Checking Behavior, Fear of Recurrence, and Daily Triggers in Breast Cancer Survivors

Checking Behavior, Fear of Recurrence, and Daily Triggers in Breast Cancer Survivors Abstract Background Fear of cancer recurrence (FCR) is a top ongoing concern of breast cancer (BC) survivors and thus the focus of recent intervention development. The Self-Regulation Model of FCR (Lee-Jones C, Humphris G, Dixon R, Hatcher MB. Fear of cancer recurrence–a literature review and proposed cognitive formulation to explain exacerbation of recurrence fears. Psychooncology. 1997;6:95–105.) states that everyday cancer-related events trigger FCR, which, in turn, leads to specific behavioral responses, including checking the body for signs or symptoms of cancer. Links between triggering events, FCR, and checking behavior have not yet been studied in the context of daily life or at the within-person level. Purpose The goal of this study was to examine whether FCR has a within-person link with daily checking behavior and whether FCR mediates the link between triggering events and checking behavior. Methods Seventy-two early-stage BC survivors completed daily diaries over a 21-day period approximately 5 months after BC surgery. FCR, checking behavior, and triggering events were assessed each evening. Results Multilevel modeling results indicated that FCR predicted greater odds of same-day, but not next-day, checking behavior. We found that daily FCR significantly mediated the same-day effect of triggering events on checking behavior. These average within-person effects varied substantially between patients and were not explained by momentary negative affect. Conclusions Findings support the within-person relationship between triggering events, FCR, and checking behavior posited by guiding theory, and can inform FCR intervention development. Breast cancer, Fear of cancer recurrence, Survivorship, Reassurance, Checking behavior, Breast self-examination Fear of cancer recurrence (FCR) is defined as “fear, worry, or concern relating to the possibility that cancer will come back or progress” [1]. Particularly for cancers with high survival rates, such as breast cancer (BC), FCR can become a permanent fixture in the lives of patients whose cancer has been treated successfully. For the 3.6 million BC survivors in the USA [2], the possibility of recurrence is often the top ongoing concern [3–5]. FCR can be best understood as a chronic rather than acute experience, as it has been shown to persist years following BC treatment [6, 7]. A growing literature points to numerous negative outcomes of FCR, including psychological distress [6, 8], worse quality of life [4, 9], and anxiety [6]. The Self-regulation Model of FCR The most prominent theoretical framework for conceptualizing FCR is an adaptation of Leventhal’s Self-Regulation Model of Illness (SRM [10]) by Lee-Jones and colleagues [11]. The SRM posits a network of processes, beginning with antecedents or triggers of FCR and ending with its consequences. Posited triggers of FCR include internal cues, such as experiencing pain and other physical symptoms, and external cues, such as annual cancer screenings and commercials for cancer treatment. The SRM points to both behavioral and psychological consequences of FCR, including checking for signs or symptoms of cancer and anxiety. It is important to note that some view these triggers and consequences as focal aspects of the clinical manifestation of FCR, such that clinical FCR is defined and measured, in part, by their presence [1, 12–14]. Although there is currently a lack of consensus as to the best way to define and measure both clinical and non-clinical FCR, we conceptualize triggers and consequences as important and related constructs, but distinct from FCR [12, 15, 16]. Rather, the central feature of FCR is viewed as thoughts and fears related to the possibility of recurrence. Psychological consequences (or correlates) of FCR are one well-studied aspect of the SRM framework [4]. However, there have been few empirical tests of the proximal triggers and behavioral consequences of FCR. In terms of the theorized triggers, studies have found cross-sectional associations between FCR and internal triggers including pain, physical symptoms, fatigue, and changes in appearance (for a review, see [4, 17]). These studies indicate that patients who experience more physical symptoms, treatment side effects, or changes to their appearance tend to be the same patients who report greater FCR. Research on external triggers of FCR is less developed. Qualitative work suggests that common external triggers include doctor appointments, hearing about someone else having cancer, and annual mammograms [8, 18]. Behavioral consequences of FCR have also received little empirical attention. One theorized behavioral consequence is checking the body for signs or symptoms of cancer [11]. It is thought that when a patient experiences FCR, one way she may attempt to regulate her distress is by seeking confirmation that no new signs or symptoms have surfaced. When she gains this confirmation by checking her body, she may be reassured in the moment that her cancer has not progressed or recurred. Thus, theoretically, checking behavior temporarily attenuates the distress associated with FCR. For patients who experience FCR frequently, checking behavior may become excessive, problematic, and even compulsive. Over the past several decades, scholars have frequently described and drawn upon this theorized process [19–23]. In fact, excessive checking behavior is an explicit target of an intervention designed to reduce FCR [24]. Although checking behavior may attenuate FCR, excessive use of this strategy may lead to other negative outcomes. Guidelines from the American Society of Clinical Oncology recommend monthly breast self-examination for BC survivors [25] based on evidence that a greater frequency increases rates of invasive procedures resulting in benign findings, but does not improve rates of mortality [26]. Self-examination is not necessarily a certain or straightforward method for assessing signs or symptoms of cancer progression or recurrence, especially when performed in an emotionally aroused state [19, 27, 28]. Thus, checking in response to FCR may actually evoke more uncertainty about disease status or lead to unnecessary doctor appointments and procedures. Fear of Recurrence and Checking Behavior as Within-Person Phenomena The strong theoretical and intuitive interest in checking behavior does not currently correspond to its empirical support as a consequence of FCR. Only two quantitative studies, to our knowledge, have assessed FCR and checking behavior in BC patients. In a cross-sectional study, Thewes and colleagues [29] found that greater FCR was associated with more frequent self-reported breast self-examination. In another cross-sectional study, survivors who had more severe FCR reported more frequent use of checking as a strategy to cope with FCR [16]. In related research on healthy women with elevated BC risk, several studies have found positive links between BC worry and self-exam frequency [30, 31], although others have not [32]. Although preliminary support for the SRM is encouraging, it is exclusively based on cross-sectional data. Cross-sectional designs lend themselves well to the study of between-person associations, which describe the comparison of individuals to one another based on their global or average levels of a particular variable. For example, the cross-sectional study reviewed above suggests that patients with greater global FCR tend to report a higher frequency of breast self-examination than patients with lower global FCR [29]. However, typically also of interest are within-person associations, which describe the comparison of individuals to themselves (rather than to other individuals) based on their within-person variability captured in repeated measures. For example, is a patient more likely to report checking behavior on a day in which she has higher FCR compared with a day in which she has lower FCR? Longitudinal studies are required to assess both between- and within-person effects. Echoing Ryu, West, and Sousa [33], we argue that the within-person association between FCR and checking behavior is of substantive theoretical and clinical interest, and it is important to examine whether the between-person effects found previously extend to the within-person level. The presence of a between-person effect does not imply the presence of a within-person effect and vice versa; in fact, these two effects can differ in magnitude and even direction [34]. Although prior cross-sectional findings suggest that individuals with higher FCR tend to report checking more frequently, global and retrospective assessment may only distally capture this link. Global cross-sectional methods are vulnerable to retrospective biases [35] and do not rule out unmeasured between-person confounds or address the directionality of an association [36–38]. Behavioral consequences of FCR, including checking behavior, have not yet been examined prospectively and at the within-person level, despite the fact that checking behavior is targeted in FCR interventions. Moreover, FCR triggers hypothesized to initiate the SRM sequence, while well-studied cross-sectionally, have also not yet been examined at this level. Intensive longitudinal methods, which involve repeated measurements of individuals over time (e.g., daily diaries [37]), can provide valuable information that maps onto our theoretical and intuitive understanding of FCR. This approach can limit the retrospective biases associated with global self-report of behaviors or feelings over a longer span of time and provide a clearer picture of the processes flowing into and from FCR [35–38]. Furthermore, intensive longitudinal data offer the potential of identifying the directionality and sequence of effects [36, 37]. For example, one can examine whether a particular effect carries over from one assessment to the next, which could provide insight into the potency and duration of the effect. This level of insight can be easily translated and applied to the development and refinement of FCR interventions. The Present Study We examined the associations between triggering events, FCR, and checking behavior on a daily, within-person level, which has not been previously examined. Daily diary data from early-stage BC patients were used to accomplish several aims. Our first aim was to determine whether FCR was associated with checking behavior on the same day (concurrently). Based on the SRM, we hypothesized that patients who reported greater FCR on one day would be more likely to check for signs or symptoms of cancer on that same day. Note that a concurrent within-person association is not synonymous with a cross-sectional between-person association—the former models within-person day-to-day variability while the latter does not. We also sought to explore the directionality of the association between FCR and checking behavior by examining carryover of the effect—in both directions—from one day to the next. No prior research has attempted to delineate the timeline for these effects. Furthermore, it has been suggested that checking behavior may reinforce and strengthen FCR, but this has not been examined empirically. Thus, an exploratory aim was to investigate whether FCR was associated with a greater likelihood of checking the next day, and vice versa. Finally, we examined the posited sequence of triggers predicting FCR, in turn, leading to checking behavior. Our second aim was to determine the extent to which the effect of daily triggers on checking behavior is mediated by FCR. Based on the SRM, we hypothesized that FCR would mediate the within-person association between triggers and checking behavior. To rule out the possibility that feelings of negativity could explain the hypothesized links among these variables, we included momentary negative affect as a within-person covariate in all tests. Method Participants Participants were female early-stage BC patients recruited from a community cancer center in the mid-Atlantic region of the USA for a larger longitudinal study with Institutional Review Board approval. Patients were asked to participate if they met the following criteria: (a) diagnosed with Stage 0 (lobular/ductal carcinoma in situ), I, II, or IIIA BC; (b) had recent BC surgery; (c) in a committed romantic relationship with a partner who also agreed to participate; (d) English-speaking; (e) lived within an hour of the recruitment site; and (f) did not have prior cancer diagnoses. Patients who had a recent positive breast biopsy and appeared eligible per electronic medical records were sent a letter about the study and contacted by phone. Note that although the larger study obtained data from patients and their partners, only patient data are examined here. Of the 1,161 patients who had a recent positive breast biopsy, 698 were identified via medical records as ineligible. The 463 potentially eligible patients were mailed a letter and contacted by phone. Of those who were contacted, 117 verbally agreed to participate and were mailed a consent form, 110 were ineligible based on patient-provided information (78% ineligible due to being unpartnered), 82 could not be reached, and 154 declined to participate (most frequent reasons were “not enough time” and “spouse does not wish to participate”). Seventy-nine patients provided informed consent and participated, 4 provided consent but did not participate, and 34 did not return the consent form. This article examined one assessment period of the larger longitudinal study (discussed in more detail below), to which 72 patients contributed data, resulting in a final sample of 72 patients described here. Approximately 89% of patients identified as Caucasian, 10% as African American, 1% as Asian, and all as non-Hispanic. The average age was 54.54 years (SD = 9.48) and 94% were married. In terms of occupational status, about 43% reported not working, 39% reported working full time, and 18% reported working part time. Most (73%) reported an annual household income over $60,000. Fifteen percent of patients were diagnosed with Stage 0 BC, 49% Stage IA, 24% Stage IIA, 11% Stage IIB, and 1% Stage IIIA. The majority of patients did not receive adjuvant chemotherapy (67%) but did receive adjuvant radiation (71%) and hormonal therapy (81%). Procedure Patients were recruited around the start of adjuvant treatment. After providing informed consent, patients completed a series of baseline questionnaires. The baseline assessment took place an average of 3 months after surgery (SD = 1.56, range = 1–8). Data for the current study came from a daily diary period that took place after the baseline assessment, as soon as possible after adjuvant treatment. This time period was targeted for study based on evidence suggesting that FCR emerges soon after the end of adjuvant treatment, when patients see their health care providers less frequently and often feel as though they are no longer actively fighting the cancer [39, 40]. Patients who did not receive adjuvant chemotherapy or radiotherapy completed the diary period as soon as possible after the baseline assessment. Patients started the diary period an average of 5 months after surgery (SD = 2.09, range = 2–12). The daily diary period took place over 21 consecutive days. During this period, patients were asked to complete a short survey every evening within about an hour of going to sleep. Before the diary period began, patients were provided with a web link to access this survey from their home each evening. The evening survey was designed to be brief and took patients 13 min on average to complete. Because several items asked patients about the events of their day, compliance with the requested completion time was monitored via the survey software. Data were only included in analyses if the survey was completed between 6:00 p.m. and 3:00 a.m. On average, patients completed the surveys on 17 out of the 21 days (81% compliance rate). Measures Daily FCR Six items from the Fear of Cancer Recurrence Inventory (FCRI [8]), which were among the highest-loading items in past analytic work [8], were adapted for daily use. One item (from the FCRI severity subscale) assessed how much time patients spent thinking about the possibility of cancer recurrence that day, with responses ranging from 0 (didn’t think about it at all) to 4 (several hours). Four items (from the distress subscale) assessed negative emotional reactions (worried, afraid, or anxious; sad, discouraged, or disappointed; frustrated, angry, or outraged; helpless or resigned) related to the possibility of recurrence. The fifth item, drawn from the insight subscale, assessed the extent to which one felt that he/she “worried excessively” about the possibility of recurrence. The latter five items were rated from 0 (not at all) to 4 (extremely). The six items were summed to create a daily FCR composite, which had strong within-person reliability (ω = .91). The average daily FCR score was 2.09 (within-person SD = 2.53, between-person SD = 2.20). FCR was positively skewed with a large proportion (57.8%) of zero scores. As a result, FCR was modeled as a multilevel count outcome with a Poisson distribution in all analyses. Daily triggers Eight daily triggers were assessed. As described earlier, although some include triggers in their measurement of overall FCR, we conceptualize these as distinct constructs. In the absence of a previously validated daily measure, these items were developed through consultation with a clinical psycho-oncologist and a nurse navigator specializing in cancer survivorship needs. Several items overlapped with items from the FCRI triggers subscale [8]. Patients indicated whether each trigger occurred that day. The following five internal triggers were assessed: noticed skin irritation, tingling or loss of sensation in hands or feet, had swelling or trouble moving the arm, cognitive difficulties (e.g., issues with memory, everyday decision making, focusing, finding words), and noticed changes in physical appearance (e.g., scars, hair loss, change in complexion). Three external triggers were assessed: had a long wait in doctor’s office, received negative news from physician, and frustrating interaction with medical professional. The eight items were summed to represent the number of triggers experienced daily. The average number of daily triggers was 0.86 (within-person SD = 0.76, between-person SD = 0.84). Although it could be argued that internal and external triggers should be examined separately, the current sample size, limited number of items, and relatively low frequency of triggers led to the decision to utilize a single trigger composite. Daily checking behavior One item assessed whether patients engaged in daily checking behavior: “Today, did you examine yourself physically for signs or symptoms of cancer?” Patients responded yes or no. On average, patients reported checking on 19% of the diary days (within-person SD = 0.32, between-person SD = 0.23). Momentary negative affect Momentary negative affect was measured using seven standard items from the negative affect subscale of the Positive and Negative Affect Schedule Expanded Form [41], which were averaged to create a composite. Only seven negative affect items (sad, angry, afraid, lonely, blue, scared, and frightened) were included to keep the daily diaries as brief as possible. Patients indicated the extent they experienced each item “at this moment” from 0 (not at all) to 4 (extremely). Coefficient omega estimated acceptable within-person reliability of the momentary negative affect scale (ω = .79). The average momentary negative affect score was 0.17 (within-person SD = 0.26, between-person SD = 0.24). Results Data Analytic Approach See Supplementary Material 1 for additional descriptive statistics and correlations. Multilevel modeling was used to accommodate the nested structure of the daily diary data (i.e., days nested within patients). Multilevel analyses were conducted in R (version 3.3.2; R Foundation for Statistical Computing, Vienna, Austria) using the lme4 package (version 1.1–12 [42]). The linear effect of time and momentary negative affect were both included as within-person covariates in all models (note that the exclusion of negative affect as a covariate resulted in the same pattern of findings for all analyses). Time is included as a covariate to control for autocorrelation in Level 1 residuals as well as any systematic increase or decrease in the outcome over the diary period (e.g., as a result of reactivity to study procedures [37]). As mentioned earlier, since, by definition, person-level (i.e., between-person) factors, such as global or baseline levels of FCR severity, cannot confound within-person relationships, such factors need not be included as covariates in these models. A random intercept was estimated in each model, allowing for person-to-person variability in levels of the outcome. Random slope effects for focal predictors were estimated when possible. Time-varying predictors were person-mean centered and person-level predictors were grand-mean centered [37]. The multilevel model provides valid inferences assuming data are missing at random [43]. To address our first aim, we examined the concurrent daily association between FCR and checking behavior (a binary outcome variable) by estimating a multilevel logistic regression model. Daily FCR was specified as a predictor of daily checking behavior (the outcome variable), for which a random slope effect was estimated. In addition to the within-person effect of interest, we estimated the between-person effect by including average FCR over the diary period as a person-level predictor. To explore the directionality of the association between FCR and checking behavior, we examined whether FCR predicted not only concurrent checking behavior, but also future checking behavior, and vice versa. An additional set of models were estimated in which the outcome was set one day after the predictor. First, same-day FCR was specified as a predictor of next-day checking behavior (the outcome variable), while controlling for same-day checking behavior. We then specified same-day checking behavior as a predictor of next-day FCR (the outcome variable), while controlling for same-day FCR. A random slope effect for the focal predictor of each model (i.e., FCR or checking behavior) was also estimated. For our second aim, we conducted a 1→1→1 multilevel mediation analysis [44–46], in which the predictor (triggering events), mediator (FCR), and outcome (checking behavior) were all time-varying variables (i.e., measured daily). The hypothesized mediation model is depicted in Fig. 1. Because FCR was modeled as a count outcome and checking behavior as a binary outcome, the traditional approach to mediation analysis could not be used—the standard test of the mediated effect (i.e., the product of the indirect paths [47]) would produce a biased estimate [48]. To obtain a valid estimate of the indirect effect for our model, we used a more general approach to mediation analysis [48] based on the counterfactual, or potential outcomes, framework [49, 50]. This more general approach to mediation analysis accommodates conditions that traditional approaches do not, including noncontinuous outcome and mediator distributions (e.g., here, checking behavior is a binary outcome modeled using multilevel logistic regression [51]). The mediation package in R was used to test the hypothesized mediated effect and obtain valid estimates of the average direct, mediated, and total effects in the multilevel context (v4.4.5 [52]). For readers interested in a detailed description of this approach and how it differs from traditional mediational analysis, see Supplementary Material 2 and [48, 49, 51, 52]. Fig. 1. View largeDownload slide Hypothesized within-person multilevel mediation model. The a path was estimated by regressing fear of cancer recurrence (FCR) on daily triggers, controlling for momentary negative affect, average triggers, and time, with a random slope estimated for daily triggers. The b and c′ paths were estimated by regressing checking behavior on daily FCR and daily triggers. Although not depicted, we controlled for momentary negative affect, average FCR, average triggers, and time. Fig. 1. View largeDownload slide Hypothesized within-person multilevel mediation model. The a path was estimated by regressing fear of cancer recurrence (FCR) on daily triggers, controlling for momentary negative affect, average triggers, and time, with a random slope estimated for daily triggers. The b and c′ paths were estimated by regressing checking behavior on daily FCR and daily triggers. Although not depicted, we controlled for momentary negative affect, average FCR, average triggers, and time. Daily Link Between FCR and Checking Behavior The results of our first aim (determine whether FCR was associated with same-day checking behavior) are displayed in Table 1. FCR emerged as a significant within-person predictor of checking behavior, indicating that when a patient experiences a one-unit increase in FCR (compared with what is typical for her), the odds are 32% greater that she will check herself for signs or symptoms of cancer the same day. The random effect variance of this slope was 0.03 (SD = 0.18), indicating that the slopes of about 95% of patients fell between −0.07 (odds ratio [OR] = 0.93) and 0.64 (OR = 1.89). Thus, for some patients, the effect of FCR on checking behavior approached zero, while for others, the effect was greater than the average effect. In addition, average FCR over the diary period was a significant predictor of average checking behavior (γ = 0.37, OR = 1.45, z = 3.18, p = .002). This between-person effect indicates that patients who tend toward higher average daily FCR also tend toward more frequent checking. Table 1 Multilevel regression results of concurrent and prospective link between FCR and checking behavior Effect  Estimate  SE  Odds/rate ratio  p  95% CI  Lower  Upper  Concurrent link: FCR → Same-day checking behaviora  FCR  0.28  0.06  1.32  <.001  0.162  0.398  Negative affect  −0.57  0.42  0.57  .171  −1.393  0.253  Time  0.01  0.02  1.01  .546  −0.029  0.049  Prospective link: FCR → Next-day checking behaviora  FCR  0.04  0.06  1.04  .473  −0.078  0.158  Negative affect  −0.24  0.40  0.79  .574  −1.024  0.544  Time  −0.01  0.02  0.99  .710  −0.049  0.029  Prospective link: Checking behavior → Next-day FCRb  Checking behavior  0.31  0.20  1.37  .112  −0.082  0.702  Negative affect  0.28  0.06  1.32  <.001  0.162  0.398  Time  −0.01  0.00  0.99  .005  −0.018  −0.002  Effect  Estimate  SE  Odds/rate ratio  p  95% CI  Lower  Upper  Concurrent link: FCR → Same-day checking behaviora  FCR  0.28  0.06  1.32  <.001  0.162  0.398  Negative affect  −0.57  0.42  0.57  .171  −1.393  0.253  Time  0.01  0.02  1.01  .546  −0.029  0.049  Prospective link: FCR → Next-day checking behaviora  FCR  0.04  0.06  1.04  .473  −0.078  0.158  Negative affect  −0.24  0.40  0.79  .574  −1.024  0.544  Time  −0.01  0.02  0.99  .710  −0.049  0.029  Prospective link: Checking behavior → Next-day FCRb  Checking behavior  0.31  0.20  1.37  .112  −0.082  0.702  Negative affect  0.28  0.06  1.32  <.001  0.162  0.398  Time  −0.01  0.00  0.99  .005  −0.018  −0.002  While not shown, same-day level of the outcome was also included as a covariate in prospective models. FCR = daily fear of cancer recurrence. aPoisson regression estimates are exponentiated to obtain rate ratios. bLogistic regression estimates are exponentiated to obtain odds ratios. View Large Table 1 Multilevel regression results of concurrent and prospective link between FCR and checking behavior Effect  Estimate  SE  Odds/rate ratio  p  95% CI  Lower  Upper  Concurrent link: FCR → Same-day checking behaviora  FCR  0.28  0.06  1.32  <.001  0.162  0.398  Negative affect  −0.57  0.42  0.57  .171  −1.393  0.253  Time  0.01  0.02  1.01  .546  −0.029  0.049  Prospective link: FCR → Next-day checking behaviora  FCR  0.04  0.06  1.04  .473  −0.078  0.158  Negative affect  −0.24  0.40  0.79  .574  −1.024  0.544  Time  −0.01  0.02  0.99  .710  −0.049  0.029  Prospective link: Checking behavior → Next-day FCRb  Checking behavior  0.31  0.20  1.37  .112  −0.082  0.702  Negative affect  0.28  0.06  1.32  <.001  0.162  0.398  Time  −0.01  0.00  0.99  .005  −0.018  −0.002  Effect  Estimate  SE  Odds/rate ratio  p  95% CI  Lower  Upper  Concurrent link: FCR → Same-day checking behaviora  FCR  0.28  0.06  1.32  <.001  0.162  0.398  Negative affect  −0.57  0.42  0.57  .171  −1.393  0.253  Time  0.01  0.02  1.01  .546  −0.029  0.049  Prospective link: FCR → Next-day checking behaviora  FCR  0.04  0.06  1.04  .473  −0.078  0.158  Negative affect  −0.24  0.40  0.79  .574  −1.024  0.544  Time  −0.01  0.02  0.99  .710  −0.049  0.029  Prospective link: Checking behavior → Next-day FCRb  Checking behavior  0.31  0.20  1.37  .112  −0.082  0.702  Negative affect  0.28  0.06  1.32  <.001  0.162  0.398  Time  −0.01  0.00  0.99  .005  −0.018  −0.002  While not shown, same-day level of the outcome was also included as a covariate in prospective models. FCR = daily fear of cancer recurrence. aPoisson regression estimates are exponentiated to obtain rate ratios. bLogistic regression estimates are exponentiated to obtain odds ratios. View Large Prospective Link Between FCR and Checking Behavior Results of our exploratory aim (determine whether FCR was associated with greater odds of checking the next day, and vice versa) are shown in Table 1. First, we did not find evidence that FCR predicted next-day checking behavior on average (OR = 1.04, p = .473). Thus, our results suggest that FCR is a strong predictor of same-day checking behavior, but does not have a detectable effect on next-day checking behavior (above and beyond its same-day effect). The FCR random slope effect estimated a variance of 0.03 (SD = 0.17), indicating that 95% of individual patient slopes in the population fall between −0.29 (OR = 0.75) and 0.38 (OR = 1.46). Second, results suggested a larger but nonsignificant average effect of checking behavior on next-day FCR (rate ratio [RR] = 1.37, p = .112). When a patient checked, regardless of her FCR score that day, she was predicted to have a 1.37 times greater FCR score the next day. The random slope of checking behavior also pointed to substantial variability (variance = 0.73, SD = 0.85), indicating that 95% of patient slopes fall between −1.40 (RR = 0.25) and 2.02 (RR = 7.53). Within-Person Mediation Model The results of the tests of the hypothesized mediation model (depicted in Fig. 1) are displayed in Table 2. The a path was estimated by regressing FCR on daily triggers, momentary negative affect, average triggers, and time, with a random slope effect of daily triggers also estimated. As expected, triggering events were a significant and positive predictor of same-day FCR. Specifically, on a day a patient experienced one additional triggering event (compared with her average number of daily triggers), she was predicted to have a 1.51 times greater FCR score that same-day. The random effect variance of this fixed effect was 0.14 (SD = 0.37), indicating that about 95% of patient slopes in the population fall between −0.32 (RR = 0.72) and 1.15 (RR = 3.16). Average number of triggers over the diary period also significantly predicted higher average FCR, γ = 0.71, RR = 2.04, z = 3.76, p < .001. Table 2 Results of within-person multilevel mediation: FCR triggering events → FCR → checking behavior Effect  Estimate  SE  Odds/rate ratio  p  95% CI  Lower  Upper  Mediator model: FCR triggers → FCRa  FCR triggers (a path)  0.41  0.08  1.51  <.001  0.259  0.568  Negative affect  0.42  0.05  1.52  <.001  0.313  0.518  Time  −0.02  0.00  0.98  <.001  −0.023  −0.009  Outcome model: Triggers and FCR → Checking behaviorb  FCR (b path)  0.28  0.04  1.33  <.001  0.203  0.365  FCR triggers (c′ path)  0.10  0.13  1.11  .414  −0.146  0.354  Negative affect  −0.60  0.39  0.55  .119  −1.363  0.155  Time  0.01  0.02  1.01  .448  −0.019  0.042  Mediation effectsc  Average mediated effect  0.02  –  –  <.001  0.01  0.04  Average direct effect  0.01  –  –  .42  −0.02  0.04  Total effect  0.04  –  –  .03  0.00  0.07  Proportion mediated effect  0.66  –  –  .03  0.22  2.65  Effect  Estimate  SE  Odds/rate ratio  p  95% CI  Lower  Upper  Mediator model: FCR triggers → FCRa  FCR triggers (a path)  0.41  0.08  1.51  <.001  0.259  0.568  Negative affect  0.42  0.05  1.52  <.001  0.313  0.518  Time  −0.02  0.00  0.98  <.001  −0.023  −0.009  Outcome model: Triggers and FCR → Checking behaviorb  FCR (b path)  0.28  0.04  1.33  <.001  0.203  0.365  FCR triggers (c′ path)  0.10  0.13  1.11  .414  −0.146  0.354  Negative affect  −0.60  0.39  0.55  .119  −1.363  0.155  Time  0.01  0.02  1.01  .448  −0.019  0.042  Mediation effectsc  Average mediated effect  0.02  –  –  <.001  0.01  0.04  Average direct effect  0.01  –  –  .42  −0.02  0.04  Total effect  0.04  –  –  .03  0.00  0.07  Proportion mediated effect  0.66  –  –  .03  0.22  2.65  FCR = daily fear of cancer recurrence. aEstimates from Poisson regression are exponentiated to obtain rate ratios. bEstimates from logistic regression are exponentiated to obtain odds ratios. cMediation effect estimates are displayed in probability units. View Large Table 2 Results of within-person multilevel mediation: FCR triggering events → FCR → checking behavior Effect  Estimate  SE  Odds/rate ratio  p  95% CI  Lower  Upper  Mediator model: FCR triggers → FCRa  FCR triggers (a path)  0.41  0.08  1.51  <.001  0.259  0.568  Negative affect  0.42  0.05  1.52  <.001  0.313  0.518  Time  −0.02  0.00  0.98  <.001  −0.023  −0.009  Outcome model: Triggers and FCR → Checking behaviorb  FCR (b path)  0.28  0.04  1.33  <.001  0.203  0.365  FCR triggers (c′ path)  0.10  0.13  1.11  .414  −0.146  0.354  Negative affect  −0.60  0.39  0.55  .119  −1.363  0.155  Time  0.01  0.02  1.01  .448  −0.019  0.042  Mediation effectsc  Average mediated effect  0.02  –  –  <.001  0.01  0.04  Average direct effect  0.01  –  –  .42  −0.02  0.04  Total effect  0.04  –  –  .03  0.00  0.07  Proportion mediated effect  0.66  –  –  .03  0.22  2.65  Effect  Estimate  SE  Odds/rate ratio  p  95% CI  Lower  Upper  Mediator model: FCR triggers → FCRa  FCR triggers (a path)  0.41  0.08  1.51  <.001  0.259  0.568  Negative affect  0.42  0.05  1.52  <.001  0.313  0.518  Time  −0.02  0.00  0.98  <.001  −0.023  −0.009  Outcome model: Triggers and FCR → Checking behaviorb  FCR (b path)  0.28  0.04  1.33  <.001  0.203  0.365  FCR triggers (c′ path)  0.10  0.13  1.11  .414  −0.146  0.354  Negative affect  −0.60  0.39  0.55  .119  −1.363  0.155  Time  0.01  0.02  1.01  .448  −0.019  0.042  Mediation effectsc  Average mediated effect  0.02  –  –  <.001  0.01  0.04  Average direct effect  0.01  –  –  .42  −0.02  0.04  Total effect  0.04  –  –  .03  0.00  0.07  Proportion mediated effect  0.66  –  –  .03  0.22  2.65  FCR = daily fear of cancer recurrence. aEstimates from Poisson regression are exponentiated to obtain rate ratios. bEstimates from logistic regression are exponentiated to obtain odds ratios. cMediation effect estimates are displayed in probability units. View Large Next, we tested the b and c′ paths by regressing checking behavior on daily FCR, daily triggers, momentary negative affect, average FCR, average triggers, and time. Results are shown in Table 2. No random slope effects were able to be estimated in this model due to nonconvergence. Because we were unable to estimate a random effect for this slope using lme4, we also used Bayesian and generalized estimating equation (GEE) methods to confirm that the focal fixed effect results remained after estimating random slope effects (Bayesian) or correcting standard errors (GEE). These fixed effect results remained statistically significant. Consistent with the results of the first aim, results revealed a significant positive effect of FCR on checking behavior (b path). However, the effect of triggers on checking behavior was not statistically significant (c′ path). The R mediation package uses the results of these models to estimate the mediated effects (provided in probability units; shown in Table 2). The average mediated causal effect was 0.02, indicating that when a patient experiences one more trigger relative to what is typical for her, there is a 2% increase in the probability of her checking on the same day that is due solely to increases in FCR. The average direct effect was 0.01, suggesting that a one-unit increase in daily triggers results in a 1% increase in the probability that the patient will exhibit checking behavior that is not explained by increases in FCR. The average proportion mediated effect is 0.66, indicating that 66% of the total effect of triggers on checking behavior is explained by FCR. Discussion Although a growing literature points to FCR as a critical component of BC adjustment, we are aware of no published work examining antecedents and consequences of FCR within the context of daily life. The present study examined daily checking behavior, a consequence of FCR according to the prominent SRM [11] and a target of FCR interventions [24]. Checking behavior has long been theorized to function as a distress-reducing strategy and resemble a compulsion in patients with very high FCR [11, 19–23]. Surprisingly, this link has only been directly examined in one cross-sectional study [29]. Despite the paucity of empirical support, emerging FCR interventions draw heavily on the SRM and explicitly target excessive checking behavior [24]. Using daily diary data from BC survivors, the current study addresses this gap by examining the proximal (FCR) and distal (FCR-triggering events) predictors of daily checking behavior proposed by the SRM. The current study provides the first test of the within-person processes outlined in the SRM and targeted in FCR interventions but not captured in the existing literature, which is based almost exclusively on global and cross-sectional assessments [16]. Results provided support for our hypothesis that patients who experience greater FCR on one day would be more likely to report checking their bodies for signs or symptoms of cancer. We found that on a day that a patient reports a one-unit increase in FCR, the odds are 32% greater that she will check for signs or symptoms of cancer the same-day. Moreover, this effect was not explained by momentary negative affect during the evening survey. Although these results are based on concurrent data and do not permit inferences about the causal nature of this relationship, they complement prior cross-sectional findings by indicating the presence of a significant within-person association between FCR and checking behavior, which has not been demonstrated previously. Furthermore, our approach minimizes biases associated with typical cross-sectional studies that require patients to retrospectively report on far longer periods of time (e.g., past month or year). In addition, person-level (i.e., between-person) factors, such as global or baseline FCR severity, cannot confound within-person associations between FCR and checking behavior; therefore, any stable factors that vary only from person to person are not plausible alternative explanations for the current within-person findings. We also examined the extent to which the association between FCR and checking behavior holds across different patients, which has not been described previously. The significant fixed effect was found to vary between patients, such that for some, the estimated effect was near zero, while for others, it was over twice that of the average fixed effect. These results point to potential moderating variables that attenuate or intensify the effect of FCR on checking behavior. Although the current study was not designed or sufficiently powered to examine moderators of this effect, future studies should investigate variables that may explain why some patients engage in checking behavior when they experience FCR and others do not. Given the lack of empirical or theoretical work on the timescale of the SRM, we explored carryover effects of FCR on checking behavior the next day. We did not find evidence of an effect of FCR on next-day checking behavior (controlling for same-day checking behavior). This finding does not necessarily support or conflict with the hypothesis that FCR plays a causal role, but does suggest that future studies must implement more frequent assessments to determine the directionality and temporal sequence of these momentary experiences. Intuitively, it seems likely that the effect of FCR on checking behavior occurs within a shorter time frame than one day, making it impossible for a daily diary design to capture. Importantly, there was evidence of variability in this effect, such that greater FCR on one day predicted increased odds of checking the next day for some patients and decreased odds for others. It will be important to investigate moderators that explain these individual differences in future research. Scholars have suggested that checking behavior may maintain or exacerbate FCR [11, 19–23]. We explored this by examining whether checking behavior predicted next-day FCR, controlling for FCR and momentary negative affect the previous day. Although this effect was not statistically significant, there was substantial variability in this effect, such that for some it was strongly negative and others strongly positive. Although checking behavior has been conceptualized as an FCR-reinforcing compulsion, more research is required to determine for whom and under what conditions this rings true. Again, more frequent assessments are likely needed to clarify the sequence and cycle of FCR and checking behavior over time. Finally, we tested a multilevel within-person mediation model representing the full sequence delineated in the SRM—triggering events predict FCR, which, in turn, predicts checking behavior. The results supported this mediation model, indicating that the effect of daily triggers on checking behavior was significantly mediated by daily FCR. The direct effect of daily triggers on checking behavior, not explained by FCR, was not statistically significant. These results support the antecedents, FCR, and consequences posited by the SRM at the same momentary, within-person level that inherently defines the model. Furthermore, the methods employed minimize retrospective bias and eliminate the possibility of person-level confounding variables. Taken together, the results and methodological strengths of the current study provide strong support for the SRM that has not been previously documented. Limitations and Future Directions We attempted to address directionality and temporal precedence in the association between FCR and checking behavior by examining the effects from one day to the next. Because we did not find strong support for these next-day carryover effects, it is possible that checking behavior temporally precedes FCR. Similarly, we cannot rule out the possibility that checking behavior leads to increased sensitivity or awareness of daily triggering events. As described above, it seems plausible that the lack of strong evidence of temporal precedence resulted from an incongruence between the timeframe of assessments and unfolding of effects. It is possible to engage in several acts of checking behavior within a day, which, furthermore, may occur with or without conscious awareness. Future studies should be designed to capture these processes that may unfold over minutes or hours. Our use of intensive longitudinal methodology allowed us to quantify heterogeneity in the within-person effects among BC patients. Although this study was not designed to explain this variability, our results suggest that this is a promising avenue for future research. It will be important to understand the individual characteristics and contextual factors that attenuate or intensify the effect of FCR on checking behavior. For example, patients who perceive themselves at higher risk of recurrence may be more likely to actively check for signs of recurrence [53]. Trait anxiety may also differentiate the patient who has this behavioral response to FCR from the patient who does not [22, 54, 55]. Patients who use adaptive coping strategies, such as disclosing concerns about recurrence to a supportive and responsive loved one, may be less likely to engage in checking behavior. Furthermore, if checking behavior resembles a compulsion, then one may expect for those who check more frequently to exhibit a stronger daily link between FCR and checking behavior. Finally, variability in the association between triggering events and FCR suggests individual differences in reactivity to certain events. For example, patients with higher trait anxiety may be more reactive to events. Relatedly, although our daily triggers measure was based on the widely used and validated FCRI [8], given the paucity of prior studies of FCR in daily life, it is unlikely to capture all possible daily triggers. Future studies with large samples are required to investigate whether between-person factors, such as trait anxiety, moderate the within-person associations between these variables. It is possible that insufficient power contributed to the null finding for checking behavior on next-day FCR. A larger and more diverse sample may be necessary to determine whether checking behavior functions as a FCR-reinforcing compulsion. Patients in the current sample were largely high-functioning and typically reported low levels of daily FCR. The nature of the associations under study may differ across a range of FCR severity, and our findings may not generalize to those suffering from very high levels of FCR. At the same time, our findings highlight the importance of examining the normative experience and dimensional nature of FCR, as they revealed meaningful and theory-driven associations on a day-to-day basis between FCR, triggering events, and checking behavior in our sample. Nonetheless, future work should address these limitations because understanding the potential harm of checking behavior has important implications for mental and physical health care. Ideally, future research will define the bounds between psychologically adaptive and maladaptive checking behavior. In the absence of such work, an empirically-based definition of excessive checking behavior from a psychological standpoint is unavailable. Thus, our current conceptualization of excessive checking behavior relies on the medical literature, which shows that checking more than monthly is maladaptive for physical health [26]. However, FCR interventions that target checking behavior will be more effective if guided by empirical support for the distinction between psychologically adaptive and maladaptive checking behavior, and how and for whom this behavior perpetuates a cyclical process. Such work will also inform the guidance and education about self-examination provided to patients. Little is known about other potential outcomes of checking behavior. The excessive checker may be more vulnerable to false positives that result in unnecessary health care utilization [26]. Theory also points to health care utilization as another consequence of FCR itself, although extant research is limited. The few studies that have examined FCR and health care utilization suggest inconsistent associations across different forms of health care [29, 56, 57]. For example, Thewes and colleagues [29] found that FCR was associated with more frequent self-checking behavior but less frequent formal screenings (e.g., mammograms). Another study found that FCR predicted more outpatient and emergency room visits, but not specialist visits or other health care appointments [56]. More research is needed to gauge the extent and domains of the consequences of FCR, as well as the associations among the consequences themselves—for example, does FCR lead to health care utilization in the absence of excessive checking behavior? Finally, the current sample was fairly homogeneous in terms of racial/ethnic makeup (mostly Caucasian) and income (relatively high on average). These socio-cultural characteristics relate to the low levels of distress observed in this sample, which limit the generalizability of our findings to patients with greater distress and impairment, as noted above. In addition, culture plays a role in the nature of adaptation to cancer [58], and therefore, the reported associations between FCR, triggering events, and checking behavior may not generalize to other socio-cultural groups. More cross-cultural research on FCR and cross-culturally validated measures are sorely needed [59]. The associations studied here must be tested in samples characterized by greater distress and impairment, as well as greater socio-cultural diversity. In addition, patients with prior cancer diagnoses were excluded from the study. Therefore, future research is necessary to determine whether the antecedents and consequences of FCR posited by the SRM hold for patients with longer and more complex histories of cancer-related experiences. Conclusion The SRM is a widely used framework for the conceptualization of FCR and development of FCR interventions. This was the first study to move beyond global and cross-sectional tests of the model and examine the daily within-person processes that are inherent in the framework. Our findings support the posited sequence of triggering events, FCR, and checking behavior. These findings have potential implications for health care utilization, self-checking instruction for BC patients, and FCR intervention development. Supplementary Material Supplementary material is available at Annals of Behavioral Medicine online. Compliance with Ethical Standards Authors’ Statement of Conflict of Interest and Adherence to Ethical StandardsThe authors declare that they have no conflict of interest. Authors' Contributions E. C. Soriano wrote the manuscript with input from all authors. E. C. Soriano and J. P. Laurenceau conducted the analyses. E. C. Soriano, R. Valera, E. C. Pasipanodya, and J. P. Laurenceau conceptualized the paper. E. C. Soriano, E. C. Pasipanodya, and A. K. Otto collected data. S. D. Siegel and J. P. Laurenceau designed the parent study. Ethical Approval All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed Consent Informed consent was obtained from all individual participants included in the study. Acknowledgments Work presented in this manuscript was generously supported by the National Cancer Institute (grant number R21CA171921-02). References 1. Lebel S, Ozakinci G, Humphris Get al.  ; University of Ottawa Fear of Cancer Recurrence Colloquium attendees. From normal response to clinical problem: Definition and clinical features of fear of cancer recurrence. Support Care Cancer . 2016; 24( 8): 3265– 3268. Google Scholar CrossRef Search ADS PubMed  2. Miller KD, Siegel RL, Lin CCet al.   Cancer treatment and survivorship statistics, 2016. CA Cancer J Clin . 2016; 66( 4): 271– 289. Google Scholar CrossRef Search ADS PubMed  3. Baker F, Denniston M, Smith T, West MM. Adult cancer survivors: How are they faring? Cancer . 2005; 104( S11): 2565– 2576. Google Scholar CrossRef Search ADS PubMed  4. Simard S, Thewes B, Humphris Get al.   Fear of cancer recurrence in adult cancer survivors: A systematic review of quantitative studies. J Cancer Surviv . 2013; 7( 3): 300– 322. Google Scholar CrossRef Search ADS PubMed  5. Vickberg SM. The Concerns About Recurrence Scale (CARS): A systematic measure of women’s fears about the possibility of breast cancer recurrence. Ann Behav Med . 2003; 25( 1): 16– 24. Google Scholar CrossRef Search ADS PubMed  6. Deimling GT, Bowman KF, Sterns S, Wagner LJ, Kahana B. Cancer-related health worries and psychological distress among older adult, long-term cancer survivors. Psychooncology . 2006; 15( 4): 306– 320. Google Scholar CrossRef Search ADS PubMed  7. Mellon S, Kershaw TS, Northouse LL, Freeman-Gibb L. A family-based model to predict fear of recurrence for cancer survivors and their caregivers. Psychooncology . 2007; 16( 3): 214– 223. Google Scholar CrossRef Search ADS PubMed  8. Simard S, Savard J. Fear of cancer recurrence inventory: Development and initial validation of a multidimensional measure of fear of cancer recurrence. Support Care Cancer . 2009; 17( 3): 241– 251. Google Scholar CrossRef Search ADS PubMed  9. Mehnert A, Berg P, Henrich G, Herschbach P. Fear of cancer progression and cancer-related intrusive cognitions in breast cancer survivors. Psychooncology . 2009; 18( 12): 1273– 1280. Google Scholar CrossRef Search ADS PubMed  10. Leventhal H, Diefenbach M, Leventhal EA. Illness cognition: Using common sense to understand treatment adherence and affect cognition interactions. Cogn Ther Res . 1992; 16( 2): 143– 163. Google Scholar CrossRef Search ADS   11. Lee-Jones C, Humphris G, Dixon R, Hatcher MB. Fear of cancer recurrence–a literature review and proposed cognitive formulation to explain exacerbation of recurrence fears. Psychooncology . 1997; 6: 95– 105. Google Scholar CrossRef Search ADS PubMed  12. Mutsaers B, Jones G, Rutkowski Net al.   When fear of cancer recurrence becomes a clinical issue: A qualitative analysis of features associated with clinical fear of cancer recurrence. Support Care Cancer . 2016; 24( 10): 4207– 4218. Google Scholar CrossRef Search ADS PubMed  13. Sharpe L, Thewes B, Butow P. Current directions in research and treatment of fear of cancer recurrence. Curr Opin Support Palliat Care . 2017; 11( 3): 191– 196. Google Scholar CrossRef Search ADS PubMed  14. Thewes B, Lebel S, Seguin Leclair C, Butow P. A qualitative exploration of fear of cancer recurrence (FCR) amongst Australian and Canadian breast cancer survivors. Support Care Cancer . 2016; 24( 5): 2269– 2276. Google Scholar CrossRef Search ADS PubMed  15. Costa DS, Smith AB, Fardell JE. The sum of all fears: Conceptual challenges with measuring fear of cancer recurrence. Support Care Cancer . 2016; 24( 1): 1– 3. Google Scholar CrossRef Search ADS PubMed  16. Custers JAE, Gielissen MFM, de Wilt JHWet al.   Towards an evidence-based model of fear of cancer recurrence for breast cancer survivors. J Cancer Surviv . 2017; 11( 1): 41– 47. Google Scholar CrossRef Search ADS PubMed  17. Crist JV, Grunfeld EA. Factors reported to influence fear of recurrence in cancer patients: A systematic review. Psychooncology . 2013; 22( 5): 978– 986. Google Scholar CrossRef Search ADS PubMed  18. Gill KM, Mishel M, Belyea Met al.   Triggers of uncertainty about recurrence and long-term treatment side effects in older African American and Caucasian breast cancer survivors. Oncol Nurs Forum . 2004; 31( 3): 633– 639. Google Scholar CrossRef Search ADS PubMed  19. Fardell JE, Thewes B, Turner Jet al.   Fear of cancer recurrence: A theoretical review and novel cognitive processing formulation. J Cancer Surviv . 2016; 10( 4): 663– 673. Google Scholar CrossRef Search ADS PubMed  20. Ghazali N, Cadwallader E, Lowe D, Humphris G, Ozakinci G, Rogers SN. Fear of recurrence among head and neck cancer survivors: Longitudinal trends. Psychooncology . 2013; 22( 4): 807– 813. Google Scholar CrossRef Search ADS PubMed  21. Lasry JC, Margolese RG. Fear of recurrence, breast-conserving surgery, and the trade-off hypothesis. Cancer . 1992; 69( 8): 2111– 2115. Google Scholar CrossRef Search ADS PubMed  22. Stark DP, House A. Anxiety in cancer patients. Br J Cancer . 2000; 83( 10): 1261– 1267. Google Scholar CrossRef Search ADS PubMed  23. Ziner KW, Sledge GW, Bell CJ, Johns S, Miller KD, Champion VL. Predicting fear of breast cancer recurrence and self-efficacy in survivors by age at diagnosis. Oncol Nurs Forum . 2012; 39( 3): 287– 295. Google Scholar CrossRef Search ADS PubMed  24. Humphris G, Ozakinci G. The AFTER intervention: A structured psychological approach to reduce fears of recurrence in patients with head and neck cancer. Br J Health Psychol . 2008; 13( 2): 223– 230. Google Scholar CrossRef Search ADS PubMed  25. Khatcheressian JL, Hurley P, Bantug Eet al.  ; American Society of Clinical Oncology. Breast cancer follow-up and management after primary treatment: American Society of Clinical Oncology clinical practice guideline update. J Clin Oncol . 2013; 31( 7): 961– 965. Google Scholar CrossRef Search ADS PubMed  26. Thomas DB, Gao DL, Ray RMet al.   Randomized trial of breast self-examination in Shanghai: Final results. J Natl Cancer Inst . 2002; 94( 19): 1445– 1457. Google Scholar CrossRef Search ADS PubMed  27. Miller SM. Monitoring versus blunting styles of coping with cancer influence the information patients want and need about their disease. Implications for cancer screening and management. Cancer . 1995; 76( 2): 167– 177. Google Scholar CrossRef Search ADS PubMed  28. Taylor C, Richardson A, Cowley S. Surviving cancer treatment: An investigation of the experience of fear about, and monitoring for, recurrence in patients following treatment for colorectal cancer. Eur J Oncol Nurs . 2011; 15( 3): 243– 249. Google Scholar CrossRef Search ADS PubMed  29. Thewes B, Butow P, Bell MLet al.  ; FCR Study Advisory Committee. Fear of cancer recurrence in young women with a history of early-stage breast cancer: A cross-sectional study of prevalence and association with health behaviours. Support Care Cancer . 2012; 20( 11): 2651– 2659. Google Scholar CrossRef Search ADS PubMed  30. Cohen M. Breast cancer early detection, health beliefs, and cancer worries in randomly selected women with and without a family history of breast cancer. Psychooncology . 2006; 15( 10): 873– 883. Google Scholar CrossRef Search ADS PubMed  31. Erblich J, Bovbjerg DH, Valdimarsdottir HB. Psychological distress, health beliefs, and frequency of breast self-examination. J Behav Med . 2000; 23( 3): 277– 292. Google Scholar CrossRef Search ADS PubMed  32. Posluszny DM, McFeeley S, Hall L, Baum A. Stress, breast cancer risk, and breast self-examination: Chronic effects of risk and worry. J Appl Biobehav Res . 2004; 9( 2): 91– 105. Google Scholar CrossRef Search ADS   33. Ryu E, West SG, Sousa KH. Distinguishing between-person and within-person relationships in longitudinal health research: Arthritis and quality of life. Ann Behav Med . 2012; 43( 3): 330– 342. Google Scholar CrossRef Search ADS PubMed  34. Hamaker EL. Why researchers should think “within-person”: A paradigmatic rationale. In: Mehl MR, Conner TS, Csikszentmihalyi M, eds. Handbook of Research Methods for Studying Daily Life . Paperback ed. New York, NY: Guilford; 2012: 43– 61. 35. Shiffman S, Stone AA, Hufford MR. Ecological momentary assessment. Annu Rev Clin Psychol . 2008; 4( 1): 1– 32. Google Scholar CrossRef Search ADS PubMed  36. Bolger N, Davis A, Rafaeli E. Diary methods: Capturing life as it is lived. Annu Rev Psychol . 2003; 54( 1): 579– 616. Google Scholar CrossRef Search ADS PubMed  37. Bolger N, Laurenceau J-P. Intensive Longitudinal Methods: An Introduction to Diary and Experience Sampling Research . New York, NY: Guilford Press; 2013. 38. Laurenceau J-P, Bolger N. Using diary methods to study marital and family processes. J Fam Psychol . 2005; 19( 1): 86– 97. Google Scholar CrossRef Search ADS PubMed  39. King MT, Kenny P, Shiell A, Hall J, Boyages J. Quality of life three months and one year after first treatment for early stage breast cancer: Influence of treatment and patient characteristics. Qual Life Res . 2000; 9( 7): 789– 800. Google Scholar CrossRef Search ADS PubMed  40. McKinley ED. Under toad days: Surviving the uncertainty of cancer recurrence. Ann Intern Med . 2000; 133( 6): 479– 480. Google Scholar CrossRef Search ADS PubMed  41. Watson D, Clark LA, Harkness AR. Structures of personality and their relevance to psychopathology. J Abnorm Psychol . 1994; 103( 1): 18– 31. Google Scholar CrossRef Search ADS PubMed  42. Bates D, Mächler M, Bolker B, Walker S. Fitting linear mixed-effects models using lme4. J Stat Softw . 2015; 67( 1): 1–48. 43. Enders CK. Applied Missing Data Analysis . New York: Guilford Press; 2010. 44. Bauer DJ, Preacher KJ, Gil KM. Conceptualizing and testing random indirect effects and moderated mediation in multilevel models: New procedures and recommendations. Psychol Methods . 2006; 11( 2): 142– 163. Google Scholar CrossRef Search ADS PubMed  45. Kenny DA, Korchmaros JD, Bolger N. Lower level mediation in multilevel models. Psychol Methods . 2003; 8( 2): 115– 128. Google Scholar CrossRef Search ADS PubMed  46. Krull JL, MacKinnon DP. Multilevel modeling of individual and group level mediated effects. Multivariate Behav Res . 2001; 36( 2): 249– 277. Google Scholar CrossRef Search ADS PubMed  47. Baron RM, Kenny DA. The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. J Pers Soc Psychol . 1986; 51( 6): 1173– 1182. Google Scholar CrossRef Search ADS PubMed  48. Imai K, Keele L, Tingley D. A general approach to causal mediation analysis. Psychol Methods . 2010; 15( 4): 309– 334. Google Scholar CrossRef Search ADS PubMed  49. Imai K, Jo B, Stuart EA. Using potential outcomes to understand causal mediation analysis: Comment on. Multivariate Behav Res . 2011; 46( 5): 861– 873. Google Scholar CrossRef Search ADS PubMed  50. Rubin DB. Causal inference using potential outcomes: Design, modeling, decisions. J Am Stat Assoc . 2005; 100( 469): 322– 331. Google Scholar CrossRef Search ADS   51. Valeri L, Vanderweele TJ. Mediation analysis allowing for exposure-mediator interactions and causal interpretation: Theoretical assumptions and implementation with SAS and SPSS macros. Psychol Methods . 2013; 18( 2): 137– 150. Google Scholar CrossRef Search ADS PubMed  52. Tingley D, Yamamoto T, Hirose K, Keele L, Imai K. Mediation: R package for causal mediation analysis. J Stat Softw . 2014; 59( 5): 1– 38. Google Scholar CrossRef Search ADS PubMed  53. McCaul KD, Schroeder DM, Reid PA. Breast cancer worry and screening: Some prospective data. Health Psychol . 1996; 15( 6): 430– 433. Google Scholar CrossRef Search ADS PubMed  54. Cameron LD, Leventhal H, Love RR. Trait anxiety, symptom perceptions, and illness-related responses among women with breast cancer in remission during a tamoxifen clinical trial. Health Psychol . 1998; 17( 5): 459– 469. Google Scholar CrossRef Search ADS PubMed  55. Cohen M. First-degree relatives of breast-cancer patients: Cognitive perceptions, coping, and adherence to breast self-examination. Behav Med . 2002; 28: 15– 22. Google Scholar CrossRef Search ADS PubMed  56. Lebel S, Tomei C, Feldstain A, Beattie S, McCallum M. Does fear of cancer recurrence predict cancer survivors’ health care use? Support Care Cancer . 2013; 21( 3): 901– 906. Google Scholar CrossRef Search ADS PubMed  57. Sarkar S, Sautier L, Schilling G, Bokemeyer C, Koch U, Mehnert A. Anxiety and fear of cancer recurrence and its association with supportive care needs and health-care service utilization in cancer patients. J Cancer Surviv . 2015; 9( 4): 567– 575. Google Scholar CrossRef Search ADS PubMed  58. Curran L, Sharpe L, Butow P. Anxiety in the context of cancer: A systematic review and development of an integrated model. Clin Psychol Rev . 2017; 56: 40– 54. Google Scholar CrossRef Search ADS PubMed  59. Lebel S, Ozakinci G, Humphris Get al.   Current state and future prospects of research on fear of cancer recurrence. Psychooncology . 2017; 26( 4): 424– 427. Google Scholar CrossRef Search ADS PubMed  © Society of Behavioral Medicine 2018. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Annals of Behavioral Medicine Oxford University Press

Checking Behavior, Fear of Recurrence, and Daily Triggers in Breast Cancer Survivors

Loading next page...
 
/lp/ou_press/checking-behavior-fear-of-recurrence-and-daily-triggers-in-breast-Rfg9JzNVm5
Copyright
© Society of Behavioral Medicine 2018. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.
ISSN
0883-6612
eISSN
1532-4796
D.O.I.
10.1093/abm/kay033
Publisher site
See Article on Publisher Site

Abstract

Abstract Background Fear of cancer recurrence (FCR) is a top ongoing concern of breast cancer (BC) survivors and thus the focus of recent intervention development. The Self-Regulation Model of FCR (Lee-Jones C, Humphris G, Dixon R, Hatcher MB. Fear of cancer recurrence–a literature review and proposed cognitive formulation to explain exacerbation of recurrence fears. Psychooncology. 1997;6:95–105.) states that everyday cancer-related events trigger FCR, which, in turn, leads to specific behavioral responses, including checking the body for signs or symptoms of cancer. Links between triggering events, FCR, and checking behavior have not yet been studied in the context of daily life or at the within-person level. Purpose The goal of this study was to examine whether FCR has a within-person link with daily checking behavior and whether FCR mediates the link between triggering events and checking behavior. Methods Seventy-two early-stage BC survivors completed daily diaries over a 21-day period approximately 5 months after BC surgery. FCR, checking behavior, and triggering events were assessed each evening. Results Multilevel modeling results indicated that FCR predicted greater odds of same-day, but not next-day, checking behavior. We found that daily FCR significantly mediated the same-day effect of triggering events on checking behavior. These average within-person effects varied substantially between patients and were not explained by momentary negative affect. Conclusions Findings support the within-person relationship between triggering events, FCR, and checking behavior posited by guiding theory, and can inform FCR intervention development. Breast cancer, Fear of cancer recurrence, Survivorship, Reassurance, Checking behavior, Breast self-examination Fear of cancer recurrence (FCR) is defined as “fear, worry, or concern relating to the possibility that cancer will come back or progress” [1]. Particularly for cancers with high survival rates, such as breast cancer (BC), FCR can become a permanent fixture in the lives of patients whose cancer has been treated successfully. For the 3.6 million BC survivors in the USA [2], the possibility of recurrence is often the top ongoing concern [3–5]. FCR can be best understood as a chronic rather than acute experience, as it has been shown to persist years following BC treatment [6, 7]. A growing literature points to numerous negative outcomes of FCR, including psychological distress [6, 8], worse quality of life [4, 9], and anxiety [6]. The Self-regulation Model of FCR The most prominent theoretical framework for conceptualizing FCR is an adaptation of Leventhal’s Self-Regulation Model of Illness (SRM [10]) by Lee-Jones and colleagues [11]. The SRM posits a network of processes, beginning with antecedents or triggers of FCR and ending with its consequences. Posited triggers of FCR include internal cues, such as experiencing pain and other physical symptoms, and external cues, such as annual cancer screenings and commercials for cancer treatment. The SRM points to both behavioral and psychological consequences of FCR, including checking for signs or symptoms of cancer and anxiety. It is important to note that some view these triggers and consequences as focal aspects of the clinical manifestation of FCR, such that clinical FCR is defined and measured, in part, by their presence [1, 12–14]. Although there is currently a lack of consensus as to the best way to define and measure both clinical and non-clinical FCR, we conceptualize triggers and consequences as important and related constructs, but distinct from FCR [12, 15, 16]. Rather, the central feature of FCR is viewed as thoughts and fears related to the possibility of recurrence. Psychological consequences (or correlates) of FCR are one well-studied aspect of the SRM framework [4]. However, there have been few empirical tests of the proximal triggers and behavioral consequences of FCR. In terms of the theorized triggers, studies have found cross-sectional associations between FCR and internal triggers including pain, physical symptoms, fatigue, and changes in appearance (for a review, see [4, 17]). These studies indicate that patients who experience more physical symptoms, treatment side effects, or changes to their appearance tend to be the same patients who report greater FCR. Research on external triggers of FCR is less developed. Qualitative work suggests that common external triggers include doctor appointments, hearing about someone else having cancer, and annual mammograms [8, 18]. Behavioral consequences of FCR have also received little empirical attention. One theorized behavioral consequence is checking the body for signs or symptoms of cancer [11]. It is thought that when a patient experiences FCR, one way she may attempt to regulate her distress is by seeking confirmation that no new signs or symptoms have surfaced. When she gains this confirmation by checking her body, she may be reassured in the moment that her cancer has not progressed or recurred. Thus, theoretically, checking behavior temporarily attenuates the distress associated with FCR. For patients who experience FCR frequently, checking behavior may become excessive, problematic, and even compulsive. Over the past several decades, scholars have frequently described and drawn upon this theorized process [19–23]. In fact, excessive checking behavior is an explicit target of an intervention designed to reduce FCR [24]. Although checking behavior may attenuate FCR, excessive use of this strategy may lead to other negative outcomes. Guidelines from the American Society of Clinical Oncology recommend monthly breast self-examination for BC survivors [25] based on evidence that a greater frequency increases rates of invasive procedures resulting in benign findings, but does not improve rates of mortality [26]. Self-examination is not necessarily a certain or straightforward method for assessing signs or symptoms of cancer progression or recurrence, especially when performed in an emotionally aroused state [19, 27, 28]. Thus, checking in response to FCR may actually evoke more uncertainty about disease status or lead to unnecessary doctor appointments and procedures. Fear of Recurrence and Checking Behavior as Within-Person Phenomena The strong theoretical and intuitive interest in checking behavior does not currently correspond to its empirical support as a consequence of FCR. Only two quantitative studies, to our knowledge, have assessed FCR and checking behavior in BC patients. In a cross-sectional study, Thewes and colleagues [29] found that greater FCR was associated with more frequent self-reported breast self-examination. In another cross-sectional study, survivors who had more severe FCR reported more frequent use of checking as a strategy to cope with FCR [16]. In related research on healthy women with elevated BC risk, several studies have found positive links between BC worry and self-exam frequency [30, 31], although others have not [32]. Although preliminary support for the SRM is encouraging, it is exclusively based on cross-sectional data. Cross-sectional designs lend themselves well to the study of between-person associations, which describe the comparison of individuals to one another based on their global or average levels of a particular variable. For example, the cross-sectional study reviewed above suggests that patients with greater global FCR tend to report a higher frequency of breast self-examination than patients with lower global FCR [29]. However, typically also of interest are within-person associations, which describe the comparison of individuals to themselves (rather than to other individuals) based on their within-person variability captured in repeated measures. For example, is a patient more likely to report checking behavior on a day in which she has higher FCR compared with a day in which she has lower FCR? Longitudinal studies are required to assess both between- and within-person effects. Echoing Ryu, West, and Sousa [33], we argue that the within-person association between FCR and checking behavior is of substantive theoretical and clinical interest, and it is important to examine whether the between-person effects found previously extend to the within-person level. The presence of a between-person effect does not imply the presence of a within-person effect and vice versa; in fact, these two effects can differ in magnitude and even direction [34]. Although prior cross-sectional findings suggest that individuals with higher FCR tend to report checking more frequently, global and retrospective assessment may only distally capture this link. Global cross-sectional methods are vulnerable to retrospective biases [35] and do not rule out unmeasured between-person confounds or address the directionality of an association [36–38]. Behavioral consequences of FCR, including checking behavior, have not yet been examined prospectively and at the within-person level, despite the fact that checking behavior is targeted in FCR interventions. Moreover, FCR triggers hypothesized to initiate the SRM sequence, while well-studied cross-sectionally, have also not yet been examined at this level. Intensive longitudinal methods, which involve repeated measurements of individuals over time (e.g., daily diaries [37]), can provide valuable information that maps onto our theoretical and intuitive understanding of FCR. This approach can limit the retrospective biases associated with global self-report of behaviors or feelings over a longer span of time and provide a clearer picture of the processes flowing into and from FCR [35–38]. Furthermore, intensive longitudinal data offer the potential of identifying the directionality and sequence of effects [36, 37]. For example, one can examine whether a particular effect carries over from one assessment to the next, which could provide insight into the potency and duration of the effect. This level of insight can be easily translated and applied to the development and refinement of FCR interventions. The Present Study We examined the associations between triggering events, FCR, and checking behavior on a daily, within-person level, which has not been previously examined. Daily diary data from early-stage BC patients were used to accomplish several aims. Our first aim was to determine whether FCR was associated with checking behavior on the same day (concurrently). Based on the SRM, we hypothesized that patients who reported greater FCR on one day would be more likely to check for signs or symptoms of cancer on that same day. Note that a concurrent within-person association is not synonymous with a cross-sectional between-person association—the former models within-person day-to-day variability while the latter does not. We also sought to explore the directionality of the association between FCR and checking behavior by examining carryover of the effect—in both directions—from one day to the next. No prior research has attempted to delineate the timeline for these effects. Furthermore, it has been suggested that checking behavior may reinforce and strengthen FCR, but this has not been examined empirically. Thus, an exploratory aim was to investigate whether FCR was associated with a greater likelihood of checking the next day, and vice versa. Finally, we examined the posited sequence of triggers predicting FCR, in turn, leading to checking behavior. Our second aim was to determine the extent to which the effect of daily triggers on checking behavior is mediated by FCR. Based on the SRM, we hypothesized that FCR would mediate the within-person association between triggers and checking behavior. To rule out the possibility that feelings of negativity could explain the hypothesized links among these variables, we included momentary negative affect as a within-person covariate in all tests. Method Participants Participants were female early-stage BC patients recruited from a community cancer center in the mid-Atlantic region of the USA for a larger longitudinal study with Institutional Review Board approval. Patients were asked to participate if they met the following criteria: (a) diagnosed with Stage 0 (lobular/ductal carcinoma in situ), I, II, or IIIA BC; (b) had recent BC surgery; (c) in a committed romantic relationship with a partner who also agreed to participate; (d) English-speaking; (e) lived within an hour of the recruitment site; and (f) did not have prior cancer diagnoses. Patients who had a recent positive breast biopsy and appeared eligible per electronic medical records were sent a letter about the study and contacted by phone. Note that although the larger study obtained data from patients and their partners, only patient data are examined here. Of the 1,161 patients who had a recent positive breast biopsy, 698 were identified via medical records as ineligible. The 463 potentially eligible patients were mailed a letter and contacted by phone. Of those who were contacted, 117 verbally agreed to participate and were mailed a consent form, 110 were ineligible based on patient-provided information (78% ineligible due to being unpartnered), 82 could not be reached, and 154 declined to participate (most frequent reasons were “not enough time” and “spouse does not wish to participate”). Seventy-nine patients provided informed consent and participated, 4 provided consent but did not participate, and 34 did not return the consent form. This article examined one assessment period of the larger longitudinal study (discussed in more detail below), to which 72 patients contributed data, resulting in a final sample of 72 patients described here. Approximately 89% of patients identified as Caucasian, 10% as African American, 1% as Asian, and all as non-Hispanic. The average age was 54.54 years (SD = 9.48) and 94% were married. In terms of occupational status, about 43% reported not working, 39% reported working full time, and 18% reported working part time. Most (73%) reported an annual household income over $60,000. Fifteen percent of patients were diagnosed with Stage 0 BC, 49% Stage IA, 24% Stage IIA, 11% Stage IIB, and 1% Stage IIIA. The majority of patients did not receive adjuvant chemotherapy (67%) but did receive adjuvant radiation (71%) and hormonal therapy (81%). Procedure Patients were recruited around the start of adjuvant treatment. After providing informed consent, patients completed a series of baseline questionnaires. The baseline assessment took place an average of 3 months after surgery (SD = 1.56, range = 1–8). Data for the current study came from a daily diary period that took place after the baseline assessment, as soon as possible after adjuvant treatment. This time period was targeted for study based on evidence suggesting that FCR emerges soon after the end of adjuvant treatment, when patients see their health care providers less frequently and often feel as though they are no longer actively fighting the cancer [39, 40]. Patients who did not receive adjuvant chemotherapy or radiotherapy completed the diary period as soon as possible after the baseline assessment. Patients started the diary period an average of 5 months after surgery (SD = 2.09, range = 2–12). The daily diary period took place over 21 consecutive days. During this period, patients were asked to complete a short survey every evening within about an hour of going to sleep. Before the diary period began, patients were provided with a web link to access this survey from their home each evening. The evening survey was designed to be brief and took patients 13 min on average to complete. Because several items asked patients about the events of their day, compliance with the requested completion time was monitored via the survey software. Data were only included in analyses if the survey was completed between 6:00 p.m. and 3:00 a.m. On average, patients completed the surveys on 17 out of the 21 days (81% compliance rate). Measures Daily FCR Six items from the Fear of Cancer Recurrence Inventory (FCRI [8]), which were among the highest-loading items in past analytic work [8], were adapted for daily use. One item (from the FCRI severity subscale) assessed how much time patients spent thinking about the possibility of cancer recurrence that day, with responses ranging from 0 (didn’t think about it at all) to 4 (several hours). Four items (from the distress subscale) assessed negative emotional reactions (worried, afraid, or anxious; sad, discouraged, or disappointed; frustrated, angry, or outraged; helpless or resigned) related to the possibility of recurrence. The fifth item, drawn from the insight subscale, assessed the extent to which one felt that he/she “worried excessively” about the possibility of recurrence. The latter five items were rated from 0 (not at all) to 4 (extremely). The six items were summed to create a daily FCR composite, which had strong within-person reliability (ω = .91). The average daily FCR score was 2.09 (within-person SD = 2.53, between-person SD = 2.20). FCR was positively skewed with a large proportion (57.8%) of zero scores. As a result, FCR was modeled as a multilevel count outcome with a Poisson distribution in all analyses. Daily triggers Eight daily triggers were assessed. As described earlier, although some include triggers in their measurement of overall FCR, we conceptualize these as distinct constructs. In the absence of a previously validated daily measure, these items were developed through consultation with a clinical psycho-oncologist and a nurse navigator specializing in cancer survivorship needs. Several items overlapped with items from the FCRI triggers subscale [8]. Patients indicated whether each trigger occurred that day. The following five internal triggers were assessed: noticed skin irritation, tingling or loss of sensation in hands or feet, had swelling or trouble moving the arm, cognitive difficulties (e.g., issues with memory, everyday decision making, focusing, finding words), and noticed changes in physical appearance (e.g., scars, hair loss, change in complexion). Three external triggers were assessed: had a long wait in doctor’s office, received negative news from physician, and frustrating interaction with medical professional. The eight items were summed to represent the number of triggers experienced daily. The average number of daily triggers was 0.86 (within-person SD = 0.76, between-person SD = 0.84). Although it could be argued that internal and external triggers should be examined separately, the current sample size, limited number of items, and relatively low frequency of triggers led to the decision to utilize a single trigger composite. Daily checking behavior One item assessed whether patients engaged in daily checking behavior: “Today, did you examine yourself physically for signs or symptoms of cancer?” Patients responded yes or no. On average, patients reported checking on 19% of the diary days (within-person SD = 0.32, between-person SD = 0.23). Momentary negative affect Momentary negative affect was measured using seven standard items from the negative affect subscale of the Positive and Negative Affect Schedule Expanded Form [41], which were averaged to create a composite. Only seven negative affect items (sad, angry, afraid, lonely, blue, scared, and frightened) were included to keep the daily diaries as brief as possible. Patients indicated the extent they experienced each item “at this moment” from 0 (not at all) to 4 (extremely). Coefficient omega estimated acceptable within-person reliability of the momentary negative affect scale (ω = .79). The average momentary negative affect score was 0.17 (within-person SD = 0.26, between-person SD = 0.24). Results Data Analytic Approach See Supplementary Material 1 for additional descriptive statistics and correlations. Multilevel modeling was used to accommodate the nested structure of the daily diary data (i.e., days nested within patients). Multilevel analyses were conducted in R (version 3.3.2; R Foundation for Statistical Computing, Vienna, Austria) using the lme4 package (version 1.1–12 [42]). The linear effect of time and momentary negative affect were both included as within-person covariates in all models (note that the exclusion of negative affect as a covariate resulted in the same pattern of findings for all analyses). Time is included as a covariate to control for autocorrelation in Level 1 residuals as well as any systematic increase or decrease in the outcome over the diary period (e.g., as a result of reactivity to study procedures [37]). As mentioned earlier, since, by definition, person-level (i.e., between-person) factors, such as global or baseline levels of FCR severity, cannot confound within-person relationships, such factors need not be included as covariates in these models. A random intercept was estimated in each model, allowing for person-to-person variability in levels of the outcome. Random slope effects for focal predictors were estimated when possible. Time-varying predictors were person-mean centered and person-level predictors were grand-mean centered [37]. The multilevel model provides valid inferences assuming data are missing at random [43]. To address our first aim, we examined the concurrent daily association between FCR and checking behavior (a binary outcome variable) by estimating a multilevel logistic regression model. Daily FCR was specified as a predictor of daily checking behavior (the outcome variable), for which a random slope effect was estimated. In addition to the within-person effect of interest, we estimated the between-person effect by including average FCR over the diary period as a person-level predictor. To explore the directionality of the association between FCR and checking behavior, we examined whether FCR predicted not only concurrent checking behavior, but also future checking behavior, and vice versa. An additional set of models were estimated in which the outcome was set one day after the predictor. First, same-day FCR was specified as a predictor of next-day checking behavior (the outcome variable), while controlling for same-day checking behavior. We then specified same-day checking behavior as a predictor of next-day FCR (the outcome variable), while controlling for same-day FCR. A random slope effect for the focal predictor of each model (i.e., FCR or checking behavior) was also estimated. For our second aim, we conducted a 1→1→1 multilevel mediation analysis [44–46], in which the predictor (triggering events), mediator (FCR), and outcome (checking behavior) were all time-varying variables (i.e., measured daily). The hypothesized mediation model is depicted in Fig. 1. Because FCR was modeled as a count outcome and checking behavior as a binary outcome, the traditional approach to mediation analysis could not be used—the standard test of the mediated effect (i.e., the product of the indirect paths [47]) would produce a biased estimate [48]. To obtain a valid estimate of the indirect effect for our model, we used a more general approach to mediation analysis [48] based on the counterfactual, or potential outcomes, framework [49, 50]. This more general approach to mediation analysis accommodates conditions that traditional approaches do not, including noncontinuous outcome and mediator distributions (e.g., here, checking behavior is a binary outcome modeled using multilevel logistic regression [51]). The mediation package in R was used to test the hypothesized mediated effect and obtain valid estimates of the average direct, mediated, and total effects in the multilevel context (v4.4.5 [52]). For readers interested in a detailed description of this approach and how it differs from traditional mediational analysis, see Supplementary Material 2 and [48, 49, 51, 52]. Fig. 1. View largeDownload slide Hypothesized within-person multilevel mediation model. The a path was estimated by regressing fear of cancer recurrence (FCR) on daily triggers, controlling for momentary negative affect, average triggers, and time, with a random slope estimated for daily triggers. The b and c′ paths were estimated by regressing checking behavior on daily FCR and daily triggers. Although not depicted, we controlled for momentary negative affect, average FCR, average triggers, and time. Fig. 1. View largeDownload slide Hypothesized within-person multilevel mediation model. The a path was estimated by regressing fear of cancer recurrence (FCR) on daily triggers, controlling for momentary negative affect, average triggers, and time, with a random slope estimated for daily triggers. The b and c′ paths were estimated by regressing checking behavior on daily FCR and daily triggers. Although not depicted, we controlled for momentary negative affect, average FCR, average triggers, and time. Daily Link Between FCR and Checking Behavior The results of our first aim (determine whether FCR was associated with same-day checking behavior) are displayed in Table 1. FCR emerged as a significant within-person predictor of checking behavior, indicating that when a patient experiences a one-unit increase in FCR (compared with what is typical for her), the odds are 32% greater that she will check herself for signs or symptoms of cancer the same day. The random effect variance of this slope was 0.03 (SD = 0.18), indicating that the slopes of about 95% of patients fell between −0.07 (odds ratio [OR] = 0.93) and 0.64 (OR = 1.89). Thus, for some patients, the effect of FCR on checking behavior approached zero, while for others, the effect was greater than the average effect. In addition, average FCR over the diary period was a significant predictor of average checking behavior (γ = 0.37, OR = 1.45, z = 3.18, p = .002). This between-person effect indicates that patients who tend toward higher average daily FCR also tend toward more frequent checking. Table 1 Multilevel regression results of concurrent and prospective link between FCR and checking behavior Effect  Estimate  SE  Odds/rate ratio  p  95% CI  Lower  Upper  Concurrent link: FCR → Same-day checking behaviora  FCR  0.28  0.06  1.32  <.001  0.162  0.398  Negative affect  −0.57  0.42  0.57  .171  −1.393  0.253  Time  0.01  0.02  1.01  .546  −0.029  0.049  Prospective link: FCR → Next-day checking behaviora  FCR  0.04  0.06  1.04  .473  −0.078  0.158  Negative affect  −0.24  0.40  0.79  .574  −1.024  0.544  Time  −0.01  0.02  0.99  .710  −0.049  0.029  Prospective link: Checking behavior → Next-day FCRb  Checking behavior  0.31  0.20  1.37  .112  −0.082  0.702  Negative affect  0.28  0.06  1.32  <.001  0.162  0.398  Time  −0.01  0.00  0.99  .005  −0.018  −0.002  Effect  Estimate  SE  Odds/rate ratio  p  95% CI  Lower  Upper  Concurrent link: FCR → Same-day checking behaviora  FCR  0.28  0.06  1.32  <.001  0.162  0.398  Negative affect  −0.57  0.42  0.57  .171  −1.393  0.253  Time  0.01  0.02  1.01  .546  −0.029  0.049  Prospective link: FCR → Next-day checking behaviora  FCR  0.04  0.06  1.04  .473  −0.078  0.158  Negative affect  −0.24  0.40  0.79  .574  −1.024  0.544  Time  −0.01  0.02  0.99  .710  −0.049  0.029  Prospective link: Checking behavior → Next-day FCRb  Checking behavior  0.31  0.20  1.37  .112  −0.082  0.702  Negative affect  0.28  0.06  1.32  <.001  0.162  0.398  Time  −0.01  0.00  0.99  .005  −0.018  −0.002  While not shown, same-day level of the outcome was also included as a covariate in prospective models. FCR = daily fear of cancer recurrence. aPoisson regression estimates are exponentiated to obtain rate ratios. bLogistic regression estimates are exponentiated to obtain odds ratios. View Large Table 1 Multilevel regression results of concurrent and prospective link between FCR and checking behavior Effect  Estimate  SE  Odds/rate ratio  p  95% CI  Lower  Upper  Concurrent link: FCR → Same-day checking behaviora  FCR  0.28  0.06  1.32  <.001  0.162  0.398  Negative affect  −0.57  0.42  0.57  .171  −1.393  0.253  Time  0.01  0.02  1.01  .546  −0.029  0.049  Prospective link: FCR → Next-day checking behaviora  FCR  0.04  0.06  1.04  .473  −0.078  0.158  Negative affect  −0.24  0.40  0.79  .574  −1.024  0.544  Time  −0.01  0.02  0.99  .710  −0.049  0.029  Prospective link: Checking behavior → Next-day FCRb  Checking behavior  0.31  0.20  1.37  .112  −0.082  0.702  Negative affect  0.28  0.06  1.32  <.001  0.162  0.398  Time  −0.01  0.00  0.99  .005  −0.018  −0.002  Effect  Estimate  SE  Odds/rate ratio  p  95% CI  Lower  Upper  Concurrent link: FCR → Same-day checking behaviora  FCR  0.28  0.06  1.32  <.001  0.162  0.398  Negative affect  −0.57  0.42  0.57  .171  −1.393  0.253  Time  0.01  0.02  1.01  .546  −0.029  0.049  Prospective link: FCR → Next-day checking behaviora  FCR  0.04  0.06  1.04  .473  −0.078  0.158  Negative affect  −0.24  0.40  0.79  .574  −1.024  0.544  Time  −0.01  0.02  0.99  .710  −0.049  0.029  Prospective link: Checking behavior → Next-day FCRb  Checking behavior  0.31  0.20  1.37  .112  −0.082  0.702  Negative affect  0.28  0.06  1.32  <.001  0.162  0.398  Time  −0.01  0.00  0.99  .005  −0.018  −0.002  While not shown, same-day level of the outcome was also included as a covariate in prospective models. FCR = daily fear of cancer recurrence. aPoisson regression estimates are exponentiated to obtain rate ratios. bLogistic regression estimates are exponentiated to obtain odds ratios. View Large Prospective Link Between FCR and Checking Behavior Results of our exploratory aim (determine whether FCR was associated with greater odds of checking the next day, and vice versa) are shown in Table 1. First, we did not find evidence that FCR predicted next-day checking behavior on average (OR = 1.04, p = .473). Thus, our results suggest that FCR is a strong predictor of same-day checking behavior, but does not have a detectable effect on next-day checking behavior (above and beyond its same-day effect). The FCR random slope effect estimated a variance of 0.03 (SD = 0.17), indicating that 95% of individual patient slopes in the population fall between −0.29 (OR = 0.75) and 0.38 (OR = 1.46). Second, results suggested a larger but nonsignificant average effect of checking behavior on next-day FCR (rate ratio [RR] = 1.37, p = .112). When a patient checked, regardless of her FCR score that day, she was predicted to have a 1.37 times greater FCR score the next day. The random slope of checking behavior also pointed to substantial variability (variance = 0.73, SD = 0.85), indicating that 95% of patient slopes fall between −1.40 (RR = 0.25) and 2.02 (RR = 7.53). Within-Person Mediation Model The results of the tests of the hypothesized mediation model (depicted in Fig. 1) are displayed in Table 2. The a path was estimated by regressing FCR on daily triggers, momentary negative affect, average triggers, and time, with a random slope effect of daily triggers also estimated. As expected, triggering events were a significant and positive predictor of same-day FCR. Specifically, on a day a patient experienced one additional triggering event (compared with her average number of daily triggers), she was predicted to have a 1.51 times greater FCR score that same-day. The random effect variance of this fixed effect was 0.14 (SD = 0.37), indicating that about 95% of patient slopes in the population fall between −0.32 (RR = 0.72) and 1.15 (RR = 3.16). Average number of triggers over the diary period also significantly predicted higher average FCR, γ = 0.71, RR = 2.04, z = 3.76, p < .001. Table 2 Results of within-person multilevel mediation: FCR triggering events → FCR → checking behavior Effect  Estimate  SE  Odds/rate ratio  p  95% CI  Lower  Upper  Mediator model: FCR triggers → FCRa  FCR triggers (a path)  0.41  0.08  1.51  <.001  0.259  0.568  Negative affect  0.42  0.05  1.52  <.001  0.313  0.518  Time  −0.02  0.00  0.98  <.001  −0.023  −0.009  Outcome model: Triggers and FCR → Checking behaviorb  FCR (b path)  0.28  0.04  1.33  <.001  0.203  0.365  FCR triggers (c′ path)  0.10  0.13  1.11  .414  −0.146  0.354  Negative affect  −0.60  0.39  0.55  .119  −1.363  0.155  Time  0.01  0.02  1.01  .448  −0.019  0.042  Mediation effectsc  Average mediated effect  0.02  –  –  <.001  0.01  0.04  Average direct effect  0.01  –  –  .42  −0.02  0.04  Total effect  0.04  –  –  .03  0.00  0.07  Proportion mediated effect  0.66  –  –  .03  0.22  2.65  Effect  Estimate  SE  Odds/rate ratio  p  95% CI  Lower  Upper  Mediator model: FCR triggers → FCRa  FCR triggers (a path)  0.41  0.08  1.51  <.001  0.259  0.568  Negative affect  0.42  0.05  1.52  <.001  0.313  0.518  Time  −0.02  0.00  0.98  <.001  −0.023  −0.009  Outcome model: Triggers and FCR → Checking behaviorb  FCR (b path)  0.28  0.04  1.33  <.001  0.203  0.365  FCR triggers (c′ path)  0.10  0.13  1.11  .414  −0.146  0.354  Negative affect  −0.60  0.39  0.55  .119  −1.363  0.155  Time  0.01  0.02  1.01  .448  −0.019  0.042  Mediation effectsc  Average mediated effect  0.02  –  –  <.001  0.01  0.04  Average direct effect  0.01  –  –  .42  −0.02  0.04  Total effect  0.04  –  –  .03  0.00  0.07  Proportion mediated effect  0.66  –  –  .03  0.22  2.65  FCR = daily fear of cancer recurrence. aEstimates from Poisson regression are exponentiated to obtain rate ratios. bEstimates from logistic regression are exponentiated to obtain odds ratios. cMediation effect estimates are displayed in probability units. View Large Table 2 Results of within-person multilevel mediation: FCR triggering events → FCR → checking behavior Effect  Estimate  SE  Odds/rate ratio  p  95% CI  Lower  Upper  Mediator model: FCR triggers → FCRa  FCR triggers (a path)  0.41  0.08  1.51  <.001  0.259  0.568  Negative affect  0.42  0.05  1.52  <.001  0.313  0.518  Time  −0.02  0.00  0.98  <.001  −0.023  −0.009  Outcome model: Triggers and FCR → Checking behaviorb  FCR (b path)  0.28  0.04  1.33  <.001  0.203  0.365  FCR triggers (c′ path)  0.10  0.13  1.11  .414  −0.146  0.354  Negative affect  −0.60  0.39  0.55  .119  −1.363  0.155  Time  0.01  0.02  1.01  .448  −0.019  0.042  Mediation effectsc  Average mediated effect  0.02  –  –  <.001  0.01  0.04  Average direct effect  0.01  –  –  .42  −0.02  0.04  Total effect  0.04  –  –  .03  0.00  0.07  Proportion mediated effect  0.66  –  –  .03  0.22  2.65  Effect  Estimate  SE  Odds/rate ratio  p  95% CI  Lower  Upper  Mediator model: FCR triggers → FCRa  FCR triggers (a path)  0.41  0.08  1.51  <.001  0.259  0.568  Negative affect  0.42  0.05  1.52  <.001  0.313  0.518  Time  −0.02  0.00  0.98  <.001  −0.023  −0.009  Outcome model: Triggers and FCR → Checking behaviorb  FCR (b path)  0.28  0.04  1.33  <.001  0.203  0.365  FCR triggers (c′ path)  0.10  0.13  1.11  .414  −0.146  0.354  Negative affect  −0.60  0.39  0.55  .119  −1.363  0.155  Time  0.01  0.02  1.01  .448  −0.019  0.042  Mediation effectsc  Average mediated effect  0.02  –  –  <.001  0.01  0.04  Average direct effect  0.01  –  –  .42  −0.02  0.04  Total effect  0.04  –  –  .03  0.00  0.07  Proportion mediated effect  0.66  –  –  .03  0.22  2.65  FCR = daily fear of cancer recurrence. aEstimates from Poisson regression are exponentiated to obtain rate ratios. bEstimates from logistic regression are exponentiated to obtain odds ratios. cMediation effect estimates are displayed in probability units. View Large Next, we tested the b and c′ paths by regressing checking behavior on daily FCR, daily triggers, momentary negative affect, average FCR, average triggers, and time. Results are shown in Table 2. No random slope effects were able to be estimated in this model due to nonconvergence. Because we were unable to estimate a random effect for this slope using lme4, we also used Bayesian and generalized estimating equation (GEE) methods to confirm that the focal fixed effect results remained after estimating random slope effects (Bayesian) or correcting standard errors (GEE). These fixed effect results remained statistically significant. Consistent with the results of the first aim, results revealed a significant positive effect of FCR on checking behavior (b path). However, the effect of triggers on checking behavior was not statistically significant (c′ path). The R mediation package uses the results of these models to estimate the mediated effects (provided in probability units; shown in Table 2). The average mediated causal effect was 0.02, indicating that when a patient experiences one more trigger relative to what is typical for her, there is a 2% increase in the probability of her checking on the same day that is due solely to increases in FCR. The average direct effect was 0.01, suggesting that a one-unit increase in daily triggers results in a 1% increase in the probability that the patient will exhibit checking behavior that is not explained by increases in FCR. The average proportion mediated effect is 0.66, indicating that 66% of the total effect of triggers on checking behavior is explained by FCR. Discussion Although a growing literature points to FCR as a critical component of BC adjustment, we are aware of no published work examining antecedents and consequences of FCR within the context of daily life. The present study examined daily checking behavior, a consequence of FCR according to the prominent SRM [11] and a target of FCR interventions [24]. Checking behavior has long been theorized to function as a distress-reducing strategy and resemble a compulsion in patients with very high FCR [11, 19–23]. Surprisingly, this link has only been directly examined in one cross-sectional study [29]. Despite the paucity of empirical support, emerging FCR interventions draw heavily on the SRM and explicitly target excessive checking behavior [24]. Using daily diary data from BC survivors, the current study addresses this gap by examining the proximal (FCR) and distal (FCR-triggering events) predictors of daily checking behavior proposed by the SRM. The current study provides the first test of the within-person processes outlined in the SRM and targeted in FCR interventions but not captured in the existing literature, which is based almost exclusively on global and cross-sectional assessments [16]. Results provided support for our hypothesis that patients who experience greater FCR on one day would be more likely to report checking their bodies for signs or symptoms of cancer. We found that on a day that a patient reports a one-unit increase in FCR, the odds are 32% greater that she will check for signs or symptoms of cancer the same-day. Moreover, this effect was not explained by momentary negative affect during the evening survey. Although these results are based on concurrent data and do not permit inferences about the causal nature of this relationship, they complement prior cross-sectional findings by indicating the presence of a significant within-person association between FCR and checking behavior, which has not been demonstrated previously. Furthermore, our approach minimizes biases associated with typical cross-sectional studies that require patients to retrospectively report on far longer periods of time (e.g., past month or year). In addition, person-level (i.e., between-person) factors, such as global or baseline FCR severity, cannot confound within-person associations between FCR and checking behavior; therefore, any stable factors that vary only from person to person are not plausible alternative explanations for the current within-person findings. We also examined the extent to which the association between FCR and checking behavior holds across different patients, which has not been described previously. The significant fixed effect was found to vary between patients, such that for some, the estimated effect was near zero, while for others, it was over twice that of the average fixed effect. These results point to potential moderating variables that attenuate or intensify the effect of FCR on checking behavior. Although the current study was not designed or sufficiently powered to examine moderators of this effect, future studies should investigate variables that may explain why some patients engage in checking behavior when they experience FCR and others do not. Given the lack of empirical or theoretical work on the timescale of the SRM, we explored carryover effects of FCR on checking behavior the next day. We did not find evidence of an effect of FCR on next-day checking behavior (controlling for same-day checking behavior). This finding does not necessarily support or conflict with the hypothesis that FCR plays a causal role, but does suggest that future studies must implement more frequent assessments to determine the directionality and temporal sequence of these momentary experiences. Intuitively, it seems likely that the effect of FCR on checking behavior occurs within a shorter time frame than one day, making it impossible for a daily diary design to capture. Importantly, there was evidence of variability in this effect, such that greater FCR on one day predicted increased odds of checking the next day for some patients and decreased odds for others. It will be important to investigate moderators that explain these individual differences in future research. Scholars have suggested that checking behavior may maintain or exacerbate FCR [11, 19–23]. We explored this by examining whether checking behavior predicted next-day FCR, controlling for FCR and momentary negative affect the previous day. Although this effect was not statistically significant, there was substantial variability in this effect, such that for some it was strongly negative and others strongly positive. Although checking behavior has been conceptualized as an FCR-reinforcing compulsion, more research is required to determine for whom and under what conditions this rings true. Again, more frequent assessments are likely needed to clarify the sequence and cycle of FCR and checking behavior over time. Finally, we tested a multilevel within-person mediation model representing the full sequence delineated in the SRM—triggering events predict FCR, which, in turn, predicts checking behavior. The results supported this mediation model, indicating that the effect of daily triggers on checking behavior was significantly mediated by daily FCR. The direct effect of daily triggers on checking behavior, not explained by FCR, was not statistically significant. These results support the antecedents, FCR, and consequences posited by the SRM at the same momentary, within-person level that inherently defines the model. Furthermore, the methods employed minimize retrospective bias and eliminate the possibility of person-level confounding variables. Taken together, the results and methodological strengths of the current study provide strong support for the SRM that has not been previously documented. Limitations and Future Directions We attempted to address directionality and temporal precedence in the association between FCR and checking behavior by examining the effects from one day to the next. Because we did not find strong support for these next-day carryover effects, it is possible that checking behavior temporally precedes FCR. Similarly, we cannot rule out the possibility that checking behavior leads to increased sensitivity or awareness of daily triggering events. As described above, it seems plausible that the lack of strong evidence of temporal precedence resulted from an incongruence between the timeframe of assessments and unfolding of effects. It is possible to engage in several acts of checking behavior within a day, which, furthermore, may occur with or without conscious awareness. Future studies should be designed to capture these processes that may unfold over minutes or hours. Our use of intensive longitudinal methodology allowed us to quantify heterogeneity in the within-person effects among BC patients. Although this study was not designed to explain this variability, our results suggest that this is a promising avenue for future research. It will be important to understand the individual characteristics and contextual factors that attenuate or intensify the effect of FCR on checking behavior. For example, patients who perceive themselves at higher risk of recurrence may be more likely to actively check for signs of recurrence [53]. Trait anxiety may also differentiate the patient who has this behavioral response to FCR from the patient who does not [22, 54, 55]. Patients who use adaptive coping strategies, such as disclosing concerns about recurrence to a supportive and responsive loved one, may be less likely to engage in checking behavior. Furthermore, if checking behavior resembles a compulsion, then one may expect for those who check more frequently to exhibit a stronger daily link between FCR and checking behavior. Finally, variability in the association between triggering events and FCR suggests individual differences in reactivity to certain events. For example, patients with higher trait anxiety may be more reactive to events. Relatedly, although our daily triggers measure was based on the widely used and validated FCRI [8], given the paucity of prior studies of FCR in daily life, it is unlikely to capture all possible daily triggers. Future studies with large samples are required to investigate whether between-person factors, such as trait anxiety, moderate the within-person associations between these variables. It is possible that insufficient power contributed to the null finding for checking behavior on next-day FCR. A larger and more diverse sample may be necessary to determine whether checking behavior functions as a FCR-reinforcing compulsion. Patients in the current sample were largely high-functioning and typically reported low levels of daily FCR. The nature of the associations under study may differ across a range of FCR severity, and our findings may not generalize to those suffering from very high levels of FCR. At the same time, our findings highlight the importance of examining the normative experience and dimensional nature of FCR, as they revealed meaningful and theory-driven associations on a day-to-day basis between FCR, triggering events, and checking behavior in our sample. Nonetheless, future work should address these limitations because understanding the potential harm of checking behavior has important implications for mental and physical health care. Ideally, future research will define the bounds between psychologically adaptive and maladaptive checking behavior. In the absence of such work, an empirically-based definition of excessive checking behavior from a psychological standpoint is unavailable. Thus, our current conceptualization of excessive checking behavior relies on the medical literature, which shows that checking more than monthly is maladaptive for physical health [26]. However, FCR interventions that target checking behavior will be more effective if guided by empirical support for the distinction between psychologically adaptive and maladaptive checking behavior, and how and for whom this behavior perpetuates a cyclical process. Such work will also inform the guidance and education about self-examination provided to patients. Little is known about other potential outcomes of checking behavior. The excessive checker may be more vulnerable to false positives that result in unnecessary health care utilization [26]. Theory also points to health care utilization as another consequence of FCR itself, although extant research is limited. The few studies that have examined FCR and health care utilization suggest inconsistent associations across different forms of health care [29, 56, 57]. For example, Thewes and colleagues [29] found that FCR was associated with more frequent self-checking behavior but less frequent formal screenings (e.g., mammograms). Another study found that FCR predicted more outpatient and emergency room visits, but not specialist visits or other health care appointments [56]. More research is needed to gauge the extent and domains of the consequences of FCR, as well as the associations among the consequences themselves—for example, does FCR lead to health care utilization in the absence of excessive checking behavior? Finally, the current sample was fairly homogeneous in terms of racial/ethnic makeup (mostly Caucasian) and income (relatively high on average). These socio-cultural characteristics relate to the low levels of distress observed in this sample, which limit the generalizability of our findings to patients with greater distress and impairment, as noted above. In addition, culture plays a role in the nature of adaptation to cancer [58], and therefore, the reported associations between FCR, triggering events, and checking behavior may not generalize to other socio-cultural groups. More cross-cultural research on FCR and cross-culturally validated measures are sorely needed [59]. The associations studied here must be tested in samples characterized by greater distress and impairment, as well as greater socio-cultural diversity. In addition, patients with prior cancer diagnoses were excluded from the study. Therefore, future research is necessary to determine whether the antecedents and consequences of FCR posited by the SRM hold for patients with longer and more complex histories of cancer-related experiences. Conclusion The SRM is a widely used framework for the conceptualization of FCR and development of FCR interventions. This was the first study to move beyond global and cross-sectional tests of the model and examine the daily within-person processes that are inherent in the framework. Our findings support the posited sequence of triggering events, FCR, and checking behavior. These findings have potential implications for health care utilization, self-checking instruction for BC patients, and FCR intervention development. Supplementary Material Supplementary material is available at Annals of Behavioral Medicine online. Compliance with Ethical Standards Authors’ Statement of Conflict of Interest and Adherence to Ethical StandardsThe authors declare that they have no conflict of interest. Authors' Contributions E. C. Soriano wrote the manuscript with input from all authors. E. C. Soriano and J. P. Laurenceau conducted the analyses. E. C. Soriano, R. Valera, E. C. Pasipanodya, and J. P. Laurenceau conceptualized the paper. E. C. Soriano, E. C. Pasipanodya, and A. K. Otto collected data. S. D. Siegel and J. P. Laurenceau designed the parent study. Ethical Approval All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed Consent Informed consent was obtained from all individual participants included in the study. Acknowledgments Work presented in this manuscript was generously supported by the National Cancer Institute (grant number R21CA171921-02). References 1. Lebel S, Ozakinci G, Humphris Get al.  ; University of Ottawa Fear of Cancer Recurrence Colloquium attendees. From normal response to clinical problem: Definition and clinical features of fear of cancer recurrence. Support Care Cancer . 2016; 24( 8): 3265– 3268. Google Scholar CrossRef Search ADS PubMed  2. Miller KD, Siegel RL, Lin CCet al.   Cancer treatment and survivorship statistics, 2016. CA Cancer J Clin . 2016; 66( 4): 271– 289. Google Scholar CrossRef Search ADS PubMed  3. Baker F, Denniston M, Smith T, West MM. Adult cancer survivors: How are they faring? Cancer . 2005; 104( S11): 2565– 2576. Google Scholar CrossRef Search ADS PubMed  4. Simard S, Thewes B, Humphris Get al.   Fear of cancer recurrence in adult cancer survivors: A systematic review of quantitative studies. J Cancer Surviv . 2013; 7( 3): 300– 322. Google Scholar CrossRef Search ADS PubMed  5. Vickberg SM. The Concerns About Recurrence Scale (CARS): A systematic measure of women’s fears about the possibility of breast cancer recurrence. Ann Behav Med . 2003; 25( 1): 16– 24. Google Scholar CrossRef Search ADS PubMed  6. Deimling GT, Bowman KF, Sterns S, Wagner LJ, Kahana B. Cancer-related health worries and psychological distress among older adult, long-term cancer survivors. Psychooncology . 2006; 15( 4): 306– 320. Google Scholar CrossRef Search ADS PubMed  7. Mellon S, Kershaw TS, Northouse LL, Freeman-Gibb L. A family-based model to predict fear of recurrence for cancer survivors and their caregivers. Psychooncology . 2007; 16( 3): 214– 223. Google Scholar CrossRef Search ADS PubMed  8. Simard S, Savard J. Fear of cancer recurrence inventory: Development and initial validation of a multidimensional measure of fear of cancer recurrence. Support Care Cancer . 2009; 17( 3): 241– 251. Google Scholar CrossRef Search ADS PubMed  9. Mehnert A, Berg P, Henrich G, Herschbach P. Fear of cancer progression and cancer-related intrusive cognitions in breast cancer survivors. Psychooncology . 2009; 18( 12): 1273– 1280. Google Scholar CrossRef Search ADS PubMed  10. Leventhal H, Diefenbach M, Leventhal EA. Illness cognition: Using common sense to understand treatment adherence and affect cognition interactions. Cogn Ther Res . 1992; 16( 2): 143– 163. Google Scholar CrossRef Search ADS   11. Lee-Jones C, Humphris G, Dixon R, Hatcher MB. Fear of cancer recurrence–a literature review and proposed cognitive formulation to explain exacerbation of recurrence fears. Psychooncology . 1997; 6: 95– 105. Google Scholar CrossRef Search ADS PubMed  12. Mutsaers B, Jones G, Rutkowski Net al.   When fear of cancer recurrence becomes a clinical issue: A qualitative analysis of features associated with clinical fear of cancer recurrence. Support Care Cancer . 2016; 24( 10): 4207– 4218. Google Scholar CrossRef Search ADS PubMed  13. Sharpe L, Thewes B, Butow P. Current directions in research and treatment of fear of cancer recurrence. Curr Opin Support Palliat Care . 2017; 11( 3): 191– 196. Google Scholar CrossRef Search ADS PubMed  14. Thewes B, Lebel S, Seguin Leclair C, Butow P. A qualitative exploration of fear of cancer recurrence (FCR) amongst Australian and Canadian breast cancer survivors. Support Care Cancer . 2016; 24( 5): 2269– 2276. Google Scholar CrossRef Search ADS PubMed  15. Costa DS, Smith AB, Fardell JE. The sum of all fears: Conceptual challenges with measuring fear of cancer recurrence. Support Care Cancer . 2016; 24( 1): 1– 3. Google Scholar CrossRef Search ADS PubMed  16. Custers JAE, Gielissen MFM, de Wilt JHWet al.   Towards an evidence-based model of fear of cancer recurrence for breast cancer survivors. J Cancer Surviv . 2017; 11( 1): 41– 47. Google Scholar CrossRef Search ADS PubMed  17. Crist JV, Grunfeld EA. Factors reported to influence fear of recurrence in cancer patients: A systematic review. Psychooncology . 2013; 22( 5): 978– 986. Google Scholar CrossRef Search ADS PubMed  18. Gill KM, Mishel M, Belyea Met al.   Triggers of uncertainty about recurrence and long-term treatment side effects in older African American and Caucasian breast cancer survivors. Oncol Nurs Forum . 2004; 31( 3): 633– 639. Google Scholar CrossRef Search ADS PubMed  19. Fardell JE, Thewes B, Turner Jet al.   Fear of cancer recurrence: A theoretical review and novel cognitive processing formulation. J Cancer Surviv . 2016; 10( 4): 663– 673. Google Scholar CrossRef Search ADS PubMed  20. Ghazali N, Cadwallader E, Lowe D, Humphris G, Ozakinci G, Rogers SN. Fear of recurrence among head and neck cancer survivors: Longitudinal trends. Psychooncology . 2013; 22( 4): 807– 813. Google Scholar CrossRef Search ADS PubMed  21. Lasry JC, Margolese RG. Fear of recurrence, breast-conserving surgery, and the trade-off hypothesis. Cancer . 1992; 69( 8): 2111– 2115. Google Scholar CrossRef Search ADS PubMed  22. Stark DP, House A. Anxiety in cancer patients. Br J Cancer . 2000; 83( 10): 1261– 1267. Google Scholar CrossRef Search ADS PubMed  23. Ziner KW, Sledge GW, Bell CJ, Johns S, Miller KD, Champion VL. Predicting fear of breast cancer recurrence and self-efficacy in survivors by age at diagnosis. Oncol Nurs Forum . 2012; 39( 3): 287– 295. Google Scholar CrossRef Search ADS PubMed  24. Humphris G, Ozakinci G. The AFTER intervention: A structured psychological approach to reduce fears of recurrence in patients with head and neck cancer. Br J Health Psychol . 2008; 13( 2): 223– 230. Google Scholar CrossRef Search ADS PubMed  25. Khatcheressian JL, Hurley P, Bantug Eet al.  ; American Society of Clinical Oncology. Breast cancer follow-up and management after primary treatment: American Society of Clinical Oncology clinical practice guideline update. J Clin Oncol . 2013; 31( 7): 961– 965. Google Scholar CrossRef Search ADS PubMed  26. Thomas DB, Gao DL, Ray RMet al.   Randomized trial of breast self-examination in Shanghai: Final results. J Natl Cancer Inst . 2002; 94( 19): 1445– 1457. Google Scholar CrossRef Search ADS PubMed  27. Miller SM. Monitoring versus blunting styles of coping with cancer influence the information patients want and need about their disease. Implications for cancer screening and management. Cancer . 1995; 76( 2): 167– 177. Google Scholar CrossRef Search ADS PubMed  28. Taylor C, Richardson A, Cowley S. Surviving cancer treatment: An investigation of the experience of fear about, and monitoring for, recurrence in patients following treatment for colorectal cancer. Eur J Oncol Nurs . 2011; 15( 3): 243– 249. Google Scholar CrossRef Search ADS PubMed  29. Thewes B, Butow P, Bell MLet al.  ; FCR Study Advisory Committee. Fear of cancer recurrence in young women with a history of early-stage breast cancer: A cross-sectional study of prevalence and association with health behaviours. Support Care Cancer . 2012; 20( 11): 2651– 2659. Google Scholar CrossRef Search ADS PubMed  30. Cohen M. Breast cancer early detection, health beliefs, and cancer worries in randomly selected women with and without a family history of breast cancer. Psychooncology . 2006; 15( 10): 873– 883. Google Scholar CrossRef Search ADS PubMed  31. Erblich J, Bovbjerg DH, Valdimarsdottir HB. Psychological distress, health beliefs, and frequency of breast self-examination. J Behav Med . 2000; 23( 3): 277– 292. Google Scholar CrossRef Search ADS PubMed  32. Posluszny DM, McFeeley S, Hall L, Baum A. Stress, breast cancer risk, and breast self-examination: Chronic effects of risk and worry. J Appl Biobehav Res . 2004; 9( 2): 91– 105. Google Scholar CrossRef Search ADS   33. Ryu E, West SG, Sousa KH. Distinguishing between-person and within-person relationships in longitudinal health research: Arthritis and quality of life. Ann Behav Med . 2012; 43( 3): 330– 342. Google Scholar CrossRef Search ADS PubMed  34. Hamaker EL. Why researchers should think “within-person”: A paradigmatic rationale. In: Mehl MR, Conner TS, Csikszentmihalyi M, eds. Handbook of Research Methods for Studying Daily Life . Paperback ed. New York, NY: Guilford; 2012: 43– 61. 35. Shiffman S, Stone AA, Hufford MR. Ecological momentary assessment. Annu Rev Clin Psychol . 2008; 4( 1): 1– 32. Google Scholar CrossRef Search ADS PubMed  36. Bolger N, Davis A, Rafaeli E. Diary methods: Capturing life as it is lived. Annu Rev Psychol . 2003; 54( 1): 579– 616. Google Scholar CrossRef Search ADS PubMed  37. Bolger N, Laurenceau J-P. Intensive Longitudinal Methods: An Introduction to Diary and Experience Sampling Research . New York, NY: Guilford Press; 2013. 38. Laurenceau J-P, Bolger N. Using diary methods to study marital and family processes. J Fam Psychol . 2005; 19( 1): 86– 97. Google Scholar CrossRef Search ADS PubMed  39. King MT, Kenny P, Shiell A, Hall J, Boyages J. Quality of life three months and one year after first treatment for early stage breast cancer: Influence of treatment and patient characteristics. Qual Life Res . 2000; 9( 7): 789– 800. Google Scholar CrossRef Search ADS PubMed  40. McKinley ED. Under toad days: Surviving the uncertainty of cancer recurrence. Ann Intern Med . 2000; 133( 6): 479– 480. Google Scholar CrossRef Search ADS PubMed  41. Watson D, Clark LA, Harkness AR. Structures of personality and their relevance to psychopathology. J Abnorm Psychol . 1994; 103( 1): 18– 31. Google Scholar CrossRef Search ADS PubMed  42. Bates D, Mächler M, Bolker B, Walker S. Fitting linear mixed-effects models using lme4. J Stat Softw . 2015; 67( 1): 1–48. 43. Enders CK. Applied Missing Data Analysis . New York: Guilford Press; 2010. 44. Bauer DJ, Preacher KJ, Gil KM. Conceptualizing and testing random indirect effects and moderated mediation in multilevel models: New procedures and recommendations. Psychol Methods . 2006; 11( 2): 142– 163. Google Scholar CrossRef Search ADS PubMed  45. Kenny DA, Korchmaros JD, Bolger N. Lower level mediation in multilevel models. Psychol Methods . 2003; 8( 2): 115– 128. Google Scholar CrossRef Search ADS PubMed  46. Krull JL, MacKinnon DP. Multilevel modeling of individual and group level mediated effects. Multivariate Behav Res . 2001; 36( 2): 249– 277. Google Scholar CrossRef Search ADS PubMed  47. Baron RM, Kenny DA. The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. J Pers Soc Psychol . 1986; 51( 6): 1173– 1182. Google Scholar CrossRef Search ADS PubMed  48. Imai K, Keele L, Tingley D. A general approach to causal mediation analysis. Psychol Methods . 2010; 15( 4): 309– 334. Google Scholar CrossRef Search ADS PubMed  49. Imai K, Jo B, Stuart EA. Using potential outcomes to understand causal mediation analysis: Comment on. Multivariate Behav Res . 2011; 46( 5): 861– 873. Google Scholar CrossRef Search ADS PubMed  50. Rubin DB. Causal inference using potential outcomes: Design, modeling, decisions. J Am Stat Assoc . 2005; 100( 469): 322– 331. Google Scholar CrossRef Search ADS   51. Valeri L, Vanderweele TJ. Mediation analysis allowing for exposure-mediator interactions and causal interpretation: Theoretical assumptions and implementation with SAS and SPSS macros. Psychol Methods . 2013; 18( 2): 137– 150. Google Scholar CrossRef Search ADS PubMed  52. Tingley D, Yamamoto T, Hirose K, Keele L, Imai K. Mediation: R package for causal mediation analysis. J Stat Softw . 2014; 59( 5): 1– 38. Google Scholar CrossRef Search ADS PubMed  53. McCaul KD, Schroeder DM, Reid PA. Breast cancer worry and screening: Some prospective data. Health Psychol . 1996; 15( 6): 430– 433. Google Scholar CrossRef Search ADS PubMed  54. Cameron LD, Leventhal H, Love RR. Trait anxiety, symptom perceptions, and illness-related responses among women with breast cancer in remission during a tamoxifen clinical trial. Health Psychol . 1998; 17( 5): 459– 469. Google Scholar CrossRef Search ADS PubMed  55. Cohen M. First-degree relatives of breast-cancer patients: Cognitive perceptions, coping, and adherence to breast self-examination. Behav Med . 2002; 28: 15– 22. Google Scholar CrossRef Search ADS PubMed  56. Lebel S, Tomei C, Feldstain A, Beattie S, McCallum M. Does fear of cancer recurrence predict cancer survivors’ health care use? Support Care Cancer . 2013; 21( 3): 901– 906. Google Scholar CrossRef Search ADS PubMed  57. Sarkar S, Sautier L, Schilling G, Bokemeyer C, Koch U, Mehnert A. Anxiety and fear of cancer recurrence and its association with supportive care needs and health-care service utilization in cancer patients. J Cancer Surviv . 2015; 9( 4): 567– 575. Google Scholar CrossRef Search ADS PubMed  58. Curran L, Sharpe L, Butow P. Anxiety in the context of cancer: A systematic review and development of an integrated model. Clin Psychol Rev . 2017; 56: 40– 54. Google Scholar CrossRef Search ADS PubMed  59. Lebel S, Ozakinci G, Humphris Get al.   Current state and future prospects of research on fear of cancer recurrence. Psychooncology . 2017; 26( 4): 424– 427. Google Scholar CrossRef Search ADS PubMed  © Society of Behavioral Medicine 2018. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Annals of Behavioral MedicineOxford University Press

Published: May 16, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off