Healing Children's GriefSurviving a Parent's Death From CancerChapter 4 - Expanded STUDY SAMPLE, INTERVENTION, METHODOLOGYSAMPLE
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| TABLE 4 - 1: DEMOGRAPHIC CHARACTERISTICS OF PARENTS | |||
| Mothers | Fathers | ||
| AGE: |  |
||
| > 40 | 33% | 24% | |
| 40 - 49 | 63% | 63% | |
| 50+ | 4% | 14% | |
| RELIGION | |
||
| Catholic | 52% | 45% | |
| Protestant | 22% | 21% | |
| Jewish | 18% | 20% | |
| Other | 8% | 14% | |
| EDUCATION | |
||
| < HS | 6% | 5% | |
| HS Graduation | 23% | 23% | |
| College | 56% | 48% | |
| Post-graduate | 15% | 24% | |
| MEDIAN FAMILY INCOME: $53,000 | |
||
| RACE | |
||
| White | 80% | ||
| Hispanic | 7% | ||
| Black | 7% | ||
| Mixed | 5% | ||
| Asian | 1% | ||
Single-parent families were not recruited for this study because the stresses and responses of children who lose their only parent are quite different from those faced by children who have a surviving parent. For the same reason, families in which a parent died from AIDS or from unexpected causes such as homicide, suicide, or accident were not recruited (Pynoos, Steinberg, & Wraith, 1995). Although these restrictions reduced the generalizability of the findings, the NIMH reviewers agreed with us that at this early stage of knowledge and understanding of the effect of this family tragedy, the utility of these findings to other researchers would be best enhanced with this more restricted population.
CHILDREN'S DEMOGRAPHIC CHARACTERISTICS: Of the 157 children, 79 were girls and 78 boys. There were 19 in the pre-school cluster, and 32 to 38 in each of the four older age clusters. The methods used to assign children to these clusters are described below under Developmental dimension. The children,s characteristics are summarized in Table 4.2.
| TABLE 4 - 2: CHILDREN AND ADOLESCENTS IN THIS SAMPLE | |||
| AGE | |||
| 3 - 5: | 11 girls | 7 boys | 18 children |
| 6 - 8: | 16 girls | 16 boys | 32 |
| 9 -11: | 18 girls | 19 boys | 37 |
| 12 -14: | 21 girls | 17 boys | 38 |
| 15 -17: | 13 girls | 19 boys | 32 |
| TOTALS: | 79 girls | 78 boyx | 157 children |
INTERVENTION DESIGN
The psycho-educational intervention was developed to facilitate the adjustment of children to the terminal illness and subsequent death of a parent,
emphasizing a parent-guidance approach. More specific details about this intervention have been published (Christ, Siegel, Mesagno, & Langosch, 1991; Siegel, Mesagno, & Christ, 1990). A
telephone supportive intervention was also developed as a control condition (Christ, Raveis, Siegel, & Karus, 2000). For this intervention a social worker telephoned the well parent every four to
eight weeks. The goal of this intervention was to maintain contact with the well parent between psychological evaluations, to provide referrals to community based therapists or support groups when
such a referral was requested, or to appropriate hospital personnel when questions such as uncertainty about planned treatment procedures, billing, or untoward reactions of the ill parent were raised
by the well parent. Since the data generated by this intervention were insufficient for qualitative analyses, only data from families who participated in the psycho-educational intervention were used
for the qualitative arm of the analyses.
The interventions were planned to start during the terminal illness. This decision was based on clinical experience. We found that the family member,s responses differed substantially during the terminal stage of the illness from responses following the death. This clinical experience was confirmed by the quantitative analyses of depression and anxiety measures that indicated that children were significantly more anxious and depressed during the pre-death period than at the end of the reconstitution stage (Siegel, Karus, & Raveis, 1996; Siegel et al., 1992).
A typical psycho-educational parent guidance intervention spanned about 14 months and included six or more 60- to 90- minute therapeutic interviews during the terminal stage of the illness and six or more after the death. The sequence of interviews, most of which were carried out in the family home, included one or two initial interviews with the well parent, depending on the urgency created by the expected survival time of the ill parent. The first interview emphasized background data about each parent and child and about the cancer illness, the second emphasized the therapeutic engagement. These interviews could include the patient if so desired. Possible ways of handling emerging problems, including problems with the children, were discussed at each meeting. A separate interview was then done with each child. This was followed by an informing interview with the parent(s) in which the parent(s) was given an assessment of the children,s adaptation to the illness. A family interview that included the well parent and all of the children in that family was then done. This was followed by two or more bi-weekly to monthly parent interviews. A similar schedule of interviews was followed starting 2 to 4 weeks after the death of the parent. When appropriate and requested, additional child and/or family interviews were scheduled. After the final interview, the social worker initiated bi-monthly to monthly telephone contacts with the surviving parent until the final post-death psychological assessment was completed about 14 months after the death of the parent. If significant family crises emerged either during the psychologist,s final assessment or the social worker,s final telephone contact, additional telephone contacts were scheduled, and, if necessary, individual parent, child and/or family sessions were offered.
Interviewer training and supervision was an integral part of the project. Each of the social workers who were part of the intervention team had at least five years' post masters experience working in a medical or mental health setting. The social workers received extensive written and verbal descriptions about the goals of the intervention, their role in meeting these goals, the types of material that would be elicited from children and parents, and the type and quality of notes that would be kept after each scheduled and unscheduled contact. Ongoing training was provided through weekly individual and group supervisory meetings. The author coordinated and supervised the clinical program.
The analysis of each child,s experience was guided by what William Runyan (1982 p. 100) described as "STAGE-STATE analysis of the life course." (To
reduce confusion, when the words STAGE or STATE are used to refer to Runyan's STAGE-STATE analysis of the life course, they will be capitalized.) Runyan suggested that the temporal approach of a
life-course orientation is more effective in analyzing situations that require one to consider the interaction of personal, behavioral, and situational variables over time. The life-course
orientation is presented as an alternative to other orientations that focus primarily on how traits, psycho-dynamic processes, situational influences, or all of these impact on affects, symptoms and
behavior. The predictive limitations of these other approaches have been recognized (Mischel, 1968; Mischel, 1973). The STAGE-STATE analytic approach provides "a method . . . for investigating the
alternative routes or sequences of processes through which initial person-situation configurations may be linked with an array of potential outcomes" (Runyan, 1982, p. 101). In this approach, STAGE
is defined broadly to indicate periods in a process (e.g., stages of the illness). STATES within a STAGE are "specific kinds of persons behaving in particular ways in particular social and historical
circumstances" (Runyan, 1982, p. 102).
Under each of the STAGES defined below, the related behavior-person-situation determining processes that might contribute to the child,s adjustment were identified. There were three sources for identifying potential person-situation configurations: clinical knowledge of the demands of the illness experience on patients, children and their families, significant variables from the quantitative analyses of our own and other published findings, and the data in the family files. The sequence of processes or interactions (" behavior-determining processes, . . . person-determining processes, . . . situation determining processes" ; Runyan, 1982, p. 84) that link the potential themes with the adaptive efforts of the children and the parents were sought.
DATA SOURCE: Two hundred twelve (72%) of the 275 families who met all eligibility criteria agreed to participate and to be randomly assigned to one of two interventions. One hundred eighty-four, (87%) completed one of the interventions, and of these, 104 (56%) completed the initial and the second and/or third psychological evaluation. After eligible families were informed that they would be randomly assigned to one of the two interventions that were described to them, the well parent agreed to participate, signed the consent forms, and their children gave their assent, two psychologists from the evaluation team scheduled a meeting with the family to administer a battery of tests and conduct an interview with each child and with the well parent. (These assessments were repeated 8 and 14 months after the death of the ill parent). After the psychological assessment was completed, the family was randomly assigned to the psycho-educational intervention (75%) or to the telephone supportive intervention (25%). The family was then assigned to one of a group of social workers who, along with their supervisors, made up the clinical team. If the child had checked off one of the "thinking of self harm" questions in the psychologist,s assessment and the interventionist, in collaboration with the parent, assessed this to be a possible problem, an appointment with a child psychiatrist was offered as a part of the intervention program.
When permitted, all interviews were audio-taped. The data for analysis relied primarily on open ended therapeutic interviews. The looser framework of such unstructured interviews places greater demands on the researcher to find important instances of cross-case comparability. "In multiple-case research ... the looser the initial framework, the more each researcher can be receptive to local idiosyncrasies - but cross-case comparability will be hard to get, and the costs and the information load will be colossal" (Miles & Huberman, 1994, p.17). Since we knew that cross-case analyses would be important in this study, we felt that a tight framework through the development of semi-structured forms would enhance cross-case comparability.
The social work interventionist filled out a lengthy 10-14 page semi-structured form following each interview. Different forms were developed for each
pre-death and post-death interview that reflected the intent of that interview. These ten forms and their sub-headings were developed as the intervention was being developed, then modified during the
pilot part of the study (Christ et al., 1991; Siegel et al., 1990). The intent of the forms was not to serve as headings for a semi-structured interview, but rather to remind interviewers about the
general topic areas that should be covered and to serve as a pedagogic guide for the interventionists. Table 4.3 lists the ten forms. As indicted in the table, a number of them might require two or
more sessions to cover that material. For example, one major heading in the child,s post death session form (Form 7) is titled "evaluation", and has 13 sub-headings (e.g. "child,s
adaptation", "school performance and behavior", "coping strategies" ), some with additional sub subheadings (e.g., "process of being informed", "changes to the family life style"). Another example is
the final pre-death parent interview form (Form 6) titled "Preparation for death and issues related to loss", which has 15 subheadings (e.g. "discuss severity of parent,s illness", "explore what
each child knows and believes about the parent,s death", "educate parent about value of children,s participation in rituals such as wake, funeral, burial"). These notes were used to aid in
the supervisory process, but were also intended to facilitate the qualitative analysis. Detailed written notes were kept of all telephone contacts with or about the family.
TABLE 4 - 3: THERAPEUTIC INTERVIEW FORMS
| Pre-death interview titles: | |
| Form 1a-n* | Well parent interview. Establish therapeutic alliance, obtain comprehensive history, initiate psycho-educational and supportive intervention (continued in each session). |
| Form 2a-n: | Each child - individual interview. |
| Form 3a-n: | Well parent, feedback of children's current state, discuss parental role. |
| Form 4: | Family session, well parent and all children. |
| Form 5a-n: | Well parent, preparation for death and issues related to loss. |
| Post-death interview titles: | |
| Form 6a-n: | Well parent. Discuss death, funeral, each family member's reactions, mourning. |
| Form 7 a-n: | Each child - individual session. |
| Form 8 a-n: | Well parent, feedback and discuss parent's role. |
| Form 9: | Family session, well parent and all children, may include significant caretaker(s). |
| Form 10 (a-n): | Well parent, preparation and planning for future. |
| *1a-n indicates that the same form may be used for more than one interview. | |
DATA ENTRY: The author transcribed all interviews of 31 of the 88 families in order to assess the accuracy of written entries as compared to the transcribed material. The 31 families who had finished all planned interviews, all three of the psychological evaluations, and where each session was audible were selected for complete transcription. Additional family tapes were transcribed to insure the presence of families from each of the six interviewers, others were selected to insure that boys and girls from each age (3 to 17) were represented. Individual interviews were also transcribed when the interviewer,s notes seemed unclear or a more complete record of the verbatim interview seemed indicated.
The notes on the semi-structured forms, the transcribed interviews, the psychologist,s pre and post death summaries, all telephone contacts, notes from teachers, and the psychiatric evaluation reports were prepared for entry into the Ethnograph program for computer-assisted analysis (Seidel, Kjolseth, & Seymour, 1988). To facilitate retrieval of information, a problem in any very large data base (Huberman & Miles, 1998, see especially pp 182-184) , a separate file was created for each family. The entries were entered sequentially by date, and the title of each entry included the date of the contact, the type and site of contact, and the names of all participants. Table 4.4 summarizes the forms used to keep track of all material.
TABLE 4 - 4:
| FORMS USED TO TRACK DATA | |
| a: | Informed consent forms. |
| b: | Registration form: Patient, well parent, and children information. |
| c: | Year calendar grid, for entry of sessions held, no show, cancellations, telephone calls and date of death. |
| d: | Medical record verification form (diagnosis and procedure summary). |
| e: | Research interview schedule, psychologist interview evaluation form, psychologist interview summary. |
| f: | All telephone calls with and about family. |
| g: | Consent form for teacher to fill out the Achenbach Teacher Report Form. |
DATA ORGANIZATION: Once family files had been prepared, the first step was to write a 1 to 2 page precis of each family to serve
as a reference and as a brief reminder of important aspects of each family. The next step was to organize the data to facilitate a "data display", an important step in qualitative analyses (Huberman
& Miles, 1998, pp.180-182). Eighty eight families with 176 parents and their 157 three to seventeen year old children represent a very large data base for qualitative analysis. A random sample of
families or interviews was not done because in order to develop a deeper understanding of this experience, not only within-case analyses but also cross-case and cross-cluster analyses of an unknown
number of clusters of children were indicated. Twenty to forty children in each cluster is a reasonable number for this qualitative study design (Morse, 2000).
The division of the data required logical subdivisions in order to make the results of the analyses more precise and meaningful. Further, it was necessary that such "conceptual frameworks" (Huberman & Miles, 1998) not introduce confusing distortions. Several seemingly logical subdivisions of families (by gender of parent) or clusters of children (i.e. under and over 12) were found to be unhelpful. Rather than clarifying similarities and differences in the processes or interactions that affected the children,s states ( e.g. parental warmth, understanding of children,s reactions, sensitivity to children,s needs), the overlap in the processes or interactions of interest using these subdivisions or clusters resulted in unclear and confusing results. As described below, the data was subdivided within each file by 1) illness stages and 2) by clustering files using the children,s developmental attributes. The children were also divided 3) by outcome groups.
ILLNESS STAGES: Terminal illness, death, reconstitution. Each family file was divided into four sections: History of the illness and of each family member prior to the terminal illness, and then into the three illness stages - terminal illness, death and death related rites and rituals, and family reconstitution as a single parent family.
DEVELOPMENT: Five development derived age clusters. Combining children into logical and meaningful clusters presented a more difficult problem. The use of age to indicate development has been seriously challenged. Rutter (1989a) wrote "it is necessary to appreciate that age is an ambiguous variable ..." p. 28. Speece and Brent (1996a) noted that "Age by itself explains nothing ... It is rather a convenient general, omnibus index of a wide range of loosely correlated biological and environmental variables." (p 43). Trembley & Israel (1998) warned of the problem of using age to understand children,s adaptation to parental death.
The study of the family files showed numerous examples of developmentally specific responses of children to the stresses of the parent,s illness and death. Development is a theoretically meaningful way to group children, yet the challenge to find a way to do so remained unmet. While age can serve as an index to a group of identified developmental attributes, it can not be used as an indicator of development (Rutter, 1989; Rutter, 1994; Speece & Brent, 1996a; Speece & Brent, 1996b).
To address this problem, I first searched for the developmental attributes that were relevant to the experiences of the children
through a careful search of each file in our sample. The developmental characteristics that emerged consistently were cognitive, emotional and ecological. The responses of the children to the illness
and death, their explanations about the illness and death, the content of their questions, their three wishes, their recommendations to other children were some of the types of information that were
useful in ascertaining their developmental attributes. So too were their different needs and responses to parental interventions, the type of involvement they had with peers and other significant
adults, and the level of their involvement in their ecological world. The discussions with the parents during the therapeutic interviews, when they described their efforts to understand the
children,s behavior and interact with them and the type of comfort they needed to ameliorate their stress further clarified the emotional and cognitive developmental attributes of the children.
As the narrative chapters show, the illness and death of the parent affected the whole ecological system of which the child was a part. Highly relevant was that the ecological system changed,
becoming more complex with older age clusters. Relevant also were the differences in the interactions where the child affected the members of the ecological system as profoundly as the members of
that system affected the child. The ecological theory briefly summarized in Chapter 2 (Bronfenbrenner, 1979; Bronfenbrenner, 1989; Bronfenbrenner, 1993) and the eco-system theory (Meyer, 1983) were
helpful in understanding this developmental component. The three developmental areas that were required to understand this sample are defined and discussed in Chapter 2 and amplified at the start of
each Themes chapter.
Once the three relevant developmental areas were identified, all data that was related to each child was coded to identify all examples of their emotional, cognitive and ecological developmental attributes. Given that the data in this study derives from a parent-guidance intervention, much of what was learned about the children came not only from their interviews, but also from the parent, their siblings, teachers, and from the psychological and psychiatric reports. For example, an insightful account of a 4-year-old girl,s reaction that occurred when she was told that her father had died (Chapter 6, pp 61-70) came from her 7-year-old sister. "She does not know what dead is. ... she thought something exciting had happened." (Weeks later) "she kept asking when Daddy would come back." Information about a person as told by another has been called "shadow data" (Morse, 2000). The verification of shadow data is important. The numerous accounts by her mother verified that this four-year-old was another example of a mid pre-operational child.
Several attempts to cluster children with similar developmental attributes to maximize within cluster similarities and between cluster differences resulted in five clusters. As parents, educators and mental health workers know so well, there are differences in the maturational rates of children, some based on gender, some on socio-demographic characteristics, others on innate capacities, to name only three. No effort was made to correct for slower or more rapid developers in assigning them to age clusters. Finally, each child,s age was assigned as their age the day the parent died. The ages of the children in each cluster (3-5, 6-8, 9-11, 12-14, 15-17) are an index to the best compromise of similar developmental attributes. It is quite likely that a different sample with a different mix of less advantaged and/or more troubled children would have a different age distribution that would correspond to these developmental attributes.
OUTCOME: Although most of the children were doing quite well when last seen, there were others who were not. The next step in the analysis was to identify processes or interactions that were associated with a timely and with a more compromised reconstitution in the children when last seen 8-14 months after the death of the parent as described under Second STAGE-STATE analysis below. To do so, it was first necessary to identify the children,s outcome without introducing a circularity, that is, by not using the processes or interactions that were related to different outcomes we were interested in identifying. A board certified child psychiatrist and I assessed the full record of each child,s reconstitution and made judgements about the child,s return to their pre-terminal illness level when last seen using five emotional and behavioral domains. These domains included psychological state and behavior (e.g emotional state, self esteem, internalizing and externalizing symptoms), family relationships (e.g. with well and ill parent, siblings, grand-parents, care-takers), school competence (e.g. grades) and behavior (e.g. relation to teacher and classmates), after school activities (e.g. sports), and peer relationships. The child was assessed by comparing their functioning on these five domains before the terminal illness as noted in the history part of each chart with their reported functioning at the end of our contact.
TABLE 4 - 5: CHILD OUTCOME BY GENDER OF CHILD
AND GENDER OF SURVIVING PARENT
| Surviving mothers | Surviving fathers | ||
| (N = 51) | (N = 37) | ||
| Daughters | Timely reconstitution | 21 | 16 |
| N = 68 | Delayed reconstitution | 7 | 8 |
| Compromised reconstitution | 4 | 8 | |
| Symptomatic reconstitution | 3 | 1 | |
| Sons | Timely reconstitution | 34 | 13 |
| N = 71 | Delayed reconstitution | 8 | 9 |
|
Compromised reconstitution
|
0 | 1 | |
| Symptomatic reconstitution | 2 | 4 |
We identified four outcome categories: Timely reconstitution (improved reconstitution in all five emotional-behavioral domains),
delayed reconstitution (ongoing problems in one or more domains that were improving when last seen), compromised reconstitution (ongoing and unimproved problems in one or more of the five domains),
and symptomatic reconstitution (showing evidence of the emergence of new affects, behaviors, and/or interactions after the death of the parent that were consistent with a DSM IV diagnosis). The
outcome by gender of child and gender of surviving parent is summarized in Table 4.5. No psychological evaluations were done with the pre-school children because they were not a part of the
quantitative analysis arm of this study. There was sufficient data from their interviews by the interventionists, their participation in the family sessions, and from descriptions given during the
interviews of the parents and their older siblings to understand their responses through qualitative analyses. However, since these 19 (11 girls and 7 boys) did not receive a battery of psychological
tests, nor a comprehensive evaluation by the psychologists at 8 and 14 months after the death, they were not assigned an outcome rating.
FIRST STAGE-STATE ANALYSIS: within-cluster between-case analyses: After each family file was divided into three illness stages and the children were clustered into five development-related age groups, the author coded for all death related and unrelated stressful and stress reducing processes or interactions that took place between the children, their family and the larger ecological system of which they were a part. This was greatly facilitated by the headings we had developed in the semi-structured forms under which the interventionists entered notes at the end of each session. The transcribed audio-taped interviews were compared with the notes entered by the social work interventionists, and coded to identify relevant quotes that would exemplify the categories. They were also used to identify additional interactions with significant others, emotional reactions, and types of meaningful interactions with their larger ecological world. I searched for processes and interactions identified in the literature as important (e.g. "parental warmth"), in the psychologist,s evaluations, the interventionist,s telephone calls, in notes from teachers and in the psychiatric assessments for possible suicidality when such an assessment occurred. The codes were then organized by similarity of underlying concepts. This process of conceptual mapping (Fullilove, 1998) or data display (Huberman & Miles, 1998) yielded three areas of processes or interactions: stress/support components ( positive and negative mediators and moderators), concurrent and death related occurrences, and mourning, defined as the adaptation to the death of a loved one.
Reviews of the transcripts of the children and their families supported that clustering children with similar developmental attributes was helpful in understanding children,s different responses to the illness and death in each of the development generated age clusters. Similarly, different stressors and responses were identified in each of the three illness stages. For example, the surviving parent,s ability to understand and respond to children,s reactions at different developmental stages were clarified. Many, but not all of the fathers had difficulty understanding their younger children,s thinking processes and emotional reactions, while many, again not all of the mothers had difficulty understanding and dealing with mid-adolescent boy,s need for emotional distance and independence. The responses of the well parent to the terminal illness of their spouse was highly variable, ranging from parents who devoted themselves exclusively to the patient,s care and relied on others to take care of their children to those who carefully, though often at great personal cost, balanced the needs of the children and of the dying spouse. While a few remained detached from the children during the reconstitution stage, most conscientiously sought to "make up" for this temporary abandonment. The effectiveness of the grouping by illness stage and clustering by developmental attributes was further exemplified by identifying the emergence of anticipatory mourning in the older groups, the higher levels of anxiety and depression during the terminal illness stage than after the death, the differing expressions of mourning and in the creation of a legacy during the reconstitution stage. The range of, and difference in, the positive and negative mediators and moderators and the range of different responses of the children in each of the development derived clusters and in the different illness stages also became clear. The results of these analyses are presented in the five Themes chapters, and in the concluding chapter.
SECOND STAGE-STATE ANALYSIS: Within-case analysis: The method of analysis to search for and code for all categories that were (causally) related to each
child,s outcome, that is, to the sequence of processes or interactions (" behavior-determining processes, . . . person-determining processes, . . . situation determining processes" (Runyan,
1982, p. 84) with the adaptive efforts were sought. This more focused within-case analysis facilitated finding the relationships between the more specific positive and negative mediators and
moderators as well as the important concurrent and death related occurrences in each child and in each illness stage. A discussion of the factors that were associated with each outcome in each of the
four development derived age clusters is provided at the end of the Narratives chapters for the four older clusters, and is summarized in the final chapter and in Table 4.3.
THIRD STAGE-STATE ANALYSIS: Between - age cluster analyses: This third stage-state analysis was done to clarify similarities and differences between different development derived age clusters of children. Where my previous analyses looked for similarities and differences in the experience of children within age clusters, the focus now was on identifying similarities and differences between development derived age clusters. Of interest was the way in which ongoing development affected children,s adaptive processes, the difference in how they responded to similar stressors and how different the stressors were for children in different clusters, the way different developmental attributes and need