Good Adherence Moderate Adherence Poor Adherence
Both, CyA levels and diary completion were satisfactory
CyA levels were satisfactory, but diary completion was unsatisfactory
CyA levels were unsatis factory regardless of diary completion
Reliabilitv-Results: Child’s Adjustment and Family Measures
Child’s Mental Status (Rutter & Graham 1968).
All psychiatric interviews were audio-recorded and the child's mental status was rated (no abnormality/psychiatric illness) separately by a child psychiatrist (BL) and the author. Cohen's kappa for inter-rater reliability was 0.90. When disparity was found between the two raters consensus was reached by discussion.
Child’s Global Assessment o f Functioning (GAF Scale).
The scale has shown high inter-rater reliability, test-retest stability, and discriminant validity for children and adolescents (Shaffer et al 1983; Steinhausen 1987). The child psychiatrist and the investigator scored the Child's Global Assessment o f Functioning independently. The mean score was calculated when the ratings between the two raters differed by ten points or less. If the difference between the raters was more than ten points consensus was reached by discussion. The intraclass correlation coefficient between the ratings was 0.73.
All interviews were audio recorded, transcribed and subsequently rated by two independent raters. The reliability between the raters’ ratings was established. As EE ratings are on a quantitative scale, an analysis of variance (ANOVA) intraclass correlation approach was thought to be desirable (Bartko and Carpenter 1976; Dunn 1995, personal communication). The following coefBcients were obtained regarding parental attitudes to the child: Overinvolvement Warmth Hostility Positive Remarks Critical Comments Mother 0.91 Mother 0.93 Mother 0.95 Mother 0.88 Mother 0.97 Father 0.89 Father 0.95 Father 0.98; Father 0.76 Father 0.86
Prior to entering the scores in the computer, both raters discussed those ratings in which there was a disagreement, and consensus was reached by discussion.
During the pilot study conducted prior to the current research, the raters encountered some obstacles regarding the ratings o f two scales, overinvolvement and warmth. This was discussed at length with an expert in the Expressed Emotion field. Dr. Christine Vaughn, and consensus about the criteria for rating these scales was reached prior to the collection of data o f the present study. The difficulties were due to the following factors:
1) At the time o f the first interviews, and in order to survive, many of the children had to be pampered and were completely dependent on their parents. Parents' own needs had to be 'on hold' in order to accommodate the child's needs. Sometimes it was difficult to assess to what extent parents' self-sacrificing behaviour was appropriate or exaggerated given the child's physical condition.
2) Some parents' emotional life was in a turmoil due to the fact that their children were in a pre-terminal state and they had to continue all their efforts to keep their child alive at the same time as having to come to terms with the likelihood o f death in the face o f progressive deterioration. This situation sometimes raised inner conflicts in the parents and ambivalence
mechanisms, exaggerated emotional responses and their attitudes towards the child.
3) A significant number o f these children were initially misdiagnosed and the parents' concerns about the child's health had not been taken seriously. At the time o f the initial interview, parents had not ventilated their feelings o f anger and fiustration related to their earlier experiences o f the child's illness, and those negative feelings were very vivid in their minds during the interviewing process.
4) Some parents felt ambivalence towards the demands o f the child and his/her illness on them. This ambivalence manifested itself in a mixture o f emotions such as overwhelming feelings o f anger, denial, desperation and guilt about having given birth to an ill child and about having negative feelings regarding the child's condition and care. At the same time, parents had strong feelings of warmth, sympathy, care, and love. In addition, they were desperate to keep their disabled child alive. Understandably, parents had difficulties integrating all those feelings. Some parents coped by denying the existence o f negative feelings and "idealising" the child. In these situations, it was difficult to distinguish between warmth and idealisation o f the child. All these factors together made it difficult to rate warmth, and to judge to what extent parental devotion, self-sacrificing behaviour (putting the children's need ahead of their own), protectiveness and strong emotional responses were appropriate given the child's condition.
In order to rate overinvolvement, the patient's age and his/her degree o f disability were taken into account. This applied to both the initial assessment and the follow-up. Individual parents' behaviours and emotions were judged against what was considered to be appropriate for the child's age, in the context o f the child’s physical status. For example, parental high involvement with a 15 years old (i.e. carrying him to the toilet, bathing him, feeding him, etc.) may be adaptive if the child is severely disabled and unable to do things by himself, but that same interaction style may become smothering after successful treatment, if the child is not disabled and has the ability to become more age-appropriate independent.
Attitudes towards the Partner (The Expressed Emotion Partner Interview Schedule EEPIS))'. As with the CFI, all interviews were audiotaped and scored by two raters. The following intraclass correlation coefficients between the two raters were obtained regarding parental attitudes to their partner:
Positive Remarks Mother 0.81 Critical Comments Mother 0.91
Father 0.96 Father 0.95
The high level o f agreement across the raters, and the discrimination o f these scales between well and moderate-poorly adjusted families (see TABLE XXV, page 115) at T l are evidence o f the concurrent validity o f this new instrument. Furthermore, significant associations were found between warmth, critical comments and positive remark ratings at T l and marital adjustment (GRIMS) at T l.
-W arm th: mothers: r=-. 62, p<. 001; fathers - Critical Comments : mothers: r=.35, p<.001; fathers - Positive Remarks: mothers: r=-.24, p=.01; fathers
r=-.71, p<001 r= .27, p<001 .31, p=.001
In order to explore whether these scales were predictors o f fixture marital breakdown (predictive validity) a series o f X^ were conducted between the EE variables at T l and the marital scale at T2. The variables were categorised as follows: Warmth: none-some/moderate- high; Positive remarks: present/absent; Critical comments: present/absent; Marital relationship: very good-average/poor-severe problems. Results showed that the scales at T l discriminate between good and poor marital relationship (GRIMS) at T2:
- Warmth: mothers: X^=13.1, df=2, p=.001; fathers: X^=12.2, df=2, p=.002, - Positive Remarks: mothers: X^= 5.3, df=l, p=.02, - Critical Comments: mothers: X^= 4.4, df=l, p=.03
Clinical Ratings on Family Adjustinent (Schneiderman categories):
Using a clinical protocol, as part o f the transplant programme routine assessment, families of children who were referred for transplantation were also evaluated by a child psychiatrist, nurse specialist, liaison sister or social worker. Following their assessment, the clinicians rated family adjustment using the same global scale as the investigator. The agreement between the clinicians and the investigator on this rating was 0.91 (intraclass correlations).
Analyses of Data
The statistical package for the social sciences (SPSS) programmes were utilised on an IBM computer system. Although non-parametric statistical tests have been used, parametric statistical methods were mainly used for the analysis as these are more powerful and can take account o f more complex data structures. The data have been analysed as follows:
i) Differences in demographic characteristics, family organisation, children’s and parents’ adjustment between the groups presenting different psychosocial typology o f illness (hypotheses 1 and 2):
A series o f descriptive statistics, Chi-square tests and One-way analyses o f variance (using an a priori contrast method), were conducted to identify differences in the demographic and psychosocial characteristics between the following five specified group combinations (see FIGURE 7) - in all contrasts children and family characteristics were compared.
With multiple tests o f significance, the probability o f making at least one type I error is higher than 1 in 20 (Fleiss 1986). To overcome the problem, the p value corresponding to the required level o f significance can be lowered using the Bonferroni multiple comparison adjustment. Because this study is exploratory, attempts to control for such an error were not made; this means that extra caution is needed in interpreting the results.
Figure 7: Specified Group Comparisons
Group Comparisons
Referred for Transplantation (Tx)
Admitted for Conventional Cardiac Surgery (Conv Surg)
Heart Tx Heart-Lung Tx Low-risk (LR) High Risk (HR)
HTx HLTx PvHL
Contrast 2: Heart transplant active list (HTx) Vs Heart-lung transplantation active and provisional list (HLTx + PvHL).
Contrast 3: Heart transplantation active list (Htx) Vs Heart-lung transplantation active list