The online version of this article has been published under an open access model.

Users are entitled to use, reproduce, disseminate, or display the open access
version of this article for non-commercial purposes provided that: the original authorship is properly and fully attributed; the Journal and Oxford University Press
are attributed as the original place of publication with the correct citation details given; if an article is subsequently reproduced or disseminated not in its entirety
but only in part or as a derivative work this must be clearly indicated. For commercial re-use, please contact journals.permissions@oxfordjournals.org
Published by Oxford University Press on behalf of the International Epidemiological Association. International Journal of Epidemiology 2007;36:504–511
ß The Author 2007; all rights reserved. Advance Access publication 13 March 2007 doi:10.1093/ije/dym016

COHORT PROFILE

Cohort Profile: Mandela’s children: The 1990

birth to twenty study in South Africa
Linda Richter,1* Shane Norris,2 John Pettifor,2 Derek Yach3 and Noel Cameron4

Accepted 26 January 2007

How did the study come about? first cohort born into a democratic South Africa. At the start,
the study was called Birth to Ten (BT10), but changed to Birth to

Downloaded from ije.oxfordjournals.org by guest on March 3, 2011

The late 1980s were a period of profound sociopolitical change
in South Africa. It was clear that the Apartheid state was
crumbling and, amongst acts of civil disobedience which
characterized ‘the struggle’, Black Africansy began to disregard
restrictive legislation that constrained where they lived and
worked. Very rapid unplanned urbanization began, and shanty
towns mushroomed around formerly White cities and towns.
It was anticipated that this rapid urbanization, with urban
growth estimated at the time to be 3.5% per year1,2 would have
profound effects on children’s health and development. While
improved access to health care, education and employment in
urban areas could decrease preventable childhood morbidity
and mortality, the inability of government to establish and
maintain services to meet the needs of the growing urban
population could exacerbate existing infectious diseases, such
as HIV/AIDS and tuberculosis. Non-infectious conditions related
to the interaction of lifestyle, urban stressors and socio-cultural
changes—for example, childhood injuries, substance use and
obesity—were predicted to increase.
In 1988, as a result of discussions on these issues, Noel
Cameron at the University of the Witwatersrand, and Derek
Yach at the South African Medical Research Council (MRC)
approached the then MRC President, Andries Brink, for funds
to begin a birth cohort study in Soweto-Johannesburg with
Lucy Wagstaff, also from the University of the Witwatersrand.
The aim was to track a group of urban children for 10 years,
not knowing at the time, that these children would also be the
1
Human Sciences Research Council and University of KwaZulu-Natal,
South Africa.
2
University of the Witwatersrand, South Africa.
3
Rockefeller Foundation, USA.
4
Loughborough University, UK.
y
The Apartheid regime created a system of so-called race classification which
legally differentiated between Whites (of European origin), Indians
(a collection of different people from the South East Asian region, mainly
from India), Coloureds (people of mixed ancestry) and Blacks (people of
African descent). Participation in society was differentiated in freedom and
quality on a continuum from Whites to Indians to Coloureds to Black people.
The terminology is retained because it carries the legacy of decades of
oppression and discrimination, the effects of which are still evident.


Corresponding author. Child Youth Family and Social Development, Human
Sciences Research Council, Private Bag X07, Dalbridge 4014, South Africa.
E-mail: lrichter@hsrc.ac.za
Twenty (BT20) in 2000 when we reached the 10-year follow-up
goal and resolved to continue the study to age 20. The children
came to be known colloquially as Mandela’s Children3 because
they were born in the 7 weeks following Nelson Mandela’s
release from prison on the February 11, 1990, an event that
heralded in radical social and political changes. Linda Richter
was part of the original investigator group. Shane Norris has
been the Project Manager since 2001.
With seed funding, we undertook several pilot studies to
determine, amongst other things, the seasonality of births in
the area and the optimum months for recruitment into a birth
cohort study; the nature and accuracy of routinely collected
health service data; and follow-up rates of children from birth
to six months.4 The date for enrolling the birth cohort was
set for February, but a national hospital strike delayed
enrolment, which occurred from April 23 to June 8, 1990.
What does the study cover—and
how has this changed?
From its inception, BT20 was planned to be multidisciplinary,
tracking the growth, health, well-being and educational
progress of urban children across the first decade of their life.
To do so, we had to innovate in a number of areas, such
as establishing tracking systems under circumstances where
few people had a street address or telephone, create a flexible
dataset because names were inaccurately translated and
transcribed from record to record, translating and adapting
questionnaires developed in the West and establishing norms
for many of the measures used.
We began the study during the antenatal period, collecting
information on pregnancy and birth. Because we adopted
a lifecycle approach, the study covers many of the major issues
confronting the particular developmental phase of children and
young people in which data is collected. In the early years
we concentrated on environmental influences (poverty, migra-
tion and political violence), access to health services, nutrition,
child care and growth and development.
In the first few years of school, the emphasis was on cognitive
ability and school performance, as well as social adjustment

504

and participation. Later questionnaires began to explore
early lifestyle risk factors—diet and weight gain, parental
monitoring and supervision, educational failure, and sexual
experimentation, and a wide range of physical (body composi-
tion and bone mineral density scans) and physiological
measures (pubertal development, biochemical markers of
insulin resistance) were introduced. Now that the cohort has
turned 16, data collection is focused on the prediction and
measurement of risk, particularly in three domains: (i) sexual
and reproductive behaviour (sexual debut, sexual violence,
unplanned pregnancy and sexually transmitted infections), (ii)
early expression of the metabolic syndrome (obesity, hyperten-
sion and insulin resistance) and (iii) social marginalization
(school drop out, psychosocial difficulties, substance abuse,
violence and conflict with the law).
The major change in the study occurred as we neared our
10-year goal. BT10, as it was then called, was a collaborative
enterprise involving more than 20 researchers, some of whom
had interests in only one specialized area or particular age
group. No single individual had responsibility for the overall
direction and resourcing of the study; there was a constant
struggle to make ends meet and to enter and clean data
requested by individual scientists. In response to this situation,
Linda Richter and John Pettifor took the decision to raise
research funding for two specific hypothesis-driven longitudinal
studies, one on sexual risk and the other on bone health. BT10,
as it had existed, was closed down and BT20 took its place,
incorporating all the staff and data from the first ten years.
As from October 2005, the next generation of children has
started to be born.z Our first young mother was 14-years-old
when she delivered her baby. There were three 14-year-old
mothers in the original cohort of parents. Deliveries are speed-
ing up; more than a third of women in South Africa have their
first child before their 20th birthday. While we intend to repeat
all the questionnaires used in BT20 with 3G children, we
are also doing in-depth studies of young parenting (including
fathering) and we will be expanding physiological measures
in the new cohort, as well as our fledgling genetic interests.
Who is in the sample?
BT20 includes all singleton children born to women resident in
Soweto-Johannesburg during the 7-week enrolment period.
In late 1989, we began interviewing women in public antenatal
clinics in an effort to collect data from pregnant women who
were predicted to deliver their babies during the enrolment
period. More than 2000 women participated; however, because
the cohort enrolment dates shifted as a result of the hospital
strike, only 1594 of these women delivered within the enrol-
ment period. A further selection criterion was that the mother
and baby had to remain in the area until the child was
six months old. The reason for this criterion was that the pilot
studies had demonstrated that large numbers of women came
from rural areas to deliver babies in the city, either because
they believed the services were better, to ensure their
child a birth certificate entitling them to live and work in an
z Called 3G because they are the third generation; 3G also refers to high
speed wireless internet connection, a fitting association for these children,
born to parents half of whom have a cell phone.
BIRTH TO TWENTY STUDY IN SOUTH AFRICA 505
urban area under Apartheid law, or because they wanted the
child’s father, a migrant labourer, to witness his child’s birth
and thus secure the child’s support. All births are registered in
the municipal area under a local ordinance, but to ensure that
our records were complete, we also checked mortuaries and
back-tracked children brought to clinics for their first 6-week
postnatal check-up. According to the records we compiled, 5449
births were registered during the 7-week enrolment period,
3273 of which met the cohort entry criterion of continued
residence in the area. Of this group, we are still in contact with
more than 70% of the cohort at age of 16 years.
Birth notification forms contain basic information from which
we could derive characteristics of total births. In terms of so-
called race groups, the total cohort is roughly representative of
the South African population, except for an under-representa-
tion of White children. Whites comprise 9.2% of the population
of the country.5 The reason for their under-representation is 2-
fold. First, it was not possible to enrol children through private
health practitioners and facilities, which White families
Downloaded from ije.oxfordjournals.org by guest on March 3, 2011
predominantly used at the time. As the study has progressed,
White families, who tend to be better off, have shown higher
attrition than other families. Cohort studies in developing
countries have generally found, contrary to what pertains in
better-resourced countries, that higher socioeconomic class
individuals are harder to enrol and maintain, principally
because they perceive fewer benefits and more disadvantages
to participation in research.6 To compensate for this, at age 10
years in the bone health study, we recruited a supplementary
sample of 120 White children born during the cohort enrolment
dates, but not in the area. Most migrated into the metropole
between 1990 and 1997. Table 1 shows some basic character-
istics of the cohort.7,8 Although retrospective data has been
collected from the supplementary group, they are not included
in the description of the original cohort in the table.
How often have they been followed
up—and what is attrition like?
Table 2 shows the year, place of data collection, numbers con-
tacted, from whom data was collected and reasons for attrition
at all data collection waves to date.
We distinguish three types of attrition; absolute attrition due,
for example, to the death of a child. To determine this, we have
relied on reports of family and neighbours when visiting the
last known address of the family. Second, there is intermittent
attrition due to circular migration between rural and urban
areas. This is evident in the large numbers of people (n ¼ 1449)
who are reported to have left the study area during any partic-
ular data collection wave (in total 44% of the cohort), some
of whom return to the urban area and are traced during a later
data collection wave. Last, there is study attrition, due to the
incapacity of the study to either trace the family or interview
the participants in a data collection wave, because of a lack of
resources, including time. For example, the 9–10-year data
collection wave was never completed. A proportionately larger
number of families were reported to have moved during the
data collection years 3 and 4 because we devoted a large
amount of time to tracing families who had not been found
during the previous few years.
506 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

Table 1 Characteristics of the BT20 cohort were recruited from local communities, some of whom have
BT20 cohort been employed in the study from its inception. Staff is also
Characteristics (n ¼ 3273) trained in cohort maintenance, as well as Good Clinical Practice
Population group and ethical research conduct.
African 2568 (78%) The maintenance strategy, especially in the first decade, was
directed to parents and other primary caregivers. Dissemination
White 207 (6%)
of information about the study was crucial to ensure that the
Coloured 383 (12%)
requirements were well understood and participants were
Indian 115 (4%) motivated to take part.9 Information is disseminated by staff
Maternal age at each contact, as well as through newsletters and workshops.
<17 years 92 (3%) Compensation for transport expenses is provided for families
17–19 392 (12%) who come into one of the two data collection sites, at the rate
of R50.00 a visit.§ Tracing of families is done through multiple
20–38 2692 (82%)
contact addresses and telephone numbers collected from all
39þ 95 (3%)
participants at each contact.
Gravidity Contact with families is maintained between data collection
1 1094 (33%) phases through birthday cards, reports on the study in
2–4 1875 (57%) newspapers and on radio and television and regular newsletters
55 304 (9%)
for parents as well as for child/adolescent participants and

Downloaded from ije.oxfordjournals.org by guest on March 3, 2011

Gestational age
<37 weeks 388 (12%)
37–41 2773 (85%)
42þ 11 (0.3%)
Birth weight
<1500 g 30 (1%)
1500–2499 322 (10%)
2500–3999 2827 (86%)
4000þ 89 (3%)
Follow-up in the early years was severely handicapped by a
lack of resources. At the time, we restricted follow-up of
children to the original cohort area, some 400 km2. With the
collapse of Apartheid, the urban landscape changed substan-
tially, because families had the freedom to live and work where
they chose (Figure 1). Currently, we trace children and families
throughout the Gauteng province (an area of 17 000 km2).
In 1993 (cohort age 3), we suspended data collection in order
to devote all our capacity to tracing ‘lost’ children and families,
and in 2003 (cohort age 13), we conducted a cross-sectional
survey of same-aged children in 622 primary schools in the
area, primarily to trace families lost to follow-up. Through the
school survey, we re-connected with close to 400 children
previously thought to be lost because of multiple address
changes. This is a generally very mobile population, with about
a quarter of families reporting a change of address during each
major data collection period. There are 441 families who we
saw once within the first year of life but not again. These
families probably moved and are now considered absolute
losses to follow-up.
We use multiple strategies to maintain participation. When
the study began, we created relationships with the community
and early data collection was planned in cooperation with the
health services. A Community Advisory Board (CAB) was
created whose membership has changed over time to reflect
current concerns in the study. Currently, education officials and
parents are strongly represented. In addition, field workers
school principals. A limited social and health referral service is
incorporated into the study—for example, a toll-free telephone
number and a free-post address to enable families to contact
the office for advice and information; and referral notes to local
services are given to families when health or social problems
are detected.
As the children have grown older, the cohort maintenance
strategy has been re-orientated to the adolescent participants.
On average, study participants spend 3–4 h at the data
collection site, during which time refreshments are provided.
The study team provides letters to schools explaining children’s
absence if they miss school. We also promote awareness of
BT20 at schools. However, the most significant initiative has
been to shift data collection to weekends to make it more
convenient for caregivers and adolescents to visit the site.
What has been measured?
While some rounds of data collection contain questionnaires or
measures of special interest, common themes run through the
data collection rounds, as shown in Table 3. These include
demographic, socioeconomic and household information; com-
munity, neighbourhood and school environments; health and
nutrition; childcare, supervision and monitoring; growth and
physical activity; cognitive development and school perfor-
mance; social and psychological adjustment; risk behaviours;
and a range of physiological measures. Data has been collected
at 17 points across the 16 years of the study—at antenatal,
delivery, 6 months, 1 year, 2 years, 3 years, 4 years, 5 years,
7 years, 9/10 years, 11/12 years, 13 years twice in the year,
14 years twice and 15 years twice—as well as through a school
survey at age 13 years.
What has been found?
A full list of publications is available on the Birth to Twenty
website (http://www.wits.ac.za/birthto20). Under the categories
§ Exchange rate $1¼R7.
 BIRTH TO TWENTY STUDY IN SOUTH AFRICA 507

& Homes

278

596

1794 (55%) 1730 (53%) 2080 (64%) 2100 (64%)

2345

54

3273

15
Study sites

103 (4%)
2005–2006
used on the web site, highlights from each of five areas are
15 years briefly summarized as follows.

Methodological issues

54

25
Study sites

96 (4%)
& Homes

270

609
2340

3273
2004–2005
14 years

At the start of the study, several analyses examined the

reliability and validity of routinely collected health service data,
including reproductive histories reported in response to ques-
tionnaires10 and the accuracy of birth data recorded on health
51

33
Study sites

70 (3%)
& Homes

268

591
2363

3273
2001–2003 2003–2004

and administrative records.11 The information collected was

Incomplete 11/12 years 13 years

used to design subsequent questionnaires. Several papers

engaged vigorously with the issue of race in the new South
Africa,12 as well as implications for care. For example, we
0

169
4 (0.2%)
2470

31

3273

49
Study
sites

found that White women using private facilities were ten times
more likely to have a Caesarean section delivery than African
women using a public health service.13 More recently,
methodological substudies have concentrated on ensuring
4
Homes &

28

16

58
16 (1.0%)

3273
Schools

1517

1498 (46%)
accurate measurement of pubertal status14 and maximizing
9/10 years

1999–2001

disclosure of sensitive information by young adolescents.15

Downloaded from ije.oxfordjournals.org by guest on March 3, 2011

Nutrition
Homes &

27

3273

27
Schools

2087

138

182
9 (0.4%)

1594 (49%) 1907 (58%) 2213 (68%) 1839 (56%) 1858 (57%) 1586 (48%) 2023 (62%)
1997–1999
7/8 years

From longitudinal food frequency analyses—conducted in 1995,

1997, 1999 and 2000—we consistently found that calcium, iron,
zinc and biotin fell below 65% of the Recommended Daily
Allowance. In addition to these nutrients, vitamin A, riboflavin,
28

21
Homes

157

283
15 (0.9%)
1798

3273
1995–1996

nicotinic acid and pantothenic acid fell below the RDA in 2003.
5 years
Table 2 Number of study participants contacted and lost to follow-up over the past 10 data collection waves

From 2000 to 2003 there was an increase in the percentage of

children falling below the recommended intakes for most
micronutrients.16,17
Homes

298

576
16 (.07%)
2173

25

3273

41
1993–1995
3/4 years

Growth and bone health

The cohort shows improved growth in comparison with
Clinics &

1873

20

3273

18
Homes

199

238
3 (.09%)

children born in the 1970s, but with little reduction in the

1992–1993

difference between Blacks and Whites. White children continue

2 years

to be taller and heavier.18 Analyses examining ethnic differ-

ences in bone mass indicate that pre-pubertal Black children,
Clinics &

13

56

51

15
Homes

2 (1.0%)
2229

3273

despite adverse environmental factors (low calcium intakes

1991–1992

<400 mg/day, less physical activity, lower socio-economic

1 year

status) have greater hip bone mass than White children, and
similar bone mass at the spine and appendicular skeleton after
adjusting for differences in stature, likely due to genetic rather
Homes

0
Clinics &

20 (1.0%)
1943

46

39

3273

21
6 months

than environmental factors.19 Physical activity, as measured by

the amount of sport played, walking, and leisure activities, is
1990

lower in Black than White children, mainly because of poor

physical education and sporting opportunities in traditionally
Clinics

1594

3273
–

–
Antenatal
Year reported in this data wave 1989–1990

Black schools.20 The amount of physical activity undertaken by

Black children is insufficient to have an osteogenic effect on
bone mass. Peak bone mass is established in childhood and this
influences the likelihood of minimal trauma fractures in later
life. A lack of physical activity in childhood may also contribute
that changed residential address
Data collected from participants
(Caregiver wishes not to participate

to obesity in young adults.

Percentage (%) of the cohort
(Participants’ home; primary
school, data collection site)

Family moved to unknown

Reported child or caregiver

Family moved out of the

Place of data collection

Infant, child and adolescent well-being

Established contact

Similar to their counterparts in resource-rich countries,

young children in South Africa show age-related increases
in this wave
in the study)

and decreases in adjustment profiles, gender differences in

study area
mortality

externalizing and internalizing behaviours, and early onset

address

and persistence of maladjustment amongst a small group of

Total

children whose first signs of difficulties manifest as

508 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
Figure 1 Johannesburg-Soweto in the Gauteng Province, South Africa
parental perceived ‘difficultness’ and poor peer relations.21
Exposure to violence, in the home and the community,
has been found to have a pervasive relationship with children’s
emotional and social adjustment, independent of the impact of
poverty.22,23
Sexual and lifestyle risk
Sexual risk
Unwanted teen pregnancy and a high prevalence of sexually
transmitted infections (STI), including HIV, are significant
health threats in South Africa, both increasing as BT20
children move into their late teens. To date, 15 girls in
the cohort have delivered a child, the first just after turning
14 and a further 13 are pregnant. Sexual intercourse
was reported among 5% of 13-year-olds, equally by boys and
girls; it picks up rapidly at 14 (17% boys, 6% girls) and 15 (27%
boys, 11% girls). However, amongst both boys and girls
more than half of all sexual activity reported, including
foreplay and oral sex, is indicated to be non-consensual.24
Two BT20 adolescents have tested HIV positive through a
voluntary screening programme and several have been treated
for other sexually transmitted infections.
Metabolic disease risk
Children who are born small and then grow quickly appear to
be at increased risk of obesity and risk factors for Type II
diabetes.25,26 We found that lower birth weight was associated
with greater insulin production and higher glucose concen-
trations at age 7. This suggests that the ability of insulin
Downloaded from ije.oxfordjournals.org by guest on March 3, 2011
to clear glucose from the circulation is reduced in children with
low birth weights. Body mass index at 7 years was positively
associated with both insulin concentrations and insulin resist-
ance, as was the rate of weight change or weight velocity from
birth to 7 years. Finally, high subcutaneous fat levels at the
subscapular and triceps sites, measured at age 7, were also
positively related to both insulin concentrations and insulin
resistance.25
What are the main strengths
and weaknesses of the study?
A critical strength has been the perseverance of a small group
of people determined to maintain the study, with little funding
and against some opposition. At the start, several prominent
public health specialists warned that long-term follow-up was
neither possible nor particularly useful. The core group, who
also provide continuity, drove a collaborative, multidisciplinary
approach and a strong association with health services, schools
and the community.
As resources have grown, most significantly with long-term
funding from the Wellcome Trust, stable systems have been
established for the management of the study. A team of very
committed people, responsible for operations, administration,
laboratory and data, report to the investigators on a weekly
basis. Sophisticated bar code, filing, and electronic systems
have been designed to print address lists, weekly appointments,
tracking participants through the study components, data

Table 3 Data collected in Bt20
General category
Residence and contact details
Growth
Socio-economic
circumstances
Child health
Maternal/caregiver
stress and health
Child nutrition
Child care
Cognitive development
Education
Psychological assessment
Neighborhood, community
Child blood pressure
Physical activity
Bone mass & body
composition
Pubertal development
Biochemical tests
Child risk behaviour
Sexually transmitted
infections, including HIV
Genetics
BIRTH TO TWENTY STUDY IN SOUTH AFRICA 509
Example Data collection wave
Address, relatives/friends contact numbers All
Birth weight, height, weight, skinfold circumferences All
Head circumference 3 months to 7/8 years
Hip and waist circumferences From 9 years onwards
Education; occupation; housing type and ownership; water, sanitation & 0–2 years; 5 years; 7/8 years;
electricity; refuse removal; household composition and density; household assets 11/12 years; 13/14 years
Morbidity; health histories; immunization; accidents; injuries; developmental 6 months; 1 year; 2 years;
problems; medication; accessing health services 4 years; 7/8 years; 11/12 years;
13 years; 14 years; 15 years
Stress and social support; post-natal depression; exposure to violence; family Antenatal: 6 months, 5 years,
medical history; alcohol and tobacco use; lifestyle; weight; height; 7 years, 10 years, 11/12 years,
body composition; blood pressure 13 years, 14 years, 15 years
Breastfeeding practices, dietary intake (food frequency) retrospective recall 0–2 years; 5 years; 7/8 years;
9 years; 10 years; 13 years
Caregiving; day care; after school activities; monitoring and supervision; 0–2 years; 5 years; 7/8 years;
perception of parenting and family relationships 11/12 years; 13–15 years
Downloaded from ije.oxfordjournals.org by guest on March 3, 2011
Pre-speech development; Bayley Scales of Infant Development; Griffiths 6 months, 1 year, 2 years,
Scales of Mental Development; speech, language and hearing development; 4 years, 5 years, 7 years
Denver Developmental Screeing Questionnaire; Ravens Coloured
Progressive Matrices
Child care; preschool attendance; school start age; repetitions and failure; 7/8 years onwards
annual school reports; teacher ratings of children; school infrastructure;
bullying; perceptions of school environment
Richman & Graham Behaviour Rating; South African Child Assessment 2 years, 4 years, 5 years,
Scale (SACAS); Draw-A-Person; Youth Self-Report (YSR); Rosenberg 7 years, 10 years, 11/12 years,
Self-Esteem; body image; future aspirations; eating attitudes; identity 13 years, 14 years, 15 years
Community rating; perceptions of safety and connectedness; 5 years, 11/12 years, 13 years,
community violence 14 years, 15 years
Diastolic, systolic, pulse rate 5 years, 7 years, 11/12 years,
13 years, 14 years, 15 years
Transport to school; participation in school sports; sedentary behaviour; 7 years, 11/12 years, 13 years,
extra-mural activities; casual work 14 years, 15 years
DXA-derived bone mass and body composition 9 years, 10 years, 11 years,
12 years, 13 years; 14 years;
15 years
Menstruation; Tanner pubertal development scales Annually from 11 years
Lead At birth
Lipids 5 years; 13 years; 15 years
Glucose and insulin 13 years; 15 years
Cotinine 13 years; 14 years; 15 years
STIs and HIV 15 years
Vitamin D 9 years, 10 years, 11 years,
12 years, 13 years; 14 years;
15 years
Bone turnover markers 9 years, 10 years, 11 years,
12 years, 13 years; 14 years;
15 years
Tobacco, drug and alcohol use; sexual experience; violence 10 years, 11/12 years, twice
a year from 13-years
Through screening—reported sexual behaviour, pregnancy, ULE Annually since 13 years
DNA bank (n ¼ 2200 child samples, 2200 one biological parent)
510 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
completeness and quality, entry, cleaning and the construction
of analytical datasets.
Considerable energy goes into tracking and cohort main-
tenance. There are few street names and numbers in Soweto,
informal housing springs up quickly and they confuse the
visual landscape, many people have temporary work away from
home and there are few fixed line telephones. Mobile phone
usage is burgeoning, making it easier to contact caregivers and
children. To date, attrition has been low in comparison to other
developing country cohorts,6,8 and occurred largely in children’s
early years. It is difficult maintaining better-off families who
see little benefit in participating in a long-term study with twice
yearly data collection, including fasting urine and blood draws.
However, most African participants identify strongly with
the study, and report being happy to participate if the
information collected can be used to improve the quality of
life of South Africans.
The study is unique and has been a source of reference
for several major policy decisions in the country, particularly in
the absence of other data. For example, information on
children’s recognition of cigarette brands was used by the
Minister of Health from 1997 to 1999 to push forward
progressive tobacco legislation, preventing public advertising
and the sale of cigarettes to minors. Early age of school entry
revealed by longitudinal data was used by the Ministry of
Education in 2004 to legislate a minimum age for school
enrolment.
Looking backwards at what could have been done better is
always with 20:20 vision. We could have over-enrolled White,
Coloured and Indian families so that more systematic compar-
isons could have been made between the so-called Apartheid
race groups. However, both at the time the study was initiated,
and now, there is considerable ambivalence about treating
these political classifications as biological, or even social,
categories. Consultations with experts in longitudinal
child studies¥ advised us that the study would be maximally
useful if it focused on the heterogeneity among a cohort of
urban children, even if they were largely African. In the bone
health study, an additional 120 White families were enrolled
to ensure adequate power to test for ethnic differences in
bone mass.
We certainly would benefit from having data that are
more complete at each point. Circular migration and the
difficulty in finding people means that some cases have missing
data waves. Given the frequency of data collection, we can
impute data from nearby time points for many variables, such
as socioeconomic status or maternal education, but this is not
possible for growth or developmental status. Since age 12, we
have managed to achieve much more complete data at each
round, and this has been helped by twice yearly data collection
waves.
From the beginning, we have struggled to find the appro-
priate level of statistical expertise needed to do justice to
longitudinal data. There are very few experienced applied
statisticians in South Africa and statistics is regarded as a
scarce skill throughout the science and technology sector.
Collaborations with colleagues in the United States and the
United Kingdom have brought with them the required expertise
but, until very recently, this was ad hoc rather than an integral
part of the staffing structure.
Can I get hold of the data?
Where can I find out more?
The outline of the study, history and questionnaires are
available on the website and collaborators can pull down data
sets from the web through password access (http://www.wits.
ac.za/birthto20). Collaborations are established through formal
agreements with the principal investigators (lrichter@hsrc.
ac.za), and include funds needed for the collaboration, staff
requirements and mutual obligations, including the return of
analytical datasets, notification of publications, etc. Our current
grants cover costs involved in planned data collection and
analysis. New initiatives that require additional staff or resour-
ces to collect or clean data have to be additionally funded.
Apart from a core group of South Africans at local universities
Downloaded from ije.oxfordjournals.org by guest on March 3, 2011
and research institutions, such as Nigel Crowther from
the University of the Witwatersrand who works on insulin
resistance, we have active collaborations with Noel Cameron
from Loughborough University on aspects of growth; Alan Stein
and Paul Ramchadani from the University of Oxford on
maternal depression, child adjustment and adolescent body
image; Linda Clarke and Vicky Hosegood from the London
School of Hygiene and Tropical Medicine on the role of fathers
in children’s development; Marie-Louise Newell from the
Institute of Child Health in London and the Africa Centre for
Health and Population Studies in South Africa and Audrey
Pettifor from the University of North Carolina, on adolescent
sexual risk behaviour; the Lancet Child Development Series on
a paper coordinated by Sally McGregor from the Institute for
Child Health in London; and the Lancet Nutrition Series on a
paper coordinated by Cesar Victora at the Universidade Federal
de Pelotas.
Acknowledgements
From its inception, BT10 has been supported by the South
African Medical Research Council. Other funders include
the Institute for Behavioural Sciences at the University
of South Africa. Since 1998, the Wellcome Trust has been
BT20’s major funder, with additional support from the Human
Sciences Research Council of South Africa, the Medical
Research Council, the University of the Witwatersrand, the
Mellon Foundation, the South-African Netherlands Programme
on Alternative Development and the Anglo American
Chairman’s Fund.
Special thanks are due to Sharon Fonn, Sylvia van Rensburg
and Thea de Wet, who coordinated the study in its early phases.
Many colleagues have been involved in various aspects of the
study over the years, including Peter Cleaton-Jones, Peter
Cooper, George Ellison, Dev Griesel, Allan Herman, Justice
Hofmeyer, Carel IJsselmuiden, Angie Mathee, James McIntyre,
¥ Dr Jean Golding, from the Avon Longitudinal Study of Pregnancy and
Childhood (ALSPAC) Bristol University and Sir Michael Rutter at the
University of London.

Nicky Padayachee, Krisela Steyn, Lucy Wagstaff and Dinky
Levitt.
The study would never have survived or thrived without the
dedication of staff, especially those field staff who have been
with the study since it began: Barbara Moneypote, Eliza Tseou,
Thabile Sibiya, Mantoa Langa and Andreas Phake.
References
1
Yach D, Padayachee N, Cameron N, Wagstaff L, Richter L. ‘‘Birth to
Ten’’ - A study of children of the 1990’s living in the Johannesburg-
Soweto area. South Afr Med J 1990;77:325–26.
2
Yach D, Cameron N, Padayachee L, Wagstaff L, Richter L, Fonn, S.
Birth to ten: Child health in South Africa in the 1990s. Rationale and
methods of a birth cohort study. Paed Perinat Epidemiol 1991;5:211–33.
3
Barbarin O, Richter L. Mandela’s children: Growing up in post-apartheid
South Africa. New York: Routledge, 2001.
4
Fonn S, De Beer M, Kgamphe S et al. ‘‘Birth to Ten’’ - Pilot studies to
test the feasibility of a birth cohort study investigating the effects of
urbanization in South Africa. South Afr Med J 1991;79:449–54.
5
Statistics South Africa. Mid-year Population Estimates, South Africa, 2006.
Pretoria: Statistics South Africa.
6 African children at age 9 years. J App Physiol 2004;97:1006–12.
Harpham T, Huttly S, Wilson I, de Wet T. Linking public issues with
private troubles: Panel studies in developing countries. J Int Develop
2003;15:353–63.
7
Richter L, Yach D, Cameron N, Griesel R, De Wet T. Enrolment into
Birth to Ten (BT20): Population and sample characteristics. Paed
Perinat Epidemiol 1995;9:109–20.
8
Richter L, Norris SA, de Wet T. Transition from Birth to Ten to Birth
to Twenty: the South African cohort reaches 13 years of age. Paed
Perinat Epidemiol 2004;18:290–301.
9
Fogas B, Osterheld J, Shader R. A retrospective study of children’s
perceptions of participation as clinical research subjects in a minimal
risk study. J Develop Behav Paed 2001;22:211–6.
10
Anderson A, Richter L. ‘‘Birth to Ten’’: Error detection and correction
of a longitudinal address database. South Afr J Epidemiol Infec
1994;9:16–9.
11
Ellison G, Richter L, de Wet T, Harris H, Griesel R, McIntyre J.
Improving the accuracy of birth notification data: lessons from the
Birth to Ten study. South Afr J Epidemiol Infec 1997;12:91–6.
12
Ellison G, De Wet T, IJsselmuiden C, Richter L. Segregated health
statistics perpetuate ’racial’ stereotypes. Br Med J 1997;314:1485.
13
Matshidze K, Richter L, Ellison G, Levin J, McIntyre J. Caesarean
section rates in South Africa: Evidence of bias among different
‘‘population groups’’. Ethnic Health 1998;3:71–79.
BIRTH TO TWENTY STUDY IN SOUTH AFRICA 511
14
Norris SA, Richter L. Usefulness and reliability of Tanner pubertal
self- rating to urban Black adolescents in South Africa. J Res Adol
2005;15:609–24.
15
Norris SA, Richter LM. HIV/AIDS risk research among young
people: a methodological study of the reporting of sensitive issues
among adolescents in the Birth to Twenty Study, South
Africa. Proceedings of the XVIth International AIDS Conference, Toronto,
13–18 August, 2006.
16
MacKeown JM, Cleaton-Jones PE, Norris SA. Nutrient intake among
a longitudinal group of urban black South African children at four
interception between 1995 and 2000 (Birth-to-Ten Study). Nutr Res
2003;23:185–97.
17
MacKeown JM, Pedro TM, Norris SA. Energy, macro-and micro-
nutrient intake among a true longitudinal group of South African
adolescents at two interceptions (2000 and 2003): the Birth to
Twenty (Bt20) study. Pub Health Nut (in press).
18
Cameron N. Physical growth in a transitional economy: the
aftermath of South African apartheid. Econom Human Biol
2003;1:29–42.
19
Vidulich L, Norris SA, Cameron N, Pettifor JM. Ethnic differences in
Downloaded from ije.oxfordjournals.org by guest on March 3, 2011
bone mass in 10 year old South African children. Osteopor Internat
2006;17:433–40.
20
McVeigh JA, Norris SA, Cameron N, Pettifor JM. Associations
between physical activity and bone mass in black and white South
21
Richter L, Griesel R, Barbarin O. Behavioural problems among
preschool children in South Africa: A six-year longitudinal perspec-
tive from birth to age five. In: Singh N, Leung J, Singh A (eds).
International perspectives on child and adolescent mental health.
Amsterdam: Elsevier Press, 2000, pp.159–82.
22
Barbarin O, Richter L. Adversity and psychological functioning of
children growing up in black townships in South Africa: Effects of
community danger and socioeconomic status. Amer J Orthopsych
1999;11:16–25.
23
Barbarin OA, Richter LM, De Wet T. Exposure to violence, coping
resources and psychological adjustment of South African children.
Amer J Orthopsych 2001;71:16–25.
24
Richter LM, Norris SA. Sexual activity on early teens: longitudinal
data from adolescents 13 to 15 years of age from the Birth to Twenty
study. Proceedings of South Africa. XVIth International AIDS Conference,
Toronto, 13–18 August, 2006.
25
Crowther NJ, Cameron N, Trusler J, Gray IP. Association between
poor glucose tolerance and rapid post natal weight gain in seven-
year-old children. Diabetol 1998;41:1163–67.
26
Crowther NJ, Trusler J, Cameron N, Toman M, Gray IP.
Relation between weight gain and beta cell secretory activity and
non-esterified fatty acid production in 7-year-old African children:
Results from the Birth to Ten study. Diabetol 2000;43:978–985.