Professional Documents
Culture Documents
DOI: 10.1002/pon.5138
PAPER
Mareike Ernst1 | Elmar Brähler1 | Eva M. Klein1 | Claus Jünger2 | Philipp S. Wild2,3,4 |
1
Department of Psychosomatic Medicine and
Psychotherapy, University Medical Center of Abstract
the Johannes Gutenberg‐University Mainz,
Objective: A child's cancer diagnosis and treatment affect the whole family. While
Mainz, Germany
2 it has been recognized that parents are an important resource for their children, little
Preventive Cardiology and Preventive
Medicine, Center for Cardiology, University is known about the specifics of parenting in the face of serious illness.
Medical Center of the Johannes Gutenberg‐
University Mainz, Mainz, Germany Methods: We used the Recalled Parental Rearing Behavior Questionnaire in a
3
Center for Thrombosis and Hemostasis register‐based cohort of adult childhood cancer survivors (CCS) (N = 951) and a rep-
(CTH), University Medical Center of the
resentative population sample of the same age range (N = 2042). The questionnaire
Johannes Gutenberg‐University Mainz, Mainz,
Germany assesses behavior of mothers and fathers with three scales (emotional warmth,
4
German Center for Cardiovascular Research rejection/punishment, and control/overprotection) by querying the (former) child.
(DZHK), Mainz, Germany
5
We compared the two groups using general linear models. With a hierarchical linear
Department of Pediatric
Hematology/Oncology/Hemostaseology, regression analysis, we tested associations of recalled rearing behavior with
Center for Pediatric and Adolescent Medicine, disease‐ and treatment‐related factors.
University Medical Center of the Johannes
Gutenberg‐University Mainz, Mainz, Germany Results: Compared with the general population, CCS remembered both parents as
6
Institute for Medical Biostatistics, emotionally warmer, more overprotective, and less punishing/rejecting and less ambi-
Epidemiology and Informatics, University
Medical Center of the Johannes Gutenberg‐
tious. The regression analysis showed that having received radiotherapy (β = 0.092;
University Mainz, Mainz, Germany P = .009) and chemotherapy (β = 0.077; P = .027) was positively related to memories
Correspondence
of maternal emotional warmth.
Mareike Ernst, Department of Psychosomatic Conclusions: CCS remembered parenting styles which are generally deemed more
Medicine and Psychotherapy, University
Medical Center Johannes Gutenberg, positive. The extent of recalled control and overprotection deviated from the popula-
University Mainz, Untere Zahlbacher Str. 8, tion in different directions, suggesting that parenting in childhood cancer entails
55131 Mainz, Germany.
Email: mareike.ernst@unimedizin‐mainz.de more complex adaptations than being affectionate and giving comfort. The results
suggest an adaptation of parental behavior to particularly challenging treatments.
Funding information
Federal Ministry of Education and Research, They highlight potential vulnerability and resilience factors, some of which were
Grant/Award Number: BMBF01EO1503; Ger- sex‐dependent.
man Cancer Aid, Grant/Award Number:
70112165; German Research Foundation,
Grant/Award Numbers: FA1038/2‐1&2, K E Y W OR D S
SP1381/2‐1&2 and WI3881/2‐1&2 adaptation, cancer, childhood cancer survivors, family relations, oncology, parenting, pediatric,
survivorship
Psycho‐Oncology. 2019;28:1663–1670. wileyonlinelibrary.com/journal/pon © 2019 John Wiley & Sons, Ltd. 1663
1664 ERNST ET AL.
and clinical phenotyping, self‐administered questionnaires, and a of the dependent variable's variance in a statistically significant way
computer‐assisted personal interview (CAPI).21,23 after accounting for all other variables. We did this separately for
CCS and the representative sample completed the German ultra- maternal and paternal rearing behavior. Missing values were not
short screening version of the Recalled Parental Rearing Behavior imputed and are reported in the table legends. Sensitivity analysis
Questionnaire.16 It was derived from the Swedish scale Egna Minnen was performed using the calculator provided by Soper.27 As we are
av Barndoms Uppfostran (EMBU; my memories of upbringing). Previ- not aware of comparable investigations, we tested the sample size
ous studies using the short version have attested to its satisfactory required to observe a small effect ( f 2 = 0.02) on recalled rearing
to good psychometric properties and measurement invariance with behavior taking all five predictors of the final model into account,
respect to sex and age.17,24 The selected items had previously shown yielding a minimum number of 91 participants. Regression models
the highest factor loadings and had a high, positive correlation were checked for multicollinearity using the variance inflation factor
with the total value. Its subscales had relevant associations with (VIF). No observed VIF was larger than 4 (10 being the critical thresh-
relationship status, resilience, and depression and anxiety symptoms old28), indicating no concerning levels of multicollinearity. Effect sizes
17,24,25
in representative samples from the German population and (d) and regression coefficients are interpreted after Cohen.29 P values
26
with resilience in a clinical Korean sample. correspond to two‐tailed tests. Statistical analyses were performed
The scale consists of 12 items (3 dimensions with 2 questions per using SPSS 23 for Windows.
parent) referring to parenting style and behavior: “Did your
mother/father comfort you when you were sad?” and “Was your
mother/father able to smooch with you?” assessing emotional 3 | RESULTS
warmth, “Did your mother/father spur you to become the best?” (in
the following called ambition) and “Do you think that your
3.1 | Sample description
mother's/father's anxiety that something might happen to you was
exaggerated?” (in the following called overprotection), assessing Table 1 gives an overview of the CCS' and the population sample's
control/overprotection, “Have you been punished hard by your demographic characteristics and mean values for the different facets
mother/father, even for trifles (small offenses)?” and “Did it happen of parental rearing behavior. The most common treatment exposure
that your mother/father gave you corporal punishment without rea- among CCS was the combination of radiotherapy and chemotherapy.
son?”, and assessing rejection/punishment. Participants judge the The population sample was less likely to have a high level of educa-
respective frequency of occurrence on a Likert scale ranging from tion. This was true for men (d = 0.47) and women (d = 0.42). Cancer
0 = no, never to 3 = yes, always. In the present study, reliability esti- survivors were less likely to be married (men: d = 0.26; women:
mates of the scales were ω = 0.82 (emotional warmth), ω = 0.76 d = 0.50) (see Table SS1 for univariate comparisons).
(rejection/punishment), and ω = 0.71 (control/overprotection).
The representative sample's age (initial age range 14‐92 years) was We observed consistent, but small group effects (Table 2): CCS
aligned to CCS by excluding younger (<23 years) and older (>48 years) recalled their fathers and mother as emotionally warmer and less
participants. This reduced the initial sample of 5036 participants to rejecting/punishing. Regarding the control/overprotection scale,
2042. To robustly compare means, we conducted a general linear group affiliation had no impact on the maternal sum score; the
model (GLM). We modeled group (CCS vs population) and sex as fixed paternal sum score was lower in CCS. The single items were differ-
effects and age as a covariate. Criterion variables were the three ently affected by group affiliation: Maternal and paternal ambition
dimensions per parent (emotional warmth, rejection/punishment, and was reported as weaker by CCS, and overprotection as more
control/overprotection). Items of the control/overprotection scale pronounced. Age effects pertained to emotional warmth and
were investigated separately to offer more insight into the inconclu- control/overprotection (sum score and ambition item): Older
19
sive results of previous research. As this resulted in three GLMs participants remembered their parents as less warm, and less
investigating the dependent variable(s) control/overprotection (per controlling/overprotecting, in particular as less ambitious. There were
parent), for these analyses, the threshold for P values considered sig- also effects of the participants' sex: Women remembered mothers and
nificant was lowered to P = .017 (in all other instances, it was fathers as emotionally warmer. Women also remembered mothers as
P = .05). To investigate the relevance of disease‐related characteris- more rejecting, whereas men remembered fathers as more rejecting.
tics, we tested them as factors in separate hierarchical linear There was one interaction effect between group and sex: Male CCS
regression models (one per each scale of the Recalled Parental reported fathers as less rejecting than men of the general population.
Rearing Behavior Questionnaire and one per item of the control/ This effect was not present in women. There were different effects of
overprotection scale). Hierarchical linear regression allows for testing sex regarding the separate items of the control/overprotection scale.
whether the introduction of new predictors adds to the explanation Men remembered both parents as more ambitious than female
1666 ERNST ET AL.
TABLE 1 Demographic, disease‐, and treatment‐related information and recalled rearing behavior
Women Women
All N = 951 Men N = 526 N = 425 All N = 2042 Men N = 880 N = 1162
M SD M SD M SD M SD M SD M SD
Age 34.05 5.56 34.50 5.50 33.48 5.58 36.50 7.26 36.40 7.52 36.59 7.05
Age at diagnosis 6.14 4.28 6.56 4.32 5.63 4.18
Parental rearing behavior
Emotional warmth (mother)b 3.92 1.45 3.77 1.33 4.12 1.58 3.46 1.42 3.30 1.38 3.58 1.44
Emotional warmth (father)b 3.05 1.61 2.82 1.52 3.35 1.66 2.42 1.57 2.15 1.48 2.61 1.61
b
Rejection/punishment (mother) 0.34 0.73 0.31 0.63 0.38 0.83 0.47 0.89 0.42 0.80 0.51 0.94
b
Rejection/punishment (father) 0.37 0.76 0.37 0.72 0.36 0.82 0.65 1.03 0.76 1.09 0.56 0.97
Control/overprotection (mother)b 1.57 1.34 1.59 1.30 1.54 1.39 1.61 1.30 1.59 1.25 1.64 1.34
c
Overprotection 1.07 0.99 1.06 0.94 1.10 1.05 0.87 0.85 0.83 0.83 0.91 0.87
Ambitionc 0.50 0.75 0.54 0.77 0.45 0.71 0.74 0.81 0.76 0.78 0.73 0.83
b
Control/overprotection (father) 1.15 1.15 1.16 1.11 1.14 1.19 1.32 1.17 1.38 1.16 1.28 1.18
c
Overprotection 0.65 0.77 0.61 0.73 0.70 0.81 0.52 0.68 0.46 0.65 0.57 0.71
Ambitionc 0.50 0.75 0.54 0.76 0.44 0.73 0.80 0.84 0.91 0.86 0.71 0.82
n % n % n % n % n % n %
Demographics
High educationa 390 41.01 223 42.4 167 39.3 406 19.89 183 20.8 223 19.2
Married 354 37.23 200 38.0 154 36.2 1,188 58.18 453 51.5 735 64.8
Treatment‐related variables
Chemotherapy and radiation 444 46.7 248 47.1 196 46.1
Chemotherapy only 325 34.2 179 34.0 146 34.4
Radiation only 27 2.8 17 3.2 10 2.4
None of the two 74 7.8 41 7.8 33 7.8
Note. Missing values: marital status: 1 female CCS, educational level: 2 female CCS.
a
German Abitur/equivalent qualification ranked 4 in the European Qualifications Framework (EQF), required for postsecondary education at universities.
b
Range 0‐6.
c
Range 0‐3.
participants. However, women reported their fathers as more parents, corroborating the results of the previously reported GLM.
overprotective than men. Disease‐related variables associated with maternal warmth were che-
motherapy and radiotherapy: Participants who had undergone these
treatments remembered their mothers as emotionally warmer. Female
3.3 | Relations of disease characteristics and rearing sex was also positively related to the recalled paternal warmth.
behavior However, neither age at diagnoses nor therapy variables added
predictive power to the model.
We conducted hierarchical linear regressions to test whether disease‐
related characteristics (age at diagnosis, chemotherapy, and radiation)
were related to recalled parental rearing behavior (Table 3). Only the 4 | DISCUSSION
regression models with the dependent variables maternal warmth
(final model: R2 = 0.053, F (5, 820) = 10.236, P < .001) and paternal Drawing from a large, register‐based sample of long‐term CCS and a
warmth (final model: R2 = 0.036, F (2, 785) = 6.978, P < .001) representative sample from the community, the present study con-
explained statistically relevant proportions of variation in recalled rear- firmed differences between CCS and the general population in
ing behavior. This suggests that the predictors were not related to recalled rearing behavior using a validated questionnaire. The results
recalled rejection/punishment and control/overprotection. Female speak to an adaptation of parents' rearing behavior to the special,
sex was positively associated with recalled emotional warmth of both taxing situation of caring for a child afflicted by a serious illness. Our
ERNST ET AL. 1667
TABLE 2 Results of the general linear models on parental rearing results corroborate previous research showing that CCS recalled both
behavior their parents as emotionally warmer and less rejecting.19 They are in
Maternal Paternal line with other surveys and interviews of CCS9,15 and mirror parents'
statements about what they perceive as helpful parenting styles in
F P η2 F P η2
the face of childhood cancer.30
Emotional warmth
We add to previous research by showing that the subjective mem-
Sex 28.169 <.001 0.009 58.124 <.001 0.020 ories of ambition and overprotection were affected in different ways.
Group 60.173 <.001 0.020 103.236 <.001 0.034 Ours is the first study using the EMBU in a large CCS sample with this
Age 25.578 <.001 0.009 15.397 <.001 0.005 approach. However, as CCS remembered less maternal ambition and
Sex × group 0.143 .706 0.000 0.122 .727 0.000 more maternal overprotection than the general population, the single
Rejection/punishment items' mean scores might cancel each other out when calculating the
sum score. With regard to maternal control/overprotection, this was
Sex 6.448 .011 0.002 6.886 .009 0.002
the case in our investigation. Concerning this scale's sum score, there
Group 10.238 .001 0.003 52.139 <.001 0.018
were no differences between CCS and the general population,
Age 3.772 .052 0.001 3.260 .071 0.001
mirroring results reported by Lehmann, Hagedoorn.19
Sex × group 0.132 .716 0.000 6.305 .012 0.002 Regarding fathers' control/overprotection, however, the whole
Control/overprotection sum score differed between CCS and the population. This finding con-
Sex 0.011 .917 0.000 1.915 .167 0.001 trasts previous research19 using a longer version of the same question-
Group 1.500 .221 0.001 17.826 <.001 0.006 naire. Thus, possible differences might be attributed to the different
Age 5.883 .015 0.002 9.548 .002 0.003 questionnaire version or sample characteristics (such as older age at
TABLE 3 Hierarchical regression analyses of recalled emotional warmth in childhood cancer survivors
Note. Sex: men = 0, women = 1. Chemotherapy: no = 0, yes = 1. Radiation: no = 0, yes = 1. Values for B and 95% CI B are taken from the model in which the
respective predictor was introduced.
1668 ERNST ET AL.
Similarly, differences between men and women regarding ambitious one item each) and to the specificity of our CCS sample. Due to the
and overprotective behavior might be traced back to gendered expec- long follow‐up times in our study, its results might not reflect the
tations of a child's individual capacities and vulnerabilities. Along these experiences of childhood cancer patients with shorter survival times.
lines, the experiences of childhood cancer survivors today might be Furthermore, they might not reflect the experiences of those long‐
different—on the one hand because of improved survival rates1 and term cancer survivors who did not respond to the study invitation.
on the other hand due to more liberal parenting styles.33 The absence of data assessed at the time of cancer diagnosis and
Lastly, the results caution against a reductionist categorization of treatment is another limitation. Retrospectively assessing parental
recalled parental rearing behavior as “positive” or “negative”: On the rearing behavior by querying the former child entails a number of
one hand, emotional warmth has been implicated as a powerful pro- caveats. Most likely, these internal conceptions have been shaped by
tective force, fostering resilience and adaptation. However, in combi- recurring experiences throughout development. As they depend on
nation with fearful overprotection, and more forgiving parenting the individual's subjective appraisal, they might be influenced by mood
behavior, CCS might encounter conflicts in navigating developmental or recent events. However, memories of parenting styles have shown
tasks.34 More longitudinal research is needed which links variables of to be stable across mood changes.38,39 Lastly, a limitation is the
the context of the experience of childhood cancer to well‐being later absence of data on the family constellation, eg, growing up in a one‐
in life in a meaningful way to test the clinical relevance of the small parent household or experiencing a divorce/separation.
group differences observed in the present study. The aim should be
to advance the understanding of resilience factors, to detect those
CCS who are at risk for unfavorable outcomes, and to offer 5 | CO NC LUSIO NS
adequate interventions to pediatric cancer patients, survivors, and
their families. The present study details particularities of parenting as recalled by
survivors of childhood cancer. It expands previous research by using
a quantitative approach to compare a large sample of adult long‐term
4.1 | Clinical implications survivors with a representative population sample. CCS remembered
more affectionate parenting than participants from the general popu-
The present results suggest that parenting in the context of childhood
lation. They also remembered more overprotection and less ambition.
cancer is more complex than just giving solace as cancer modulated
Our results speak to an adaptation of parents' rearing behavior in the
other parental behaviors as well. This finding can be helpful in guiding
face of an extremely taxing situation and shine a light on powerful
supportive care efforts during and after treatment. There is abundant
predictors of vulnerability and resilience over the life span.
evidence that parental behavior has far‐reaching mental health conse-
quences, so it should also be explored in the context of childhood can-
cer survivors' long‐term adaptation and coping. Especially discussing ACKNOWLEDGEMENTS
experiences of parents' overprotective behaviors could give insight We thank all former childhood cancer patients who underwent clinical
into survivors' individual struggles and decisions revolving around examination for this study, all participating and supporting medical
autonomy. Along these lines, more research is needed to link parent- centers, the study centers of the GPOH, the staff of the GHS,
ing styles with long‐term survivors' quality of life and other relevant and the staff of the treatment data retrieval team. The CVSS is
lived outcomes. funded by the German Research Foundation (DFG) (SP1381/2‐1&2,
Importantly, changes in parental behavior can indicate a way in FA1038/2‐1&2, WI3881/2‐1&2). PSYNA is funded by the German
which parents' troubles affect the child. Bearing in mind the high levels Cancer Aid (DKH) (70112165). P.S. Wild is funded by the Federal
of psychological distress among parents caring for childhood cancer Ministry of Education and Research (BMBF01EO1503).
patients, especially those undergoing intensive treatments,3,35 not
every parent might be able to offer their child the needed emotional
support. For example, depressed mothers have previously shown less CONFLIC T OF INT E RE ST
36
sensitivity towards their children and less observant caregiving. None. This work is part of the dissertation of the first author.
Routine screening and targeted interventions should be available in
order to detect and alleviate parents' distress. Ultimately, this will also
benefit pediatric cancer patients or long‐term survivors (who previ- DATA AVAILABILITY STATEMENT
ously showed increased susceptibility towards mental distress37).
The written informed consent of the study participants is not suitable
for public access to the data, and this concept was not approved by
4.2 | Study limitations the local data protection officer and ethics committee. Access to data
at the local database in accordance with the ethics vote is offered
Limitations pertain to the limited information available regarding pre- upon request at any time. Interested researchers make their requests
vious treatment, the brevity of the Recalled Parental Rearing Behavior to the principal investigators of the CVSS/PSYNA study (philipp.
Questionnaire (ie, that overprotection/ambition were assessed using wild@unimedizin‐mainz.de).
ERNST ET AL. 1669