|Home | About | Journals | Submit | Contact Us | Français|
Recent work suggests that perceptions of the impact of cancer on survivors’ lives are associated with physical and mental health and quality of life (QOL) outcomes. This study examines the association between the Impact of Cancer version 2 scales (IOCv2) and these outcomes in a large sample of survivors of adult non-Hodgkin lymphoma (NHL).
Participants completed a mailed survey to assess physical and mental health (SF-36), cancer-specific quality of life (FACT-G) and perceived impact of cancer (IOCv2). Hierarchical multiple regression models, in which demographic, clinical, psychosocial and IOCv2 measures were added sequentially, were employed to evaluate their contribution to explaining variance in SF-36 and FACT-G scores.
A total of 652 post-treatment NHL survivors participated. Survivors with comorbidities and negative appraisals of life threat and treatment intensity reported worse physical and mental health and QOL (all P<.05). After controlling for demographic and clinical characteristics, younger respondents reported better physical but worse mental health and QOL (all P<.01). Lower IOCv2 Negative Impact (all P<.001) and higher Positive Impact (all P<.05) scores were associated with better physical and mental health and QOL after controlling for demographic, clinical and psychosocial characteristics.
Findings suggest that perceptions of cancer’s impact on survivors’ lives may influence or be influenced by health status and functioning and QOL. Longitudinal research is needed to establish causality, which could lead to the development of interventions targeting survivors’ impact of cancer concerns, and ultimately to the enhancement of overall health and QOL.
The National Cancer Institute  estimates that there are 11.4 million cancer survivors in the US, representing about 3.8 percent of the population. Coinciding with this growing population is an increased interest in the quality of life (QOL) of individuals living long-term with a history of cancer. Studies have shown that younger age [2-4], lower income [5-7], and less education [3, 8] and social support [3, 6, 8], for example, are associated with psychological distress and/or poorer QOL in adult cancer survivors. In addition, clinical characteristics such as time since diagnosis [3, 9], comorbidity [3, 7] and chemotherapy treatment [9, 10] have been associated with QOL-related outcomes in cancer survivor samples.
Recent work suggests that positive and negative perceptions of the impact of cancer on health and well-being are also associated with physical and mental health functioning and QOL outcomes . These associations have been observed in a mixed cancer survivor sample using an 81-item instrument, the Impact of Cancer (IOC), designed specifically to measure the unique issues (notably, both positive and negative impacts of cancer) associated with long-term survivorship . Studies that employed the IOC stand in contrast to many other survivorship studies to date, which have largely focused on pathological or negative responses (e.g., depression and anxiety), identified through the use of general and cancer-related QOL instruments.
A refined and shortened version of the IOC instrument, the 47-item IOCv2, was recently validated in breast cancer survivors  and found to be robust in non-Hodgkin lymphoma (NHL) survivors . With the availability of a large sample of off-treatment survivors of NHL, in this paper we assessed the contribution of survivors’ perceptions of the positive and negative impacts of cancer as measured by the IOCv2 scales to explaining QOL-related outcomes.
Patients were eligible if they were diagnosed with adult NHL (≥19 years old), ≥2 years post-diagnosis, and currently off-treatment and in remission or cured. The Duke Cancer Center and University of North Carolina Lineberger Tumor Registries were used to identify potential participants following approval by their Institutional Review Boards. Prospective subjects were mailed a package which included a consent form, letter of introduction from their oncologist, and self-administered questionnaire; a $2 bill was provided as an incentive. Non-respondents were sent replacement mailings and later telephoned to confirm receipt of the mailed survey.
Demographic and clinical characteristics were self-reported by participants. Histology was categorized as indolent or aggressive based on the REAL/WHO classification system . A 12-item self-report version of the Charlson Index (Self-administered Co-morbidity Questionnaire) was used to assess other health-related problems .
The 20-item Medical Outcomes Study-Social Support Survey (MOS-SSS) measures perceived availability of social support  and has been used in various populations, including long-term breast cancer survivors [6, 8]. The seven-item Appraisal of Life Threat and Treatment Intensity Questionnaire (ALTTIQ) assesses the extent to which cancer and its treatment are perceived to be life-threatening and intense in the past and currently . A 24-item instrument developed by the CALGB clinical research group (TOTRELAT)  was used to assess cancer-related employment and insurance-related issues.
The IOCv2 was used to measure perceptions of positive and negative impacts of cancer on aspects of the survivors’ lives. The IOCv2 uses 37 items to measure four positive (Altruism/Empathy, Health Awareness, Meaning of Cancer, and Positive Self-Evaluation) and four negative (Appearance Concerns, Body Change Concerns, Life Interferences, and Worry) subscales, which total to two summary scores (Positive and Negative Impact) . Development of the IOC was spurred by the need to measure aspects of survivorship not addressed by existing QOL measures, such as health-related worries and post-traumatic growth . In this study, internal consistency for the Positive and Negative Impact summary scores were α=0.90 and 0.91, respectively. Higher scores on the Positive Impact indicate greater positive impacts, while higher scores on the Negative Impact indicate greater negative impacts. Correlation coefficients for the IOCv2 subscales and the outcome measures used in this study are reported elsewhere .
A general health measure, the Medical Outcomes Study (MOS) SF-36, was used to assess physical and mental health status and functioning. The 36 items represent eight sub-scales (physical functioning, role limitations due to physical problems, bodily pain, general health perceptions, vitality, social functioning, role-limitations due to emotional problems, and mental health) and two summary scores, the physical component (PCS) and the mental component (MCS) . The 27-item Functional Assessment of Cancer Therapy (FACT-G) was used to assess cancer-specific QOL. The FACT-G contains four subscales (physical, social/family, emotional, and functional well-being) which may be summed to produce a total score .
Bivariate analyses (t-tests, analysis of variance) compared mean values of the outcome measures (PCS, MCS, FACT-G) and IOCv2 Positive and Negative Impact Summary scores by demographic and clinical characteristics. Pearson correlations were calculated between the outcome measures and IOCv2 summary scores and all continuous covariates. Hierarchical multiple regression models in which sets of covariates were added sequentially were employed to evaluate their contribution to explaining variance in PCS, MCS and FACT-G scores, yielding a series of five regression models for each outcome variable. Model 1 included demographic characteristics, Model 2 added clinical variables, Model 3 added psychosocial variables (MOS-SSS, ALTTIQ and TOTRELAT), Model 4 added IOCv2 Negative Impact Summary score, and Model 5 added IOCv2 Positive Impact Summary score. Partial F-tests were conducted to compute the significance of each set of added variables in explaining the remaining variance in the outcomes after controlling for the other variables previously added to the model. All statistical analyses were conducted using SAS Version 9.1 (SAS Institute Inc., Cary, NC).
Sample characteristics. Of the 1312 eligible survivors who were mailed a survey, 117 (9%) packages were returned undelivered and untraceable. Of the remaining 1195 survivors, 886 (74%) completed and returned their surveys, 258 (22%) did not respond to the mailing, and 51 (4%) refused participation Those who reported being off-treatment and in remission or cured were included in the analyses (n=652). Sample bias analyses using demographic information from the registries indicated that participating survivors were less frequently African American (8% vs. 20%, P<.001), older at study enrollment (mean age 62.7 vs. 58.8 years, P<.001), and older at diagnosis (51.9 vs. 48.1 years, P<.001) than non-participants.
As indicated in Table 1, a similar number of females and males participated, 12% were non-Caucasian, 26% earned <$30,000 annually, 40% had a college degree, and 41% were employed. Mean age at study enrollment was 62.7 and almost half (46%) were older adults (≥ 65 years of age). The mean time from diagnosis to study enrollment was 10.8 years (SD 7.5) while the mean age at lymphoma diagnosis was 51.9 (SD 14.2). Almost half of the participants (48%) had received radiation therapy and the majority (82%) had received chemotherapy. Most participants (55%) were diagnosed with an aggressive form of NHL. Participants cited an average of 2.9 (SD 2.1) co-morbid conditions. Scores for other scales were: ALTTIQ 23.8 (SD 6.4), PCS 46.1 (10.8), MCS 49.6 (11.1), FACT-G 87.8 (15.2), IOCv2 Positive 3.5 (0.8), and IOCv2 Negative 2.1 (0.7).
Bivariate associations between physical (PCS) and mental (MCS) health status and functioning and QOL (FACT-G) scores and the independent variables are shown in Table 1. Among categorical demographic and clinical variables, those who had an annual income under $30,000, did not obtain a college degree, were not married, were not employed or retired, or were older at study enrollment had lower PCS scores (all P<.05). Concerning mental health status, those who had an annual income under $30,000, were younger at study enrollment, or had received a bone marrow or stem cell transplant had lower MCS scores (all P<.05). Regarding QOL, those who were non-Caucasian, had an annual income under $30,000, did not obtain a college degree, were younger at study enrollment, or had received a bone marrow or stem cell transplant had lower FACT-G scores (all P<.05). Among the continuous variables, the strongest PCS relationships were for age at diagnosis (r=−0.27) and co-morbidity (r=−0.57) and the strongest MCS and FACT-G relationships were for co-morbidity (r=−0.26; −0.38) and social support (r=0.37; 0.51), respectively (all P<.0001).
Bivariate associations between IOCv2 Positive and Negative Impact scores and the independent variables are shown in Table 1. Among demographic and clinical variables, those who were male, were Caucasian, had an annual income ≥ $30,000, had a college degree, were not employed or retired, were older at study enrollment, did not receive chemotherapy, or did not receive a bone marrow or stem cell transplant had lower IOCv2 Positive Impact scores (all P<.05). In addition, those who were younger at study enrollment, were Stage IV at diagnosis, had received chemotherapy, had undergone a bone marrow or stem cell transplant, had received biologic therapy, or had one or more NHL recurrences had higher IOCv2 Negative Impact scores (all P<.05). Among the continuous variables, the strongest IOCv2 Positive Impact association was with appraisal of life threat and treatment intensity (r=0.34; P<.001). The strongest IOCv2 Negative Impact associations (all P<.0001) were with social support (r=−0.29), appraisal of life threat and treatment intensity (r=0.44), and employment and insurance issues related to having had cancer (r=0.28).
Tables Tables22--44 display the results of hierarchical regressions showing the contribution of demographic, clinical and psychosocial variables and IOCv2 Negative and Positive Impacts Summary scores to explaining variance in SF-36 PCS, MCS and FACT-G scores. The following demographic variables were independently associated (all P<.05) with at least one outcome: race, college and employment status, and age at enrollment. Only two clinical (received chemotherapy and comorbidity) and one psychosocial (social support) variables were significantly related to an outcome after all of the covariates were entered into the regression (Model 5). Both IOCv2 Negative and Positive Impact summary scores contributed significantly to the amount of variance explained (R2) in all three models (all P<.05; Tables Tables22--44).
As indicated in Table 2, education, employment status, age at study enrollment, chemotherapy receipt and comorbidity score remained significant predictors for PCS in all models after controlling for other variables. ALTTIQ was a significant predictor in Model 3 but became non-significant when IOCv2 Negative Impact Summary score was added to the model. IOCv2 Negative Impact Summary (P<.0001) and Positive Impact Summary (P<.05) scores explained a highly significant proportion of the variance in PCS after controlling for demographic, clinical and psychosocial variables. In total, 46% of the variance was accounted for by the covariates in Model 5.
Per Table 3, survivors who were younger at study enrollment, had comorbid health conditions and less social support were more likely to report worse mental health status after controlling for other variables. Having received a bone marrow or stem cell transplant and negative appraisals of life threat and treatment intensity were associated with worse mental health status in Model 3 (all P<.05), but the associations became non-significant with the addition of the IOCv2 scores. Similar to the pattern seen with PCS, IOCv2 Negative (P<.0001) and Positive Impact (P<.01) Summary scores added a highly significant proportion of the variance in MCS after controlling for demographic, clinical and psychosocial variables (Model 5). The full model explained 36% of the variance in MCS scores.
As displayed in Table 4, survivors who were non-Caucasian, did not receive a college degree, were not employed, were younger at study enrollment, had comorbid health conditions and less social support reported worse QOL when controlling for other variables (Model 5). Male gender, years since diagnosis, having received a bone marrow or stem cell transplant, and negative appraisals of life threat and treatment intensity were associated with worse QOL in some models (all P<.05), but these associations became statistically non-significant with the addition of the IOCv2 scores. The amount of variance explained by the full model (67%) was significantly increased with the addition of the IOCv2 summary scales (all P<.0001), with both the Negative and Positive Impact scales making a significant contribution while controlling for the other scale.
This study provides the first examination of the relationship between the recently validated IOCv2 and health status and functioning and QOL in a large sample of post-treatment NHL survivors. This analyses extends prior IOC research with mixed  and breast  cancer samples by using a robust set of covariates (demographic, clinical, psychosocial) and employing an NHL sample with good gender, race, and diagnosis (indolent and aggressive NHL) distributions, thereby enhancing generalizability.
Findings from the bivariate analyses indicate statistically significant relationships between QOL-related outcomes and select demographic (race, income, education, marital and employment status, age), clinical (treatment types, comorbidity), and psychosocial (social support, appraisals, and insurance and employment issues related to cancer) characteristics, which is consistent with previous studies of long-term cancer survivors [2-10]. However, many of these relationships became non-significant with the introduction of the IOCv2 scores in the multivariate models (Tables (Tables22--4).4). For example, only age, comorbidity and social support remained independently associated with the mental health status and functioning component summary score (MCS) after the IOCv2 Positive and Negative Impact Summary scores were entered in the hierarchical regression. A possible explanation for older respondents reporting better QOL than the younger respondents is that they are likely to have more experience coping with stressful events and fewer family-related and other demands than younger individuals.
In only one model (FACT-G) did a variable (race) gain significance after accounting for the IOCv2 scores (Positive Impact). It is possible that differences remained between the Caucasian and non-Caucasian cancer survivors that are not accounted for in the set of covariates used in this study and that were evident only after considering the IOCv2 in the hierarchical regression model. Further, it seems that the positive IOC that was statistically significantly related to race was largely operative for QOL among non-Caucasians.
Consistent with Zebrack et al. , our findings indicate that individuals’ perceptions of the positive and negative IOC on various life domains are related to their health status and functioning and QOL. In addition, accounting for IOCv2 scores reduced or diminished associations between demographic, clinical and psychosocial variables and the outcome measures in most instances. Furthermore, our findings indicate that positive IOC was significantly associated with PCS (P<.05; Table 2) after controlling for all other variables, which was the only deviation from the previous study . Finally, in only one instance did a demographic or clinical variable meet the criteria for minimally important difference  on the SF-36 scales after accounting for all variables; being presently employed added 3 points (P<.001) to the PCS score (Table 2).
Our findings suggest that how individuals perceive the cancer experience may influence or be influenced by their health status and functioning and QOL. As our colleagues  suggest, further research in the area of cognitive and behavioral interventions (e.g., cognitive re-framing) is needed to determine if positive reinterpretation of a negative traumatic event could subsequently lead to improved QOL and decreased distress. Based upon findings reported here and elsewhere , the IOCv2, with strong reliability and validity psychometrics, shows evidence of potential utility for monitoring intervention effects longitudinally.
As is typical for any cross-sectional study, we are unable to establish a cause-effect relationship between the IOCv2 and health status and functioning and QOL. For example, we cannot determine if perceptions of the cancer experience preceded the outcome variables of interest (health status and functioning, QOL), whether subjective QOL-related symptoms adjusted perceptions of how cancer impacted the survivors’ lives, or if both are true. The ability to assess a cohort over time, as would be possible in a longitudinal design, would help determine the direction(s) of causality. As a second limitation, the inclusion of NHL survivors from only two NC cancer centers may limit the generalizability of our results to survivors living in other regions and treated at smaller hospitals. However, our demographic profile closely mirrors that of the national population of NHL survivors , thereby strengthening the generalizability of our analyses. Further, some of the findings may reflect the fact that the IOCv2 and outcome measures assess similar constructs, although psychometric evaluation indicates that while overlapping, the measures have distinct aspects . Finally, study strengths include a high response rate (74%) and balanced gender ratio.
In closing, our findings suggest that positive and negative perceptions of cancer’s impact on survivors’ lives may influence or be influenced by self-reported physical and mental health and QOL. Longitudinal intervention research is needed to establish evidence of causality related to the IOCv2, which could lead to the development of treatments targeting perceptions of the cancer experience, ultimately to enhance health and QOL in survivors. In addition, results suggest that targeted screening, aimed at survivors who are younger, have comorbidities, and less social support would be beneficial to identify QOL-related issues that might be amenable to intervention.
Supported by: NCI (CA016042 and CA101492), the American Cancer Society (DSW-0321301-SW), AHRQ NRSA Training Grant (T32-HS000032), Cancer Care Quality Training Grant (CA116339), and the University of North Carolina Research Council