Mini-clusters of potentially prodromal symptoms may
identify psychiatrically well Amish children at higher risk of
developing bipolar I disorder
Egeland JA, Shaw JA, Endicott J, et al. Prospective study of prodromal features for bipolarity in well
Amish children. J Am Acad Child Adolesc Psychiatry 2003;42:786–96.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Q Do frequencies of potential early prodromal clinical features for bipolar I (BPI) disorder in psychiatrically well Amishchildren correlate with family history, and therefore inferred risk, of BPI?

METHODS

Design: Prospective cohort study; outcome assessors were
blinded as to family history, however, it was impossible to blind
data collectors.

Follow up period: Seven years.

Setting: Amish families in rural Pennsylvania, USA; 1994 to
2000.

People: 210 children from 27 families. 14 BPI families (100
children) had one parent with BPI. The families of same sex,
psychiatrically unaffected siblings of each bipolar parent were
selected as matched controls where possible. Control siblings
were selected to be as close in age, marriage date, and number
of children to the bipolar sibling. In the absence of a suitable
matched sibling control, the family of a non-related Amish
participant matched for sex, age, and family size, but without a
family history of BPI, was selected. There were 9 sibling control
families (77 children) and 4 family history negative control
families (33 children).

Risk factors: The Child and Adolescent Research Evaluation
(CARE) interview was administered annually to parents of
participating children. The CARE interview comprised three parts:
part A (yes/no questions) pertained to the pregnancy/delivery,
medical, and developmental history of the child; part B was
narrative and asked parents whether they had any concerns
about their child’s health, wellbeing, or behaviour; part C (yes/
no questions) was used in alternate years and comprised 40
questions about the child being ‘‘noticeably different’’ from other
children for potentially prodromal or antecedent items, and about
the child’s functioning and ‘‘wellness’’.

Outcomes: Presence of prodromal symptoms and predicted risk
of developing BPI disorder as determined by independent,
blinded assessors (one psychiatrist, two child psychiatrists, and
one clinical psychologist).

MAIN RESULTS
Children with one BPI parent were significantly more likely to have 7
or more potentially prodromal symptom/behaviour items than
children with well parents (AR: 12% for BPI family child v 3% for
control family child; p = 0.0007). Symptoms significantly more

frequent in children with one BPI parent than in children with well
parents were: anxiety, poor attention span in school, low energy,
excitability, hyper-alertness, mood lability, school role impairment,
sensitivity, somatic complaints, and stubbornness (p(0.05).
Assessors identified more of the children with one BPI parent as
‘‘at risk’’ of BPI than children with well parents (AR: 38% for BPI
family child v 17% for control family child; overall p value for liability
class distribution ,0.003).

CONCLUSIONS
Mini-clusters of early potential prodromal features occur at higher
frequencies in psychiatrically well Amish children with one parent
with BPI than in those with well parents.

NOTES
The authors note that the major limitation of this interim report of
their study is that although these findings do identify potential early
prodromal features for BPI, the true test of the outcome will be
whether children designated as ‘‘at risk’’ do eventually develop BPI.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
For correspondence: Dr Egeland, University of Miami North Research Office,
Briarcrest Square, Hershey, Pennsylvania, USA; OOAmish@aol.com

Sources of funding: The CARE (Child and Adolescent Research and
Evaluation) programme was supported in part by the Stanley Medical
Research Institute and with funding from the University of Miami School of
Medicine.

Commentary

T
his paper describes a prospective study of 210 healthy old-order
Amish children, designed to determine the pattern of psychiatric
symptoms that may be prodromal for bipolar disorder. Data are the

result of 7 years of annual interviews with families begun in 1994 as part
of the Child and Adolescent Research Evaluation (CARE) programme,
designed for the above purpose. It is a novel study design producing
useful data for several reasons.

While several studies have collected retrospective data from inpatients
on selected prodromal bipolar symptoms or prospective data on such
symptoms in children from affected families, none has used a community
sample including children of varying levels of risk for bipolar disorder.
Further, as they have studied an Amish group, they have been able to
collect multigenerational data in large families useful for genetic studies,
within a cultural community with much social stability. These group data
on early bipolar risk may not be applicable to a general population, as
symptoms of disruptive behaviour disorders—common in bipolar
children—are possibly mitigated by cultural influences. However, this
study has allowed the Egeland team to describe a spectrum of prodromal
symptoms that may be specific for bipolar disorder and prove to be of
relevance to clinical practice.

Of clinical importance, they found that children with bipolar parents
have high risk of developing bipolar disorder—38%, similar to rates cited
in other risk studies. Children of unaffected siblings of a bipolar adult had
intermediate risk compared with control children, suggesting reduced
penetrance of bipolar symptomatology in families. Symptoms of children
of a bipolar parent, which are significantly more common than in
children of non-bipolar parents, are described. It is suggested that some
prodromal symptoms for bipolar disorder are episodic in 50% of the
children, such as mood and energy changes, need for sleep, and anger/
temper outbursts. These findings will be helpful for clinicians assessing
bipolar risk in children with subsyndromal symptoms combined with
known or suspected family history of the illness.

Khrista Boylan, MD
Department of Psychiatry and Behavioural Neurosciences, McMaster

University, Hamilton, Ontario, Canada

26 AETIOLOGY

www.ebmentalhealth.com

 o
n
 A

p
ril 5

, 2
0
2

1
 b

y g
u

e
st. P

ro
te

cte
d

 b
y co

p
yrig

h
t.

h
ttp

://e
b
m

h
.b

m
j.co

m
/

E
vid

 B
a

se
d

 M
e

n
ta

l H
e

a
lth

: first p
u

b
lish

e
d

 a
s 1

0
.1

1
3

6
/e

b
m

h
.7

.1
.2

6
 o

n
 9

 F
e
b
ru

a
ry 2

0
0
4
. D

o
w

n
lo

a
d
e
d
 fro

m
 

http://ebmh.bmj.com/