Positive and negative psychosocial impacts on cancer survivors

The purpose of this study is to understand psychosocial impacts on cancer survivors using the patient-reported outcomes measurement information system (PROMIS) Psychosocial Illness Impact banks. Cancer survivors (n = 509; age: 59.5 ± 1.4; 51.5% men) completed the PROMIS positive and negative illness impact items consisting of four sub-domains: self-concept (SC), social impact (SI), stress response (SR), and spirituality (Sp). Illness impact was defined as changed scores from items measuring “current” experiences to recalled experiences prior to cancer diagnosis. Descriptive statistics, effect sizes (ES), and coefficient of variation (CV) were calculated at item and sub-domain levels. Analysis of variance was used to identify potentially influential factors on the impacts. Our study found survivors reported stronger positive than negative impacts (overall ES mean: 0.30 vs. 0.23) in general; and more moderate (ES ≧ 0.30) positive than negative impacts at the item level, 54.3% (25 of 46) and 40% (16 of 40) for positive and negative items, respectively. Participants reported more positive impacts on SI and Sp but more negative impacts on SR. The CV results showed more individual differences appeared on positive SC items. Younger survivors reported stronger positive and negative impacts. Women reported higher positive impacts. Survivors with higher education levels tended to have higher positive SI impacts, while those with a lower family income reported higher negative SI and negative SR impacts. We conclude positive and negative psychosocial impacts coexisted—the strength of impacts varied across sub-domains. Age, gender, education, and family income influenced the psychosocial impacts reported by survivors. These findings provide a foundation to develop interventions to strengthen positive and minimize negative impacts and improve cancer survivors’ overall well-being.


Methods
We presented our study design and results following the reporting guideline from the EQUATOR Network [34][35][36] .

Participants. Participants included 509 cancer survivors recruited from the Duke Cancer Care Research
Program in Durham, NC (n = 72), the Duke Tumor Registry (n = 283), and NexCura, a nationwide online registry of more than 500,000 cancer survivors (n = 154).Survivors were eligible if they were 18 years or older, had a cancer diagnosis, and were fluent in English.This study was approved by the Institutional Review Board of Northwestern University, and all participants provided informed consent.All methods performed in this study follow the relevant guidelines and regulations.
Measures.This study was conducted as a cross-sectional study, and participants completed the following measures only once.The PROMIS Psychosocial Illness Impact item banks 27 consist of four conceptual subdomains (see Fig. 1 for the structure of the item banks): Self-Concept (SC), Social-Impact (SI), Stress-Response (SR), and Spirituality (Sp).Across these four sub-domains are 46 items measuring positive psychosocial illness impacts and 40 items measuring negative psychosocial illness impacts, including 11 positive and 9 negative SC items; 12 positive and 11 negative SI items; 11 positive and 10 negative SR items; and 12 positive and 10 negative Sp items (see Tables 1, 2 for item contents).Participants describe the extent to which the concept, as reflected in each item, affected their lives before their cancer diagnosis and/or treatment (How true was this before your illness?) and currently (How true is this now, since your illness?).In the following analyses, "before" referred to "How true was this before your illness?" and "current" referred to responses to "How true is this now, since Figure 1.The structure of the psychosocial illness impact item banks.your illness?"A 5-point rating scale is used: 0 = not at all, 1 = a little bit, 2 = somewhat, 3 = quite a bit, and 4 = very much."Illness impact" was defined as the changed scores from "before" to "current" ("now").Larger changed scores on positive and negative items indicated positive and negative impacts, respectively.Data analysis.Descriptive statistics were conducted at both item-and sub-domain levels on "illness impact" ("current"-"before") (range: -4 to 4); that is, response to "How true is this now, since your illness?"-response to "How true was this before your illness?"Effect size (ES; mean of impact divided by standard deviation) was used to estimate the averaged magnitudes and direction of impacts.An ES is considered trivial when the absolute value < 0.1, small when ES is between 0.1 and 0.3, moderate when ES is between 0.3 and 0.5, and large when ES ≧ 0.5 37 .The coefficient of variation (CV; standard deviation divided by the impact mean) was used to estimate the variability of responses on individuals.An item with high variability implies its potential to capture a wide range of differences in individual responses to this item that are often overlooked during the averaging process 38 .Items with absolute CVs > 10 are considered to have significant variability 39 .

Consent to participate.
Informed consent was obtained from all individual participants included in the study.
Descriptive statistics.Descriptive statistics and CV and ES of impact scores are shown in Tables 1 and 2 for positive and negative measures, respectively.At the item level, means of impact scores were 0.26 (range − 0.10 to 0.66) and 0.23 (range − 0.54 to 0.77) for positive and negative items, respectively.For items capturing positive aspects, patients reported the largest change since cancer diagnosis on "I see what is really important in my life" (Sp12; impact score = 0.66), followed by "I am aware of the love and support available from other people" (SI05; impact score = 0.62), and "I am comfortable receiving help from others" (SI07; impact score = 0.59).For those capturing negative aspects, patients reported the largest change since cancer diagnosis on "I fear what will happen in the future" (SR06; impact score = 0.77), followed by "I worry about the future" (SR01; impact score = 0.73) and "Worry about my health interferes with my life" (SR05; impact score = 0.73).At the sub-domain level, survivors reported almost no impact on positive SC (mean = 0.06) and negative Sp (mean = 0.07), yet large impacts on positive SI (mean = 0.39) and negative SR (means = 0.33).These results corresponded to the results at the item level.It was noted that patients reported minimal impacts on positive SC and negative Sp, where 10 (of 11) and 8 (of 10) items had an absolute impact mean≦ 0.2 for positive SC and negative Sp items, respectively.
Standardized impacts-effect size.As shown in Tables 1 and 2, the overall ES means (SDs) were 0.30 (± 0.23) and 0.23 (± 0.25 for positive and negative impacts, respectively.Among all 86 positive and negative items, 26 (30.2%) items and 15 (17.4%) items had moderate or strong impacts (absolute ES value ≥ 0.3) towards a positive and negative direction, respectively.This suggests survivors generally reported more positive than negative impacts from their cancer experiences.At the sub-domain level, survivors reported small or moderate impacts towards a positive direction on positive SI (mean ES = 0.47), positive SR (ES = 0.27), and positive Sp (mean ES = 0.37).A negligible impact was reported on positive SC (mean ES = 0.07).Survivors reported small impacts towards a negative direction on all negative sub-domains: SC (mean ES = 0.27), SI (mean ES = 0.27), SR (mean ES = 0.29), and Sp (mean ES = 0.11).These results supported our previous findings that positive and negative impacts coexisted with different strengths across sub-domains 26 .
At the item level, for positive items, thirteen larger positive ES (see Fig. 2) suggested that survivors had more appreciation of their physical health (SC06, ES = 0.45), life (Sp03, ES = 0.50), important things in life (Sp12, ES = 0.70) and people in their life (SI08, ES = 0.66); found more love and support from others (SI05, ES = 0.64), more compassion (SI02, ES = 0.52), more strength in prayer (Sp11, ES = 0.46) and close to God (Sp10, ES = 0.45); realized who their friends are (SI06, ES = 0.59); they were more comfortable asking others for help (SI03, ES = 0.53), receiving help (SI07, ES = 0.64), and expressing emotion (SR01, ES = 0.50); and took better care of themselves (SR08, ES = 0.60) after the cancer diagnosis.For negative items, ten larger positive ES (see Fig. 3) indicated that survivors experienced greater worry (SR01, SR05, SR09, ES > 0.58), fear (SR06, ES = 0.64), Table 1.Descriptive statistics for impact scores on positive sub-domains (n = 509).SC, self-concept; SI, social impact; SR, stress response; Sp, spirituality; NA, not available.† The bold value of Mean and SD for each sub-domain was calculated from survivors' scores on the items within the sub-domain.‡ The bold values of coefficient of variation (CV) and effect size (ES) for each sub-domain were calculated from the mean of item scores within the sub-domain.**Impact Scores: changed scores from before cancer to current.

Variability of items.
The variability of impact scores was small across all sub-domains with CVs < 10.It was noticed that 15 items (9 positive and 6 negative) had an absolute mean ≦ 0.1 but had an absolute CV greater than 10.Of these 9 positive items, 6 were positive Self-Concept items.These findings suggest significant individual variations on these items.However, changes in individual participants were inconsistent in direction resulting in the mean offsets impact changes in positive and negative directions.

Influential factors of illness impact.
Analyses of variance (ANOVA) showed significant results on age, gender, education, and income factors (p < 0.05).The following post-hoc comparison results are presented in Table 3. Age was a significant factor in 7 of 8 sub-domains across negative and positive impacts; younger survivors reported higher positive and negative impacts.Gender was a significant factor only in four positive subdomains but not in negative ones; women had higher positive impact scores than men.Survivors with higher education levels tended to have higher positive SI impact scores.Survivors with a lower family income had higher negative SI and negative SR impact scores.No significant effects were observed on ethnicity and relationship status.

Discussion
The experience of cancer can be devastating but can also bring an opportunity for personal growth and new perspectives on life [40][41][42][43][44] .In this study, we examined the psychosocial impacts of cancer and its treatment using concepts raised by cancer survivors, which were then formatted to capture the psychosocial impacts due to cancer diagnosis in depth.Our results indicated positive and negative psychosocial impacts coexisted, coinciding with the literature 26,[40][41][42][43][44][45] .We suggest that interventions should consider both positive and negative impacts.
Our findings can serve as a foundation to develop targeted, individualized interventions for whom increasing facilitators are needed by emphasizing positive impacts and minimizing barriers are needed by focusing on decreasing negative impacts.Survivors reported different levels of impact upon contents addressed by individual items.A brief form that consists of items with large ES can be created for measuring survivor psychosocial impacts upon cancer if administering full-length PROMIS Psychosocial Illness Impact item banks or computerized adaptive tests is not feasible.Past research proposed three important facets of impact from severely stressful life events: self-perceptions, social relationships, and personal growth and life priorities [42][43][44]46 . Tedechi and Calhoun 47 studied individual growth after encountering trauma such as cancer using Posttraumatic Growth Inventory.They found five factors: a greater appreciation of life, a changed sense of priorities, warmer and more intimate relationships, a greater sense of personal strength, and recognition of new possibilities or paths for one's life and spiritual development.Our results confirm these findings. Paticularly, we found that survivors reported more social support (mean ES = 0.47 for positive items) and better spiritual well-being (mean ES = 0.37 for positive items) but had relatively more challenges with stress (mean ES = 0.29 for negative items) and worse self-concept (mean ES = 0.27 for negative items).These findings were also supported by Park and Blank's study, in which cancer survivors reported larger positive impacts than negative ones 43 .However, their measure was not limited to psychosocial impact.
Individual differences were noted on 15 items with absolute CVs > 10, most related to positive SC.This result indicates that compared with other items, these SC items with larger CV reflected that the participants had more relative variation, either decrease or increase in their perception of "self " before/after the disease diagnosis.Yet the different directions of these item impact scores were canceled at the sub-domain level resulting in negligible impact scores.As items with larger CVs have the potential to discriminate against survivors with different levels of impact, these items could be considered candidates to detect individual differences over time.However, we should use CVs with caution.When the denominator (i.e., impact mean) is close to zero, the CV becomes very sensitive to small changes in the mean 48 .This condition occurs when participants' impact scores show variation, but positive and negative scores offset the mean of impact scores.
Our results on age, gender, and income level are consistent with the literature 43,45,47,[49][50][51][52][53][54][55] .We found younger survivors reported greater life disruptions yet simultaneously reported a more positive attitude towards the disease.Female survivors reported a more positive attitude, but there is no gender difference in negative attitudes toward the disease.Survivors with lower income report more negative psychosocial impacts from cancer.However, the literature does not provide a consistent pattern about education, ethnicity, and relationship status in the psychosocial impacts of cancer 45,50,52,54,55 .Our study found survivors with higher education have a more positive attitude toward the disease.No significant differences were found in ethnicity and relationship status.Future studies on different sample groups should be conducted to evaluate the replicability of our findings.
This study had some limitations.Our sample was not nationally representative; there was an over-representation of well-educated and White survivors.Replication of our results with a more diverse set of survivors is needed.Another limitation is that we relied on survivors to recall their experiences before diagnosis, as it is not practical to conduct a prospective study, enrolling people before a cancer diagnosis.We attempted to minimize recall bias by implementing an appropriate question format by asking survivors to answer each item content with a before/after format; subsequently, the "before" question could be the reference for the "current" question.Our sample had an average time since cancer diagnosis of 4.7 years; thus, the recall accuracy was questionable.Future studies that evaluate psychosocial impacts at different stages of the disease continuum longitudinally, e.g., every year or every six months post-diagnosis, should be conducted to establish trajectory patterns of impact over time.In addition, we collected in this study from participants' self-reported perceived change, which might not be their veridical change.Boals and colleagues distinguished perceived and veridical stress-related growth and presented four possible constructs to shape perceived growth: (a) adherence to a cultural script, (b) reappraisal coping through secondary control or self-enhancement; (c) changes in narrative identity; and (d) violation of post-recovery expectations 56 .In other words, participants' perceived responses might be influenced by their cultures; they regulate their beliefs and reaction to fit the world's expectations by secondary control; they maintain personal continuity over time to face adverse experiences by autobiographical reasoning; and they attribute their experience by a convincing explanation.Perceived change may be a mix of these possible constructs.Future studies are needed to evaluate the replicability of our findings.This study result may have implications for strategies in managing cancer survivors to promote their healthy adjustment to cancer throughout the disease continuum.The clinical focus of this research includes the impacts of cancer on self-concept, stress responses, social relationships, and spirituality; for example, to promote survivors' quality of life after a cancer diagnosis, clinical practitioners can facilitate survivors' positive consequences and alleviate negative ones.
In conclusion, this study expanded our prior work on the development of positive and negative psychosocial impact measures by requesting cancer survivors to report psychosocial impacts comparing before and after their

Figure 2 .
Figure 2. The effect size (ES) for positive items.

Figure 3 .
Figure 3.The effect size (ES) for negative items.

Table 2 .
Descriptive statistics for impact scores on negative sub-domains (n = 509).SC, self-concept; SI, social impact; SR, stress response; Sp, spirituality; NA = not available.†Thebold value of Mean and SD for each sub-domain was calculated from survivors' scores on the items within the sub-domain.‡Thebold values of coefficient of variation (CV) and effect size (ES) for each sub-domain were calculated from the mean of item scores within the sub-domain.**Impact scores: changed scores from before cancer to current.