SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. TRIANGULATION OF METHODS
  4. DISCUSSION AND CONCLUSIONS
  5. Acknowledgements
  6. REFERENCES

We describe the triangulation of qualitative and quantitative research methods used to develop and test the Adolescent Resilience Model (ARM). The differences in meaning-based and function-based health-related quality of life (HRQL) are discussed, and method triangulation is presented as a means of developing models of HRQL that represent the perspectives of the adolescent and family. Qualitative methods of phenomenology, simultaneous concept analysis, focus groups and thematic analysis were used to generate the ARM. Quantitative instrumentation and structural equation model development and testing were used to evaluate the ARM. A decision-making process for combining qualitative and quantitative research, so that both approaches are equally valued and used, is also presented. Int. J. Cancer Suppl. 12:125–131, 1999. ©1999 Wiley-Liss, Inc.

The theoretical work on health-related quality of life (HRQL) is not well developed and often is derived from a “shopping list” of concepts that lack clear identification of the theory or assumptions underlying their selection (Haase and Braden, 1998). In addition, the pressing need to measure HRQL as an outcome for children and adolescents seems to be leading to an acceptance of function-based models without a critical evaluation of the underlying assumptions for both function and meaning as the basis for HRQL assessment. Our purpose is to describe the processes used to develop and test the Adolescent Resilience Model (ARM). The ARM was developed from a meaning-based perspective. It was developed also through research triangulation, the use of multiple research approaches to enrich understanding of a phenomenon. By framing HRQL within a meaning-based perspective and by triangulating research approaches, our aim was to identify intervention foci that were based on child/adolescent and family-derived experience. Meaning-based models and triangulation may lead also to identification of intervention strategies that are important, appropriate and acceptable to the child/adolescent and family.

The two predominant types of conceptual model emerging in oncology are function-based and meaning-based (Costain et al.,1993). The major characteristics as well as the underlying assumptions of these 2 model types are quite different. The focus of function-based models is predominantly from a biomedical perspective, emphasizing maintenance of functional abilities by management of clinical concerns, toxicity of treatments and functioning across several domains. Meaning-based models focus on the patterns and experience of illness from a subjective and holistic perspective. Meanings are derived from patients' understanding of the situation and their autonomy, beliefs, choices and relationships with others. The perceptions of the child and the family about the meaning of illness, including changes in functioning, are considered essential research data within a meaning-based model.

An example may help to illustrate differences in function- and meaning-based models. When a child's or adolescent's social function is assessed from the function-based perspective by instrument items asking about ability to be involved in a social activity, involvement in the activity is assumed to be positive. Thus, when items asking if an adolescent can attend school or visit friends are answered “never” or “rarely”, a decreased quality of life in the social dimension is assumed. Missing from the items is the meaning that staying at home has for the adolescent. Remaining at home and being unable to see friends may indicate decreased HRQL for many adolescents with cancer. However, qualitative data, from our study of the experience of completing cancer treatment, indicate that at least some adolescents and families view the time at home positively, as an opportunity to be together, strengthen family ties and focus on friendships that endure in spite of isolation (Haase and Rostad, 1994). Thus, without additional items or instruments to evaluate concepts that are meaningful from the adolescent's perspective, items and instruments focused solely on functional performance would indicate poor social functioning erroneously and result in potentially misleading conclusions about HRQL.

While function-based indicators are often related directly to the nature and course of the cancer diagnosis and treatment regimen, many patients are able to “rise above” their discomforts. Physical functioning can fluctuate markedly during the course of treatment but may or may not indicate HRQL from the patient's or family's perspective because patients often reframe experiences positively that are assumed to be negative. A salutogenic or health focus on positive adjustment is gaining recognition in research on cancer and other chronic illnesses. Theories such as resilience (Rutter, 1987), hardiness (Kobasa, 1979) and self-efficacy (Bandura, 1977) are used to focus on and explain positive adjustment to illness based on the belief that such theories will yield useful information about interventions (Forsyth et al.,1984; Sinnema, 1991). Of such concepts, resilience is most often studied in children and adolescents. Resilience is defined as the process of identifying or developing resources and strengths to manage stressors flexibly and gain a positive outcome, i.e., a sense of confidence or mastery, self-transcendence and self-esteem. As a process and outcome, resilience is congruent with the goal of extending cancer treatment beyond mere survival. Resilience can lead to expanded growth potential and enhanced quality of life.

TRIANGULATION OF METHODS

  1. Top of page
  2. Abstract
  3. TRIANGULATION OF METHODS
  4. DISCUSSION AND CONCLUSIONS
  5. Acknowledgements
  6. REFERENCES

Methodological triangulation is the use of quantitative or qualitative research approaches either sequentially or simultaneously to refine, evaluate and/or extend theory. However, research using qualitative and quantitative approaches can be complementary in important ways only if sufficient thought is given to the philosophical assumptions of each and if the methods chosen are not contradictory to those assumptions (Myers and Haase, 1989). For development of the ARM, 2 general types of study were conducted to maximize the understanding gained from each approach (Table I). Six qualitative studies, referred to as Model Generation (MGen) studies, were the basis for model generation, qualitative model validation or model concepts clarification. Most of these MGen studies were conducted to obtain the adolescent perspective; other perspectives studied were those of parents and health-care providers. These studies also helped to clarify some of the more elusive concepts in the model in terms of antecedents, critical attributes and outcomes. Four quantitative studies, referred to as Model Evaluation (MEval) studies, served as the basis for developing and testing the psychometric properties of instruments to measure factors in the ARM and provide preliminary quantitative testing of the ARM. The MGen and MEval studies were done interactively, though they are listed chronologically in Table I. For example, as understanding was gained from an MEval study, findings from previously conducted MGen studies were examined in light of the new information. Likewise, data from MGen studies were used to interpret the findings from MEval studies.

Table I. CHRONOLOGY OF MODEL GENERATION (MGen) AND MODEL EVALUATION (MEval) STUDIES
StudyPurposeMethod(s)
1985 MGen 1: Components of courage in adolescents with chronic illnessDescribe experience of couragePhenomenology
1986 MGen 2: Qualitative perspectives for instrument development on courageClarify specific themes from previous studyPhenomenology
1988–89 MEval 1: Courage in chronically ill adolescents: model and instrument developmentDevelop latent variable model and instrumentsDecision trees and content validity analysis
1989–90 MGen 3: Simultaneous concept analysis of hope, spirituality, self-transcendence and acceptanceClarify elusive concepts in emerging modelSimultaneous concept analysis
1989–90 MEval 2: Pilot study of instruments and model specification for the Becoming Courageous Model (BCM)Evaluate psychometrics of BCM instruments and conduct confirmatory and exploratory model testsPsychometric testing, structural equation modeling
1990–91 MGen 4: Psycho-social support resources: focus groupDescribe existing adolescent cancer support programs for instrument developmentFocus groups
1991–94 MEval 3: Adolescent Resilience Model (ARM): instrumentation and model testingConduct exploratory measurement and structural equation model testing of the ARMInstrumentation and structural equation modeling
1993–94 MGen 5: Experiences of completing cancer therapy: 3 perspectivesEvaluate applicability of the ARM to the experience of completing cancer treatmentsComparative phenomenology
1995 MGen 6: Isolating functional performance in chronic illnessDescribe the nature and process of functional performanceThematic analysis
1996–present MEval 4: Resilience and quality of life in adolescents with cancerExamine patterns of change in resilience and quality of life and evaluate the ARM pilot interventionProspective, longitudinal descriptive study followed by quasi-experimental time lag

MGen studies

To discover the meanings adolescents with cancer ascribe to their experience, 2 studies (MGen 1 and MGen 2) were done using a phenomenological approach (Giorgi, 1970). While it is beyond our scope to describe phenomenology fully from a philosophical or methodological perspective, a brief description of the method is provided here. In phenomenology, data are the raw descriptions provided by individuals who have experienced the phenomenon. The data are obtained usually from open-ended interviews using a single data-generating question that asks the participant to “please tell me about your experience of (topic of study)”. The goal is to obtain a rich narrative of the experience with as many details as possible but without leading the participant by asking direct questions.

Data in phenomenological studies are analyzed through a systematic step-by-step process (Colaizzi, 1978). The 1st step is “bracketing”, which involves reflecting on, explicitly stating and then putting aside one's own theoretical ideas about a phenomenon. The 2nd step is listening reflectively to the audiotaped interviews to gain a feel for their meanings as whole stories. The 3rd step is extracting statements (words, phrases or sentences) that are important. To illustrate, the following statement by a 17-year-old cancer survivor regarding the future was extracted: “Twenty years down the line when they know that I'm cured . . . you (I) can say ′I had cancer'. And then you (I) talk about it like old people do at a bar or something, but, you know, not now”. In the 4th step, a restatement is developed. This is done to move incrementally to the language of science and to ensure that meaning is derived through progressive subjectivity rather than intuitive leaps. For the above example, the following restatement was derived: “Twenty years in the future, when people know that one is cured, one can tell others about having had cancer and talk about it as old people might discuss the past at a bar or other gathering place. But, the cancer experience cannot be discussed now”. In the 5th step, a meaning for the restatement is formulated, taking into consideration the full context of the participant's experience. For the adolescent's statement above, the following meaning was derived: “A long time is required after cancer treatments are over to believe that one is finally cured. One can envision sharing the cancer experience with others when older, much like old war stories, but the memories of the cancer experiences are too painful to share casually with others at this time”. The 6th step is to derive themes from the formulated meaning within and across different participant interviews. For the example, themes included Waiting and Hoping for the Cure, Comfort in Talking about the Experience as Anticipated in the Future and Cancer Memories are Painful. These themes are sorted into clusters and categories. From the list of clustered and categorized themes, an extensive narrative description of the experience is developed. Next, the narrative description is distilled into the essential structure of the experience that captures the essence of all participants' stories. Finally, the essential structure is validated by presenting it to the participants and asking for feedback as to whether it reflects their own experiences accurately and fully.

Participants in the first 2 phenomenological studies included adolescents between 11 and 21 years of age with life-threatening and non-life-threatening chronic illnesses, such as cancer, cystic fibrosis, scoliosis and asthma (Haase, 1987). Courage was a major focus of these investigations, and the Becoming Courageous Model, a precursor for the ARM, evolved from the work. The Becoming Courageous Model included the following key concepts and propositions derived from the phenomenological studies. Courage is a dynamic, progressive–regressive process that occurs through coping with daily encounters of “mini-situations of courage” such as medical procedures. Awareness of the situation as one of courage occurs gradually. Supportive relationships with others with the same or similar condition, family and health-care providers aid progress toward resolution of the situation as one of courage. Spirituality and hope also assist progress toward resolution. When a sense of resolution is reached, it is characterized by a sense of mastery, confidence and accomplishment as well as motivation to maintain or improve the situation and to help and inspire others (Haase, 1987).

A 3rd study that contributed to model generation (MGen 3) was conducted using a process called Simultaneous Concept Analysis (Haase et al.,1992). To increase the power of theoretical models to explain variance, concepts measured must be defined clearly. Three concepts identified in the phenomenological studies (hope, self-transcendence and spirituality) seemed to be somewhat theoretically elusive, and existing instruments to measure them had redundancy in items and subscales. The outcome of Simultaneous Concept Analysis is a series of matrices comparing the antecedents, critical attributes and outcomes across concepts. The matrices comparing hope, self-transcendence and spirituality helped to identify redundancy. These matrices assisted also with interpretation of the results in the measurement phases of the MEval studies. For example, in our model tests, when several instruments loaded on more than 1 factor, results of the Simultaneous Concept Analysis helped us identify sources of theoretical overlap and assisted with analysis decisions (e.g., when to use subscales rather than total scales in subsequent statistical examinations of the model).

In a 4th study (MGen 4), we used focus group techniques to identify the nature and extent of existing psycho-social support resources. While the phenomenological studies helped identify that relationships between health-care providers and adolescents with cancer were important for adjustment, very little research was found about the nature and extent of such relationships or the psycho-social resources available for adolescents with cancer. A series of focus group sessions were held in collaboration with 4 pediatric oncology clinical nurse specialists and social workers from 4 sites in the United States and Canada. The findings from the focus group sessions were used to develop an instrument to evaluate existing adolescent support program participation and satisfaction (Haase et al.,1994).

Two additional studies are mentioned here briefly for their contributions to our emerging understanding of the processes of resilience and HRQL. In a phenomenological study (MGen 5), we examined children's, adolescents' and their parents' perspectives of the experience of completing cancer treatment. These perspectives provided further insights into the process of resilience and HRQL over time. Findings indicated that resilience and HRQL were embedded in the cancer experience, continue after treatments and were qualitatively different for the children/adolescents and parents (Haase and Rostad, 1994). Insights gained from the thematic analysis of functional status in 12 elderly persons with chronic obstructive pulmonary disease (MGen 6) yielded surprisingly useful information about functional status/performance that may be generalizable to children/adolescents (Leidy and Haase, 1996, 1999). Illness-related events and circumstances were found to be challenges to personal integrity, specifically challenges to the sense of being able (functional status) and being with (connectedness or relationships with others). These results have led to further theoretical work on the concept of connectedness that goes beyond the initial phenomenological work on relationships with others. For example, the concepts of connected, unconnected and disconnected are relevant to investigations about the influence of relationships with health-care providers on the resilience of patients.

The MGen studies used a variety of qualitative research methods and provided a way of including the adolescent and family perspectives as the foundation of the ARM, unlike models that are developed primarily from evaluation of existing publications. The adolescents, their parents and health-care providers were asked specifically for their perspective on their cancer experiences, and these were incorporated into the ARM MEval studies at both the full-model specification and instrument levels. The phenomenological studies also helped to distinguish differences in perspective within and between adolescent, parent and health-care provider groups. The Simultaneous Concept Analysis helped to identify and clarify salient concepts, to identify conceptual redundancy and to increase the explanatory power of the ARM. Focus group methods helped to generate instruments that were generalizable across geographical groups and different clinical settings. Finally, the thematic analysis provided insights into distinctions and processes of functional performance and connectedness that go beyond mere functional capacity.

MEval studies

Three model evaluation studies were conducted and a 4th is in progress. These studies represent a progressive refinement of our theorizing about how adolescents adjust to the cancer experience with outcomes of resilience and HRQL.

The first 2 MEval studies were conducted to specify and evaluate the Becoming Courageous Model and instruments. Model specification for the Becoming Courageous Model (MEval 1) was accomplished using results from the phenomenological studies of courage and a decision-making process for combining qualitative findings with existing theory (Fig. 1a). The strategy consisted of developing labels and definitions for each factor in the model through an iterative process which sought to retain the inductively derived meanings from the phenomenological studies while taking advantage of existing theory. Hence, depending on the goodness-of-fit with the qualitative findings, some variables were taken from existing theory, such as that on coping, while other variables were labeled and defined on the basis of the phenomenological findings, such as relationships with others. The resulting theoretical model consisted of 6 latent (indirectly measured) variables: (i) Awareness of Condition, the extent to which adolescents with chronic illness feel the health-related situation is real, comprehensible and requires active participation; (ii) Ways of Coping, the constantly changing cognitive and behavioral effort to manage specific external/internal demands appraised as taxing or exceeding the resources of the person; (iii) Relationships with Others, the extent to which adolescents with chronic illness perceive social support, are comfortable informing others of the condition and are influenced by or perceive themselves influencing others with the same or similar condition; (iv) Spirituality, Hope and Spiritual Perspective; (v) Being Courageous, the extent of effort to live out situations related to having a chronic illness in ways that integrate commitment to values with the difficult realities of the situation; (vi) the outcome variable quality of life was developed from the phenomenologically derived theme category called Resolution and consisted of 3 components: self- transcendence, mastery or confidence and self-esteem.

thumbnail image

Figure 1. (a) Decision-making process part 1: combining qualitative data and existing theory. (b) Decision-making process part 2: instrumentation from qualitative data.

Download figure to PowerPoint

After the latent variables were defined, a latent variable measurement model was developed. To conduct latent variable model analysis, each latent factor requires at least 2 manifest variables (instruments) to measure it (Bentler, 1989). To assist in this task, another decision-making process (Fig. 1b) was developed to guide instrument selection so that the instruments would be grounded in the qualitative study findings while taking advantage of existing instruments. As a first step in this process, an instrument was developed for each major theme category derived from the phenomenological studies (MGen 1 and 2). Items were worded to reflect the words and phrases actually used by the adolescents. These new instruments were then examined for congruence with existing instruments, using criteria such as consistency with the qualitative data, adequate psychometric properties and clinical appropriateness. The result of this decision process was a set of 15 instruments to measure manifest variables: 8 existing instruments meeting established criteria for reliability and validity and 7 new instruments.

For the new instruments, readability, clarity and consistency (Imle and Atwood, 1988) were evaluated with 3 panels: 18 6th graders, 3 doctorally prepared nurses with instrumentation expertise and 3 adolescents with chronic illness. Criterion agreement was met at the 90% to 95% level for all panels, and revisions were made on the basis of panel member feedback. Content validity was addressed in 3 ways: (i) qualitative identification of dimensions of the variable and comparison of the dimensions with existing publications, (ii) generation of items using data from the qualitative study and comparison of items with existing instruments and (iii) attention to refinement of the items based on feedback from the panels of judges (Lynn, 1986).

After the model was specified and instruments identified and/or developed, the measurement and full model were evaluated in MEval 2 (Haase et al.,1991, 1993). This multi-site study was conducted at 3 medical centers in the United States and Canada using a convenient sample of 73 non-hospitalized adolescents with cancer, cystic fibrosis and asthma between 11 and 21 years of age. Reliability and validity estimates met established criteria (Cronbach's alpha > 0.70) for all but 2 instruments (Nunnally and Bernstein, 1994). The adolescent instrument that did not meet the criteria had a theta internal consistency of 0.67. This instrument was retained. The second instrument that did not meet the criteria was one completed by the parents. It was not used in further model evaluation. Support of the predicted factor structure and significant moderate to strong correlations between the new instruments and pre-existing instruments measuring the same latent variables provided further evidence of construct validity.

Once the internal consistency of the instruments was established, their relationships to the hypothesized latent variables constituting the Becoming Courageous Model were analyzed. Exploratory analysis of measurement models using EQS (Bentler, 1989) led to further reconceptualization of the latent variables. The best-fitting empirical measurement model (comparative fit index = 0.92) included subscales of instruments, rather than total scales, as indicators of 5 latent variables (Table II). The Jalowiec Coping Scale (Jalowiec et al.,1984; Jalowiec, 1988) and Being Courageous–Adolescent Report factors had lower factor loadings on HRQL than many other factors, indicating that these instruments explained variance in HRQL as well as in Courageous Coping.

Once the Becoming Courageous Model was settled upon, structural relationships among all 5 latent variables were explored. The best-fitting structural model (comparative fit index = 0.95) is displayed in Figure 2. Relationship Derived Meaning was, surprisingly, affected positively by Uncertainty in Illness (beta = 0.42), indicating that the greater the uncertainty, the more hope is sought/used by the adolescent. Courageous Coping was affected positively both by Defensive Coping (beta = 0.41) and by Relationship Derived Meaning (beta = 0.49). This finding supported our qualitatively derived hypothesis that adolescents usually used defensive coping at first and then learned more effective coping strategies. We hypothesized further that those who continued to use primarily defensive coping over time were at higher risk. Quality of life was affected negatively by Defensive Coping (beta = –0.40) and affected positively by Relationship Derived Meaning (beta = 0.70).

thumbnail image

Figure 2. Revised Becoming Courageous Model.

Download figure to PowerPoint

Table II. MEval 2 BECOMING COURAGEOUS MEASUREMENT MODEL (n = 73)1
Factor(s)Factor loading
  • 1

    Comparative fit index = 0.92.–Parentheses indicate secondary factor loadings.

Uncertainty in illness
 Complexity0.916
 Ambiguity0.814
Defensive Coping
 Jalowiec Coping Scale
  Evasive Subscale0.736
  Fatalistic Subscale0.688
  Emotive Subscale0.642
  Palliative Subscale0.578
 Herth Hope Index
  Mutuality Subscale−0.254
Relationship Derived Meaning
 Herth Hope Index
  Action Orientation0.940
  Mutuality0.778
  Temporality0.829
 Informing others0.352
 Wandersman Perceived Social Support0.490
 Herth Hope Scale0.940
Courageous Coping
 Influence of others with same or similar condition0.713
 Jalowiec Coping Scale
  Confrontive0.776
  Optimistic0.732
  Supportant0.723
  Palliative(0.500)
 Haase Coping with Chronic Illness0.596
 Being Courageous–Adolescent Report(0.485)
 Reed Self-transcendence(0.445)
Quality of Life
 Reed Self-transcendence0.627
 Nowatny Confidence0.829
 Rosenberg Self-esteem0.863
 Jalowiec Coping Scale
  Optimistic0.255
  Self-reliant0.497
  Palliative(0.789)
 Reed Spiritual Perspective0.358
 Being Courageous–Adolescent Report0.534

Following MEval 2, the Becoming Courageous Model was refined as the ARM-Version I based on several sources of input: the results of MEval 1; a critical analysis of the literature on resilience and related concepts; and MGen 3, the Simultaneous Concept Analysis. The theoretical ARM-Version 1 consisted of 5 classes of influencing factor and 1 outcome factor. The manifest variables to measure each factor are listed in Table III. The 3 classes of factor hypothesized to affect resilience positively were labeled “protective” factors, those factors that help to resist or recover from adversity (Jacelon, 1997). These included individual, family and social protective factors. The hypothesized Individual Protective Factors consisted of 2 latent factors: Courageous Coping and Derived Meaning. The hypothesized Family Protective Factors consisted of two latent factors: Family Atmosphere and Family Support/Resources. The hypothesized Social Protective Factors consisted of Health Care Resources and Social Integration.

Table III. MEval 3 ADOLESCENT RESILIENCE MODEL-VERSION 1 CLASSES OF FACTORS, LATENT FACTORS AND MANIFEST VARIABLES
Class of factorLatent factorManifest variables
Individual protective factorsCourageous CopingJalowiec Coping Scale—Revised (Jalowiec et al.,1984; Jalowiec, 1988) and subscales: Confrontive, Optimistic, Supportant
Being Courageous–Adolescent Report
 Derived MeaningHerth Hope Index (Herth, 1991)
Reed Spiritual Perspectives Scale (Reed, 1986)
Family protective factorFamily AtmosphereFamily Adaptability and Cohesion Scale II
Parent–Adolescent Communication Scale (Olson et al.,1985)
Perceived Social Support–Family (Procidano and Heller, 1983)
 Family Support/ResourcesFamily Strengths (Olson et al.,1985)
SES = Socioeconomic status Variables
Social protective factorsHealth-Care ResourcesPerceived Social Support–HCP = Health care provider
Adolescent Support Program Participation and Satisfaction
Adolescent Support Program Site Evaluation
 Social IntegrationPerceived Social Support–Friends (Procidano and Heller, 1983)
Individual risk factorDefensive CopingJalowiec Coping Scale—Revised and subscales: Avoidant, Evasive, Emotive, Fatalistic
Illness-related risk factorsIllness PerspectiveMishel Uncertainty in Illness Scale—Revised (Mishel, 1981)
Short Form Health Survey—MOS = Medical Outcomes Study. Health subscales: Perceptions and Mental Health
 Symptom Distress/Severity of IllnessMcCorkle Symptom Distress Scale (McCorkle and Young, 1978)
Short Form Health Survey—MOS (Stewart et al.,1988) and subscales: Physical Functions, Health Role, Pain
Frequency and length of hospitalizations
Resilience outcomeConfidence or MasteryNowotny Confidence Subscale (Nowotny, 1989)
Self-transcendence Reed Self-transcendence Scale (Reed, 1989)
Self-esteem Rosenberg Self-esteem (Rosenberg, 1989)

In addition to the 3 protective factors, 2 Risk Factors were hypothesized to affect the development of resilience negatively. These included the hypothesized Individual Risk Factor of Defensive Coping and the hypothesized Illness-Related Risk Factors of Illness Perspective and Symptom Distress/Severity of Illness.

The Quality of Life outcome factor from the Becoming Courageous Model was relabeled Resilience in the ARM-Version 1 since the concepts were consistent with descriptors of resilience outcomes in the literature. HRQL was not tested as an outcome variable in MEval 3. In MEval 4, HRQL was defined as a sense of well-being and an outcome of Resilience in ARM-Revision II model testing.

After specifying the model, the next step in MEval 3 was to further evaluate the psychometric properties of the new and existing instruments for adolescents with cancer and to conduct measurement and full-model tests of the ARM. MEval 3 was conducted at 4 sites in the United States and Canada with a sample (n = 130) of adolescents and young adults with cancer between 11 and 24 years of age. The best-fitting measurement model, as indicated by low off-diagonal standardized residuals (0.21) and a comparative fit index of 0.99, is depicted in Table IV. This measurement model was used to test the goodness-of-fit of the hypothesized full latent variable model. Results of statistical tests indicated that the estimated parameters were unstable, though the goodness-of-fit index for most exploratory models was >0.96. The relatively small sample size (n = 81), due to list-wise deletion of cases with missing data, is believed to be responsible for the observed instability (poor reproducibility) of parameters.

Table IV. MEval 3: ARM MEASUREMENT MODEL (n = 81)1
FactorsFactor loadings
  • 1

    Factor Loadings >0.30; comparative fit index = 0.99.

  • 2

    Significant factor loadings (p < 0.05).

Individual protective factors
 Courageous Coping (F1)
  Jalowiec Confrontive0.762
  Jalowiec Optimistic0.902
  Jalowiec Supportant0.84
 Derived Meaning—Hope (F2)
  Herth Mutuality Subscale0.46
 Spirituality (F3)
  Reed Spiritual Perspective0.442
Family protective factor (F4)
 FACES II: Cohesion0.922
 FACES II: Adaptability0.962
 Parent Communication—Problem: Mother0.842
 Parent Communication—Open: Mother0.862
 Parent Communication—Problem: Father0.672
 Family Strength0.732
 Perceived Social Support—Family0.592
Social protective factor: social integration
 Perceived Social Support—Friends (F5)0.902
 Perceived Social Support—Health-Care Provider (F6)1.002
Individual risk factors
 Defensive Coping (F7)
  Jalowiec Evasive0.972
  Jalowiec Emotive0.832
 Severity of Illness (F8)
  Symptom Distress0.862
  MOS Physical Function0.642
  MOS Role Function0.742
  MOS Social Function0.672
  MOS—Health Perception0.802
 Illness Perspective (F9)
  MOS—Mental Health0.832
  Mishel Uncertainty in Illness Scale—Ambiguity0.37
Subscale
Resilience (F10)
 Nowotny Confidence Subscale0.672
 Self-esteem1.002
 Self-transcendence0.302

The MEval studies provided support for the validity and reliability of ARM-related instruments. All instruments in the ARM-Version 1 demonstrated properties that met criteria for new or established instruments and evidence of concurrent and/or construct validity based on correlations and predicted factor structures. Stability (reproducibility) in the measurement model was evident for most, but not all, factors: (i) indicators of Defensive Coping, Courageous Coping and Resilience loaded together consistently and significantly across the MEval studies; (ii) indicators of the Family Protective Factors and Severity of Illness loaded as predicted in each exploratory model run in MEval 3; (iii) indicators of Illness Perspectives, Derived Meaning, Health Care Resources and Social Integration were unstable in exploratory measurement model runs. The instability may be attributed to small sample size and/or the fact that stages of illness, not controlled for, may have more impact on these variables. In addition, in some cases, higher-order factors may provide a better fit in the model. For example, the latent factors Health-Care Resources and Social Integration might explain more variance if combined into a larger Social Protective Factor and Illness Perceptions, and Symptom Distress and Severity of Illness might be combined into Illness-Related Distress.

The full structural equation model for the Becoming Courageous Model, although stable, did not include all factors important to resilience. Several good-fit models, with parameters in the direction predicted, were identified in full structural equation model evaluations of the ARM data. However, the parameters are unstable, and further work is being done with a larger sample in our current study, MEval 4.

DISCUSSION AND CONCLUSIONS

  1. Top of page
  2. Abstract
  3. TRIANGULATION OF METHODS
  4. DISCUSSION AND CONCLUSIONS
  5. Acknowledgements
  6. REFERENCES

The triangulation of qualitative and quantitative research approaches was a useful way of gaining understanding of the process of resilience and HRQL, but several issues must be raised. The results of a theory generated from qualitative studies are more complex, but they have the potential also to provide a more explanatory and meaningful model. Triangulation is costly in time and resources and requires researchers who have skill in both qualitative and quantitative methods. However, without development of theory that is grounded in the experiences of patients, we are likely to continue the “shopping list” approach to HRQL.

To adequately represent the factors derived from the meaning-based approach to understanding resilience and HRQL, we used a profile of instruments, rather than an existing generic HRQL instrument. The profile of instruments has the advantage that the adolescents seemed to derive meaning from being able to think about and discuss the issues raised in the instruments. Despite the length of the questionnaire, which took approximately 2 hr to complete, we had very little missing data and only 4 adolescents were unwilling to participate in the MEval studies. Perhaps the most compelling evidence that the instruments were meaning-based are the qualitative responses from some of the adolescents. Several adolescents wrote thank you notes for the opportunity to think about the issues raised by questions in the booklet, and some returned the monetary appreciation gift, indicating that they wanted someone else to be offered the inducement to participate. Two adolescents provided elaborate drawings on the envelopes, indicating their appreciation for participating.

The thoughtful completion of the questionnaires by adolescents is a key issue in studying HRQL. Additional strategies that we used to obtain complete data sets include designing an attractive illustrated booklet, using a project logo, reviewing instructions and format of each questionnaire with subjects, providing a “thank you” in the form of a movie ticket or cash and conducting a follow-up phone call to answer questions and encourage completion of the booklet within 48 hr. Adolescents were encouraged to take their time to complete the questionnaire, to take breaks when getting tired and to answer the questions without input from others. The long duration for questionnaire completion resulted in reduced control over the data collection process. It is possible that some adolescents received help from family members, and we know that a few adolescents took more than 48 hr to complete the questionnaire.

It is not adequate to simply assess HRQL. Interventions to improve life quality for pediatric patients is imperative. To that end, as part of MEval 4, we are developing and pilot testing an intervention in which we use baseline data to develop a psycho-social profile of the adolescent and family. The profile is used by an interdisciplinary health-care team to develop a psycho-social care plan based on the ARM-Version 2. The care plan is then discussed and revised based on input from the adolescent and family. In the care plan, we are emphasizing and building on the strengths identified in the ARM protective factors. To date, 6 subjects have been included in the intervention pilot study and the profiles for these 6 have been extremely useful for the health-care team. While the profile seems to validate the initial clinical impressions, it also provides a means of identifying specific strengths upon which to build and it guides development of interventions that are meaningful from the adolescent's and family's perspectives. Clinicians are valuing the systematic and holistic approach to psycho-social assessment and planning. Adolescents and their families have indicated that they value and appreciate the opportunity to connect with health-care providers through a discussion of their profile.

The use of triangulation to develop meaning-based models requires a flexibility in thinking and a tolerance for ambiguity as model refinement occurs. Although the pressing need for HRQL assessments in the context of clinical trials may preclude widespread use of this strategy, triangulation can make significant contributions to a fuller understanding of HRQL.

Acknowledgements

  1. Top of page
  2. Abstract
  3. TRIANGULATION OF METHODS
  4. DISCUSSION AND CONCLUSIONS
  5. Acknowledgements
  6. REFERENCES

The authors acknowledge the contribution of the following individuals: Ms. D. Berry; Ms. C. Doner; Drs. A. Kuperberg, N. Kline Leidy and S. Myers; and Ms. M. Rostad. Their knowledge of and dedication to children and adolescents with cancer have contributed greatly to our current thinking about the ARM.

REFERENCES

  1. Top of page
  2. Abstract
  3. TRIANGULATION OF METHODS
  4. DISCUSSION AND CONCLUSIONS
  5. Acknowledgements
  6. REFERENCES
  • Bandura, A., Self-efficacy: toward a unifying theory of behavioral change. Psychol. Rev., 84, 191-215 (1977). Medline
  • Bentler, P., EQS: structural equations program manual, BMDP Statistical Software, Los Angeles (1989).
  • Colaizzi, P.F., Psychological research as the phenomenologist views it. In: R.S.Valle and M.King (eds.), Existential-phenomenological alternatives for psychology, pp. 48-71, Oxford University Press, New York (1978).
  • Costain, K., Hewison, J. and Hawes, M., Comparison of a function-based model and a meaning-based model of quality of life in oncology: multidimensionality examined. J. pychosoc. Oncol., 11, 17-37 (1993).
  • Forsyth, G., Delaney, K. and Gresham, M., Vying for a winning position: management style of the chronically ill. Res. Nurs. Health, 7, 181-188 (1984). Medline
  • Giorgi, A., Psychology as a human science: a phenomenologically based approach, Harper and Row, New York (1970).
  • Haase, J., The components of courage in chronically ill adolescents. Adv. Nurs. Sci., 9, 64-80 (1987).
  • Haase, J., Berry, D. and Stutzer, C., Instruments for the Becoming Courageous model [Abstract]. In: National conference on instrumentation in nursing, p. 41, College of Nursing, Tucson (1991).
  • Haase, J. and Braden, C.J., Guidelines for achieving clarity of concepts related to quality of life. In: C.King (ed.), Quality of life: theory, research, and practice, pp. 54-73, Jones and Bartlett, Boston (1998).
  • Haase, J.E., Britt, T., Coward, D.D., Leidy, N.K. and Penn, P.E., Simultaneous concept analysis of spiritual perspective, hope, acceptance, and self-transcendence. Image, 24, 141-147 (1992).
  • Haase, J., Doner, C., Heiney, S., Ruccioni, K., Kuperberg, A. and Stutzer, C., Characteristics of cancer support programs for adolescents [Abstract]. In: American Cancer Society third national conference on cancer nursing research, p. 35-C, American Cancer Society, Atlanta (1994).
  • Haase, J. and Rostad, M., Experiences of completing cancer treatments: child perspectives. Oncol. Nurs. Forum, 21, 1483-1494 (1994). Medline
  • Haase, J., Stutzer, C. and Berry, D., Quality of life outcomes for chronically ill adolescents based on the adolescent resilience model [Abstract]. In: Sigma Theta Tau International Nursing Research Congress, p. 367, Sigma Theta Tau, Indianapolis (1993).
  • Herth, K., Development and refinement of an instrument to measure hope. Nurs. Pract., 5, 39-51 (1991).
  • Imle, M. and Atwood, J., Retaining qualitative validity while gaining quantitative reliability and validity: development of the transition to parenthood concerns scale. Adv. Nurs. Sci., 11, 61-75 (1988).
  • Jacelon, C., The trait and process of resilience. J. adv. Nurs., 25, 123-129 (1997). Medline
  • Jalowiec, A., Confirmatory factor analysis of the Jalowiec Coping Scale. In: C.F.Waltz and O.L.Strickland (eds.), Measurement of nursing outcomes. Vol. 1, Measuring client outcomes, pp. 287-308, Springer, New York (1988).
  • Jalowiec, A., Murphy, S.P. and Powers, M.J., Psychometric assessment of the Jaloweic Coping Scale. Nurs. Res., 33, 157-161 (1984). Medline
  • Kobasa, S., Stressful life events, personality and health: an inquiry into hardiness. J. pers. soc. Psychol., 37, 1-11 (1979). Medline
  • Leidy, N. and Haase, J., Functional performance in people with chronic obstructive pulmonary disease: a qualitative analysis. Adv. Nurs. Sci., 18, 77-89 (1996).
  • Leidy, N. and Haase, J., Functional status from the patient's perspective: The challenge of preserving personal integrity. Res. Nurs. Health 22, 67-77 (1999). Medline
  • Lynn, M., Determination and quantification of content validity. Nurs. Res., 35, 382-385 (1986). Medline
  • McCorkle, R. and Young, K., Development of a symptom distress scale. Cancer Nurs., 1, 373-378 (1978). Medline
  • Mishel, M., The measurement of uncertainty in illness. Nurs. Res., 30, 258-263 (1981). Medline
  • Myers, S. and Haase, J., Guidelines for combining qualitative and quantitative research approaches. Nurs. Res., 38, 299-301 (1989). Medline
  • Nowotny, M.L., Assessment of hope in patients with cancer: development of an instrument. Oncol. Nurs. Forum, 16, 57-61 (1989). Medline
  • Nunnally, J. and Bernstein, I., Psychometric theory (3rd ed.), McGraw-Hill, New York (1994).
  • Olson, D.H., McCubbin, H.I., Barnes, H., Larsen, A., Muxen, M. and Wilson, M., Family inventories: inventories used in a national survey of families across the family life cycle, Family Social Science, St. Paul (1985).
  • Procidano, M.E. and Heller, K., Measures of perceived social support from friends and from family: three validation studies. Amer. J. Community Psychol., 11, 1-24 (1983).
  • Reed, P., Spirituality and well-being in terminally ill and healthy adults. Res. Nurs. Health, 9, 35-42 (1986). Medline
  • Reed, P., Mental health of older adults: initiating research based on a developmental conceptual framework. West. J. Nurs. Res., 11, 143-157 (1989). Medline
  • Rosenberg, M., Society and the adolescent self-image, Wesleyan University Press, Middletown (1989).
  • Rutter, M., Psychosocial resilience and protective mechanisms. Amer. J. Orthopsychiatry, 57, 316-331 (1987).
  • Sinnema, G., Resilience among children with special health-care needs and among their families. Pediatr. Ann., 20, 483-486 (1991). Medline
  • Stewart, A., Hays, R. and Ware, J., The MOS Short Form General Health Survey: reliability in a patient population. Med. Care, 26, 724-735 (1988). Medline