1 INTRODUCTION

Distress in patients with cancer is associated with various negative outcomes, like reduced adherence to treatment,1 treatment complications,2 lower quality of life,3, 4 and mortality.4, 5

Psychosocial distress often accompanies the psychological adaptation process to the diagnosis of cancer. However, identifying patients with clinical levels of psychosocial distress and investigating the associated determinants is crucial for patient-oriented and timely support. As hospitalization often follows or precedes initial diagnosis or other significant crossroads (e.g. discontinuation/change of treatment goals, disease progression), distress screening in the inpatient setting is of high importance. Accordingly, implementation of a distress screening procedure is mandatory in certified comprehensive cancer centers (CCC) in Germany.6

Actually, studies differentiating between treatment settings are scant, and only very few studies explicitly examined psychosocial distress in hospitalized patients with various tumor entities. In these studies, prevalence rates of high psychosocial distress amounted to more than 60%.7, 8

Regarding determinants of distress, previous studies demonstrated the relevance of sociodemographic characteristics: Female gender8-10 and younger age11, 12 were associated with higher distress. Some studies that investigated the clinical characteristics of patients with cancer showed that patients with poor physical performance status are more likely to be distressed13 as are those at advanced disease states.10 The prevalence of distress also varies with cancer site; however, associations are less clear.8High rates of clinical distress were found for example among patients with genitourinary cancer, hematologic, lung, and head and neck cancer.10, 14

Numerous psychosocial variables have been investigated as determinants of distress in patients with cancer. For instance, recent research found negative associations between cancer coping self-efficacy,15 acceptance of cancer16 and elevated distress. However, as people's dispositional characteristics impact psychological well-being, it seems useful to control for such variables before assigning unique predictive power to clinical variables and more volatile personal characteristics.

Research has shown that personality traits and especially neuroticism (usually referring to the Big Five personality trait model), are highly relevant regarding the onset and chronicity of numerous mental disorders.17 With regard to cancer patients, there is evidence that personality traits are associated with distress in outpatients at time of diagnosis,18 in out-patients undergoing treatment19 as well as in cancer survivors beyond primary treatment.20 However, the interplay between individual dispositions and current clinical and psychological variables in determining psychosocial distress in the inpatient setting which is characterized by, among others, reduced mobility, lack of personal space, frequent diagnostic and therapeutic procedures and interaction with multiple health professionals, remains unclear.

Therefore, this study aimed to investigate the relevance of clinical and psychosocial determinants for elevated distress in cancer inpatients when controlling for dispositional determinants (e.g. personality traits).

2 METHODS 2.1 Study design and participants

This was a secondary analysis of the study by Pichler et al.21 that investigated factors associated with decline of psychological support during hospital stay. For this cross-sectional study, inpatients were recruited from the two university hospitals of the Comprehensive Cancer Center Munich (CCC Munich), Germany. The data collection took place between August 2016 and October 2017 at the departments of gynecology, urology, and radiation oncology in each hospital. Participants were eligible if they were ≥18 years, German-speaking, and had a verified diagnosis of a malignant tumor. Exclusion criteria were verbal, mental, or physical impairments that were incompatible with giving informed consent and filling out a self-report questionnaire (physician's assessment). After giving written informed consent, patients completed the questionnaires during their hospital stay. The Ethics Committee of the Technical University of Munich (238/16 S) and the Ethics Committee of the University of Munich (402-16) approved this study. For a more detailed description of the study methods, please see Pichler et al.21

2.2 Measures 2.2.1 Sociodemographic characteristics and clinical information

Sociodemographic data (age, sex, marital status, education, and work situation) and clinical information (tumor entity, date of initial diagnosis, disease status, metastases, and current treatment) were assessed using standardized sheets. Moreover, patients indicated their current physical condition on a visual analog scale (1 = "very good physical condition", 10 = "very bad physical condition").

2.2.2 Psychosocial distress

Psychosocial distress was assessed using self-reporting questionnaires (distress screenings). In certified cancer centers in Germany, the assessment of psychosocial distress via distress screening is mandatory and part of the clinical routine care pathway.

Following the different screening measures implemented at the two university hospitals of the CCCM, we used the Questionnaire on Stress in Cancer Patients-Revised (QSC-R10),22 a self-reporting questionnaire that is routinely used at one partner site of the CCC Munich, and the Distress Thermometer (DT),23 which is routinely used at the other partner site.

The 10 items of the QSC-R10 relate to potentially distressing cancer-related experiences. For each item, patients indicate whether it applies to them or not. If the item applies, they indicate how severely distressed they feel about it (0 = “the problem does not apply to me” to 5 = “the problem does apply and causes severe distress”). A sumscore of ≥15 is recommended as a cut-off for clinically significant distress (Cronbach's Alpha in the current sample: α = 0.87).

The German version of the DT contains a single-item visual analog scale (0 = “no distress” to 10 = “extreme distress”) to assess a global level of psychosocial distress. Further, patients answer a 34-item problem-checklist by checking “yes” or “no” for each item. Different cut-off scores were used for the DT.24 Based on previous studies examining mixed samples of cancer patients, we choose a cut-off of ≥6 at the visual analog scale, which indicates clinically significant distress.12, 25, 26

2.2.3 Depressive symptoms

We used the 2-item-version of the Patient Health Questionnaire (PHQ-2),27 which is a screening instrument for depressive symptoms. A sum score ≥3 differentiates between no/low and clinical level of depressive symptoms (Cronbach's Alpha in the current sample: α = 0.73).

2.2.4 Self-efficacy

We assessed self-efficacy using the short form of the German version of the General Self-Efficacy Scale (ASKU).28 It consists of three items. Participants indicate for each statement to which extent it applies to them (1 = “doesn't apply at all” to 5 = “applies completely”). A higher mean score indicates higher self-efficacy (Cronbach's Alpha in the current sample: α = 0.89).

2.2.5 Personality

The Big Five Inventory-Short Form (BFI-10)29 was applied to assess personality traits. The 10 items of this instrument measure extraversion, agreeableness, conscientiousness, neuroticism, and openness. Higher subscale-values are associated with stronger manifestations of the corresponding personality trait.

2.2.6 Information about psychological support

We assessed informational need regarding psychological support via the question “Do you feel well informed about the psychological support offered in this hospital?” (response options: “yes” and “no”).

2.2.7 Previous use of psychological treatment

Previous uptake of psychological treatment was determined by the item “Have you ever been in psychological treatment?” (response options: “yes, due to my cancer”, “yes, because of other problems” and “no”). For further analysis, we categorized the answers into “yes” and “no”.

2.3 Statistical analysis

We report frequencies (total numbers and percentages), mean values, and standard deviations for sociodemographic, clinical and psychological characteristics, and personality traits for descriptive purposes. The distress scores of the DT and the QSC-R10 were standardized (z-scores) in order to combine the two samples for intercorrelation analysis and linear regression analysis. We performed Pearson correlations for the following variables: distress, age, personality traits (extraversion, neuroticism, openness, conscientiousness, and agreeableness), physical condition, depressive symptoms, and self-efficacy. Finally, we calculated a multiple linear regression analysis with psychosocial distress as dependent variable. The selection of independent variables was based on theoretical considerations. We included sociodemographic characteristics (sex, age, marital status, and education), personality traits (extraversion, neuroticism, openness, conscientiousness, and agreeableness), clinical characteristics (illness duration, metastases, current treatment, and physical condition), and psychosocial characteristics (depressive symptoms, self-efficacy, informational need, and previous use of psychological treatment). We dummy-coded variables if necessary. For missing values within variables with high numbers of missings (education, illness duration, metastases, depressive symptoms, and informational need), we performed a separate category. All statistical tests were two-tailed, and the level of significance was set at p < 0.05. We calculated analyses restricted to patients with complete data regarding distress and performed those using IBM SPSS Statistics for Windows, Version 25.0.30

3 RESULTS 3.1 Sample characteristics

Out of 2999 inpatients with cancer eligible during the study period, we contacted n = 1737 for study participation. Of these, 972 (55.6%) participated and 879 (90.4%) of the participants provided complete data regarding psychosocial distress (for a detailed study flowchart please see Supporting Information). The analyzed sample comprised 48.1% women, mean age was 61.9 years (standard deviation = 11.8). The most frequent cancer diagnoses were prostate (27%), breast (18.1%), and kidney/urinary passages/bladder (11.9%). For half of all patients, illness duration was up to 3 months (50.5%, n = 433); 27.9% had metastases (53.3% had no metastases and 18.8% did not know). Patients were most frequently treated with surgery 64.1% (n = 562), followed by radiotherapy (33.6%, n = 295), chemotherapy (24.4%, n = 214), and hormone therapy (7.3%, n = 64). Further sociodemographic and clinical characteristics are presented in Table 1 and psychosocial characteristics are presented in Table 2.

TABLE 1. Sociodemographic and clinical variables of the study participants (n = 879) Total sample Sociodemographic characteristics M SD Age (n = 878) 61.9 11.8 n % 879 100 Sex (n = 879) Female 423 48.1 Male 456 51.9 Age group (n = 878) ≤50 151 17.2 51–65 364 41.5 66–75 261 29.7 76 and older 102 11.6 Marital status (n = 877) Single 98 11.2 Married/living with partner 617 70.4 Divorced/separated 99 11.3 Widowed 63 7.2 Education level (n = 872) None/elementary school 214 24.5 Junior high 236 27.1 High school 115 13.2 Graduated 285 32.7 Other 22 2.5 Work situation (n = 878) Employed 371 42.3 Unemployed 30 3.4 Retired 416 47.4 Homemaker 45 5.1 Other 16 1.8 Clinical characteristics n % Disease condition (n = 859) First occurrence 626 72.9 Recurrence 123 14.3 Second tumor 84 9.8 Unknowna 26 3.0 Entities (n = 875) Brain 30 3.4 Head & neck 59 6.7 Gastrointestinal 30 3.4 Breast 158 18.1 Female reproductive organs 81 9.3 Kidney/urinary passages/bladder 104 11.9 Prostate 236 27.0 Testicles 7 0.8 Bone/soft tissue 17 1.9 Lung 34 3.9 Others 44 5.0 Multiple entities 75 8.6 Metastases (n = 860) Yes 240 27.9 No 458 53.3 Unknowna 162 18.8 Illness duration (n = 858) Up to 3 months 433 50.5 4 to 12 months 149 17.4 More than 1 year to 5 years 138 16.1 More than 5 years 138 16.1 Current treatment (agree)b Chemotherapy (n = 877) 214 24.4 Radiotherapy (n = 877) 295 33.6 Surgery (n = 877) 562 64.1 Hormone therapy (n = 877) 64 7.3 No therapy (n = 877) 42 4.8 Other therapy (n = 876) 39 4.5 M SD Physical condition (n = 841)c 4.5 2.0 Abbreviations: M, means; SD, standard deviations. a patients who did not know their status and answered that item with “I do not know”. b multiple response possible. c Visual analog scale (1 = “very good physical condition”, 10 = “very bad physical condition”). TABLE 2. Psychological variables of the study participants (n = 879) Total sample Psychosocial characteristics n % Feeling well informed about psych. support (n = 842) Yes 580 68.9 No 262 31.1 Previous psychological treatment (n = 875) Yes 236 27.0 No 639 73.0 Depressive symptoms (n = 847) Significantly elevated 182 20.7 No symptoms/low levels 665 75.7 Missing data 32 3.6 M SD Depressive symptoms (n = 847) 1.64 1.53 Self-efficacy (n = 868) 3.99 0.70 Personality traits Extraversion (n = 869) 3.42 1.03 Neuroticism (n = 867) 2.78 0.92 Openness (n = 866) 3.62 0.99 Conscientiousness (n = 865) 4.14 0.74 Agreeableness (n = 868) 3.36 0.79 Abbreviations: M, means; SD, standard deviations. 3.2 Prevalence and correlates of psychosocial distress

Applying the QSC-R10, 48.2% (n = 191) of the participants reported clinically significant distress. At the second university hospital, which routinely used the DT, 44.5% (n = 215) of the patients were above the cut-off. Intercorrelations of distress, age, personality traits (extraversion, neuroticism, openness, conscientiousness, and agreeableness), physical condition, depressive symptoms, and self-efficacy are shown in Table 3.

TABLE 3. Intercorrelations between distress, age, personality traits (BFI-10), physical condition (VAS), depressive symptoms (PHQ-2), and self-efficacy (AKSU); (n = 812–878) M SD 1 2 3 4 5 6 7 8 9 10 1 Distress (z-score) – – – −0.121** −0.168** 0.365** −0.125** −0.107** −0.004 0.559** 0.612** −0.204** 2 Age 61.89 11.84 – – −0.067* −0.106** 0.034 0.020 −0.064 −0.073* −0.064 −0.033 3 Extraversion 3.42 1.03 – – – −0.197** 0.162** 0.245** 0.140** −0.156** −0.183** 0.226** 4 Neuroticism 2.78 0.92 – – – – −0.160** −0.101** −0.072* 0.193** 0.324** −0.349** 5 Openness 3.62 0.99 – – – – – 0.153** 0.080* −0.089* −0.127** 0.262** 6 Conscientiousness 4.14 0.74 – – – – – – 0.097** −0.104** −0.123** 0.333** 7 Agreeableness 3.36 0.79 – – – – – – – −0.010 −0.031 0.086* 8 Physical condition 4.54 2.00 – – – – – – – – 0.465** −0.195** 9 Depressive symptoms 1.64 1.53 – – – – – – – – – −0.216** 10 Self-efficacy 3.99 0.70. – – – – – – – – – – *p ≤ 0.05; **p ≤ 0.01. 3.3 Determinants of psychosocial distress

Inspection of the intercorrelations did not suggest problems with multicollinearity. Accordingly, the variance inflation factor (VIF) ranged between 1.05 and 2.08, indicating that inclusion of each variable was adequate. Results of the multiple linear regression model showed that younger age (β = −0.061, p = 0.033), higher neuroticism (β = 0.178, p = <0.001), having metastases (β = 0.091, p = 0.002), being in a worse physical condition (β = 0.380, p = <0.001), depressive symptoms (β = 0.270, p = <0.001), not feeling well informed about psychological support (β = 0.054,