Abstract
Purpose:
To develop and evaluate a new instrument that measures unique aspects of long-term survivorship for people diagnosed with cancer as Adolescent and Young Adult (AYA), not measured by existing tools.
Methods:
A new candidate instrument—the Impact of Cancer for Adolescent and Young Adult (IOC-AYA) cancer survivors—was developed and administered to AYA cancer survivors aged 18–35 years who were 15–29 years old when diagnosed with cancer. Psychometric properties of newly derived scales were assessed.
Results:
Factor analyses of items derived seven new and specific subscales: Social Life; Uncertainties, Worries and Wonders; Cognitive Function; Sense of Purpose/Life Goals; Identity; Health Behaviors; and Health Literacy. Two separate and conditional subscales were formed: Relationship Concerns (partnered/unpartnered) and Having Children (at least one child/no children). Internal consistency measurements for these subscales ranged from 0.70 to 0.90. Expected associations within and among IOC-AYA subscales and standardized measures of quality of life were observed.
Conclusion:
Psychometric analyses indicated that this initial version of the IOC-AYA measures distinct and relevant constructs for survivors diagnosed with cancer in adolescence and young adulthood. Future work is needed to confirm the responsiveness to change and further validate the instrument in multiple and representative samples. Use of the IOC-AYA instrument in research and clinical practice will inform the development of psychosocial and supportive care interventions that not only minimize or prevent long-term deleterious effects of cancer but also promote positive adaptation, resilience, and the achievement of age-specific developmental tasks.
Introduction
T
Cancer survivorship studies most often involve homogenous samples of older adult patients or heterogeneous samples inclusive of patients across a wide range of ages. Emerging investigations among age-specific populations of AYA cancer patients demonstrate extensive psychosocial and symptom burdens of disease, for example, diminished feelings of autonomy, problems with identity development, disruptions of social life and school/employment, relationship, sexuality, and fertility issues, which are different from both pediatric and older adult cancer populations.4–6 However, most survivor studies are limited by a reliance on generic standardized measures of quality of life (QoL) and/or cancer-specific measures designed specifically for either adult or pediatric populations that are not validated in the AYA cancer population. 7 Specific instruments used to asses QoL in AYAs are the Pediatric Quality-of-Life (PedsQL) Generic Core (with cancer-specific modules for ages 13–18 and 18–25 years)8,9 and the Minneapolis–Manchester measure (with an adolescent module for ages 13–20 years and a nonvalidated young adult module for ages 21–45 years). 10 A cancer-specific QoL measure for AYAs has recently been developed. 5 However, all these cancer-specific measures assess issues that are salient during or shortly after cancer treatment and do not capture the distinct long-term experiences of AYA cancer survivors nor their attributions about how current conditions in their lives may in some way be related to, or influenced by, having had cancer. 11
Development of the IOC measure for adult cancer survivors (all ages) was spurred by the need to measure unique aspects of survivorship not addressed by existing QoL measures. 12 Whereas QoL measures generally focus on physical, cognitive, social, and emotional functioning, the IOC is intended to assess off-treatment survivorship experiences as they relate to employment, life outlook and concerns for the future, body and healthcare, feelings about and meanings attributed to cancer, and involvement in social activities and relationships. 12 A childhood cancer survivor version assesses a different set of survivorship issues and challenges of relevance to teenagers and young adults who were diagnosed with cancer as children and treated on pediatric cancer protocols.13,14 The aim of the current study is to develop a new module of the IOC that assesses the varied and unique ways in which cancer impacts the lives of young people who were diagnosed with cancer between the ages of 15 and 29 years.
Methods
This study describes the development and pilot testing of candidate items for an AYA version of the IOC instrument, followed by administration and psychometric evaluation of these new items in a sample of AYA survivors. All procedures were approved by the University of Michigan Comprehensive Cancer Center Institutional Review Board.
Instrument development
Candidate items were derived through an iterative process of expert item review, AYA survivor focus groups, item reduction and refinement, and cognitive testing. 15
Expert review
Seventeen clinicians and researchers involved in AYA oncology care and nine AYA survivor advocates were identified through the principal investigator's professional network and invited via US mail to complete a survey in which they were instructed to rate the importance of each of 82 items from the childhood cancer survivor version of the IOC instrument (IOC-CS).13,14 On a four-point Likert scale ranging from 0 (“not at all”) to 3 (“very much”), participants evaluated the IOC-CS items in terms of their importance in understanding cancer's impact on the lives of AYA cancer survivors. Respondents were also asked to rank order the top 3 items of relevance in each 10 number of thematic categories, as well as to note items that they felt were not relevant to AYA cancer survivors. They also were asked to suggest any other topics related to any category that they felt should be asked of AYA cancer survivors.
Mean scores for each of 82 items were calculated. Items with a mean score <2 or a standard deviation >1 were eliminated from future psychometric consideration. Also, 17 new items were developed for categories in which reviewers suggested topics of relevance to AYA cancer survivors that they thought were not assessed by the existing items within the category.
AYA survivor review and focus groups
A focus group consisting of eight off-treatment survivors aged 19–34 years was then conducted at the University of Michigan Comprehensive Cancer Center. Participants reviewed the items, one section at a time, and provided commentary as to their relevance in terms of learning something important about how cancer affects the lives of AYA cancer survivors. Because eliminating items based on expert review only is limited, participants in the focus groups also were asked to review the items rejected by the expert reviewers and provide an argument for retaining them if they in fact believed them to be relevant. Focus group participants also provided suggestions as to better age-appropriate wording and phrasing of items. In determining retention of items, the investigative team assessed general consensus from the focus group as to whether an individual item should be retained or jettisoned.
Cognitive testing
Fourteen AYA cancer survivors recruited through the University of Michigan Cancer Center participated in individual interviews in which they reviewed 99 retained items and responded to interviewer prompts to identify items that were incomprehensible or difficult to understand. To establish face validity of items, investigators asked respondents to describe what they believed particular items meant, and noted when respondents could not explain an item or provide a response that was consistent with the investigators' intent of the item. Feedback from respondents assisted investigators with the rewording of items and elimination of eight items.
Administration and evaluation of the IOC-AYA
Ninety-one items (see Appendix) were selected for inclusion in an initial administration of a candidate IOC-AYA instrument to a new sample of AYA cancer survivors. A potential pool of 551 respondents derived from the tumor registry at the University of Michigan Comprehensive Cancer Center received survey questionnaires, informed consent forms, and a self-addressed and stamped return envelope via US mail. Study eligibility criteria included AYA cancer survivors who were 15–29 years old at time of diagnosis, 18–35 years old at time of study, no more than 10 years postdiagnosis, and diagnosed with one of the following invasive cancer types: Hodgkin lymphoma, non-Hodgkin lymphoma, leukemia, testicular cancer, soft tissue sarcoma, and bone tumors (i.e., Ewings, osteosarcoma). These cancers comprise 41% of all invasive cancers in the 15–29-year-old age range. 16
Survey booklet
IOC-AYA version
All 91 candidate items were organized and presented to survey respondents in 10 hypothesized subcategories: Your body and your health; Cancer treatment and healthcare; Who are you? About cancer; Meaning of cancer; Memory and thinking; Health insurance, money, school, work; Parents; Socializing and being with friends; Life goals (see Appendix). Two additional subcategories involved items related to “Relationships and Dating” and “Having Children.” Respondents answered different subsets of items depending upon their relationship status and whether or not they had children.
To assess the extent of impact of each item on AYA cancer survivors' lives, respondents were instructed to indicate their level of agreement with each item on a five-point Likert scale (0 = strongly disagree; 1 = disagree; 2 = neither agree nor disagree; 3 = agree; 4 = strongly agree). Higher scores indicated a greater IOC with regard to the item. In addition to the candidate items, the mailed questionnaire also included (1) the SF-36, a widely used and well-validated measure of QoL in healthy populations, ill populations, and cancer patients/survivors specifically 17 and (2) the Brief Symptom Inventory (BSI-18), another widely used and well-validated instrument that assesses mental health status with regard to depression, anxiety, and somatic distress. 18 Both measures have demonstrated utility, validity, and reliability in AYA cancer populations. 19
Sociodemographic data included gender, race, employment status, educational attainment, marital/relationship status, age at study, age at diagnosis, and years since diagnosis. Surveillance, Epidemiology, and End Results program (SEER) codes were used to categorize cancer type and then organize them by severity of disease. 20 To approximate severity of disease, three categories were generated for this study based on epidemiological evidence specific to AYA cancer patients, diseases with (1) expected 5-year survival rates >80% (e.g., Hodgkin lymphoma); (2) expected 5-year survival rates of 50%–80% (e.g., osteosarcoma); (3) expected 5-year survival rates <50% (e.g., leukemia). 21
Statistical analyses
Item descriptive statistics and the response distributions for each item were calculated to examine central tendency, variability, and symmetry. Thereafter, theoretically and statistically derived dimensions and scales were determined by engaging a five-step iterative process consisting of (1) a priori categorization of items; (2) exploratory factor analyses to test psychometric properties of hypothesized scales; (3) item reduction based on factor loadings; (4) rescaling and retesting (factor analyses); and (5) derivation and scoring of subscales. Scales were evaluated for both theoretical and statistical consistency. Cronbach's alpha was calculated for each subscale to evaluate internal reliability, a minimum score of 0.70 is preferred. 22 Construct validity was evaluated by examining Pearson product-moment correlations among the various derived subscales. Multitrait scaling was used to confirm the scale structure of the questionnaire. To test for item-scale convergent validity, correlations of 0.40 or greater were sought between an individual item and its scale, corrected for overlap. Item-scale discriminant validity was examined by comparing the correlation of each item with its own scale versus other scales. An item was expected to correlate significantly higher with its own scale than with other scales.
Based on a priori hypotheses, we evaluated external (concurrent) validity by examining correlations of each of the IOC-AYA subscale with the Physical Component Score (PCS) and Mental Component Score (MCS) of the SF-36, and the Global Severity Index (GSI) scores of the BSI-18. We expected that scales that are conceptually related would correlate moderately to high with one another (r ≥ 0.40). 22 All analyses were performed using SPSS software, version 22.0 (SPSS, Inc., Chicago, IL).
Results
Subject characteristics
One hundred seventy-three AYA cancer survivors completed and returned useable surveys (31% response rate). Table 1 summarizes demographic and clinical characteristics for respondents and nonrespondents. Respondents were significantly more likely to be of white race (80% vs. 70%) and diagnosed with Hodgkin lymphoma (34% vs. 25%) compared to nonrespondents. In short, mean age of respondents at time of diagnosis was 21.6 years and at time of survey was 27.3 years, 58% were male, 52% had a partner, 36% had less than high school education, 18% were not employed or not in school, and the majority of the survivors had a survival rate between 80% and 100% (53%).
NA, not applicable; SD, standard deviation.
Scaling of the IOC-AYA
We performed an iterative process of exploratory factor analyses and subsequent item reduction only on the 65 items answered by all participants (Table 2). The items of the “Relationship Concerns” and “Having Children” scales (26 items in total) were excluded from these factor analyses since these were conditional items. For each iteration, principal axis factoring was followed by a varimax (orthogonal) rotation. For the first iteration, we forced a 10-factor solution as a potential representation of a priori hypothesized domains. Together, these factors explained 56% of the common variance. Ten items with factor loadings <0.4 on more than one factor were eliminated during this initial run. In subsequent iterations, the number of factors retained was based on three criteria: (1) evaluation of screen tests and factor loading coefficients; (2) consistent cumulative proportion of common variance explained with the results of the first iteration (i.e., >56%); and (3) theoretical interpretability. A final iteration yielded a 7-factor solution explaining 66% of the common variance, after the elimination of 24 additional items due to absolute factor loading coefficients that were <0.4 on all factors or else ≥0.4 on multiple factors. Scale content and summary statistics for the 7-factor-derived subscales consisting of 31 items are described in Table 2. Three subscales—(1) Social Life; (2) Uncertainties, Worries, and Wonders; and (3) Cognitive Function, suggested a negative or deleterious effect of cancer on AYA cancer survivors. Four others (4) Sense of Purpose/Goals; (5) Identity; (6) Health Behaviors; and (7) Health Literacy suggested a positive valence.
Category in which the item appeared in the mailed survey.
Where item loaded above 0.40 on own factor and no greater than 0.40 on any other factor and maintained a clear theoretical interpretation.
Mean scores derived from the following response categories: 0 = strongly disagree; 1 = disagree; 2 = neither agree nor disagree; 3 = agree; 4 = strongly agree.
Scaled separately.
Scoring on this item was reversed before calculating scale mean scores.
IOC, Impact of Cancer.
Reliability and validity
Cronbach's alphas ranged from 0.70 to 0.90 (Table 2). None of the scales showed improvement in internal consistency when any item was removed; thus, no additional items were eliminated from the instrument. The “Relationship Concerns” and “Having Children” items were scaled separately after content review and assessment of internal reliability. Cronbach's alphas for the “Relationship Concerns” subscales were 0.85 for those without partner and 0.70 for AYAs with partner. Cronbach's alphas for the “Having Children” subscales were 0.71 for AYAs who had one or more children and .77 for those without children. No ceiling or floor effects (disproportionately high or low [90%] numbers of respondents scoring at one of the two ends of an item) were observed.
To evaluate internal construct validity, we examined Pearson product-moment correlations among the subscale scores (Table 3). Based on literature suggesting interrelationships among physical, psychological, social, and spiritual domains of QoL, 23 we expected to observe such associations among related subscales of the IOC-AYA. Indeed, the scales assessing “Social Life” and “Uncertainties, Worries, and Wonders” were significantly correlated in expected directions with subscales assessing “Cognitive function,” “Relationship Concerns,” and “Having Children.” As hypothesized, “Health Behavior” was significantly correlated with “Sense of Purpose/Life Goals” and with “Social Life.”
Bold indicates correlation ≥ 0.40.
Pearson correlation between items and hypothesized scale (corrected for overlap).
Number of item-scale correlations that meet minimum standard for convergent validity (≤0.40).
Pearson correlations between items and other scales.
Correlations higher between items and other scales in comparison with hypothesized scale.
Based on only three AYA cancer survivors having children without a partner.
AYA, Adolescent and Young Adult; NA, not applicable.
Scales: 1, Social Life; 2, Uncertainties, Worries and Wonders; 3, Sense of Purpose/Goals; 4, Identity; 5, Health Behavior; 6, Health Literacy; 7, Cognitive Function; R-UP, Relationships and Dating (Unmarried or unpartnered); R-P, Relationships (Married or Partnered); HV, Having Children; NHV, Not Having Children.
p < 0.05; **p < 0.01.
Results of multitrait scaling analysis suggested that scaling assumptions were adequately met for six of the seven scales (Table 3).
With regard to concurrent validity, as expected, reporting a negative IOC (social life, worries, and cognitive function) was moderately to highly associated with lower MCS of the SF-36 and higher levels of distress, as measured by the BSI-18 (Table 4). No high correlations were found between the IOC-AYA subscales with the PCS of the SF-36. Reporting a more positive impact, in terms of sense of purpose/life goals and health behaviors, was associated with significantly better MCS and lower distress scores.
Bold values represent substantial to high correlations (r ≥ 0.4). 22
p < 0.05; **p < 0.01.
Discussion
This work fills a research and clinical care gap by offering an instrument that assesses the perceived IOC for a unique and underserved age-defined population of cancer survivors. It is significant in that prior research suggests that subjective perceptions of one's experiences are better predictors of QoL outcomes than are objective clinical conditions or sociodemographic indicators.24–26 Indeed, when comparing IOC-AYA subscales to existing standardized QoL measures, we found that the IOC-AYA subscales were always more strongly correlated with measures of mental health function and psychological well-being than with measures of physical health function. This finding suggest that the IOC-AYA is more closely related to survivors' psychological and subjective experience, and perhaps reflects survivors' perceptions of the way cancer has affected them and continues to affect them. Our results also suggest that the IOC-AYA consists of reliable and valid subscales for which we observed expected significant and substantial associations (convergence and divergence) among IOC-AYA subscales and existing standardized QoL measures.
Compared to the childhood survivor version of the IOC, the IOC-AYA version contains more items related to social life (social isolation; feeling different compared to people without cancer; uncertainty how to move on with life; and not able to do work of interest), sense of purpose/goals, and relationships and family planning, whereas less emphasis is placed on body image (attractiveness, self-confident), talking with parents, sibling concerns, and financial issues. This suggests that cancer survivors diagnosed as younger children experience cancer differently than do those diagnosed as AYA, which may have to do with the different phase of life when diagnosed with cancer. Childhood cancer survivors are living with their parents (and siblings) at time of diagnosis. At this young age, cancer treatment can have a significant negative impact on the growth and development of the body, which may negatively influence the development of a healthy body image and normal confidence levels. Most AYAs, on the contrary, are living on their own or with a partner/friends at time of diagnosis, a phase of life where romantic and/or intimate relationships, family planning, and establishing a personal set of values and goals are probably more salient. The difference in items and scales of the IOC-CS and IOC-AYA again emphasizes the need for an AYA-specific measure.
The response rate (31%) and inclusion of only a selected group of tumor types (only 41% of all invasive cancers in the AYA age range were covered) and age range (15–29 years at diagnosis) limit the generalizability of our findings, although respondents were representative of the entire sample pool in terms of age at diagnosis, age at study, time since diagnosis, and gender. In addition, the small subsample answering the “Relationships Concerns” and “Having Children” subscales limited our capacity to test validity. Given the small sample size of this study, we emphasize the need for future work to confirm the factor structure of the IOC-AYA questionnaire and establish the validity for its scales, and to test their criterion validity (known-group analyses) and predictive validity (test–retest analyses) in larger heterogeneous samples of AYA survivors.
For AYA cancer survivors, the ability to prevent deleterious long-term effects and promote adaptive psychosocial functioning require an understanding of how the perceived IOC intersects with the unique developmental challenges of adolescence and young adulthood. 27 The work proposed here extends current stress-coping and QoL models in cancer survivorship research by placing this investigation of cancer-related outcomes in the context of human psychological/cognitive growth and life span development. It offers an opportunity to identify aspects of cancer survivorship for AYAs that are yet to be well established and measured empirically. Its ultimate product—a field tested and psychometrically sound instrument appropriate for AYAs—has the potential to advance the field of psycho-oncology research by offering investigators a means for obtaining a clear and comprehensive understanding of the IOC on the lives of an age-defined subpopulation of cancer survivors. These data will subsequently inform the development of psychosocial and supportive care interventions that not only minimize or prevent long-term deleterious effects of cancer but also promote positive adaptation, resilience, and the achievement of age-specific developmental tasks. This knowledge also can serve as a foundation for deriving testable hypotheses regarding the myriad variables that may influence the quality of survivors' lives throughout the remainder of their lives. In addition, once established, the IOC-AYA has the potential for identifying salient survivorship issues in clinical care settings, where survivors access ongoing healthcare monitoring for long-term effects.
Footnotes
Acknowledgments
We thank the participating AYA cancer survivors, and Sun Teng for preparing the data. This work was supported by the Vivian A. and James L. Curtis School of Social Work Research and Training Center, University of Michigan (Ann Arbor, MI), Grant No. U029841. Dr. Olga Husson (KUN2015-7527) is supported by a Social Psychology Fellowship from the Dutch Cancer Society.
Author Disclosure Statement
No competing financial interests exist.