An Exploration into the Relationship between Insomnia and Repetitive Negative Thinking among Cancer Survivors

Abstract

Objective:

Insomnia and repetitive negative thinking (RNT) are both prevalent among cancer survivors, yet little work has investigated their interrelationship. To explore the hypothesis that RNT and insomnia are related, we conducted secondary analyses on data from a pilot clinical trial of cognitive behavioral therapy for insomnia (CBT-I) for cancer survivors to highlight areas of commonality and identify potential targets for future interventions.

Methods:

This study analyzed survey data from 40 cancer survivors with insomnia who participated in a pilot randomized trial of CBT-I. Correlations and linear regression models were used to determine associations between aspects of RNT and related constructs (fear of cancer recurrence [FCR], cancer-specific rumination, worry, and intolerance of uncertainty) and sleep (insomnia and sleep quality), while accounting for psychiatric symptoms such as anxiety and depression. Treatment-related change in RNT was examined using a series of linear mixed models.

Results:

Evidence for an association between RNT and insomnia among cancer survivors emerged. Higher levels of FCR and cancer-related rumination were correlated with more severe insomnia symptoms and worse sleep quality; FCR continued to predict insomnia controlling for anxiety and depression. Results identified potential benefits and limitations of CBT-I in addressing RNT that should be examined more thoroughly in future research.

Conclusions:

RNT is a potential target to consider in insomnia treatment for cancer survivors.

Introduction:

An estimated 30–50% of cancer survivors experience chronic insomnia persisting at least three months.^1,2 Among cancer survivors, insomnia has been linked with impairments in daily living, mood, health, and quality of life.³ Insomnia is commonly perpetuated by late effects of cancer treatments (e.g., hot flashes, pain, fatigue), physiological factors (e.g., circadian rhythm), psychological factors (e.g., worry), and behavioral factors (e.g., sleep schedule disruptions, disrupted sleep hygiene),^2,4–6 several of which are modifiable with evidence-based interventions.

Cognitive behavioral therapy for insomnia (CBT-I) is the gold standard for treating insomnia in cancer survivors and the general population.^3,7–9 Meta-analyses by our team and others have found that existing CBT-I programs tested with cancer survivors lead to medium sized improvements in insomnia severity, sleep efficiency, sleep onset latency, and wake after sleep onset. However, other sleep-related outcomes have more modest effects (e.g., sleep duration continuity, perceived cognitive functioning depression, anxiety).^10–14 In particular, it there is room to optimize CBT-I treatment to impact mood.³

Repetitive negative thinking (RNT), which includes worry and rumination, is both a risk factor for and a common symptom of anxiety and depression^15–17 RNT has been linked with sleep difficulties in various populations, both with and without psychiatric diagnoses.¹⁵ Cancer survivors may be prone to RNT due to their higher risk for emotional distress and psychiatric disorders compared to the general population.¹⁸

Worry experienced by cancer survivors is often (although not always) tied to uncertainty inherent to their status as survivors of serious illness.^19,20 Common sources of worry experienced by survivors include: fear of cancer recurrence (FCR), anxiety about overlooking symptoms of recurrence, upcoming doctor’s appointments, tests or scans, perceptions of burdening others, and late and long-term effects of cancer and its treatment.^19–23 As many as two-thirds of cancer survivors live with elevated FCR.²³ Observational studies have found that FCR accounts for up to 44% of distress in cancer survivors.²⁴ Like insomnia, the FCR severity is independent of cancer type, treatment history, or time since treatment.^25–33

Rumination, common in depression, is a cognitive style characterized by a tendency to dwell on the causes, meanings, and consequences of one’s negative life experiences and associated distress.¹⁶ For cancer survivors, ruminative thoughts may center on the reasons why one developed cancer, difficult aspects of one’s cancer treatment, and/or the ways in which cancer has negatively impacted one’s life.³⁴ Though there is a paucity of large-scale studies on the prevalence of rumination among survivors, one study of breast and lung cancer patients found that 40–50% endorsed both general and cancer-specific rumination.³⁵ Cancer-related rumination has been linked with increased distress and poorer quality of life across a variety of domains.^36–38

The presence and impact of worry and rumination in survivors are relevant to cancer-related insomnia since many studies link these forms of RNT to insomnia in other populations. In a recent meta-analysis of non-clinical populations, perseverative cognitions including worry and rumination were significantly associated with worse sleep quality, delayed sleep onset, and shorter sleep duration.³⁹ Research indicates that RNT may also maintain sleep disturbance among individuals with primary insomnia, major depression, and generalized anxiety disorder.^8,40 Within clinical populations, associations between rumination and insomnia may be particularly strong when the content of ruminative thoughts are symptom-related,^41,42 which may also be true for cancer survivors.

Although both RNT and insomnia are common among cancer survivors, associations between these constructs in this population are understudied. Cross-sectional studies suggest that sleep-related worry and rumination are more common and severe among survivors with insomnia (versus without)^9,28,35 and that sleep impairment is associated with cancer-related worry interference among women attending a cancer risk evaluation program.⁴³ Two recent FCR interventions that included a CBT skill for managing interference from worry (i.e., worry time) also found improvements in survivors’ sleep quality.^44,45 While these findings point towards a potential role of RNT in insomnia, further investigation is needed to explore the relationship between specific aspects of RNT and insomnia and to develop an understanding of the extent to which CBT-I ameliorates RNT.

To identify whether RNT represents a potential target for future insomnia interventions, we investigated the associations between different aspects of RNT, insomnia, and sleep quality in survivors with insomnia who participated in a randomized clinical trial of CBT-I vs. enhanced usual care. Specifically, associations were examined at a baseline assessment (Aim 1), as well as in response to treatment (Aim 2).

Method:

Preliminary Investigation.

Prior to conducting the analyses associated with Aims 1 and 2, we explored cancer survivors’ perceptions of the relationships between RNT and insomnia by examining data from semi-structured qualitative interviews conducted with 10 cancer survivors with insomnia between April-August 2020. Interviews did not specifically inquire about RNT, but included questions asking survivors to describe their sleep quality, factors that interfere with sleep, and the perceived utility of CBT-I for addressing insomnia and related concerns (for more information on interview procedures and primary qualitative results, see Hall et al., 2022).⁴⁶ Two independent primary coders (BCJ and MHW) identified RNT-related themes; discrepancies were resolved by consensus through discussion with first (KAH) and senior (DLH) authors. As seen in Supplement Table 1, multiple survivors identified worry, and to a lesser extent, rumination as bothersome symptoms that interfered with their sleep and quality of life. These results provided support for the exploratory investigation presented below.

Participants and Procedures.

The current study was a secondary analysis of a pilot randomized controlled trial (NCT04566068) comparing a 4-session CBT-I protocol to enhanced usual care (EUC) for insomnia. Primary results from the study found that the CBT-I protocol was feasible, acceptable, and preliminarily efficacious for addressing self-reported insomnia symptom severity in cancer survivors with insomnia⁴⁶ All procedures were approved by the Dana Farber Harvard Cancer Center IRB (#20–170) and informed consent was obtained prior to participation. Inclusion criteria were: age ≥18; history of non-metastatic, localized, or regional solid or blood malignancy(ies); completed primary cancer treatment (i.e., radiation, surgery, and/or chemotherapy); and chronic insomnia indicated by Diagnostic and Statistical Manual of Mental Disorders, 5^th Edition (DSM-5) criteria and Insomnia Severity Index score ≥ 15.⁴⁷ Patients who completed their primary cancer treatment and were on hormonal and/or other long-term maintenance therapy agents were eligible.

Forty cancer survivors with insomnia were enrolled and randomized to either the Survivorship Sleep Program (SSP; n = 20) or an EUC control group (EUC; n = 20). The SSP is a CBT-I intervention from an evidence-based protocol⁴⁸ modified to integrate feedback from qualitative research with cancer survivors (see Preliminary Investigation above).⁴⁶ Using secure videoconferencing technology, CBT-I was delivered virtually over four weekly synchronous sessions (~45 min each) covering CBT-I skills (stimulus control, “efficient sleep” [i.e., sleep restriction], sleep hygiene, relaxation training, and cognitive therapy [i.e., worry time, reframing negative beliefs about sleep]). Participants also received ~15 min between-session check-ins to problem-solve barriers to practicing CBT-I skills. The EUC condition consisted of an electronic sleep hygiene handout and referral to the academic medical center’s behavioral sleep medicine clinic. Assessments occurred at baseline (T0), post-intervention (T1), and one-month follow up (T2).

Measures.

Sociodemographic and medical characteristics.

Sociodemographic and medical characteristics were self-reported and clinical data were verified by a review of medical records.

Fear of Cancer Recurrence.

FCR was assessed as an index of cancer-specific worry using the validated 9-item Fear of Cancer Recurrence Inventory-Severity (FCRI-Severity) subscale.^49,50 Scores ranged from 0–36 with higher scores indicating greater FCR severity. Internal consistency was good, α = 0.86.

Cancer-Related Rumination.

Cancer-related rumination was measured using a modified version of the 8-item rumination subscale of the Responses to Intrusions Questionnaire (RIQ-R).⁵¹ Participants were instructed to think of their experiences since completing their cancer treatment and report what they did when memories of their experience with cancer popped into their mind. Total scores ranged from 0–24 with higher scores indicating greater cancer-related rumination. Internal consistency was good, α = 0.83.

General Worry.

The general tendency to worry was measured using the validated 16-item Penn State Worry Questionnaire (PSWQ).⁵² Total scores ranged from 16–39, with higher scores indicating greater worry. Internal consistency was excellent, α = 0.92.

Intolerance of Uncertainty.

Intolerance of uncertainty, a construct closely related with worry, was measured using the validated 12-item Intolerance of Uncertainty Scale (IUS-12).⁵³ Total scores ranged from 12–60, with higher scores indicating greater uncertainty intolerance. Internal consistency was also excellent, α = 0.92.

Anxiety and Depression Symptoms.

The Hospital Anxiety and Depression Scale anxiety (HADS-A) and depression (HADS-D) subscales are 7-item validated symptom severity measures.⁵⁴ Across subscales, scores of 0–7, 8–10, and 11–21 indicate normal, borderline, and clinical levels of symptoms. Internal consistency was α = .87 and α =.78 for the anxiety and depression subscales, respectively.

Insomnia.

The Insomnia Severity Index (ISI) is a validated measure frequently used to assess eligibility and outcomes of CBT-I interventions.⁴⁷ Scores of 0–7, 8–14, 15–21, and 22–28 indicate absent, sub-threshold, moderate, and severe insomnia, respectively. Internal consistency was acceptable, α = 0.67.

Sleep Quality.

Sleep quality was assessed using a single item from the Pittsburgh Sleep Quality Index (PSQI).⁵⁵ Scores range from 0–3 with higher scores indicating worse sleep quality.

Analyses.

Analyses were conducted with SPSS v.25. All variables were assessed for normality. Pearson’s r correlations were used to test the degree to which RNT and related constructs (i.e., FCRI, RIQ-R, PSWQ, IUS-12) were associated with insomnia (ISI) and sleep quality (PSQI item) at baseline [T0] for all study participants (N=40). For aspects of RNT significantly associated with sleep, linear regression models were used to test these associations while controlling for anxiety and depression symptom severity (HADS-A, HADS-D). In the parent trial, survivors randomized to EUC (vs SSP) were more likely to be employed (p < .001) and have private insurance (p < .01) at T0. Thus, regression analyses were conducted unadjusted and adjusting for these factors. Outcomes were then analyzed using intent-to-treat, with separate linear mixed models examining the effects of CBT-I (n=20) vs. EUC (n=20) over time. For each outcome, we conducted a random intercept model with group (reference = EUC) and time (reference = baseline [T0]) treated as categorical variables to obtain estimated marginal means and standard errors. Partial r² values were calculated to assess the magnitude of fixed effects.

Results:

Detailed information on the participant characteristics is published elsewhere.⁴⁶ Across the sample (N=40), participants were an average of 53.2 (SD = 14.1) years old and 85% were female. Most participants self-identified as White (85%), 10% identified as Hispanic/Latino, 5% as Asian, and 2.5% as African. Participants had various types of cancers, including breast (n = 26), hematologic (n = 5), thyroid (n = 4), prostate (n = 3), skin (n = 3), bladder (n = 2), ovary (n = 2), lung (n = 1), testicular (n = 1), urethra (n = 1), and/or uterus (n = 1). Participants were an average of 61.6 (SD = 61.5) months since treatment and received various treatments, most commonly surgery (n = 37), radiation therapy (n = 23), IV chemotherapy/other IV cancer therapy (n = 22), and hormone or endocrine therapy (n = 18). The intervention conditions did not differ on the majority of sociodemographic or medical variables (ps > .05). However, as noted earlier, survivors randomized to EUC were more likely to be employed and have private insurance than survivors randomized to the SSP.

Correlations between baseline measures of RNT and related constructs, insomnia, and sleep quality are presented in Table 1 (Aim 1). Greater FCR and cancer-related rumination were associated with greater insomnia severity and poorer subjective sleep quality. Effects were moderate to large. In contrast, the general tendency to worry and trait intolerance of uncertainty were not significantly associated with insomnia severity or sleep quality. As expected, symptoms of anxiety and depression were also significantly positively correlated with both sleep metrics (Table 1). After controlling for anxiety and depression symptom severity, FCR continued to predict insomnia (β=.35, t=2.12, p=.04), but not sleep quality (β=.15, t=.92, p=.36), whereas cancer-related rumination was no longer a significant predictor of insomnia (β=.24, t=1.31, p=.20) or sleep quality (β=.07, t=.40, p=.70).

Table 1:

Correlations between repetitive negative thinking, anxiety and depression, and sleep at baseline (N = 40)

Variable	ISI	PSQI Sleep Quality
FCRI	.49***	.37*
RIQ-R	.39*	.31*
PSWQ	.23	.14
IUS-12	.08	.15
HADS-A	.34*	.37*
HADS-D	.45**	.50***

Note:

^*p ≤ .05

^***p ≤ .001.

FCRI = Fear of Cancer Recurrence Inventory severity subscale; RIQ-R = modified rumination subscale of the Responses to Intrusions Questionnaire; PSWQ = Penn State Worry Questionnaire; IUS-12 = 12-item Intolerance of Uncertainty Questionnaire; HADS = Hospital Anxiety and Depression Scale

Changes in RNT in response to treatment are shown in Table 2. At baseline, CBT-I (n=20) and EUC (n=20) conditions did not differ significantly on any of the RNT-related measures (p’s >.05). Relative to participants in the EUC condition, participants in the CBT-I condition demonstrated a greater decrease in FCR from baseline to T1, but not from baseline to T2, and the effect remained significant after controlling for employment status and insurance type. CBT-I did not significantly change cancer-related rumination, the general tendency to worry, or intolerance of uncertainty.

Table 2:

Change in repetitive negative thinking and related constructs in participants receiving cognitive behavioral therapy for insomnia versus enhanced usual care

Fear of Cancer Recurrence – Severity
	B	SE	df	t	p
Intercept	19.65	1.52	55.60	12.95	< .001
t0 – t1	−.17	1.18	73.60	−.14	.89
t0 – t2	−1.76	1.18	73.60	−1.49	.14
CBT-I (t0)	2.60	2.15	55.60	1.21	.23
t0 – t1 * Condition	−3.63	1.63	72.83	−2.23	.03
t0 – t2 * Condition	−2.18	1.64	72.93	−1.33	.19
Cancer-Related Rumination
	B	SE	df	t	p
Intercept	6.00	.96	59.85	6.22	< .001
t0 – t1	−.51	.84	73.60	−.61	.55
t0 – t2	−1.08	.82	73.33	−1.31	.19
CBT-I (t0)	1.45	1.36	59.85	1.06	.29
t0 – t1 * Condition	−1.12	1.14	72.60	−.98	.33
t0 – t2 * Condition	−1.07	1.14	72.57	−.94	.35
Worry
	B	SE	df	t	p
Intercept	49.30	2.84	48.74	17.33	< .001
t0 – t1	−.25	1.79	71.66	−.14	.89
t0 – t2	−.30	1.75	71.54	−.17	.87
CBT-I (t0)	3.50	4.02	48.74	.87	.39
t0 – t1 * Condition	−.47	2.44	71.11	−.19	.85
t0 – t2 * Condition	−2.71	2.44	71.10	−1.11	.27
Intolerance of Uncertainty
	B	SE	df	t	p
Intercept	30.10	2.02	57.85	14.89	< .001
t0 – t1	−.76	1.68	73.46	−.46	.65
t0 – t2	−2.20	1.65	73.22	−1.33	.19
CBT-I (t0)	−2.35	2.86	57.85	−.82	.41
t0 – t1 * Condition	.09	2.30	72.54	.04	.97
t0 – t2 * Condition	−.47	2.29	72.51	−.20	.84

Note. t₀ = baseline, t₁ = post-treatment, and t₂ = 1-month follow-up; fear of cancer recurrence= FCRI; cancer-related rumination= RIQ-R; worry= PSWQ; intolerance of uncertainty= IUS-12

Discussion:

Results of the current investigation indicated that aspects of RNT were associated with patient-reported insomnia and sleep quality and suggest that RNT may be an important target to consider within insomnia interventions for this population. Our findings are consistent with previous research demonstrating that worry and rumination are common experiences of cancer survivors that may negatively impact sleep.^{23,24,34–38}

Aim 1 results revealed significant associations between pre-intervention measures of RNT and insomnia and poor sleep quality. However, effects were only significant for cancer-specific measures (i.e., the FCRI and RIQ-R), as opposed to general measures (i.e., PSWQ, IUS) of RNT and related constructs. FCR was a robust predictor of insomnia even after controlling for anxiety and depression. The attenuated association between cancer-related rumination and insomnia in regression analyses suggests that this aspect of RNT is related to comorbid anxiety or mood disturbance and may need to be addressed as a symptom of these conditions.

Aim 2 findings indicated that CBT-I, while shown to be efficacious for insomnia in our previous research,⁴⁶ did not improve RNT beyond short-term changes in FCR. In line with these results, previous research has found that although CBT-I demonstrates large effects for insomnia and sleep, it only produces small to medium effects on secondary outcomes such as fatigue, depression, anxiety, and QOL among cancer survivors.¹⁰ Thus, there may be additional opportunities to refine CBT-I to address RNT and enhance its effects for this population.

Clinical Implications.

The extent to which standard CBT-I reduces RNT remains unclear. Although RNT has been proposed as one important factor perpetuating insomnia,⁵⁶ it has not been clearly delineated as a mechanism by which CBT-I reduces insomnia symptoms. RNT tends to be addressed as part of cognitive restructuring in CBT-I rather than specifically targeted.⁵⁷ Meta-analyses of CBT-I in non-oncological populations show moderate-to-large effects of CBT-I on worry (Hedges’ g range: −.41 to −.71),^58,59 but only small and unreliable effects on rumination.⁵⁸ These findings, taken together with the results of the current study, suggest that additional, larger-scale investigations of the impact of CBT-I on RNT are warranted. Results also provide preliminary support for the hypothesis that tailoring CBT-I to address FCR and cancer-related rumination may improve its efficacy for cancer survivors. In addition to the worry time strategy that was used in the current protocol, potential adjuncts to CBT-I that may be examined as methods for reducing RNT include mindfulness, meta-cognitive, and acceptance-based strategies.^8,60–64 Future qualitative and optimization trials could compare the efficacy of these strategies to reduce RNT and insomnia among cancer survivors. Potentially, routine FCR screening could identify survivors who may benefit from insomnia treatment with CBT-I.⁶⁵ Additionally, future studies should evaluate the impact of sequencing (e.g., timing of techniques to address RNT) within the context of CBT-I implementation.

Limitations.

It is important to note that the current study was an exploratory secondary analysis of a pilot randomized controlled trial; thus, results should be interpreted with caution and replication in a larger sample is encouraged. The purpose of the present study was to identify areas for future research rather than to definitively report estimates of the strength and direction of the relationship between RNT and insomnia. Nevertheless, the strength of this work is that it lays a foundation for CBT-I intervention research which may ultimately improve the quality and impact of treatment for cancer survivors with insomnia.

Conclusions.

Preliminary evidence indicates that certain types of RNT, particularly FCR and cancer-related rumination, are associated with insomnia severity and poorer sleep quality, but do not appear to be strongly influenced by CBT-I in its standard format. From these results, we hypothesize that cancer survivors with insomnia who also present with high FCR and/or cancer-related rumination may benefit from augmented CBT-I that addresses these factors. Future research should evaluate the potential for cognitive behavioral techniques that address RNT to augment CBT-I for cancer survivors.

Data Availability Statement:

The data that support the findings of this study are available from the corresponding author upon reasonable request.