population; a previous report, however, suggests it is
around 32%.2° Many of these unplanned pregnancies
will be terminated; in our study the figure of 19%
agrees with evidence to suggest that one in five
pregnancies are terminated in the United Kingdom.

Conclusion
The confused medical awareness of which contra-

ceptive methods are appropriate may partly explain
why advice to women with sickle cell disease is often
lacking. We cannot exclude a possible risk of crises and
thrombotic episodes with the use of the combined
contraceptive pill. Despite this, we do not consider
it to be contraindicated in this group of women as
any complications should be balanced against the
substantial risks of pregnancy. We feel that all methods
of contraception may be considered in women with
sickle haemoglobinopathies, though with appropriate
caution.

We gratefully acknowledge generous financial support
from SCAR (Sickle Cell Anaemia Relief) and the invaluable
cooperation of the North London Haemoglobinopathy Group
and the Sickle Cell Centres of north London.

1 Tuck SM, Studd JWW. Pregnancy in sickle cell disease in the UK.
BrJ Obstet Gynaecol 1983;90:112-7.

2 Foster HW. Contraceptives in sickle cell disease. South MedJ 1981;74:543-5.
3 Evans DIK. Sickle cells and hormonal contraception. British Journal of

Family Planning 1984;1O:80-1.

4 PerkinsRP.Contraceptionforsicklers. NEngl_Med 1971;285:296.
5 Morrison JC. Hemoglobinopathies and pregnancy. CGin Obstet Gynecol

1979;22:8 19-42.
6 Louden N, ed. Handbook of family planning. 2nd ed. London: Churchill

Livingstone, 1991.
7 Sergeant GR. Sickle haemoglobin and pregnancy. BMJ 1983;287:628-30.
8 Petrou M, Modell B, Darr A, Old J, Kin E, Weatherall D. Antenatal

diagnosis. How to deliver a comprehensive service in the United Kingdom.
Ann N YAcad Sci 1990;612:251-63.

9 Loukopoulos D, Hadji A, Papadakis M, Karababa P, Sinopoulou K, Boussiou
M, et al. Prenatal diagnosis of thalassaemia and of the sickle cell syndromes
in Greece. Ann N YAcad Sci 1990;612:226-36.

10 Anionwu EN, Patel N, Kanji G, Renges H, Brozovic M. Counselling for
prenatal diagnosis of sickle cell disease and thalassaemia major: a four year
experience. JMed Genet 1987;25:769.

11 Rowley PT, Loader S, Sutera CJ, Walden M, Kozyra A. Prenatal screening for
haemoglobinopathies. I. A prospective clinical trial. Am J Hum Genet
1991 ;48:439-46.

12 Samuels-Reid JH, Scott RB, Brown WE. Contraceptive practices and
reproductive patterns in sickle cell disease. J Natl Med Assoc 1984;76:
879-83.

13 Charache S, Niebl JR. Pregnancy and sickle cell disease. Clin Haematol
1985;14:729-46.

14 De Ceulaer K, Gruber C, Hayes R, Serjeant GR. Medroxyprogesterone acetate.
and homozygous sickle cell disease. Lancet 1 982;ii:229-3 1.

15 Freie HMP. Sickle cell diseases and hormonal contraception. Acta Obstet
Gynecol Scand 1983;62:21 1-7.

16 Lutcher CL, Milner PF. Contraceptive-induced vascular occlusive events in
sickle cell disorders-fact or fiction? Clin Res 1986;34:21 7A.

17 Greenwald JG. Stroke in a woman with sickle cell trait taking oral contracep-
toves. Conn Med 1971;35:231-2.

18 Hargus EP, Shearin R, Colon AR. Pulmonary embolism in a female adolescent
with sickle cell trait and oral contraceptive use. Am J Obstet Gynecol
1 977;129:697-9.

19 Haynes RL, Dunn JM. Oral contraceptives, thrombosis, and sickle cell
haemoglobinopathies. JAMA 1967;200:186-8.

20 Fleissig A. Unintended pregnancies and the use of contraception: changes
from 1984 to 1989. BM3 1991;302:147.

21 Smith T. Unwanted pregnancies. BMJ 1990;300:1154.

(Accepted 8April 1993)

Why have child pedestrian death rates fallen?

Ian Roberts

Injury Prevention Research
Centre, Department of
Community Health,
University ofAuckland,
Private Bag 92019,
Auckland, New Zealand
Ian Roberts, Health Research
Council trainingfellow

BMJ 1993;306:1737-9

Pedestrian injuries are a leading cause of childhood
mortality and disability. Over the past two decades in
Britain child pedestrian death rates have fallen
despite large increases in traffic volume. In this
paper Roberts examines the likely reasons for this
decline. He argues that neither prevention pro-
grammes nor improvements in medical care are
a plausible explanation and that the decline is
most likely the result of a substantial reduction in
children's traffic exposure. He believes, however,
that restricting children's traffic exposure exacer-
bates socioeconomic differentials in childhood
mortality and denies children their right to mobility.
Roberts is convinced that one answer is for British
transport policy to be aimed at providing mobility
equitably rather than struggling to meet the ever
increasing demands ofcar travel.

Between 1980 and 1990 in England and Wales an
average of 239 children were killed each year on the
roads as pedestrians. For every death around 10
children required hospital admission for injuries.'
Pedestrian injuries were among the leading causes
of childhood admission to intensive care facilities.2
Between 60% and 80% of these children have severe
head injuries and are likely to experience long term
disability.'
With the more widespread recognition of the public

health importance of child pedestrian injuries in
recent years several epidemiological studies have been
mounted aimed at identifying modifiable risk factors.
These studies have used case-control or cohort methods
in an attempt to identify the factors which place some
children at high risk of pedestrian injury."6 However,
an effective public health response to the problem of
child pedestrian injury demands that a second aetio-

logical issue should be addressed. This concerns
the identification of determinants of the incidence,
attempting to understand why child pedestrian death
rates change over time. Answering this question is
likely to be as important as identifying risk factors,
because if the determinants of the incidence could be
identified it might be possible to control them, with
gains for the whole child population.7
Over the past two decades child pedestrian death

rates have fallen in many developed countries. In
England and Wales between 1968 and 1987 the pedes-
trian mortality among children aged 0-4 years fell by
67%, and among children aged 5-14 years it fell by
39%.8 To gain an insight into why child pedestrian
death rates have fallen changes in the potential deter-
minants of the child pedestrian death rate must be
considered.

Potential determinants ofchild pedestrian deaths
The determinants of incidence are most readily

identified when their prevalence changes abruptly. For
this reason the effect of traffic volume on child
pedestrian mortality was most evident during the
energy crisis, when an increase in the price of petrol
had a measurable effect on the growth in traffic
volume. In New Zealand government restrictions on
car use in the aftermath of the energy crisis effectively
arrested the growth in traffic volume for seven years,
during which there was a 46-4% reduction in child
pedestrian mortality.9 Before this the death rate had
been increasing in parallel with increasing traffic
volumes. That the rate fell, as opposed to levelling
off in line with traffic volume, suggests that there
were other processes operating, tending to reduce the
mortality, but which became evident only when traffic

BMJ voLuME 306 26JUNE1993 1737

 o
n
 5

 A
p
ril 2

0
2

1
 b

y g
u
e
st. P

ro
te

cte
d
 b

y co
p
yrig

h
t.

h
ttp

://w
w

w
.b

m
j.co

m
/

B
M

J: first p
u

b
lish

e
d

 a
s 1

0
.1

1
3

6
/b

m
j.3

0
6

.6
8

9
4

.1
7

3
7

 o
n

 2
6
 Ju

n
e
 1

9
9
3
. D

o
w

n
lo

a
d
e
d
 fro

m
 

http://www.bmj.com/


volume was unchanging. In Britain since 1968 the
child pedestrian death.rate has fallen despite an almost
exponential increase in traffic volume (fig 1).' This
implies' even mQre substantiWl changes in other, more
powerful determinants of incidence. These changes
have been particularly important for children aged 0-4,
for 'whom reductions in mortality have been most
dramatic.

PEDESTRIAN SKILLS TRAINING

It would be comfortng if there were evidence to'
show that at least some, of the reduction in. child
pedestrian mortality could be attributed"to pedestrian
injury prevention programmes, but'this evidence is
lacking. In Britain the main thrust of preventive
strategy has comprised efforts to improve childL pedes-
trian behaviour through -pedestrian skills training
programmes-for example, the Green Cross Code.-
However, data from evaluation studies suggest that
these programmes are of limited, value.'0 Some North
American studies have claimed reductions in injury
rates following pedestrian education programmes but
evaluation in these studies 'was based on before and:
after comparisons,-so that the apparent reductions may
have been confounded by changes in the background
rate." 2 Studies which have employed more rigorous-
methods have concluded that even large efforts to
improve pedestrian behaviour are rewarded by only
small gains."3 Furtiermore, as programmes have been
aimed predominantly' at schioolchildren they cannot
explain the large -reductioi in pedestrian mortality
among preschool children. It therefore seems im-
plausible that pedestrian education programmes have
had a substantial effect on--child pedestrian mortality.

In Britain between 1968 and 1987 the number of
police reported child pedestrian casualties fell by 49%
but the number of deaths fell by 61/% (J Broughton,
Transport Research Laboratory, personal communica-
tion, 1992), suggesting that some ofthe decline in child
pedestrian mortality may be due -to reduction in case
fatality. These- data must ';be-viewed with caution;'
however. Child pedestrian- injuries aare significantly
underreported -in p6lice aciident databases, and'
changes over time it casualty'numbers might simply
reflect changes in the extent of underreporting.'4
Furthermore, as children under 5 have a higher case
fatality rate posgily fthe apparent, overall reduction Ain
case fatality is- due to a change in the age distribution of
injured childn, young ch n now maldng up a
smaller- proportion of the total casualties." -If there has
been a reduction in case fatality this may have resulted
from a decrease in injury severity or fromn improve-

FIG z-Lur own;, ,y L X Lowry

ments in medical care. Regardless, the 49% decrease in
casualty numbers implies that most of the reduction
in child pedestrian mortality is due to a reduction
in collisions between child pedestrians and motor
vehicles.

TRAFFIC EXPOSURE

The most plausible explanation for the downward
trend in child pedestrian mortality is that it reflects a
decline in children's traffic exposure. As the volume of
road traffic has increased streets have become more
dangerous and increasingly children have been kept
away fronr them.'6 Until recently the only testimony
was old photographs or paintings depicting towns and
cities as they were before the heyday ofthe motorcar, in
which children would often be seen playing in the
street. The urban landscapes of L S Lowry starkly
portray the grime- and pollution of industrial England,
but the street scenes convey a vivacity and a sense of
both safety and 'community which is absent today
(fig 2).
A recent British study which examined changes in

children's independent mobility over the past two
decades has provided compelling evidence for a decline
in children's traffic exposure. ' For example, it was
found that whereas in 1971, 80% of 7 and 8 year old
English children were allowed to travel to school
unaccompanied, in 1990 the -figure was 90/o. The
principal reason parents gave for their reluctance to
allow their children to travel 'unaccompanied was
traffic danger. Similarly, the age at which children
,receive parental "licence" to cross roads has increased
considerably, and there was a pronounced decline in
the proportion of children who were allowed to cross
roads alone in 1990 compared with 1971. These
findings might explain why the decline in pedestrian
mortality has been least in the 5-14 year age group, as it
would be more difficult for parents to limit the traffic
exposure of older children.'8 Moreover, a study in
Manchester found that the pedestrian injury rate
among children aged 11-14 had increased by 15%
bet-ween 1969 and 1987.'9
T'.he importance of traffic exposure is also apparent

in case centred epidemiological studies. A North
American case-control study found that the absence of
a play area adjacent to the home was associated with a
fivefold increased risk of pedestrian injury (odds ratio
5-3; 95% confidence interval 2*6 to 1 1.0).4 The effect of
this risk factor is likely to be mediated through the
greater traffic exposure of children who have fewer
altematives to playing in the street.

INDICES OF SAFETY

-'A consideration of children's traffic exposure has
been the missing element in virtually all of the political
deliberations on road safety for children over the past
two decades.'7 As a result the effect on child pedestrian

> safety of the massive increase in traffic volume which
has taken place over this period has been completely
obscured. In the absence of an exposure measure

-thepedestrian injury rate has been used as an index
o.af- safety. The fallacy' inherent in this reasoning is
-jopparent from a consideration of age specific child
pedesinan injury rates.

h In the absence of'information on traffic exposure,
! using pedestrian injury hospitalisation rates as a

measure of safety would lead us to conclude that a
3 year old child (65 9/100 000 admissions/year) is safer
in a traffic setting than a 6 year old (68-9/100 000/year)

- (unpublished observations, New Zealand data). In
o contrast, for motor vehicle occupants the established
; safety measure is deaths per vehicle mile travelled, the
| rationale for which was given in the motor vehicle
8 safety- position paper of a recent American injury

control conference-"highway transportation has been

BMJ VOLUME 306 26 uNE1993

Traffic volume
o- Fatality rate

6

.35Q~~~~~~
083004

3
E 250
20 2
~200
150 0

1970 75 80 85 90
Year

FIG 1-Childpedestnan death
rates and traffic volume in
Britain 1967-90. (Sources-
Fatality rates: Office of
Population Censuses and
Surveys, unpublished data,
traffic volume: Transpoq
Research Laboratory,
unpublished data)

-'1738

 o
n
 5

 A
p
ril 2

0
2

1
 b

y g
u
e
st. P

ro
te

cte
d
 b

y co
p
yrig

h
t.

h
ttp

://w
w

w
.b

m
j.co

m
/

B
M

J: first p
u

b
lish

e
d

 a
s 1

0
.1

1
3

6
/b

m
j.3

0
6

.6
8

9
4

.1
7

3
7

 o
n

 2
6
 Ju

n
e
 1

9
9
3
. D

o
w

n
lo

a
d
e
d
 fro

m
 

http://www.bmj.com/


created to facilitate mobility and mobility is of value."20
Strategies which reduce the need for car travel or
substitute car travel with safer forms of transport
would substantially reduce population death rates.2'
Thus deaths per vehicle mile travelled clearly signals
that improvements in safety must not limit car use.
Because children's mobility is not valued the decline
in child pedestrian mortality has taken place at its
expense.

Implications oflimiting child mobility
As a means of preventing child pedestrian injuries

limiting children's traffic exposure has powerful disad-
vantages. In particular, it will exacerbate the striking
socioeconomic inequalities in child pedestrian injury
morbidity and mortality that already exist. Both
British and overseas studies have shown that poor
children have a risk of pedestrian injury some three
times greater than the least poor, and census area
unemployment has been shown to be strongly corre-
lated with census area child pedestrian injury rates.22-25
Indeed, pedestrian injuries are a major contributor to
socioeconomic differentials in childhood mortality.22
As one third of British households do not have access to
a car escorting children is likely to be far more difficult
for some families. Poor children are likely to have
fewer alternatives to playing in the street, and super-
vising children will be far more demanding in single
parent families.
The British epidemiologist Geoffrey Rose has

argued that in disease prevention two broad types
of strategy can be distinguished: those that restore
biological normality and those that take us further
from the conditions for which we were genetically
adapted.26 Restoration of biological normality is gener-
ally safe whereas the consequences of moving away
from our biological condition are often unknown and
the potential for harm may easily be overlooked. In this
context limiting children's independent mobility is a
clear departure from biological normality. Changes in
childhood exercise patterns may impact on children's
cardiovascular health, and with children increasingly
confined to the home because of traffic danger height-
ened family tensions may erupt as family violence.
Historically, public health investigators have been
slow to recognise the importance of such social pro-
cesses and as a result have witnessed the waxing and
waning of mortality from many diseases, although
often ignorant of the reasons and powerless to inter-
vene.27

The future
Traffic volume is expected to increase by 142% by

the year 2025, and no doubt the rampant individualism
of free market economic policies will ensure that this
prophecy is realised.28 The increase in traffic volume
will inevitably result in a more hostile urban living
environment for children. More fortunate children will
reap the benefits of increased car travel while being
largely immune from the accompanying dangers. But
for the increasing number of British children living in
poverty29 paramount among threats to health are
vehicles that most will only ever have the opportunity
to steal. Further reductions in children's independent
mobility may offset the increase in danger but already
child pedestrian death rates seem to be levelling off and
in some countries-for example, the United States-
may be rising.8

It must be recognised that unfettered market forces
will inevitably disregard the rights of the more vulner-

able members in society and that a more collective
social responsibility is appropriate. British transport
policy should aim to provide mobility equitably and for
all people rather than continually struggle to meet
the ever increasing needs of car users. Of course,
challenging the dominant position of the private
passenger car in the transport system will inevitably
confront powerful vested interests. The car lobby is
strong and well organised.'8 The lessons leamt from
previous public health struggles with pecuniary power
will be invaluable.30 Mobility is of value but so is
equity. The transport system of the future should
embrace both of these concems.

1 Rivara FP. Child pedestrian injuries in the United States. Am J Dis Child
1990;144:692-6.

2 Roberts I, Streat S, Judson J, Norton R. Critical injuries in paediatric
pedestrians. NZMedJ 1991;104:247-8.

3 Roberts I, Norton R, Hassall I. Child pedestrian injury 1978-87. N Z Med J
1992;105:5 1-2.

4 Mueller BA, Rivara FP, Shyh-Mine L, Weiss NS. Environmental factors
and the risk of childhood pedestrian-motor vehicle collision occurrence.
AmJEpidemiol 1990;132:550-60.

5 Pless IB, Verreault R, Tenina S. A case control study of pedestrian and
bicyclist injuries in childhood. AmJ Public Health 1989;79:995-8.

6 Pless IB, Peckham CS, Power C. Predicting traffic injuries in childhood: a
cohort analysis. J Pediatr 1989;115:932-8.

7 Rose G. Sick individuals and sick populations. Int7Epidemiol 1985;14:32-8.
8 Roberts I. Intemational trends in pedestrian injury mortality. Arch Dis Child

1993;68: 190-2.
9 Roberts I, Marshall R, Norton R. Child pedestrian mortality and traffic

volume in New Zealand. BMJ 1992;305:283.
10 Organisation for Economic Co-operation and Development. Traffic safety of

children. Paris: OECD, 1983.
11 Preusser DF, Blomberg RD. Reducing child pedestrian accidents through

public education.J7ournal ofSafety Research 1984;15:47-56.
12 Fortenberry JC, Brown DB. Problem identification, implementation and

evaluation of a pedestrian safety programme. Accid Anal Prev 1982;14:
315-22.

13 Rivara FP, Booth CL, Bergman AB, Rogers LW, Weiss J. Prevention of
pedestrian injuries to children: effectiveness of a school training program.
Pediatrics 199 1;88:770-5.

14 Maas MW, Harris S. Police recording of road accident inpatients. Accid Anal
I4ev 1984;16:167-84.

15 Pitt R, Guyer B, Hsieh C, Malek M. The severity of pedestrian injuries in
children: an analysis of the pedestrian injury causation study. Accid Anal
1rev 1990;22:549-59.

16 Adam JG. Risk homeostasis and the purpose of safety regulation. Ergonomics
1988;31:407-28.

17 Hillman M, Adams J, Whitelegg J. One false move: a study of children's
independent mobility. London: Policy Studies Institute, 1991.

18 Quick A. Unequal risks: accidents and social policy. London: Socialist Health
Association, 1991.

19 Preston B. Child pedestrian casualties with special reference to casualties
on the joumey to or from school in Manchester and Salford, England.
AccidAnalPrev 1989;21:291-301.

20 US Department of Health and Human Services. Third national injury control
conference. Position papers. Washington, DC: US Govemment Printing
Office, 1992.

21 Trinca GW, Johnston IR, Campbell BJ, Haight FA, Knight PR, Mackay GM,
et al. Reducing traffic injury: a global challenge. Melboume: Royal Australasian
College of Surgeons, 1988.

22 Office of Population Censuses and Surveys. Occupational mortality: child
supplement. England and Wales 1979-80, 1982-83. London: HMSO, 1988.
(Series DS No 8.)

23 Pless IB, Verreault R, Arsenault L, Frappier J, Stulginskas J. Am J Public
Health 1987;77:358-60.

24 Dougherty G, Pless IB, Wilkins R. Social class and the occurrence of traffic
injuries and deaths in urban children. Can JPublic Health 1990;81:204-9.

25 Roberts I, Marshall R, Norton R, Borman B. An area analysis of child injury
morbidity in Auckland. J Pediatr Child Health 1992;28:438-41.

26 Rose G. Strategy of prevention: lessons from cardiovascular disease. BMJ
1981;282:1847-5 1.

27 Rose G. Reflections on the changing times. BMJ 1990;301:683-7.
28 Godlee F. Children harmed by traffic fumes. BMJ 1991;303:329.
29 Spencer NJ. Child poverty and deprivation in the UK. Arch Dis Child

1991;66:1255-7.
30 Beaglehole R. Science, advocacy and health policy: lessons from the New

Zealand tobacco wars.J Public Health Policy 1991;12:175-83.

(Accepted 25 March 1993)

Correction

How can we best prolong life? Benefits ofcoronary risk
factor reduction in non-diabetic and diabetic subjects
A typesetting error occurred in table IV of this paper by John S
Yudkin (15 May, pp 1313-8). In the bottom line of the table
(lowering systolic blood pressure 2 142 mm Hg) the entry in the
second column of figures is wrong: it should read 0-38 (-0 07 to
1-79), not 9-83 (-0 07 to 1-79).

BMJ VOLUME 306 26JUNE1993 1739

 o
n
 5

 A
p
ril 2

0
2

1
 b

y g
u
e
st. P

ro
te

cte
d
 b

y co
p
yrig

h
t.

h
ttp

://w
w

w
.b

m
j.co

m
/

B
M

J: first p
u

b
lish

e
d

 a
s 1

0
.1

1
3

6
/b

m
j.3

0
6

.6
8

9
4

.1
7

3
7

 o
n

 2
6
 Ju

n
e
 1

9
9
3
. D

o
w

n
lo

a
d
e
d
 fro

m
 

http://www.bmj.com/