Negative illness perceptions are related to more fatigue among haematological cancer survivors: a PROFILES study

Abstract

Objectives: The common sense model provides a theoretical framework for understanding substantial fatigue among (haematological) cancer survivors based on their illness perceptions. We therefore examined the associations between modifiable illness perceptions and substantial fatigue while controlling for sociodemographic, clinical, and psychological factors (symptoms of depression and anxiety) among haematological cancer survivors.

Methods: Data from the population-based PROFILES registry were used. Survivors diagnosed between 1999 and 2013 with Hodgkin lymphoma (N = 164), non-Hodgkin lymphoma (N = 655) and chronic lymphocytic leukaemia (N = 174) were included. Survivors completed the Brief Illness Perception Questionnaire (B-IPQ), the Fatigue Assessment Scale (FAS), and Hospital Anxiety and Depression Scale (HADS). Multivariable logistic regressions analyses were performed for the total group and three haematological cancers separately relating illness perceptions to substantial fatigue (>21 FAS).

Results: Haematological cancer survivors with illness perceptions that represent more negative consequences (consequences, OR = 1.27; 95%CI = 1.13–1.42); attribute more symptoms to their illness (identity, OR = 1.29; 95%CI = 1.17–1.43); and have a poorer illness understanding (coherence, 1.13; 1.04–1.22) were more often substantially fatigued. For the remaining five illness perceptions, no significant association was found. Non-Hodgkin lymphoma survivors who reported a poor illness understanding (coherence, OR = 1.35; 95% CI = 1.06–1.72) and chronic lymphocytic leukaemia survivors who reported that treatment can control (OR = 1.25; 95%CI = 1.01–1.55) the illness experienced more often substantial fatigue.

Conclusion: Those who experience more consequences of their disease, attribute more symptoms to their illness, and have a poorer illness understanding, have a higher risk to experience substantial levels of fatigue even years after diagnosis. Psychological interventions changing these illness perceptions may be beneficial in reducing fatigue among haematological cancer survivors.

Introduction

The life expectancy of haematological cancer patients has increased considerably over the past decades due to the use of new and more effective treatments [1]. However, with improved survival and prognosis, these patients face long-term physical and psychosocial adverse effects from cancer and its treatment [2,3]. One of the most frequently reported and most distressing long term effects is fatigue [4,5]. Overall, approximately up to 50% of haematological cancer survivors have to deal with persistent fatigue after treatment completion [4], although the use of different instruments and threshold criteria across studies results in a large variability in prevalence estimates.

Research regarding the aetiology of fatigue in cancer survivors has mainly focussed on the assessment of demographic and clinical determinants of fatigue. According to some studies, a younger age (≤ 65 years), being female, a lower educational level, absence of a partner, having comorbidities, and shorter time since diagnosis are related to more fatigue [4,6], although these findings are inconsistent [5,7,8]. Although extensively studied, almost unanimously no relations were found between cancer treatment (including type and dosage of chemotherapy and radiation) and fatigue among various cancer survivors [8]. Rather the interaction of biological, psychosocial, and behavioural factors seem to be particularly important in initiation and prolongation of fatigue among cancer survivors [8].

Leventhal’s common sense model provides a theoretical framework for understanding individual differences in experiencing persistent fatigue. This model describes that individuals respond to illness and threat, by formulating illness perceptions, that is, cognitive and emotional representations.[9] These illness perceptions comprise several dimensions: consequences, timeline, personal control, treatment control, identity, concern, coherence and emotional representation. The dimensions personal and treatment control express the belief an individual has to what extent their illness can be controlled personally or by treatment, whereas coherence for example indicates the level of disease understanding. According to Leventhal’s common sense model, these illness perceptions guide certain behaviour, which are in turn linked to physical and psychosocial outcomes [10]. A good understanding of your illness (e.g. ‘it is normal to feel fatigued, yet moderate exercise can help’) may guide one into moderate activity which can help in managing the level of fatigue. Whereas illness perceptions have been consistently linked to fatigue in other diseases [11–13], the number of studies relating illness perceptions to fatigue among cancer populations is limited. One study among 155 survivors of various cancers (e.g. breast, haematological, gynaecological, genitourinary and gastrointestinal cancer) found that more negative illness perceptions on the dimensions consequences, identity and emotional representations were related to more fatigue [14]. Additionally, another study found similar relations among 147 relapsed/refractory chronic lymphocytic leukaemia (CLL) patients [15].

Surprisingly, the two aforementioned studies are the only studies that have investigated the potential contribution of illness perceptions to fatigue among cancer survivors. However, both studies did not include all known relevant information on factors related to fatigue among cancer survivors, that is, sociodemographic, clinical, and psychological factors such as depression and anxiety [14,15]. Furthermore, they included various malignancies pooled together [14] or recruited patients in a trial design, reducing generalizability to other CLL patients [15]. We know that illness perceptions differ considerably by cancer type [16] and patients with solid cancers, and it is expected that living with cancer may differ from living after cancer. The aim of this study was however exploratory in nature as we examined the relation between illness perceptions and fatigue, while controlling for sociodemographic, clinical, and psychological factors (symptoms of depression and anxiety) among a group of haematological cancer survivors diagnosed with (non-)Hodgkin lymphoma (NHL and HL) or CLL and for each cancer group separately. Knowledge on the relation between illness perceptions and fatigue may provide useful guidance for cognition-behavioural-based interventions aimed at reducing fatigue via changing the negative illness perceptions among haematological cancer survivors.

Method

Study design and setting

This study used data collected via the PROFILES (‘Patient Reported Outcomes Following Initial treatment and Long term Evaluation of Survivorship’) registry [17] for secondary data analyses. PROFILES is a registry facilitating individual studies by ongoing data collection of patient-reported outcomes which are linked with clinical data from the NCR of all individuals newly diagnosed with cancer in the Netherlands.[17].

Study population

The NCR was used to select adult patients (≥18 years) with HL, NHL, or CLL that were diagnosed between 1999 and 2013. Exclusion criteria were: inability to fill out a questionnaire (due to language problems, severe cognitive deficits such as dementia, severe psychopathology, those with terminal cancer or severe illness) as determined by their (ex)-attending specialist; and being deceased, as determined by information retrieved from NCR, hospital records and Central Bureau for Genealogy (which collects information on all deceased Dutch people from civil municipal registries). Ethical approval (#0734) was obtained from a local Dutch certified medical ethics committee.

Data collection

Description of the data collection has been described previously [17]. In short, survivors were informed about the study via a letter by their (ex-)attending specialist. This letter contained a secured link to a web-based informed consent form and an online questionnaire. Alternatively, survivors could return a postcard to request a paper-and-pencil questionnaire. For each participant, informed consent was obtained.

Measures

Sociodemographic and clinical factors

Gender and age at diagnosis were obtained from the NCR. Educational level (lower, secondary, higher), marital (married/living together versus divorced/widowed/never married) and employment (employed versus not employed) status were assessed via self-designed questions.

The clinical factors; date of diagnosis, cancer type, cancer stage and primary treatment were obtained from the NCR. Cancer type was assessed according to the International Classification of Diseases for Oncology – 3^rd edition (ICD-O-3) [18]. Cancer stage was classified according to Ann Arbour Code for NHL and HL [19], whereas no stage was registered for the CLL survivors. Primary treatments were divided into surgery, systemic treatments (chemotherapy, immune therapy or targeted therapy), radiotherapy, or active surveillance/no therapy. Self-reported weight and height were used to calculate body mass index (BMI). Comorbidities at the time of survey were assessed by the Self-administered Comorbidities Questionnaire (SCQ) [20] and categorised into no comorbidity, 1 comorbidity or >1 comorbidity.

Illness perception

Illness perceptions were assessed by the Brief Illness Perception Questionnaire (BIPQ) assessing eight dimensions: (1) How much does your illness affect your life (consequences); (2) How long do you think your illness will continue (timeline); (3) How much control do you feel you have over your illness (personal control); (4) How much do you think your treatment can help your illness (treatment control); (5) How much do you experience symptoms from your illness (identity); (6) How concerned are you about your illness (concern); (7) How well do you understand your illness (coherence); (8) How much does your illness affect you emotionally (emotional representation) [21]. All questions are rated on a 10-point Likert scale. Answers on item 3 (personal control), 4 (treatment control) and 7 (coherence) are reversed, so that a high score will indicate a more negative illness perception on all items.

Fatigue

The Fatigue Assessment Scale (FAS) is a questionnaire to assess physical and mental fatigue [22]. All items are rated on a 5-point scale and the total score ranges from 10 to 50, with a higher score indicating more fatigue. A score >21 indicates substantial fatigue [23].

Symptoms of depression and anxiety

The Hospital Anxiety and Depression Scale (HADS) consists of two 7-item subscales measuring symptoms of anxiety and depression, respectively [24]. Items are rated on a four-point Likert scale (0–3), with subscale scores ranging from 0 to 21, where higher scores indicate more symptoms.

Statistical analyses

Patient characteristics, illness perceptions, fatigue, and symptoms of depression and anxiety scores were described using frequencies (percentages) and means (standard deviation) for the total sample and stratified by haematological cancer groups (HL, NHL and CCL). To provide insight into the profile (sociodemographic and clinical factors) and scores on fatigue, illness perceptions and psychological variables of haematological cancer groups, continuous scores were compared by means of one-way ANOVA, whereas categorical variables were compared via Chi Square tests. To examine which cancer groups scored different on continuous variables, Games-Howell post-hoc tests were performed as sample sizes were unequal.

The associations between the eight illness perceptions and fatigue (substantially fatigued score >21 on the FAS versus not substantially fatigued score ≤21 on the FAS) were examined by hierarchical logistic regression analyses. In the first model, all BIPQ items were entered in the analysis while controlling for sociodemographic (age, sex, educational level, marital status, and employment status) and clinical factors (time since diagnosis, cancer stage, BMI, and number of co-morbidities) known to be related to illness perceptions and/or fatigue [8]. As cancer stage and cancer treatment are closely related, adding treatment information as covariates to multivariable analyses would result in bias due to over adjustment. Furthermore, it is well-known that cancer-treatment is unrelated to fatigue [8]. In the second model, symptoms of depression and anxiety were added. To provide information on clinical importance, unadjusted difference scores on illness perception scales were calculated for those with versus without substantial fatigue, where a mean difference of at least half a standard deviation is considered clinically important [25]. All logistical regression analyses were performed for the total group of haematological cancer survivors and each cancer group (HL, NHL, CLL) separately. Statistical analyses were performed using SPSS Statistics 23 and were tested two-sided and statistically significant if p < 0.05.

Results

Patient characteristics

A total of 993 haematological cancer survivors were included, of which the majority was diagnosed with NHL (n = 636), 163 with HL and 166 with CLL. For the total sample, the average age at diagnosis was 56.3 years (SD = 14.6), participants were predominantly male (n = 608, 61.2%), married or living together (n = 772, 79.2%) and unemployed (n = 648, 69.8%), Table 1. The majority received systemic therapy (n = 674, 67.9%). Questionnaires were administered after on average 4.4 years post-cancer diagnosis (SD = 2.7). Information on sociodemographic and clinical factors for each haematological cancer group and whether groups are statistically different is presented in Table 2.

Table 1. Patient characteristics in frequencies and percentages or means and standard deviations.

		Subgroups of haematological cancer survivors
	Total group (N = 993)	HL (N = 164)	NHL (N = 655)	CLL (N = 174)	p Value^a
Sociodemograpgics
Age at diagnosis (SD)	56.3 (14.7)	41.6 (15.4)	58.4 (13.1)	62.6 (10.1)	<.01*^a,b,c
Gender
Male	608 (61.2)	89 (54.3)	403 (61.5)	116 (66.7)	.06
Educational level
Lower	142 (14.7)	13 (8.0)	99 (15.6)	30 (17.6)	.01*
Secondary	586 (60.5)	99 (60.7)	387 (60.8)	100 (58.8)
Higher	241 (24.9)	51 (31.3)	150 (23.6)	40 (23.5)
Marital status
Married/living together	772 (79.2)	122 (74.8)	513 (80.0)	137 (80.1)	.33
Divorced/widowed/never married	203 (20.8)	41 (25.2)	128 (20.0)	34 (19.9)
Employment status
Not employed	648 (69.8)	85 (56.7)	447 (73.0)	136 (81.9)	<.01*
Employed	280 (30.2)	65 (43.3)	165 (27.0)	30 (18.1)
Clinical factors
Years since diagnosis (SD)	4.4 (2.7)	4.9 (2.9)	4.3 (2.7)	4.3 (2.6)	.03*^a
Stage at diagnosis
I	224 (29.4)	30 (18.4)	194 (32.4)	NA	<.01*
II	202 (26.5)	83 (50.9)	119 (19.9)
III	142 (18.6)	33 (20.2)	109 (18.2)
IV	194 (25.5)	17 (10.4)	177 (29.5)
Treatment
Systemic therapy	674 (67.9)	159 (97.0)	472 (72.1)	43 (24.7)	<.01*
Radiotherapy	295 (29.7)	98 (59.8)	190 (29.0)	7 (4.0)	<.01*
No therapy	210 (21.1)	1 (0.6)	85 (13.0)	124 (71.3)	<.01*
BMI (SD)	26.3 (4.3)	25.9 (4.2)	26.5 (4.4)	25.9 (4.0)	.12
Comorbidity
None	355 (36.6)	83 (50.9)	224 (35.2)	48 (28.2)	<.01*
1	285 (29.4)	46 (28.2)	195 (30.7)	44 (25.9)
≥2	329 (34.0)	34 (20.9)	217 (34.1)	78 (45.9)

CLL: chronic lymphocytic leukaemia; HL: Hodgkin lymphoma; NA: not available; NHL: Non-Hodgkin lymphoma; SD: standard deviation.

*p < 0.05. Post-hoc tests were conducted for continuous variables.

^a Significant difference (p < 0.05) between HL and NHL patients;

^b Significant difference (p < 0.05) between HL and CLL patients;

^c Significant difference (p < 0.05) between NHL and CLL patients.

Table 2. Differences in fatigue, illness perceptions and symptoms of depression and anxiety scores according to haematological cancer group.

		Subgroups of haematological cancer survivors
	Total (n = 965)	HL (n = 163)	NHL (n = 636)	CLL (n = 166)	p Value^a
Fatigue
FAS total	21.5 (7.7)	21.2 (7.6)	21.8 (7.8)	20.8 (7.5)	.25
Substantial fatigue	412 (42.7)	65 (39.9)	287 (45.1)	60 (36.1)	.08
Illness perceptions
Consequences	4.2 (2.5)	4.4 (2.5)	4.3 (2.6)	3.7 (2.3)	.02*^b,c
Timeline	6.1 (3.6)	3.9 (3.0)	5.9 (3.6)	8.8 (2.3)	<.01*^a,b,c
Personal control	6.1 (3.2)	5.7 (3.2)	6.2 (3.2)	6.4 (3.3)	.20
Treatment control	3.2 (2.5)	2.3 (1.7)	3.2 (2.5)	4.3 (3.1)	<.01*^a,b,c
Identity	3.7 (2.6)	3.9 (2.6)	3.8 (2.7)	3.2 (2.4)	.01*^b^,^c
Concern	4.1 (2.7)	4.0 (2.7)	4.2 (2.7)	4.0 (2.5)	.78
Coherence	4.2 (2.9)	4.0 (2.8)	4.3 (3.0)	4.2 (3.0)	.57
Emotional representation	3.7 (2.6)	4.0 (2.7)	3.8 (2.6)	3.3 (2.5)	.04*^b
Psychological variables
Symptoms of depression	4.2 (3.9)	3.7 (3.9)	4.4 (3.9)	4.1 (3.8)	.09
Symptoms of anxiety	4.5 (3.9)	4.7 (4.3)	4.6 (3.8)	4.2 (3.7)	.49

Scores are presented in mean scores (standard deviation). Substantial fatigue (>21 FAS) in number and percentages. Mean scores of the three haematological cancers are compared by ANOVA- and post-hoc test. CLL: chronic lymphocytic leukaemia; HL: Hodgkin lymphoma; NHL: non-Hodgkin lymphoma.

^a Significant difference (p < .05) between HL and NHL patients;

^b Significant difference (p < .05) between HL and CLL patients;

^c Significant difference (p < .05) between NHL and CLL patients.

Fatigue, illness perceptions, symptoms of depression and anxiety

Haematological cancer groups did not differ on fatigue, depressive symptoms, and symptoms of anxiety (Table 2). There were statistically significant differences across haematological cancer groups on five of the eight BIPQ dimensions: consequences, timeline, treatment control, identity and emotional representation (all p’s < .05). Post-hoc tests revealed that the three groups all differed significantly from each other on the dimensions timeline and treatment control, with CLL survivors reporting highest (i.e. most negative) and HL survivors reporting lowest scores (least negative perceptions). Additionally, CLL survivors reported significantly less negative perceptions for the dimensions consequences and identity than HL and NHL survivors. Finally, HL survivors had higher scores on emotional representations than CLL survivors.

Associations between illness perceptions and fatigue

Total group of haematological cancer survivors

Survivors reporting more negative perceptions on the consequences [odds ratio (OR) = 1.20; 95% confidence interval (95% CI) = 1.06–1.37), identity (OR = 1.26; 95% CI = 1.20–1.41), or coherence scale (OR = 1.11; 95% CI = 1.02–1.21)] reported more often substantial levels of fatigue which was of clinical importance (Figure 1).

Figure 1. Means scores on illness perceptions stratified by those with (>21 FAS) and without (≤21 FAS) substantial fatigue. Illness perception scores (range 0–10) for those with (>21 FAS) and without substantial fatigue. *p < .05 in model 2 of the hierarchical multivariable logistic regression analyses. ^§Clinically important difference, if the difference is ≥ 0.5 standard deviation [25].

For the total group of haematological cancer survivors, those who reported more negative consequences of their illness (consequences, OR = 1.27; 95% CI = 1.27; 1.13–1.42); attributed more symptoms to their illness (identity, OR = 1.29; 95% CI = 1.17–1.43); or had a poorer understanding of their illness (coherence, OR = 1.13; 95% CI = 1.04–1.22) were more often substantially fatigued. In model 2, those reporting more symptoms of anxiety (OR = 1.16; 95% CI = 1.05–1.28) and depression (OR = 1.33; 95%CI = 1.20–1.48) reported more often substantial fatigue.

Hodgkin, non-Hodgkin, and CLL survivors separately

Among all three haematological cancer groups (Table 3), those who reported more consequences of the illness (consequences) (HL: OR = 1.80; 95% CI = 1.19–2.72; NHL: OR = 1.23; 95% CI = 1.08–1.40; CLL: OR = 1.56; 95% CI = 1.08–2.24) were more often substantially fatigued, although only the associations among HL (OR = 2.59; 95% CI = 1.19–5.64) and NHL (OR = 1.17; 95% CI = 1.01–1.34) remained statistically significant and of clinical importance after addition of experiencing symptoms of anxiety and depression.

Table 3. Summary of hierarchical multivariable logistic regression analyses on associates of fatigue for the total group and each haematological cancer group.

	Total group		HL (N = 132)		NHL (N = 501)		CLL (N = 128)
	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2
	OR (95%CI)	OR (95%CI)	OR (95%CI)	OR (95%CI)	OR (95%CI)	OR (95%CI)	OR (95%CI)	OR (95%CI)
Illness perceptions
Consequences	1.27 (1.13–1.42)*	1.20 (1.06–1.37)*	1.80 (1.19–2.72)*	2.59 (1.19–5.64)*	1.23 (1.08–1.40)*	1.17 (1.01–1.34)*	1.56 (1.08–2.24)*	1.39 (0.86–2.24)
Timeline	0.99 (0.92–1.07)	0.99 (0.91–1.07)	0.92 (0.72–1.17)	0.76 (0.51–1.13)	0.99 (0.91–1.07)	0.98 (0.89–1.07)	0.94 (0.73–1.21)	1.03 (0.76–1.40)
Personal control	1.00 (0.93–1.07)	1.02 (0.94–1.10)	0.88 (0.69–1.11)	0.87 (0.60–1.26)	1.01 (0.93–1.09)	1.03 (0.94–1.13)	0.87 (0.70–1.08)	0.80 (0.61–1.05)
Treatment control	1.09 (0.98–1.21)	1.07 (0.95–1.20)	1.16 (0.75–1.80)	1.19 (0.60–2.33)	1.07 (0.96–1.19)	1.04 (0.92–1.18)	1.25 (1.01–1.55)*	1.38 (1.04–1.84)*
Identity	1.29 (1.17–1.43)*	1.26 (1.20–1.41)*	1.57 (1.17–2.10)*	1.21 (0.78–1.85)	1.26 (1.12–1.42)*	1.26 (1.11–1.43)*	1.30 (0.92–1.85)	1.30 (0.82–2.05)
Concern	1.03 (0.91–1.16)	1.00 (0.88–1.15)	0.74 (0.51–1.08)	0.55 (0.29–1.02)	1.09 (0.95–1.25)	1.06 (0.91–1.24)	0.90 (0.63–1.28)	0.98 (0.65–1.49)
Coherence	1.13 (1.04–1.22)*	1.11 (1.02–1.21)*	1.35 (1.06–1.72)*	1.48 (0.99–2.21)	1.10 (1.00–1.20)	1.09 (0.99–1.20)	0.99 (0.80–1.22)	0.82 (0.60–1.11)
Emotional representations	1.10–0.98–1.24)	0.94 (0.82–1.08)	1.25 (0.89–1.77)	1.15 (0.68–1.95)	1.07 (0.94–1.22)	0.93 (0.80–1.08)	1.24 (0.91–1.70)	0.98 (0.64–1.52)
Psychological variables
Symptoms of anxiety	—	1.16 (1.05–1.28)*	—	1.20 (0.87–1.67)	—	1.15 (1.03–1.29)*	—	1.12 (0.84–1.49)
Symptoms of depression	—	1.33 (1.20–1.48)*	—	2.55 (1.50–4.34)*	—	1.28 (1.15–1.43)*	—	1.67 (1.24–2.26)*
Sociodemographics
Age	1.24 (0.79–1.96)	1.56 (0.93–2.64)	1.02 (0.97–1.07)	1.01 (0.94–1.09)	1.01 (0.99–1.04)	1.00 (0.98–1.03)	0.98 (0.91–1.04)	0.93 (0.85–1.01)
Sex	1.01 (0.99–1.03)	1.00 (0.98–1.02)	1.71 (0.48–6.07)	3.10 (0.53–18.07)	1.17 (0.70–1.94)	1.39 (0.78–2.47)	1.83 (0.61–5.44)	3.11 (0.71–13.64)
Education (ref. = lower)
Secondary	0.82 (0.41–1.62)	0.97 (0.45–2.11)	0.19 (0.01–2.75)	0.11 (0.00–14.37)	0.88 (0.42–1.85)	0.95 (0.42–2.15)	3.34 (0.67–16.81)	2.97 (0.37–23.59)
High	0.68 (0.31–1.51)	1.25 (0.50–3.13)	0.25 (0.02–4.08)	0.34 (0.00–42.72)	0.75 (0.31–1.80)	1.30 (0.49–3.45)	1.20 (0.18–7.90)	1.03 (0.09–12.29)
Marital status	1.43 (0.84–2.44)	1.78 (0.97–3.25)	0.97 (0.27–3.45)	2.50 (0.35–18.03)	1.42 (0.76–2.66)	1.81 (0.90–3.64)	0.96 (0.28–3.35)	0.51 (0.10–2.70)
Employment	1.07 (0.61–1.90)	1.57 (0.83–2.96)	1.33 (0.27–6.57)	1.30 (0.12–14.18)	1.30 (0.66–2.55)	1.82 (0.87–3.78)	1.30 (0.26–6.56)	2.22 (0.31–15.72)
Clinical variables
Time diagnosis	1.10 (1.01–1.19)*	1.00 (0.98–1.02)	1.01 (0.82–1.23)	1.01 (0.75–1.37)	1.12 (1.02–1.23)*	1.08 (0.97–1.20)	1.25 (0.99–1.57)	1.19 (0.90–1.57)
Disease stage (ref. = 1)
Stage 2	0.85 (0.48–1.52)	0.95 (0.50–1.79)	0.88 (0.20–3.82)	0.45 (0.05–4.21)	0.78 (0.39–1.55)	0.98 (0.47–2.05)	Na	Na
Stage 3	1.66 (0.88–3.14)	1.39 (0.69–2.82)	1.20 (0.18–8.16)	0.14 (0.01–2.75)	1.71 (0.84–3.47)	1.58 (0.73–3.44)	Na	Na
Stage 4	1.70 (0.96–3.04)	1.75 (0.92–3.32)	3.19 (0.30–33.70)	7.47 (0.29–190.52)	1.75 (0.95–3.21)	1.81 (0.93–3.53)	Na	Na
BMI	1.09 (1.03–1.14)*	1.13 (1.07–1.20)*	1.17 (1.03–1.33)*	1.44 (1.12–1.86)*	1.08 (1.02–1.14)*	1.11 (1.05–1.18)*	0.94 (0.82–1.08)	0.93 (0.77–1.14)
# Comorbidities (ref = 0)
1	1.99 (1.18–3.38)*	1.86 (1.04–3.32)*	6.23 (1.40–27.82)*	9.94 (1.01–97.36)*	1.65 (0.91–2.99)	1.57 (0.83–2.98)	2.01 (0.49–8.25)	1.02 (0.19–5.56)
≥ 2	2.54 (1.45–4.43)*	1.86 (0.99–3.47)	6.68 (1.01–44.00)*	6.76 (0.18–253.89)	2.24 (1.21–4.12)*	1.73 (0.88–3.37)	3.88 (0.98–15.32)	3.01 (0.60–15.10)

For the CLL group no tumour stage was available, to include this patient sample in the total group analyses we provided a fictitious missing value for them. CLL: chronic lymphocytic leukaemia; HL: Hodgkin lymphoma; NHL: non–Hodgkin lymphoma; ref.: reference group.

* p < .05.

HL and NHL survivors who attributed more symptoms to the illness (identity, HL: OR = 1.57; 95% CI = 1.17–2.10 and NHL: OR = 1.26; 95% CI = 1.12–1.42) were more often substantially fatigued, where only the association among NHL survivors (identity, OR = 1.26; 95% CI = 1.11–1.43) remained significant and was of clinical importance after adding experienced symptoms of anxiety and depression to the model.

HL survivors who reported a poorer understanding of the illness (coherence, OR = 1.35; 95% CI = 1.06–1.72) were more often substantially fatigued, although this association was no longer significant after controlling for experienced symptoms of anxiety and depression.

Additionally, CLL survivors with a strong believe that treatment could control their illness were more often substantially fatigued (treatment control, OR = 1.25; 95% CI = 1.01–1.55) which remained significant and of clinical importance after addition of experiencing symptoms of depression and anxiety (OR = 1.38; 95% CI = 1.04–1.84).

Discussion

In this study, we observed that haematological cancer survivors who experience more negative consequences (consequences), attribute more symptoms to their illness (identity) and have a poorer illness understanding (coherence) were more often substantially fatigued. The found significant associations are in line with previous research, showing that these so-called threat perceptions of experiencing consequences and symptom attributing (identity) [10] are the most consistent predictors of fatigue which was found in a cross-sectional study among cancer patients [14] and among 147 CLL patients included in a clinical trial [15]. Additionally, a review and meta-analysis concluded that these perceptions were also related to maladaptive outcomes in general, like poorer quality of life and higher levels of distress [26].

Associations between other dimensions of illness perceptions and experiencing substantial fatigue seemed to be more haematological cancer group specific. That is, having a poorer disease understanding (coherence) in the NHL group and believing treatment can control the illness in the CLL group were related to experiencing substantial fatigue. Since the NHL group had the largest sample size of all cancer groups, differences in statistical power could explain why a relatively small but significant association between believing you have a limited understanding of your illness (high score on coherence) and more fatigue was found in the NHL group but not in the other cancer groups. An explanation for the CLL-specific association between treatment control and fatigue could be found in the fairly incurability of the disease [27]. As the common sense model proposed treatment curability to be a moderator in predicting the effects of illness perceptions on outcomes [10], this may explain why less perceived treatment control was associated with more fatigue among CLL survivors.

Furthermore, no significant associations were observed for the following illness perceptions; how long you think your illness will continue (timeline), believing you can personally control your illness, how concerned you are regarding your illness and how much you are emotionally affected by your illness (emotional representations) and substantial fatigue. Our study sample consisted of (long-term) lymphoma survivors on average 4.5 years after diagnosis. Therefore, it could be argued that, for HL and many NHL survivors treatment has been finished. Hence, beliefs regarding how long your illness (cancer) will continue (timeline), whether you have personal control over it, and experienced concerns and emotional consequences of your cancer are less relevant with respect to currently present fatigue. Nevertheless, CLL and some indolent NHL lymphomas are more chronic in nature requiring active surveillance and treatment, which can explain the found association between perceiving the illness can be controlled by treatment and more substantial fatigue among survivors of CLL.

The current findings have important clinical implications with respect to cognitive-behavioural interventions aimed at reducing (persistent) fatigue. In detail, these interventions can focus on challenging negative illness perceptions in order to create alternative, more positive perceptions, especially with respect to experienced consequences, identity (or symptom attribution), and increasing coherence (illness understanding). A review among patients with coronary heart disease has shown beneficial effects for cognitive-behavioural therapy together with psycho-education and counselling in changing illness [28]. Among cancer survivors, studies have shown comparable results [29,30]. A longitudinal 1-year follow-up study found that positive changes in illness perceptions improved emotional well-being among 57 breast cancer patients who took part in a psychosocial aftercare programme [29]. Similarly, changes in illness perceptions were related to positive changes in psychological well-being over time within a longitudinal 1-year follow-up study including a sample of 189 oesophageal cancer survivors [30]. This provides promising evidence for the effectiveness of interventions aimed at helping individuals acquiring more positive illness perceptions, thereby lessening their substantial fatigue among hematological cancer survivors.

Several limitations of this study should be acknowledged. First, the participants consisted of survivors up to ten years after their cancer diagnosis and the illness perception questions were not specifically aimed at fatigue but rather general reflecting on their cancer. It could be insightful to gather specific information on illness perceptions regarding their fatigue and the experienced fatigue. Furthermore, it is unsure whether participants reflected on their cancer or whether perceptions regarding co-morbidities were taken into account, although we did correct for number of co-morbidities in our analyses. Third, the inclusion of survivors up to ten years after their diagnosis induces survivorship bias. High fatigue levels have been associated with increased mortality (86). Hence, the more fatigued patients may have been underrepresented in the current sample as they had already deceased. Also, highly fatigued patients are probably less likely to participate since their lack of energy. However, no responder analyses could be performed so there is no information on if and how selection bias may influenced our results. Finally, the cross-sectional data precludes judgements regarding directionality of the relations, hence it cannot be ruled out that being substantially fatigued may also influence patients’ illness perceptions. Therefore, future studies should further elucidate underlying mechanisms by which illness perceptions can affect fatigue, preferably by using longitudinal data and by taking baseline fatigue levels—preferably prior to cancer diagnosis and treatment—into account.

A major strength of the current study is its large population-based sample increasing generalizability of our findings. Furthermore, the availability of a wide range of sociodemographic and clinical characteristics and the inclusion of symptoms of depression and anxiety made it possible to control for other known contributors of substantial fatigue. Finally, subgroup analyses for HL, NHL and CLL revealed differences in illness perceptions and their associations with substantial fatigue according to haematological cancer type.

With more people surviving haematological cancers, there should be more attention to and prevention of long-term effects of cancer such as substantial fatigue. Although fatigue may start as a side-effect during cancer treatment, it can develop into a serious long-term health issue among many haematological cancer survivors. Our cross-sectional study shows that those survivors who experience more consequences of their disease and attribute more symptoms to their illness have a higher risk to experience substantial levels of fatigue even years after diagnosis, although causality is unclear. As research into the underlying pathways of fatigue among haematological cancer survivors continues, clinicians should pay attention to not only symptoms of depression and anxiety but also negative illness perceptions. Psychological interventions, like cognitive behavioural therapy changing these illness perceptions may be beneficial in reducing substantial fatigue among haematological cancer survivors.

Disclosure statement

No potential conflict of interest was reported by the author(s).