564061 research-article2014
JFNXXX10.1177/1074840714564061Journal of Family NursingKobayashi et al.
Article
Interrelations Between Siblings and Parents in Families Living With Children With Cancer
Journal of Family Nursing 2015, Vol. 21(1) 119–148 © The Author(s) 2014 Reprints and permissions: sagepub.com/journalsPermissions.nav DOI: 10.1177/1074840714564061 jfn.sagepub.com
Kyoko Kobayashi, RN, PHN, PhD1, Akira Hayakawa, MD, PhD2, and Naohiro Hohashi, RN, PHN, LSN, PhD2
Abstract Having a child diagnosed with cancer is a stressful event for the family. This exploratory multimethod study utilized both quantitative and qualitative multiinformant methodologies to investigate the relationships between parental family functioning and siblings’ health-related quality of life (HRQOL) and to describe interrelations between the experiences of parents and siblings of children with childhood cancer. A total of 14 Japanese families participated in the quantitative study, and 4 families of the 14 participated in the qualitative study. In-depth, semistructured interviews revealed three family-unit stages during the time course of the ill child’s treatment that included particular parent–sibling interrelations. We also found strong correlation between parental family functioning and siblings’ HRQOL in the quantitative study. The results suggest the importance of family nursing interventions directed to individual family members and the family unit that focus on strengthening the parent–sibling relationship and supporting families who are experiencing childhood cancer.
1St.
Luke’s International University, Tokyo, Japan University, Kobe, Japan
2Kobe
Corresponding Author: Kyoko Kobayashi, Associate Professor, Department of Child and Family Nursing, St. Luke’s International University School of Nursing, 10-1, Akashicho, Chuo, Tokyo 104-0044, Japan. Email: [email protected]
Downloaded from jfn.sagepub.com at GEORGIAN COURT UNIV on May 3, 2015
120
Journal of Family Nursing 21(1)
Keywords siblings, sibling–parent relationship, family functioning, interrelations, childhood cancer, quality of life, Japan
The prevalence of childhood cancer is approximately 150 per million in Japan (Baba et al., 2010). Although advances in treatment have accomplished improved survival rates exceeding 70% in both Japan and the West (Baba et al., 2010; Feltbower et al., 2009; Fujita et al., 2011), childhood cancer is the second most common cause of death for children under 15 years of age, after accidents (Statistics and Information Department, Minister’s Secretariat, Ministry of Health, Labour and Welfare, 2014). In other words, cancer is the leading cause of death among diseases in Japan. Having a child diagnosed with cancer is a stressful event for the family and forces them to face many challenges (Alderfer & Kazak, 2006; Deatrick et al., 2006). Along with the fear of a life-threatening disease and the associated emotional reactions, families must make many difficult adjustments. The parents have heavy responsibilities: accompanying the ill child to the hospital, caring for the child with side effects, doing chores at home, and earning money at work. Furthermore, the unpredictable course of the illness takes away their control of their daily lives and causes parents to struggle to adjust their family life (Clarke-Steffen, 1993, 1997; McGrath, 2001; Woodgate & Degner, 2004). Siblings are also affected by the diagnosis and experience various family life changes. At least one of the parents stays at the hospital with the ill sibling, and healthy siblings frequently stay home alone or stay with their relatives (usually grandparents), separated from their ill sibling and parents (Harding, 1996). The separation can be quite lengthy, as children with cancer in Japan are generally hospitalized for longer periods compared with those in most Western countries. These changes may trigger numerous difficulties for the siblings in their daily lives and family relationships, affecting their health-related quality of life (HRQOL).
Siblings of Children With Cancer Many previous studies dealing with the effects of childhood cancer on siblings have focused on the siblings’ psychological functioning by investigating such aspects as behavioral problems, depression, anxiety, loneliness, and guilt (Alderfer et al., 2010; Buchbinder et al., 2011; McGrath, 2001; Nolbris, Enskar, & Hellstrom, 2007; Patterson, Holm, & Gurney, 2004; Sloper, 2000; Woodgate, 2006). The majority of previous studies did not find indications of
Downloaded from jfn.sagepub.com at GEORGIAN COURT UNIV on May 3, 2015
Kobayashi et al.
121
obvious depression or anxiety in siblings of children with cancer. However, some studies have reported more depression among siblings of survivors of childhood cancer (Buchbinder et al., 2011), and have revealed adjustment problems, loneliness, guilt, and shock in young siblings (McGrath, 2001; Nolbris et al., 2007; Patterson et al., 2004). Other studies have described siblings’ school and social functioning by investigating school problems, problems of social relationships, and poor academic achievement (Alderfer & Hodges, 2010; Houtzager, Grootenhuis, Hoekstra-Weebers, & Last, 2005; Lahteenmaki, Sjoblom, Korhonen, & Salmi, 2004). Still other studies have revealed physical problems such as sleep deprivation and eating problems (Zeltzer et al., 1996). In addition, a few studies have reported symptoms of posttraumatic stress as a long-term effect after the completion of a sibling’s cancer treatment (Kaplan, Kaal, Bradley, & Alderfer, 2013). It is evident from the research that siblings experience comprehensive difficulties related to psychological, social, and physical functioning. HRQOL is a multidimensional concept that comprises elements of physical, functional, social, and psychological health, as well as the individual’s perceived health status and well-being (de Hann, Aaronson, Limburg, Hewer, & van Crevel, 1993). A few studies have reported HRQOL of siblings of children with cancer, and Houtzager et al. (2004) reported poor HRQOL of siblings compared with other children. HRQOL is affected by social and cultural factors (Franco et al., 2012). However, as far as we know, there is no research reporting the HRQOL of Japanese siblings of children with cancer.
The Sibling–Parent Relationship In studies by Sloper (2000) and Woodgate (2006), siblings reported closer family relationships as a result of the process of coping with an illness situation; nevertheless, they experienced losses of attention, family routine, security, and a sense of well-being within their families. Similarly, McGrath (2001) found that parents experienced role conflict between responsibility for the sick child and caring for siblings at home, and they had to relinquish the roles of nurturing the siblings and maintaining the home and daily family life. Although parents recognize that their family cohesion is strengthened through the experience of one child’s cancer, at the same time, they feel guilt for neglecting their other children (Brown, Pikler, Lavish, Keune, & Hutto, 2008; D. S. Cohen, Friedrich, Jaworski, Copeland, & Pendergrass, 1994). D. S. Cohen et al. (1994) administered the Child Behavior Checklist to primary caregivers of children with cancer to investigate variables related to sibling adaptation and found that high levels of family cohesion and adaptability were associated with better adaptation of siblings. These studies indicate that
Downloaded from jfn.sagepub.com at GEORGIAN COURT UNIV on May 3, 2015
122
Journal of Family Nursing 21(1)
the interrelations among family members affect siblings’ well-being. Also, an association between social support and more positive emotional, behavioral, and academic achievement has also been reported (Alderfer & Hodges, 2010). However, the relationship between siblings’ self-reported HRQOL and family functioning and social support, and the interrelations of siblings’ and parents’ experiences has, to date, not been thoroughly investigated. Most previous research on siblings of children with cancer has been conducted using child mono-informant or parent proxy-reported data. Even in studies in which several family members acted as informants, each family member’s data were analyzed separately. Family research is conducted from a family-unit perspective, and measures from multiple family members are considered necessary to capture the complexity of the family system (Cromwell & Olson, 1975). HRQOL is a measure of an individual’s perception, and proxy evaluation of children’s HRQOL is influenced by the informant’s mental health and well-being (Cremeens, Eiser, & Blades, 2006; Kobayashi & Kamibeppu, 2011). Therefore, the HRQOL of siblings of childhood cancer patients should be evaluated using the siblings’ self-reported measures. At the same time, family functioning is an index of the family unit as a whole, and a multiple-informant quantitative study is indicated. Although a quantitative approach is utilized for research that aims to predict, control, and explain family phenomena (Moriarty, 1990), quantitative research on family nursing is limited in its ability to create data that will reflect the family as a unit (Fisher et al., 1985). Furthermore, the findings of quantitative researches can reveal the magnitude and spread of the problem (Lavera et al., 2008) but fail to capture the richness and complexity of contextual factors (Patton, 1990). Qualitative research can overcome these weaknesses by providing a “human context” behind the numbers, even if such research has limited ability to assess the extent of the problem (Lavera et al., 2008). In addition, the results of quantitative research and null-hypothesis significance testing are sensitive to sample size and have limited ability to determine the practical significance of statistical relationships. Methodological problems, therefore, arise because of the low prevalence of childhood cancer. Moreover, small sample sizes in previous research (McGrath, 2001; Nolbris et al., 2007) have not always included siblings and multiple family members. The strength of a qualitative approach is its power to generate rich data about families’ subjective experiences and relationships (Moriarty, 1990), thereby allowing sibling–parent interrelations to be more richly described. Similar to quantitative research of sibling–parent interrelations, it is necessary in qualitative research to investigate experiences that are described by multiple informants to understand both the siblings’ and parents’ perspectives.
Downloaded from jfn.sagepub.com at GEORGIAN COURT UNIV on May 3, 2015
Kobayashi et al.
123
Therefore, this exploratory study, utilizing multimethods to overcome the weaknesses of single-method research and provide more complete data (Creswell & Plano Clark, 2006), used multiinformant methodologies to investigate the interrelations between siblings’ HRQOL and parents’ family functioning in families with a child diagnosed with cancer. We conducted quantitative and qualitative research concurrently, using quantitative research methods along with qualitative research methods to complement, explain, and expand the quantitative findings. The quantitative research was conducted to clarify the HRQOL of siblings of children with cancer and the correlation between siblings’ HRQOL and family functioning and social support, whereas the qualitative research was conducted to describe how siblings’ experiences and parents’ experiences interrelate to maintain siblings’ HRQOL within the family context.
Purposes of the Study Our specific aims were (a) to clarify the self-reported HRQOL of Japanese siblings of children with childhood cancer and family functioning from the siblings’ and parents’ perspectives (quantitative), (b) to clarify the state of agreement or disagreement in siblings’ self-reported family functioning and HRQOL and parents’ reports of family functioning and proxy-reported sibling HRQOL (quantitative), (c) to examine interrelations between siblings’ HRQOL and family functioning and social support by utilizing multiinformant data (quantitative), and (d) to describe how sibling and parent experiences were interrelated by using a multiinformant qualitative study (qualitative).
Method Research Design A cross-sectional, multimethod study was conducted in Japan using multiinformant data. Prior to the data collection, ethical approval was obtained from the institutional review board of Kobe University, with which two study authors are affiliated.
Sample Healthy siblings of children who were outpatients of a pediatric oncology clinic in Japan and their mothers and fathers were invited to participate. The inclusion criteria were (a) siblings who were 8 to 18 years old; (b) siblings who lived with a brother or sister who had been diagnosed
Downloaded from jfn.sagepub.com at GEORGIAN COURT UNIV on May 3, 2015
124
Journal of Family Nursing 21(1)
with cancer; and (c) siblings, mothers, and fathers who were not physically or mentally impaired at the time of the study. All families who met the inclusion criteria were approached during a seven-month period from 2009 to 2010. A clinic physician gave an explanation sheet about the study to all parents of children with cancer who had siblings when they accompanied children with cancer to the pediatric oncology clinic. Parents who were interested in the study visited one of the researchers (first author) in a private room within the clinic. The researcher explained the purpose of the study by means of a written explanation. If the parent agreed to participate, a questionnaire for parents and siblings were given to the parents (quantitative). At this time, parents were asked whether they were willing to be interviewed (qualitative); if they were, their telephone number was obtained. After reconfirmation of agreement of parents and siblings to participate in the interview, the date and place of the interview were arranged according to the family’s preferences. For ethical reasons, parents’ positive or negative decisions to visit the researcher and to participate in the research were not reported to the physician. A total of 18 families were approached, and 14 families participated in the quantitative study (a response rate of 77.8%). Among the 14 families, 13 siblings participated; 7 were female (53.8%), and the mean age of the siblings was 12.8 years (range = 8-18 years). The mean time since discharge of the ill child was 26.7 months (range = 3-35 months), and children had remained in the hospital for a mean duration of 10.8 months (range = 5-25 months). Of the 14 ill children, 6 had undergone treatment for leukemia, 4 had been treated for solid tumor, and 1 for lymphoma (data missing for 3 children). Of the parents, 12 mothers and 10 fathers returned their questionnaire sheets. For the qualitative study, 4 of 13 families were interested in participation (totaling 11 family members; 4 siblings, 4 mothers, and 3 fathers). All were families of four: two parents, the ill child, and the healthy sibling. Regarding the siblings’ relationships to the ill children, one was a younger sister, one an older sister, and two were older brothers.
Quantitative Study Measures The Pediatric Quality of Life Inventory (Japanese version; PedsQL-J). Informants: siblings, mothers, and fathers. The Japanese version of the Pediatric Quality of Life Inventory 4.0 Generic Core Scales (PedsQL-J; Kobayashi & Kamibeppu, 2010) was used to assess sibling HRQOL. Both the self- and proxy-reports comprise 23 items covering four subscales: physical, emotional, social, and
Downloaded from jfn.sagepub.com at GEORGIAN COURT UNIV on May 3, 2015
Kobayashi et al.
125
school functioning. Each item is scored on a five-point Likert-type scale from 0 (never) to 4 (almost always). The scores are calculated by using a linearly transformed, reversed score range from 0 to 100 (0: 100, 1: 75, 2: 50, 3: 25, 4: 0). Higher scores indicate better HRQOL. Scale scores are calculated as the sum of the items for that scale divided by the number of items answered; however, when more than 50% of answers for the scale are missing, the scale score is not calculated (Varni, Seid, & Kurtin, 2001). The psychometric properties of the Japanese version of the PedsQL have been empirically tested (Kobayashi & Kamibeppu, 2010; Kobayashi, Okano, & Hohashi, 2011); the Japanese version has acceptable internal consistency (Cronbach’s α = .85 for children’s self-reports and .92 for parents’ proxy-reports) and good clinical validity in differentiating HRQOL among different health statuses. Social Support Survey for Children (SSSC). Informant: siblings. Generally, the concept of social support is defined as having three dimensions—social embeddedness, perceived support, and enacted support (Barrera, Sandler, & Ramsay, 1981). Perceived support is most important for children because their social network is generally limited compared with adults, and they depend for enacted support on the adults who are closest to them, such as parents or teachers (S. Cohen & Hoberman, 1983). In this study, perceived social support was assessed using the SSSC (Shimada, 1998), originally developed as a Japanese-language instrument. The SSSC contains five items such as “He/she cheers you up when you are depressed.” For each item, respondents give their perceptions of the degree of support they expect from five resources: father, mother, siblings, teachers, and friends. A four-point Likert-type scale (from 1, definitely no, to 4, definitely yes) is utilized, with higher scores representing a higher expectation of support from a particular source. The psychometric properties of the SSSC have been tested, and satisfactory Cronbach’s α for each resource (ranging from .90 to .97) and a well-validated factor structure were revealed (Shimada, 1998). Family Adaptability and Cohesion Evaluation Scale (FACESKG). Informants: siblings, mothers, and fathers. Family functioning was assessed by using the Japanese version of the FACESKG (Tachigi, 1999). The FACESKG is a selfreport questionnaire that is based on the Circumplex Model of Marital and Family Systems (Olson, 1991). This model consists of two dimensions of family systems: cohesion and adaptation. Cohesion is defined as “the emotional bonding that family members have towards one another” (Olson, 1989, p. 9). Adaptability is defined as “the ability of a marital or family system to change
Downloaded from jfn.sagepub.com at GEORGIAN COURT UNIV on May 3, 2015
126
Journal of Family Nursing 21(1)
its power structure, role relationships, and relationship rules in response to situational and developmental stress” (Olson, 1989, p. 12). The model identifies four types of cohesion (disengaged [very low], separated, connected, and enmeshed [very high]) and four types of adaptation (rigid [very low], structured, flexible, and chaotic [very high]). A four-point Likert-type scale from 1 (never true) to 4 (always true) is utilized. To describe family functioning, we identified the types for each dimension as well as calculating the sum of the score. The FACESKG showed good internal consistency (alphas of each dimension for each family member ranged from .72 to .89) and good clinical validity in differentiating among different health statuses (Tachigi, 1999). Feetham Family Functioning Survey–Japanese (FFFS-J). Informants: mothers and fathers. The Japanese version of the FFFS-J (Hohashi, Maeda, & Sugishita, 2000), originally developed by Roberts and Feetham (1982), was used to assess parental perceptions of family functioning. The FFFS-J is based on the Family Ecological Model (Bronfenbrenner, 1977) and assesses functioning not only of intrafamily relationships but also of interfamily relationships, including those with subsystems (e.g., neighborhood) and the broader society according to the definition of the Family Ecological Model (Hohashi, Honda, Hiratani, & Feetham, 2008). The FFFS-J is a self-administered questionnaire consisting of 27 items divided into two sections: 25 family-function items and two open-ended questions. In this study, we analyzed the 25 family-function items only. These questions ask about family functioning in three areas: the relationship between the individual and family members (relationship with family members; 10 items), relationship between family and subsystems (eight items), relationship between family and society (six items), and one item that does not belong to any of the three areas. For each item, respondents are asked to answer three questions: (a) “How much is there now?” (b) “How much should there be?” and (c) “How important is this to you?”—on a seven-point Likerttype scale from 1 (little) to 7 (much). The absolute value of the difference between the responses to the score for (a) and the score for (b) is calculated for each family function (d score). The d score represents the discrepancy between the perceived actual family function and the perceived expected family function, and a higher d score indicates greater perceived dissatisfaction with a function (Hohashi, Honda, et al., 2008; Roberts & Feetham, 1982). The overall family-functioning score (total d score) is calculated by summing the item d scores, and three area d scores are calculated by summing the items within each area and dividing by the number of items answered. Cronbach’s α for the three areas and the total d score of the FFFS-J were .66, .77, .82, and .80, respectively.
Downloaded from jfn.sagepub.com at GEORGIAN COURT UNIV on May 3, 2015
Kobayashi et al.
127
Analysis. Descriptive statistics for each measurement were calculated. The means of the PedsQL-J scores were compared with published Japanese general population norms (Kobayashi et al., 2011) using the Mann–Whitney U test. Mean values for norm data for healthy children were extracted from a previous study (Kobayashi et al., 2011), and analyses showed no differences in mean ages of the children (Mann–Whitney U test) or frequency of gender (chi-square test) between norm data and the present study. Also, effect sizes (Cohen’s d) were calculated by dividing each mean score difference by the standard deviation of the general population norm (Kobayashi et al., 2011). Although null-hypothesis significance testing is sensitive to sample size and has limited ability to determine the practical significance of statistical relationships, effect sizes estimate the magnitude of an effect or the association between two or more variables and are resistant to sample-size influence. Therefore, in gauging the practical importance, effect size provides an accurate measure of the magnitude of effect between variables (Ferguson, 2009). Effect sizes were interpreted according to J. Cohen (1988) as small (.20), medium (.50), and large (.80). Furthermore, distributions of siblings’ and parents’ perceived HRQOL compared with percentiles of the general population were calculated. The agreement levels of the PedsQL-J between siblings and parents were examined with intraclass correlation coefficients (ICCs) and mean difference testing using the Wilcoxon signed-rank test and Spearman rank order correlation (ρ). The effect sizes of the ICC and ρ were categorized as small (.5; J. Cohen, 1988). The magnitude of the discrepancies between sibling and parent PedsQL-J scores was examined by calculating the means of the absolute parent–sibling differences. In addition, the means of the directional differences (parent proxy-reported scores minus sibling self-reported scores) were determined. The directions of the differences can indicate the bias of the parent proxy-reported scores relative to the sibling self-reported scores, and negative differences indicate parental underestimation. The mean values of the directional differences were standardized (d) by relating a given score to its standard deviation (SD) to determine the statistical magnitude of any observed systematic biases, which were labeled negligible (
JFNXXX10.1177/1074840714564061Journal of Family NursingKobayashi et al.
Article
Interrelations Between Siblings and Parents in Families Living With Children With Cancer
Journal of Family Nursing 2015, Vol. 21(1) 119–148 © The Author(s) 2014 Reprints and permissions: sagepub.com/journalsPermissions.nav DOI: 10.1177/1074840714564061 jfn.sagepub.com
Kyoko Kobayashi, RN, PHN, PhD1, Akira Hayakawa, MD, PhD2, and Naohiro Hohashi, RN, PHN, LSN, PhD2
Abstract Having a child diagnosed with cancer is a stressful event for the family. This exploratory multimethod study utilized both quantitative and qualitative multiinformant methodologies to investigate the relationships between parental family functioning and siblings’ health-related quality of life (HRQOL) and to describe interrelations between the experiences of parents and siblings of children with childhood cancer. A total of 14 Japanese families participated in the quantitative study, and 4 families of the 14 participated in the qualitative study. In-depth, semistructured interviews revealed three family-unit stages during the time course of the ill child’s treatment that included particular parent–sibling interrelations. We also found strong correlation between parental family functioning and siblings’ HRQOL in the quantitative study. The results suggest the importance of family nursing interventions directed to individual family members and the family unit that focus on strengthening the parent–sibling relationship and supporting families who are experiencing childhood cancer.
1St.
Luke’s International University, Tokyo, Japan University, Kobe, Japan
2Kobe
Corresponding Author: Kyoko Kobayashi, Associate Professor, Department of Child and Family Nursing, St. Luke’s International University School of Nursing, 10-1, Akashicho, Chuo, Tokyo 104-0044, Japan. Email: [email protected]
Downloaded from jfn.sagepub.com at GEORGIAN COURT UNIV on May 3, 2015
120
Journal of Family Nursing 21(1)
Keywords siblings, sibling–parent relationship, family functioning, interrelations, childhood cancer, quality of life, Japan
The prevalence of childhood cancer is approximately 150 per million in Japan (Baba et al., 2010). Although advances in treatment have accomplished improved survival rates exceeding 70% in both Japan and the West (Baba et al., 2010; Feltbower et al., 2009; Fujita et al., 2011), childhood cancer is the second most common cause of death for children under 15 years of age, after accidents (Statistics and Information Department, Minister’s Secretariat, Ministry of Health, Labour and Welfare, 2014). In other words, cancer is the leading cause of death among diseases in Japan. Having a child diagnosed with cancer is a stressful event for the family and forces them to face many challenges (Alderfer & Kazak, 2006; Deatrick et al., 2006). Along with the fear of a life-threatening disease and the associated emotional reactions, families must make many difficult adjustments. The parents have heavy responsibilities: accompanying the ill child to the hospital, caring for the child with side effects, doing chores at home, and earning money at work. Furthermore, the unpredictable course of the illness takes away their control of their daily lives and causes parents to struggle to adjust their family life (Clarke-Steffen, 1993, 1997; McGrath, 2001; Woodgate & Degner, 2004). Siblings are also affected by the diagnosis and experience various family life changes. At least one of the parents stays at the hospital with the ill sibling, and healthy siblings frequently stay home alone or stay with their relatives (usually grandparents), separated from their ill sibling and parents (Harding, 1996). The separation can be quite lengthy, as children with cancer in Japan are generally hospitalized for longer periods compared with those in most Western countries. These changes may trigger numerous difficulties for the siblings in their daily lives and family relationships, affecting their health-related quality of life (HRQOL).
Siblings of Children With Cancer Many previous studies dealing with the effects of childhood cancer on siblings have focused on the siblings’ psychological functioning by investigating such aspects as behavioral problems, depression, anxiety, loneliness, and guilt (Alderfer et al., 2010; Buchbinder et al., 2011; McGrath, 2001; Nolbris, Enskar, & Hellstrom, 2007; Patterson, Holm, & Gurney, 2004; Sloper, 2000; Woodgate, 2006). The majority of previous studies did not find indications of
Downloaded from jfn.sagepub.com at GEORGIAN COURT UNIV on May 3, 2015
Kobayashi et al.
121
obvious depression or anxiety in siblings of children with cancer. However, some studies have reported more depression among siblings of survivors of childhood cancer (Buchbinder et al., 2011), and have revealed adjustment problems, loneliness, guilt, and shock in young siblings (McGrath, 2001; Nolbris et al., 2007; Patterson et al., 2004). Other studies have described siblings’ school and social functioning by investigating school problems, problems of social relationships, and poor academic achievement (Alderfer & Hodges, 2010; Houtzager, Grootenhuis, Hoekstra-Weebers, & Last, 2005; Lahteenmaki, Sjoblom, Korhonen, & Salmi, 2004). Still other studies have revealed physical problems such as sleep deprivation and eating problems (Zeltzer et al., 1996). In addition, a few studies have reported symptoms of posttraumatic stress as a long-term effect after the completion of a sibling’s cancer treatment (Kaplan, Kaal, Bradley, & Alderfer, 2013). It is evident from the research that siblings experience comprehensive difficulties related to psychological, social, and physical functioning. HRQOL is a multidimensional concept that comprises elements of physical, functional, social, and psychological health, as well as the individual’s perceived health status and well-being (de Hann, Aaronson, Limburg, Hewer, & van Crevel, 1993). A few studies have reported HRQOL of siblings of children with cancer, and Houtzager et al. (2004) reported poor HRQOL of siblings compared with other children. HRQOL is affected by social and cultural factors (Franco et al., 2012). However, as far as we know, there is no research reporting the HRQOL of Japanese siblings of children with cancer.
The Sibling–Parent Relationship In studies by Sloper (2000) and Woodgate (2006), siblings reported closer family relationships as a result of the process of coping with an illness situation; nevertheless, they experienced losses of attention, family routine, security, and a sense of well-being within their families. Similarly, McGrath (2001) found that parents experienced role conflict between responsibility for the sick child and caring for siblings at home, and they had to relinquish the roles of nurturing the siblings and maintaining the home and daily family life. Although parents recognize that their family cohesion is strengthened through the experience of one child’s cancer, at the same time, they feel guilt for neglecting their other children (Brown, Pikler, Lavish, Keune, & Hutto, 2008; D. S. Cohen, Friedrich, Jaworski, Copeland, & Pendergrass, 1994). D. S. Cohen et al. (1994) administered the Child Behavior Checklist to primary caregivers of children with cancer to investigate variables related to sibling adaptation and found that high levels of family cohesion and adaptability were associated with better adaptation of siblings. These studies indicate that
Downloaded from jfn.sagepub.com at GEORGIAN COURT UNIV on May 3, 2015
122
Journal of Family Nursing 21(1)
the interrelations among family members affect siblings’ well-being. Also, an association between social support and more positive emotional, behavioral, and academic achievement has also been reported (Alderfer & Hodges, 2010). However, the relationship between siblings’ self-reported HRQOL and family functioning and social support, and the interrelations of siblings’ and parents’ experiences has, to date, not been thoroughly investigated. Most previous research on siblings of children with cancer has been conducted using child mono-informant or parent proxy-reported data. Even in studies in which several family members acted as informants, each family member’s data were analyzed separately. Family research is conducted from a family-unit perspective, and measures from multiple family members are considered necessary to capture the complexity of the family system (Cromwell & Olson, 1975). HRQOL is a measure of an individual’s perception, and proxy evaluation of children’s HRQOL is influenced by the informant’s mental health and well-being (Cremeens, Eiser, & Blades, 2006; Kobayashi & Kamibeppu, 2011). Therefore, the HRQOL of siblings of childhood cancer patients should be evaluated using the siblings’ self-reported measures. At the same time, family functioning is an index of the family unit as a whole, and a multiple-informant quantitative study is indicated. Although a quantitative approach is utilized for research that aims to predict, control, and explain family phenomena (Moriarty, 1990), quantitative research on family nursing is limited in its ability to create data that will reflect the family as a unit (Fisher et al., 1985). Furthermore, the findings of quantitative researches can reveal the magnitude and spread of the problem (Lavera et al., 2008) but fail to capture the richness and complexity of contextual factors (Patton, 1990). Qualitative research can overcome these weaknesses by providing a “human context” behind the numbers, even if such research has limited ability to assess the extent of the problem (Lavera et al., 2008). In addition, the results of quantitative research and null-hypothesis significance testing are sensitive to sample size and have limited ability to determine the practical significance of statistical relationships. Methodological problems, therefore, arise because of the low prevalence of childhood cancer. Moreover, small sample sizes in previous research (McGrath, 2001; Nolbris et al., 2007) have not always included siblings and multiple family members. The strength of a qualitative approach is its power to generate rich data about families’ subjective experiences and relationships (Moriarty, 1990), thereby allowing sibling–parent interrelations to be more richly described. Similar to quantitative research of sibling–parent interrelations, it is necessary in qualitative research to investigate experiences that are described by multiple informants to understand both the siblings’ and parents’ perspectives.
Downloaded from jfn.sagepub.com at GEORGIAN COURT UNIV on May 3, 2015
Kobayashi et al.
123
Therefore, this exploratory study, utilizing multimethods to overcome the weaknesses of single-method research and provide more complete data (Creswell & Plano Clark, 2006), used multiinformant methodologies to investigate the interrelations between siblings’ HRQOL and parents’ family functioning in families with a child diagnosed with cancer. We conducted quantitative and qualitative research concurrently, using quantitative research methods along with qualitative research methods to complement, explain, and expand the quantitative findings. The quantitative research was conducted to clarify the HRQOL of siblings of children with cancer and the correlation between siblings’ HRQOL and family functioning and social support, whereas the qualitative research was conducted to describe how siblings’ experiences and parents’ experiences interrelate to maintain siblings’ HRQOL within the family context.
Purposes of the Study Our specific aims were (a) to clarify the self-reported HRQOL of Japanese siblings of children with childhood cancer and family functioning from the siblings’ and parents’ perspectives (quantitative), (b) to clarify the state of agreement or disagreement in siblings’ self-reported family functioning and HRQOL and parents’ reports of family functioning and proxy-reported sibling HRQOL (quantitative), (c) to examine interrelations between siblings’ HRQOL and family functioning and social support by utilizing multiinformant data (quantitative), and (d) to describe how sibling and parent experiences were interrelated by using a multiinformant qualitative study (qualitative).
Method Research Design A cross-sectional, multimethod study was conducted in Japan using multiinformant data. Prior to the data collection, ethical approval was obtained from the institutional review board of Kobe University, with which two study authors are affiliated.
Sample Healthy siblings of children who were outpatients of a pediatric oncology clinic in Japan and their mothers and fathers were invited to participate. The inclusion criteria were (a) siblings who were 8 to 18 years old; (b) siblings who lived with a brother or sister who had been diagnosed
Downloaded from jfn.sagepub.com at GEORGIAN COURT UNIV on May 3, 2015
124
Journal of Family Nursing 21(1)
with cancer; and (c) siblings, mothers, and fathers who were not physically or mentally impaired at the time of the study. All families who met the inclusion criteria were approached during a seven-month period from 2009 to 2010. A clinic physician gave an explanation sheet about the study to all parents of children with cancer who had siblings when they accompanied children with cancer to the pediatric oncology clinic. Parents who were interested in the study visited one of the researchers (first author) in a private room within the clinic. The researcher explained the purpose of the study by means of a written explanation. If the parent agreed to participate, a questionnaire for parents and siblings were given to the parents (quantitative). At this time, parents were asked whether they were willing to be interviewed (qualitative); if they were, their telephone number was obtained. After reconfirmation of agreement of parents and siblings to participate in the interview, the date and place of the interview were arranged according to the family’s preferences. For ethical reasons, parents’ positive or negative decisions to visit the researcher and to participate in the research were not reported to the physician. A total of 18 families were approached, and 14 families participated in the quantitative study (a response rate of 77.8%). Among the 14 families, 13 siblings participated; 7 were female (53.8%), and the mean age of the siblings was 12.8 years (range = 8-18 years). The mean time since discharge of the ill child was 26.7 months (range = 3-35 months), and children had remained in the hospital for a mean duration of 10.8 months (range = 5-25 months). Of the 14 ill children, 6 had undergone treatment for leukemia, 4 had been treated for solid tumor, and 1 for lymphoma (data missing for 3 children). Of the parents, 12 mothers and 10 fathers returned their questionnaire sheets. For the qualitative study, 4 of 13 families were interested in participation (totaling 11 family members; 4 siblings, 4 mothers, and 3 fathers). All were families of four: two parents, the ill child, and the healthy sibling. Regarding the siblings’ relationships to the ill children, one was a younger sister, one an older sister, and two were older brothers.
Quantitative Study Measures The Pediatric Quality of Life Inventory (Japanese version; PedsQL-J). Informants: siblings, mothers, and fathers. The Japanese version of the Pediatric Quality of Life Inventory 4.0 Generic Core Scales (PedsQL-J; Kobayashi & Kamibeppu, 2010) was used to assess sibling HRQOL. Both the self- and proxy-reports comprise 23 items covering four subscales: physical, emotional, social, and
Downloaded from jfn.sagepub.com at GEORGIAN COURT UNIV on May 3, 2015
Kobayashi et al.
125
school functioning. Each item is scored on a five-point Likert-type scale from 0 (never) to 4 (almost always). The scores are calculated by using a linearly transformed, reversed score range from 0 to 100 (0: 100, 1: 75, 2: 50, 3: 25, 4: 0). Higher scores indicate better HRQOL. Scale scores are calculated as the sum of the items for that scale divided by the number of items answered; however, when more than 50% of answers for the scale are missing, the scale score is not calculated (Varni, Seid, & Kurtin, 2001). The psychometric properties of the Japanese version of the PedsQL have been empirically tested (Kobayashi & Kamibeppu, 2010; Kobayashi, Okano, & Hohashi, 2011); the Japanese version has acceptable internal consistency (Cronbach’s α = .85 for children’s self-reports and .92 for parents’ proxy-reports) and good clinical validity in differentiating HRQOL among different health statuses. Social Support Survey for Children (SSSC). Informant: siblings. Generally, the concept of social support is defined as having three dimensions—social embeddedness, perceived support, and enacted support (Barrera, Sandler, & Ramsay, 1981). Perceived support is most important for children because their social network is generally limited compared with adults, and they depend for enacted support on the adults who are closest to them, such as parents or teachers (S. Cohen & Hoberman, 1983). In this study, perceived social support was assessed using the SSSC (Shimada, 1998), originally developed as a Japanese-language instrument. The SSSC contains five items such as “He/she cheers you up when you are depressed.” For each item, respondents give their perceptions of the degree of support they expect from five resources: father, mother, siblings, teachers, and friends. A four-point Likert-type scale (from 1, definitely no, to 4, definitely yes) is utilized, with higher scores representing a higher expectation of support from a particular source. The psychometric properties of the SSSC have been tested, and satisfactory Cronbach’s α for each resource (ranging from .90 to .97) and a well-validated factor structure were revealed (Shimada, 1998). Family Adaptability and Cohesion Evaluation Scale (FACESKG). Informants: siblings, mothers, and fathers. Family functioning was assessed by using the Japanese version of the FACESKG (Tachigi, 1999). The FACESKG is a selfreport questionnaire that is based on the Circumplex Model of Marital and Family Systems (Olson, 1991). This model consists of two dimensions of family systems: cohesion and adaptation. Cohesion is defined as “the emotional bonding that family members have towards one another” (Olson, 1989, p. 9). Adaptability is defined as “the ability of a marital or family system to change
Downloaded from jfn.sagepub.com at GEORGIAN COURT UNIV on May 3, 2015
126
Journal of Family Nursing 21(1)
its power structure, role relationships, and relationship rules in response to situational and developmental stress” (Olson, 1989, p. 12). The model identifies four types of cohesion (disengaged [very low], separated, connected, and enmeshed [very high]) and four types of adaptation (rigid [very low], structured, flexible, and chaotic [very high]). A four-point Likert-type scale from 1 (never true) to 4 (always true) is utilized. To describe family functioning, we identified the types for each dimension as well as calculating the sum of the score. The FACESKG showed good internal consistency (alphas of each dimension for each family member ranged from .72 to .89) and good clinical validity in differentiating among different health statuses (Tachigi, 1999). Feetham Family Functioning Survey–Japanese (FFFS-J). Informants: mothers and fathers. The Japanese version of the FFFS-J (Hohashi, Maeda, & Sugishita, 2000), originally developed by Roberts and Feetham (1982), was used to assess parental perceptions of family functioning. The FFFS-J is based on the Family Ecological Model (Bronfenbrenner, 1977) and assesses functioning not only of intrafamily relationships but also of interfamily relationships, including those with subsystems (e.g., neighborhood) and the broader society according to the definition of the Family Ecological Model (Hohashi, Honda, Hiratani, & Feetham, 2008). The FFFS-J is a self-administered questionnaire consisting of 27 items divided into two sections: 25 family-function items and two open-ended questions. In this study, we analyzed the 25 family-function items only. These questions ask about family functioning in three areas: the relationship between the individual and family members (relationship with family members; 10 items), relationship between family and subsystems (eight items), relationship between family and society (six items), and one item that does not belong to any of the three areas. For each item, respondents are asked to answer three questions: (a) “How much is there now?” (b) “How much should there be?” and (c) “How important is this to you?”—on a seven-point Likerttype scale from 1 (little) to 7 (much). The absolute value of the difference between the responses to the score for (a) and the score for (b) is calculated for each family function (d score). The d score represents the discrepancy between the perceived actual family function and the perceived expected family function, and a higher d score indicates greater perceived dissatisfaction with a function (Hohashi, Honda, et al., 2008; Roberts & Feetham, 1982). The overall family-functioning score (total d score) is calculated by summing the item d scores, and three area d scores are calculated by summing the items within each area and dividing by the number of items answered. Cronbach’s α for the three areas and the total d score of the FFFS-J were .66, .77, .82, and .80, respectively.
Downloaded from jfn.sagepub.com at GEORGIAN COURT UNIV on May 3, 2015
Kobayashi et al.
127
Analysis. Descriptive statistics for each measurement were calculated. The means of the PedsQL-J scores were compared with published Japanese general population norms (Kobayashi et al., 2011) using the Mann–Whitney U test. Mean values for norm data for healthy children were extracted from a previous study (Kobayashi et al., 2011), and analyses showed no differences in mean ages of the children (Mann–Whitney U test) or frequency of gender (chi-square test) between norm data and the present study. Also, effect sizes (Cohen’s d) were calculated by dividing each mean score difference by the standard deviation of the general population norm (Kobayashi et al., 2011). Although null-hypothesis significance testing is sensitive to sample size and has limited ability to determine the practical significance of statistical relationships, effect sizes estimate the magnitude of an effect or the association between two or more variables and are resistant to sample-size influence. Therefore, in gauging the practical importance, effect size provides an accurate measure of the magnitude of effect between variables (Ferguson, 2009). Effect sizes were interpreted according to J. Cohen (1988) as small (.20), medium (.50), and large (.80). Furthermore, distributions of siblings’ and parents’ perceived HRQOL compared with percentiles of the general population were calculated. The agreement levels of the PedsQL-J between siblings and parents were examined with intraclass correlation coefficients (ICCs) and mean difference testing using the Wilcoxon signed-rank test and Spearman rank order correlation (ρ). The effect sizes of the ICC and ρ were categorized as small (.5; J. Cohen, 1988). The magnitude of the discrepancies between sibling and parent PedsQL-J scores was examined by calculating the means of the absolute parent–sibling differences. In addition, the means of the directional differences (parent proxy-reported scores minus sibling self-reported scores) were determined. The directions of the differences can indicate the bias of the parent proxy-reported scores relative to the sibling self-reported scores, and negative differences indicate parental underestimation. The mean values of the directional differences were standardized (d) by relating a given score to its standard deviation (SD) to determine the statistical magnitude of any observed systematic biases, which were labeled negligible (
