Correlation of affective temperament and psychiatric symptoms in palliative care cancer patients

This study was approved by the institutional Ethics Committee of the Medical University of Vienna and Vienna General Hospital (AKH) under the reference number EC 1472/2015.

Data were collected between May and September 2015. During this period a total of 94 inpatients were admitted to the PCU. All were subsequently approached to participate in the study. In total, 55 patients (response rate 58.5%) participated and filled in the questionnaires, which included general sociodemographic questions, the brief TEMPS-M questionnaire (TEMPS: Temperament Evaluation of Memphis, Pisa, Paris and San Diego-autoquestionnaire version) (n = 48, response rate 51%) and the Patient Health Questionnaire (PHQ-9, n = 53, response rate 56%). The inclusion criteria were age over 18 years, ability and willingness to participate by signing an informed consent declaration. Exclusion criteria were mental impairment or patients who could not sign the informed consent due to the severity of the disease.

The general information about the patients included sex, age, height, weight, education, housing and financial situation, smoking status and alcohol consumption, physical diseases, symptoms, pain, medication and a history of suicidal ideation within the past month and the past 12 months.

For the assessment of the patient’s temperament the brief TEMPS-M auto-questionnaire was used [14], which differentiates five basic types of temperament: depressive, irritable, hyperthymic, anxious and cyclothymic [11, 12]. These affective dispositions are seen as biologically determined and time-stable traits of a person’s core personality [11, 12]. The brief TEMPS-M short version consists of 35 items, including 7 items for each type of temperament. It was developed and validated in in two independent German-speaking samples in Germany [14] and Austria [15]. Previous studies showed good internal consistency (alpha = 0.69–0.90) and sufficient validity (r = 0.49–0.72) for all subscales [14, 15].

The short 9‑item German version of the PHQ-9 adapted by Spitzer et al. in 1999 [16] was used to assess depressive and anxiety symptoms. For this study the analysis was based on the sum-score method. A PHQ-9 cut-off score ≥10 showed a pooled sensitivity of 85% and a pooled specificity of 89% for MDD [17]. In another meta-analysis, the pooled sensitivity for a cut-off score of 10 was 0.78 and the pooled specificity was 0.87, indicating acceptable diagnostic properties for MDD [18].

The statistical analysis was performed by IBM SPSS v. 24.0 (Microsoft, New Mexico, USA). The nonparametric Mann-Whitney U‑test was used for rank order between two groups. Associations between interval-scaled variables were tested by Spearman’s correlation coefficient.

Of the 55 participants 30 were female (54.5%) and 25 were male (45.5%), 46 (83.6%) participants were over 50 years and 9 (16.4%) were between 31 and 50 years old. The highest completed education was compulsory school in 4 participants (7.3%), vocational training in 22 participants (40%), secondary school in 15 participants (27.3%) and university degree in 10 participants (18.2%). Of the participants 4 (7.3%) did not indicate their educational status (Table 1) and 10 (18.2%) lived alone. None of the patients stated the financial situation as “not enough” and three participants (5.5%) did not indicate their financial situation (Table 1).

As expected, higher education significantly correlated with a better financial situation (r = 0.482, p < 0.001). While the financial situation was significantly associated with depression scores measured by PHQ-9 (r = 0.348, p = 0.016), there were no associations with other somatic or mental symptoms including PHQ-9 anxiety (p < 0.05).

Gender differences were not significant for self-reported psychiatric and somatic symptoms, PHQ-9, suicidal ideation and temperament (all p < 0.05), with the exception of the cyclothymic temperament, which was significantly more frequent among males (Z = −2.416, p = 0.016).

In total only 7.3% (n = 4) were smokers, 30 participants (54.5%) had stopped smoking and 20 participants (36.4%) had never smoked. Of the participants 5.5% (n = 3) indicated drinking alcohol often but most patients reported drinking alcohol less frequently (n = 16, 29.1%) or never (n = 33, 60%; Table 1).

Of the 55 patients 11 (20.8%) were diagnosed with pancreatic cancer, 9 (17.0%) suffered from lung cancer, 8 (15.1%) were diagnosed with breast cancer, 4 (7.6%) with renal cancer, 3 (5.7%) suffered from colorectal or ovarian cancer, 2 patients (3.8%) had either head and neck, gastric or esophageal cancer, biliary or prostate carcinoma. Furthermore, two patients suffered from cancer of unknown primary (CUP) and one patient each (1.9%) suffered from either malignant melanoma, multiple myeloma, hepatocellular cancer, sarcoma or mesothelioma.

The assessment of psychiatric and somatic symptoms as shown in Table 2 suggests a high burden, with pain, anorexia, constipation and nausea/vomiting being the most prevalent self-reported symptoms in the sample. Interestingly, self-reported symptoms of depression were less commonly reported (21.8%) than unveiled by the structured assessment with PHQ-9.

Of the 55 patients, 53 PHQ-9 questionnaires on depressive symptoms were completed. Of these 26.4% (n = 14) reported mild symptoms of depression and 64.1% (n = 34) had a MDD. Only 8.5% (n = 4) showed no symptoms of depression. Out of 52 completed PHQ-9 questionnaires on anxiety symptoms, 34.6% (n = 18) showed symptoms of anxiety and 3.9% panic disorder. The patients also frequently reported suicidal ideations within the past month (23.6%, n = 13) and within the past 12-months (29.1%; N = 16).

As shown in Table 3 the depressive and cyclothymic temperaments showed significant associations with both depressive (both p < 0.001) symptoms and anxiety (p = 0.002; p = 0.036) as measured by PHQ-9. Furthermore, the anxious temperament was significantly associated with symptoms of depression (p = 0.027). All associations remained significant after controlling for the financial situation (all p < 0.05).

Fig. 1 shows the significant differences in temperament scores (depressive Z = −3.127, p = 0.002; cyclothymic Z = −2.810, p = 0.005; anxious Z = −2.013, p = 0.044) between depressive and non-depressive individuals as defined by the PHQ-9 cut-off ≥10. The irritable (Z = −1.191, p > 0.05) and hyperthymic temperaments (Z = −0.774, p > 0.05) showed no significant differences.

With its cross-sectional design this study has some limitations. The rate of depression in this study is higher than in comparable studies. Lie et al. examined psychiatric symptoms in 969 palliative patients across Europe using the PHQ-9. They found a total rate of 45.3% for any depressive disorder [26]. This might be explained by the fact that self-report instruments generally show higher depression rates or also by the small sample size of consecutive patients within one clinic. Furthermore, patients were evaluated on arrival at the PCU. It is therefore possible that the time shortly after first admission to the palliative care unit might be a period of higher burden and may overlap with symptoms of adjustment disorder. Unfortunately the PHQ-9 cannot appropriately discriminate between depression and adjustment disorders, as many symptoms of both conditions overlap; however, from the perspective of clinical decision making, both adjustment disorder and MDD require appropriate psychiatric interventions. Another observation was the subjective overestimation that occurs when patients are answering questionnaires without an additional interview [27, 28]. To get an accurate psychiatric diagnosis a clinical interview should follow up the questionnaire [29]. Ryan et al. [30] reported that short screening instruments have a high sensitivity of 100% for distressed patients but relatively poor specificity (49–60%), and adding a short question to the screening about the need for help (e. g. “Is this something for which you would like help?”) could considerably increase the specificity in oncological settings. Further studies should also validate self-reported symptoms in a clinical face to face interview.

To the best of our knowledge, this is the first study which investigated the incidence of depression at a PCU with a focus on the individual temperament type. As a clinical implication depression is highly prevalent in terminally ill patients and due to its association with mortality and quality of life, screening for and appropriate treatment of depression within a multidisciplinary team is essential. This study could contribute to increase awareness for psychiatric vulnerabilities and it might be important to consider temperamental traits as risk factors in the etiology and therapy of depressive patients in palliative care. Further investigations to elaborate the concept of temperament in context to psychiatric disorders in palliative care are needed.