR is often used to compare mortality rates in different population groups. The analysis in this paper also made use of the SMR. This analysis follows previous studies in migrant health in which mortality in migrant groups was compared with the national figures for the Netherlands. The SMR is sensitive to differences in age distribution between groups (McMahon, & Trichopoulos, 1996; Mausner, & Kramer, 1985; Shryrock & Siegel, 1973). This paper will study the question to what extent problems related to age distribution may have influenced previous studies.
The data consists of statistics on population size and number of deaths collected by the civil registry in Amsterdam during the period 1-1-1994 to 1-1-2000.
Origin is determined in the following way: each resident of Amsterdam is classified as a member of a certain nationality group. The analysis was carried out on five nationality groups among residents of Amsterdam: those of Dutch descent, those of Mediterranean or Caribbean descent, and those from a non-industrialised or an industrialised country. For the sake of this analysis, people from Morocco, Turkey and Southern Europe were combined to form a Mediterranean nationality group, and people from Surinam and the Netherlands Antilles to form a Caribbean nationality group.
The analysis made use of standard demographic analyses such as life table analysis and the calculation of standardised mortality ratios (SMRs). Life table analysis was carried out according to the abridged life table analysis method discussed by Chiang (Chiang, 1968). The data are divided into five-year age categories running from 0-4, 5-9, and so on, up to 80+. In calculating the SMRs indirect standardisation was used, i.e. the national Dutch or Amsterdam age-specific mortality ratios were applied to the age categories of the 5 nationality groups.
Table one. Population sizes and number of deaths in the Netherlands, Amsterdam and five nationality groups living in Amsterdam. Percentages between brackets.
Table 1 gives an overview of the data used in the analysis. In the first column, population sizes on 1-1-2000 are shown. Among both males and females, individuals from the Netherlands constitute the largest group. The second largest group consists of people of Mediterranean origin (about 15%), followed by those of Caribbean origin (about 11%). In the second column the sum of the mean populations for the years 1994-2000 is given. This sum forms the denominator in the calculations that follow. In the third column the total number of deaths during the period from 1-1-1994 to 1-1-2000 is given for each nationality group. Note the high proportion of deaths among individuals from the Netherlands and the industrialised world: these account for about 90% of all deaths in Amsterdam among both males and females. This high level can be explained by the fact that taken together these two groups form a large proportion of the total population and that they have a higher average age than the other nationality groups (see the last column of Table 1).
Table 2. Life expectancy for the Netherlands population, Amsterdam population and five nationality groups living in Amsterdam for the period from 1-1-1994 to 1-1-2000.
Table 2 presents the life expectancy at age nil and at age fifty for males and females, and for various nationality groups. Males from the Mediterranean and Caribbean nationality group have a higher life expectancy than the national average, while males from all other nationality groups have a lower life expectancy than the national average. For Amsterdam males of Mediterranean and Dutch descent and for those from industrialised countries, the difference is statistically significant. Life expectancy is lowest for citizens from the Netherlands and from the industrialised countries. Males born in the Mediterranean area or their immediate descendants have a life expectancy which is 2.5 years longer than those born in the Netherlands. At age fifty, immigrants from Caribbean, Mediterranean and non-industrialised countries all have a higher life expectancy compared with the national figures. The difference is significant for Mediterranean and Caribbean males; these two groups can look forward to a long remaining period of life. Males from the Netherlands and the industrialised countries have a significantly shorter life expectancy.
For females, immigrants from the Mediterranean and non-industrialised countries have a higher life expectancy compared with the national figures. For Mediterranean females the difference is statistically significant. Females of Mediterranean origin score highest in terms of life expectancy, while those from the Netherlands and the industrialised countries score lowest (a significant outcome). At age fifty, all nationality groups except those from the Netherlands and the industrialised countries have a longer life expectancy than the female Dutch population in general. These relationships are statistically significant for females of Mediterranean and Dutch origin and those from industrialised countries.
Table 3. Mortality rates and SMRs for the Netherlands population, Amsterdam population and five nationality groups living in Amsterdam for the period from 1-1-1994 to 1-1-2000.
Table 3 presents mortality rates and standardised mortality ratios. As can be seen in the first column, crude mortality rates differ distinctly between the various nationality groups. People from the Netherlands have by far the highest mortality rate, followed by those from industrialised countries. In the second column the age-specific mortality figures for Amsterdam are applied to each of the age groups in each of the nationality groups. The resulting SMR will be interpreted in terms of 'benefit'. With the exception of males and females from the Netherlands and males from the industrialised countries, none of the groups would benefit from having a mortality pattern which was more like the overall Amsterdam mortality pattern. This outcome is statistically significant for the Mediterranean group: their mortality rate is far lower than it would be if it matched the general pattern.
The third column in Table 3 shows the SMRs based on the national Dutch mortality figures. For males, all nationality groups except the Mediterranean group would benefit from having a pattern more like the national pattern. The groups from industrialised and non-industrialised countries and from the Caribbean, i.e. from Surinam and the Netherlands Antilles, would benefit most. Among females, all groups except the Mediterranean group would benefit from having a mortality pattern similar to the national pattern.
This analysis raises a number of questions. On the basis of the Amsterdam mortality figures, those from the Netherlands have the highest relative mortality. The results of the analysis using SMRs confirm the results of the life table analysis. However, if the analysis is based on national figures, the outcome is that citizens from the industrialised and non-industrialised countries and those of Caribbean origin have a high relative mortality. This contradicts the life table analysis, but bears out the results of previous studies in migrant mortality: the SMRs obtained here are very similar to previous findings. How can this contradiction between the life expectancy measure and the SMR be explained?
The explanation is to be found in the complex interaction between the age distributions in the
population studied and the mortality patterns in the standard population. In order to study this interaction, in Figure 1 age-specific mortality risks are compared with the national mortality pattern. The horizontal line at Y-value 1 (one) denotes the national mortality pattern, the wobbly lines denote the relative risks.
In general there is relatively high mortality in children 0 to 4 years of age, particularly among children from the non-industrialised countries. Up to the age group 25-29 the pattern is less consistent: the total mortality in these age groups is low in general, so that chance fluctuation may play a role. From 20-24 years of age onwards relative mortality increases, and then drops to lower levels in higher age groups. Amsterdam residents from non-industrialised countries are not particularly disadvantaged as far as this trend is concerned: from age group 55-59, the age from which mortality really starts to bite, their relative mortality tends to be low.
The high SMRs observed for males in Table 3 in the comparison between the Netherlands and the Amsterdam nationality groups can now be explained. Citizens of Amsterdam tend to have a higher mortality in the young adult and mid-age population than those of the Netherlands as a whole. Many migrant groups also have a lower average age than the overall population of the Netherlands and are moreover concentrated in these age groups. Interaction with high relative mortality rates takes place in precisely those age groups in which the number of migrants is relatively large; this leads to a higher SMR for a number of nationality groups than might be expected on the basis of the national pattern.
In this study mortality patterns in several migrant groups in Amsterdam were examined. Approximately half of the population of Amsterdam is of Dutch descent and the other half consists of migrants or their immediate descendants. The Dutch population has a high average age, whereas the migrants have a low average age. In general, residents of Amsterdam have a low life expectancy in comparison with the population of the Netherlands as a whole; the difference is 1.7 years for males and 1.6 years for females. With respect to life expectancy, Amsterdam residents of Dutch origin or from an industrialised country score lowest, whereas migrant groups score higher. However, many studies using the SMR have shown high mortality in groups of migrants (Wild & McKeigue, 1997; Balarajan, 1991; Uniken Venema, Garretsen & van der Maas, 1995; Bollini & Siem, 1995; Sundquist & Johanson, 1997b; Sundquist, 1995). An SMR study in this paper also shows that if the national Dutch population is used as the standard, migrant groups living in Amsterdam have a high relative mortality. However, if the subgroups of the population of Amsterdam are compared with the general population in Amsterdam, the SMRs change considerably; the outcome is then that the Dutch have the highest relative mortality. Further study of the data reveals that migrants, who predominantly live in urban areas, have an urban pattern of mortality. This pattern is characterised by a high relative mortality in adult and mid-age groups and a low mortality at high age. For some groups, the SMR is inflated by the interaction between this urban pattern of mortality and the low average age of migrants. In short, migrants have a relatively high chance of dying in the age groups in which they are also represented in large numbers. The conclusion is that the SMR is unreliable for use in migrant studies because the age structures of the different groups vary so widely. This is not a new conclusion (McMahon & Trichopoulos, 1996; Mausner & Kramer, 1985; Shryrock & Siegel, 1973), but its implications demand further consideration. For example, there are studies which compare SMRs from different periods in order to study changes in migrant mortality (Wild & McKeigue, 1997; Balarajan, 1991). But to what extent are the observed changes truly the result of a changed disease prevalence and to what extent are they the result of ageing populations and other demographic phenomena?
Balarajan, R. (1991). Ethnic differences in mortality from ischaemic heart disease and cerebrovascular disease in England and Wales. British Medical Journal 302, 560-4.
Bollini, P., & Siem, H. (1995). No real progress towards equity. Health of migrants and ethnic minorities on the eve of the year 2000. Social Science and Medicine 41, 819-28.
Chiang, C. L. (1968). The life table and its construction. In, Chiang C.L. Introduction to Stochastic Processes in Biostatistics (pp.189-214). New York, John Wiley & Sons Inc.
Hoogenboezem, J., & Israëls, A. Z. (1990). Sterfte naar doodsoorzaak onder Turkse en Marokkaanse ingezetenen in Nederland 1979-1988. Maandbericht gezondheid (CBS). (August), 5-20.
Kocken, P. L., Mackenbach, J. P., Oers, J. A. M., & Uniken Venema., H. P. (1994). Sterfte, ervaren gezondheid en gerapporteerd voorzieningengebruik van Rotterdamse Surinamers. Tijdschrift voor gezondheidswetenschappen 72, 231-6
Mausner, J. S., & Kramer, S. (1985). Epidemiology-An introductory text. Philedelphia, W. B. Saunders Company.
McMahon, B. & Trichopoulos, D. (1996). Epidemiology, principles and methods (2nd edition). Boston, Little., Brown and company.
O&S. De Amsterdammers in acht ethnische groepen. Amsterdam, Het Amsterdamse Bureau voor onderzoek en statistiek, (annual).
Pressat, R. (1969). Demographic Analysis. Methods., Results Applications. London, Edward Arnold.
Reijneveld, S. A. (1998). Reported health, lifestyles & use of health care of first generation immigrants in The Netherlands: do socio-economic factors explain their adverse position? Journal of Epidemiology and Community Health 52, 298-304.
Shryrock, H. S. & Siegel, J. S. Mortality. In Shryrock, H.S., & Siegel, J.S. (1973). The methods and materials of demography (pp. 389-428). Washington, U. S. Department of Commerce, Bureau of the Census.
Shryrock, H. S. & Siegel, J. S. (1973). The Life Table. In Shryrock, H.S., & Siegel, J.S. The methods and materials of demography (pp. 429-460). Washington, U. S. Department of Commerce, Bureau of the Census.
Sundquist, J. (1995). Ethnicity, social class and health. A population-based study on the influence of social factors on self-reported illness in 223 Latin American refugees., 333 Finnish and 126 south European labour migrants and 841 Swedish controls. Social Science and Medicine 40 (6),777-87.
Sundquist, J. & Johansson, S. E. (1997a). Long-term illness among indigenous and foreign-born people in Sweden. Social Science and Medicine 44 (2),189-98.
Sundquist, J. & Johansson, S. E. (1997b) The influence of country of birth on mortality from all causes and cardiovascular disease in Sweden 1979-1993. International Journal of Epidemiology 26 (2),279-87.
Uniken Venema, H. P., Garretsen, H. F. L. & van der Maas, P. J. (1995). Health of migrants and migrant health policy, The Netherlands as an example. Social Science and Medicine 41, 809-18.
Veugelers, P. J., Erkens, C. G. M., Verhoeff, A. P. (1998). Toenemende verschillen in levensverwachting tussen Amsterdam en Nederland. Tijdschrift voor gezondheidswetenschappen. 8, 479-80.
Wild, S. & McKeigue, P. (1997). Cross sectional analysis of mortality by country of birth in England and Wales., 1970-92. British Medical Journal 314, 705-710.
SISA Research Paper