The role of the healthy parent in adolescents' adjustment to parental physical illness.


Material Information

The role of the healthy parent in adolescents' adjustment to parental physical illness.
Physical Description:
vi, 121 leaves : ill. ; 29 cm.
Houck, Christopher D.
Publication Date:


Subjects / Keywords:
Research   ( mesh )
Parent-Child Relations   ( mesh )
Adaptation, Psychological -- Adolescence   ( mesh )
Family Health   ( mesh )
bibliography   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph.D.)--University of Florida, 2002.
Bibliography: leaves 112-120.
Statement of Responsibility:
by Christopher D. Houck.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
oclc - 82103362
System ID:

Full Text








I would like to thank all of my family, friends, and colleagues for their support

throughout my pursuit of this degree. 1 am extremely lucky to have had such great

people in my life.

I would like to specially thank my grandmother, Gerda Houck, for her support of

my goals, for her belief in the importance of education, and for her constant




ACKNOW LEDGM ENTS.............. .. ............................... .......... ...ii

ABSTRACT...................................... ........................ ... ...... v


1 INTRODUCTION ............................... .................. 1

2 REVIEW OF LITERATURE ............................... 4

Prevalence of Chronic Illness in Parent Populations ............................. 4
A adolescents and Stress....... .. ...... .... ................... ............... 5
Parental Physical Illness and Child/Adolescent Functioning...................... 8
Models of Parental Physical Illness and Child Functioning....................... 22
Critique of Current M odels............................................ ................. 28
Factors Associated with Child/Adolescent Functioning............................ 30
Child Factors............... ...................................... ....... 30
Parent Factors .............. .................. .. ..... ... ... ...................... ... 35
Illness Factors ................. ... ................................................... 39
F am ily F actors..................... .................................................... .. 40

3... AIMS AND STUDY JUSTIFICATIONS ....................................... 49

Purpose of the Study ...... ... ....... ... .......................... 49
Hypotheses ................................... ........... ... ...........49
Hypotheses Related to Levels of Adolescent Adjustment ......................49
Hypotheses Related to Parent-Adolescent Communication .................... 49
Hypotheses Related to Ill vs. Healthy Parent Reports of Adolescent
A djustm ent............. .... ........... ... .................. ....... .......... 50

4 METHOD........ ................................................ 51

Participants........ .................. .. ... ...................... ............... 51
Ill Parents ................................................. 51
Healthy Parents ..... ....... .......... ............................... ... 53
Adolescents......... ............................................ .. 53
Family Demographics... ....................................... .........53

Procedure ............. .............................................. .......... 54
M measures .......................... ............................ ... ....... 55
D em ographics.......................... ........ ................. ....... ...... 55
Perceptions of Illness Questionnaire................... .. ......... ......... 56
Reynolds Adolescent Depression Scale (RADS)............................... 56
Children's Manifest Anxiety Scale- Revised (RCMAS) ......... .......... 57
Impact of Event Scale (IES)....... ............... .................. 57
Parent-Adolescent Communication Scale (PACS)............................. 60
Child Behavior Checklist (CBCL) ................. ......... ...................62

5 RESULTS.............. .. ............. .............................. .......... 64

Prelim inary A nalyses...................................... ............ ...... ...... 64
Prim ary A analyses .................. ................ .......... .................. 67
Hypothesis 1: Adolescent Adjustment to Parental Physical Illness.......... 67
Hypothesis 2: Effects of Same Sex Ill Parents..................................... 72
Hypothesis 3: Adolescent Perceptions of Ill vs. Healthy Parent
Com m unication........................................................................ 73
Hypothesis 4: Relationship Between Parent-Adolescent Communication
and A djustm ent...................................................................... ...75
Hypothesis 5: Effects of the Presence of Parental Positive Relationships.. 79
Hypothesis 6: 111 vs. Healthy Parent Reports of Adolescent Adjustment... 80

6 D ISC U SSIO N ....... ..... ... .......................... .............................. 84

Discussion of Major Findings..................... ................. ................ 84
Hypothesis 1: Adolescent Adjustment to Parental Physical Illness..........84
Hypothesis 2: Effects of Same Sex Ill Parents.................................... 92
Hypothesis 3: Adolescent Perceptions of Ill vs. Healthy Parent
C om m unication..........................................................................93
Hypothesis 4: Relationship Between Parent-Adolescent Communication
and A djustm ent....................................... .............................. 94
Hypothesis 5: Effects of the Presence of Parental Positive Relationships..98
Hypothesis 6: 111 vs. Healthy Parent Reports of Adolescent Adjustment...99
Sum m ary of Findings ............................. ........................................ 101
Contributions and Weaknesses of the Present Study............................... 102

APPENDIX..... .............................................................. ..... 110

R E FE R E N C ES ............................. ............ .................................... .... 112

BIOGRAPHICAL SKETCH.................................................................. 121

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy



Christopher D. Houck

August 2002

Chair: James R. Rodrigue
Major Department: Clinical and Health Psychology

The objective of the present study was to examine the relationship between

parent-adolescent communication and psychological functioning among adolescents who

live with parents who have chronic physical health conditions. Three members of each

family (healthy parents, ill parents, and adolescents) completed questionnaires. Results

indicated that nearly one-third of the sample reported clinically meaningful levels of

stress-response symptoms related to their parent's illness. Adolescent communication

with the healthy parent was significantly associated with symptoms of anxiety but not

depression. However, communication with the ill parent was not associated with

adolescent adjustment. Finally, evidence from the present study does not provide

adequate support for the hypothesis that biases exist in the reporting of adolescent

behavior problems by ill parents in comparison to healthy parents. These findings

suggest that adolescents with severely ill parents may experience symptoms of anxiety

not typically assessed in this population. Furthermore, the relationship with the healthy

parent may serve an important and unique function for these adolescents and may be a

useful target for clinical intervention.


Obesity, substance abuse, depression, and chronic physical health problems are

commonly experienced by adults of parenting age. Inasmuch as a parental health

problem may affect numerous aspects of family life, exposure to this stressor may have

negative effects on millions of children. However, "despite the high incidence and

clinical relevance of parental health problems, the impact of such problems on children

and families has not been well described" (Drotar, 1994, p. 526). The amount of

scientific knowledge available varies depending on the health problem. While there is a

fair amount of research on the impact of alcoholism on children's development,

relatively little information exists related to parental chronic physical illness as it affects


The effects of parental physical illness on children and adolescents continue to

grow in importance. Thousands of people in parenting age are diagnosed with chronic

conditions each year. Advances in medicine have increased the life expectancy for many

individuals with chronic illnesses and have allowed them to continue their lives at home

rather than in health care facilities. These factors, coupled with great strides in fertility

treatments for adults who once may not have been able to become parents, have increased

the number of children living with a parent who has a chronic health condition.

While such medical advances are positive, data suggest that children of

chronically ill parents may exhibit more symptoms of psychological distress than

children of healthy parents. Investigators have explored several variables to determine

their relationships with outcomes for children in these families. One variable that has

been related to poorer outcomes is age. Specifically, findings suggest that adolescents

may be at greater risk than younger children for experiencing psychological distress when

a parent is ill. Drawing from findings in the general adolescent stress literature,

researchers have begun to explore the role of parent-child relationships on children's

psychological functioning in these families, generally finding an association between less

conflicted parent-adolescent relationships and better adolescent psychological

functioning. However, it seems likely that in families in which one parent is ill and one

is healthy, adolescents may maintain different relationships with these two parents.

While investigators have developed models of child functioning considering both parents

separately, no study has simultaneously considered the influence of both parent-

adolescent relationships in predicting children's adjustment.

The purpose of the present study was to further delineate key variables in

predicting adjustment for a population that appears to be at risk for psychological

distress, adolescents of physically ill parents. Specifically, the focus was to investigate

the simultaneous influence of the ill parent-adolescent relationship and healthy parent-

adolescent relationship on adolescent adjustment. It was hypothesized that the healthy

parent relationship would be more strongly related to adolescent adjustment than the ill

parent-adolescent relationship. Understanding the relative importance of this relationship

is a contribution to the literature that has direct clinical implications.

The current document begins with a description of the prevalence of parental

illness, followed by a review of studies examining the impact of stressors during


adolescence. Next, literature on the impact of parental physical illness on children's

functioning is reviewed, including theoretical models describing mechanisms by which

this effect may be exerted. An examination of studies investigating these variables

follows. Finally, a study conducted to test the stated hypotheses is described, with

findings from the data reported and implications of those data discussed.


Prevalence of Chronic Illness in Parent Populations

Parental illness can represent a highly salient stressor for children and

adolescents. A disease and its treatment can be an emotionally intense experience for

both the patient and family members and can be characterized by a variety of emotions

including fear, anxiety, anger, confusion, and depression. Illness can result in extensive

changes throughout the family and give rise to a great deal of uncertainty regarding such

factors as disease course, heritability, and outcome. Children of ill parents are often

faced with the potential for both short-term losses, such as parental hospitalizations and

disrupted routines, and long-term losses, including parental death and ongoing family

disruption (Leedham & Meyerowitz, 1999). Rait and Lederberg (1989), commenting on

families of cancer patients specifically, referred to family members as "second-order

patients" as a result of the stressors to which they are exposed. Yet despite recognition of

these experiences, children of ill parents have received little attention in the research


Statistics describing the prevalence of chronic illness in adults in their child-

rearing years are impressive. Six of the ten leading causes of death for adults age 25 to

44 are chronic conditions, including malignant neoplasms, heart disease, and HIV. In

1997, these resulted in over 59,000 deaths in this parenting age group (Hoyert, Kochanek,

& Murphy, 1999). In addition, there are thousands of other adults who live with chronic

diseases affecting the heart, liver, kidney, lung, and pancreas. For instance, the United

Network for Organ Sharing (UNOS) reported that as of January 17, 2000, the number of

individuals between the ages of 18 and 49 on the waiting list for solid organ transplants

was over 35,000 (, and it is estimated that over half of patients

evaluated for transplantation are not listed for various medical and psychological reasons.

In addition, from 1994 to 1998 transplant centers reported performing 56,523 solid organ

transplants on patients in this age range. In short, significant numbers of families are

affected by health conditions serious enough to warrant organ transplantation. Millions

more are affected by other conditions that are debilitating but not necessarily life-

threatening. For instance, for every 1000 adults in the United States between the ages of

18 to 44, there are 51.6 people with asthma, 46.9 with arthritis, 11.4 with diabetes, and

5.8 with epilepsy (Benson & Marano, 1998).

Adolescents and Stress

The effects of stress in adults have been the focus of scientific study for many

years. In comparison, research on the effects of stressful life circumstances in

adolescents began only recently (Compas, Slavin, Wagner, & Vannatta, 1986). A life

events approach, focusing on significant changes in an individual's life, has provided the

conceptual framework for much of the work in this area. Empirical research has

suggested that stressful life events are significantly related to child and adolescent

adjustment in several areas, including internalizing behaviors, externalizing behaviors,

and school performance (Johnson, 1986). This finding has been demonstrated in both

cross-sectional research (e.g., Forehand, Middleton, & Long, 1987) and prospective

studies (e.g., Siegel & Brown, 1988).

Studies indicate that as the number of concurrent stressors in an adolescent's life

increases, adolescent functioning decreases. This relationship has been demonstrated

across several domains, including academic functioning (Forehand et al., 1991; Forehand

et al., 1987; Simmons, Burgeson, Carlton-Ford & Blyth, 1987), internalizing symptoms

(Compas et al., 1986; Forehand et al., 1991; Siegel & Brown, 1988), somatic complaints

(Compas et al., 1986; Siegel & Brown, 1988), self-esteem (Simmons et al., 1987),

externalizing problems (Forehand et al., 1991), and participation in extracurricular

activities (Simmons et al., 1987). These findings have been demonstrated with both

familial stressors, such as divorce and parent-adolescent conflict (Forehand et al., 1991;

Forehand, Middleton, & Long, 1987), and normal developmental events, such as dating

and transition to junior high school (Simmons et al., 1987). These data generally show

linear relationships. However, some data suggest that one or two stressors have relatively

few effects on adolescents, but three or more stressful life events are associated with a

sharp increase in problems in areas of adjustment noted above (Forehand et al., 1987;

Simmons et al., 1987). There is also some evidence to suggest that the negative impact

of family stressors on adolescents may not be fully seen until young adulthood

(Forehand, Biggar, & Kotchick, 1998).

Simmons and colleagues (1987) proposed that the stress of life events can be

eased for the adolescent if he or she has "arenas of comfort" (p. 1232) where the

individual can withdraw to regroup. The parent-adolescent relationship may be one such

arena for adolescents experiencing significant stressors. However, the proposition by

Simmons et al. (1987) implies that adolescent satisfaction with the "arena," here, the

parent-adolescent relationship, is important. Indeed, adolescent satisfaction with the

social support, not the number of support mechanisms available, has been related to

fewer psychological symptoms (Compas et al., 1986). Research has also specifically

supported the notion that parent-adolescent relationships with less conflict are associated

with better adjustment for adolescents under stress (Forehand et al., 1987; Forehand et

al., 1991; Wierson, Forehand, Fauber, & McCombs, 1989). For example, Neighbors,

Forehand, and McVicar (1993) found that a less conflictual relationship with their

mothers distinguished adolescents rated as resilient to stress from those who were not. In

somewhat conflicting findings, Forehand et al. (1991) found that adolescent perceptions

of the father-adolescent relationship exerted a main effect on externalizing problems

regardless of the number of family stressors experienced. The same was true for

adolescent perceptions of the mother-adolescent relationship and its relationship with the

adolescent's grade point average. However, an interaction effect was noted for

adolescent internalizing problems. As the number of stressors in the adolescent's life

increased, a positive father-adolescent relationship was associated with fewer

internalizing symptoms. Forehand and colleagues, citing research by Montemayor and

Hanson (1985), suggested that the mother-adolescent relationship is typically more

conflictual than the father-adolescent relationship. Therefore, adolescents may rely on

fathers more than mothers in times of stress. Regardless of the source of this finding, it

supports the notion that a good parental relationship can have a stress-buffering effect for

an adolescent.

However, one flaw of many of the studies examining adolescent stress is that

stressors are equated. Unlike adult research on life stress, which uses "life change units"

to weight the presumed impact of various stressors (Johnson, 1986), adolescent stress

research has not accounted for differing effects of adverse life events. For example, in

the study by Simmons and colleagues (1987), parental death and entering puberty were

each counted as "one stressor" despite vast differences in the impact of these two events.

Because of this, research on the effects of different types of stressors remains unclear.

Some work exploring specific adolescent life events, such as parental divorce, has

been conducted. The divorce literature has reported findings similar to those of the

general adolescent stress research. For example, in a study of male adolescents, Wierson

et al. (1989) reported that adolescents who felt they had good relationships with both

divorced parents were rated by teachers as exhibiting better academic performance and

fewer conduct problems than did adolescents who described poor relationships with both

divorced parents. The researchers detected no significant differences between

adolescents from divorced families and intact families as long as the adolescent perceived

good relationships with both parents.

Collectively, these data highlight two important ideas for the current

investigation. First, adolescents appear to be at risk for negative consequences when

experiencing several stressful life events. Second, parent-adolescent relationships may

have a critical role in buffering these negative effects.

Parental Physical Illness and Child/Adolescent Functioning

Despite the staggering number of parents who have chronic health conditions, our

knowledge of the impact of parental illness on children's psychological adjustment

remains limited (Armistead, Klein, & Forehand, 1995; Drotar, 1994). Nevertheless,

recent reviews of the existing literature concluded that parental physical illness appears to

be associated with problems in children's functioning (Armistead et al., 1995; Worsham,

Compas, & Ey, 1997), although the extent of these problems remains somewhat

controversial (Kahle & Jones, 1999). Several hypotheses have been offered regarding the

relationship between parental illness and children's adjustment problems. Wellisch et al.

(1992), for instance, suggested that parental illness can draw attention away from

children, as well as drain emotional and material resources. Moreover, threats to security

in the parent-child relationship may be felt by children and adolescents through increased

parental unavailability, changes in daily routines, or fear of parental death (Armsden &

Lewis, 1993). Concern for the parent's well being may also cause adjustment difficulties

for children with a chronically ill parent. Various possibilities have been proposed, and

explanations for these problems are likely to be influenced by the child's stage of

development. Regardless of the mechanism by which parental illness may put children at

risk, studies suggest that children of ill parents are more likely to exhibit higher levels of

psychological symptoms than children of healthy parents (see Table 1).

In a study comparing children of mothers who had either chronic pain, diabetes,

or no illness, self-reports of depression, but not anxiety, were significantly higher for

children of mothers with chronic pain than those with no illness (Dura & Beck, 1988).

Scores for children of mothers with diabetes fell between those of the other two groups,

but were not significantly different from either of them. A nonsignificant trend emerged

toward higher levels of parent-reported total behavior problems for children of parents

with chronic pain or diabetes when compared to children of parents with no illness.

However, since each group in the study contained only seven subjects, the sample size

may have hindered detection of effects. In addition, for parent reports only total behavior

problem scores were reported; therefore it is not possible to determine whether these

Table 1

Summary of Findings from Empirical Studies of Children of Parents with Chronic
Physical Illness
Authors Parent/ n Age Main findings
illness range
Armistead et al. Fathers/ 67 3-18 A more positive parent-child
(1997) hemophilia relationship was related to less child
and HIV depression and externalizing problems
and better school performance.
Biggar et al. Mothers/ 85 6-11 Children of HIV-positive mothers
(1998)* HIV reported more depressive symptoms
than children of HIV-negative
mothers. Maternal depressive
symptoms were not related to
children's depression. Children's
depression scores did not differ based
on whether the mother was
Chun et al. Mothers 35 6-16 Teachers, but not parents, reported
(1993) and fathers/ that children of chronic pain patients
chronic had more behavior problems and less
pain social competence. Parent functional
disability, but not the patient's gender,
depressive symptoms, or marital
satisfaction, predicted parent-reported
child behavior problems.
Compas et al. Mothers 110 6-32 Disease characteristics were generally
(1994) and fathers/ not related to anxiety/depression or
cancer stress-response symptoms for
adolescents. Perceived seriousness
and perceived stressfulness were
associated with stress-response
symptoms. Self-reports of anxiety/
depression were significantly higher
for adolescents than for younger
children. Adolescent girls with ill
mothers reported the most difficulties.
Compas et al. Mothers 134 6-32 Adolescents reported using more
(1996) and fathers/ emotion-focused coping than younger
cancer children. Emotion-focused coping
was related to symptoms of adolescent
Dorsey et al. Mothers/ 100 6-11 Presence of additional adults in the
(1999) HIV home was not predictive of better
child psychosocial functioning.

Table 1. Continued
Authors Parent/ n Age Main findings
illness range
Dura & Beck Mothers/ 14 7-13 Self-reports of depression, but not
(1988)* chronic anxiety, were significantly higher for
pain, children of mothers with chronic pain
diabetes than those of healthy moms. Scores
for children with diabetes were
between the two groups. A
nonsignificant trend toward greater
parent-reported behavior problems
emerged for children whose parents
had chronic pain or diabetes.
Forehand et al. Fathers/ 137 3-18 Parent-reported internalizing
(1997) hemophilia symptoms differed between children
(some with whose parents had hemophilia only
HIV) and those whose parents were also
HIV-positive but asymptomatic. More
conflict was reported in the mother-
child relationship when the father's
HIV status was positive, and worse
still when he was HIV symptomatic.
Hemophilia severity was not
associated with parent reports of child
Forehand et al. Mothers/ 87 6-11 Self-reported depression and
(1998)* HIV aggression were significantly higher
for children of HIV-infected mothers.
Parent-reported social competence and
cognitive competence were
significantly lower for children of
infected mothers. Several mean
scores for children of infected mothers
were in the borderline clinical or
clinical range.
Forsyth et al. Mothers/ 26 6-16 Significant differences were observed
(1996)* HIV between children of infected vs.
uninfected mothers on parent-reported
withdrawal, anxiety/depression, social
competence, and attention problems.
Significant differences were observed
between children of symptomatic and
asymptomatic HIV-positive mothers
on self-reported anxiety and parent-
reported anxiety/depression.

Table 1. Continued
Authors Parent/ n Age Main findings
illness range
Grant & Mothers 55 11-18 Girls whose mothers were ill reported
Compas (1995) and fathers/ more family responsibilities, which
cancer accounted for the interaction between
sex of ill parent and sex of adolescent
in predicting anxiety/depression.
Hirsch, Moos, "Parents"/ 16 12-18 Adolescents of parents with arthritis
& Reischl rheumatoid participated in fewer school activities
(1985)* arthritis and reported lower self-esteem than
controls. For many variables there
were also no significant differences
between adolescents of parents with
arthritis and adolescents of parents
with depression.
Howes et al. Mothers/ 32 4-18 Mothers who reported more
(1994) breast psychological distress rated their
cancer children as having more emotional and
behavior problems.
Kotchick, Mothers/ 86 6-11 HIV-infected mothers reported more
Forehand et al. HIV parent-child conflict and less
(1997)* monitoring of their children than
uninfected mothers. These two
variables were important predictors of
child-reported internalizing and
externalizing problems and parent-
Ireported social competence.
Kotchick, Fathers/ 53 7-18 Perceived parental social support was
Summers et al. hemophilia significantly predictive of parent-
(1997) (some with reported externalizing problems, self-
HIV) reported depression, and academic
competence. In addition, child
perceptions of parental support
significantly predicted parent-reported
internalizing symptoms and self-
reported depression in families where
parents reported high physical and
_psychological impact of illness.
Lewis et al. Mothers/ 40 6-12 The peer relations of children whose
(1993) breast mothers had breast cancer were
cancer correlated with the quantity and
quality of exchanges between the
healthy parent and the child, but were
not correlated with ill parent-child

Table 1. Continued
Authors Parent/ n Age Main findings
illness range
Mikail & von Unstated/ 24 9-17 Children of chronic headache sufferers
Baeyer (1990)* migraine exhibited more problems in general
headache adjustment, internalizing problems,
somatic complaints, externalizing
problems, and social skills than
children of healthy controls. Mean
scores were within normal limits.
Peters & Esses Mothers 33 12-18 Compared to controls, adolescents
(1985) and fathers/ from families with a parent with MS
multiple rated their families higher on conflict
sclerosis and lower on cohesion.
Rodrigue & Mothers 33 11-15 Presence of a chronic health condition
Houck (2001)* and fathers/ in fathers, but not mothers, was related
various to more behavior problems in
illnesses adolescents. Nonsignificant trend
toward interaction between sex of ill
parent and sex of adolescent in
predicting behavior problems.
Rotherham- Mothers 239 11-19 Parental ill health was related to
Borus & Stein and fathers/ adolescent self-reports of internalizing
(1999) AIDS symptoms and somatic complaints.
Rotherham- Mother and 211 11-18 Better adolescent adjustment predicted
Borus, Stein, & fathers/ by lower parental distress and physical
Lin (2001) HIV ___symptoms in a two-year follow-up.
Siegel et al. Mothers 42 7-16 Target children had significantly
(1992) and fathers/ higher levels of self-reported
terminal depression and anxiety and more
cancer parent-reported internalizing and
externalizing behavior problems than
controls. They also had lower self-
esteem and lower parent-reported
social competence.
Steele, Fathers/ 69 3-18 Healthy parent-child relationship
Forehand, & hemophilia problems accounted for nearly twice
Armistead (some with as much variance in children's
(1997) HIV) adjustment as ill parent-child
relationship problems, although both
were significantly associated with
child internalizing problems.

Steele, Tripp, et
al. (1997)

(some with



The child's uncertainty about the
father's illness predicted children's
self-reports of both anxiety and

Table 1. Continued
Authors Parent/ n Age Main findings
illness range
Stein et al. Mothers 183 11-19 Taking on adult roles was associated
(1999) and fathers/ with having an ill mother, being
AIDS female, and greater parent drug use.
Follow-up 3-9 months later indicated
that taking on adult roles was
associated with more behavior
Wellisch et al. Mothers/ 30 18-65 No differences found on measures of
(1991)* breast long-term psychological adjustment.
cancer Differences were noted on measures
of sexual functioning.
Wellisch et al. Mothers/ 60 18-65 Retrospective accounts of discomfort
(1992) breast regarding their mother's illness were
cancer greater for women who were
adolescents at the time than for
women who were adults.
Zahlis (2001) Mothers/ 16 11-18 Retrospective accounts indicated that
breast children had many worries during
cancer their mother's illness and made
considerable effort to make sense of
the experience.
Note. = Sample size listed for between subjects studies reflects only number of subjects
in the target group (i.e., control children are not included).

problems were of an internalizing or externalizing nature. The authors also do not state

which parent or parents rated their child, which may be relevant to interpreting findings.

Finally, the clinical meaningfulness of the findings is questionable since all mean scores

fell in the nonclinical range.

Chun, Turner, and Romano (1993), using a somewhat larger sample, examined

behavior problems and social competence of children whose parents were either healthy

or living with chronic pain. Combined parent ratings did not indicate significant

differences on either construct. However, a strength of this study was its use of teacher

ratings, which showed a pattern of results that differed from parent reports. Teachers

rated children of pain patients as having significantly more behavior problems and being

significantly less socially competent than children of control families, although mean

scores were within nonclinical limits. The authors suggest that behavior problems in

these children may be manifested more often outside the home or that they may be more

noticeable by outside observers. Like Dura and Beck (1988), these authors reported total

behavior problems, thus leaving the nature of the problems in this sample unclear.

Another methodological weakness of this study is the fact that parents with more than

one child were allowed to select which child they rated for the study. Parents may have

chosen children they perceived as having few problems, thus leading to a conservative

estimate of problems in the sample. The lack of self-report measures also weakens the

conclusions that can be drawn.

In another study, children of parents with chronic headaches were reported by

their ill parent to exhibit significantly more internalizing behaviors, externalizing

behaviors, somatic complaints, and social skills deficits than children whose parents were

healthy (Mikail & von Baeyer, 1990). However, mean scores were in the nonclinical

range, and no significant group differences on anxiety or withdrawal were reported.

Hirsch, Moos, and Reischl (1985) compared adolescents of parents with arthritis

to those of parents with depression and those without a health condition. Adolescents of

patients with arthritis reported significantly lower self-esteem and less participation in

school activities. Scores on measures of psychological symptoms for adolescents of the

parents with arthritis were consistently between those of teens with depressed parents and

those of healthy parents and not significantly different from those of children of

depressed parents. The study also examined the impact of life events on psychological

symptomatology. Their results revealed that children of ill parents reported significantly

more negative events in the past twelve months than did healthy parents and that both

positive and negative life events were associated with increases in symptoms for

adolescents whose parents had arthritis. This was not true for children of healthy parents.

This may indicate that any changes in the lives of these young people can negatively

disrupt their well being.

Studies of the adjustment of children with physically ill parents are not limited to

the chronic pain literature. Wellisch et al. (1991), in a study of adult daughters of

mothers who had breast cancer, found no differences on a variety of psychological

dimensions, including depression and somatization, when compared to a matched control

group of women. The only differences emerged on measures of sexual satisfaction and

frequency of sexual intercourse. However, this study explored the long-term adjustment

of individuals with an ill parent and does little to illuminate concurrent adjustment. In

addition, the sample consisted entirely of upper class Caucasian women, which may have

provided resources that aided in the adjustment of these individuals.

In a separate retrospective study of adult women who had had a parent with

cancer, Leedham and Meyerowitz (1999) also found no significant differences on

measures of psychological adjustment when compared to matched controls.

Qualitatively, the women recalled several ways in which their experiences with their

parent's cancer had affected their lives. In addition to negative outcomes such as feeling

more fearful of death, these grown children noted positive changes as well, including

feeling they were stronger people and appreciating others more. However, they also

reported significant disruption in family functioning caused by the disease during the

illness and its treatment.

Concurrent reports of children whose parents have terminal cancer have suggested

more psychological difficulties when compared to matched controls (Siegel et al., 1992).

Children of ill parents, whose prognosis was less than six months survival, reported

significantly more depressive symptoms, more anxiety symptoms, and lower self-esteem

than community control children. Parent-report measures, completed by the healthy

parent, also indicated significantly higher levels of both internalizing and externalizing

behavior problems and lower social competence. Average scores on all measures for this

sample were in the subclinical range, however, several children reported clinically

significant depressive symptoms. Nearly twice as many of the target children (27%)

were in the clinical range on self-reported depression. In addition, significantly more

target children were in the clinical range for parent-reported externalizing problems, total

behavior problems, and social competence problems. Interpreting these findings for

purposes of the current investigation is problematic, since it is difficult to disentangle the

impact of illness from the impact of the impending death of a parent. However, given

that many patients pursuing transplants are given a two-year prognosis for survival

without a transplant, the issue of severity can be likened to some degree. Also,

participants in this study were predominantly middle to upper class, a problem common

to studies in this literature (e.g., Lewis, Hammond, & Woods, 1993; Wellisch et al.,


Families with lower socioeconomic status (SES) have been studied, mostly in

describing the adjustment of healthy children whose mothers are HIV-positive. For

instance, Forehand et al. (1998) conducted a study of inner city, low SES, African-

American children whose mothers were HIV-positive. Only 20% of the children in the

study were aware of their mother's diagnosis. Nonetheless, the investigators found

significant differences between children of HIV-positive parents and children of healthy

parents on measures of child-reported depression and aggression. Maternal reports of

these constructs did not reveal statistically significant differences; average scores for both

groups were in the borderline clinical and clinical ranges. However, maternal reports of

social competence and cognitive competence indicated significantly less competence for

children of infected mothers when compared to uninfected mothers. No measures of

anxiety were included in the study. That significant findings for self-report measures

were obtained in a sample in which a small number of children were aware of their

mother's diagnosis would suggest that circumstances that accompany the disease, not the

disease itself, put children at risk for psychological difficulties. However, an important

caution should be considered with this study. Although control families were from the

same community, matching procedures were limited to the school, age, and gender of the

child. The authors note that this omits maternal variables important to this population

(e.g., maternal drug use) that may also differentiate these two groups. Also relevant for

the current study is the fact that the vast majority of the mothers in this sample were

single. Presumably, a healthy parent was not available from whom children might seek

support, which may be an additional consideration in the unusually high scores on

measures of child maladjustment.

Forsyth et al. (1996) also reported psychological difficulties in this population.

While no differences in self-reported anxiety between children of HIV-positive and HIV-

negative mothers were revealed, statistically significant differences in anxiety symptoms

were observed between children of symptomatic and asymptomatic mothers. In addition,

parents reported more symptoms of anxiety/depression, withdrawal behavior, and

attention problems in children of infected mothers compared to uninfected mothers.

Parent reports of somatic complaints, externalizing problems, and social problems were

not significantly different between the two groups. Although there was only a

nonsignificant trend toward greater parent-reported total behavior problems, 52% of the

target children was in the clinical range on this construct, whereas 32% of the control

group was in this range. No differences were detected between children of symptomatic

and asymptomatic mothers on any parent-report measure, except anxiety/depression

symptoms. Similar to the Forehand et al. (1998) findings, less than half of the children in

the study had been told of their mother's disease.

Biggar et al. (1998), using the same sample as Forehand et al. (1998), and Forsyth

et al. (1996) found higher self-reported depression scores for children of infected vs.

uninfected women. Average scores in these studies were similar to the findings of Siegel

et al. (1992). Also similar to that study, greater numbers of target children than control

children were in the clinical range for depression. Thirteen percent and 19% of children

with infected mothers were in the clinical range in the Biggar et al. (1998) and Forsyth et

al. (1996) studies, respectively, compared with 4% from control groups in each study.

This difference was statistically significant in the former study but nonsignificant in the


Studies using between-subjects designs to determine whether the concurrent

psychological functioning of children with an ill parent differs from that of children with

healthy parents typically note discrepancies. Many children from families with the

stressor of an ill parent appear to function relatively well, yet it is clear that some do not.

While scores on measures of psychological functioning for children of ill parents are

typically in the nonclinical range of measure norms, they are also often significantly

higher than scores of children of healthy parents. In addition, several studies reported

that more children with ill parents exhibited clinical levels of psychological symptoms

than children of healthy parents (Biggar et al., 1998; Forsyth et al., 1996; Siegel et al.,


While other reviews have concluded that these differences exist mainly in

internalizing symptoms (Kotchick, Summers, Forehand, & Steele, 1997; Worsham et al.,

1997), the present review highlighted other areas of functioning that may be affected as

well. Although results are mixed, several studies reported significantly higher levels of

self-reported (Forehand et al., 1998) and parent-reported (Mikail & von Baeyer, 1990;

Seigel et al., 1992) externalizing problems. In addition, lower social competence was

described in several studies using both teacher (Chun et al., 1993) and parent (Forehand

et al., 1998; Mikail & von Baeyer, 1990; Siegel et al., 1992) reports. Nonetheless,

internalizing symptoms were often associated with parental physical illness. However,

differences in internalizing problems were more uniformly observed in children's self-

reports than parent reports. Within children's self-reports, depression revealed consistent

differences (Dura & Beck, 1988; Forehand et al., 1998; Forsyth et al., 1996; Siegel et al.,

1992), while self-reports of anxiety were less consistent. However, one possibility is that

severity of illness may be an important predictor of anxiety reactions. Forsyth et al.

(1996) found no differences in anxiety between children of ill and healthy parents;

however, differences did emerge between symptomatic and asymptomatic parents.

Siegel et al. (1992) studied children of parents with a terminal illness and found

differences on measures of anxiety as well. It may be that as children are provided with

more salient cues that their parent's illness is worsening, symptoms of anxiety may be

more likely to arise in response. This hypothesis is consistent with conclusions that

particular family or illness factors may be related to specific child adjustment domains

(Armistead et al., 1995; Steele, Tripp, Kotchick, Summers, & Forehand, 1997). In

reference to the present study, it is important to note that the decision to pursue

transplantation, radiation, or dialysis may have signified a cue to adolescents that their

parent's condition was worsening, thus increasing the possibility of an anxious reaction.

Finally, these findings suggest important considerations of reporter issues. First,

the consistent findings of elevated internalizing symptoms in children's self-reports,

combined with only mixed findings on parent reports, may suggest that some of the

adjustment problems in these children are often covert. If so, these easily could be

overlooked in families experiencing more salient stressors, such as parental physical

frailty or economic distress. Dorsey, Watts Chance, Forehand, Morse and Morse (1999)

have suggested that as a parent's condition worsens, his or her ability to detect their

child's difficulties may decline. Indeed, Worsham and colleagues' (1997) review of this

literature concluded that evidence for maladjustment is clearer from children's and

adolescents' self-reports than from parents' reports. Second, three of the studies

reviewed above used ill parents as reporters, one used the healthy parent, one averaged

two parents' ratings, and one used unclear methodology. The only study that consistently

found more problems on all domains measured was that using the healthy parent as a

reporter (Siegel et al., 1992). It may be that healthy parents are in a special position to

detect difficulties for children in these families. While this is a very tentative suggestion,

given the limited data and other unique aspects of the study such as the terminal

population examined, rarely is the role of the reporter in the family and potential biases

from this role discussed when considering results. There is research to suggest that

mothers and fathers do not report different numbers of problems (Achenbach, 1991b;

Stanger & Lewis, 1993) and that there is no sex of parent by sex of adolescent effect in

the number of problems described by parents (Achenbach, Howell, Quay, & Conners,

1991; Stanger & Lewis, 1993). Given that these factors do not appear to have a

significant impact on reports, the potential impact of the reporter's health status should be


In summary, these data highlight three important conclusions relevant to the

current investigation. First, children and adolescents whose parents have chronic health

conditions appear to be at risk for internalizing symptoms. It can be argued that these

symptoms may be of greater concern for these young people because other stressors in

the family may prevent adults from detecting and/or intervening. Second, anxiety

symptoms especially may be relevant when increasing illness severity becomes salient to

the child or adolescent. Finally, reporter issues may influence findings, therefore

multiple sources of information (e.g., child, ill parent, and healthy parent) are critical to

research in this area.

Models of Parental Physical Illness and Child Functioning

Four models explaining the impact of parental illness on child functioning have

been proposed in the literature. Two of these models have been explicitly tested. This

section presents these models with critiques specific to each model. In the next section, a

broader conceptual critique of the models is presented.

Lewis et al. (1993) developed and tested a model of child and family functioning

when the mother has breast cancer (Figure 1). This complex model hypothesized that

environmental factors (e.g., socioeconomic status, social support) predicted perceived

illness demands, which predicted parental depression, marital adjustment, and family

coping. The latter variables were hypothesized to indirectly predict child psychosocial

functioning through the parent-child relationship. In addition, marital adjustment and

family coping were suggested to directly predict children's adjustment.

This model was tested separately for mothers and fathers, and results for both

models were similar. Illness demands and, for mothers only, social support significantly

predicted symptoms of depression, which predicted marital adjustment. Lower marital

satisfaction predicted, for both parents, poorer family coping and, in the mother model

only, poorer parent-child relationships. In the mother (ill parent) model, family coping,

but not the parent-child relationship, significantly predicted child psychosocial

functioning. In the father (healthy parent) model, the parent-child relationship

significantly predicted child adjustment. This suggests that the two parent-child

relationships served different functions for the adolescents in this study, such that their

interactions with the ill parent had little impact on adjustment, while relationships with

the healthy parent were highly correlated with positive functioning.

This study represents the first published conceptual model of children's

adjustment to an ill parent. It includes a number of important variables that previously

had been unstudied. However, one major criticism of this model is its lack of attention to

child variables (e.g., age, sex) as predictors of child functioning, despite evidence

suggesting that these variables may have important roles. In addition, child adjustment in

Figure 1. Model of how maternal breast cancer influences family functioning proposed
by Lewis, Hammond, and Woods (1993).

the study was measured by a single questionnaire subscale, "peer relations," and the

study was limited to preadolescent children. Therefore, the generalizability of

conclusions drawn from this rather specific domain of functioning and from this age

group is questionable and limits interpretation of the findings.

Based on their review of the research, Armistead et al. (1995) proposed a model

of how parental physical illness influences child adjustment (Figure 2). In this model,

disrupted parenting is the key mechanism by which parental illness affects child

functioning. Parental physical illness is hypothesized directly and indirectly to predict

parenting. Indirect mechanisms include increased relationship conflict and greater

parental depression, which are also hypothesized to predict parenting disruptions. This

model is consistent with some of the research from the general adolescent stress literature

(e.g., Ge et al., 1994). While not incorporated into the model, another strength of the

authors' discussion of the proposed model is the issue of reciprocal relationships. The

authors note that constructs such as parental depression and relationship conflict may

have an impact on the parent's physical health.

Relationship Parental
Conflict Divorce

Parental Disrupted Child
Physical --- Parenting


Figure 2. Model of how parental physical illness influences child functioning proposed
by Armistead, Klein, and Forehand (1995).

In what appeared to be a revision of this model, Steele, Forehand, and Armistead

(1997) reported empirical findings for a proposed model for predicting child internalizing

problems when a parent is chronically ill (Figure 3). This model does not focus on

disrupted parenting as the primary mechanism affecting children, but rather maintains

that illness severity affects family process variables (i.e., marital conflict, parental

maladjustment, and parent-child relationship problems). Family process variables, in

turn, have both a direct impact on children's internalizing problems as well as an indirect

impact through children's coping strategies. This model was examined separately for

mothers and fathers, using families in which the fathers had hemophilia, some of whom

were also HIV-positive. Variables in the mother (healthy parent) model accounted for

45% of the variance in children's internalizing symptoms. Significant paths existed from

illness severity to maternal depression and from maternal depression to child

internalizing problems. The paths from mother-child relationship problems, measured

here by parent-child conflict, to children's avoidant coping and to children's internalizing

problems were also significant. Finally, the path from children's avoidant coping to

children's internalizing problems was significant. No paths to or from marital adjustment

were significantly related to any other factors in the model. Variables in the father (ill

parent) model accounted for 41% of the variance in internalizing symptoms of the

children in this sample. Significant paths from illness severity to marital adjustment and

paternal depression emerged. Paternal depressive symptoms were significant predictors

of marital adjustment, father-child relationship problems, and child internalizing

problems. As with the mother model, the paths from father-child relationship problems

to both children's avoidant coping and to children's internalizing problems were

significant, as was the path from children's avoidant coping to children's internalizing

problems. Notably, the path from parent-child relationship problems accounted for

nearly twice the variance in the mother model as it did in the father model, again

suggesting a relatively more important role for the healthy parent-child relationship in

predicting children's adjustment.

Figure 3. Model of relationships among parental physical illness
problems proposed by Steele, Forehand, and Armistead (1997).

Another important conceptual conclusion can be drawn from the results of this

study. The authors stress that while illness severity was indirectly related to child

internalizing problems, the direct path to child adjustment was not significant. This

suggests that it is not the illness itself, but rather the effects it has on the family, that

affect child psychological functioning. However, the measurement of the illness

severity construct in this study is of concern. First, illness severity was determined by

combining diverse objective measures of illness (HIV status, bleeds per year, per cent

of clotting factor) and subjective parent ratings of the impact of the illness. Therefore,

the roles of objective severity indicators and illness perceptions cannot be teased apart

to determine their respective impacts on the model. Second, children's perceptions of

their parent's illness severity were not considered in the illness severity construct.

Presumably, a child's cognitions about their parent's condition would be useful in

examining their adjustment.

A final model from the Forehand research group (Family Health Project

Research Group, 1998) has been proposed explicitly for families of mothers with HIV

(Figure 4). Because of the nature of HIV and its associated patient demographic

characteristics in the sample chosen, this model is significantly different from the other

models. Environmental stressors and economic resources are more fully incorporated

into this model. Interestingly, the parent-child relationship, which was strongly

associated with functioning in the previous study, is not included in this proposal.

However, parenting, absent from the previous revision, appears again in this model.

Child factors, including coping strategies, self-sufficiency, and knowledge of the

illness, are viewed as important predictors.

Figure 4. Model of factors associated with how maternal HIV influences child
psychosocial functioning proposed by The Family Health Project Research Group (1998).

Critique of Current Models

These models have provided a framework for research in this area thus far. All of

them share a common element in that they are focused on parent and family functioning

as the critical predictors of child adjustment. However, the emphasis on illness and child

factors varies from model to model. More recent findings have improved our knowledge

of child functioning in families of an ill parent, and several limitations of these models

should be noted. First, in light of our current knowledge, each model omits variables

empirically demonstrated to be correlated with child adjustment. For example,

developmental factors, child sex, and patient sex are not included in current models. The

first two models by the Forehand group are relatively simplistic; each includes five

variables considered to be predictive of child adjustment. The last model omits the

parent-child relationship completely and, because of the sample studied, is a single-parent

model. While it may be useful in predicting the adjustment of children in families with a

mother who has HIV, this model likely would not generalize to families with other


This raises a second concern. With the exception of Armistead et al. (1995), the

models proposed have been developed for specific diseases, not illness in general. While

specific disease characteristics should not be ignored, the impact of parental illness as a

general stressor on the adjustment of children and adolescents has been understudied.

Mechanisms common to diseases remain unclear, limiting the generalizability of this


Both of the models that have been empirically examined have tested the impact of

parents on child adjustment for patients and partners separately. Neither model includes

the co-occurring influences of both parents. Given that evidence suggests that the ill and

healthy parents may serve different roles in these families, it would appear to be

beneficial to include separate pathways in the same model to study relative differences in

prediction of adjustment. Simultaneously considering the influences of each parent on

the adolescent, as well as on each other, is important and remains unexplored in the

current literature.

Finally, the nonrecursive nature of these models is problematic. As previously

noted, Armistead et al. (1995) suggested the importance of considering reciprocal

relationships when considering the context of the family. For example, they suggest that

parental physical illness can influence parental depression, and, in turn, depression can

influence physical illness. There are many other relationships in this literature that

should be considered reciprocal that are typically not. Child adjustment is viewed as the

dependent variable of parental independent variables, such as depression or parent-child

relationships. However, child adjustment easily can be viewed as influencing these

parental variables as well, creating vicious cycles for maladjustment. While the

methodology of the present study did not allow for empirical tests of this argument, the

interpretation of findings should consider the possibility of reciprocal relationships.

The models presented have provided valuable guides for the literature in testing

variables believed to be related to child and adolescent adjustment. Studies testing these

and other relevant constructs have provided new information about predictors of

adjustment in children with ill parents.

Factors Associated with Child/Adolescent Functioning

Several factors have been examined as predictors of children's psychological

adaptation to parental physical illness. While few variables have been examined with

multiple samples, preliminary conclusions can be drawn regarding the relevance of

factors potentially affecting child outcomes.

Child Factors

Children's psychological adjustment to parental illness has been found to be

associated with the child's age. For example, Compas et al. (1994) found that

adolescents reported significantly more anxiety/depression symptoms than did young

children. Wellisch et al. (1992) found that adults who were adolescents at the time of

their mother's breast cancer diagnosis retrospectively reported feeling greater discomfort

about involvement with their mother's illness than did women who were over twenty

years old. In contrast, some researchers have found no relationship between child age

and adjustment to parental illness (e.g., Kotchick, Summers, et al., 1997).

Theoretically, it has been suggested that parental illness may interfere with

classical conceptualizations of the adolescent developmental tasks of separation and

individuation (Armsden & Lewis, 1993; Wellisch et al., 1992). Armsden and Lewis

(1993) suggested that the family's needs for cooperation and cohesiveness may conflict

with the adolescent's needs to become more involved in roles outside of the family and to

express oneself as an individual. Adolescents with an ill parent may feel drawn back to

the family due to the illness, causing distress or self-loathing over these conflicting roles.

In a qualitative study of adolescents of terminal cancer patients, Christ, Siegel, and

Sperber (1994) reported that typical adolescent themes, such as sibling conflict and

complaints regarding parental supervision, seemed accentuated in this population. In

addition, their interviews revealed that adolescents in these families often felt divided

between achieving developmentally appropriate goals and the need to deal with the

practical aspects of having a terminally ill parent. The authors also reported that

adolescents were more likely than younger children to describe guilt about

developmentally appropriate feelings of anger toward their parent for being ill.

One reason for difficulties in separating from the family may be increasing

responsibilities at home. When a parent is ill, adolescents, especially girls, may be more

likely than younger children to be called upon to take on new roles in the family (Compas

et al., 1994; Grant & Compas, 1995; Stein et al., 1999). Family responsibility stressors,

such as chores, have been associated with higher internalizing symptoms in adolescents

whose parents have cancer (Grant & Compas, 1995) or AIDS (Stein et al., 1999).

Parental role-taking, such as helping with important family decisions, has been associated

with higher externalizing behavior problems in adolescents of parents with AIDS,

including sexual activity, substance use, and conduct problems (Stein et al., 1999).

Greater cognitive awareness also may contribute to elevated distress responses in

adolescents (Christ et al., 1994; Compas et al., 1994; Worsham et al., 1997). Christ et al.

(1994) reported that adolescents in their study appeared to comprehend both the illness

and its ramifications better than young children. It may be that a more thorough

understanding of the implications for the ill parent and for the family leads to more

difficulties for adolescents than for younger children who are unable to process the

nuances of the situation. In addition to understanding the immediate impact on one's life,

it has been suggested that daughters of mothers with breast cancer may be at risk for

adjustment difficulties because of the realization of the possible genetic risk for cancer

themselves (Wellisch et al., 1991). However, one empirical test of this hypothesis did not

reveal significant differences in depression/anxiety symptoms in adolescent girls whose

mothers did vs. did not have sex-linked cancers (Grant & Compas, 1995).

It has also been hypothesized that adolescents may be given more information by

adults than young children or asked by adults to provide information about the illness, yet

they may lack the necessary coping strategies for handling this information (Compas et

al., 1994). This may lead to maladaptive coping strategies. In fact, adolescents of

parents with cancer have been found to use more emotion-focused coping than

preadolescent children, which has been associated with more avoidance reactions

(Compas et al., 1996). This coping strategy was related to more symptoms of anxiety and


Main effects for child sex have not typically been found in studies reporting such

analyses (e.g., Chun, et al., 1993; Kotchick, Summers, et al., 1997; Rodrigue & Houck,

2001; Steele, Forehand, & Armistead, 1997). However, differences were detected in a

study of adolescents of parents with HIV, in which adolescent boys reported more

problem behaviors than girls and girls reported more emotional symptoms than boys

(Rotherham-Borus, Stein, & Lin, 2001). While atypical, these findings suggest that boys

and girls in this situation may manifest distress in different ways.

While not present in all studies (e.g., Chun et al., 1993), a significant interaction

between the sex of the child and the sex of the ill parent has been found in others.

Specifically, female adolescents whose mothers had cancer have more symptoms of

anxiety and depression than females with ill fathers or males whose mothers or fathers

were ill (Compas et al., 1994). Rodrigue and Houck (2001) found a similar, but

statistically insignificant, pattern of results for both female and male adolescents.

Multiple hypotheses regarding this interaction should be considered. First, there are

indications from the general parent-adolescent relationship literature that suggest that

differences exist in parent-child dyads. Steinberg (1987) has described these dyads as a

continuum of emotional involvement in which mother-daughters are most intense and

father-daughter relationships are most flat. It may be that these emotional interactions are

exacerbated by parental illness. Second, there is evidence to suggest that female

adolescents with ill mothers take on the responsibilities of that parent in the home, adding

to their life stress and internalizing symptoms (Grant & Compas, 1995; Stein et al.,

1999). Unfortunately, most studies in this area have used single-sex parent populations,

limiting the conclusions that can be drawn regarding the importance of the interaction

between these two variables.

In addition to age and sex, children's perceptions of their parent's illness appear

to be related to their psychological adjustment. For example, Compas et al. (1994)

reported that greater perceived illness severity was associated with higher levels of stress

response symptoms (avoidance and intrusive thoughts) in adolescents, but was not related

to symptoms of internalizing problems. The perceived stressfulness of the parent's

cancer was correlated with both stress response and internalizing symptoms. It is

important to note that the methodology used to assess children's perceptions (i.e., single

Likert-type ratings) and the timing of the assessment (i.e., near the time of the patient's

initial diagnosis) may explain the higher levels of stress response symptoms but no

difference for more stable anxiety and depression symptoms. No other studies have

explicitly examined the role of children's perceptions of their parent's illness in

predicting adjustment, although the perceived ambiguity children experience about their

parent's illness may be related to anxiety and depression (Steele, Tripp, et al., 1997).

While this finding does not specifically support the notion that perceived illness severity

is important to adjustment, it does indicate that children's perceptions of the situation do

have an impact on their functioning. While preliminary, these two studies suggest that

children's perceptions of their parent's illness should be examined further.

How a child copes with their parent's illness is another factor that may be

associated with their adjustment. For instance, Compas et al. (1996) found that

adolescents reported more emotion-focused coping and dual-focused coping, coping

strategies aimed at controlling both internal reactions and external characteristics of their

parent's cancer. The use of emotion-focused coping among children of ill parents was

correlated with more anxiety/depression symptoms and perceptions that the illness was

more stressful. Specifically, avoidance strategies were associated with worse disease

stage, perceptions of greater seriousness, perceptions of greater stressfulness, and more

anxious/depressive symptoms. Avoidant coping strategies were also predictive of child

internalizing problems in a study of fathers with hemophilia, some of whom also had

HIV (Forehand et al., 1997). The only significant predictor of child avoidant coping in

this study was problems in parent-child relationships, suggesting that these relationships

may be as important in these families as has been found in the general adolescent stress


Parent factors

Researchers (e.g., Kotchick, Forehand, et al., 1997; Lewis et al., 1993)

consistently assert that parental physical illness indirectly affects children's adjustment

by operating through family variables. Parental adjustment has been associated with

outcomes for these children. Numerous studies in the general clinical literature suggest

that children of parents with an affective disturbance are at greater risk for psychological

problems than children whose parents do not exhibit such disturbances (Beardslee,

Versage, & Gladstone, 1998). The same may be true for children of physically ill parents

since both patient and healthy parent depressive symptoms have been found to be directly

(Steele, Forehand, & Armistead, 1997) and indirectly (Lewis et al., 1993) predictive of

child psychosocial functioning. This has been demonstrated longitudinally as well.

Structural equation modeling data by Rotherham-Borus, Stein, and Lin (2001) suggested

that the emotional distress of parents with HIV is predictive of adolescent emotional

distress two years later, again indicating a relationship between parent and adolescent


Other measures of the ill parent's adjustment to the illness have demonstrated

relationships to child psychological symptomatology as well. Fathers' ratings of their

own greater physical and global psychological symptoms significantly predicted

combined parent ratings of child internalizing problems as well as children's self-reports

of depression (Kotchick, Summers, et al., 1997). Howes, Hoke, Winterbottom, &

Delafield (1994) found similar results between the self-reports of psychological

adjustment to illness of women with breast cancer and their reports of their children's

total psychological symptoms, although the validity of these ratings is of concern given

that the ill parent completed both measures. Nonetheless, Howes et al. (1994) suggested

that the parents' perceptions of the illness and their reactions to it might influence the

way the child interprets the illness. There is some empirical evidence to support this

hypothesis in addition to results related to parental depression. For example, Rodrigue

and Houck (2001) found that mothers' perceptions of poorer health-related quality of life

were associated with more adolescent behavior problems, while the objective presence of

a chronic condition was not. However, this relationship was stronger and more consistent

in parent reports of behavior problems than adolescent reports of their own behavioral


Similarly, in a study of children of parents with chronic pain, patient

psychological symptoms were significantly correlated with parent reports of child

anxiety, somatic complaints, and general adjustment (Mikail & von Baeyer, 1990). Chun

et al. (1993) also found that greater parent-reported functional disability in chronic pain

accounted for nearly twenty percent of the variance in children's behavior problems.

Lewis et al. (1993) reported that perceived illness demands were directly related to

parental psychological functioning, which was indirectly related to child psychosocial

functioning, defined by a measure of peer relationships. These findings should be

qualified, however, by noting that in all of these studies parents completed ratings of both

their own symptoms and those of their children, thus possibly confounding these


The healthy parent's adjustment to a partner's illness can be hypothesized to

affect children's adjustment. For example, in a study of ill fathers, maternal uncertainty

about a spouse's illness was associated with child-reported anxiety (Steele, Tripp, et al.,

1997). There is also evidence that spouses perceive illness as more stressful than patients

themselves do (Compas et al., 1994). However, the impact of parent-perceived

stressfulness on children's adjustment has not been described. Nonetheless, these

findings, in combination with findings related to children's perceptions of the parental

illness, suggest that perceptions of illness are important to the adjustment of both parents

and children.

As discussed previously, the sex of the ill parent may interact with the sex of the

child in predicting child adjustment. However, main effects for parent sex have also been

noted. Rotherham-Borus, Stein, and Lin (2001) reported that adolescents of parents with

HIV described more emotional symptoms and sexual risk behaviors at two year follow-

up when their mothers were ill than when their fathers were ill. However, more often ill

fathers have been associated with more negative outcomes. Chun et al. (1993) found that

children of ill fathers had less social competence on combined parent ratings than

children of ill mothers, although no sex differences were detected on parent- or teacher-

reported behavior problems. Rodrigue and Houck (2001) also found that adolescents of

fathers with a chronic health condition had significantly higher behavior problem scores

on both maternal and self-report questionnaires than did adolescents of fathers without a

chronic health condition. Adolescents of mothers with a chronic health condition did not

differ from adolescents of mothers without a chronic health condition on any of the

measures. However, it is important to note that none of the mean scores for these groups

were in the clinical range.

One explanation for these sex-based findings is that many families rely on fathers

as primary sources of financial support and decision making (Chun et al., 1993; Rodrigue

& Houck, 2001). Role changes, as well as the impact of the illness on the mother, may

lead to greater family instability when the father is ill (Compas et al., 1994). Indeed,

there is some evidence to suggest that women respond to illness in their spouses more

negatively than men do. Perceived severity of a spouse's cancer has been shown to be

greater for wives whose husbands have cancer than for husbands whose wives are ill

(Compas et al., 1994). In addition, Romano, Turner, and Clancy (1989) found that the

relationship between spouse adjustment and patient disability was stronger when the

patient was male.

Taken together, these findings indicate that the sex of the ill parent may be critical

to understanding the impact parental illness has on children's adjustment. The

experience of having an ill mother may particularly affect adolescent daughters because

of potential redistribution of household responsibilities. Families with ill fathers may

experience greater changes in family roles, thus putting all members at greater risk for

problems in adjustment.

Illness factors

The impact of objective illness factors on the adjustment of children and

adolescents has also been examined. Most commonly, researchers have examined the

relationship between objective illness severity and child adjustment, yielding mixed

results. Compas and colleagues (1994) found that the stage of a parent's cancer was

unrelated to children's symptoms of internalizing problems and stress. They suggest that

this result is consistent with cognitive models of stress in which appraisals are more

important than actual disease characteristics. Other studies of objective illness

characteristics have supported this conclusion. For instance, in a study of fathers with

hemophilia, some of whom had contracted HIV, neither hemophilia severity nor HIV

stage was significantly associated with child outcomes (Armistead, Klein, Forehand, &

Wierson, 1997). Moreover, in a sample of inner city HIV-positive mothers, children's

depression scores did not differ whether the women were symptomatic or asymptomatic

(Biggar et al., 1998).

In contrast, other studies have linked parent reports of disease severity with more

child adjustment difficulties. In a sample of low income, HIV-positive mothers, Forsyth

and colleagues (1996) found that children whose mothers demonstrated symptoms of

their disease reported significantly more symptoms of anxiety than did children whose

mothers were asymptomatic. In another sample of adolescents with a parent who had

AIDS, Rotherham-Borus and Stein (1999) found more self-reported internalizing

symptoms as parent-reported ill health increased. However, this could also be interpreted

as resulting from adolescents' perceptions of their parents' health rather than direct

effects of worsening illness.

Other characteristics of illness have been investigated less frequently. Length of

diagnosis does not appear to be predictive of psychosocial functioning for children of

parents with cancer (Compas et al., 1994; Lewis et al., 1993). Compas and colleagues

(1994) noted that worse prognosis was associated with more adolescent stress-response

symptoms, but was not related to symptoms of internalizing. Overall, the role of

objective illness factors in children's psychological adjustment remains equivocal.

However, it appears that objective illness factors may influence adults' and children's

perceptions of the illness, which in turn may affect psychological adaptation.

Family factors

Several authors have emphasized the importance of family processes in children's

adjustment to a family member's illness (e.g., Forehand et al., 1997; Kazak, 1989), thus

implicating roles for dyadic relationships and family coping styles. At the same time, it

has been suggested that family relationships may be placed at risk by physical illness

(Armistead et al., 1995). While direct relationships between marital adjustment and child

adjustment have not been found in this population (Chun et al., 1993; Lewis et al., 1993;

Steele, Forehand, & Armistead, 1997), other family factors have been implicated in the

functioning of children with chronically ill parents. Mothers with breast cancer who

rated their families as more adaptable have been found to describe fewer behavioral and

emotional difficulties in their children (Howes et al., 1994). Data have also indicated that

ill parents from cohesive families report significantly less perceived physical and

psychosocial impact of chronic illness than families characterized by conflict

(Stuifbergen, 1990). At the same time, adolescents of ill parents have been shown to

perceive their families as having more conflict, lower cohesion, and poorer organization

than adolescents of healthy mothers (Peters & Esses, 1985). Dura and Beck (1988) also

found that members of families in which the mother had chronic pain or diabetes reported

significantly less family cohesiveness than families of healthy parents. In addition,

families of healthy parents also reported less conflict than families in which the mother

had chronic pain, but no differences were noted in the amount of conflict in families with

a mother who had diabetes. Finally, in a study of families in which a parent had

rheumatoid arthritis, greater cohesion and expressiveness and less conflict were

associated with decreased psychological symptoms (Hirsch, Moos, & Reischl, 1985).

These factors were not related to functioning for children of healthy parents or those with

a depressed parent. Given the importance of illness perceptions discussed earlier, it could

be hypothesized that, for families coping with parental illness, cohesive families are at

less risk for difficulties in adjustment than families in which conflict is prevalent because

relationships in these families may be marked by more open communication styles.

Specifically, cohesive parent-child relationships may be one compensatory factor

that lessens the risk for children and adolescents. As discussed earlier, research from the

general stress literature has repeatedly supported the hypothesis that parent-child

relationships are important to the adjustment of adolescents during stressful life events.

Studies have substantiated this finding for adolescents and children facing the specific

stressor of a chronically ill parent.

It appears that it is the quality, not the quantity, of adult relationships that is

critical to adjustment. The mere presence of adults in the homes of mothers in the

advanced stages of HIV was not sufficient to predict better psychosocial functioning in

children (Dorsey et al., 1999). This suggests that perceptions of relationship quality may

be more relevant to children's functioning. For example, Kotchick, Summers, et al.

(1997) found significant effects of perceived parental support for children who had an ill

parent. Specifically, higher perceived parental social support predicted less parent-

reported externalizing problems, less child-reported depression, and higher grade point

averages. Extrafamilial support did not have a main effect on any of the child adjustment

domains measured in the study. Also, children's reports of anxiety symptoms were not

significantly related to social support. Interestingly, parental social support was also

significantly related to parent reports of internalizing behaviors and child reports of

depression when fathers reported that their illness had had a large impact on their lives,

physically or psychologically. This is especially relevant to the families in the present

study, since the pursuit of transplantation often coincides with illness that has had a

significant psychological and physical impact on the parent.

Parent-child relationships have been identified as an important mechanism for

promoting child adjustment in other studies of families coping with an ill parent (e.g.,

Lewis et al., 1993; Steele, Forehand, & Armistead, 1997). Greater conflict in parent-

child relationships has been associated with more depressive symptoms, externalizing

problems, and academic difficulties in children of fathers with HIV infection (Armistead

et al., 1997). Kotchick, Forehand, et al. (1997) also found that more conflict in the

mother-child relationship was associated with more child-reported internalizing and

externalizing behaviors, as well as parent-reported social competence. Mothers with HIV

reported significantly poorer relationship quality than did uninfected mothers, and both

mother- and child-reported psychosocial symptomatology was higher among children

whose mothers were HIV-positive.

However, despite the important role parental support plays in the adjustment of

children with a chronically ill parent, there are reasons to believe that they may be at risk

for not having the support they need. In a retrospective study of grown children of

parents who had had cancer, 20% of the women in the study reported negative feelings

toward their ill parent and 15% described having problems with their ill parent's behavior

during the illness (Leedham and Meyerowitz, 1999). Parent-child relationships may be

affected when a parent is ill because the usual ways of interacting with parents can be

altered by the effects of the illness, which may include emotional distress or impaired

parenting (Lewis et al., 1993). For example, Kotchick, Forehand, et al. (1997) found that

HIV-infected mothers reported less monitoring of their children than did uninfected

women. Physical restrictions imposed by the illness may also threaten the quality of the

ill parent-child relationship by limiting the amount and quality of interactions spent with

the child (Kotchick, Summers, et al., 1997). Child perceptions of not wanting to bother

an ill parent or of limited physical availability due to illness may also be detrimental to

the parent-child relationship.

Sources of social support other than the ill parent would seem to take on increased

importance to children when a parent is ill. However, the Kotchick, Summers, et al.

(1997) study suggested only a limited role for extrafamilial social support. Therefore, a

relationship with a healthy parent, when available, may prove to be critical for

adolescents in families with such stressors. However, the healthy parent-child

relationship may be compromised by demands of caring for an ill spouse and the effects

on that parent's psychological functioning and energy level (Kotchick, Summers, et al.,

1997). Leedham and Meyerowitz (1999) found that 22% of their sample recalled

having had negative feelings toward their healthy parent, similar to the percentage that

recalled negative affect toward their ill parent. However, 33% of the women noted

having had problems with their healthy parent's behavior (e.g. being too demanding)

during the illness; this represents twice the number who recalled difficulties with their

ill parent. This information suggests that while it seems probable that with regards to

communication, relative to an ill parent, children may perceive a healthy parent as a

more available source of support and may rely on this relationship to a greater degree,

problems in this relationship are common and can represent threats to adolescent well

being in this time of stress.

Unfortunately, much of the research in this area has focused on children of

parents with HIV who often come from single-parent households (e.g., Forehand et al,

1998; Forsyth et al., 1996; Rotherham-Borus & Stein, 1999). Therefore, information on

the effects of a healthy parent has been unavailable in many studies, thus limiting our

knowledge. However, published research in other areas supports the idea that these

relationships may be critical.

Studies have indicated that a single positive parental relationship may be

sufficient to buffer adolescents from the negative effects of stress. In studying

adolescents of divorced families, Forehand et al. (1987) found that adolescents with

conflict-ridden relationships with both parents were rated by teachers as having lower

cognitive competence, more externalizing behavior problems, and more internalizing

behavior problems than adolescents with a good relationship with either one or both

parents. The notion that one positive parental relationship may be all that is necessary to

foster good adolescent adjustment is an important consideration in families with an ill

parent, given the factors that may interfere with the usual ways of interacting with an ill

parent. A positive relationship with a healthy parent may be very beneficial to children's

adjustment; however, this is not a foregone conclusion for many families. Forehand and

colleagues (1997) found that children of fathers with hemophilia experienced poorer

relationships with their mothers than with their fathers. Compas et al. (1994) found that

wives perceived their spouses' illnesses as more severe than did husbands. Greater

distress in wives of ill husbands may impair this healthy parent-child relationship and

therefore place children of ill fathers at greater risk.

The relative importance of these relationships has been asserted in isolated

studies. In a study of families of mothers with breast cancer, parents reported their

children as having higher psychosocial functioning when the healthy parent (father)

reported a better parent-child relationship, conceptualized as the frequency of interactions

or "togetherness" experienced in the dyad (Lewis et al., 1993). The frequency of

interactions between the ill parent and the child was not predictive of child functioning.

The researchers suggest that a healthy parent may be an important route for information

for the child about the parent's illness. This hypothesis is plausible and has been

supported by findings in which parental feelings of uncertainty about a partner's illness

were shown to be predictive of children's reports of anxiety beyond the predictive power

of the children's own uncertainty (Steele, Tripp, et al., 1997).

Steele, Forehand, and Armistead (1997) further examined this relationship in a

study of fathers with hemophilia, some of whom also were HIV-positive. High conflict

in both the mother (healthy parent)- and father-child relationships independently

predicted child internalizing problems. However, the mother-child relationship

accounted for nearly twice as much of the variance (19%) in child adjustment as did the

father-child relationship (10%). These authors suggest that mothers in families

subjected to severe stressors take on a role of "gatekeeper" whereby they protect the

children from distress. Therefore, when this relationship is jeopardized by conflict,

children may be at greater risk for psychological problems. However, considering this

in combination with the Lewis et al. (1993) finding, it seems unlikely that it is the

mother per se who serves as gatekeeper for families coping with parental illness.

Rather, it is the individual in the role of the healthy parent, who may be perceived as

more available or psychologically stable. However, our current understanding of the

importance of the healthy parent is limited by the dearth of research on the topic and

the use of measures examining the frequency of contact (Lewis et al., 1993) or the

amount of conflict in the relationship (Steele, Forehand, & Armistead, 1997), which

may not be the most relevant constructs for these families. Instead, it is proposed that

parent-adolescent communication may be most meaningful.

Communication has often been considered an important, but understudied,

element in the functioning of families coping with chronic illness. Studies of adolescents

have indicated that better parent-adolescent communication is associated with better

perceived family dynamics, such as family satisfaction and family cohesion (Barnes &

Olson, 1985). These qualities may be buffers against the stress of having an ill parent.

Indeed, Compas and colleagues (1996) called for more research on factors affecting

children's perceptions of cancer, specifically identifying the amount and type of

information received and abilities to process these facts as potentially relevant variables.

Other researchers have suggested that children's uncertainty about parental illness and

the negative consequences associated with this ambiguity are influenced by parents' own

uncertainty about the illness, communicated directly or indirectly to the child (Steele,

Tripp, et al., 1997).

Therefore, it may be that adolescent adjustment is affected by his or her cognitive

understanding of the illness and its meaning to the family. This, in turn, implicates

parent-adolescent communication as critical for providing information and support to the

adolescent. These outcomes would also be expected to improve the parent-child

relationship, earlier noted to be related to measures of adolescent adjustment. Indeed,

better parent communication skills have been associated with less parent-child conflict

(Armistead et al., 1997; Wierson & Forehand, 1992), which is often used to define

parent-child relationships. Others have used this line of thinking to suggest that parental

illness can interfere with usual interactions with the child, thus leading to behavior

problems (Dorsey et al., 1999).

Indeed, in a qualitative, retrospective study of adult women who had had a parent

with cancer, when asked what had hindered their coping with the stress of the illness,

22% identified a lack of honesty or disclosure (Leedham & Meyerowitz, 1999). When

asked for advice they would give children with an ill parent, 69% of the women

suggested that children keep lines of communication open. Further emphasizing the

importance of communication in this population, 87% of participants in the open-ended

interview format advised parents to be honest with their children, and 71% recommended

talking to their children in this situation. In summary, the responses of the grown

children in this study suggest that they perceived open family communication as a critical

factor in their adjustment to parental illness.

It has been recommended that chronically ill parents provide their children with as

much information as developmentally appropriate because "information and open

communication may work to reduce the child's uncertainty" (Steele, Tripp, et al., 1997,

p. 587). Other authors have recommended that increasing parental emotional availability

and assisting the child's self-expression are important clinical recommendations for

helping children adjust to having a physically ill parent (Armsden & Lewis, 1993). Open

parent-adolescent communication may facilitate the implementation of these

recommendations. In addition, good communication would appear to signify a positive

parent-child relationship characterized by openness and a lack of major conflicts.

Finally, having satisfying parent-adolescent relationships may simply represent one less

stressor that adolescents in these families must contend with in maintaining their own

mental health, thus contributing to better adjustment. However, little research regarding

the impact of parent-adolescent communication exists for this population despite calls for

research on "the underlying processes leading to health and disorder" (Hirsch, Moos, &

Reischl, 1985, pp. 163)


Purpose of the Study

The primary purposes of the study were to better describe adolescent perceptions

of their relationships with healthy and ill parents and to better delineate the relative

influence of these parent-adolescent relationships on the adjustment of adolescents in

families with an ill parent. Previous studies have suggested that adolescents may be at

elevated risk for adjustment difficulties, but they have not been the focus of considerable

study. The present study examined the following specific hypotheses.


Hypotheses Related to Levels of Adolescent Adjustment

1. It was hypothesized that because parental illness can draw attentional, emotional, and

material resources away from an adolescent, mean scores on measures of

psychological symptoms of depression and anxiety would be elevated relative to

normative data.

2. It was hypothesized that, in agreement with previous studies (e.g., Compas et al.,

1994; Rodrigue & Houck, 2001), adolescents with ill same sex parents would have

more adjustment difficulties than those whose opposite sex parents were ill.

Hypotheses Related to Parent-Adolescent Communication

3. It was hypothesized that because illness may represent a threat to security in the

parent-adolescent relationship through perceived unavailability or fear of parental

death and because of adolescent concerns for the effects of stress on an ill parent,

adolescents would perceive poorer communication quality with their ill parents

compared to their healthy parents.

4. It was hypothesized that because healthy parents may have more contact with

adolescents, be perceived by their children as more emotionally available than ill

parents, and therefore be relied upon to a greater degree for emotional support,

communication with the healthy parent would be a stronger predictor of adolescent

adjustment than communication with the ill parent. It was hypothesized that this

pattern would remain after testing for effects of adolescent age, sex, perceived illness

severity, and socially desirable responding.

5. It was hypothesized that, consistent with the finding of Forehand et al. (1987),

adolescents with at least one positive parental relationship would have fewer reported

symptoms of depression and anxiety than those with average or negative parent


Hypotheses Related to Ill vs. Healthy Parent Reports of Adolescent Adjustment

6. Because it was hypothesized that adolescents communicate better with healthy

parents than with ill parents, thus allowing healthy parents greater awareness of

adolescent functioning, it was also hypothesized that healthy parent reports of

adolescent symptoms would better reflect adolescents' self-descriptions of adjustment

than ill parent reports. It was expected that healthy parents would endorse more

problem behaviors and that healthy parent reports of symptomatology would be more

strongly associated with adolescent reports of their own symptoms.



Thirty-eight two-parent families with adolescents participated in the study. To be

included in the study, families were comprised of at least one parent with a chronic, life-

threatening illness, one parent who did not have such a condition, and an adolescent

between the ages of 12 and 17. In families with more than one adolescent, only one was

randomly chosen to be invited to participate. Because the major hypotheses of the study

were related to communication between parent and child through a consistent

relationship, only families in which all three members had shared the same residence for

the previous six months were included. All adolescents were aware of their parent's

illness and treatment.

Ill Parents

Chronically ill parents were identified through transplant, radiation oncology, and

dialysis clinics and had a variety of diagnoses, including liver disease (32%), cancer

(29%), kidney disease (24%), heart disease (11%), and lung disease (5%). Fathers

comprised 66% of the sample. This statistic is partly a result of the fact that more males

than females are evaluated for transplantation at the university medical center where the

study was conducted. Men comprise 60% of liver candidates, 85% of heart candidates,

60% of kidney candidates, 49% of lung candidates, and 50% of bone marrow candidates.

While variation in nature and severity of diseases in the present sample occurred,

all patients had active, significant illnesses requiring ongoing medical treatment.

Because the objective of the study was not to examine the stress of a treatment (e.g.,

chemotherapy, transplant) itself, but rather the experience of having an ill parent, all

patients with the aforementioned illnesses seen in participating clinics at the hospital

were eligible for inclusion in the study.

The age of the ill parents in the study ranged from 32 to 63 years, with a mean age

of 44.74 years and a standard deviation of 6.23 years. Educationally, 13% of the patient

sample had completed less than the twelfth grade, while 45% had graduated high school.

Thirty-four percent had attended at least one year of college, and an additional 8% had

pursued post-graduate education. The patient was a biological parent to the participating

adolescent in 82% of the families. The parent with a chronic illness was a biological

father in 18 families, a biological mother in 13 families, a step-father in 6 families, and a

live-in boyfriend in one family. One ill father experienced a health crisis shortly after the

family began participation in the study and was unable to complete the CBCL regarding

his daughter.

Parents had been hospitalized between 0 and 15 times prior to participation in the

study. The average number of hospitalizations was 3.84 with a standard deviation of

3.28. Time since diagnosis ranged from 3 months to 32 years, 7 months. Thirty-four

percent of the sample had been diagnosed within the past year. An additional 26% had

been diagnosed between one and three years prior to participation in the study, 13% had

been diagnosed for between three and five years, and 26% had been diagnosed for more

than five years.

Healthy Parents

The age of the healthy parents in the present sample ranged from 28 to 59 years.

The average age was 43.67 years, with a standard deviation of 7.40 years. Five percent

of the healthy partners had not completed high school, while 49% had graduated the

twelfth grade. Forty-one percent had completed some college, and 5% had attended

graduate school. Healthy partners were biological parents to 84% of the adolescents in

the study. Of the 38 participating families, 24 healthy parents were biological mothers, 8

were biological fathers, 2 were step-mothers, and 4 were step-fathers. Three healthy

parents were unable to be contacted for study participation.


Females represented 66% of the adolescents participating in the study. The

average age was 14.92 years with a standard deviation of 1.65 years. Of the adolescents

participating in the study, 3% were in the sixth grade, 13% in the seventh grade, 32% in

the eighth grade, 16% in the ninth grade, 21% in the tenth grade, 5% in the eleventh

grade, and 11% in the twelfth grade. The sample was primarily Caucasian (84%);

thirteen percent of the sample was Black, and 3% was of Hispanic descent.

Family demographics

Family composition included only the two parents and adolescent in 32% of the

families studied, while 50% had four individuals living in the home. Eighteen percent of

the families had five or more members residing together. Family income was less than

$10,000 for 5% of participating families, between $10,000 and $19,999 for 8%, between

$20,000 and $29,999 for 19%, between $30,000 and $39,999 for 16%, between $40,000

and $49,999 for 14%, and greater than $50,000 for 38% of the families.

For 63% of the families, parents reported that their adolescent was immediately

told of the parent's illness at the time of diagnosis. On average, adolescents in the current

study had been aware of their parent's illness for 34 months prior to the study interview.

However, there was a great deal of variability in the amount of time adolescents had been

coping with this stressor, with a range of 2 to 165 months. Forty percent of the sample

had been aware for 12 months or less, 29% had been aware for 13 to 36 months, 13% had

had knowledge of it for 37 to 60 months, and 18% had been coping with this stressor for

more than five years.

Of the 38 families who took part in the study, 18 included an ill father with an

adolescent daughter, 7 had an ill mother with an adolescent daughter, 7 had an ill father

with an adolescent son, and 6 had an ill mother with an adolescent son.


All study procedures were approved by the university's institutional review board.

Families were recruited for the study during routine patient clinic appointments. Patients

(and spouses, when available) were approached at the time of their appointment when

informed consent procedures, including permission to contact their adolescent child, were

conducted. Ill parents completed a family demographic questionnaire and the Child

Behavior Checklist (CBCL; Achenbach, 1991a). When present, healthy parents also

completed a CBCL at this appointment. When parents were not available to complete

questionnaires in person, these instruments were conducted via phone interviews.

In all but three cases, adolescents were not present during their parent's

appointment and were subsequently contacted by phone. The study was explained to all

adolescents and verbal assent was obtained. Adolescents completed the Reynolds

Adolescent Depression Scale (RADS; Reynolds, 1987), Revised Children's Manifest

Anxiety Scale (RCMAS; Reynolds & Richmond, 1985), Impact of Event Scale (IES;

Horowitz, Wilner, & Alvarez, 1979), Parent-Adolescent Communication Scale (PACS;

Olson, 1985), and a single item regarding the perceived severity of their parent's illness

(Compas et al., 1994). While the majority of responses were obtained by telephone,

copies of these measures were sent home with participating parents for the adolescents to

reference during the telephone administration of the questionnaires. Adolescents who

completed the interview received a $10 gift certificate.


The outcome measures described below were selected for the present study for

several reasons. First, they were designed to assess both general psychological

adjustment in adolescents (RADS and RCMAS) as well as stressor-specific distress

(IES). Second, adequate reliability and validity have been demonstrated for all of the

measures and are described below. Finally, the potential range of scores for each of the

instruments is large, and, with the exception of the RADS, each of the outcome measures

had been used in previous studies examining adjustment to parental illness and had

demonstrated significant effects.

Demographics (Appendix)

A demographic questionnaire was completed by the patient and obtained

background information regarding the patient, his or her partner, and the adolescent.

Medical history information, such as diagnosis, illness duration, and how long the

adolescent was aware of the illness, was also obtained via this questionnaire. To assess

other concurrent stressors, parents also reported whether their adolescent had experienced

any of nine events in the past year. These included the death of a friend, death of a

relative, academic problems, change in residence, family financial problems, school

change, legal problems, parental divorce, or parental remarriage.

Perceptions of Illness Questionnaire

Consistent with Compas and colleagues (1994), adolescent perceptions of illness

severity were assessed using the question, "How bad do you think your parent's illness

is?" Adolescents responded on a 5-point Likert scale ranging from 1 = not at all bad to 5

= extremely bad.

Reynolds Adolescent Depression Scale (RADS)

The RADS (Reynolds, 1987) is a 30-item self-report measure of depressive

symptoms for adolescents. Adolescent total scores were compared using the

standardization sample percentile rankings based on sex and grade (Reynolds, 1987).

These were converted to T scores for analyses.

The measure has demonstrated good reliability and validity. For the

standardization sample, the internal consistency reliability coefficient was .92. Test-

retest reliability at six weeks was .80, at three months was .79, and at one year was .63.

The standard error of measurement is approximately four raw score points (Reynolds,

1987). Convergent validity has been demonstrated with other self-report depression

instruments, such as the Beck Depression Inventory, Center for Epidemiological Studies-

Depression scale, and the Children's Depression Inventory. Each of these measures

yielded correlation coefficients with the RADS of.68 or higher (Reynolds, 1987).

Similarly, concurrent validity with a depression interview, the Hamilton Rating Scale, has

been established with a correlation coefficient of.83 (Reynolds, 1987).

The RADS contains an item regarding self-injurious behavior, "I feel like hurting

myself," which was not asked during adolescent interviews. A conservative approach

was taken with this deletion and the lowest possible score ("almost never") was used to

calculate total depression scores.

Children's Manifest Anxiety Scale- Revised (RCMAS)

The RCMAS (Reynolds & Richmond, 1985) is a 37-item self-report questionnaire

for children 6 to 19 years of age. It assesses the presence or absence of anxiety-related

symptoms, yielding a total anxiety scale score, three subscales scores (Physiological,

Worry and Oversensitivity, and Concentration), and a Lie scale. The Lie scale contains

seven items and provides an index of socially desirable responding. As with the RADS,

all scale scores were standardized using age and sex norms, and T scores were used in all

analyses, except where noted.

Adequate internal consistency reliability for the total anxiety scale score is

established, with alpha greater than .80 (Reynolds & Paget, 1983). In addition, it is

highly correlated (r = .85) with the trait scale of the State-Trait Anxiety Inventory for

Children, suggesting good convergent validity (Reynolds, 1980).

Impact of Event Scale (IES)

The IES (Horowitz, Wilner, & Alvarez, 1979) is a 15-item questionnaire

assessing the subjective distress of a particular event. In addition to yielding a Total

Stress score, the scale is commonly divided into two subscales, Intrusion and Avoidance,

as reported by the authors. The validity of this factor structure in adults has been

supported by Zilberg, Weiss, and Horowitz (1982), however the presence of a third

factor, Emotional Numbing (Schwarzwald, Solomon, Weisenberg, & Mikulincer, 1987)

or Sleep Disturbance (Larsson, 2000), has been suggested in other replication studies

with adults. Despite this, the original two factor structure remains commonly used in the

trauma literature.

Horowitz, Wilner, and Alvarez (1979), using a sample of adults experiencing

bereavement or personal injury, reported test-retest reliability of .87 for the Total stress

score, .89 for the Intrusion subscale, and .79 for the Avoidance subscale. Specific to the

present study, Compas and colleagues (1994) used the IES in a study of children of

parents with cancer and found internal consistency reliability to be adequate for

adolescents (q = .67). Adequate validity data has been presented as well. Horowitz,

Wilner, and Alvarez (1979) found lower scores on the measure after patients had

undergone psychotherapy, suggesting that the measure is sensitive to decreases in the

stressfulness of an event resulting from time and therapy.

The IES, while originally used with adults, has been incorporated into studies of

children and adolescents. Psychometric properties appear similar to those found among

adults, although the third factor, Emotional Numbing, has been identified in two studies

analyzing the measure's properties with adolescents and young adults (Sack, Seeley,

Him, & Clarke, 1998; Yule, Bruggencate, & Joseph, 1994). Because of the ambiguous

findings regarding this third factor, the present study used the Total Stress score as the

main dependent variable for analyses involving the IES.

Because of the event-specific nature of the IES, normative data do not exist.

However, several studies have used the instrument with children and adolescents, thus

providing information for comparison to other samples (see Table 2). The majority of

these studies have administered the IES to victims of large-scale accidents (e.g., cruise

ship sinking, dormitory fire) or violence (e.g., war, parental homicide). In addition, one

study investigated lifetime prevalence of significant negative life events (e.g., family

member with a substance abuse problem, parental separation or divorce) and reported

mean IES scores for their sample of adolescents who had experienced each event (Joseph,

Mynard, & Mayall, 2000). This provides additional data by which to compare post-

traumatic stress symptoms among adolescents experiencing significant stressors.

Table 2

Comparison Scores on the Impact of Event Scale for Children and Adolescents
Event Author/ Year Age Time since Sex N Mean
Taken hostage at Vila et al. 7.5-9.5 6-8 wks. Both 21 33.6
school by armed (1999) yrs. 16-18 wks. 20 24.5
mental patient 28-30 wks. 16 17.3
82-84 wks. 18 20.1

Indirectly exposed to 7.5-9.5 16-18 wks. Both 21 25.5
hostage incident yrs. 21 14.8
18 18.0
Cruise ship sinking Yule & 14-16 10 days Female 24 35.5
Udwin (1991) yrs. 5 mos. 24 35.33
Bus accident Stallard & 14 yrs., 6 mos. Both 7 31.0
Law (1993) 9 mos.-
16 yrs.,
3 mos.
Dormitory fire Jones & 14-19 4 mos. Male 25 28.0
Ribbe(1991) yrs.
Ferry sinking Yule & 12-14 3-6 mos. Both 7 46.9
Williams yrs.
(1990) 12-15 mos. Male 6 34.1
__Female 4 43.8
Witnessed parent Malmquist 5-10 Unreported Both 16 56.9
murder/ attempted (1986) yrs.
Coach accident Curie & 6-18 2 yrs. Both 25 16.08
Williams yrs. Male 11 10.6
(1996) 1 Female 14 20.4

Table 2. Continued
Event Author/ Year Age Time since Sex n Mean
Life threat to family Joseph et al. 11-16 Unreported Male 90 22.15
member or friend (2000) yrs. Female 122 27.74
(accident, injury, or

Life threat to self Male 46 16.95
(accident, injury, or Female 42 25.06

Witnessed attack or Male 55 12.76
physical assault Female 48 19.48

Personal assault Male 36 16.63
Female 25 34.40

Fire or natural Male 25 12.87
disaster Female 12 25.82

Parental separation or Male 49 23.29
divorce Female 69 29.64

Family member with Male 21 23.41
substance abuse Female 31 31.64

Trouble with the law Male 44 15.16
Female 26 17.65

Note. aMean female score significantly higher than males in Joseph, Mynard, and Mayall
(2000) sample

Parent-Adolescent Communication Scale (PACS)

The PACS (Olson, 1985) measures perceptions of both positive and negative

communication between adolescent and parent. The PACS focuses on the exchange of

ideas between adolescent and parent, trust between them, and the emotional tone of

interactions (Jacob & Tennenbaum, 1988). The adolescent form used in this study

contains 20 items that adolescents separately rated in relation to his or her mother and

father. The measure was administered in its entirety in relation to one parent and then

completed for the other parent; the order in which it was administered (mother vs. father)

was counterbalanced.

Each item was rated on a 5-point Likert scale from "strongly disagree" to

"strongly agree." Items on the PACS are phrased in both negative and positive terms.

The PACS generates a total score as well as two subscale scores. The Open Family

Communication subscale reflects feelings of free expression and understanding in parent-

adolescent interactions. Sample items include "When I ask questions, I get honest

answers from my mother/father" and "1 find it easy to discuss problems with my

mother/father." The second subscale, Problems in Family Communication, measures

negative interaction patterns and hesitancy to disclose concerns. Examples of items on

this subscale include "My mother/father has a tendency to say things to me that would be

better left unsaid" and "I don't think I can tell my mother/father how I really feel about

some things." Good internal consistency of the entire measure has been established ( =

.88; Olson, 1985). Cronbach's alphas for Open Family Communication and Problems in

Family Communication were .87 and .78, respectively. Total score test-retest reliability

over 4 to 5 weeks was .60.

Olson (1985) has reported a factor analysis supporting the construct validity of the

measure scales (Jackson, Bijstra, Oostra, & Bosma, 1998). Concurrent validity of the

PACS has been supported by consistently significant and elevated correlations (.43 .73)

between adolescent reports on the Cohesion subscale of the Family Adaptability and

Cohesion Evaluation Scale (FACES-Ill) and both the PACS total (Brown & Mann, 1990)

and subscale scores (Morrision & Zetlin, 1988; Tulloch, Blizzard, & Pinkus, 1997).

Jackson et al. (1998) also found that the PACS correlated significantly with the Family

Satisfaction Scale.

This same study found that the PACS was significantly correlated with adolescent

reports of the outcomes of arguments with both their mothers and fathers. Poorer

communication (less open and more problems) was associated with greater frustration,

greater escalation, and less intimacy. Poorer communication was also associated with an

aggressive approach to disagreement and negatively correlated with an approach to

disagreement characterized by seeking compromise, suggesting that conflict is associated

with scores on the PACS. In a separate study by Baer (1999), higher scores on the Open

Communication scale of the PACS significantly predicted decreased family conflict in

three different ethnic groups of adolescents.

Child Behavior Checklist (CBCL)

The CBCL (Achenbach, 1991a) measures 4- to 16-year-old children's

competencies and behavior problems as reported by their parents. For this study, only the

behavior problem scales were collected. Parents rated 112 behaviors as "not true,"

"somewhat or sometimes true," or "very true or often true" in the past six months.

Responses on the CBCL yield scores on three broad scales: internalizing behaviors,

externalizing behaviors, and total behavior problems. Eight subscales have also been

established. Only the 14-item Anxious/Depressed subscale was used for the present


Adequate reliability and validity data have been reported (Achenbach, 1991a).

Seven-day test-retest reliability correlations for the Internalizing, Externalizing, and Total

Problems scales were .89, .93, and .93, respectively. Achenbach (1991a) also reported a

correlation of .76 for interrater reliability between parents for the Total Problems scale.

Parent correspondence of .66 and .80 were reported for the Internalizing and

Externalizing scales, respectively. Construct validity has been assessed by comparing

scores on the CBCL to scores on the Conners Parent Questionnaire, yielding a correlation

of .81 for total problems, .72 for internalizing symptoms, and .88 for externalizing

behaviors. The correlation between the Total Problems scores on each scale was .82.

Standardized T scores are based on normative data for age and gender; these

standardized scores were used in all analyses. Clinical cutoffs have been established for

the Total Problems, Internalizing, and Externalizing scales whereby a T score of 60 or

greater is considered deviant. The clinical cutoffs for these scales have been shown to

significantly discriminate between children referred for psychological services and those

not referred. For the Anxious/Withdrawn subscale, a T score of 67 is recommended as

the clinical cutoff.


Preliminary analyses

To assess the reliability of adolescent-completed measures in the present sample,

alpha coefficients were calculated. Coefficients ranged from .72 to .90 for the total

sample suggesting adequate internal consistency reliability for all instruments used. In

addition, alpha coefficients were calculated for males and females separately; these

values ranged from .60 to .91 (see Table 3). Alpha coefficients were not calculated for

the CBCL as it is an established measure and its reliability was assumed.

Table 3

Alpha coefficients of adolescent-completed measures for total sample, males, and
Total Males Females
Measure n ao n a n a
RADS, Total score 37 .89 13 .83 24 .91
RCMAS, Total score 38 .90 13 .87 25 .91
IES, Total score 38 .82 13 .68 25 .86
PACS, Total score- mother 38 .85 13 .81 25 .86
PACS, PFC score- mother 38 .78 13 .61 25 .80
PACS, OFC score- mother 38 .87 13 .87 25 .87
PACS, Total score- father 38 .88 13 .90 25 .88
PACS, PFC score- father 38 .80 13 .79 25 .81
PACS, OFC score- father 38 .87 13 .87 25 .88
Note. RADS = Reynolds Adolescent Depression Scale; RCMAS = Revised Children's
Manifest Anxiety Scale; IES = Impact of Event Scale; PACS = Parent- Adolescent
Communication Scale, PFC = Problems in Family Communication, OFC = Open Family

Data were examined for outliers by examining ranges. To determine the retention

and deletion of possible outliers, Hair et al. (1998) recommend that data should be

eliminated only if there is proof that they are completely aberrant and do not represent

any occurrence in the population. Using this guideline, no values in this data set were

believed to be nonrepresentative of the population being considered.

To assess normality, z values for skewness and kurtosis were calculated (see

Table 4). A critical value of +1.96, corresponding to a .05 error level, was chosen.

Values for all measures were below this critical value, with the following exceptions,

which suggested abnormal distributions. Both ill and healthy parent reports on the CBCL

Anxious/ Depressed subscale were significantly positively skewed. In addition, ill parent

reports on this subscale were more sharply peaked than a normal curve. This distribution

is likely a function of both the Achenbach (1991a) scoring system, which does not allow

for subscale scores below 50, and the fact that few parents reported significant levels of

symptoms on this subscale. Due to its poor statistical properties with this sample, further

analyses were not conducted using this subscale as normed by Achenbach.

The ratings of perceived illness severity were significantly negatively skewed.

The range of values available on the Likert scale was five, and 90% of the adolescents in

the sample rated their parent's illness as "3" or higher on the scale. Because the variable

was not a continuous one, no transformation was conducted.

Despite assurances given during adolescent interviews, the possibility was

considered that adolescents might consider their responses relevant to their parents'

candidacy for transplant, and thus attempt to present in a favorable light. As noted

earlier, the RCMAS Lie scale is an index of socially desirable responding. Four of the

respondents were above the 901' percentile on this measure. Correlations were computed

between the RCMAS Lie scale raw scores and adolescent self-report independent (ill and

healthy parent PACS T scores) and dependent (RCMAS and RADS T scores, IES raw

scores) variables to determine whether adolescent responding appeared to have been

influenced by social desirability. Using a one-tailed test, findings indicated a significant

positive correlation between ill parent (r = .29), but not healthy parent (r = .13),

communication and socially desirable responding on the RCMAS. No significant

relationships emerged on correlations between dependent measures and the RCMAS Lie

scale. Despite its correlation with ill parent communication, because the social

desirability measure was not also associated with any of the dependent variables, it was

not included in the final regression models.

Table 4

Shape descriptors of variable distributions
Skewness Kurtosis
Measure Statistic z value Statistic z value
RCMAS, Total score .343 .864 -.911 -1.146
RCMAS, Lie score -.339 .854 -.181 -.228
RADS, Total score .151 .380 -.397 -.499
IES, Total score .745 1.877 .346 .435
CBCL, Ill parent- Total Problems .120 .302 -.380 -.478
CBCL, Ill parent, Internalizing .100 .252 -.224 -.282
CBCL, 11 parent, Externalizing .425 1.071 -.285 -.358
CBCL, III parent, Anxious/ Depressed 1.915 4.824 3.389 4.263
CBCL, Healthy parent, Total Problems -.421 -1.060 .012 .015
CBCL, Healthy parent, Internalizing -.310 -.781 -.537 -.675
CBCL, Healthy parent, Externalizing .144 .363 -.621 -.781
CBCL, Healthy parent, Anxious/ Depressed 1.071 2.700 -.198 -.249
Perceived illness severity -.751 -1.892 .378 .475
PACS, Ill parent, Total score .094 .237 -.903 -1.136
PACS, Healthy parent, Total score -.525 -1.322 -.187 .235
Adolescent age .223 .562 -1.101 -1.385
Note. Z values derived by dividing the statistics by the appropriate standard errors of.397
skewnesss) and .795 kurtosiss).

Primary analyses

Hypothesis 1: Adolescent Adjustment to Parental Physical Illness

It was hypothesized that mean scores on measures of depression and anxiety

would be elevated relative to normative samples. This hypothesis was partially

supported. Means and standard deviations for adolescents' self-reports of depression

(RADS) and anxiety (RCMAS and IES) are presented in Table 5. Means and standard

deviations for parent reports of internalizing, externalizing, total problems, and symptoms

of depression/anxiety on the CBCL are presented in Table 6.

Table 5

Means and standard deviations of self-report measures of adolescent adjustment

Measure n M SD
RADS, Total score 38 48.10 8.72
RCMAS, Total score 38 49.21 10.84
RCMAS, Lie Scaleb 38 10.68 2.43
IES, Total score 38 27.87 14.30
Note. RADS = Reynolds Adolescent Depression Scale; RCMAS = Revised Children's
Manifest Anxiety Scale; IES = Impact of Event Scale.
aT score conversion. bScaled score conversion.

Table 6

Means and standard deviations of parent-report measures of adolescent adjustment
Healthy (n = 35) Ill (n = 37)
Measure M SD M SD
CBCL, Total Problems 49.11 10.73 49.46 8.70
CBCL, Internalizing Problems 48.63 9.96 48.38 10.08
CBCL, Externalizing Problems 49.57 9.74 49.65 8.03
Note. Values represent T score conversions. CBCL = Child Behavior Checklist.

Results related to symptoms of adolescent depression did not support the

hypothesis. The mean score on the RADS was in the average range and represented a

subclinical level of responding. A Welch's v test was conducted to compare the raw

score measure mean of the present sample with that of the normative sample. There was

no significant difference between the current sample and the standardization sample,

v(10) = 1.41, p > .05. On the RADS a raw score of 77 or above has been established as a

clinical cutoff identifying a level of symptomatology representative of clinical depression

(Reynolds, 1987). Five percent of adolescents in the present study responded at this level

of clinical significance. This is lower than anticipated, as expected percentages in the

clinical range of scores have been reported as 8% for boys and 14% for girls (Reynolds,


Adolescent responses on the RCMAS also did not support the hypothesis. The

average score was in the subclinical range for the Total score. Welch's v test indicated

that there was no significant difference between the current sample of adolescents and the

normative group for the RCMAS Total score, v(10) = .055, g > .05. Consistent with the

recommendation of Reynolds and Richmond (1997), scores greater than one standard

deviation of the mean were interpreted as outside the normal range on the RCMAS.

Eighteen percent of the present sample was beyond this cutoff. While the measure

manual does not indicate expected percentages in the clinical range, using the cutoff of

one standard deviation suggests that 16% of the sample would be expected to score at this


Results for the IES partially supported the hypothesis. Adolescent responses on

the IES yielded a mean total score below the clinical cutoff of 35 recommended by Neal,

Busuttil, Rollins, Herepath, Strike, and Turnbull (1994). As discussed earlier, due to the

context-specific nature of the measure, normative data for the IES for the population

being considered does not exist. In addition, the study by Compas et al. (1994) did not

use the complete IES with children and adolescents, therefore no comparison sample is

available for the IES. However, using the standard of Neal et al. (1994), 31.6% of the

adolescents in the present sample reported significant levels of stress-related symptoms.

Parent-reports of psychological symptoms did not support the hypotheses. As

noted earlier, Achenbach (1991a) has identified a T-score of 60 or above as clinically

significant on the CBCL Total Problems, Internalizing, and Externalizing scales. For

subscales of the CBCL, such as the Anxious/Depressed subscale, a T-score of 67 or

above is the recommended clinical cutoff (Achenbach, 1991a). The means on all scales

(Total Problems, Internalizing, Externalizing, and Anxious/Depressed) were within

normal limits for both healthy and ill parents. Fourteen percent of healthy parents (n =

35) and 11% of ill parents (n=37) described their adolescents in the clinical range of

scores on the Total Problems scale. On the Internalizing scales, 9% of healthy parents

and 14% of ill parents reported clinically elevated levels of symptoms, while 17% and

14% of healthy and ill parents, respectively, indicated clinically significant elevations on

the Externalizing scale. None of the healthy parents and 5% of the ill parents reported

clinically meaningful elevations on the Anxious/ Depressed subscale of the CBCL. The

clinical range of the CBCL represents the highest 2% of scores from the normative


Secondary analyses were conducted to assess the influence of demographic

influences on the adjustment scores observed and to determine variables associated with

dependent measures for inclusion as covariates in subsequent regression analyses. For

this reason, a family wise error rate was not used and a more liberal error rate (.05) was

determined as a cutoff

Adolescent sex differences on measures of adjustment were examined using

independent samples t-tests. No significant sex differences emerged on adolescent self-

report measures (RADS, RCMAS, or IES). On the parent-report CBCL, ill parents rated

males as having significantly higher levels of total problems, t(35) = 2.45, p< .05, and

internalizing problems, t(35) = 2.71, p < .05, than females. No other significant sex

differences were reported on parent-report measures of adolescent behavior (see Table 7).

Differences based on the sex of the ill parent were examined using independent

samples t-tests. Adolescents whose mothers were ill did not differ significantly from

adolescents whose fathers were ill on adolescent-report or parent-report measures of


Table 7

Means on measures of adolescent adjustment by sex
Males (n = 13) Females (n = 25)
Measure M SD M SD
RCMAS, Total score 49.77 10.58 48.92 11.17
RADS, Total score 50.14 7.27 47.04 9.35
IES, Total score 29.62 11.58 26.96 15.67
CBCL, Ill parent- Total Problems* 53.92 8.27 47.04 8.09
CBCL, Ill parent, Internalizing* 54.00 10.16 45.33 8.81
CBCL, Ill parent, Externalizing 52.31 7.19 48.21 8.22
CBCL, Healthy parent, Total Problems 50.08 8.40 48.61 11.91
CBCL, Healthy parent, Internalizing 47.17 7.32 49.39 11.18
CBCL, Healthy parent, Externalizing 48.75 6.74 50.00 11.10
Note. Values represent T score conversions for all measures, except IES. RCMAS =
Revised Children's Manifest Anxiety Scale; RADS = Reynolds Adolescent Depression
Scale; IES = Impact of Event Scale; CBCL = Child Behavior Checklist.
* Significant difference between males and females, p < .05

Correlation coefficients were calculated between adolescent age and dependent

(RCMAS, RADS, IES, and CBCL Total and Internalizing Problems for both ill and

healthy parents) variables. One correlation emerged as significant in one-tailed tests; IES

scores were positively correlated with age (r = .36, p < .05). All other correlation

coefficients were not significant at a .05 probability level (see Table 8).

Table 8

Correlations between demographic variables and study measures

Measure Adolescent age Time since diagnosis
PACS- Ill -. 11 .08
PACS- Healthy -.22 -.20
RCMAS .21 .04
RADS .20 .14
IES- Total .36* .10
CBCL Total- 11 .18 .26
CBCL Total- Healthy .09 .23
CBCL Internalizing- 11 .13 .19
CBCL Internalizing- Healthy .01 .05
Perceived Illness Severity .19 .13
Note. p < .05. All significance tests are one-tailed. PACS = Parent-Adolescent
Communication Scale; RCMAS = Revised Children's Manifest Anxiety Scale; RADS =
Reynolds Adolescent Depression Scale; IES = Impact of Event Scale; CBCL = Child
Behavior Checklist

Correlation coefficients were calculated between the amount of time the

adolescent had been aware of his or her parent's illness and outcome measures. These

were conducted with two-tailed significance tests as there were no hypotheses regarding

whether higher scores would be associated with the acute knowledge phase (i.e., less

time) or with greater time spent coping with the stressor. No correlation coefficients

approached significance on any parent report measures of behavior problems (CBCL

Total Problems or Internalizing) or self-report measures of adjustment (RADS, RCMAS,

or IES) (see Table 8).

To further ensure that no differences existed between those who had learned of

their parent's illness relatively recently and those who had been aware of it for a longer

period of time, independent samples t-tests were conducted between adolescents who had

been aware of the disease for one year or less and those who had been aware for more

than one year. No significant differences emerged.

The impact of adolescents' beliefs regarding the severity of their parents' illness

was examined. On the five-point Likert scale of perceived illness severity, 5% of

adolescents rated their parent's illness as 1, "not at all bad", 5% rated it 2, 26% rated it 3,

37% rated it 4, and 26% rated it 5, "extremely bad." The mean rating for the total sample

was 3.72, with a standard deviation of 1.08. Ratings differed by sex as well, with girls

(M = 3.94, SD = .74) describing their parent's illness as more severe than boys (M =

3.31, SD = 1.49). However, this difference was not statistically significant, t(15)= -1.44,

p > .05. In addition, adolescent ratings of perceived illness severity were not

significantly correlated with any self-report measures (i.e., RADS, RCMAS, IES, PACS).

Using previous research that has suggested that adolescents who experience three

or more stressors appear to be more vulnerable to psychological distress, adolescents

whose parents reported two or more stressors (in addition to parental illness) on the

demographic questionnaire were compared to those whose parents did not. Using

independent samples t-tests, no significant differences emerged for the RCMAS, RADS,

IES, or either ill or healthy parent CBCL total problem reports. No differences emerged

for healthy parent reports of internalizing behaviors; however, adolescents with two or

more additional stressors were rated by parents as having significantly more internalizing

symptoms than those who did not.

Hypothesis 2: Effects of Same Sex Ill Parents

It was hypothesized that adolescents with ill same sex parents would have more

psychological symptoms than those whose opposite sex parents were ill. This hypothesis

was not supported. Two (sex of adolescent) x 2 (sex of ill parent) analyses of variance

(ANOVAs) were conducted to assess differences on adolescent adjustment measures as a

function of the interaction between the sex of the adolescent and the sex of the ill parent.

Because of the number of tests conducted, the Bonferroni-Holm procedure was used to

correct for possible Type 1 errors, yielding an error rate of.007.

Main effects of adolescent sex and ill parent sex have been previously reported;

adolescent sex was associated with higher scores on ill parent reports of total problems

and internalizing symptoms. No significant interactions were detected for models testing

the RCMAS, F(1, 34) = .012, p > .007; RADS, F(1, 34) = .350, p > .007; or IES, F(1, 34)

= .040, p > .007; CBCL Total Problems as reported by the ill parent, F(1, 33)= .229, p >

.007; CBCL Total Problems as reported by the healthy parent, F(1, 31) = .091, R > .007;

CBCL Internalizing Problems as reported by the ill parent, F(1, 33) = 1.161, 2 > .007; or

CBCL Internalizing Problems as reported by the healthy parent, F(1, 31)= .053, p > .007.

Hypothesis 3: Adolescent Perceptions of Ill vs. Healthy Parent Communication

It was hypothesized that adolescents would perceive poorer communication

quality with their ill parents compared to their healthy parents. This hypothesis was not

supported. Means and standard deviations on the PACS are presented in Table 9 for the

total sample as well as separately for boys and girls. A 2 (sex of adolescent) x 2 (health

status of parent) repeated measures ANOVA was conducted on PACS scores to

determine whether adolescents perceived poorer communication quality with their ill

parent and whether this was affected by the sex of the adolescent. The main effect of

parent health status was nonsignificant, F(1, 36) = 2.07, p > .05. Results for the test of

the main effect of adolescent sex were nonsignificant, F(1, 36) = 2.64, p > .05. The

interaction effect between adolescent sex and health status of parent on communication

scores was also nonsignificant, F(l, 36) = 2.21, p > .05.

Table 9

Means and standard deviations on Parent-Adolescent Communication Scale (PACS)
Males (n = 13) Females (n = 25) Total (n = 38)
Parent group M SD M SD M SD
Healthy parents 46.70 10.94 55.27 12.58 52.34 12.59
Il1 parents 54.82 11.72 55.13 10.12 55.02 10.53
Mothers 51.99 10.97 58.15 10.56 56.04 10.96
Fathers 49.53 13.01 52.24 11.44 51.31 11.90
Same sex parents 49.53 13.01 58.15 10.56 55.20 12.02
Opposite sexparents 51.99 10.97 52.24 11.44 52.16 11.14
Note. Values represent T score conversions.

Secondary analyses were then conducted to assess for effects of other

demographic variables on parent-adolescent communication. A 2 (sex of adolescent) x 2

(sex of parent) repeated measures ANOVA was conducted on PACS scores to evaluate

whether adolescents reported differences in communication based on the sex of the parent

or the match between adolescent sex and parent sex. Results for the main effect of

adolescent sex were nonsignificant, F(I ,36) = 2.64, 2 > .05. The main effect of parent

sex was nonsignificant, F(l, 36) = 2.30, p > .05. Finally, the interaction effect between

adolescent sex and parent sex revealed no significant differences on the measure of

communication, F(1, 36) = .39, p> .05.

Adolescent sex differences in communication were further examined using

independent samples t-tests. The Bonferroni-Holm adjustment was used to control the

error rate, requiring a significance level of .008 for the first (lowest) p-value. No

significant differences emerged between girls and boys on ratings of perceived

communication with mothers, fathers, ill parents, healthy parents, same sex parents, or

opposite sex parents. In addition, PACS scores were not significantly associated with

adolescent age or the amount of time the adolescent had been aware of the parent's

illness (see Table 8).

Finally, Welch's v tests were conducted on raw scores of the current sample

compared to the normative sample collected by Olson (1985). Adolescents in the current

sample rated communication with their mothers significantly better than did adolescents

in the normative sample, v(9) = 3.09, p < .05. No significant difference emerged for

adolescent ratings of their fathers, v(7) = .55, 2 > .05. Comparisons based on health

status are not available as the measure has not been normed in this way.

Hypothesis 4: Relationship Between Parent-Adolescent Communication and

Adolescent communication with the healthy parent was hypothesized to be a

stronger predictor of adolescent adjustment than ill parent communication after testing

for age, sex, perceived illness severity, and socially desirable responding. This

hypothesis was partially supported. First, one-tailed bivariate correlation coefficients

were calculated to examine the relationships between the subscales of the PACS,

Problems in Family Communication (PFC) and Open Family Communication (OFC), and

measures of adolescent adjustment (see Table 10). Because normative information based

on parent sex is available only for total PACS scores, unadjusted raw scores were used

during these initial correlations. Adjusted scores were used in regression analyses,

reported next.

Perceived problems in the relationship with the healthy parent were significantly

negatively correlated with RCMAS and IES scores, such that lower scores (more

problems) were associated with more psychological symptoms. Open communication

with the healthy parent as reported by adolescents was significantly negatively associated

with RCMAS scores and ill parent reports on the CBCL only. Total healthy parent raw

scores were significantly associated with the RCMAS, IES, and ill parent CBCL scores.

Table 10

Correlations between measures of adjustment and PACS raw scores
Measure Ill parent PACS Healthy parent PACS

RCMAS -.16 .06 -.06 -.38* -.44* -.48*
RADS -.27 .01 -.15 -.25 -.10 -.20
IES- Total -.01 .18 .10 -.41* -.07 -.28*
CBCL Total- ll -.15 .10 -.03 -.23 -.37* -.35*
CBCL Total- Healthy .14 .12 .15 -.27 .00 -.17
CBCL Int.- Ill -.18 .14 -.02 -.24 -.44* -.40*
CBCL Int.- Healthy .13 -.04 .05 -.06 .09 .01
Note. p < .05. All significance tests are one-tailed. PACS = Parent-Adolescent
Communication Scale; PFC = Problems in Family Communication; OFC = Open Family
Communication; RCMAS = Revised Children's Manifest Anxiety Scale; RADS =
Reynolds Adolescent Depression Scale; IES = Impact of Event Scale; CBCL = Child
Behavior Checklist; Int. = Internalizing

Adolescent descriptions of problems in their relationships with their ill parents

were not significantly associated with scores on any adolescent adjustment measures.

Similarly, open communication with the ill parent was not associated with any of the

instruments assessing adolescent adjustment, nor were total raw scores for ill parent


To determine whether adolescent reports of communication were more strongly

correlated with adolescent adjustment, Williams tests for significant differences between

dependent correlations were conducted using the PACS raw scores. Each dependent

variable was treated as a family of tests, thus creating three tests for each variable. The

Bonferroni-Holm procedure was used to adjust for the number of analyses conducted

using one-tailed tests. Two comparisons were significantly different. The relationship

between open parent communication and generalized anxiety symptoms was significantly

stronger for healthy parents than ill parents, t = 2.24, p < .016. Also, the relationship

between open parent communication and ill parent reports of internalizing symptoms was

significantly stronger for healthy parents than ill parents, t = 2.64, p < .016. For a

complete list of these results, refer to Table 11.

Table 11

Differences between correlations of adjustment measures and PACS raw scores for ill
and healthy parents
Measure PFC OFC Total
t P t p t P

RCMAS 1.06 .14 2.24* .01 1.91 .03
RADS .10 .46 .45 .33 .21 .42
IES- Total 1.97 .03 1.05 .15 1.64 .06
CBCL Total- Ill .37 .36 2.07 .02 1.39 .09
CBCL Int.- Ill .28 .39 2.64* .01 1.68 .05
CBCL Total- Healthy 1.97 .03 .50 .31 1.36 .09
CBCL Int.- Healthy .87 .19 .54 .30 .17 .44
Note. Significant difference. All significance tests are one-tailed. PACS = Parent-
Adolescent Communication Scale; PFC = Problems in Family Communication; OFC =
Open Family Communication; RCMAS = Revised Children's Manifest Anxiety Scale;
RADS = Reynolds Adolescent Depression Scale; IES = Impact of Event Scale; CBCL =
Child Behavior Checklist; Int. = Internalizing

To assess whether perceived communication with the healthy parent was a

stronger predictor of adolescent adjustment than communication with the ill parent,

regression analyses were performed using total PACS scores, adjusted for norms based

on parent sex. An initial regression model was conducted, entering adolescent sex,

adolescent age, perceived illness severity, social desirability, and whether the

adolescent had experienced more than two stressors in the past year in a forward

manner to determine their relevance to the primary regression model of interest. No

correction for error rate was administered, as a more liberal approach to include any

potentially significant variables in the models of primary interest, those which examined

the influences of parental communication, was desired.

No variables remained in the models of the RCMAS, RADS, or healthy parent

CBCL Total Problems or Internalizing scores. The regression model for IES scores

indicated that adolescent age was the only significant predictor in the model (R2 = .13),

F(1, 36) = 5.49, p < .05. The regression model for ill parent reports of Total

Problems on the CBCL indicated that adolescent sex was the only significant predictor

in the model (R2 = .15), F(1, 36) = 6.00, p < .05. The model for ill parent reports

of internalizing symptoms indicated that both adolescent sex (being male) and having

experienced more than two stressors in the past year were significant predictors ( =

.32), F(2, 36) = 7.84, p < .05.

Next, regressions were conducted entering any significant demographic

predictors, followed by adolescent reports of perceived communication with the ill

parent (PACS T-scores), and, finally, by adolescent reports of perceived

communication with the healthy parent. Forward selection and stepwise regression

models yielded the same set of significant predictors; only the forward selection results

will be reported. Using the seven adolescent adjustment measures as dependent

variables, separate regression analyses were performed and the Bonferroni-Holm

adjustment was used. The RCMAS model resulted in a significant predictive model (Rj

= .23), F(1, 36) = 10.96, p < .007. Higher healthy parent communication was

associated with lower self-reported anxiety scores. No other models resulted in

significant predictors at the established p-value.

Hypothesis 5: Effects of the Presence of Positive Parental Relationships

It was hypothesized that adolescents with at least one positive parental

relationship, as defined by the PACS, would have fewer reported symptoms of depression

and anxiety than those with average or negative scores on the PACS. Scores one

standard deviation or more above the mean were determined to represent positive

communication in the parent-adolescent relationship. Adolescents were grouped

according to how many parental relationships they had rated as significantly more

positive than the normative sample: zero (n = 20), one (n = 11), or two (n = 7). Initial t-

tests were conducted to compare those adolescents who reported at least one positive

parental relationship with those who had not. No significant differences were detected on

adolescent self reports of anxiety on the RCMAS, t(36) = 1.519, p > .007, depression on

the RADS, t(36) = 1.184, p > .007, or stress response symptoms on the IES, t(36) = -.165,

p > .007. Similarly, no differences emerged on descriptions of total problems in

adolescent behavior by either ill parents, t(35) = 1.455, p > .007, or healthy ones, t(33) = -

.188, p > .007, on the CBCL Total Problems scale. No significant differences emerged

for descriptions of internalizing symptoms by ill parents, t(30) = 1.40, p > .007, or

healthy parents, t(33)= -. 18, p > .007

Exploratory analyses were conducted to examine whether adolescents with two

positive parental relationships appeared to fare better on measures of adjustment than

those with one or no positive relationship. Comparisons indicated no significant

differences between adolescents who rated communication with both parents as

significantly better than average when compared to those who rated one or none of their

parental relationships as positive (see Table 12).

Table 12

Means on measures of adjustment by number of positive parental relationships
0 1 2
Measure (n = 20a) (n = 1b) (n =7)
RCMAS, Total score 51.70 48.73 42.86
RADS, Total score 49.68 47.62 44.32
IES, Total score 27.50 30.00 25.57
CBCL, III parent, Total Problems 51.35 47.30 47.14
CBCL, Il1 parent, Internalizing 50.40 46.00 46.00
CBCL, Healthy parent, Total Problems 48.78 50.20 48.43
CBCL, Healthy parent, Internalizing 48.33 48.30 49.86
Note. Values represent T score conversions, except for IES. aSample size for CBCL,
healthy parent = 18; bSample size for CBCL, ill parent and CBCL, healthy parent = 10;
RCMAS = Revised Children's Manifest Anxiety Scale; RADS = Reynolds Adolescent
Depression Scale; IES = Impact of Event Scale; CBCL = Child Behavior Checklist

Hypothesis 6: Ill vs. Healthy Parent Reports of Adolescent Adjustment

It was hypothesized that healthy parents would endorse more problem behaviors

than ill parents would report. To compare differences in the levels of symptoms endorsed

by ill and healthy parents, dependent samples t-tests were conducted on the problem scale

ratings of the CBCL. These tests indicated no significant differences on the Total

Problems, t(33) = -.325, p > .017; Internalizing, t(33) = -.673, p > .017; or Externalizing

scales, t(33)= .059, p >.017.

Correlation coefficients were calculated to assess amount of parent agreement in

symptom ratings. Significant correlation coefficients emerged between ill parent T

scores and healthy parent T scores on both the Total Problems (r = .48, p < .01) and

Externalizing (r = .54, p < .01) scales of the CBCL. On the Internalizing scale of the

measure, healthy parent and ill parent ratings were not significantly associated (r = -.03, P

> .05).

It was further hypothesized that healthy parent reports of symptomatology would

be more strongly associated with adolescent reports of their own symptoms. While tests

indicated that no significant differences existed between the level of problems reported

by healthy and ill parents, further examination of the data was performed to evaluate

whether parental health status was related to the strength of association between parent

reports on the full scales of the CBCL and adolescent self-reports on the RADS,

RCMAS, and IES Correlations between parent-report measures of adolescent

adjustment and adolescent self-reports are presented in Table 13. Results indicated that

ill parent reports of both Total Problems (r = .31, p < .05) and Internalizing symptoms (r

= .31, p <.05) were associated with RCMAS scores in one-tailed significance tests. Ill

parent reports of Total Problems were also correlated with RADS scores (r = .36, p <

.05). No other significant relationships emerged between ill parent descriptions of

adolescent behavior problems and adolescent self-reports. Healthy parent reports were

not significantly associated with adolescent reports of their emotional functioning.

Table 13

Correlations between measures of adjustment

Measure 1 2 3 4 5 6 7 8 9
1. RCMAS, Total -- .60** .16 .31* .31* .12 .24 .24 .17
2. RADS, Total -- .11 .36* .25 .26 .25 .10 .22
3. IES, Total -- .12 .11 .05 .19 .18 .20
4. Ill- Total -- .79** .79** .48** .17 .47**
5. Ill- Int. -- .38** .17 -.03 .11
6. Ill- Ext. -- .46** .14 .54**
7. Healthy- Total -- .84** .86**
8. Healthy- Int. .60**
9. Healthy- Ext. --
Note. All significance tests are one-tailed. Variables 4 9 are Child Behavior Checklist
(CBCL) scales. RCMAS = Revised Children's Manifest Anxiety Scale; RADS =
Reynolds Adolescent Depression Scale; IES = Impact of Event Scale; Total = Total
Problems scale; Int. = Internalizing Problems scale; Ext = Externalizing Problems scale

Because parent reports and adolescent reports were not strongly correlated, it was

hypothesized that differences in the items forming the two measures may have yielded

differing reports of internalizing symptomatology. Therefore, measures were compared

to identify items that were shared on both parent-report and adolescent-report

questionnaires. These items are listed in Tables 14 and 15.

Table 14

Corresponding Items Used to Construct Matching Subset Between CBCL
Anxious/Withdrawn Subscale and RADS
CBCL item RADS item
12. Complains of loneliness. 3. I feel lonely.
14. Cries a lot. 8. I feel like crying.
33. Feels or complains that no one loves 12. I feel loved.
35. Feels worthless or inferior. 20. 1 feel I am no good.
103. Unhappy, sad, or depressed. 7. I feel sad.
112. Worries. 26. I feel worried.

Table 15

Corresponding Items Used to Construct Matching Subset Between CBCL
Anxious/Withdrawn Subscale and RCMAS
CBCL item RCMAS item
12. Complains of loneliness. 15. 1 feel alone even when there are
people with me.
32. Feels he/she has to be perfect. 8. I get nervous when things do not go
the right way for me.
34. Feels others are out to get him/her. 35. A lot of people are against me.
45. Nervous, highstrung, or tense. 34. 1 am nervous.
50. Too fearful or anxious. 7. 1 am afraid of a lot of things.
71. Self-concious or easily embarrassed. 14. I worry about what other people think
about me.
112. Worries. 6. I worry a lot of the time.

A "subset" score was generated for the RCMAS using only those seven items that

matched symptoms described on the Anxious/ Depressed subscale of the parent-report

CBCL. Similarly, only those seven items on the CBCL considered similar to the

RCMAS items were used to generate a short-version score. The same procedure was

used with six items on the RADS that were felt to match items on the CBCL Anxious/

Depressed subscale. Because of the nature of the items on the IES, few items

approximated those on the CBCL, therefore subset scores were not constructed for the

measure. Reliability coefficients were in the moderate range and are reported in Table


To determine whether adolescents and parents showed higher agreement when

asked about the same internalizing symptoms, bivariate correlations were performed

between adolescent subset scores for the RADS and RCMAS and healthy and ill parent

subset scores for the corresponding CBCL items. One-tailed significance tests indicated

that the correlation between the adolescent RADS subset and parent CBCL subset for

corresponding items was nonsignificant for both ill (r = .19, p > .05) and healthy (r = .28,

p > .05) parents. Similarly, the RCMAS subset was not significantly associated with

subset scores of ill parents (r = 10, p > .05) or healthy parents (r = .25, P > .05).

Table 16

Reliability coefficients for parent and adolescent subset scores of RCMAS and RADS

Adolescent .69 .75
Ill parent .71 .63
Healthy parent .65 .62
Note. RCMAS Revised Children's Manifest Anxiety Scale; RADS = Reynolds
Adolescent Depression Scale


Discussion of Major Findings

Hypothesis 1: Adolescent Adjustment to Parental Physical Illness

The present study had the advantage of using multiple reporters to assess

adolescent adjustment and emotional functioning in adolescents with severely chronically

ill parents. For the most part, with regards to average symptom levels, adolescents', ill

parents', and healthy parents' reports were in agreement that, overall, adolescents were in

the subclinical range of scores on measures of internalizing and externalizing symptoms.

While in contrast to the stated hypothesis, these findings are consistent with the majority

of prior research in this area.

However, a larger proportion of adolescents were in the clinical range of scores

on the measures of anxiety, the RCMAS and the IES, than on the depression measure, the

RADS. This pattern of results contrasts previous work in this area, which has found

more consistent findings related to depression than anxiety when comparing children of

ill parents to those of healthy parents. These data may be attributable to differences

between the sample used in this study and those of previous studies and support the

hypothesis that for teens with severely ill parents, anxiety symptoms are more prominent

than depressive ones.

First, parents in the study were severely ill and were receiving or considering

invasive treatments for their diseases, of which their children were aware. In studies of

parents with diseases that typically require less dramatic treatments, such as HIV or

chronic pain, children may have been less concerned about their parent's illness or less

aware of the treatment regimen, and thus symptoms of anxiety may not have been as

notable. It seems likely that children and adolescents, who presumably lack more

sophisticated understandings of medical information to make decisions, may rely on

external cues, like level of prescribed treatment, to make cognitive determinations

regarding the level of seriousness of a parent's illness. With the current group of

adolescents, their parents' needs for transplant, dialysis, or radiation may have served as

evidence of the severity of their parents' conditions. This is supported by the fact that

almost all of the adolescents reported that their parent's illness was moderately to

extremely severe. It may be that in this group of adolescents, the perceptions of severe

illness contributed to uncertainty about the future and, in turn, negative psychological

consequences as suggested by Steele, Tripp, and colleagues (1997).

The age range of the children in the present study may have contributed to these

findings as well. Most studies reviewed earlier included a broader age range that

included younger children. It may be that adolescents, with greater cognitive

development and abilities for abstract thinking, are more aware of the potential

ramifications of their parent's health condition than are school age children. They may

also be more likely to be provided with more information related to the parent's illness,

thus contributing to worries as a result. The present sample, entirely composed of

children more likely to be in this developmental range, may be at greater risk for

experiencing anxiety as a result than a sample including younger children.

As noted earlier, the IES is a stressor-specific measure that examined anxiety in

the form of intrusive thoughts and avoidance behaviors specifically related to the parent's

health condition. The IES assesses symptoms distinct from those of generalized anxiety

captured by the RCMAS. Unfortunately, IES normative data for a population similar to

that in the present study was not available. However, the mean total IES scores for the

current sample, while below the cutoff indicated by Neal et al. (1994), were comparable

to scores reported in previous studies of children who had been exposed to traumatic

events (see Table 2). These include children who had experienced a bus accident

(Stallard & Law, 1993), a dormitory fire (Jones & Ribbe, 1991), and war (Kuterovac,

Dyregrov, & Stuvland, 1994). In addition, several of the scores were similar to or higher

than those reported for multiple life events in the sample of adolescents obtained by

Joseph, Mynard, and Mayall (2000). Comparable to this study, these include adolescents

who reported a history of life threat to a family member or friend through accident,

injury, or illness. Other stressors rated similarly included life threat to self, experiencing

a natural disaster, parental divorce, and parental substance abuse.

It is important to note that adolescent females' reports of symptoms on the IES in

the study by Joseph, Mynard, and Mayall (2000) were significantly higher than males'

reports on several events, including life threat to a family member or friend. This

suggests that females may be more likely to report greater symptomatology in response to

traumatic events. Given that the composition of the present sample was two-thirds

female, the potential for somewhat elevated scores due to demographic characteristics

was considered, as adolescent females appear to report more symptoms on the IES.

However, examination of scores by sex indicated that this pattern was not present in the

current study; males reported slightly higher scores on the IES than females. This

suggests that the experience of having a severely ill parent is distressing regardless of sex

and can result in symptoms similar to those of children and adolescents who have

experienced an acute trauma, such as an accident or disaster.

Adolescent age was moderately positively correlated with IES scores. Again, this

may be a function of greater cognitive development that facilitated older adolescents'

access to more troubling information provided by adults. This information, in

combination with greater capacity to understand the possible ramifications of the illness,

may have caused these higher scores. However, IES scores were the only adjustment

measure significantly correlated with age. Reasons why this cognitive development

would differentially affect scores on the IES and not more general measures of anxiety

and depression are difficult to surmise. In addition, Compas et al. (1994), in using the

IES with 6- to 30-year-old children of cancer patients, found the opposite pattern of

results such that children had higher scores than adolescents, who had higher scores than

young adults. Further research is needed to replicate this finding to determine the nature

of the relationship between age and IES scores when used with children.

The scores on the IES are also noteworthy in that the comparison studies typically

reported levels of symptoms within weeks of exposure to a traumatic event. Sixty

percent of the adolescents in the present study had been aware of this stressor, their

parent's illness, for more than one year. Two studies using the IES at multiple time

points indicated that IES scores tend to decrease over time following acute exposure to

trauma (Vila, Porche, & Mouren-Semeoni, 1999; Yule & Williams, 1990). In addition,

findings from a two-year follow-up study of survivors of a coach accident (Curie &

Williams, 1996) indicated lower scores on the IES relative to other studies of children

and adolescents who were victims of accidents when those studies were conducted more

proximally to the time of the event.

Nonetheless, the findings for mean scores on the IES in this study remained

comparable even after longer time since exposure to the information. In addition,

measures of adjustment, including IES scores, were not associated with the amount of

time that the adolescent had been aware of the diagnosis. It is also noteworthy that the

scores in the present study were comparable to those studies measuring post-traumatic

stress symptoms in adolescents experiencing ongoing stressors such as war (Kuterovac,

Dyregrov, & Stuvland, 1994) or parental separation or divorce (Joseph, Mynard, &

Mayall, 2000).

Terr (1991) has suggested a conceptualization of childhood trauma that includes

mental harm inflicted by a single, unexpected event, referred to as a "type I" trauma, or

by a chronic, repetitive series of events, defined as a "type II" trauma. Stress response

symptoms have been most often studied following examples of "type I" traumas, such as

natural disasters or transportation accidents, but less is known about these symptoms

among those exposed to "type I1" traumas, especially those in which the child is not the

primary person experiencing the illness, but rather is a more secondary victim. For some

adolescents, it appears that parental chronic illness may represent a "type II" trauma in

which they experience traumatic reactions not in response to direct attacks on themselves,

but rather indirect threats to their security through threats to their parents. Results from

the IES in this study indicate that, even though adolescents do not directly experience the

traumatic event (i.e., illness) per se, they do experience a stress response pattern similar

to children who do, such as direct victims of disasters. These symptoms appear to persist

over time, as the stressor typically does not fade and remains via constant reminders of

poor health status.

Given that nearly one-third of the present sample obtained scores at or beyond the

recommended cutoff, it seems reasonable to conclude that a stress-response syndrome

related to their parent's illness that interferes with daily functioning is a significant issue

for many of these adolescents. It appears that the ongoing presence of parental illness

can continue to elicit symptoms of avoidance and intrusion even after its initial

presentation, causing distress similar to that typically associated with acute or

extraordinary traumatic events. These results suggest that these effects are chronic and

are likely grossly underestimated, contributing to distress among this population directly

through their presence and indirectly through their effects on parent relationships.

Symptoms of avoidance may have a profound impact on adolescents' relationships with

ill parents if they avoid reminders of the illness and fragile health status that are likely

associated with the parent's everyday existence. By not having social support from

parents, the anxiety reactions described may be exacerbated further.

Interestingly, IES scores were not correlated with parent reports of adjustment.

This suggests that, while parents may have some awareness of symptoms of depression

and anxiety, they may have little knowledge of the intrusive cognitions and behavioral

impact occurring for their adolescents. On the other hand, the IES may reflect a construct

completely unassociated with the problems tapped on the CBCL, the parent-report

measure used in this study. However, given the level of symptoms present on the IES in

this sample, it seems unlikely that this would be completely unrelated to CBCL items.

This raises concern about the use of generalized measures of anxiety such as the

RCMAS, which is more commonly used in clinical settings than the IES. This likely

results in the underidentification of stress-response symptoms among children and

adolescents even when chronic stressful life events are identified by clinicians. Broadly,

these findings suggest that by widening the definition of"trauma" among adolescents and

assessing for symptoms of a stress reaction, clinicians may be more likely to identify

these difficulties and assist in their treatment.

Both healthy and ill parent reports indicated between nine and seventeen percent

of adolescents in the clinical ranges on global measures of behavior problems (i.e. Total

Problems, Internalizing, and Externalizing). However, despite numerous adolescents

who reported clinical levels of anxiety on the RCMAS or IES, few parents reported

clinically meaningful symptoms on the Anxious/ Depressed subscale of the CBCL

despite several adolescents' reports of significant anxiety symptoms on the RCMAS and

IES. The CBCL lacks an anxiety subscale; rather it combines anxiety and depression into

one subscale. It may be that CBCL items on this subscale reflect depressive symptoms

more strongly than anxious ones, and therefore do not identify the concerns of these

adolescents. The nature of internalizing symptoms may also have led to lower scores, as

parents who are coping with multiple disease-related stressors may not be as attuned to

their children's emotional status.

Parent reports of adolescent functioning were relatively consistent between ill and

health parents with regards to mean scores and percentages of teens in the clinical ranges.

In the present study, the percentages of adolescents rated by parents as having significant

levels of symptoms was consistently higher than the expected 2% of the sample. Of note,

ill and healthy parents reported 14 and 17% of adolescents, respectively, in the clinical

range on externalizing problem behaviors. While externalizing behaviors were not the

focus of the present study, this evidence suggests that a significant number of families

observed behaviors of concern, warranting further investigation into acting out behaviors

among adolescents with chronically ill parents. However, it should be noted that the

number of participants in the present study was small and that each adolescent

represented 2.6% of the sample.

Demographic characteristics were generally not associated with parent reports of

adolescent functioning. However, ill parents described adolescent boys as having more

total problems and exhibiting more signs of internalizing difficulties than ill parents of

adolescent girls. The reasons for this difference are unclear, and findings related to

communication do not illuminate this result. In addition, these results contrast those of

Rotherham-Borus, Stein, and Lin (2001), who found that adolescent girls of parents with

HIV reported higher levels of emotional distress while boys reported higher levels of

behavior problems. The current sample showed no differences in externalizing behavior

problems, which may be a function of the reporter (i.e., parent-report measure used in the

present study).

Except where already noted, demographic factors, such as age and adolescent sex,

were not associated with self-report measures of adolescent adjustment. The amount of

time the adolescent had known about their parent's disease was not associated with

adjustment. This result complements previous studies that have found no relationship

between adjustment and length of diagnosis (Compas et al., 1994; Lewis et al., 1993) and

supports neither the hypothesis that the acute coping stages are most difficult or that

longer time spent dealing with the stressor are associated with greater problems.

In addition, perceived illness severity was also not associated with any adjustment

measures. This is in contrast to the findings from Compas and colleagues (1994), who

found that perceived illness severity was associated with self-reports of IES symptoms

although not generalized depression/anxiety symptoms. Several explanations for the

current finding may exist. First, statistically, the range of responses describing illness

severity was limited, with most of the adolescents indicating that their parent was

severely ill. This restriction of range may have prevented detection of such an

association. However, the scale used in this study was identical to that used in the

Compas et al. (1994) research. Second, the range of objective illness severity, while not

directly measured in this study, may have also been limited, as patients were selected

because of their medical status. Finally, it may be that no such relationship exists and

perceived illness severity is not associated with adolescent adjustment.

Hypothesis 2: Effects of Same Sex Ill Parents

The sex of the ill parent was not found to be associated with adolescent

adjustment, as in some previous studies (e..g Rodrigue & Houck, 2001). Also, unlike

other studies in this area (Compas et al., 1994; Rodrigue & Houck, 2001) adolescents

whose same sex parent was ill did not show evidence of greater adjustment problems than

other adolescents. It may be that the severity of the illnesses in the present study made

the sex of the parent less relevant from an emotional standpoint. However, the study by

Compas and colleagues was conducted with families of patients with cancer, which is a

severe illness associated with high levels of uncertainty like those in the present study,

making explanations for this finding somewhat unclear.

Hypothesis 3: Adolescent Perceptions of Ill vs. Healthy Parent Communication

The effects of parental communication on adolescent adjustment in families in

which one parent has a severe chronic illness was the focus of the present study, and

several findings related to communication should be noted. First, no significant

differences existed when comparing adolescents' communication ratings of ill parents to

those of healthy parents. In contrast to the hypothesis that adolescents might have more

access to healthy parents, feel less burdensome when disclosing information to them, and

feel more comfortable asking questions, the total sample of adolescents did not report

differences in communication based on the health status of the parent. The quality of

parent-adolescent communication does not appear to have been affected by parental

illness such that adolescents perceive significant differences in their relationships with

their parents. This is positive information when considering factors affecting

interpersonal dynamics for families with ill parents in that it suggests that adolescents do

not feel more difficulties opening up to them versus healthy parents. It does not appear

that the notion that ill parents might be perceived as emotionally unavailable or fragile is

supported by these data.

Several reasons for this finding may exist. Ill parents may make special effort at

effective communication with their children to compensate for other areas in which they

may feel less able to participate. Adolescents may spend more time with healthy parents,

which may allow for more openness for some and more conflict for others. Ill parents

may avoid these conflicts to a larger extent because adolescents may feel less

comfortable arguing with them or because they are less available and choose to use time

spent with their children in positive interactions. Another explanation for this finding

may be that adolescents were reluctant to disclose negative feelings about their

relationship with an ill parent. Guilt or shame may have led to socially desirable

responses, thus inflating the scores. Supporting this hypothesis is the fact that the only

adolescent-report measure of the study that significantly correlated with the social

desirability index of the RCMAS was that assessing ill parent communication.

Previous research on parent-adolescent relationships has found that adolescents

tend to perceive more openness (Barnes & Olson, 1985; Youniss & Smollar, 1985) and

greater closeness (Paulson, Hill, & Holmbeck, 1991) with mothers than fathers but also

tend to report more conflict with them as well (Montemayor & Hanson, 1985). While no

significant differences emerged between mothers and fathers, it is noteworthy that the

majority of healthy parents were mothers, making firm conclusions difficult to draw, as

health status may partially account for results.

Hypothesis 4: Relationship Between Parent-Adolescent Communication and

While no significant differences were reported in overall level of communication

between ill and healthy parents, differences in their predictive qualities for adolescent

adjustment did emerge. Adolescents' perceptions of communication with their healthy

parents were significantly associated with their adjustment, while their relationships with

ill parents showed less of a relationship. When the two subscales, openness and

communication problems, were combined to form the total communication scale,

communication with the healthy parent was related to generalized anxiety, stressor-

specific anxiety, and ill parent reports of problem behaviors. The hypothesis that