Personality and choice of coping predict quality of life in head and neck cancer patients during follow-up

Abstract

Introduction. The aim of the present study was to investigate to what extent personality and choice of coping predicted self-reported quality of life (QoL) in successfully treated head and neck squamous cell carcinoma (HNSCC) patients. Materials and methods. We determined QoL by the European Organization for Research and Treatment of Cancer Quality of life Questionnaire (EORTC-QLQ) C30/H&N35, personality by the Eysenck Personality Inventory and coping by the COPE questionnaire. All patients younger than 80 years who had been diagnosed with HNSCC in Western Norway in the period from 1992 to1997, and who had survived until 1999, were sampled. Ninety-six patients (90% response rate) were included 48±2 months after diagnosis. Fifty-five of 58 eligible patients were interviewed a second time 47±1 months after the first interview where neuroticism and QoL questionnaires were answered. Results. Numerical T stage was inversely associated with the second QoL scores (CV: 10–24%). High neuroticism generally predicted low secondary QoL scores both directly (common variance: 17–25%) and adjusted by the QoL values measured simultaneously as the neuroticism (CV: 11–25%). Avoidance focused, problem focused, drinking to cope and coping by humor all predicted QoL scores (CV: 8.5–15%). The present association pattern could still be shown when adjusted for gender, age and educational level when studied by multiple regression analyses. Conclusion. In conclusion, a high T stage, high neuroticism, coping by humor and coping by problem solving directly predicted low QoL whereas neuroticism was also associated with QoL through avoidance coping.

To be diagnosed with, and treated for head and neck squamous cell carcinoma (HNSCC) must be a frightening experience [1]. Furthermore, life as a former HNSCC patient with fear of recurrence, new cancer disease, as well as experiencing sequels caused by the cancer treatment, place considerable demand on the patient. However, most investigators have shown that reported general health-related quality of life (HRQoL) of surviving HNSCC patients is remarkably similar to that of a comparable, general population [2], [3]. Increasing disease load justifies more intense HNSCC tumor treatment with expected worsened QoL following treatment. A reduced QoL should also be expected following increasing TNM stage. This is, however, not well established when referring to global and functional QoL scales beyond the first year following successfully completed HNSCC treatment [2], [4–8]. Yet, QoL differs between individuals – and some have unacceptable low QoL. The aim of the present investigation has been to study some of the underlying reasons for this.

Several definitions of QoL have been proposed. Generally, the definitions have changed from being observer-based to being based on the perspective of the patients [9], [10]. The multiple dimensions of QoL and the importance of symptom-specific as well as general aspects of QoL have also been emphasized [9], [10]. The European Organization for Research and Treatment of Cancer (EORTC) has developed QoL questionnaires aimed at cancer patients [10]. These QoL questionnaires have a non-specific part that is common to all cancer diseases [10], together with disease-specific parts of which one is aimed at HN cancer [11], [12]. These QoL questionnaires fulfill the requirements of a contemporary QoL questionnaire, [6], [7], [13–15] and have therefore been implemented in the present investigation.

Demographic characteristics of the HNSCC patients predict to some extent QoL [16]. It has furthermore, been shown that psychological variables, like personality [17] and choice of coping [18], is associated with self reported QoL. It is therefore of interest to further study the relations between QoL versus personality and QoL versus choice of coping as reported by the patients.

Personality may be defined as those characteristics of the person that account for consistent patterns of feeling, thinking and behavior [19]. There are several approaches to conceptualizing personality [19]. The Eysenck's model [20] utilizes the personality dimensions of “Extraversion” and “Neuroticism”, together with a “Lie dimension”. These dimensions are furthermore 2 of 5 dimensions of the big 5 personality theory [19].

Neuroticism is a broad pervasive dimension of normal personality whereby people vary in their tendency to experience dysphoric emotional states [21]. Individuals high on neuroticism are assumed to predispose to worry regardless of the presence or absence of threats, and to report more subjective health complaints than do stable individuals [21]. We and others have shown a close inverse association between level of neuroticism and level of QoL in HNSCC patients [17], [22].

Lazarus [23] defines coping as ongoing cognitive and behavioral efforts to manage specific external and/or internal demands that are judged to tax or exceed the resources of the person. The fear of cancer-death, the physical symptoms caused by the disease and its treatment, may be viewed as demands made on the HNSCC patients. Coping contribute to enable cancer patients to live with these demands, and experience general well being, despite having been treated for cancer. The principal coping styles identified are problem-focused, emotional-focused and avoidance-focused coping style [23], [24]. Problem-focused coping may be actively to do something to reduce the demand. Emotion-focused coping may be to change the attitudes toward the demand by social support or by a cognitive re-interpretation like: “I have grown as a human being because of having had cancer”. Avoidance coping may be to behave, or to think, as if the cancer disease had never occurred [23], [24].

Drinking to cope (DTC) is a relatively new example of task-specific coping introduced around 1980 [25], [26]. DTC has proven a very interesting concept, which is reasonable reliable as well as stable over time and well connected to problematic (increased) drinking [25], [26]. We have previously shown that as measured by response to the COPE questionnaire, QoL levels are inversely associated with avoidance coping and drinking to cope (DTC) and complexly associated with levels of emotional coping [27].

When self-rated quality of life is correlated with other characteristics of the individual, it is often difficult to define what constitutes the general association. The association with treatment should be relatively easy to interpret, such as associations between increasingly dry mouth and radiation therapy level. This is more difficult with psychological, social and demographic variables. How many of the EORTC C30/H&N35 34 scales and items should be associated with a characteristic of the patients before there is a general association? Our aims have therefore been to define domains of several QoL scores that could be analyzed simultaneously where associations to the variables mentioned above are investigated. In addition, such general domain scores reduce the multiplicity of independent correlations in a QOL study, and thus minimize coincidental significant correlations. In this way interpretations may become easier.

The QoL prediction of personality scores and choice of coping have not been studied with longitudinal design to any extensive degree despite the importance of this question. We have therefore presently aimed to investigate these questions. We have studied to what extent personality, as measured by the Eysenck personality inventory (EPI) scores, and choice of coping as measured by COPE responses, predict EORTC QLQ QoL scores measured at about 4 years interval in patients successfully treated for HNSCC.

Patients and methods

We have included all patients diagnosed with HNSCC in Western Norway in the period from July 1, 1992 to December 31, 1997 and below 80 years of age (n = 122) who had been disease free for at least one year following therapy by December 31, 1999. The first interviews were conducted in the period from December 10, 1998 to February 22, 2000. The mean period between diagnosis and first interview was 48±2 months. Sixteen patients were found not to be eligible for the study at the first interview time point. A further 10 refused to participate. Of the 96 patients primary included, 55 were interviewed a second time 47±1 months later between October 21^st 2002 and March 29^th 2004. These patients are the patients included in the present sample. Of the initial 96 patients 21 were excluded due to age above 78 years at the second interview point, 6 because of other new serious disease, 11 were dead and 3 patients were lost to follow up. None of the eligible firstly interviewed refused to complete the 2^nd interview. Table I shows the sites of the neoplasms (both 4 and 8 years following diagnosis) of the included patients. Table II shows the TNM stages of the twice included patients. The HNSCC treatment is given in Table I. At the first inclusion time point, 14 of the entire cohort were living alone, whereas 11 of the patients interviewed twice were living alone. The mean age of the twice interviewed patients was 56±1 year at the first interview. Twice interviewed were 46 males and 9 females whereas 75 males and 21 females were in the primary cohort.

Table I.  Primary tumor site, education level and treatment given of the included HNSCC patient; both at 48±2 and 95±3 months following diagnosis.

		months after diagnosis
		48±2	95±3
Tumor primary site of the included patient
Lip		1	1
Tongue		15	13
Gingiva		5	4
Mouth floor		5	5
Oral cavity		4	1
Oro/hypopharynx		14	9
Sinus		3	2
Larynx		42	20
Educational status of the included patients.
Primary school		22	9
Blue collar		39	25
Secondary school		14	8
College		12	9
Master degree		6	4
Treatment given to the included patients.
	No	Yes	No	Yes
Radio-sensitizing	91	5	51	4
Radiation therapy	22	74	15	40
Neck radiation therapy	48	48	28	27
Primary tumor surgery	36	60	18	37
Neck dissection	62	34	36	19
Flap reconstruction	84	12	49	6

Table II.  TNM stage of the twice included patients.

	N0	N1	N2	N3	Total
T1	25	2	2	1	30
T2	6	3	3	0	12
T3	3	1	0	0	4
T4	6	2	1	0	9
Total	40	8	6	1	55

Quality of life inventory

The QoL was determined by employing the EORTC QLQ-C30 version 3.0 [10] and the EORTC QLQ-H&N35 aimed at HN cancer patients [11–13]. The answers were given according to a 4-point Likert format, except questions about general health and quality of life, which were given according to a 7-point Likert format. The indexes were scored according to the EORTC guidelines. The C 30 functional scales and the global scale were transformed so that 100% indicates best function and 0% least function of the individual QoL indexes whereas the C 30 symptom scales and the H&N35 scales were transformed so that 0% indicates least and 100% most symptoms.

The QLQ scores that were built up of more than one response, were studied by Crohnbach's α. It was shown that all but the H&N35 “senses” and the C30 “cognitive function” had values above 0.70.

Scores of each cluster were also subjected to reliability analysis. In the original sample, the QLQ-C30 functional scores had a Crohnbach alpha of 0.83, the QLQ-C30 symptom scores had a Crohnbach alpha of 0.83 and the QLQ-H&N35 score had a Crohnbach alpha of 0.85. Thus, it is psychometrically valid to calculate the proposed cluster sum scores.

Eysenck personality inventory

The neuroticism (24 questions), extraversion (24 questions) and lie score (9 questions) dimensions of the Eysenck personality inventory [28] were determined. The subject responds YES or NO to each question. The scales are calculated as sum scores.

The neuroticism scale consists of questions related to mental symptoms such as obsessive thoughts, anxiety, depression and low self esteem, but also includes somatic symptoms like muscle pain, tachycardia and sleeplessness. The scale assesses adjustment versus emotional instability and identifies individuals prone to psychological distress, unrealistic ideas, excessive cravings or urges, and maladaptive coping responses. The low scorer is characterized as calm, relaxed, unemotional and self-satisfied [19].

Extraverted individuals are judged to be sociable, active, talkative and optimistic. The extraversion measure assesses quantity and intensity of interpersonal interaction, activity level, need for stimulation and capacity for joy. The low scoring individual will be reserved, sober, task-oriented and quiet [28].

The lie scale is based on answers to 9 questions with phrases like: “Have you ever stolen anything?” Although originally introduced as a lie scale, it has later been suggested that the response pattern to this scale may be regarded as a measurement of a personality trait [29], possibly with a focus on handling of moral questions.

COPE inventory

Carver, Scheier and Weintraub [30] have developed the COPE questionnaire based on a conceptual framework by Lazarus [23], [24]. The scores for each of 14 assessed coping indexes are calculated as the sum of the responses to 4 different questions that are scored according to a 4-point Likert format. The questionnaire assesses the level of problem-focused coping (Active, Planning, Suppression of competing activity, Restraint, Seeking social support for instrumental reasons), emotional coping (Seeking social support for emotional reasons, Positive reinterpretation and growth, Acceptance, Focusing on and venting of emotions), avoidance coping (Denial, Behavioral disengagement, Mental disengagement), coping by humor and drinking to cope (DTC). The subjects were asked to relate the responses to their cancer disease. In the original cohort, all COPE scales, except active coping and coping by mental disengagement, had Cronbach's alpha values above 0.60 (Data not shown). Factor scores were calculated based on the different coping index responses of the first data set (except coping by humor and DTC) subjected to principal component factor analysis. The correlations between each regression factor and the coping indexes supported an interpretation of the 3 factors as representing problem focused, emotional and avoidance coping respectively (Data not shown).

Education level

The level of education was also determined at the interview. The highest formal education was noted and scored according to how many years of education required at least to reach the noted level of education starting at 7 years of age (Table I; both primary cohort and twice interviewed patients).

Tumour therapy

Radiation therapy was administered by two-dimensional external beam technique using 5 fractions (2.0 Gy) per week throughout the period. Neck dissection was performed unilateral or bilateral, as modified radical or supra-omohyoidal procedure.

The treatment record of each patient was reviewed. It was determined whether or not the patient had been subjected to local surgery, re-constructive flap surgery or neck dissection; all scored as yes or no (0/1) (Table I). Neck radiation therapy was scored on a scale where one point was added if radiated to each of the fields: high left, high right, lower neck. The maximum cumulative dose of radiation therapy to a specific site in each patient was registered and categorised into 5 levels. Whether given radiation therapy to the primary tumour site was scored as 0/1.

Statistics

The statistical program package SPSS was employed (Ver. 14.0; SPSS Inc. Chicago, IL, USA). The Pearson's r, partial correlation analysis, reliability analysis and regression analysis were performed as indicated. Statistical significance was considered if p < 0.05.

In order to add generalized statements about associations with reported QoL, analyses were also performed with clusters of the QoL scores grouped together. Cluster sum scores were constructed with either the QLQ-C30 functional, the QLQ-C30 symptom, or the QLQ-H&N35 scores included. The general health/QoL score was analyzed separately.

Results

Test re-test Eysenck Personality Inventory (EPI)

Pearson's correlations for the neuroticism score re-test was 0.66 (common variance (CV) 43.5%; p < 0.000).

Test re-test QoL scores (EPI)

Table III shows the mean, median, standard deviation and range of the QoL scores 95±3 months after primary HNSCC diagnosis. The EORTC QLQ questionnaire was answered at both time points. Table IV shows the correlations between the answers at the two different investigation points. In general, a high correlation was shown between the different QoL sum scores at each testing point (CV: 36.5–81%; all p < 0.001). Furthermore, the same QoL sum score between the different time points correlated (CV: 22.5–54.5%; all p < 0.001) (Table IV).

Table III.  EORTC C30/H&N35 QoL index-scores measured 95±3 months following diagnosis.

	Mean	Median	Std. Dev.	Lowest scored	Best scored
General health/QoL	73	75	20	8	100
Physical functioning	85	90	16	50	100
Role functioning	81	100	30	0	100
Emotional functioning	85	92	17	25	100
Cognitive functioning	88	100	21	17	100
Social functioning	88	100	20	17	100
Fatigue	24	22	25	100	0
Nausea and vomiting	4	0	13	83	0
Pain	19	0	26	100	0
Dyspnoea	17	0	25	100	0
Insomnia	16	0	26	100	0
Appetite loss	11	0	25	100	0
Constipation	9	0	21	100	0
Diarrhea	13	0	26	100	0
Financial difficulties	9	0	21	100	0
Pain	9	0	16	67	0
Swallowing	8	0	15	67	0
Senses	22	0	32	100	0
Speech	13	0	20	89	0
Social eating	10	0	19	67	0
Social contact	5	0	12	60	0
Sexuality	20	0	30	100	0
Teeth	14	0	27	100	0
Open mouth	20	0	33	100	0
Dry mouth	33	33	37	100	0
Sticky saliva	28	0	34	100	0
Coughing	28	33	32	100	0
Feeling ill	11	0	26	100	0

Table IV.  Correlation between EORTC QLQ C-30/H&N35 responses measured 48±2 months following diagnosis versus QoL sum scores measured 95±3 months following diagnosis.

	Time point II			Time point I
	C30 function	C30 symptom	H&N35	C30 function	C30 symptom
C30 symptom (II)	−0.902***
H&N35 (II)	−0.660***	0.693***
C30 function (I)	0.631***	−0.552***	−0.552***
C30 symptom (I)	−0.416**	0.476***	0.526***	−0.787***
H&N35 (I)	−0.307*	0.312*	0.739***	−0.604***	0.660***

* = P < 0.05. **P<0.01 *** P<0.001.

QoL scores dependent on smoking and alcohol consumption levels

Table V shows no significant correlation between the alcohol consumption level history and the QoL sum scores of the included HNSCC patients. Both number of years smoked, as well as present level of smoking, was inversely associated with QoL scores with maximum CV at 14%.

Table V.  Patient smoked years, cigarettes smoked per week and alcohol consumption history measured 48±2 months following diagnosis versus QoL sum scores measured 95±3 months following diagnosis.

	alcohol consumption history	cigarettes smoked per week	years smoked
cigarettes smoked per week	0.11	1.00	0.30**
years smoked	0.02	0.30**	1.00
C30 general health/QoL	−0.18	−0.32*	−0.38**
C30 functional sum	−0.16	−0.35*	−0.27
C30 symptom sum	0.11	0.28*	0.25
H&N35 sum score	0.04	0.31*	0.34*

* = P < 0.05. **P<0.01 *** P<0.001.

QoL scores dependent on TNM stage and HNSCC treatment level

The numerical T stage predicted QoL level, i.e. with the C30 functional sum score (r = − 0.34, p < 0.05), the C30 symptom sum score (r = 0.32, p < 0.05) and the H&N35 sum score (r = 0.486, p < 0.001) (Table VI). None of the variables indicating the different treatments given predicted the C30 functional sum scores or the C30 symptom sum scores. Whether given neck radiation therapy (RT) and RT dose predicted H&N35 sum score with r = 0.31 (p < 0.05) and r = 0.029 (p < 0.05) respectively (Table VI).

Table VI.  Correlation adjusted by gender and age between TNM stage, indicated treatment and H&N35 sum QoL score measured 95±3 months following diagnosis.

Treatment or TNM stage	H&N35 sum score
T stage	0.49**
N stage	0.07
Radio-sensitizing	−0.03
Radiation dose	0.29*
Neck radiation therapy	0.31*
Primary tumor surgery	−0.01
Neck dissection	0.21
Flap reconstruction	0.21

* = P < 0.05. **P<0.01 *** P<0.001.

QoL scores dependent on personality scores

The neuroticism score measured at time point 1 and the QoL sum scores correlation coefficients are shown in Table VII. All sum QoL scores correlated with the neuroticism scores, i.e. the QLQ-C30 global health/QOL score (r = − 0.50; p < 0.000), the QLQ-C30 functional sum score (r = − 0.51; p < 0.000), the QLQ-C30 symptom score (r = 0.57; p < 0.000), and the QLQ-H&N35 cluster sum score (r = 0.46; p < 0.000). The higher neuroticism score, the lower QoL score was determined.

Table VII.  Correlation between EPI scores measured 48±2 months following diagnosis versus QoL sum scores measured 95±3 months following diagnosis.

	Direct scores			Scores adjusted by age and the same QoL sum score at time point 1.
	neuroticism	extraversion	lie score	neuroticism	extraversion	lie score
Global health/ QoL	−0.50***	0.17	0.18	−0.35*	−0.00	0.09
C30 functional scores	−0.51***	0.12	0.21	−0.19	−0.08	0.14
C30 symptom scores	0.57***	−0.13	−0.18	0.44***	−0.07	−0.13
H&N35 scores	0.46***	−0.05	−0.15	0.30*	0.13	−0.07

* = P < 0.05. **P<0.01 *** P<0.001.

Furthermore, when adjusted for the QoL sum scores at time point 1, a significant correlation was to some extent determined between the neuroticism scores at time point 1 and the QoL scores at time point 2; i.e. the QLQ-C30 global health/QOL score (r = − 0.35; p < 0.05), the QLQ-C30 symptom score (r = 0.44; p < 0.000), and the QLQ-H&N35 cluster sum score (r = 0.30; p < 0.05). The higher neuroticism score, the worse QoL scores developed (Table VII).

In addition, information about age, gender and educational level were included for a partial correlation analysis as control variables correlating level of neuroticism and QoL responses. No changed association was determined between the neuroticism and QoL scores by these analyses compared to the original analysis (results not shown).

At the individual scale level, all C30 indexes but “appetite loss” and “financial difficulties” correlated with the neuroticism levels (Table IX). With the H&N35 scores, the scores mostly related to social activity correlated with the neuroticism scores (Table IX).

Neither the levels of extraversion nor the lie score level were associated with the QoL levels measured 47±1 months later (Table VII).

QoL scores dependent on COPE scores

As previously determined when measured simultaneously, a high level of avoidance coping predicted lowered QoL level, i.e. specifically the QLQ-C30 functional sum score (r = − 0.32; p < 0.05), the QLQ-C30 symptom score (r = 0.38; p < 0.01), and the QLQ-H&N35 cluster sum score (r = 0.36; p < 0.01) (Table VIII). However, when employed partial correlation adjusted with the appropriate QoL score at time point 1, only the QLQ-C30 symptom score (r = 0.27; p < 0.05) retained a significant correlation (Table VIII). At the individual QoL index level, level of avoidance coping correlated with both QoL scores reflecting interaction with others, i.e. C30 “role functioning” and H&N35 “social eating”, as well as more direct symptom indexes like “dyspnoea” and “pain” (results not shown).

Table VIII.  Correlation between COPE scores measured 48±2 months following diagnosis versus QoL sum scores measured 95±3 months following diagnosis.

	COPE scores
	Problem focused	Emotion focused	Avoidance focused	DTC	Coping by humor
Sum QLQ scores	Direct correlations
Global health/ QoL	−0.34*	−0.00	−0.12	−0.29*	−0.34*
C30 functional scores	−0.39**	0.01	−0.32*	−0.28*	−0.34*
C30 symptom scores	0.38**	−0.02	0.38**	0.23	0.29*
H&N35 scores	0.39**	0.08	0.36**	0.10	0.38**
	Correlations adjusted by age and the same QoL index measured at time point 1
Global health/ QoL	−0.34*	−0.00	−0.12	−0.18	−0.23
C30 functional scores	−0.41**	0.25	−0.09	−0.11	−0.11
C30 symptom scores	0.39**	−0.10	0.27*	0.09	0.15
H&N35 scores	0.41**	−0.07	0.20	−0.11	0.24

* = P < 0.05. **P<0.01 *** P<0.001.

Interestingly, as opposed to when measured simultaneously, a high level of problem focused coping was inversely associated with QoL levels, i.e. the QLQ-C30 global health/QOL score (r = − 0.34; p < 0.05), the QLQ-C30 functional sum score (r = − 0.39; p < 0.01), the QLQ-C30 symptom score (r = 0.38; p < 0.01), and the QLQ-H&N35 sum score (r = 0.39; p < 0.01). Furthermore, the associations between problem focused coping and the QoL scores were not changed by adjusting with the same QoL index measured at time point 1 (Table VIII). At the individual QoL index level, problem focused coping level correlated with QoL indexes most reflecting interaction with other human beings (Table IX).

Table IX.  QoL indexes measure 95±3 months following diagnosis correlated with neuroticism and COPE scores measured 48±2 months following diagnosis.

QLQ indexes	Neuroticism	Problem focused coping	Avoidance focused coping	Drinking to cope	Coping by humor
	C30 QLQ scores
Physical functioning	−0.27*	−0.20	−0.26	−0.16	−0.33*
Role functioning	−0.30*	−0.28*	−0.33*	−0.17	−0.28*
Emotional functioning	−0.51***	−0.41**	−0.32*	−0.07	−0.24
Cognitive functioning	−0.51***	−0.24	−0.18	−0.47***	−0.33*
Social functioning	−0.44***	−0.42**	−0.12	−0.18	−0.17
Fatigue	0.49***	0.35**	0.30*	0.12	0.37**
Nausea and vomiting	0.51***	0.27	0.33*	−0.02	0.15
Pain	0.46***	0.39**	0.14	0.19	0.23
Dyspnoea	0.35**	0.26	0.36**	0.11	0.33*
Insomnia	0.46***	0.13	0.20	0.09	0.03
Appetite loss	0.23	0.32*	0.24	0.21	0.23
Constipation	0.31*	0.08	0.52	0.12	0.15
Diarrhoea	0.43***	0.22	0.08	0.39**	0.15
Financial difficulties	0.22	0.20	0.21	0.10	0.13
	H&N35 QLQ scores
Pain	0.33*	0.25	0.33*	0.04	0.31*
Swallowing	0.29*	0.26	0.42**	0.07	0.25
Senses	0.12	0.10	0.13	0.22	0.47***
Speech	0.21	0.37**	0.09	0.02	0.16
Social eating	0.35*	0.28*	0.34*	−0.06	0.18
Social contact	0.36**	0.29*	0.40**	0.12	0.10
Sexuality	0.49***	0.34*	0.25	0.07	0.13
Teeth	0.23	0.28*	0.31*	0.24	0.05
Open mouth	0.07	0.08	0.06	−0.03	0.17
Dry mouth	0.26	0.15	0.24	0.12	0.37**
Sticky saliva	0.37**	0.16	0.24	0.03	0.23
Coughing	0.12	0.21	−0.03	0.04	0.08
Feeling ill	0.35*	0.31*	0.22	−0.14	0.23

* = P < 0.05. **P<0.01 *** P<0.001.

Drinking to cope levels (DTC) at time point 1 was also associated with the QoL levels at time point 2. Specifically the QLQ-C30 global health/QOL score (r = − 0.29; p < 0.05) and the QLQ-C30 functional sum score (r = − 0.28; p < 0.05) correlated significantly (Table VIII). However, when adjusted by the QoL levels measured at time point 1, no significant partial sum score correlation was obtained. At the individual QoL index level, levels of DTC were especially correlated with cognitive functioning (r = 0.47; p < 0.001) (results not shown).

Coping by humor levels were also correlated with the QoL sum score, i.e. the QLQ-C30 global health/QOL score (r = − 0.34; p < 0.05), the QLQ-C30 functional sum score (r = − 0.34; p < 0.05), the QLQ-C30 symptom score (r = 0.29; p < 0.05), and the QLQ-H&N35 cluster sum score (r = 0.38; p < 0.01). Interestingly, the higher coping by humor score, the worse QoL score was determined. When adjusted by the appropriate QoL scores at time point 1, no significant sum score association was retained.

In addition, information about gender, age and educational level were included for a partial correlation analysis as control variables correlating level of COPE variables and QoL responses. No changed association was determined between the choice of coping and QoL scores (analysis not shown).

Regression analyses with QoL sum scores as dependent variables and COPE/neuroticism scores as independent variables

The COPE indexes and neuroticism scores were also subjected to regression analyses with the different QoL sum scores as dependent variable (Table X). Into these analyses were also introduced ages of the patients. These analyses showed that all the studied models to a large extent predicted the QoL scores. Furthermore, level of neuroticism in all four analyses predicted the QoL (β = − 0.39/ − 0.33/0.42/0.26; p = 0.008/0.014/0.002/0.059). Problem focused coping level uniquely predicted the 3 QLQ sum scores, i.e. the QLQ-C30 functional sum scores (β = − 0.28; p = 0.022), the QLQ-C30 symptom sum scores (β = 0.25; p = 0.036) and the QLQ H&N35 (β = 0.30; p = 0.019) scores. Coping by humor also uniquely predicted of the QLQ-C30 functional sum scores (β = − 0.25; p = 0.047) and the QLQ H&N35 (β = 0.26; p = 0.043) scores (Table X).

Table X.  Multiple regression analysis with respective QoL sum score measured 95±3 months following diagnosis as dependent variable and neuroticism, COPE scores and age of the patient measured 48±2 months following diagnosis as independent variables.

	Standardized			Standardized
	Beta	t	Sig.	Beta	t	Sig.
	General QoL/health			C30 functional scores
Age	0.15	1.22	0.228	−0.03	−0.28	0.784
Neuroticism	−0.39	−2.76	0.008	−0.33	−2.54	0.014
Problem focused coping	−0.15	−1.14	0.260	−0.28	−2.38	0.022
Emotion focused coping	0.08	0.64	0.523	0.23	1.88	0.066
Avoidance focused coping	−0.08	−0.59	0.556	−0.14	−1.17	0.247
Drinking to cope	−0.11	−0.87	0.387	−0.16	−1.32	0.192
Coping by humor	−0.21	−1.62	0.112	−0.25	−2.04	0.047
	C30 symptom scores			H&N35 scores
Age	0.01	0.06	0.952	0.10	0.87	0.389
Neuroticism	0.42	3.31	0.002	0.26	1.94	0.059
Problem focused coping	0.25	2.17	0.036	0.30	2.44	0.019
Emotion focused coping	−0.21	−1.84	0.073	−0.10	−0.81	0.420
Avoidance focused coping	0.20	1.73	0.090	0.23	1.89	0.065
Drinking to cope	0.09	0.74	0.464	−0.04	−0.32	0.748
Coping by humor	0.18	1.54	0.132	0.26	2.08	0.043

In summary, neuroticism accounted for the QoL scores, with about 10% CV direct relation and 10% CV through coping efforts. In addition, coping choice was directly associated with QoL with about 10% CV. In total, psychological factors accounted for about 30% of the CV of the QoL responses. This has been illustrated as Figure 1.

Figure 1.  This figure depicts proposed relations between QoL, neuroticism and choice of coping based on regression analyses as shown in Table X. Predicting common variances: From neuroticism to choice of coping: ∼20%. Directly from neuroticism to QoL: ∼10%. From choice of coping to QoL: ∼10%. From neuroticism via choice of coping to QoL: ∼10%.

We have in addition performed regression analyses where tumor primary sites (larynx/pharynx/oral cavity) were introduced as 3 dependent variables scored as e.g. 1/0/0 if the patients had larynx cancer originally. Such an analysis did not change the results substantially (analyses not shown).

Discussion

Quality of life (QoL) ratings were studied in a group of survivors of HNSCC carcinoma in western Norway with tumors diagnosed through a 5-year period from 1992 to 1997. Personality, coping and first set of QoL data were based on interview primary performed 48±2 months, but at least 1 year after completed therapy. Fifty-five of 58 eligible patients were interviewed a second time 47±1 months after the first interview where neuroticism and QoL questionnaires were answered.

The original cohort contained most of the surviving patients eligible to be included. The same did the group of twice interviewed HNSCC patients. None of the firstly included patients refused a second interview. The EORTC QLQ inventory was employed as the QoL measurement instrument. In order to find clusters of the QoL scores, we have sum scores of the QLQ-C30 functional, the QLQ-C30 symptom and the QLQ-H&N35 scores respectively [8], [27]. The sum scores had acceptable Cronbach alphas. The sum scores must, however, be interpreted in conjunction with the actual QoL indexes that carried the association.

We have previously reported that level of alcohol consumption was marginally inversely associated with global QoL [27]. Reported smoking level was inversely associated with C30 functional/symptom sum QoL scores; both measured simultaneously. If QoL was measured again 4 years later, principally the same relations were determined.

Personality trait scores are usually judged to be stable in adulthood [19], [31]. That has been shown also in the present investigation as regard to neuroticism. Furthermore, as well known [2], [6], [32], the QoL indexes turned out to be relatively stable throughout the investigated time period. Thus, to some extent the QoL score fill the requirements of being a trait of the person [19] and as such to some extent satisfying the definition of being a personality trait. Thus, to some extent the QoL indexes by this reason may be associated with classical personality traits.

This paper reports a relatively strong association between degree of neuroticism and the QLQ scores measured 47±1 months later, i.e. the higher the neuroticism the lower the QoL. The neuroticism scores were broadly associated with the QoL scores. Thus, the well documented association between neuroticism and QoL is further broadened [17]. The analysis of the individual scores showed that most general indexes were associated with neuroticism. Among the QLQ-H&N scores, the scales “feeling ill” and “social contact”, as expected, were correlated with the level of neuroticism. However, symptom scales closely related to expected side effects of treatment (“swallowing” and “sticky saliva”) also correlated with the level of neuroticism. A high neuroticism score furthermore predicted a further decline of the QoL scores 4 years later than should be expected solely by the primary obtained QoL scores. Thus, high neuroticism and low QoL predicts lower future QoL than with low QoL and low neuroticism.

There was no significant association between the QoL and Eysenck personality questionnaire lie or extraversion responses. We have, however not studied another personality trait called optimism which has been shown to be associated with QoL in HNSCC patients [33]. It is likely that this trait also explain some of the QoL variance.

We and others have previously shown that lowered HRQoL was generally inversely associated with an avoidance-focused coping style [18], [34]. This has presently been confirmed also when level of avoidance focused coping was measured 4 years before the QoL scores. Measured simultaneously, no relation was determined between problem focused [18] coping and QoL. However, problem focused coping levels inversely predicted QoL 4 years later. This was true also when the QoL scores at time point 2 were adjusted by the QoL scores at time point 1.

We have previously reported that level of drinking to cope (DTC) was inversely associated with QoL [27]. The present investigation shows that this also holds true in a prospective pattern as measured by C30 global and C30 functional/symptom sum QoL scores. We have also shown that degree of sense of humor predict QoL prospectively [1]. We have presently shown that this also holds true as applied to coping by humor.

Presently, we have shown that the COPE responses in general was associated with QoL were inversely associated, i.e. the worse QoL the more coping efforts, both the indexes expected, and the indexes not expected to be so. One explanation to this may be that patients who actually perceive their QoL to be below what is desired explore all possible coping strategies in order to improve their QoL.

Psycho-social factors other than neuroticism and the COPE responses may explain the association between neuroticism/COPE responses and QoL. Included to the analyses gender, age and level of education of the patients did not change the relation pattern between the QoL levels and neuroticism.

Regression analyses were also performed when the age, neuroticism scores and COPE responses were included as independent variables with the different QoL sum scores as dependent variables. These analyses showed that level of neuroticism was generally the best predictor of QoL of the tested variables. Furthermore, the level of problem focused coping also predicted QoL as measured by these analyses. Coping by humor to some extent also predicted the QoL. It should when interpreting this analysis, however, be born in mind that personality and coping to some extent represent different cognitive levels. Personality may work through choice of coping. In the present case, it can e.g. be said that neuroticism work through DTC and avoidance focused coping. It should also be noticed that the inclusion of more patients probably would have improved the sensitivity of the study.

Based on the regression analyses we have estimated that about 30% of the QoL variance was determined to be explained by personality and choice of coping, with one third as a direct relation between neuroticism and QoL, one third directly between choice of coping and QoL and one third from neuroticism via coping to QoL.

The T stage of the included patients predicted the QoL of the patients. In general, the higher the T stage the lower the QoL. This was generally true, and opposed to the results when measured QoL more closely to the treatment [8]. Thus, short term survivors seem to have a different association between QoL and T stage compared to long-term survivors. The different QoL of short and long time survivors have also been suggested as to patients with Hodgkins disease [35]. Loss of patients between first and second interview is main caused by high age, new serious disease or death, but so far within this study each of these groups represent small groups. It would, however, be of importance to study if any of these exclusion reasons skew the general response pattern of the cohort. This will hopefully be investigated when a larger cohort becomes available.

The present investigation shows that if the aim of treatment is to generally improve QoL of the HNSCC patients, psycho-therapy may be indicated as an integral part of HNSCC treatment, especially because HNSCC patients have higher neuroticism scores than controls [36]. The present findings call for studies where psychological treatment of patients is given within an experimental design in order to study the effect of the QoL levels.

To what extent the present findings differ related to different site of the HNSCC disease is an interesting question which we have tried to answer by regression analyses. So far, our results support that the results do not to any large extent depend on site. It is however needed more investigation in order to answer this question in more detail. The relation between neuroticism, choice of coping and QoL presently reported could even be present among other cancer patients. This warrants performing similar studies with other cancer patients.

Conclusion

A high T stage predicts low QoL. We have shown that high neuroticism generally predicted low QoL both directly and adjusted by the QoL values measured simultaneously as the neuroticism. Furthermore, coping by humor and coping by problem solving directly predicted low QoL whereas neuroticism was in addition associated with QoL through avoidance coping and drinking to cope. In total, the measured psychological factors accounted for about 30% of the variance of the QoL responses.

Acknowledgements

This work has been supported by The Norwegian Cancer Society.