IJE Advance Access originally published online on March 23, 2006
International Journal of Epidemiology 2006 35(2):243-249; doi:10.1093/ije/dyl042
Published by Oxford University Press on behalf of the International Epidemiological Association © The Author 2006; all rights reserved.
Cohort Profile: The Mexico City Prospective Study
Roberto Tapia-Conyer1,
Pablo Kuri-Morales2,
Jesús Alegre-Díaz2,
Gary Whitlock3,*,
Jonathan Emberson3,
Sarah Clark3,
Richard Peto3 and
Rory Collins3
1 Subsecretaría de Prevención y Promoción de la Salud, Direccion General de Epidemiologia, Mexico.
2 Secretaria de Salud, Direccion General de Epidemiologia, Mexico.
3 Clinical Trial Service Unit, University of Oxford, UK.
* Corresponding author. E-mail: gary.whitlock{at}ctsu.ox.ac.uk
 |
How did the study come about?
|
|---|
The Mexico City Prospective Study (Estudio Mexicano de Cohorte
Para Enfermedades Crónicas en una Población Metropolitana)
is a blood-based cohort study involving follow-up of 150 000
adults (50 000 men and 100 000 women) who were aged at least
35 years when recruited in 1998–2004. The study resulted
from discussions in the early 1990s about how best to measure
the changing health effects of tobacco in Mexico. These discussions
evolved into a plan to establish a prospective cohort study
that could investigate not only the health effects of tobacco
but also those of blood lipids and various other risk factors.
Ideally, such a study should be big enough to assess the effects
of risk factors not only overall but also in specific circumstances
(e.g. at particular ages and separately in each sex) and at
different levels of other risk factors (to examine possible
causal interactions). In Mexico, there have been substantial
decreases in overall mortality from infectious diseases over
the past few decades, leaving the chronic diseases of middle
age as the main causes of premature death.
1 However, the effects
of risk factors for chronic diseases can depend on the presence
or absence of other risk factors (including some that are measurable
in blood), and these could differ greatly between Mexico and
Western countries. To gain a clearer understanding of the major
determinants of morbidity and premature mortality in Mexico,
the Mexican Ministry of Health (Secretaria de Salud SSA) decided
in 1994 to help fund a large prospective cohort study of adults
in Mexico City. The Ministry staff designed the study and carried
out its baseline survey, while Oxford's Clinical Trial Service
Unit (CTSU), with major support from the UK Wellcome Trust for
this collaboration, provided technical assistance with blood
collection and storage.
|
What does it cover?
|
|---|
The chief aim of the Mexico City Prospective Study is to assess
reliably the associations of established risk factors, and of
possible new risks factors, with the common causes of death
in Mexico.
Table 1 lists Mexico's most common causes of death
in 2000 by age (35–54 and 55–74) and sex. Ischaemic
heart disease, stroke, diabetes (chiefly type 2) and alcoholic
liver disease were major causes of death in all of these age-sex
groups, as were external causes (particularly transport-related
injury and homicide) in men aged 35–54, chronic obstructive
pulmonary disease and lung cancer in men aged 55–74, and
cervical cancer and breast cancer in women aged 35–74.
Many of the deaths in which diabetes was recorded as the underlying
cause are likely also to have involved ischaemic heart disease
or stroke (and vice versa), so the numbers for each of these
causes in
Table 1 may well be underestimates. Although one of
the main aims of the study is to investigate risk factors for
mortality (which, indirectly, will provide information for certain
causes of morbidity), the study also aims to seek information
on the incidence of cancer, non-fatal vascular events, and various
diseases that do not usually cause death.
|
Who is in the sample?
|
|---|
The study is set in two contiguous urban districts (Coyoacán
and Iztapalapa) of central and north-east Mexico City (
Figure 1).
These districts contain a diverse, but settled, mix of long-term
residents and relatively recent migrants from the north and
south of the nation of Mexico. Chronic diseases often associated
with affluence (e.g. ischaemic heart disease, and diabetes)
tend to be most prevalent in the north of the country, whereas
chronic diseases normally associated with poverty (e.g. stomach
cancer and chronic obstructive pulmonary disease) tend to be
most prevalent in the south. All men and women aged at least
35 who were living in these districts were eligible to participate
in the study. The study was approved by scientific and ethics
committees within the Mexican National Council of Science and
Technology and the Mexican Ministry of Health.
View larger version (11K):
[in this window]
[in a new window]
|
Figure 1 Maps showing the locations of (a) Mexico City within Mexico, and (b) the Coyoacán and Iztapalapa districts within Mexico City
|
|
A confidential record of all households in the two districts
was compiled in 1995–97 (i.e. before recruitment in the
main study started) by census-style door-to-door interviews.
This record included information on the name and age of each
household member, and maps showing where stairs and entrances
were located. Once the main study started, the recruitment teams
(each comprising two or three specially trained nurses) visited
individual households, working systematically through this record.
During the recruitment period, individuals in 112 333 households
with eligible inhabitants were invited to take part, and one
or more individuals from 106 059 of these households provided
consent. The baseline assessment of each participant took place
in the participant's home.
Table 2 shows characteristics of the 159 546 participants who completed the baseline questionnaire. Two-thirds were women, the mean age at survey was 53 (with 92% aged 35–74), and 60% lived in Iztapalapa. About one in seven participants had attended university or college (15.3%), and a similar proportion (13.5%) had never received any formal education. A total of 37 433 people reported a history of one or more specific chronic diseases, including 21 810 people (13.7% of all participants) who reported a history of diabetes, 2503 (1.6%) a history of ischaemic heart disease (‘heart attack’ or angina), and 2023 (1.3%) a history of cancer (any site).
|
How often are they followed up, and what is the rate of loss likely to be?
|
|---|
Participants are being followed up indefinitely for cause-specific
mortality through Mexican death registries. In Mexico, almost
all adult deaths are certified by a doctor, and the underlying
and contributing causes of death are coded according to the
10th International Classification of Diseases. This information,
together with personal identifying information (e.g. name, dates
of birth and death, addresses of residence and death), is recorded
in a Ministry of Health database that was established as a consequence
of this study (as well as in other official databases). The
investigators based in the Ministry of Health have regular confidential
access to this database and use the personal identifying information
to link death records automatically to study participants (typically
within 2 or 3 months of a participant's death). Paper copies
of death certificates can be obtained for central review (as
will be done, for example, for all deaths involving diabetes,
ischaemic heart disease, or stroke as an underlying cause: see
above), and individual households can be revisited to verify
vital status. Any participant who emigrates abroad is likely
to be lost to mortality follow-up, but it is estimated that
this would apply to much <1% of study participants annually.
Slight losses to follow-up might also occur if people move out
of Mexico City into regions where death certification is not
yet complete. But, even in rural regions, an average of 98%
of adult deaths are certified by a doctor and so will be linked
automatically.
At 5 year intervals, a reasonably representative sample of at least a few thousand surviving participants will be invited for re-assessment, including the same questions, measurements, and blood collection procedures (described below) that were used at baseline. These repeat assessments will be used to help take account of biological variation and random errors in measurements made at baseline.2
|
What has been measured?
|
|---|
For a large study to be practically and economically feasible,
its procedures need to be simple and streamlined. In this study,
the full recruitment assessment (including obtaining informed
consent and collecting blood) typically took just 30–40
min to complete, and it was nearly paperless (in common with
the recently described Kadoorie Study
3). As the information
was being obtained, data from a simple interviewer-administered
questionnaire (summarized in
Table 3) and physical measurements
(summarized in
Table 4) were entered directly into a handheld
data recorder (of one of two types, described in
Table 5). At
the end of each working day, this information was automatically
uploaded to the study's electronic database. These electronic
procedures minimized the costs of data collection, processing,
and checking, while improving data accuracy and completeness.
Information was entered into the handheld recorders either by
a touch-sensitive screen or by a combination of barcode and
alphanumeric keypad. The handheld recorders had been programmed
to take the field workers through the questionnaire in the same
prescribed manner, automatically making appropriate skips (e.g.
past questions about obstetric history if the participant was
male), and enabling the field workers to modify previous entries
or to void the questionnaire entirely if necessary. In addition,
the data recorders were programmed to query moderately extreme
physical measurements and to prevent highly implausible measurements
from being recorded. After the data had been uploaded to the
study database, and multiple backup copies made, a database-checking
programme was used to identify possible data errors. Such errors
were reviewed regularly by a data monitoring team and, when
necessary, checked by field workers (e.g. by revisiting participants).
However, because of the in-field automated checks of unusual
values, there was less need for subsequent checking than originally
expected.
The blood sample (10 ml, venous) was collected into a single EDTA tube labelled with a barcode unique to the participant. This barcode was scanned, and pertinent information (e.g. date and time of collection, and time of last meal) was entered into the handheld recorder. The sample was then placed in an insulated box containing several chilled packs, reliably maintaining an internal box temperature of 4–10°C. At the end of the day, the boxes were transported to a central laboratory in Mexico City where the samples were extracted and placed in a refrigerator (4°C) for overnight storage. The next morning the samples were centrifuged (2100g at 4°C for 15 min) and then the plasma was taken off and divided equally between two 1.8 ml cryovials (Nunc A/S, Denmark), and the buffy coat (the layer, rich in white blood cells, between plasma and red blood cells) was transferred to a third cryovial of the same type. Each cryovial in a set of three was pre-printed with a barcode unique to the particular set. This barcode, together with the barcode on the original blood collection tube, was scanned into a computerized inventory program to establish a unique link between them. The three cryovials were placed in identical positions in three separate storage boxes. These boxes were stored in three separate –80°C freezers (to prevent complete loss of an individual's samples in the event of a breakdown) until freighted by air in large insulated boxes (each containing 30 storage boxes and 40 kg of dry ice, which had been shown in pilot studies to keep specimens frozen for at least 5 days) to Oxford about once every 6 weeks. Upon receipt at the CTSU laboratory in Oxford (usually within 2–3 days of dispatch) the small storage boxes were immediately unpacked and placed in tanks containing liquid nitrogen vapour at –150°C. A PC-based inventory system developed for the study records the storage location of cryovials within boxes and of boxes within tanks.
Blood is now in storage for 155 487 participants (i.e. 97.5% of those with a completed questionnaire: Table 2). The mean delay (at 4–10°C) before blood separation was 
24 h. Previous research has shown lack of significant effects for many analytes with delays of separation of more than 2–3 days, even at room temperature.4,5
|
What has the study found?
|
|---|
Analyses of associations between possible risk factors and cause-specific
mortality will begin in a few years when sufficient numbers
of cause-specific deaths have occurred. Disentangling deaths
that involve diabetes, ischaemic heart disease, or stroke will
require inspection of death certificates (see above); however,
among those aged <75 at baseline, some 2000 deaths involving
at least one of these causes could (from WHO mortality data:
e.g. see
Table 1) be expected within the first 5 years of follow-up
(i.e. by 2007–08) and, perhaps, some 4000–6000 such
deaths within the first 10 years (as well as 300–400 deaths
each from cancer of the lung, intestine, cervix, and breast).
In addition to information on possible risk factors collected at baseline (Tables 3 and 4), possible risk markers in blood—potentially including lipid subfractions, apolipoproteins, coagulation and inflammatory factors, hormones, vitamins, markers of insulin resistance, and genetic polymorphisms—will also be investigated. Analyses of possible blood markers for risk will use a nested case–control approach, whereby the stored blood from people who have died from a particular disease will be retrieved and compared with blood retrieved from otherwise similar individuals who do not have the disease. This approach is economical (because only a sample of the total number of specimens is assayed, and because the real cost of assays tends to diminish with time), while avoiding the potential problem of reverse causality (because in nearly all instances the blood will have been collected before the onset of disease). Moreover, it allows factors that only in the future may come to be regarded as potentially important also to be assayed.
Because the study sample is largely representative of the adult population in Mexico City, the baseline prevalences of disease (Table 2) and of disease risk factors (Table 6) are of interest for assessing local health needs and for disease prevention. The prevalence of current smoking was strongly related to age, with about one in two men (50.7%) and one in four women (25.0%) at ages 35–54 being current smokers, compared with just one in five men (19.8%) and one in 20 women (5.2%) aged 
75 being current smokers. Just over one-third of the men aged 35–54 (35.6%) drank alcohol at least once a month, but the proportion of men aged 75 who did so was much lower (15.8%); and, in each age group, only one-quarter to one-fifth as many women as men drank this frequently. In each age group, about one in four men and one in five women reported having recreational physical activity at least once a week.
The mean BMI overall was 27.9 kg/m
2 in men and 29.5 kg/m
2 in
women, but in both sexes, BMI tended to be lower at ages
75
than at younger ages (
Table 3). Among those aged 50–59,
the prevalence of a BMI of
30 kg/m
2 was 30% in men (which is
slightly less than the prevalence of 31% among US men at the
same age and in about the same years
6), but it was 47% in women
(or much higher than the corresponding prevalence of 37% among
US women
6).
|
What are the main strengths and weaknesses?
|
|---|
This study is probably the largest blood-based prospective cohort
study to have been established so far in Latin America. It should,
therefore, be well placed to address uncertainties about disease
causation in that part of the world, especially since it has
blood stored for nearly all of its participants. But factors
that will limit the accrual of deaths, and, hence, the ability
to investigate causes of death, include a fairly young age distribution
(median 50 years) and an unintentionally low percentage of men
(33%). Recruiting men turned out to be harder than expected
because a higher proportion of men were working long hours during
the day in the main study than in two earlier pilot studies,
a problem that made it necessary to arrange many repeat visits
during the evening and weekends. Other major challenges to recruitment
included a series of hurricanes and tropical storms that hit
Mexico in 2002–03, creating national and regional public
health disasters to which study staff (as employees of the Ministry
of Health) were seconded (partly accounting for the flat part
at the right-hand end of the study's recruitment graph:
Figure 2).
Nonetheless, the study should still be able to address many
important questions about disease causation in Mexico, and much
of the information could also be relevant to other countries.
For example, the high prevalence of obesity and diabetes may
mean that this study can provide insights into the future impact
of these conditions in countries where the epidemics of obesity
and diabetes are currently less mature but which could have
similarly high prevalences of these conditions within a decade
or two (e.g. many Western countries).
This study was affordable only because it used simple and streamlined
data and blood collection methods. For example, the handheld
data recorders allowed rapid collection of complete information,
avoided the need to process and store more than one million
individual sheets of paper, and are expected to reduce the medium-term
data collection and storage costs by about three-quarters. Paperless
data collection methods such as these (including laptop- or
internet-based methods
3) have great potential for reducing costs
while preserving or improving data accuracy.
|
Can I get hold of the data? Where can I find out more?
|
|---|
The study's data are not freely available, but specific proposals
for future collaboration, addressed (in English or Spanish)
to any of the study's Mexico-based or Oxford-based investigators,
would be welcomed.
|
Acknowledgments
|
|---|
We thank the following people for their help in establishing
the study: David Nava Menchaca, Martin Radley, Raul Ramirez
Reyes, Fernando Velazquez-Gonzalez, Clementina Magos-López,
and Linda Youngman. The study has been funded by the Secretaria
de Salud SSA (Mexican Ministry of Health), the Consejo Nacional
de Ciencia y Tecnología (Mexican National Council of
Science and Technology), and the UK Wellcome Trust. The Clinical
Trial Service Unit receives core funding from the Medical Research
Council, the British Heart Foundation and Cancer Research UK,
and additional support for international studies of smoking
from the US NIH Fogarty International Center.
|
References
|
|---|
1 World Health Organization Statistical Information System. Available at:
http://www3.who.int/whosis/menu.cfm?path=whosis,mort (Accessed January 2006).
2 Clarke R, Shipley M, Lewington S et al. Underestimation of risk associations due to regression dilution in long-term follow-up of prospective studies. Am J Epidemiol 1999;150:341–53.[Abstract/Free Full Text]
3 Chen Z, Lee L, Chen J et al. Cohort profile: The Kadoorie Study of Chronic Disease in China. Int J Epidemiol 2005;34:1243–49.[Free Full Text]
4 Clark S, Youngman LD, Palmer A et al. Stability of plasma analytes after delayed separation of whole blood: implications for epidemiological studies. Int J Epidemiol 2003;32:125–30.[Abstract/Free Full Text]
5 Clark S, Youngman LD, Chukwurah B et al. Effect of temperature and light on the stability of fat-soluble vitamins in whole blood over several days: implications for epidemiological studies. Int J Epidemiol 2004;33:518–25.[Abstract/Free Full Text]
6 World Health Organization. Available at: http://www.who.int/ncd_surveillance/infobase/web/surf2/country_list.html (Accessed January 2006).
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
|
|
 
Z. Chen and G. Whitlock
Unveiling the causes of heart disease in China
Heart,
November 15, 2009;
95(22):
1818 - 1819.
[Full Text]
[PDF]
|
|
|
Top
How did the study...
What does it cover?