Is psychological distress in men recently diagnosed with testicular cancer associated with their neuropsychological test performance?

Abstract

Objective: To study the level of cancer-related distress (CRD) and variables associated with CRD in recently diagnosed testicular cancer patients (TCPs), and to explore associations between distress levels and neuropsychological test performance at the same time-point.

 Methods: As part of a prospective study of their psychological and cognitive functioning, 135 TCPs completed the Impact of Event Scale (IES) as a measure of CRD at a median of 37 days after diagnosis. They also completed the Hospital Anxiety and Depression Scale (HADS) and the Positive and Negative Affect Schedule (PANAS). Among 135 TCPs, 131 were interviewed and 129 were also tested with a neuropsychological battery. All investigations were done after orchidectomy but before any additional treatment. The associations between neuropsychological test-scores and IES, HADS and PANAS were examined.

 Results: Twenty-four percent (95%CI 17%–31%) of the TCPs reported clinically significant CRD (IES-total score>26). No demographic or cancer-related variables were associated with the CRD-level. In univariate analyses, previous mental problems, sleeping problems, a higher level of neuroticism, daily smoking and hazardous alcohol-use were significantly associated with the CRD-level. In multivariate analysis neuroticism, smoking and alcohol-use remained significantly associated with CRD. Four out of 18 neuropsychological test-scores were significantly associated with at least one distress-measure. Increasing distress-levels were associated with decreasing test performance on some measures of attention, working memory and executive functions.

 Conclusions: In newly diagnosed TCPs, the scores on neuropsychological tests should be considered in relation to co-existing mental distress. Future studies should consider adjustment for this on relevant tests. Copyright © 2010 John Wiley & Sons, Ltd.

Introduction

Though testicular cancer (TC) has excellent prognosis 1, receiving the diagnosis is a stressful event for most patients. Our knowledge concerning cancer-related distress (CRD) and variables associated with CRD shortly after the diagnosis is limited, however. Two small studies both found that approximately 30% of recently diagnosed testicular cancer patients (TCPs) reported moderate to high levels of CRD, but did not study associated factors 2, 3. Studies of CRD in other cancer types have shown that young age, low level of education, short time since diagnosis, advanced cancer disease, heavy treatment, dissatisfaction with cancer information, low level of social support, traumatic experiences before the cancer diagnosis and an anxious personality style are significantly associated with increasing levels of CRD 4.

Research-based knowledge of the level of post-diagnosis CRD and of variables associated with such distress is needed to give the best clinical care in this vulnerable time period. However, such research must have very good reasons and be carried out with sufficient understanding of the patients' predicament. During the last decade there has been a growing concern whether systemic chemotherapy may have a detrimental effect on cognitive function (CF) in cancer patients 5. Such side effects could have serious consequences for young patients in particular, because regularly they are under education or less firmly established in work life. Eventual increased risk of reduced CF has so serious consequences that it justifies research on TCPs in their vulnerable post-diagnosis phase. This is also supported by the need for prospective designed studies with baseline assessment before chemotherapy is initiated 5. Such prospective studies have been published for breast cancer 6–9, but due to eventual gender or treatment differences in risk for reduced CF, they also have to be carried out in young male cancer patients. So far two studies of CF in TCPs have been published 10, 11 and neither of them report major differences in CF in TCPs treated with chemotherapy compared with those not treated with chemotherapy. However, both studies were cross-sectional with examination of CFs some years after primary treatment and they had limited sample sizes, which makes it difficult to detect subtle group differences.

In a prospective design focused on detecting changes in CF associated with chemotherapy, a neuropsychological test-battery has to be administered shortly after TCPs have received their diagnosis and are in a vulnerable phase with an increased level of CRD. Research on patients with anxiety disorders or depression have documented that these mood states are associated with reduced test performance on several domains of CF, mostly for executive functions 12, 13. Whether such a reduction of CF is associated with increased level of (acute) distress in non-psychiatric samples is not well established. Some studies have shown a negative impact of distress 14, others have found no significant association between distress and test performance 15, 16, and one study even report improved test performance with increasing distress 17. In TCPs the level of CRD probably differs between a post-diagnosis assessment and those performed later on in their cancer trajectory. If the level of CRD impairs neuropsychological test performance then it is important to adjust for CRD in the statistical models employed for assessing changes in CF over time 18.

Based on these considerations, we have found it justifiable to initiate a prospective study of psychological distress and cognitive functioning in TCPs. In this report we present findings at the baseline evaluation, which was done after orchidectomy, but prior to any additional treatment. We explore the following research questions: (1) What proportion of the recently diagnosed TCPs report clinically significant CRD? (2) Which variables are significantly associated with increased level of CRD? (3) Does the current distress-level significantly affect the results on any of the neuropsychological tests at baseline evaluation?

Patients and methods

Procedures

In Norway, most TCPs have orchidectomy and receive the TC-diagnosis at their local hospital. They are then referred to a university hospital for evaluation of further treatment. Newly diagnosed TCPs in South-Eastern Norway (2 600 000 individuals) are referred for evaluation at either The Norwegian Radium Hospital or Ullevaal University Hospital. TCPs aged 18–60 years referred to these hospitals between August 2006 and September 2008 were invited to the present prospective study evaluating their psychological and cognitive functioning, if they were considered eligible by the responsible clinician. Exclusion criteria were severe mental disorders (psychotic or substance dependence disorders), degenerative brain disease or previous severe brain trauma, presence of brain metastases or severe somatic dysfunction, or lack of sufficient knowledge of Norwegian language. Patients fulfilling any of the exclusion criteria were not invited to participate in the study.

Eligible patients were informed about the study by a written information letter and by oral presentation from a medical doctor or nurse at the clinic. Some patients gave their decision about participation in the study directly to the doctor/nurse, while others gave their decision to the first author (a medical doctor trained in psychiatry) after she contacted informed patients and asked for their decision. If anything was unclear or the patients had any questions, they were answered by her.

The baseline evaluation done by the first author was performed prior to any additional treatment. The evaluation consisted of an interview, some questionnaires and a neuropsychological assessment, and the duration of the evaluation varied from 90 to 120 mins. The evaluations were done at a median of 37 days (range 6–114) after TCPs had received their diagnosis. The majority of patients were not fully informed at evaluation time whether they would receive any additional treatment with chemotherapy or radiotherapy.

Interview data

The interview started with further contact establishment and an open invitation to the patient to ask questions and raise concerns. When good report was established, the following data were collected through a semi-structured interview: socio-demography, previous somatic and mental health, current medication, smoking and sleeping habits. At the end of the interview intellectual functioning was measured by the Norwegian version of the National Adult Reading Test (NART) 19, 20. On the NART, lower score indicates better intellectual functioning (range 0–50).

The level of education was dichotomized into two categories: ⩽12 years and >12 school years completed. Paired relation consisted of those being married or cohabiting. Employment status before the TC-diagnosis was categorized as followed: working full- or part-time, being full-time student, and not working when being on sickness leave ≥2 months before TC-diagnosis, on disability pension or unemployed.

Previous severe somatic diseases or injury was defined as reporting at least one episode requiring hospitalization or extensive medical care for any reason before TC-diagnosis. This information was checked against the patients' medical records. Previous mental problems were defined as reporting at least one episode requiring help from medical professionals or the use of psychotropic medication before TC-diagnosis. Daily smoking concerned any current daily consumption of cigarettes. Sleeping problems were defined as reporting frequent sleeping problems in general or current sleeping problems after the TC-diagnosis.

The patients' satisfaction with the information about TC at the local hospital was recorded as ‘satisfying’ or ‘not satisfying’. The patient was classified as being well informed about TC or not by the interviewer.

Data from medical records

Information about TC-histology (seminoma/non-seminoma), TC-stage at baseline (Royal Marsden Hospital Index 21) and days since diagnosis (days since diagnostic orchidectomy) was obtained from the medical records. TC-stage was dichotomized into ‘stage I (non-metatastic disease)’ or ‘stage II–IV (metastatic disease)’.

Questionnaire data

The Impact of Event Scale (IES) 22, 23 assesses the psychological response to a traumatic event, here getting a diagnosis of TC, and is our measure of CRD. The IES has two subscales: intrusion (IES-I) and avoidance (IES-A). The time frame is the past week. Each item is scored from 0 (‘not at all’) to 5 (‘often’), and higher scores denote more distress. The IES has been validated to measure CRD 24. We used IES-total score >26 as cut-off score for clinically significant CRD 2. Cronbach's coefficient was α = 0.92 for IES-I and α = 0.80 for IES-A in our sample.

The Hospital Anxiety and Depression scale (HADS) 25, 26 consists of two subscales, each with seven items: anxiety (HADS-A) and depression (HADS-D). Each item is scored from 0 (minimally present) to 3 (maximally present) during the past week, with higher scores representing more symptoms. Cronbach's coefficient was α = 0.86 for HADS-A and α = 0.78 for HADS-D in our sample.

The Positive and Negative Affect Schedule (PANAS) 27 was used to measure current affects, and it contains 20 mood-descriptive adjectives; 10 concerning positive affects (PA) and 10 on negative affects (NA). Rating is on a 5-point Likert scale ranging from 1 (‘not at all/very little’) to 5 (‘very much’); and higher sum-scores indicate increasing level of current affect. Both PANAS-PA and PANAS-NA showed Cronbach's α = 0.91 in our sample.

The Eysenck Personality Questionnaire (EPQ) was use to rate neuroticism, which is a basic personality trait representing the tendency to be nervous and to experience negative emotions. Neuroticism was rated by 6 items of an abridged 18-items version of the EPQ 28, 29, and the total score ranged from 0 (low) to 6 (high). Cronbach's α = 0.59 for neuroticism in our sample.

The CAGE Questionnaire assessed hazardous alcohol-use by four items 30, 31 and sum-scores range from 0 (low) to 4 (high). We used a cut-off of ≥2 for possible hazardous use. Cronbach's α = 0.69 was observed for CAGE in our sample.

Neuropsychological test-battery

A relative short neuropsychological test-battery covering the main domains of CFs was employed. Visual learning and memory, attention and working memory were assessed by three tests from the Cambridge Neuropsychological Test Automated Battery (CANTAB) 32–34 on a computer: Paired Associates Learning, Choice Reaction Time (CRT) and Spatial Working Memory (SWM). Verbal learning and memory, motor function, psychomotor speed and executive functions were assessed with conventional paper and pencil neuropsychological tests: Hopkins Verbal Learning Test-Revised (HVLT-R) 35, Grooved Pegboard 36, Trail Making Test A + B 36, Color-Word Interference Test (CW) 1–4 37 and Word Fluency Test (FAS) 38. The test-battery, with a total of 8 validated tests, yielded 15 sub-test scores (Table 1).

Table 1. Description of neuropsychological methods and sub-test scores

Cognitive domain	Neuropsychological test	Cognitive function assessed	Sub-test score
a ^aComputerized test. b ^bSupplementary test.
Learning/memory	Hopkins Verbal Learning Test-Revised (HVLT-R) 35	Verbal learning	Learning/acquisition (total words)
		Verbal memory	Delayed recall (percent retained)
	Paired Associates Learning (PAL)a33	Visual learning/memory	Total errors 6 shapes
Attention/concentration/working memory	Spatial Working Memory (SWM)a33	Working visuo-spatial memory	Between errors (total errors)
		Strategy	Strategy (efficiency)
	Choice Reaction Time (CRT)a33	Speed of choice response	Mean latency (time)
Motor function	Grooved Pegboard (GP) 36	Visual-motor coordination and speed	Dominant hand (time)
			Non-dominant hand (time)
Psychomotor speed	Trail Making Test (TMT)A 36	Psychomotor speed	TMT-A (time)
	Color-Word Interference Test (CW) 1 + 2 37	Psychomotor speed—naming	CW-1 (time)
		Psychomotor speed—reading	CW-2 (time)
Executive function	Color-Word Interference Test (CW) 3 + 4 37	Discrimination ability	CW-3 (time)
		Compound discrimination ability	CW-4 (time)
	Trail Making Test (TMT)B 36	Set-shifting	TMT-B (time)
	Word fluency (FAS) 38	Controlled oral association ability	Total words
	Stockings of Cambridge (SOC)a,b33	Spatial planning/motor control	Mean initial thinking time
			Problems solved in minimum moves
	Intra-Extra Dimensional Set Shift (IED)a,b33	Rule acquisition /set-shifting	Total errors (adjusted for completed stages)

Toward the end of the testing session the patients were asked if they were able and willing to make one or two supplementary computerized CANTAB-tests 33, 34 of executive function namely the Stockings of Cambridge (SOC) and/or Intra-Extra Dimensional Set Shifting (IED) tests, which yielded another three sub-test scores (Table 1).

Cognitive domains assessed, neuropsychological methods used and sub-test scores reported are displayed in Table 1.

Data management and statistics

Data were analyzed by the SPSS program for PC version 15.0 (SPSS Inc, Chicago, IL, USA), using standard descriptive measures as appropriate. Correlations were assessed by Pearson's correlation coefficient for parametric data and Spearman's rho for non-parametric data. Internal consistency was described by Cronbach's coefficient α. Univariate and multivariate linear regression analyses explored the association between relevant independent variables and the IES-total score as the dependent variable. Independent variables were entered in a multivariate analysis if they showed p⩽0.10 in the univariate analyses. The standardized regression coefficient (beta) and p-values were given for these analyses.

Stepwise multivariate linear regression analyses were performed to explore the association between distress-scores and neuropsychological test-results. In these analyses the scores of the IES-total, the HADS-total and the PANAS-NA were independent variables and neuropsychological test-results were the dependent variables. Some of the neuropsychological test-results were transformed (logarithmic or square root transformation) in order to achieve normal distribution. In these stepwise analyses p-values are reported, both crude and after adjustment for NART, education level and age; and standardized betas are reported for the adjusted analyses.

All tests were two-sided, and the significance level was set at p<0.05 and was not corrected for multiple comparisons due to the explorative nature of this study.

Ethics

The study was approved by the Ethical Committee of the Southern Health Region of Norway and the National Data Inspectorate. All participating TCPs delivered written informed consent.

Results

Sample

From 202 eligible patients, a total of 135 TCPs were recruited at baseline (inclusion rate 67%). Non-inclusion was either due to administrative reasons, or the patients declined to take part. An attrition analysis showed no significant differences between the included and non-included TCPs regarding age at diagnosis or TC-stage, whereas a significantly larger proportion of the included TCPs had non-seminoma compared with the non-included (47% vs 30%, p = 0.02, data not shown).

All 135 TCPs filled in the questionnaires, and among them 131 (97%) were interviewed and 129 (96%) were tested with a neuropsychological test-battery. The extended neuropsychological test-battery including the SOC and/or the IED was applied to 71% (92/129 patients) and 69% (89/129 patients) of the test sample, respectively. There were no significant differences in education level or mean distress-scores between the groups assessed with the ordinary or the extended test-battery.

Sample characteristics

The mean age of the 135 TCPs was 34.8 years (range 19–60, SD 8.9). Forty-four percent of the sample had high level of education (>12 years). Other demographic, health and cancer-related variables are described in Table 2.

Table 2. Characteristics of the sample (N = 135)

a ^aIncludes one patient with advanced extragonadal germ cell tumor.
Age at TC-diagnosis (years), mean (SD) [range]	34.8 (8.9) [19–60]
Days from diagnosis to evaluation, median [range]	37 [6–114]
NART-score, mean (SD) [range]	18.2 (8.0) [3–37]
	N(%)
Civil status
Paired relation	86 (64)
Non-paired	49 (36)
Have fathered children	68 (50)
Education
⩽12 years	75 (56)
>12 years	60 (44)
Employment status before TC
In work	117 (87)
Student	10 (7)
Not working	8 (6)
Histology
Seminoma	71 (53)
Non-seminoma	64 (47)
TC-stage
Stage I	105 (78)
Stage II	23 (17)
Stage III	0 (0)
Stage IVa	7 (5)
Previous severe somatic disease or injury	60 (44)
Previous mental problems	37 (27)
Hazardous alcohol-use	31 (23)
Daily smoking	25 (19)
Sleeping problems	47 (35)
Satisfied with information from local hospital	95 (70)
Well informed about TC	108 (80)

Level of CRD (IES-total score)

The mean level of CRD was 18.6 (SD 12.9, range 0–61; Table 3). Clinically significant CRD was observed in 24% (95%CI 17%-31%) of the sample.

Table 3. Mean scores of cancer-related distress, anxiety/depression, affects and neuroticism (N = 135)

	Mean (SD)	Range
Cancer-related distress
IES-total	18.6 (12.9)	0–61
IES-intrusion	10.3 (8.3)	0–35
IES-avoidance	8.3 (6.6)	0–29
Anxiety/depression
HADS-total	7.9 (6.0)	0–31
HADS-anxiety	5.8 (4.0)	0–19
HADS-depression	2.2 (2.7)	0–12
Affects
PANAS-positive affects	32.8 (7.1)	16–50
PANAS-negative affects	17.0 (6.8)	10–43
Neuroticism (EPQ)	1.7 (1.5)	0–6

Variables associated with level of CRD

None of the demographic or cancer-related variables was significantly associated with level of CRD in univariate regression analyses; neither was the level of information about TC (Table 4). Significantly associated with level of CRD in univariate regression analyses were daily smoking, hazardous alcohol-use, previous mental problems, sleeping problems and higher level of neuroticism (Table 4).

Table 4. Linear regression analyses with IES-total score as dependent variable (N = 135)

Independent variablesa	IES-total score
	Univariate analyses		Multivariate analysis
	Std. β	p	Std. β	p
a ^a(reference/non-reference) given for categorical variables.
Age at TC-diagnosis	0.005	0.95	—	—
Paired relation (yes/no)	−0.025	0.77	—	—
Fathered children (yes/no)	−0.006	0.95	—	—
Education (>12 years/⩽12 years)	0.114	0.19	—	—
Employment status (in work or student / not working)	0.155	0.07	0.003	0.97
Days since diagnosis	−0.076	0.38	—	—
Satisfied with information from local hospital (yes/no)	0.132	0.15	—	—
Well informed about TC (yes/no)	0.147	0.09	0.055	0.51
Histology (seminoma/non-seminoma)	0.09	0.92	—	—
TC-stage (stage I /stage II-IV)	−0.118	0.17	—	—
Previous severe somatic disease/injury (no/yes)	0.172	0.05	0.157	0.05
Previous mental problems (no/yes)	0.239	0.005	0.018	0.83
Daily smoking (no/yes)	0.272	0.002	0.244	0.004
Hazardous alcohol-use (no/yes)	0.274	0.001	0.169	0.04
Sleeping problems (no/yes)	0.273	0.001	0.142	0.08
Neuroticism	0.414	<0.001	0.245	0.006

In the multivariate regression analysis, higher level of neuroticism, daily smoking and hazardous alcohol-use were significantly associated with the CRD-level (Table 4).

Mean scores of and correlation between distress-measures

Mean scores of HADS-total, PANAS-PA, PANAS-NA and EPQ-neuroticism are given in Table 3. The correlation coefficients between the IES-total, the HADS-total, and the PANAS-NA score ranged between 0.63 and 0.67, explaining 40–45% of the common variance between these measures.

Associations between neuropsychological test-results and current distress

The 18 neuropsychological sub-test scores were tested for associations against the sum-scores of the IES-total, the HADS-total, and the PANAS-NA. After adjusting for NART, education level and age, four neuropsychological sub-test scores were significantly associated with any of these three distress-scores (Table 5). The between (total) errors on SWM was significantly associated with the scores of HADS-total and PANAS-NA, the mean latency time on CRT was significantly associated with the scores of IES-total and PANAS-NA and the time used on both CW-1 and CW-3 were significantly associated with the IES-total score. All significant associations were inversely correlated which implies that higher distress-scores were associated with lower achievements on these tests. The remaining 14 neuropsychological sub-test scores were not significantly associated with any of the distress-scores after adjustment.

Table 5. Associationsa between neuropsychological tests-scores and distress-scores (n = 129)

Neuropsychological test score (dependent variable)		Independent variable
		IES-total			HADS-total			PANAS-NA
		Crude	Adjusted		Crude	Adjusted		Crude	Adjusted
	Nb	p	Std β	p	p	Std β	p	p	Std β	p
a ^aStepwise linear regression analyses with the neuropsychological test-score as dependent variable and either IES-total, HADS-total or PANAS-NA as independent variable. p-values are reported crude and after adjustment for NART, education level and age; p-values < 0.05 are marked in bold. Standardized β s are given for adjusted analyses. b ^bNumber of participants evaluated with the test. c ^cLog transformed test-score. d ^dSquare-root transformed test-score.
HVLT-R
Learning/acquisition	124	0.12	−0.083	0.31	0.15	−0.017	0.84	0.18	−0.016	0.85
Delayed recall	123	0.11	−0.108	0.23	0.14	−0.079	0.39	0.04	−0.134	0.14
PAL
Total errors 6 shapesc	128	0.24	0.094	0.29	0.35	0.033	0.71	0.51	−0.080	0.37
SWM
Between errorsd	128	0.05	0.148	0.08	0.001	0.204	0.02	0.006	0.201	0.02
Strategy	128	0.16	0.089	0.30	0.10	0.087	0.32	0.03	0.159	0.07
CRT
Mean latency	128	0.04	0.166	0.04	0.03	0.096	0.24	0.007	0.207	0.01
GP
Dominant hand	120	0.19	0.108	0.23	0.28	0.057	0.54	0.30	0.103	0.26
Non-dominant hand	120	0.39	0.072	0.43	0.17	0.094	0.31	0.23	0.132	0.15
TMT
TMT-A	122	0.12	0.129	0.14	0.008	0.154	0.09	0.42	0.052	0.56
TMT-Bc	120	0.11	0.094	0.26	0.009	0.121	0.16	0.28	0.024	0.78
CW
CW-1	121	0.02	0.187	0.04	0.09	0.089	0.33	0.15	0.067	0.46
CW-2	121	0.29	0.072	0.41	0.33	0.004	0.97	0.94	−0.057	0.53
CW-3	121	0.02	0.189	0.03	0.03	0.108	0.22	0.29	0.029	0.74
CW-4	120	0.65	0.018	0.84	0.49	−0.011	0.90	0.79	−0.026	0.78
FAS
Total words	120	0.98	0.035	0.69	0.88	0.078	0.39	0.89	0.079	0.38
SOC
Mean initial thinking timec	92	0.44	0.111	0.28	0.72	0.057	0.60	0.81	0.032	0.76
Problems solved in minimum movesc	92	0.65	0.021	0.84	0.07	0.145	0.18	0.04	0.183	0.08
IED
Total errors (adjusted)c	89	0.49	0.064	0.55	0.85	−0.063	0.56	0.61	0.021	0.85

Discussion

We found that 24% of the newly diagnosed TCPs reported clinically significant CRD. The CRD-level post orchiedectomy was more strongly associated with neuroticism and pre-morbid lifestyle than with cancer-related variables. The distress-level was significantly associated with four out of 18 neuropsychological test-scores, affecting some measures of attention, working memory and executive functions.

The proportion of TCPs displaying clinically significant CRD in our study is in agreement with the prevalence found in the two previous studies 2, 3. However, the prevalence of significant distress is largely dependent on the research method and instruments used. We used the IES, which has been validated for cancer settings 24 and used in several studies of TCPs 2, 39, 40. However, several cut-off levels on the IES-total score have been used to define clinically significant CRD. A selection bias could also influence the prevalence rate, because we may presume that a high distress-level is a major factor when invited patients declined to take part in the study. We are also aware that our study took part in a vulnerable phase for the patients, but as we took outmost care to make the patients feel comfortable in our research setting, we hardly believe that taking part in the research raised the proportion with clinically significant CRD.

Our results indicate that some patients may be vulnerable for high distress-levels regardless of severity of disease, since personality traits, lifestyle and previous health were found to be significantly associated with the CRD-level. Despite these results we do not suggest any routine psychometric screening shortly after a TC-diagnosis, but the finding of 24% of TCPs with CRD underlines the importance of assessing each patient's situation and needs in this vulnerable phase. The associations between high short-term CRD-level and long-term emotional function in TCPs are not known so far.

No demographic variables were significantly associated with the CRD-level in the TCPs. This contrasts findings from other cancer patients 4, where younger age and lower level of education have been associated with high distress-levels. As most TCPs are relatively young at diagnosis, the lack of variability in age may explain the no association between age and CRD-level we found. As to the cancer-related variables, the finding that TCPs with metastatic disease were not significantly more distressed than those with non-metastatic disease is of clinical interest, and may indicate that the patients were not fully aware of the clinical difference between these important features of at this time-point. Somewhat surprising, neither satisfaction with the information from the local hospital nor information-level about TC was significantly associated with the CRD-level; this was surprising because good information is thought to reduce distress.

The three distress-measures (IES, HADS and PANAS) used to explore the impact of distress on neuropsychological test performance capture somewhat different aspects of distress, since more than half of their variance was not shared. We therefore could not expect to find a fully consistent relationship in their associations with the neuropsychological test-scores. We found that 4 of the 18 tests scores were associated with at least one of the distress-measures. These four scores represent measures of attention, working memory and executive functions, whereas none of the plain memory or motor function tasks displayed significant associations. The cognitive domains associated with distress in our study are partly overlapping with the domains reported possibly impaired after chemotherapy 5. However, as we did 54 adjusted regression analyses of associations between test- and distress-scores and used a 5% significance-level, we must be aware of Type I statistical errors. The majority of the 54 analyses (89%) did not display any significant association; hence we conclude that the neuropsychological measures used in this study are relatively resistant to the influence of distress-level. Nevertheless, one should be aware that certain tasks of attention and executive functions may be vulnerable for a negative influence of distress. Statistical models that allow for adjustments of the possibly confounding effect of changing distress-levels should be preferred in analyses of cognitive change in cancer patients 18.

Our study has several strengths. Compared with previous studies we had a relatively large sample of recently diagnosed TCPs who were evaluated systematically shortly after the diagnosis. The possible impact of distress on neuropsychological test-results was explored using three different distress-measures.

There were also limitations of this study. Selection bias might have occurred without our intent because the most distressed patients could have declined to participate, as well as the most well-functioning and busiest patients with no time to spare for our study. The mental and neuropsychological content of the research may also have provoked some of the patients who were in good health, but suddenly had to deal with a threatening somatic disease. Although the psychological effects of orchidectomy hardly have been investigated empirically, van Basten et al.41 have presented evidence for a considerable mental impact of this procedure. It could also be seen as a limitation that our research design is mainly focusing on psychiatric issues, and do not take a broader view on the existential situation of TCPs. However, practical time constraints and a wish to not overwhelm patients in a vulnerable phase precluded such an approach.

All these factors may have lowered the inclusion rate. Further, the Cronbach's alpha of the CAGE and the EPQ were relatively low, and results involving these measures should be interpreted with caution. Finally, the level of information about the malignancy at evaluation time was classified according to interview data by the first author, and subjective bias on her part may have influenced the ratings.

Conclusion

Approximately one-fourth of the TCPs reported high levels of CRD shortly after diagnosis. The CRD-response was significantly associated with level of neuroticism, lifestyle and previous health variables, whereas none of the cancer-related variables were associated with the current distress-level. Clinical staff should be aware that some patients, even with less advanced disease, may experience high level of distress in the first period after diagnosis. Current distress-level can possibly influence performance on some neuropsychological tests, especially on measures of attention and executive function, but the majority of test scores from this study were not significantly associated with the distress-level.

Acknowledgements

Tone Skaali, MD, holds a research grant from the Norwegian Cancer Society.