Dyadic Investigation of Posttraumatic Stress Symptoms and Daily Sleep Health in Patients with Cancer and their Caregivers

Abstract

Objective:

Cancer can be a traumatic experience affecting multidimensional aspects of sleep among patients and caregivers. This study examined the differential associations of cancer-related post-traumatic stress symptoms (PTSS) with various sleep markers in this population.

Methods:

Patients newly diagnosed with colorectal cancer (n=138, mean age=56.93 years, 31.88% female, 60.14% Hispanic, 6.53 months post-diagnosis) and their sleep-partner caregivers (n=138, mean age=55.32 years, 68.12% female, 57.97% Hispanic) completed questionnaires assessing the four PTSS clusters (intrusion, avoidance, alterations in arousal and reactivity, negative alterations in cognitions and mood). Participants also completed daily sleep diaries for 14 consecutive days, from which sleep onset latency (SOL), wake after sleep onset (WASO), and sleep duration were derived.

Results:

Actor-partner interdependence model revealed that caregivers’ greater alterations in arousal and reactivity were associated with their own longer SOL (b=14.54, p<.001) and their patients’ longer sleep duration (b=0.47, p=.040), whereas patients’ arousal and reactivity were associated with their caregivers’ shorter SOL (b=−8.34, p=.047) and WASO (b=−8.12, p=.019). Patients’ and caregivers’ greater negative alterations in cognitions and mood were associated with patients’ longer SOL (b=8.89, p=.016) and shorter sleep duration (b=−0.40, p=.038), respectively. Caregivers’ greater intrusion was related to their own shorter SOL (b=−10.92, p=.002).

Conclusions:

The four PTSS clusters, particularly arousal and reactivity and negative cognitions and mood, have distinct associations with sleep markers individually and dyadically in patients and caregivers affected by cancer. Investigations of psychosocial and biobehavioral pathways underlying these relations are warranted. Tailored trauma treatments and sleep interventions may improve the well-being of this population.

Introduction

The life-threatening nature of cancer and the repeated exposures to cancer-related threats can be highly stressful and traumatic for patients and their family caregivers (1-3). Not surprisingly, patients with cancer and their family caregivers have reported elevated levels of symptoms of post-traumatic stress disorder (PTSD; 4). According to the DSM-5 (5), PTSD is characterized by four conceptually distinct symptom clusters: intrusive trauma-related thoughts, avoidance of trauma-related reminders, alterations in arousal and reactivity, and negative alterations in cognitions and mood. Although the current PTSD diagnostic criteria exclude chronic major medical illness as a traumatic experience (5), recent work has illustrated that the four symptom clusters extend to adult patients with cancer and their family caregivers (6). Indeed, elevated levels of post-traumatic stress symptoms (PTSS) have been reported by 20%-80% of patients (7) and 20%-35% of caregivers affected by cancer (8), compared to 3.9%-6.1% in the general adult population (9, 10).

Another prevalent concern among adult patients with cancer and their caregivers is sleep disturbance, which includes prolonged time to initiate sleep (i.e., prolonged sleep onset latency), extended wake time during the night after sleep onset (prolonged wake after sleep onset), and inadequate sleep duration. Systemic reviews have indicated that 58%-63% of patients with cancer (11) and over 72% of caregivers of patients with cancer (12) experience overall sleep disturbance, which is about two times higher than what has been observed in the general population (29.8; 13). Studies have also found longer sleep onset latency (14, 15), longer wake after sleep onset (16, 17), and shorter sleep duration (14-17) among patients with cancer and their caregivers, relative to the general population (18, 19).

Substantial evidence suggests a strong association between overall PTSD symptomology and overall sleep disturbance (20). However, despite the established multidimensionality of PTSD/S (hereafter PTSS) and sleep, less is known regarding the differential associations of each PTSS cluster with distinct sleep markers. Such an investigation is critical because, despite the high correlations among PTSS clusters, studies have found differential associations between each PTSS cluster with various sleep-related health markers, including substance use (21), pain (22), as well as numerous domains of psychosocial (23) and physical functioning (e.g., 24). Additionally, distinct sleep markers in cancer populations have been differentially related to unique downstream health outcomes, such as fatigue (25), neuroendocrine dysregulation (e.g., flatter diurnal cortisol slope; 26), and even mortality (14). Thus, not only is each PTSS cluster plausibly associated with distinct sleep markers, but understanding such unique PTSS-sleep relations have the potential to tailor treatment approaches for and improve the health of trauma-exposed individuals, many of whom have different combinations of PTSS symptoms (27).

Differential associations of PTSS clusters and sleep markers

Alterations in arousal and reactivity (hereafter, arousal and reactivity) is the most studied symptom cluster and has been associated with specific sleep markers. For example, in studies of individuals exposed to sexual trauma (28) and combat-related trauma (29, 30), greater arousal and reactivity was related to more fragmented sleep (i.e., longer wake after sleep onset) and shorter sleep duration. Among patients with irritable bowel syndrome, greater arousal and reactivity was associated with shorter sleep duration and lower sleep efficiency, an overall marker of sleep consolidation (31). These findings support hyperarousal-based theories which posit that excessive physiologic activity prior to and during sleep underlie sleep disturbances (32). Thus, greater arousal and reactivity appear to be most specifically related to longer wake after sleep onset and shorter sleep duration.

Intrusive trauma-related thoughts, another component of PTSS, have been associated with longer sleep onset latency and longer wake after sleep onset among persons clinically diagnosed with PTSD (33). One study of patients with breast cancer also found that greater intrusive thoughts were related to shorter sleep duration and lower sleep efficiency (34). Findings support the proposition that unwanted reminders/thoughts of trauma are results of insufficiently processed trauma-related information (35), leading to intrusion and excessive arousal that prolong sleep onset and disrupt sleep maintenance (36-38). Existing studies therefore suggest that greater intrusive trauma-related reminders are related to prolonged sleep onset latency and wake after sleep onset, as well as shorter sleep duration.

Fewer studies have investigated the unique associations of the avoidance cluster with specific sleep markers. Conceptually, avoidance of trauma-related cues would be counter-productive because it diminishes opportunities for individuals to make sense of trauma experiences (39, 40). One study of community elderly adults found that greater avoidance of traumatic experiences was correlated with longer sleep onset latency and shorter sleep duration (41). Greater avoidant coping has also been associated with greater overall symptoms of sleep disturbance in female Veterans (42) and patients with breast cancer (43). Notably, however, one study found non-significant relations between avoidance and sleep disturbance in patients with newly diagnosed breast cancer (44). Despite the limited and mixed findings, the current literature suggests that avoidance of trauma-related stimuli appears to be related to longer sleep onset latency and shorter sleep duration.

Negative alterations in cognitions and mood (hereafter, negative cognitions and mood) make up the newest component of the PTSD diagnostic criteria (5) and thus has been least studied in relation to distinct sleep indices. Negative cognitions and mood are characterized by persistent negative emotional state and distorted cognitions about the cause and consequence of the trauma, which lead to blame of self and others, as well as reckless or self-destructive behavior (45). Although findings are limited, greater symptoms of negative cognitive and emotional states in patients with cancer have been correlated with longer sleep onset latency and longer sleep duration (46). Similar findings have also been found among non-cancer populations, which showed that greater cognitive or emotional distress was associated with prolonged sleep onset latency and nighttime awakening (47-49) and decreased sleep efficiency (48, 50). Thus, the limited findings suggest that negative cognitions and mood are related to prolonged sleep onset latency, prolonged wake after sleep onset, and longer sleep duration.

Cancer experience as a dyadic phenomenon

Increasing evidence suggests that cancer is a dyadic phenomenon that equally affects both patients and caregivers. For example, psychological distress has been repeatedly shown to be intercorrelated between patients with cancer and their family caregivers (1, 51-53). In addition, approximately 65% of the general adult population—including patients and caregivers affected by cancer—share a bed with a significant other (54), suggesting shared sleep experiences between individuals with cancer and their partner caregivers whom they share a bed with (hereafter, sleep-partner caregivers). In fact, 25% to 35% of adult cancer patients and their sleep-partner caregivers have reported sleep disturbances because of their bed-partner’s disturbed sleep (16). Furthermore, psychological distress in patients with cancer has been associated with not only their own but also their sleep-partner caregivers’ shorter sleep duration, and vice versa (55, 56). Although existing studies indicate inter-individual relations between cancer-related distress and sleep, specific dyadic associations between cancer-related PTSS and sleep markers remain understudied.

The present study

The goal of the current study was to investigate the distinct contributions of each PTSS cluster with various sleep markers in patients with cancer and their sleep-partner caregivers. We hypothesized that greater arousal and reactivity would be specifically associated with longer wake after sleep onset and shorter sleep duration; greater intrusion would be associated with longer sleep onset latency, longer wake after sleep onset, and shorter sleep duration; greater avoidance would be associated with longer sleep onset latency and shorter sleep duration; and greater negative cognitions and mood would be associated with longer sleep onset latency, longer wake after sleep onset, and longer sleep duration. Due to limited evidence, we explored the dyadic associations between PTSS clusters and sleep markers (i.e., the degree to which one’s PTSS relates to their partner’s sleep).

Methods

The reported methods and results followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement (57). This investigation was a secondary analysis of a parent study (R01NR016838: https://doi.org/10.17605/OSF.IO/P5MHU). The current study and its analytic plan were pre-registered (https://doi.org/10.17605/OSF.IO/JU9DK). De-identified data, analysis code, and research materials are not available in a public archive but will be provided upon request.

Participants

Participants were 158 patients diagnosed with colon or rectal cancer and their caregivers who were enrolled to the study between June 2017 and February 2023. Eligibility criteria for patients were (a) a new diagnosis of colon or rectal cancer (stages I-IV) within 10 months of study enrollment; (b) 18 years or older; (c) able to speak and read English or Spanish at the fifth-grade level; and (d) have a consistent caregiver who is involved in their daily activities. Eligibility criteria for caregivers were (a) 18 years or older; and (b) able to speak and read English or Spanish at the fifth-grade level. Exclusion criteria for both patients and caregivers were (a) active yet untreated psychosis, dementia, substance dependence, phobia pertaining to medical treatment procedure, and suicidal ideation in the past year; (b) active yet untreated sleep apnea, narcolepsy, and restless leg syndrome (sleep apnea: 3 on the Berlin Questionnaire without the use of positive airway pressure machines during sleep; narcolepsy: ≥ 14 on the Ullanlinna Narcolepsy Scale; 58, 59), (c) inability to see or hear; (d) poor cognitive functioning status (≤ 24 on the Mini Mental State Examination; 60); or (e) patients who are under end-of-life care or endorsed poor physical functioning status (≤ 50 on the Karnofsky Performance Status and ≥ 3 on the Eastern Cooperative Oncology Group Performance Status; 61, 62).

Procedure

The parent study was conducted in compliance with the University of Miami Institutional Review Board. Patients were identified by research associates using medical records of participating clinics and screened to determine their eligibility to participate in the study. Eligible patients were asked to nominate a caregiver, who were then screened for eligibility. Eligible participants signed a consent form before providing any study data. Participants individually completed questionnaires, including measures on demographics and the four clusters of PTSS. Participants also completed daily sleep logs for 14 consecutive days. Each participant received a $100 incentive for providing all study data. Of 158 dyads enrolled, 9 dyads did not provide any data, leaving 149 patient-caregiver dyads with any study data. Of the 149, 11 dyads reported not sharing a bed for sleep and were thus excluded, leaving 138 dyads for this secondary analysis.

Measures

Post-traumatic Stress Symptoms (PTSS).

Arousal and reactivity, intrusion, and avoidance clusters of PTSS were measured using the 22-item Impact of Event Scale – Revised (IES-R; 63). Patients responded to the items with respect to their own cancer, and caregivers responded with respect to their patients’ cancer. Participants evaluated each PTSS cluster experienced on a 5-point Likert-format (0 = not at all, 4 = extremely). One item in the arousal and reactivity cluster and one item in the intrusion cluster were removed because they were directly related to sleep (Arousal and reactivity: “I had trouble falling asleep”; Intrusion: “I had trouble staying asleep”). Responses within each PTSS cluster were averaged, with higher scores indicating greater corresponding PTSS. The three cluster scores had good internal consistency for both samples of patients and caregivers (α ≥ .76, .90, and .86, for arousal and reactivity, intrusion, and avoidance, respectively).

The fourth PTSS cluster, negative cognitions and mood, was measured using seven items that correspond to the six negative cognitions and mood items from the PTSD checklist for DSM-5 (PCL-5; 64). The selected items were obtained from the Center for Epidemiological Studies-Depression (65), the Brief COPE (66), the Profile of Mood States-Short Form (67), and the Functional Assessment of Chronic Illness – Spiritual Well Being Scale (68) to represent the underlying constructs measured by the PCL-5 items (for more information, see 6). Confirmatory factor analysis using the seven negative cognitions and mood items and the three PTSS domains from the IES-R has shown good fit for patients with cancer and their family caregivers (6). In the current sample, the selected negative cognitions and mood items had acceptable internal consistency in (α = .74 and .71 for patients and caregivers, respectively).

Sleep.

Participants self-reported their sleep each morning, immediately after waking, using the Consensus Sleep Diary (69). From these data, sleep onset latency (SOL: minutes between intending to sleep and sleep onset), wake after sleep onset (WASO: minutes awake between sleep onset and final awakening), sleep duration (hours spent asleep: time between intending to sleep and final awakening – SOL – WASO), and sleep efficiency [percentage of time asleep while in bed: (sleep duration / total time in bed)*100] were calculated. An averaged value for each sleep marker across the 14 days was determined for each person. Longer than 15 minutes of SOL (19), longer than 20 minutes of WASO (19), less than 7 hours or greater than 8 hours of sleep duration (18), and less than 85% of sleep efficiency (19) are considered suboptimal.

Covariates.

Self-reported age, gender, diagnosis of psychiatric disorder (70), general distress (Distress Thermometer: 71), and patient’s cancer stage were included as covariates. Diagnosis of at least one psychiatric disorder (diagnosis versus no diagnosis), gender (male versus female), and patient’s cancer stage [non-advanced (stages I and II) versus advanced (stages III and IV)] were dichotomized.

Statistical Analyses

Data were examined for outliers (mean ± 3 standard deviation) and normality (skewness < ±3; kurtosis < ±8; 72). Means, standard deviations, and frequencies of study variables were computed for patients and caregivers separately. Differences between patients and caregivers were tested using paired t-tests for continuous variables and either the McNemar’s test for 2-level categorical variables or the test of Marginal Homogeneity for more than 2-level categorical variables (Table 1). Zero-order correlations among study variables were tested using Pearson’s correlation coefficients for continuous and normally distributed variables, and Spearman’s correlation coefficients for dichotomous and non-normally distributed variables (see Table S1, Supplemental Digital Content). All descriptive data analysis was conducted using SPSS Version 28.0 (73).

The primary research questions were tested using actor-partner interdependence modeling within structural equation framework (Mplus Version 8.6; 74). The eight PTSS clusters (four from patients and four from caregivers) and covariates were the exogenous variables, and sleep markers (SOL, WASO, and sleep duration) of patients and caregivers were the endogenous variables. The three sleep markers of patients and caregivers were regressed on one’s own (actor effects) and their partner’s PTSS scores (partner effects) simultaneously, controlling for covariates (i.e., one’s own age, gender, psychiatric diagnosis, general distress, and patient’s cancer stage; Table 2 and Figure 1). Sleep efficiency was separately modeled because it was derived from SOL, WASO, and sleep duration, and represented a global measure of sleep consolidation (See Table S2, Supplemental Digital Content, for the results of the sleep efficiency model). For both the primary model (with SOL, WASO, and sleep duration) and the supplemental model (with sleep efficiency only), actor and partner effects were constrained to be equal across patients and caregivers, and chi-square difference test was conducted to examine the difference in fit between the unconstrained and constrained models (75). Four model fit indices and their cut-off criterion for acceptable fit were: the chi-square test of model fit (p > .05), the comparative fit index (CFI > .95), the root mean squared error of approximation (RMSEA < .06), and the standardized root mean squared residual (SRMR < .08; 76). Missing data were handled using Full Information Maximum Likelihood. Statistical significance was determined by a 2-tailed p-value < .05.

Results

As shown in Table 1, participants were primarily middle-aged, Hispanic, well-educated, and had middle- to high-income. Patients were diagnosed approximately 7 months prior to participating in the study and primarily with advanced stage cancer. The majority (98.5%) of patients had heterosexual partners. About one-tenth of participants reported a diagnosis of psychiatric disorder. Participants reported low levels of general distress and PTSS, as well as normative WASO and sleep duration but suboptimal SOL and sleep efficiency (83.21% and 83.66% for patients and caregivers, respectively). Compared to patients, caregivers were younger, more likely to be female gender, reported greater general distress, intrusion, and arousal and reactivity, and had shorter sleep duration.

Table 1.

Sample Characteristics and Study Variable Descriptives.

	Patients			Caregivers				t or χ2
Demographics	N	M (SD) or frequency (%)		N	M (SD) or frequency (%)
Age (years)	138	56.93	(10.81)	138	55.32	(11.17)		2.84**
Gender (female)	138	44	(31.88%)	138	94	(68.12%)		18.12***
Ethnicity	138			138				−0.74
Hispanic		83	(60.14%)		80	(57.97%)
Non-Hispanic White		36	(26.09%)		36	(26.09%)
Non-Hispanic Black		8	(5.80%)		10	(7.25%)
Other		11	(7.97%)		12	(8.70%)
Education	137			137				−0.37
High school degree or less		28	(20.44%)		22	(16.06%)
Vocational or some college		25	(18.25%)		33	(24.09%)
College degree or more		84	(61.31%)		82	(59.85%)
Household income	135			130				1.77
< $40,000		41	(30.37%)		37	(28.46%)
$40,000 - $89,999		30	(22.22%)		33	(25.38%)
$90,000 - $159,999		32	(23.70%)		27	(20.77%)
≥ $160,000		21	(15.56%)		25	(19.23%)
Prefer not to answer		11	(8.15%)		8	(6.15%)
Medical variables and distress
Time since diagnosis (month)	138	6.53	(3.78)			−
Cancer Stage	138
Non-advanced (Stages I-II)		42	(30.43%)			−
Advanced (Stages III-IV)		96	(69.57%)			−
Diagnosis of psychiatric disorder	135	17	(12.59%)	135	15	(11.11%)		−0.47
							Scale range
General distress	133	3.11	(2.85)	131	4.11	(3.10)	0-10	−2.73**
Post-traumatic Stress Symptoms
Arousal and Reactivity	136	0.65	(0.69)	136	0.93	(0.79)	0-4	−3.70***
Intrusion	136	0.93	(0.87)	136	1.22	(0.92)	0-4	−3.00**
Avoidance	136	1.12	(0.88)	136	1.14	(0.91)	0-4	−0.34
Negative Cognitions and Mood	137	0.85	(0.64)	136	0.85	(0.61)	0-4	0.19
Sleep markers
Sleep Onset Latency (min)	133	24.47	(21.34)	133	25.23	(21.44)		0.13
Wake After Sleep Onset (min)	133	18.15	(23.50)	133	14.78	(18.01)		1.29
Sleep Duration	133	7.49	(1.13)	133	7.24	(0.94)		2.31*

^*p < .05

^**p < .01

^***p < .001.

Note. Differences between patients and caregivers were tested using paired t-tests for continuous variables and either the McNemar’s test for 2-level categorical variables or the test of Marginal Homogeneity for more than 2-level categorical variables.

Individual and dyadic associations of PTSS clusters with sleep markers.

The model testing the individual and dyadic associations of the four PTSS clusters with SOL, WASO, and sleep duration (Figure 1) demonstrated good fit to the data: χ²(24) = 20.45, p = .671; CFI = 1.00; RMSEA < .001, 90% CI [0.00, 0.06]; SRMR = .022. The model with equal constraints across patients and caregivers fit the data acceptably [χ²(48) = 57.23, p = .170; CFI = 0.93; RMSEA = .040, 90% CI [0.00, 0.08]; SRMR = .051], but the constrained model was significantly worse than the unconstrained model [χ²_diff (24) = 36.78, p = .046], so the unconstrained model was retained.

As shown in Table 2 and Figure 1, caregivers’ greater arousal and reactivity was associated with their own longer SOL, which was not hypothesized due to lack of existing evidence. Caregivers’ greater intrusion was also associated with their own shorter SOL, which was contrary to our hypothesis. Patients’ greater negative cognitions and mood was associated with their own longer SOL, supporting our hypothesis. Additionally, patients’ greater arousal and reactivity was associated with their caregivers’ shorter SOL and shorter WASO, whereas caregivers’ greater arousal and reactivity was associated with their patients’ longer sleep duration. Finally, caregivers’ greater negative cognitions and mood was associated with their patients’ shorter sleep duration. Among the covariates, caregivers’ older age was associated with their own longer WASO, and patients’ female gender, advanced cancer stage, and lower general distress were associated with their own longer sleep duration.

Table 2.

Parameter estimates on the associations of PTSS clusters with sleep markers.

PTSS clusters	Patient SOL		Caregiver SOL		Patient WASO		Caregiver WASO		Patient Sleep Duration		Caregiver Sleep Duration
	b (SE)	95% CI	b (SE)	95% CI	b (SE)	95% CI	b (SE)	95% CI	b (SE)	95% CI	b (SE)	95% CI
Arousal_O	−1.13 (4.14)	[−9.25, 6.98]	14.54 (3.72)	[7.25, 21.82]	−3.14 (5.37)	[−13.67, 7.39]	2.76 (3.06)	[−3.24, 8.76]	−0.06 (0.25)	[−0.55, 0.44]	−0.08 (0.20)	[−0.47, 0.30]
Intrusion_O	4.71 (3.58)	[−2.32, 11.73]	−10.92 (3.45)	[−17.68, −4.15]	−5.69 (−1.22)	[−14.82, 3.45]	1.92 (2.84)	[−3.66, 7.49]	0.26 (0.22)	[−0.17, 0.68]	0.04 (0.18)	[−0.32, 0.40]
Avoidance_O	−3.07 (2.43)	[−7.84, 1.69]	−1.20 (2.75)	[−6.59, 4.20]	0.44 (3.16)	[−5.76, 6.64]	−2.19 (2.27)	[−6.63, 2.26]	−0.13 (0.15)	[−0.42, 0.16]	0.04 (0.15)	[−0.25, 0.33]
Cognitions_O	8.89 (3.68)	[1.68, 16.09]	0.38 (3.63)	[−6.74, 7.49]	8.34 (4.77)	[−1.00, 17.69]	3.15 (2.99)	[−2.70, 9.01]	0.26 (0.22)	[−0.17, 0.70]	−0.03 (0.19)	[−0.41, 0.35]
Arousal_P	3.51 (3.77)	[−3.87, 10.89]	−8.34 (4.21)	[−16.58, −0.10]	−1.39 (4.88)	[−10.94, 8.17]	−8.12 (3.46)	[−14.90, −1.33]	0.47 (0.23)	[0.02, 0.91]	−0.02 (0.22)	[−0.45, 0.42]
Intrusion_P	−0.09 (3.32)	[−6.60, 6.42]	2.19 (3.67)	[−5.00, 9.37]	3.64 (4.32)	[−4.84, 12.11]	1.16 (3.02)	[−4.75, 7.08]	−0.38 (0.20)	[−0.78, 0.02]	0.07 (0.20)	[−0.31, 0.45]
Avoidance_P	−1.06 (2.65)	[−6.26, 4.15]	0.66 (2.51)	[−4.26, 5.58]	−0.20 (3.45)	[−6.97, 6.57]	0.81 (2.07)	[−3.24, 4.86]	0.12 (0.16)	[−0.20, 0.44]	0.06 (0.13)	[−0.21, 0.32]
Cognitions_P	3.00 (3.32)	[−3.28, 9.29]	4.07 (3.62)	[−3.03, 11.16]	2.62 (4.17)	[−5.55, 10.80]	2.98 (2.98)	[−2.86, 8.82]	−0.40 (0.19)	[−0.79, −0.02]	0.15 (0.19)	[−0.23, 0.53]
Covariates
Age	0.13 (0.16)	[−0.19, 0.45]	0.23 (0.16)	[−0.08, 0.55]	0.20 (0.21)	[−0.20, 0.61]	0.33 (0.13)	[0.07, 0.59]	−0.01 (0.01)	[−0.03, 0.01]	−0.01 (0.01)	[−0.03, 0.00]
Gender	0.89 (3.54)	[−6.04, 7.83]	3.42 (3.56)	[−3.55, 10.39]	2.39 (4.56)	[−6.54, 11.31]	−5.13 (2.93)	[−10.87, 0.62]	0.45 (0.21)	[0.04, 0.87]	0.02 (0.19)	[−0.35, 0.39]
Cancer Stage	−2.07 (3.74)	[−9.39, 5.26]	1.76 (3.79)	[−5.38, 9.19]	−3.38 (4.86)	[−12.92, 6.15]	3.02 (3.12)	[−3.01, 9.14]	0.51 (0.23)	[0.06, 0.95]	0.08 (0.20)	[−0.32, 0.47]
Psychiatric	−8.93 (4.95)	[−18.64, 0.77]	0.39 (5.38)	[−10.15, 10.92]	−3.82 (6.11)	[−15.80, 8.16]	−5.61 (4.44)	[−14.32, 3.09]	0.22 (0.28)	[−0.33, 0.78]	0.13 (0.30)	[−0.45, 0.71]
Distress	1.42 (0.92)	[−0.38, 3.21]	−0.01 (0.65)	[−1.27, 1.26]	1.24 (1.14)	[−0.99, 3.46]	0.89 (0.53)	[−0.15, 1.93]	−0.15 (0.05)	[−0.25, −0.04]	−0.04 (0.04)	[−0.11, 0.03]

Note. Bolded values indicate p < .05. O = Own; Arousal = Arousal and reactivity; Cognitions = Negative cognitions and mood; P = Partner; Gender: Female = 1, Male = 0; Psyc: psychiatric morbidity = 1, no psychiatric morbidity = 0; Cancer Stage: Advanced stage cancer = 1, Early stage cancer = 0; Distress = General psychological distress. Parameter estimates and 95% confidence intervals were obtained using actor-partner interdependence modeling.

Figure 1. Path diagram of the associations of the four PTSS clusters with sleep markers in patient-caregiver dyads.

Note. PT: Patient; CG: Caregiver; SOL: Sleep onset latency; WASO: Wake after sleep onset. Black lines indicate statistically significant paths from the patient’s PTSS clusters. Grey lines indicate statistically significant paths from the caregiver’s PTSS clusters. Associations regarding covariates not included.

Discussion

This study examined the differential associations of PTSS clusters with various sleep markers among adult patients with cancer and their sleep-partner caregivers. Compared to patients, caregivers reported greater levels of cancer-related arousal and reactivity and intrusive thoughts, as well as shorter sleep duration. Both patients and caregivers reported suboptimal sleep onset latency. Findings highlighted the deleterious role of arousal and reactivity in sleep health within individuals but suggested their sleep-consolidating functions for the partners’ sleep. Results also pinpointed the adverse roles of patients’ and caregivers’ cancer-related negative cognitions and mood in patients’ sleep.

Similarities and differences in PTSS and sleep between patients and caregivers

The current study indicated comparable levels of avoidance and negative cognitions and mood between patients and caregivers but greater levels of arousal and reactivity and intrusion in caregivers than in patients. These results corroborate the cumulating evidence that caregivers experience similar (77-80), if not greater (81-83) levels of cancer-related distress, compared with patients. According to the cognitive model of PTSD (35), elevated PTSS in caregivers may be due to the insufficiently-processed recent trauma (i.e., cancer diagnosis) that leads to more arousal and unwanted reminders. Caregivers’ roles may also make them uniquely susceptible to experience of hypervigilance and intrusive thoughts because they are involved in providing ongoing attentive support to their patients, which may perpetuate beliefs that their loved one is in danger.

Regarding sleep markers, patients and caregivers reported similar levels of sleep onset latency and wake after sleep onset, but caregivers endorsed shorter sleep duration than patients. This is consistent with evidence from a systematic review of caregivers of patients with cancer, which indicated that short total sleep time is one of the primary sleep disturbances in the cancer caregiver population (12). Because of the demands of providing ongoing attentive support to patients, caregivers may not allocate sufficient time to rest (12). Given the morbidities associated with suboptimal sleep duration (84), the current evidence further highlights the health risks of shorter sleep duration among caregivers of patients with cancer.

PTSS clusters distinctly associated with various sleep markers

Of the four PTSS symptom domains, arousal and reactivity was most correlated with sleep markers, for which the relations differed between individual/within-person and dyadic/between-partner effects. At the individual level, caregivers’ greater arousal and reactivity was associated with their own longer sleep onset latency. Although this association was not hypothesized due to limited empirical evidence, it is congruent to hyperarousal-based theories of PTSD that proposed that excessive arousal associates with sleep disturbances (32), possibly by way of elevated nighttime cortisol levels (85) and imbalance in sympathovagal functioning (86).

On the other hand, at the dyadic level, greater arousal and reactivity was unexpectedly associated with better sleep outcomes. For example, patients’ greater arousal and reactivity was related to their caregivers’ shorter sleep onset latency and wake after sleep onset. Patients’ arousal and reactivity was also significantly lower than that of their caregivers’. Perhaps those caregivers whose patients had levels of arousal and reactivity that were elevated and similar to their own benefited from dyadic similarity and experienced a sense of solidarity that promoted their sleep. Indeed, previous studies have found that higher dyadic similarity in stress appraisal was associated with lower psychological distress among patients with cancer and their caregivers (87, 88). Alternatively, it may be that highly aroused patients were more likely to seek support from their caregivers as a way to reduce the challenges of cancer. This support-seeking behavior may manifest in intimacy before or during sleep and promote caregivers’ perceived closeness and sleep consolidation. Given the dyadic nature of intimate and support-seeking behaviors, the same conjecture could also apply to the association between caregivers’ greater arousal and reactivity and patients’ longer sleep duration. Specifically, inter-partner support-seeking behaviors may promote experiences of closeness and safety that lengthen patients’ sleep duration, which has been shown to be lower than the recommended amount in patients with cancer (14, 56). Supporting this perspective, studies have shown that intimate behaviors such as interpersonal touch (89, 90) and handholding (91, 92) are related to improved sleep quality and shorter nighttime awakening. Given the limited empirical evidence on dyadic associations between PTSS and sleep, future studies are warranted to empirically test each of these speculations.

Furthermore, negative cognitions and mood of both patients and caregivers were associated with patients’ poor sleep. Specifically, patients’ greater negative cognitions and mood were associated with their own longer sleep onset latency, which is consistent with findings from existing studies that used conceptually similar measures (46). We speculate that a recent cancer diagnosis may elevate patients’ negative feelings and thoughts (e.g., about the future), which may increase rumination while in bed and delay sleep onset. The current study also found that caregivers’ elevated negative cognitions and mood were related to their patients’ shorter sleep duration. This latter finding may explain existing evidence that patients with cancer tend to sleep less than the minimum recommended amount (14, 56). For example, perhaps caregivers’ direct verbal cues of distress and/or behavioral display of dysphoria elevate patients’ guilt about how their illness has affected the family, which then decreased patients’ sleep duration. More refined investigations involving the roles of verbal and non-verbal exchanges (e.g., lexical and facial expressions) between patients and their caregivers are warranted to apprehend the underlying mechanisms that link caregivers’ cancer-related cognitive-emotional distress to patients’ sleep.

The associations of the other two PTSS clusters with sleep health were limited in the current study. Caregivers’ greater intrusion was related to their own shorter sleep onset latency, which was unexpected. Caregivers’ burden of handling the day-to-day care of their patients may not only expose them to intrusive reminders but also keep them busy and exhausted, which may increase sleep drive and precipitate shorter sleep onset by the end of the day. Although existing studies have shown that role overload in caregivers of patients with dementia was associated with greater overall sleep disturbance (93, 94), future studies are needed to investigate the unique association of role overload with sleep onset latency in individuals who provide care to patients with cancer. Additionally, avoidance was not related to any sleep markers of both patients and caregivers. Although these non-significant findings were also unexpected, they are consistent with the null findings in a study by Dupont and colleagues (44), which is the only study to date that examined the specific role of avoidance in sleep among patients who were newly diagnosed with cancer. Perhaps avoiding reminders of a recent cancer diagnosis helps to reduce arousal and distress in the short-term and thus may not directly affect sleep. Future longitudinal studies on the role of avoidance in sleep are warranted to support this speculation.

Clinical implications

The current study findings suggest critical implications for tailored treatments. For example, highly aroused individuals in response to cancer in the family may benefit from interventions that specifically target their fear memory network (e.g., prolonged exposure therapy; 95). In addition, cognitive-behavioral interventions (e.g., cognitive-behavioral therapy for insomnia), which is a gold standard treatment for sleep, and cognitive processing therapy (96, 97), may be particularly efficacious to manage cancer-related negative affect and cognitive distortions to improve sleep health for both patients with cancer and their caregivers.

Moreover, our findings highlight the various psychosocial impacts of cancer in the family (1-3) and suggest the critical need for augmenting our understanding of the efficacy of existing trauma and sleep interventions that simultaneously treat patients with cancer and their caregivers. Dyadic interventions in these areas have been targeted for Veterans and their partners (e.g., 98, 99) but remain limited for patients and caregivers affected by medical illnesses, such as cancer (e.g., 100).

Limitations and future directions

Several study limitations should be considered. The cross-sectional nature of the current investigation precluded both short- and long-term inferences. Cancer is an illness characterized by repeated exposures to acute stressors that influences different psychophysiological responses across days to years. To better understand the unique nature of cancer, future studies should consider investigating the associations of PTSS with sleep markers across days (e.g., using intensive longitudinal data) and different phases of the survivorship continuum. Additionally, the investigation of the bidirectional associations between sleep and PTSS within couples affected by cancer is warranted, given that there have been related studies in non-cancer and trauma-exposed individuals (e.g., 101).

The current study was also a secondary analysis of a parent study that was conceptualized prior to the latest four-symptom model of PTSD, which led to the ad hoc assembly of measurements for the negative cognitions and mood symptom cluster. Despite the acceptable internal consistency of each of the four symptom clusters, future studies should corroborate the current findings using measures that most accurately represent the four PTSS domains (e.g., using the PCL-5: 64). Additionally, although other instruments such as actigraphy and polysomnography have been used to investigate sleep, only subjective markers of sleep were used in the current study. This is a salient consideration for future studies because existing findings have demonstrated discordance in self-reported and actigraphy-assessed sleep in medical populations, including persons affected by cancer (102).

Furthermore, generalizability of the study findings to patients of other types of cancer, to couples who do not share a bed for sleep, and to more ethnic and sociocultural diverse populations is limited and should be considered for the future. Finally, given the modest sample size, the current study did not examine the moderating effects of demographic variables such as Hispanic ethnicity and gender, which have been shown to play a role in PTSS symptomology (103, 104).

Conclusions

Notwithstanding these limitations, the current investigation offers a step forward in understanding the role of PTSS in sleep among individuals affected by cancer. Our findings primarily indicated the adverse role of arousal and reactivity in sleep health within individuals but suggested their sleep-consolidating functions between partners. Additionally, greater negative cognitions and mood of both patients and caregivers were related to poorer markers of sleep among patients only. Together, these results extend the current knowledge by elucidating the nuanced associations between cancer-related traumatic symptoms and sleep in not only individuals, but also between sleep-partners. Findings suggest the need for advancements in precise assessments and tailored treatments to reflect the multidimensional and interpersonal nature of PTSS and sleep disturbance in newly diagnosed patients with cancer and their caregivers.