Psychometric evaluation of the impact of cancer (IOC-CS) scale for young adult survivors of childhood cancer

Abstract

Introduction

Psychosocial outcomes derived from standardized and disease-specific measures are often used in pediatric oncology; however, the reliability, validity and utility of these instruments in adult survivors of childhood cancer have yet to be established.

Purpose

To develop and evaluate a new instrument that measures aspects of long-term survivorship not measured by existing tools.

Methods

A new candidate instrument—the Impact of Cancer for childhood cancer survivors (IOC-CS)—was administered to childhood cancer survivors aged 18–39 who were 21 years of age or younger when diagnosed with cancer. Psychometric properties of newly derived scales were assessed.

Results

Factor analyses of items derived eight new and specific subscales: Life Challenges, Body/Health, Talking With Parents, Personal Growth, Thinking/Memory Problems, Health Literacy, Socializing and Financial Problems. Internal consistency measurements for these subscales ranged from 0.70 to 0.86. Expected associations within and among the IOC-CS subscales and standardized measures of health-related quality of life (HRQOL) were observed, as were some unexpected findings.

Conclusion

Psychometric analyses indicated that this initial version of the IOC-CS measures distinct and relevant constructs for young adult survivors of childhood cancer. Future work is necessary to confirm the responsiveness and further validate the instrument in multiple and representative samples.

Introduction

Numerous investigations using various research methods have assessed psychological adjustment, psychosocial adaptation and health-related quality of life (HRQOL) among childhood cancer survivors. Yet, after 30 years of study, we still lack a clear and comprehensive vision of how a diagnosis of cancer in childhood continues to affect survivors later in life. Early and current studies suggest that sizable proportions of childhood cancer survivors are seriously troubled psychologically, experiencing worse HRQOL when compared to peers, and perhaps even exhibiting psychopathologic symptoms of depression or post-traumatic stress [1–5]. In contrast, some research emphasizes that most childhood cancer survivors score well and within normal ranges on standardized psychometric measures of psychological well-being and HRQOL [6, 7]. A third set of studies can be characterized as suggesting that childhood cancer survivors are significantly healthier or better adjusted (in psychosocial and HRQOL terms) than population norms or healthy controls [4, 8–13]. Some long-term survivor studies suggest the potential for psychological growth and positive adaptation related to having had cancer as a child [14–16]. Thus, the truth about being a long-term survivor of childhood cancer varies depending upon multiple and varied sources of information and methods of data collection.

Evaluating the long-term effects of cancer and its treatment requires measurements capable of assessing the impact of cancer in life domains having relevance to this young survivor population. To date, the pediatric oncology literature is replete with reports derived from the administration of generic measures assessing psychiatric symptoms (e.g., depression, anxiety, adjustment to illness, post-traumatic stress) and health-related quality of life (HRQOL) domains (e.g., physical functioning, psychological well-being, social functioning) (For reviews see Varni et al. [17], Cantrell [18], Zebrack and Cella [19]). Relatively few studies, however, have assessed issues of particular concern to a young adult cancer survivor population, such as the effects of cancer and its treatment on sexuality and fertility, future health, school and work performance, ability to achieve life goals, relationships with family, friends and intimate partners and spirituality. In most cases, study samples consist of pediatric patients in treatment and, to a lesser extent, long-term survivors.

Varni and colleagues [17] suggest that there is value in the administration of both generic standardized measures and disease-specific instruments to assess HRQOL. One such instrument, the PedsQL generic core scale is widely used in pediatric quality of life research and has well-established norms derived from pediatric patient populations [20]. Its companion cancer module has demonstrated reliability and validity in childhood cancer patient populations, as well [21]. Others, like the Minneapolis-Manchester Quality of Life instrument [22, 23], the Pediatric Oncology Quality-of-Life measure [24], the Health Utilities Indices [25] and the Miami Pediatric Quality of Life Questionnaire [26] have been developed specifically for children and adolescents. These measures, however, have received little to no attention in application among adult survivors of childhood cancer.

Generalizable knowledge about the impact of cancer in adult survivors of childhood cancer also has been limited by studies in individualized settings, with relatively small survivor populations and lacking a valid comparison group. The Childhood Cancer Survivor Study (CCSS), a multi-institutional study of over 14,000 survivors in the North America [27], stands alone as the largest and most comprehensive cohort study of childhood cancer survivors and a sibling comparison group, but it is not without its limitations. While identifying prevalence and symptoms of psychological distress, the CCSS lacks specificity in reporting how cancer has affected, or continues to affect, young adults in specific and meaningful ways (from the survivor’s perspective). Thus, this manuscript represents the first report of the administration and psychometric evaluation of a cancer-specific measure that assesses aspects of young adulthood that may be influenced by a cancer history and have not been evaluated by existing standardized measures.

Method

Instrument development

Details regarding the development of candidate items for the Impact of Cancer—Childhood Survivor (IOC-CS) Scale are described elsewhere [28]. Briefly, qualitative interviews with 64 young adult survivors of childhood cancer (aged 18–35) resulted in the derivation of 82 candidate items selected for inclusion in a mailed survey intended to examine the psychometric properties of the new instrument. The items were organized under 12 headings representing the investigators’ theoretical notions of the emerging constructs suggested by the items: “Your Body and Your Health,” “Cancer Treatment and Health Care,” “Having Children,” “Who are You,” “Talking and Thinking About Cancer,” “Meaning of Cancer,” “Memory and Thinking,” “Finances and Money,” “Family,” “Relationships,” “Socializing and Being with Friends” and “Life Goals.” All items were scored on a five-point Likert-type scale ranging from “Not at all” to “Very Much.” Higher scores indicate greater impact. A “not applicable” category was added for the section on Finances and Money only. In most instances, items were worded such that impact is either inherently and intuitively positive (“I eat a healthy diet”) or negative (“I am afraid to die”). In some instances, however, one cannot assume that the impact is positive or negative. For example, the items “I wonder why I got cancer,” “People treat me differently after they find out I have had cancer” and “I wonder why I survived and others do not” are not inherently positive or negative.

Administration and evaluation of the IOC-CS

Following selection of items for the initial version of the IOC-CS measure, we administered them to a new sample of young adult survivors of childhood cancer. After obtaining Human Subjects (IRB) approval from each institution, a potential pool of 2,869 respondents was derived from electronic records maintained at three US hospitals that treat pediatric oncology patients. Study eligibility criteria included survivors who were off treatment and disease-free at the time of questionnaire completion, were between the ages of 18–39 years at time of study, and 21 years of age or younger when diagnosed with a malignancy. Potential subjects were mailed survey questionnaires, informed consent forms and a self-addressed and stamped return envelope.

Survey booklet

In addition to the IOC-CS, the mailed questionnaire also included several well-established health status and HRQOL measures with demonstrable evidence of reliability and validity in healthy populations [29], ill populations [30–32] and childhood cancer survivors [4, 7, 33, 34]. These measures were (1) the MOS Short-Form (SF-36), a widely used instrument that assesses health status with regard to physical, social and psychological functioning [32]; (2) the Brief Symptom Inventory-18 (BSI-18), another widely used instrument that assesses mental health status with regard to depression, anxiety and somatic distress [31]; (3) the Stress Symptom Questionnaire (PDS), a 17-item self-report measure of re-experiencing, avoidance and arousal as indicators of post-traumatic stress [29]; and (4) the Ladder of Life, a visual acuity scale ranging from 0 to 10 on which respondents indicate their satisfaction with life on a “ladder” ranging from “worst possible life” to “best possible life.”[35] Respondents also provided sociodemographic descriptive data and information about any current health problems, type of cancer and year of diagnosis.

Statistical analyses

The goal of this psychometric evaluation was to establish a valid set of scales measuring issues of concern for young adult survivors of childhood cancer, with subscales composed of internally consistent items. To determine theoretically and statistically derived dimensions and scales, we engaged in an iterative process consisting of (1) a priori categorization of items by domains (as presented in the survey booklet); (2) exploratory factor analyses to test psychometric properties of hypothesized scales; (3) item reduction; (4) re-scaling and re-testing (factor analyses); and (5) derivation and scoring of subscales. To score each scale, we computed the mean of items in that scale based on response categories ranging from 1 (not at all) to 5 (very much). For the five items representing Finances and Money, “not applicable” responses were assumed to indicate no impact to the respondent and thus recoded as 1 (not at all). Mean scores for each subscale were calculated such that higher values indicated a greater impact of cancer. Scales were evaluated for both thematic and statistical consistency. Internal reliability was evaluated using Cronbach’s alpha. Finally, we examined intraclass-correlation coefficients (ICC) for test–retest reliability in a subset of participants.

Construct validity of the IOC-CS was evaluated by examining Pearson correlations among the various derived subscales. Based on a priori hypotheses, we evaluated external (concurrent) validity by examining correlations of each of the IOC-CS subscales with the Physical Component Score (PCS), Mental Component Score (MCS) and eight subscales of the SF-36, and the subscale and overall scores for the BSI-18, PDS and Ladder of Life. Recommendations of Burnand and colleagues [36] were used to classify linear associations, using r < 0.30 as a negligible correlation; 0.30 ≤ r < 0.45 as moderate; 0.45 ≤ r < 0.60 as substantial and r ≥ 0.60 as high. Criterion validity was evaluated by review of cross-sectional differences between subjects for expected or known differences. IOC-CS sub-scale scores were examined as they differed (or not) across gender, employment status, relationship status, education, income, cancer type and reporting current health problems.

Results

Subject characteristics

Of 2,869 childhood cancer survivors identified through institutional records, 576 (20.1%) could not be located and were considered lost to follow-up. An additional 22 individuals were confirmed deceased. Among 2,271 live subjects located, 666 (29.3%) consented to participate and completed a self-report questionnaire. Of 666 completed surveys, 39 were eliminated from subsequent analyses due to surveys being completed by individuals who in fact did not fit eligibility criteria with regard to age at study (18–39 years old), age at diagnosis (21 years or younger) or treatment status (not currently receiving treatment), or because the questionnaire was completed by a surrogate. An additional 103 respondents did not complete portions of the survey necessary to conduct the psychometric analyses proposed here. Thus, these analyses are based on responses from 519 eligible participants who answered all questions needed to evaluate the candidate IOC-CS instrument. A comparison of these 519 respondents to everyone else in the sample pool showed that respondents were on average 1.1 year younger (p < .05) and 1.0 fewer year beyond diagnosis (p < .05) than all others (Table 1). Respondents also were significantly more likely to be women (p < .05) and survivors of hematologic cancers (p < .05), and less likely to be brain tumor survivors (p < .05). No statistically significant differences were observed with regard to age at diagnosis.

Table 1.

Sample characteristics of respondents and non-respondents

Variables	IOC respondents n = 519 Freq (%)	Non-respondentsan = 2,350 Freq (%)
Gender*
Female	269 (51.8)	985 (41.9)
Male	249 (48.0)	1365 (58.1)
Race/ethnicity
White	341 (65.7)	NA
Black	21 (4.0)
API	34 (6.6)
Hispanic/Latino	112 (21.6)
AmerInd	4 (0.8)
Employment status
Unemployedb	67 (12.9)	NA
Employed/occupiedc	450 (86.7)
Education
HS Grad or less	99 (19.1)	NA
Some college	234 (45.1)
4-year college grad	178 (34.3)
Marital status
No	284 (54.7)	NA
Yes	235 (45.3)
Income
≤$25K	176 (33.9)	NA
$25–$75K	210 (40.5)
> $75K	120 (23.1)
Cancer type*
Hematological	322 (62.0)	1131 (48.1)
CNS/brain tumor	62 (11.9)	498 (21.2)
Solid tumors/soft tissue tumors	116 (22.4)	517 (22.0)
Other	19 (3.7)	204 (8.7)
Reports current health problems (other than cancer related)
No	235 (45.3)	NA
Yes	278 (53.6)
Age at study, in years*
Mean (SD)	26.7 (5.3)	27.8 (6.1)
Range	18–39
Age at diagnosis, in years
Mean (SD)	11.3 (6.1)	11.1 (6.2)
Range	0–21
Years since diagnosis*
Mean (SD)	15.4 (6.9)	16.7 (7.3)
Range	2–37

^*Indicates statistically significant differences at p <0.05; NA not available from participating institutions; Percentages do not equal 100% due to missing data

^aNon-respondents include 1,605 subjects who did not return a questionnaire, 576 subjects who could not be located, 22 who were confirmed deceased, 39 who completed and returned a survey but did not actually fit eligibility criteria, and 108 survey respondents for whom missing data precluded their responses from being included in subsequent psychometric testing

^bIncludes those “on leave/disability,” “unemployed,” or “permanently unable to work”

^cIncludes “full-time employment,” “part-time employment,” “student” or “homemaker”

Scaling of the IOC-CS

For these analyses, we focused on 73 questions that were pertinent to all respondents, along with two additional items that were only applicable to respondents who have siblings, three items that were applicable only to respondents who have a spouse/committed partner, and four items that were only applicable to respondents who did not have a spouse/committed partner at the time of study. Only, the 73 IOC-CS items applicable to all participants underwent an iterative process of exploratory factor analyses and subsequent item reductions. In each iteration, principal axis factoring was followed by a varimax (orthogonal) rotation. Because our a priori categorization of the items resulted in 12 hypothesized domains, our initial run forced a 12-factor solution. Together these factors explained 86% of the common variance. Four of the 73 items had absolute loading coefficients <0.3 on every factor and thus were eliminated from further analyses. In subsequent iterations, the decision to retain factors was based on the scree test, the cumulative proportion of common variance explained (i.e., >85% in keeping with the initial run) and theoretical interpretability. A second factor analytic run included the 69 items retained from the prior analysis and yielded a 10-factor solution. Each item had at least one factor loading coefficient ≥0.3, but several items loaded on more than one factor with a coefficient ≥0.30. In addition, several factors containing items with coefficients in the 0.3–0.4 range were difficult to interpret theoretically; thus, we increased the criterion for retaining items to >0.4 and eliminated 14 additional items. A final iteration used the 55 remaining items and yielded an 8-factor solution. Two items with absolute loading coefficients <0.4 and four others with loading coefficients ≥0.4 on multiple factors were eliminated. Each of the retained factors contained at least three items, and together the factors explained 88% of the common variance. The eight factor-derived subscales consisting of 45 items described distinct theoretical constructs and are summarized in Table 2.

Table 2.

Eight factor-derived subscales consisting of 45 items on n = 519 subjects and descriptive statistics of all 11 subscales

Subscales and items	Item		Subscale
	Sourceb	Factor loadingc	Mean (SD)a	Min–Max
Life Challenges (α = 0.86)			2.28 (0.81)	1.0–4.7
5. Worry about health	Your Body/Health	0.42
31. Want to forget cancer	Thinking about cancer	0.52
34. Wonder why I got cancer	Thinking about cancer	0.67
35. Wonder why I survived	Thinking about cancer	0.58
36. Something I did caused cancer	Thinking about cancer	0.48
37. Angry about cancer	Thinking about cancer	0.67
39. Cancer controls my life	Thinking about cancer	0.58
41. Time is running out	Thinking about cancer	0.53
42. Afraid to die	Thinking about cancer	0.49
43. Worry I will die at young age	Thinking about cancer	0.59
79. Missed out on life	Socializing	0.50
82. Unsure about future	Life goals	0.55
Body and Health (α = 0.85)			3.44 (0.76)	1.3–5.0
1. Lead healthy life	Your Body/Health	0.68
2. Eat healthy diet	Your Body/Health	0.64
3. Exercise	Your Body/Health	0.64
4. Healthy as those w/o cancer	Your Body/Health	0.60
8. Believe I’m attractive	Your Body/Health	0.61
9. Like my body	Your Body/Health	0.67
23. Self-confident	Who are you	0.49
29. Feel in control	Who are you	0.45
Talking with Parents (α = 0.86)			3.89 (1.11)	1.0–5.0
62. Can talk with mom about cancer	Family	0.70
63. Can talk with dad about cancer	Family	0.86
64. Mom comfortable talking about cancer w/me	Family	0.68
65. Dad comfortable talking about cancer w/me	Family	0.81
Personal Growth (α = 0.73)			3.45 (0.92)	1.0–5.0
25. Cancer part of self	Who are you	0.57
27. More mature than those without cancer	Who are you	0.47
40. Special bond with others with cancer	Thinking about cancer	0.51
44. Good things came from cancer	Meaning of cancer	0.66
45. Learned about self	Meaning of cancer	0.71
Thinking/Memory Problems (α = 0.79)			2.30 (0.89)	1.0–5.0
48. Easy to make decisionsd	Memory/thinking	−0.50
49. Easy to learnd	Memory/thinking	−0.66
50. Hard time thinking	Memory/thinking	0.68
51. Trouble w/long-term memory	Memory/thinking	0.57
52. Trouble w/short-term memory	Memory/thinking	0.63
Health Literacy (α = 0.74)			3.58 (0.85)	1.2–5.0
12. Know who to see for med problems	Cancer tx/health care	0.47
14. Feel doctor knows cancer effects	Cancer tx/health care	0.51
15. Easy to talk to doctor about cancer	Cancer tx/health care	0.40
16. Have all cancer info I need	Cancer tx/health care	0.74
17. Know where to find cancer info	Cancer tx/health care	0.70
Socializing (α = 0.75)			3.97 (0.95)	1.0–5.0
75. Make friends easily	Socializing	0.41
76. Avoid social activitiesd	Socializing	−0.48
77. Left out of friends’ livesd	Socializing	−0.45
Financial Problems (α = 0.70)			1.64 (0.93)	1.0–5.0
55. Financial Problems from cancer	Finances	0.52
56. Parents Financial Problems from cancer	Finances	0.41
57. Trouble getting assistance/services	Finances	0.49
Sibling Concernse (α = 0.62) (n = 477)			1.90 (1.00)	1.0–5.0
66. Worry how cancer affected siblings	Siblings
67. Sibling had problems related to my cancer	Siblings
Relationship Concerns (non-partnered)e (α = 0.74) (n = 280)			2.26 (1.02)	1.0–5.0
68. Worry about having no relationship	Relationships
69. Worry telling potential partner about cancer	Relationships
70. Worry telling potential partner about fertility	Relationships
71. Worry about having sex	Relationships
Relationship Concerns (partnered)e (α = 0.52) (n = 234)			1.48 (0.67)	1.0–4.0
72. Can talk to partner about health problemd	Relationships
73. Worry partner will leave if cancer recurs	Relationships
74. Worry about having sex	Relationships

^aMean scores derived from the following response categories: 1 = not at all, 2 = a little bit, 3 = somewhat, 4 = quite a bit, 5 = very much

^bCategory in which the item appeared in the mailed survey

^cWhere item loaded above 0.40 and no greater than 0.40 on any other factor and maintained a clear theoretical interpretation

^dScoring on this item was reversed prior to calculating scale mean scores

^eScaled separately

Five subscales (Body and Health, Talking With Parents, Personal Growth, Health Literacy, Socializing) clearly reflected a positive impact. Two subscales (Thinking/Memory Problems, Financial Problems) clearly reflected negative impacts. One subscale—Life Challenges—consisted primarily of items reflecting a clearly negative impact but also contained several neutrally worded items. Prior to calculating mean scores for all subscales, four items were reversed-scored to maintain a consistent direction of effect (see Table 2). Ceiling and floor effects (disproportionately high or low numbers of respondents scoring at one of the two ends of the scale) existed in some cases. Minimal-to-moderate floor effects were observed for Personal Growth (1.2%), Socializing (1.7%), Talking With Parents (2.9%), Thinking/Memory Problems (6.7%) and Financial Problems (51%). Minimal-to-moderate ceiling effects were observed for Financial Problems (1.2%), Personal Growth (2.9%), Health Literacy (5.4%), Socializing (20.8%) and Talking With Parents (33.9%). Overall, respondents endorsed a full range of responses for all items, with item distributions for positive impacts tending to be skewed toward greater positive impact, and distributions for negative impact subscales tending to be skewed toward less negative impact. Cronbach’s alphas ranged from 0.70 to 0.86, suggesting adequate reliability [37]. None of the scales showed an improvement in their internal consistency when any item was removed; thus, no additional questions were dropped from the instrument.

The sibling and relationship items in the IOC-CS were scaled separately after content review and assessment of internal reliability. Cronbach’s alphas for the Sibling Concerns, Relationship Concerns (not partnered) and Relationship Concerns (partnered) subscales were 0.62, 0.74 and 0.52, respectively. Higher scores indicate greater negative impact (worry). Scale content and summary statistics for each of these subscales are included in Table 2.

Analysis of test–retest reliability used data from 136 respondents who returned a repeated administration of the IOC-CS within 3 weeks of completing the initial survey booklet. Test–retest reliability was well-substantiated [38] with ICC ≥ 0.75 for the full instrument and for 9 of the 11 subscales (see Table 3). Test–retest reliability for the Finances and Money subscale and Relationship Concerns subscale (non-partnered) also was adequate, with ICC = 0.62 and 0.73, respectively.

Table 3.

Test–retest reliability for the IOC-CS

	Initial		Retest
	Mean	(SD)	Mean	(SD)	ICC
Life Challenges	2.22	(0.80)	2.22	(0.80)	0.897
Body/Health	3.48	(0.77)	3.48	(0.76)	0.914
Talking With Parents	4.00	(1.09)	3.94	(1.07)	0.829
Personal Growth	3.49	(0.87)	3.54	(0.93)	0.829
Thinking/Memory Problems	2.37	(0.92)	2.31	(0.93)	0.885
Health Literacy	3.63	(0.85)	3.67	(0.80)	0.755
Socializing	4.00	(0.98)	4.00	(0.91)	0.881
Financial Problems	1.54	(0.84)	1.68	(0.90)	0.622
Sibling Concerns	1.98	(1.05)	2.09	(1.00)	0.782
Relationship Concerns (non-partnered)	2.20	(0.85)	2.14	(0.96)	0.727
Relationship Concerns (partnered)	1.50	(0.63)	1.56	(0.72)	0.798

To evaluate construct validity, we examined the Pearson product-moment correlations among the subscale scores (Table 4). Based on existing literature suggesting interrelationships among physical, psychological, social and spiritual/existential domains of QOL [39, 40], we expected to observe such associations among related subscales of the IOC-CS. Indeed, the Life Challenges subscale was moderately associated with the Body and Health, Thinking/Memory Problems, Socializing and Financial Problems subscales: as respondents indicated more Life Challenges they tended to report less positive impact in terms of their Body and Health and Socializing. Life Challenges also were associated with reporting more Thinking/Memory Problems and financial concerns. The Socializing subscale was at least moderately associated with Body and Health, Talking With Parents and Thinking/Memory Problems subscales in that more positive Socializing was associated with more positive communication with parents and less Thinking/Memory Problems. Thinking/Memory Problems and Body and Health subscales were moderately correlated, with respondents reporting a more positive Body and Health impact also tending to report less thinking and memory problems. The Personal Growth subscale was not significantly associated with any other subscale: Personal Growth appeared to be independent of any other impact of cancer.

Table 4.

Subscale correlation matrix for the IOC-CS instrument

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)
(1) Life Challenges (−)	1.00
(2) Body/Health (+)	−0.42	1.00
(3) Talk with parents (+)	−0.20	0.28	1.00
(4) Personal Growth (+)	0.06	0.18	0.14	1.00
(5) Thinking problems (−)	0.39	−0.42	−0.13	−0.01	1.00
(6) Health Literacy (+)	−0.19	0.30	0.28	0.22	−0.24	1.00
(7) Socializing (+)	−0.39	0.51	0.32	0.14	−0.38	0.28	1.00
(8) Financial Problems (−)	0.36	−0.20	−0.21	0.12	0.25	−0.16	−0.21	1.00
(9) Sibling Concerns (−)	0.24	−0.13	0.00	0.31	0.14	0.01	−0.11	0.20	1.00
(10) Relationship Concerns (non-partnered) (−)	0.51	−0.30	−0.18	0.20	0.24	−0.12	−0.42	0.35	0.27	1.00
(11) Relationship Concerns (partner) (−)	0.46	−0.36	−0.34	0.03	0.20	−0.25	−0.38	0.27	0.20	N/A	1.00

Bold figures represent substantial to high correlations (r ≥ 0.45), and bold italicized figures represent moderate correlations (0.30 ≤ r <0.45), as defined by criteria determined by Burnand et al. [36]

“+” Indicates subscale suggestive of positive impact

“−” Indicates subscale suggestive of negative impact

Regardless of relationship status, reporting Relationship Concerns was moderately to highly associated with reporting more Life Challenges and worse outcome with regard to Body and Health. Relationship Concerns also appeared moderately associated with Socializing, in that survivors reporting more positive Socializing tended to report less concern about relationships.

Associations of the IOC-CS scales with other outcome measures

Concurrent validity of the IOC-CS was evaluated by examining the associations of each of the subscales with the SF-36, PDS, BSI-18 and the Ladder of Life using Pearson product-moment correlations (Table 5). We observed moderate to substantial correlations among several IOC-CS subscales (Life Challenges, Body and Health, Thinking/Memory Problems, Socializing and Relationship Concerns) and SF-36 subscales. We also noted that the linear associations between IOC-CS subscales and the Physical Component Score (PCS) of the SF-36 were always weaker (as indicated by Pearson r) than were associations of the IOC-CS subscales and the Mental Component Score (MCS).

Table 5.

Correlation of new IOC-CS subscales with existing QOL and psychological well-being outcomes

	Impact of cancer (IOC-CS) subscales
	Life Challenges (−)	Body/Health (+)	Talk w/ parents (+)	Personal Growth (+)	Think/memory problems (−)	Health Literacy (+)	Social (+)	Finance problems (−)	Sibling Concerns (−)	Relationship (non-partnered) (−)	Relationship (partnered) (−)
SF-36
Physical function	−0.26	0.29	0.04	0.00	−0.20	0.04	0.22	−0.18	−0.18	−0.18	−0.25
Role-physical	−0.33	0.29	0.11	−0.03	−0.27	0.12	0.27	−0.23	−0.18	−0.21	−0.38
Bodily pain	−0.33	0.40	0.14	−0.03	−0.25	0.21	0.25	−0.25	−0.21	−0.22	−0.29
General health	−0.44	0.60	0.16	−0.03	−0.37	0.21	0.32	−0.24	−0.25	−0.26	−0.38
Vitality	−0.40	0.62	0.19	0.00	−0.43	0.28	0.48	−0.18	−0.19	−0.35	−0.28
Social functioning	−0.47	0.47	0.15	−0.04	−0.41	0.22	0.50	−0.25	−0.18	−0.42	−0.42
Role-emotional	−0.36	0.39	0.12	−0.03	−0.36	0.17	0.41	−0.15	−0.20	−0.35	−0.35
Mental health	−0.46	0.60	0.18	0.02	−0.44	0.29	0.53	−0.17	−0.13	−0.43	−0.34
Physical component score (PCS)	−0.28	0.31	0.08	−0.03	−0.20	0.09	0.15	−0.25	−0.22	−0.12	−0.28
Mental component score (MCS)	−0.44	0.54	0.17	−0.01	−0.44	0.26	0.53	−0.15	−0.15	−0.43	−0.35
Brief symptom inventory (T-scores)
Global (GSI)	0.51	−0.56	−0.18	0.01	0.48	−0.31	−0.49	0.30	0.16	0.46	0.38
Somatic distress	0.43	−0.44	−0.09	0.06	0.41	−0.27	−0.32	0.26	0.17	0.31	0.37
Depression	0.49	−0.55	−0.22	−0.04	0.46	−0.29	−0.53	0.25	0.17	0.48	0.37
Anxiety	0.46	−0.46	−0.16	0.05	0.41	−0.23	−0.42	0.28	0.12	0.41	0.33
PDS
Overall severity	0.58	−0.47	−0.22	0.11	0.47	−0.28	−0.55	0.30	0.21	0.51	0.44
Re-experiencing	0.52	−0.26	−0.22	0.15	0.22	−0.13	−0.32	0.30	0.21	0.40	0.38
Avoidance	0.54	−0.48	−0.20	0.07	0.38	−0.29	−0.56	0.27	0.21	0.52	0.46
Arousal	0.43	−0.41	−0.15	0.07	0.54	−0.25	−0.45	0.22	0.12	0.38	0.27
Ladder of life
Present	−0.42	0.55	0.24	0.11	−0.34	0.27	0.47	−0.18	−0.06	−0.31	−0.45

Bold figures represent high correlations r ≥ 0.6; bold italicized figures represent substantial correlations (0.45 ≤ r <0.60); italicized figures represent moderate correlations (0.30 ≤ r < 0.45), as defined by criteria determined by Burnand et al. [36]

“+” Indicates subscale suggestive of positive impact

“−” Indicates subscale suggestive of negative impact

For the most part, 7 of the 11 IOC-CS subscales were moderately to substantially associated with distress, and respondents’ impact scores changed in expected directions as distress scores increased (Table 5). We also observed moderate to substantial correlations, as expected, between life satisfaction (as measured by the Ladder of Life) and 6 of the 11 IOC subscales. Life satisfaction appeared lower in relation to reporting greater Life Challenges, Thinking/Memory Problems and Relationship Concerns. Life satisfaction was higher in survivors who reported a more positive outcome in terms of Body and Health and Socializing. In addition, we observed moderate associations between the Financial Problems subscale and the global distress measures. Unexpectedly, linear associations between distress and IOC subscales assessing Talking With Parents, Personal Growth and Health Literacy were notably low, suggesting that for the most part distress is not a function of survivors’ ability to talk with parents about cancer, their reportage of Personal Growth, or Health Literacy.

Association of IOC-CS scales with other independent variables

Existing childhood cancer survivor studies report differences in psychological well-being and HRQOL outcomes by gender, socioeconomic status variables, cancer type and physical health status [5, 7, 41, 42]. Thus, in evaluating criterion validity, we hypothesized that there would be similar differences in IOC-CS subscales in that greater impacts would be experienced by respondents who were women, non-partnered, of lower socioeconomic status (SES), brain tumor survivors, and by those reporting physical health problems at time of study (Table 6).

Table 6.

Mean subscale scores on the IOC-CS instrument by demographic and health status characteristics

	Gender		Currently employed		Currently in relationship		Education			Income			Cancer type				Health problems
	Female n = 269	Male n = 249	Yes n = 450	No n = 67	Yes n = 235	No n = 284	≤HS n = 99	Some coll n = 234	≥4 yr n = 178	≤$25 K n = 176	$25– 75 K n = 210	> $75 K n = 120	Heme n = 322	Brain/CNS n = 62	Soft tissue/solid tumor n = 116	Other n = 19	No n = 262	Yes n = 175
Life Challenges (−)	2.26	2.30	2.23	2.64	2.25	2.31	2.44	2.35	2.11	2.30	2.36	2.09	2.28	2.37	2.23	2.38	2.05	2.55
Body/Health (+)	3.43	3.44	3.49	3.09	3.56	3.34	3.22	3.41	3.61	3.31	3.43	3.62	3.47	3.36	3.40	3.31	3.64	3.25
Talking With Parents (+)	3.76	4.04	3.92	3.72	3.96	3.84	3.87	3.91	3.87	3.80	3.89	4.00	3.95	3.82	3.82	3.70	3.97	3.81
Personal Growth (+)	3.57	3.31	3.48	3.25	3.51	3.40	3.24	3.49	3.52	3.38	3.47	3.51	3.51	3.43	3.27	3.53	3.40	3.52
Thinking/Memory Problems (−)	2.32	2.28	2.28	2.47	2.18	2.41	2.48	2.34	2.16	2.49	2.22	2.17	2.25	2.73	2.22	2.28	2.10	2.50
Health Literacy (+)	3.63	3.53	3.60	3.46	3.61	3.56	3.63	3.60	3.53	3.48	3.65	3.61	3.64	3.47	3.48	3.63	3.66	3.59
Socializing (+)	3.99	3.95	4.03	3.58	4.12	3.85	3.79	3.97	4.09	3.83	3.97	4.19	4.02	3.55	4.03	4.19	4.21	3.72
Financial Problems (−)	1.65	1.63	1.57	2.02	1.63	1.65	1.58	1.71	1.58	1.82	1.61	1.42	1.58	1.85	1.68	1.67	1.42	1.91
Sibling Concerns (−)	2.11	1.68	1.88	1.98	1.91	1.90	1.79	1.95	1.90	1.90	2.02	1.72	1.93	1.99	1.76	2.09	1.73	2.15
Relationship Concerns (non- partnered) (−)	2.40	2.12	2.24	2.40	N/A		1.99	2.33	2.37	2.25	2.42	2.05	2.23	2.35	2.35	1.82	1.98	2.53
Relationship Concerns (partnered) (−)	1.55	1.38	1.45	1.69	N/A		1.41	1.58	1.39	1.64	1.43	1.44	1.46	1.68	1.47	1.29	1.30	1.65

p-Values from ANOVA tests. Bold items indicate statistically significant differences at p < 0.05

Table 6 presents our findings related to gender, relationship status, SES, cancer type and physical health problems. Consistent with the literature cited above, non-partnered respondents reported significantly less positive impact on Body and Health and Socializing, and more negative impact on Thinking/Memory Problems. Female survivors reported significantly greater impact in terms of Personal Growth and relationships. For the most part, IOC-CS subscale scores did not vary across cancer type for 9 of the 11 subscales; however, brain tumor survivors reported greater negative impact on Thinking/Memory Problems and less positive impact in Socializing. In terms of SES and consistent with literature cited above, respondents who were employed and of greater educational attainment and income also tended to report fewer Life Challenges and more positive outcomes with regard to Body and Health and Socializing. Respondents of lower education and income levels reported more Thinking/Memory Problems. Greater Financial Problems were indicated by respondents who were not currently employed and of lower income. Personal Growth appeared more likely among respondents with more than a high school education. Survivors reporting health problems also reported more Life Challenges, Thinking/Memory Problems and Financial Problems and a less positive impact with regard to Body and Health and Socializing.

Discussion

In this report, we provide results from a psychometric evaluation of a new instrument to assess important aspects of cancer survivorship for young adult survivors of childhood cancer. In the development of the IOC-CS, we specifically decided not to duplicate content from other well-validated measures that include physical, emotional and social functioning domains of HRQOL, and instead focused on concepts within these domains that were specific to the experience of this young adult population and identified in prior research [28]. The psychometric evaluation presented here suggests that the IOC-CS consists of reliable and valid subscales in that we observed expected significant and substantial associations (convergence and divergence) among IOC-CS subscales and existing standardized HRQOL-related measures. For example, with regard to employment, education and income, findings reflect a well-documented correlation of socioeconomic status and HRQOL, and thus serve as evidence supporting the external validity of the IOC-CS. Significant differences in IOC-CS subscales across relationship status, cancer type and overall health criteria that are consistent with existing studies of HRQOL also suggest that subscales measuring specific aspects of young adult survivors’ lives are sensitive to differences in childhood survivors’ cancer diagnosis, as well as current health and social status.

We also noted some unexpected findings suggesting that the IOC-CS is tapping into new and previously unreported aspects of long-term survivorship. Specifically, the IOC-CS subscales measuring Health Literacy and Personal Growth had little to no statistical relationship to any other HRQOL measure included here, or to any other IOC-CS subscale. Thus, the IOC-CS appears capable of assessing aspects of survivors’ lives beyond that of previously developed generic or disease-specific measures.

In comparing IOC-CS subscales to existing standardized measures of physical, psychological and social well-being, we noted that the IOC-CS subscales were always more highly correlated with measures of psychological well-being and mental health than with measures indicative of physical health function, namely the Physical Component Score of the SF-36. Relatively higher correlations among the IOC-CS subscales and measures of mental health functioning and distress suggest that the IOC-CS is more closely related to survivors’ psychological and subjective experience, and perhaps reflects survivors’ perceptions (as opposed to objective realities) of the way cancer has affected them and continues to affect them. Indeed, perceptions of one’s experience have been demonstrated to be better predictors of HRQOL outcomes than objective measures [43, 44]. In addition, these findings suggest some questions to be examined in future studies of long-term survivors: Are changed perceptions of body image associated with decreased HRQOL? Are positive self-evaluations and life outlooks more likely to occur in survivors diagnosed as adolescents than as younger children? Might experiencing and reporting positive impacts of cancer attenuate the influence of perceived negative effects on psychosocial or HRQOL outcomes?

The floor and ceiling effects observed for several of the subscales suggest a tendency among these respondents to emphasize positive and minimize negative impacts. This finding could reflect a response bias in that respondents selectively forget, ignore or deny the negative aspects of their illness and tend to evaluate their past selves in a manner that makes them feel good about themselves now [45]. However, one also may interpret this notion of “response shift” differently. It may be that tendencies to report more positive and less negative outcomes reflect a phenomenon in which some survivors truly grow and experience enhanced HRQOL as a result of having had cancer [15]. Certainly, there is debate and discussion about the meaning and existence of Personal Growth, positive adaptation and post-traumatic growth in cancer that will continue for some time to come [46].

The relatively small response rate and underrepresentation of race/ethnic minorities and brain tumor survivors in this sample limits the generalizability of these findings. We emphasize the need for further work to establish the validity and predictive ability of the IOC for the population of young adults survivors of childhood cancer and demonstrate the extent to which the scales measure constructs that are not just correlated with respondents’ current psychological well-being but actually predict subsequent health and well-being. Replication is needed not only with better representation of underrepresented subgroups but also to determine the stability, sensitivity and specificity of the subscales. Factor analytic or latent class analyses with replication samples would help determine whether the large number of subscales can be condensed and the measure refined.

Conclusion

For young adult survivors of childhood cancer, the ability to prevent deleterious long-term effects and promote adaptive psychosocial functioning requires an understanding of how the perceived impact of cancer intersects with the unique developmental challenges of adolescence and young adulthood [47]. Used as a complement to other health status and HRQOL instruments, the IOC-CS has the potential to measure specific aspects of long-term cancer survivors’ lives, and it may be predictive of other outcomes of interest, such as psychological well-being or mental health status. A clinical application of the IOC-CS may be in its ability to assess individual survivor concerns and guide the implementation of evidence-based supportive care interventions. Further evaluation, refinement and administration of the IOC-CS in representative and culturally diverse samples of long-term survivors will help to establish its validity, reliability, sensitivity to change over time and utility, along with its potential use in cancer survivorship clinical trials.