HEALING STORIES: NARRATIVE CHARACTERISTICS IN CANCER SURVIVORSHIP NARRATIVES AND PSYCHOLOGICAL HEALTH AMONG HEMATOPOIETIC STEM CELL TRANSPLANT SURVIVORS
Survivors of hematopoietic stem cell transplant (HSCT) have experienced a life threatening disease and a physically and psychologically challenging treatment that can affect many areas of their lives. In addition to having to cope with their diagnosis (usually a hematologic malignancy), HSCT survivors undergo toxic treatment regimens that put their lives at risk (Copelan, Reference Copelan2006), including high dose chemotherapy. Studies of cancer survivors have emphasized the traumatic characteristics of cancer, including its sudden and unexpected onset and its uncontrollable nature (Tedeschi & Calhoun, Reference Tedeschi and Calhoun1995). Furthermore, research shows that HSCT precipitates responses consistent with psychological trauma (Mosher et al., Reference Mosher, Redd and Rini2009).
As a result, many studies of HSCT have focused on its negative emotional effects on survivors, including anxiety and depressive symptoms, diminished quality of life, and elevated generalized distress (Baker et al., Reference Baker, Denniston and Zabora2003; Broers et al., Reference Broers, Kaptein and Le Cessie2000; Edman et al., Reference Edman, Larsen and Hagglund2001; Kettmann & Altmaier, Reference Kettmann and Altmaier2008; Mosher et al., Reference Mosher, Redd and Rini2009). However, other studies suggest that HSCT can provide an opportunity for personal growth (Andrykowski, Brady & Hunt, Reference Andrykowski, Brady and Hunt1993; Hefferon, Grealy & Mutrie, Reference Hefferon, Grealy and Mutrie2009; Tallman, Altmaier & Garcia, Reference Tallman, Altmaier and Garcia2007). Research on personal growth following a traumatic experience has identified benefits such as development of the self, enhanced self-disclosure, emotional expressiveness in relationships, and a changed philosophy of life (Hefferon, Grealy & Mutrie, Reference Hefferon, Grealy and Mutrie2009; Tedeschi & Calhoun, Reference Tedeschi and Calhoun1995). However, research has not yet clarified what differentiates HSCT survivors who experience psychological growth from those who do not. The current study explores this question through the lens of narrative psychology theory, using the narratives of 23 HSCT survivors writing about their experience of the treatment. By analyzing these recovery narratives, we can examine whether HSCT survivors' interpretation of their experiences helps explain differences in their post-treatment psychological health.
Narrative psychology argues that the stories people tell about their lives are not mere recordings of life events but manifestations of the subjective ways they construct reality. Thus, narrative analysis offers an opportunity to understand emotions and psychological dynamics (McAdams, Reference McAdams1993; Raggatt, Reference Raggatt2007; Singer, Reference Singer, McAdams, Josselson and Lieblich2001). Examining the narrative characteristics of psychological growth and resilience after cancer treatment can help guide interventions to promote psychological health among survivors.
Narratives consist of two dimensions: content (i.e., what is being said) and form (i.e., how it is being said). Content analysis focuses on the ideas, memories, events, and themes in the story; form analysis deals with the structure of the story, including the organization of its plot, the order of the events, and the progress of the narrative in time (Benish-Weisman, Reference Benish-Weisman2009; Lieblich, Tuval Mashiah & Zilber, Reference Lieblich, Tuval-Mashiach and Zilber1998.) Form analysis differs from content analysis in two important ways. First, while narrators are usually highly aware of the content of their story, they may be less aware, and therefore less in control, of the story's form. Therefore, form analysis provides an opportunity to look at the deep and sometimes unconscious layers of the narrator's story. Second, content analysis tends not to focus on the timeline of a story, whereas form analysis does look at how an experience is discussed with respect to time elements, thereby adding a developmental point of view (Tuval-Mashiah, Reference Tuval-Mashiach, McAdams, Josselson and Lieblich2006).
For the most part, studies suggest that the content of a narrative is associated with the writer's psychological health, although evidence for a relationship between lower levels of psychological health and expression of emotional negativity remains unclear. That is, some studies suggest that narratives containing a large proportion of negative emotional words indicating sadness, fear, and unhappiness are negatively correlated with psychological health (Liehr et al., Reference Liehr, Marcus and Carroll2010; Pennebaker & Seagal, Reference Pennebaker and Seagal1999). Others claim that the repression of negative emotions can lead to increased levels of anxiety and depression (Iwamitsu et al., Reference Iwamitsu, Shimoda and Hajime2003), therefore emphasizing the therapeutic effect of expression of negative emotional effect (Graf, Gaudiano & Geller, Reference Graf, Gaudiano and Geller2008).
In addition, many studies suggest that the form of a narrative is associated with the writer's psychological health. The nature of the connection between the episodes of the story reveals the subjective way in which the cancer survivor interprets his or her illness and recovery. In a redemption sequence, for example, the storyteller describes episodes of transformation, going from an emotionally negative life event to an emotionally positive one (McAdams et al., Reference McAdams, Diamond and de St. Aubin1997). It has been shown that redemption episodes are positively related to high levels of psychological health (McAdams et al., Reference McAdams, Reynolds and Lewis2001). To our knowledge, this is the first study to examine form characteristics of narratives in HSCT survivors.
The current study hypothesizes that negative emotional expressions and redemption episodes in recovery narratives will differentiate survivors with poor psychological health from those with better psychological health. Specifically, we hypothesize that survivors with poor psychological health will tell recovery narratives with a significantly different amount of negative emotional expressions (due to inconsistencies in the literature we do not specify direction) and fewer redemptive episodes than survivors with better psychological health.
METHODS
Participants and Procedure
Twenty three participants in the present study were recruited through screening for a parent study — a multi-site trial investigating a cognitive behavioral intervention for severely distressed HSCT survivors (described in DuHamel et al., Reference DuHamel, Mosher and Winkel2010). Men and women were recruited for the trial's telephone-administered screening protocol through clinical databases at three medical centers in the northeastern United States. To participate in the trial's screening protocol, they had to be 1–3 years post-HSCT, English-speaking, and at least 18 years old. They were excluded if they were awaiting another transplant or receiving treatment for disease relapse, or if they had severe cognitive impairment, active psychosis, suicidal ideation, or substance dependence.
Participants who completed the trial's screening were approached for recruitment into this study if they were assessed as having at least elevated general distress and/or subclinical PTSD symptoms.Footnote 1 Those who agreed completed a pilot study of a psychosocial intervention that included three writing sessions, spaced one week apart, and a follow-up assessment 3 months after the last writing session (Smyth & Pennebaker, 2010). Because many HSCT survivors live far from their treatment site and have difficulty attending in-person study meetings, all pilot study procedures were completed by telephone using mailed materials. The writing sessions provided the narratives analyzed in the present study and the follow-up assessment provided the measures of psychological health. Using a telephone-based version of Pennebaker's emotionally expressive writing instructions (Pennebaker, Reference Pennebaker1997; Zakowski et al., Reference Zakowski, Ramati and Morton2004), pilot study participants were called for writing appointments and instructed to write continuously for 20 minutes about their deepest thoughts and feelings regarding their transplant experience. (A separate group of pilot study participants wrote in a non-emotionally expressive way about their daily activities; however, their narratives were not appropriate for this study and therefore are not discussed.) Each writing session focused on a specified stage of transplant (i.e., before, during, and after hospitalization). After completing each writing session, participants returned their writing in a postage-paid envelope. All procedures were approved by the Institutional Review Boards of the two participating study sites. Participants were compensated for their time.
Measures
Psychological Health
Three measures of psychological health were employed, assessing psychological aspects of health-related quality of life, depression, and anxiety. We assessed both cancer-specific and general psychological health outcomes as recommended by Mosher et al. (Reference Mosher, Redd and Rini2009).
Psychological aspects of health-related quality of life were assessed with three items from the emotional well-being subscale part of the Functional Assessment of Cancer Therapy-Bone Marrow Transplant (FACT-BMT) Version 4, a well-validated self-report instrument. Responses to questions about events in the prior week (e.g., “I worry about dying,” or “I worry that my condition will get worse”) were provided on a scale from 0 (not at all) to 4 (very much). Responses were reversed and summed so that higher scores reflected a higher level of psychological health (α = 0.73).
Depression and anxiety were measured using the depression and anxiety subscales of the well-validated Brief Symptom Inventory (BSI, Derogatis, Reference Derogatis1975). Each subscale included five items. Depression symptoms included “feeling blue” and “feeling no interest in things.” Anxiety symptoms included “suddenly scared for no reason” and “feeling fearful.” Respondents indicated how much discomfort each of these symptoms caused them in the past month on a scale from 1 (not at all) to 5 (extremely). A mean score for each subscale was calculated. Higher scores on these scales reflected higher levels of emotional distress. The internal reliability of each scale was α = 0.73 and α = 0.66, respectively.
Narrative Characteristics
All the narratives of the 23 participants were coded by two trained independent coders. Initial training included studying the content of each coding category and discussing the optional inclusion/exclusion criteria. The unit of the coding could be a few words (as illustrated by the examples of negative emotional expressions, below) or an entire paragraph (as illustrated by the example of a redemption episode, below), dependent on the context. Nvivo, a qualitative analysis program was used for coding. The coders agreed 77% of the time. Cases of disagreement were decided by a third independent judge. The categories for coding were (1) negative emotional expression and (2) redemption episodes. Emotional expressions were coded as such if they contained explicit affect. Examples of negative emotional expressions that were found in our sample narratives included: “(I felt) sadness when thinking of my family if this had not worked out” and “I was scared, angry, overwhelmed.” The coding of the redemption episodes was based on the instructions in the Coding System for Redemption Sequences (McAdams, Reference McAdams1999). The redemption episodes were defined as explicit transformation in the story from a decidedly negative affect state to a decidedly positive affect state (McAdams et al., Reference McAdams, Reynolds and Lewis2001). The transformation could take two forms: (1) a change from a negative affect state to a positive affect state, (2) the formation of a positive outcome out of a bad situation. For example, a 32 year old woman reflected on her healing experience as follows: “Finally in conclusion, I can say that this experience has definitely been an opportunity for me to reflect on my life and truly value the things that I normally would overlook, under normal circumstances. The challenges have made me think harder; appreciate the small things in nature and within me.” It is clear from this quotation that this survivor acknowledges the benefits of the hard experience she had. Moreover, from her perspective, the disease has improved her life.
RESULTS
Descriptive Statistics
Participants ranged in age from 26 to76 (M = 56.6, SD = 11.5). Nearly half (45.8%) were women, 87.6% were white, and 70% were married. With respect to psychological health, their emotional well-being subscale scores from the FACT-BMT ranged from 14 to 24 (M = 20.87, SD = 2.38), their BSI depression subscale scores ranged from 0.01 to 0.15 (M = 0.28, SD = 3.7), and their BSI anxiety subscale scores ranged from 0.01 to 0.84 (M = 0.19, SD = 0.27).
Hypothesis Testing
Our objective was to test the hypothesis that survivors who suffered from poor psychological health would tell recovery narratives with a significantly different number of negative emotional expressions and fewer redemptive episodes than survivors with better psychological health. First, we divided the sample into different levels of psychological health by performing a cluster analysis using FACT emotional well-being scores, the anxiety subscale of the BSI, and the depression subscale of the BSI. Using a two-step cluster analysis with Schwarz Beyesian Criterion and Distance Measure Log-Likehood, two clusters emerged. The first cluster (n = 19) was characterized by a high score on the FACT emotional well-being scale (M = 21.64, SD = 1.7) and low scores on the depression (M = 0.15, SD = 0.17), and anxiety subscales of the BSI (M = 0.09, SD = 0.14). We named it the high psychological health cluster. The second cluster (n = 4) was characterized by a low score on the FACT emotional well-being scale (M = 17.25, SD = 2.3) and high scores on the depression (M = 0.8, SD = 0.58), and anxiety subscales of the BSI (M = 0.67, SD = 0.24). We named it the low psychological health cluster.
In order to examine whether narrative characteristics differentiated between the two clusters, we conducted discriminant analysis, with redemption and negative emotionality episodes predicting membership in the two psychological health clusters. Discriminant Function Analysis is well-suited to identifying and mapping the narrative patterns that differentially characterize high and low levels of psychological health. The overall Wilk's lambda was significant (Λ = 0.67, χ2[2, N = 23] = 8.06, p < 0.05), indicating that the predictors significantly differentiated the two psychological health clusters. To interpret this finding, standard discriminant loadings (or discriminant function coefficients) were examined. Discriminant loadings operate as factor loadings would in factor analysis (Burns & Burns, Reference Burns and Burns2008). The positive coefficient for redemption episodes (0.7) and negative coefficient for negative emotionality episodes (−0.67) were relatively large (i.e., over 0.5 (Burns & Burns, Reference Burns and Burns2008)). In addition, the psychological health cluster group means on the discriminant loadings indicate that high numbers of redemption episodes and low numbers of negative emotionality episodes were related to high psychological health. The cross validation enabled 78.3% of the individuals in this sample to be classified correctly, which is regarded as an acceptable percentage (Burns & Burns, Reference Burns and Burns2008). I would add a summary sentence here or the first in the discussion reminding people what we found– overall, consistent with our hypothesis, we found that the greater incidence of redemptive episodes and a the lower use of negative emotionality episodes predicted better psychological health.
DISCUSSION
Living through HSCT can result in long-lasting and negative psychological outcomes for survivors. The current study examined whether HSCT survivors' post-treatment psychological health was associated with the form of their recovery narratives. Findings indicated that, compared to survivors with poor psychological health, those with better psychological health were more likely to describe redemption episodes and less likely to use negative emotional expressions.
The fact that people with better psychological health used fewer negative emotional expressions suggests that HSCT survivors' expression of their stress and fear may not serve as a cathartic mechanism (Pennebaker, Reference Pennebaker and Wilce2003). This finding contradicts traditional clinical psychology theories that encourage the expression of repressed negative emotionality in the belief that they can release the patient from his/her dark feelings (Murray, Lamnin & Carver, Reference Murray, Lamnin and Carver1989), though it is consistent with more recent models of trauma and coping, which claim that encouraging expression of negative emotions and grief as an intervention strategy might not be helpful (Bonanno, Reference Bonanno2004; Bonnano & Mancini, Reference Bonanno and Mancini2008). Although the correlational design of the study prevent us from making conclusions about causality, our findings are consistent with previous studies suggesting that ruminating on negative emotions might preserve or even aggravate negative mood (Bushman, Reference Bushman2002), thereby preventing an opportunity for change. For example, the expression of intense anxiety by breast cancer patients may contribute to and maintain feelings of helplessness (Lieberman & Goldstein, Reference Lieberman and Goldstein2006). In other words, emotional discharge may not be enough for change in psychological adjustment (Nenova et al., Reference Nenova, Duhamel and Zemon2011).
Our findings indicate that survivors who experience their recovery as built on negative points that turn into positive ones (redemption episodes) enjoy better psychological health. This is consistent with other studies that show that cognitive change through a reappraisal of the situation may be necessary in order to achieve better psychological health (Murray, Lamnin & Carver, Reference Murray, Lamnin and Carver1989; Pennebaker, Reference Pennebaker1997). The ability to construct life events positively is related to well-being (King & Miner, Reference King and Miner2000; McAdams et al., Reference McAdams, Reynolds and Lewis2001; Taylor, Reference Taylor1989), and many studies show that optimistic or hopeful people have better mental health. We argue that HSCT survivors derive their present mental health in part from their former experience of illness. In other words, in their narratives they portray past difficulties as the reason for present psychological health, and their experience of the healthy present is intensified by their difficulties in the past.
Narratives open a window on the subjective experience and thus present an ideal opportunity to study people's experiences. However, the converse is also true: narratives can affect experiences (Widdershoven, Reference Widdershoven, Josselson and Lieblich1993). Through retelling, the feeling or evaluation of an experience may change. By putting experiences into words and by creating the possibility of looking at an experience from a distance, storytellers are able to process memories, feelings, and thoughts; they can reevaluate what happened or even have new insights. For HSCT survivors, naming amorphous and sometimes overwhelming experiences allows them to connect their stories to others (Frank, Reference Frank2001), developing feelings of companionship and solidarity.
Findings from the current study should be evaluated in the context of the study's methodological limitations. First, the study has a small number of participants. Although intriguing and potentially clinically valuable, findings should be verified in a larger study examining the relation between narrative characteristics and psychological health in cancer survivors. Second, it includes HSCT survivors who are one to three years into their recovery. As longitudinal studies suggest (e.g., Hjermstad et al., Reference Hjermstad, Knobel and Brinch2004), psychological health changes over time. Therefore, future studies should investigate HSCT survivors at various points in their recovery. A longitudinal design would facilitate investigation of the plausibility of causal links between narrative structural elements and later psychological health.
Despite some limitations, the study makes a significant applied contribution. Our findings suggest detecting structural elements in HSCT survivors' stories may enable health workers and therapists to comprehend unspoken aspects of their recovery stories (Frank, Reference Frank2001). Through the insight they derive, they may be able to help their HSCT survivor clients to do more than simply express their negative emotions, as these emotional expressions might have negative effects (Murray, Lamnin & Carver, Reference Murray, Lamnin and Carver1989). Rather, they may be able to help their survivor clients interpret their stories more adaptively while retaining the facts. Moreover, a client's writing could provide a therapist with insight. For example, when a survivor's writing indicates a lack of redemption or highly negative affect, talk therapy could be used to move the individual to a more redemptive, positive understanding of the transplant experience. In addition, providing new interpretations of past events might allow the client and therapist to compose new endings to stories of suffering (Omer & Alon, Reference Omer and Alon1997).
ACKNOWLEDGMENTS
We would like to thank Jennifer Chee-Chait for her valuable research assistance and to Professor Suzanne Ouellette for enriching our knowledge on narrative analysis.
