-
The Stagnation of the Mexican Life Expectancy in the First
Decade of the Twenty
First Century: The Impact of Violent Deaths.
Vladimir Canudas-Romo* Vctor Manuel Garca-Guerrero^1
Abstract
In the first decade of the twenty first century the Mexican life
expectancy changed from a
long trend of increase to stagnation. These changes concur with
an increase in deaths by
homicides that the country experienced in that decade. There are
138,461 official reported
deaths by homicide in the period of 2000 to 2010. However, the
time trend shows an
increase in the counts of homicides in the later years from an
annual number of 10,000 to
25,000 deaths in 2010. We quantify the impact of these changes
in homicides and other
causes of death in life expectancy. Male life expectancy
remained around the value of 72
years from 2000 to 2010. However, the apparent stagnation in
life expectancy is resultant of
increase in homicides and diabetes deaths on one hand, and the
positive improvements
observed in other causes of death on the other. The negative
impact of homicides is
particularly observed at ages 15 and 50, and diabetes for ages
45 and more, and they
account for almost an entire year of the male life expectancy.
Mexican males would have
observed a 2 years increase in life expectancy if homicides and
diabetes deaths had been
avoided.
* University of Southern Denmark, Denmark; Johns Hopkins
University, USA ^ Colegio de Mexico, Mexico
-
Introduction
The demographic transition has been described as a period of
decline from high to lower
levels of fertility and mortality (Notestein 1945, Davis 1945,
Lee 2003). The mortality
aspect of this transition has been characterized by, among
others, the increase in life
expectancy at birth (LE). As mortality has changed, the cause of
death structure has shifted
from a dominance of deaths due to infectious diseases to a
preponderance of deaths due to
chronic diseases, a change denoted as the epidemiologic
transition (Omran, 1971).
As conceived initially the epidemiological transition had three
stages: (a) The age of
pestilence and famine which precedes the mortality transition
(LE is less than 40 years);
(b) The age of receding pandemics with steady decline in
mortality and increase in LE to
levels above 40 years; and (c) The age of degenerative or
human-made diseases with LE
reaching 70 years and above. Other authors have included more
stages to the transition, for
example the missing component of the violent and accidental
mortality (Rogers and
Hackenberg 1987).
During the second half of the twentieth century Mexico
transitioned at a very rapid pace
into the second and third stages of the epidemiological
transition. Life expectancy in 1950
was 49 and 45 years for females and males respectively, and
changed by 2000 to 77 years
for females and 72 for males (INEGI, 2012). These changes
correspond to annual increases
in life expectancy of 0.38 and 0.40 of a year for females and
males respectively. These
paces of increase are far higher than the long uninterrupted
increase of 0.25 per year
observed in the trend of the female record life expectancies in
the world (Oeppen and
Vaupel 2002). Thus, Mexico has greatly reduced its longevity gap
respect to the developed
world of countries holding the maximum life expectancy.
In the first decade of the twenty first century, Mexico
experienced a dramatic change in
its public health: from 2000 to 2010 life expectancy for males
stagnated and for females
experienced a modest half a year increase during the entire
decade. These changes concur
with an increase in deaths by homicides that the country
experienced in that decade. The
focus of the present study is to assess the contribution of
violent deaths, particularly
Homicides, in explicating the observed stagnation in life
expectancy.
We ask about the possible role of other causes of death in this
stagnation and quantify
the changes in the age-pattern of mortality. However, our aim
also has the international
-
perspective of comparing the Mexican situation with that of
mortality shocks observed in
other countries. For example, the triple Japanese disaster of
2011,
earthquake/Tsunami/Nuclear disaster, and its devastating effect
in the population of that
island nation and the well document Russian excess alcohol
related mortality in the last
decades (Schkolnikov et al 2013).
The following sections include the description of the Mexican
data used in the analysis,
the methods applied, results and discussion.
Data
This project uses two types of data: death counts by causes of
death, and complete life
tables for the period 2000 to 2010. The information on deaths by
cause of death comes
from the administrative registers managed by the National
Institute of Statistics and
Geography (INEGI for its acronym in Spanish). The INEGI collects
this information on an
annual basis from both the Civil Register Bureau and the
Prosecutor Agencies. The first
one collects general and fetal deaths by cause, and the latter
collects violent and accidental
deaths. The causes of death are according to World Health
Organizations International
Classification of Diseases tenth revision (ICD-10).
Information on death counts is available by sex and single age,
for each calendar year
from 2000 to 2010. Deaths in the group of Not Specified, or
without age or sex, were
prorated by age and sex. We have selected to work with ten
causes of death: Infectious
Diseases, Neoplasm, Diseases of the Circulatory System, Mental
Disorders,
Endocrine/metabolic Diseases, Diseases of the Digestive System,
Conditions of Perinatal
Period, Homicide, Other external Causes, Others causes of death
not accounted by the other
groups. We further aggregate these ten causes for the years lost
of life analysis explained
below.
The calculation of the age-specific death rates used in the
construction of the life tables
are briefly described here (more details can be found at SOMEDE,
2011). The general
death rates are estimated from different sources: vital
statistics, six different national
surveys (EMF 76-77, ENFES 87, ENADID 92, 97 and 09), three
decennial censuses (1990,
2000 and 2010) and one midterm census (2005). Specific methods
are applied for the
estimation of the mortality for each age group: 1) under one
year old (to estimate infant
mortality rates), 2) between 1 and 4 years, and 3) for ages
above 5 years.
-
The infant mortality rate (IMR) and the mortality rates between
ages 1 and 4 come from
vital statistics, surveys pregnancy and birth histories, and
census information of death
offspring and children born alive. The births in the denominator
of the IMR come from
vital statistics registers. Births are underestimated (around
2%) from this data source, but
the total number of births are corrected by adding the births
registered four years after the
year of occurrence (Mier y Tern in SOMEDE 2011).
The information to estimate the mortality rates for ages five
and more comes from vital
registers: vital statistics and censuses. The deaths and
populations are corrected for age
misspecification and smoothed. The total mid-year population
from census is interpolated
using cubic splines and its age-structure is estimated with
linear interpolation. Finally, the
life tables are calculated as shown in Preston, Heuveline and
Guillot (2001).
We used life tables elaborated by the Ministry of Health, Labour
and Welfare of Japan
to make comparable analysis as those elaborated for Mexico and
the information available
on life tables for Russia from the Human Mortality database and
World Health
Organization for the cause of death data.
Methods
The proportion of deaths at age x of cause i, denoted as c(x,i),
was calculated and
applied to the age-specific death rates, m(x), to obtain age-
and cause-specific death rates:
m(x,i)=c(x,i)m(x). These age- and cause-specific death rates
were used in both the
decomposition and years lost analysis explained below.
Decomposition techniques are standard methods for comparing life
expectancy across
populations and time, and analyzing the age-contributions and
cause-contributions to the
differences (Preston, Heuveline and Guillot 2001). We extracted
age-specific death rates,
numbers of survivors and persons-years from the life tables
described above. These values,
together with the proportion of deaths for each cause in every
age group, were used to
obtain the age and cause-specific death rates and contributions
that make up the difference
in life expectancies between years 2000 and 2010.
We calculate the years lost due to different causes of death by
focusing on the
cumulative death distribution. Let the radix of the population
be equal to one, (0) = 1. At any given age x, the survival
function, (), and the function of the distribution of deaths from 0
to the given age x, (0, ), together add to the radix of the
population:
-
() + (0, ) = 1, (1)
The function of the distribution of deaths can be further
substituted for the addition of
the deaths by the different causes of death, (0, , ). If there
are n mutually exclusive causes of death equation (1) can be
re-written as:
() + (0, , )=1 = 1. (2)
Integrating both sides from 0 to an age of interest a, returns a
relation of the total
number of possible years lived between zero and age a, namely a,
and the temporary life
expectancy between ages 0 and a, and years lost due to each of
the causes of death:
()0 + (0, , )0=1 = . (3)
or with more succinct notation as,
0 + 0=1 = , (4)
where temporary life expectancy is defined as 0 = ()0 and the
years lost to each cause of death as 0 = (0, , )0 . More details on
this method can be found in Andersen, Canudas-Romo and Keiding
(2013).
Results
Age-specific death rates
Figure 1 presents the age-specific death rates for Mexican
females and males in 2000
and 2010. In the most recent year the accident hump
characteristic of the age-group 18-25
increases for females and males. Although, for males the
increase in the risk of death is
observed up to age 40. For the rest of the ages there is,
apparently, little change except for
the slight decline in lowest death rates for males around age 10
to 12.
-
[Figure 1 about here]
Cause-decomposition analysis
The female and male age-contribution to the change in the
Mexican life expectancy
between 2000 and 2010 are observed in Figures 2A and 2B
respectively. Between 2000 and
2010 there was barely any increase in the average life span for
females, changing from 77.2
to 77.8 years. But the situation is far worse for their male
counterparts which experienced
no change in life expectancy remaining at 72 years. In Figures
2A and 2B bars above the
horizontal line at the value of zero show the contribution of
age groups to the increase in
life expectancy while those below oppose this change. The
progress made in bringing down
mortality in the first year of life is equivalent to more than
0.4 of a year of the life
expectancy change both for females and males. Other age-groups
that experienced
survivorship increases were the ages 45 to 70 for males and ages
30 to 75 for females.
However, young adults and those above 75 have observed increases
in mortality that
translate into this negative contributions.
[Figure 2A and 2B about here]
This contrast for example with the situation observed during the
triple disaster of 2011
in Japan where at every single age deterioration in mortality
was experienced. The female
and male age-contribution to the change in the Japanese life
expectancy between 2010 and
2011 are observed in Figures 3A and 3B respectively. Between
such years there was a
decrease in the average life span for females, changing from
86.4 to 85.9 years. Similarly,
there was a decrease in the average life span for males,
changing from 79.6 to 79.4 years. In
Figure 3A all bars are below zero which means that all age
groups contributed, more or less
in the same magnitude (except the age-group of 95 and more), to
the decrease in female life
expectancy at birth. Similar results are observed for Japanese
males, except for those aged
group 0-1, 45-49 and 55-65.
[Figure 3A and 3B about here]
-
The female and male age and cause-specific contributions to the
change in the Mexican
life expectancy between 2000 and 2010 are observed in Figures 4A
and 4B respectively.
Between 2000 and 2010 the main cause of female LE increase was
the improvement of
perinatal periods conditions, concentrated in the first year of
life. Its contribution was of
0.26 of a year. Advances in the prevention of both diseases of
the circulatory system and
neoplasm between ages 35 to 85, contributed increasing LE in
0.18 and 0.19 of a year
respectively. As a counterpart, endocrine/metabolic diseases,
predominantly diabetes,
(concentrated in ages of 40 and more) and homicides
(concentrated in ages between 15 and
35 years) contributed in a negative way to LE.
Similarly, the main cause of the increase in the male LE also
was the reduction in
perinatal deaths, concentrated at ages below one. Its
contribution was of 0.28 of a year.
Advances in preventing both diseases of the digestive systems
and other external causes,
between ages 25 to 70 and 40-80 respectively, contributed
combined to an increase in LE of
0.4 of a year. Nevertheless, homicides (concentrated in ages
between 15 and 55 years) and
endocrine/metabolic diseases, where most of the cases are
diabetes, (concentrated in ages of
30 and more) contributed in a negative way to LE. Homicides
alone reduced LE in 0.6 of a
year. Similar to the increase in LE gained against perinatal,
diseases of digestive system
and other external causes together.
[Figure 4A and 4B about here]
Years lost analysis
Figure 5 presents the survival function for the Mexican
population in 2005 for females
and males, as well as the death accumulation by age and cause of
death, or years lost by
causes of death. The causes of death have been further grouped
to more clearly see their
trend into external (homicide, suicide and accidents),
cardiovascular diseases or CVD,
cancers, diabetes, perinatal, and a group of remaining causes
not accounted by the other
five. After age 55 the rapid increasing trend of CVD, cancers
and diabetes mortality is
observed in the spread of these colored bands. While for females
external causes have a
moderate presence by age 55, for males this cause is among the
most important components
of years lost of life.
-
.
[Figure 5 about here]
To put this high numbers of violent deaths (or external causes)
in context with other
situation of excess mortality, we contrast the Mexican male
results with the same analysis
for their Russian counterparts. Table 1 presents the years lost
to external causes,
cardiovascular diseases, cancers and other causes. By age 55, 70
and 85 a total number of
2.9, 6.5 and 14.3 years respectively have been lost in the
Mexican male life table. The
numbers for Russian males are 5.9, 13.9 and 26.4 years for the
three mentioned ages. Over
age, the years lost to external causes in Mexico reduced from
representing 20% to 15% of
all deaths, while in Russia the reduction is from 34% to
31%.
[Table 1 about here]
Discussion
The impact of catastrophic events in the health of a population
is captured without
ambiguity in the changes in mortality, or as studied here
changes in life expectancy. The
influences of these types of shocks in mortality depend on its
intensity, durability, and the
population context where they occur. Japans triple disaster was
intense, but its duration
was short and the public health context is far better than for
many other countries in the
world. The population of Japan is of 127 million habitants,
around 20,000 deaths occurred
during the disaster of 2011, equivalent to a decline in life
expectancy of half a year for
females. As such, Japan remains as the country with the highest
life expectancy in the
world (Canudas-Romo & Dokko, 2010). Russia on the other hand
has had a long and
lasting excess mortality, which is now reducing after more than
a decade of high levels
(Schkolnikov et al 2013).
Mexico, with 112 millions habitants, has had a total of 138,461
official deaths by
homicides in the period 2000 to 2010 and alone in 2009 and 2010
above 20,000 homicides
per year. The intensity and severity of this catastrophic event
has increased over time, and
its duration although there was a clear beginning its end is
unpredictable, and there are no
signs that violence is slowing down. At the turn of the century
Mexico was equipped with a
good medical and public health system; institutions that brought
the country to life
-
expectancies levels of 77 years for women and 72 for men in the
year of 2000. These values
are higher than the mean life expectancies for eastern European
countries of 76 and 67
years for females and males respectively (own calculations from
HMD).
In 2000, Mexico stood at a critical moment of development with
favorable demographic
dividend, or having its biggest proportion of persons between
ages 15 to 65 in history.
Around 2000 the proportion of persons in the age-group 15 to 65
passed the 60% mark
respect to the total population and by 2010 it had reached 64%.
As a consequence, the total
dependency ratio decreased from 76 dependents per every 100
workers 15-64, to 56
dependents by 2010. Furthermore, in recent population
projections Garca-Guerrero (in
press) has shown that this decrease could continue until
mid-2030s. We had hypothesized
that violent deaths have had a severe impact in terms of loss
time of the window of the
Mexican demography dividend. However, its toll on this respect
is not numerically evident
and it stands at much better situation than other countries that
are also confronting health
crises as Russia.
The lack of opportunities for young people (employment,
education, etc.), added to the
great social disparity among socioeconomic status, have
catalyzed the current crisis in the
country. On top of that, at the end of twentieth century the
main drug-cartels in the country
were disbanded and those who remained started a strong fight for
territories. The drug
cartels have a deep connection with the Mexican civilian
population, so the States war
against the cartels has directly and indirectly affected the
Mexican population. This war
was intensified during the government 2006-2012. In 2012
Mexicans elected a president
from a different party than for the last twelve years. Would the
direction of the current
violence change in the coming years with the new leadership of
the country? Answering
this question is beyond the scope of this research. Furthermore,
our quantification of
violence only shows the extreme result of the development in the
latest years, namely its
deaths, but the social, psychological and physical scars that
violence has left in the country
are not possible to quantify in such simplistic ways.
-
References
Andersen, P.K., Canudas-Romo V. and N. Keiding. 2013.
Cause-specific measures of life
years lost. (Available from the authors).
Bennet, N.G. and S. Horiuchi. 1981. Estimating the completeness
of death registration in a
closed population. Population Index 47(2):207-221.
Bennet, N.G. and S. Horiuchi. 1984. Mortality estimation from
registered deaths in less
developed countries. Demography 21(2):217-233.
Canudas-Romo, V. and I. Dokko. 2010. The Contenders of the
Future Record Life
Expectancy: The Widening Gap to Japan. European Population
Conference 2010,
Vienna, Austria.
Davis, K. 1945. "The World Demographic Transition." Annals of
the American Academy of
Political and Social Science 237: 1-11.
Garca-Guerrero, V. M. (in press). The Population Projections of
Mexico, in C. Rabell
(coord.) The Population of Mexico, 2nd Ed., Fondo de Cultura
Econmica, Mxico.
INEGI, Mexican National Institute of Statistics and Geography.
2012. (accessed October,
2012). http://www.inegi.org.mx/
Lee, R. 2003. The Demographic Transition: Three Centuries of
Fundamental Change.
The Journal of Economic Perspectives 17(4):167-190.
Ministry of Health, Labour and Welfare, Tokyo, Japan. Available
at:
http://www.mhlw.go.jp/english/database/
Notestein, FW. 1945. "Population The Long View." In: Schultz,
T.W. (ed.), Food for
the World. Chicago: University of Chicago Press.
Omran, A. 1971. The Epidemiological Transition. Milbank Memorial
Fund Quarterly 49:
509-38.
Oeppen, J. and J.W. Vaupel. 2002. Broken Limits to Life
Expectancy. Science 296:
1029-1031.
Preston, S., P. Heuveline and M. Guillot. 2001. Demography.
Measuring and Modeling
Population Processes. Blackwell Press.
Rogers, R.G. and R. Hackenberg. 1987. Extending epidemiologic
transition theory.
Social Biology 34: 234-243.
-
Shkolnikov, V; Andreev, E; McKee, M; Leon, D (2013) Components
and possible
determinants of the decrease in Russian mortality in 2004-2010.
Demographic Research,
28 (32). pp. 917-950.
SOMEDE, Mexican Society of Demography (2011). Demographic
estimates for Mexico
and its States, 1990-2010, CONAPO, unpublished.
-
Table 1. Years lost of life for Mexican and Russian males in
2005 by selected ages.
Mexico
Age, X X-LifeExp(0-X) Cancer CVD External Other Sum 55 2.9 0.1
0.1 0.5 2.2 2.9 70 6.5 0.4 0.3 1.1 4.6 6.5 85 14.3 0.9 1.1 2.3 10.0
14.3
Russia Age, X X-LifeExp(0-X) Cancer CVD External Other Sum
55 5.9 0.2 0.6 2.0 3.1 5.9 70 13.9 0.7 2.0 4.4 6.8 13.9 85 26.4
1.4 4.4 8.1 12.5 26.4 Source: INEGI, SSM, HMD & WHO
-
0 20 40 60 80 100
1e0
41e
03
1e0
21e
01
1e+0
0
Age
Dea
th ra
tes
(log
scale
)Figure 1. Agespecific death rates, Mexico 2000 and 2010.
Female 2000Female 2010
Male 2000Male 2010
-
014
59
101
4
151
9
202
4
252
9
303
4
353
9
404
4
454
9
505
4
555
9
606
4
656
9
707
4
757
9
808
4
858
9
909
4
95+
Ages
Cont
ributio
n to
Life
Exp
ecta
ncy
Gap
0.2
0.0
0.2
0.4
0.6
Figure 2A. Agecontribution to the 0.59 years of difference in
Mexican female life expectanciesbetween 2000 (LE=77.2) and 2010
(LE=77.79). Mortality that contributed with the
increase accounted for 0.72 of a year and that with the
decrement 0.13.
-
014
59
101
4
151
9
202
4
252
9
303
4
353
9
404
4
454
9
505
4
555
9
606
4
656
9
707
4
757
9
808
4
858
9
909
4
95+
Ages
Cont
ributio
n to
Life
Exp
ecta
ncy
Gap
0.2
0.0
0.2
0.4
0.6
Figure 2B. Agecontribution to the 0.07 years of difference in
Mexican male life expectanciesbetween 2000 (LE=71.97) and 2010
(LE=72.04). Mortality that contributed with the
increase accounted for 0.7 of a year and that with the decrement
0.63.
-
014
59
101
4
151
9
202
4
252
9
303
4
353
9
404
4
454
9
505
4
555
9
606
4
656
9
707
4
757
9
808
4
858
9
909
4
95+
Ages
Cont
ributio
n to
Life
Exp
ecta
ncy
Gap
0.2
0.0
0.2
0.4
0.6
Figure 3A. Agecontribution to the 0.48 years of difference in
Japanese female life expectanciesbetween 2010 (LE=86.38) and 2011
(LE=85.9). Mortality that contributed with the
increase accounted for 0 of a year and that with the decrement
0.49.
-
014
59
101
4
151
9
202
4
252
9
303
4
353
9
404
4
454
9
505
4
555
9
606
4
656
9
707
4
757
9
808
4
858
9
909
4
95+
Ages
Cont
ributio
n to
Life
Exp
ecta
ncy
Gap
0.2
0.0
0.2
0.4
0.6
Figure 3B. Agecontribution to the 0.2 years of difference in
Japanese male life expectanciesbetween 2010 (LE=79.63) and 2011
(LE=79.43). Mortality that contributed with the
increase accounted for 0.04 of a year and that with the
decrement 0.23.
-
0.2
0
0.2
0.4
0.6
0 14 59
101415192024252930343539404445495054555960646569707475798084
85+Ages
Cont
ributio
n to
Life
Exp
ecta
ncy
Gap
Figure 4A. Age and causecontribution to the 0.59 years of
difference in female life expectancy between 2000 (LE=77.2) and
2010 (LE=77.79) in Mexico.
Infectious Diseases 0.16Neoplasm 0.19Diseases of the Circulatory
System 0.181Mental Disorders 0.007Endocrine/metabolic Diseases
0.209Diseases of the Digestive System 0.003Conditions of Perinatal
Period 0.261Homicide 0.046Other external Causes 0.03Others
0.031
Source: INEGI & CONAPO
-
0.2
0
0.2
0.4
0.6
0 14 59
101415192024252930343539404445495054555960646569707475798084
85+Ages
Cont
ributio
n to
Life
Exp
ecta
ncy
Gap
Figure 4B. Age and causecontribution to the 0.07 years of
difference in male life expectancy between 2000 (LE=71.97) and 2010
(LE=72.04) in Mexico.
Infectious Diseases 0.196Neoplasm 0.1Diseases of the Circulatory
System 0.088Mental Disorders 0.134Endocrine/metabolic Diseases
0.395Diseases of the Digestive System 0.219Conditions of Perinatal
Period 0.279Homicide 0.599Other external Causes 0.198Others
0.021
Source: INEGI & CONAPO
-
0 20 40 60 80
0.0
0.2
0.4
0.6
0.8
1.0
Ages
Prob
abilit
y of
Sur
vivi
ng
ExternalOthersPerinatalDiabetesCVDCancer
Figure 5.Life table distribution of survivors and deaths by
cause of death,for females from Mexico in year 2005
Source: INEGI, SSM.
8555 70
-
0 20 40 60 80
0.0
0.2
0.4
0.6
0.8
1.0
Ages
Prob
abilit
y of
Sur
vivi
ng
ExternalOthersPerinatalDiabetesCVDCancer
Figure 5. Life table distribution of survivors and deaths by
cause of death,for males from Mexico in year 2005
Source: INEGI, SSM.
8555 70
The Stagnation of the Mexican Life Expectancy in the First
Decade of the Twenty First Century: The Impact of Violent
Deaths.AbstractIntroductionDataMethodsResultsDiscussionReferences
Figure5.pdfFig1Mex2000bFFig1Mex2000bMFig1Mex2005bFFig1Mex2005bMFig1Mex2010bFFig1Mex2010bM