The current study is a secondary data analysis of the randomized-controlled trial of Managing Cancer and Living Meaningfully (CALM; Mehnert et al., 2020; Scheffold et al., 2015), which investigated the effectiveness of the CALM intervention compared to an active control group in which a non-manualized, supportive, psycho-oncological intervention (SPI) was offered. A total of 206 out of 329 eligible patients participated in the original CALM study (61.2% female; age: M = 57.9; 84.5% UICC IV stage). The participants were either assigned to the CALM intervention group (n = 99) or to the SPI group (n = 107). A reduction in depression and anxiety as well as an improvement in the patients’ quality of life was demonstrated in both groups. However, no significant difference was found between the two study groups (Mehnert et al., 2020). Thus, in the current study, both groups were analyzed as one group. The study was conducted between April 2013 and December 2018 at the Department of Medical Psychology and Medical Sociology at the Leipzig University Hospital and at the Institute and Polyclinic for Medical Psychology at Hamburg-Eppendorf University Hospital. All participants provided written informed consent prior to study participation. Adult patients with a malignant solid tumor disease, UICC disease stage III or IV, a depression score > 9 (PHQ-9 score) or a distress score = > 5 (DT score), fluency in German and no evidence of a cognitive impairment were eligible for participation. A systematic interview was conducted with all patients to check for the following exclusion criteria: lack of interest in participation, unwillingness or inability to participate in all sessions offered, acute suicidality, communicative deficits, a score of <20 on the Short-Orientation-Memory-Concentration-Test or a score of <70 on the Karnofsky-Index (Scheffold et al., 2015; Mehnert et al., 2020). The positive votes of the ethics committees are available (Hamburg PV4435; Leipzig 143-14-14042014). The study was conducted at three measurement points: at baseline (T0), 3-month follow-up (T1) and 6-months follow-up (T2). At each measurement point, patients received a psychological interview within the respective treatment alternative and were asked to complete various validated questionnaires. For more details regarding sampling, recruitment and data assessment, see Mehnert et al., 2020.

The socio-demographic characteristics (age, gender, partnership status) were collected using a standardized self-reporting questionnaire. In addition, all medical and disease-specific data (diagnosis, current metastases, UICC disease stage, previous psychotherapeutic experience) were taken from each patient’s medical records. The survey instruments consist of the following validated questionnaires:

The data was analyzed using IBM SPSS Statistics 29 (IBM Corporation, Armonk NY). Descriptive statistics were used to calculate the sample characteristics for the entire sample. Descriptive methods and mean comparisons via t-tests (between the measurement times T0, T1 and T2) were used to address the first question, thus to determine the course of death anxiety. The effect sizes for all group comparisons were given as Cohen’s d, with effects at d = 0.20 (small effect), d = 0.50 (medium effect) and d = 0.80 (large effect) (Cohen, 2016). To address the second question, a multilevel approach was used to analyze the associated factors for the reduction of death anxiety over time (T2-T0). In separate univariate linear regression models, with the dependent variable reduction of death anxiety, relevant factors were identified. The following were examined as independent variables: sociodemographic variables (age, gender, partnership status, psychotherapeutic experience in years), medical variables (diagnosis group, UICC disease stage, metastases, cancer-specific symptom burden; MSAS) and psychological variables at T0 Baseline (attachment uncertainty; ECR-M16-G, demoralization; DS, death anxiety; DADDS-G, distress; DT). Subsequently, all relevant factors from the univariate regression models (p < 0.05) were integrated into a multiple linear regression model. Results of p < 0.05 were considered statistically significant for all analyses.

Of 329 eligible patients, 206 patients were included in the study. 123 patients (37%) declined to participate in the study (response rate: 63%). In the current secondary data analysis, all patients who had not completed the Death and Dying Distress Scale at Baseline (T0) were also excluded (n = 12), as this value formed the basis for all further calculations, so that the final evaluation sample comprised n = 194 patients (Figure 1).

Study participants and non-participants did not differ in terms of age (study participants: 57.9 years, SD = 11.7; non-participants: 57.7 years, SD = 12.0) or gender (study participants: 61% female; non-participants: 63% female). However, non-participants were more frequently diagnosed with gastrointestinal tumors (60% vs. 40%), lung tumors (25% vs. 15%) and breast cancer (33% vs. 19%) (p < 0.001) compared to the study participants (cf. Mehnert et al., 2020).

There were no differences in age (study participants: 57.5 years, SD = 11.3; drop-outs: 58.6 years, SD = 12.4), death anxiety at baseline (study participants: n = 122, SD = 0.8; drop-outs: n = 72, SD = 0.8) and the UICC disease stage (study participants: 60.7%; drop-outs: 39.3%) between drop-outs and participants who completed all questionnaires. However, there were differences in gender (female drop-outs: 31%; male drop-outs: 53%, p = 0.002).

After 3 months (T1), 149 patients were still participating in the study (drop-out rate: 23%) and 121 after 6 months (T2) (drop-out rate: 37%). The most common reasons for drop-out were death of the patient, the deterioration in patient’s general condition, lack of interest in further participation in the study and drop-out for unknown reasons.

Patients in this study were on average 58 years old (range: 29–81 years); 62% were women. Most patients were married (59%) and had children (72%). The most common tumor entities were gastrointestinal tumors (31%) and gynecological tumors (26%). In total, 16% of the patients had a relapsed cancer diagnosis (Table 1).

Mean death anxiety at Baseline (T0) was M = 1.91 (SD = 0.79), at 3 months follow-up (T1) M = 1.62 (SD = 0.81) and at 6 months follow-up (T2) M = 1.54 (SD = 0.84). Death anxiety showed a significant reduction from baseline (T0) to 3 months follow-up (T1) [t (148) = 5.26, p < 0.001, d = 0.43] and from baseline (T0) to 6 months follow-up (T2) [t (120) = 5.48, p < 0.001, d = 0.50]. There was no significant difference between T1 and T2 [t (119) = −1.43, p = 0.16, d = 0.13] (Figure 2).

The linear univariate regression models showed an effect on the reduction of death anxiety over time for the following variables: patient age, UICC disease stage, subscale anxiety attachment (ECR) and death anxiety (DADDS) at baseline (T0; Table 2). The following variables were not significant: gender (p = 0.63), partnership status (p = 0.48), years of psychotherapeutic experience (p = 0.99), current metastases (p = 0.74), cancer-specific symptom burden (p = 0.84), subjective distress (p = 0.62) and demoralization (p = 0.87).

In the multiple regression analysis, an effect for death anxiety at baseline and UICC disease stage was found. Participants with a higher death anxiety at baseline and a UICC stage III showed a greater reduction in death anxiety over time (Table 2).

The study results show a significant reduction of death anxiety over time in patients with advanced cancer undergoing a short-term psychotherapy. Death anxiety decreased within 3 months after psychotherapy and remained low after 6 months. Many of the presumed predictors for a reduction in death anxiety could not be confirmed within the study. The UICC disease stage (III) and a high level of death anxiety at baseline could be confirmed as predictors for death anxiety, meaning that death anxiety was reduced the most in patients who were most severely affected at the beginning of the short-term therapy and had a lower UICC stage (III).

Death anxiety is a common problem in people with advanced cancer, which causes significant psychological stress. Therefore, it is important to identify or further explore psychotherapeutic interventions, such as CALM, to improve people’s quality of life at the end of their life, especially since alleviating discomfort is the primary goal in palliative care (Grossman et al., 2018). The results on short-term CALM psychotherapy have shown that treatment approaches tailored to the individual are effective in reducing death anxiety in patients with advanced cancer.

Within the main randomized controlled trial of Managing Cancer and Living Meaningfully (CALM; Mehnert et al., 2020; Scheffold et al., 2015), no differential effects were found between the CALM and the control group (SPI), not even with regard to death anxiety. The latter was demonstrably reduced in both active psychotherapy groups. This seems plausible, as the CALM therapy is based on psychotherapeutic interventions that have already been researched and proven to be effective for patients with advanced cancer. It is therefore conceivable that the psycho-oncologists or palliative psychologists working in the control group (SPI), with several years of clinical experience, use treatment concepts similar to those in the CALM manual, so that there were no group differences (Koranyi et al., 2020). However, it remains unclear whether this is due to a lack of treatment differentiation between the two groups or whether patient characteristics, such as gender, age or symptoms, might have an influence on therapist’s adherence and thereby on the implementation of the therapy sessions. The latter cannot be answered in the current study, as no stratified approach was followed in the study. Koranyi et al. (2020) suggested examining therapists’ adherence in future studies using CALM RCT audio recordings to see whether this was creating differential effects. At the same time, as suggested by Mehnert et al. (2020), it could be worthwhile to analyze in a future study whether there are different results when comparing CALM short-term therapy with a control group that is also undergoing a manualised supportive or supportive-expressive psycho-oncological therapy. Nevertheless, it could also not be ruled out, neither in the main study nor in the present secondary data analysis, that the effect was not a psychotherapy effect but a time effect.

The study showed that a high level of death anxiety at the beginning of treatment, as well as an UICC disease stage of III, were associated with a greater reduction in death anxiety over time. It seems plausible that patients who have only ‘tipped over’ into a palliative setting or who have also experienced strong or possibly unexpressed death anxieties benefit most from psychotherapeutic interventions and thus their death anxiety is reduced most in comparison. This is in line with the research of Grumann and Spiegel (2003), who interviewed 12 women with advanced cancer. They offered all women at least one conversation, which was sufficient for most of them. They reported that most women found it relieving to talk openly about their thoughts about death, but also about their death anxiety (Grumann and Spiegel, 2003). Accordingly, in the present study, the first psychotherapy sessions in particular were associated with the strongest reduction of death anxiety, since it was above all the act of addressing and expressing thoughts of death as well as the shared endurance that had the strongest and clarifying effect. In addition, the positive effects of the psychotherapeutic interventions were long-lasting and persisted even after 6 months.

Nevertheless, only a few of the previously assumed predictors for the reduction of death anxiety could be confirmed. Contrary to previous research findings (Soleimani et al., 2020a; Assari and Moghani Lankarani, 2016; Philipp et al., 2017; An et al., 2018; Mercadante et al., 2019; Su et al., 2022; Neel et al., 2015; You et al., 2022; Faller et al., 2016), no effects were found for gender, partnership status, degree of metastasis, demoralization, experienced distress or cancer-specific symptom burden within the patient group. Also, previous psychotherapeutic experiences of the patients had no direct influence on a reduction of the experienced death anxiety; this may be the case, as the experienced death anxiety is not a neurotic but an existential fear, for which it is difficult to prepare. In addition to the targeted predictors addressing death anxiety, it could be promising to look at possible defense mechanisms that patients use to counter existential fears, such as repression, regression or intellectualisation. Hence, it could be instructive to examine whether the choice of defense mechanisms is related to the desire for psychological conversations to counter the experienced death anxiety and could lead to a broader understanding of the interaction between psychological interventions and death anxiety.

Although attachment and age could not be confirmed as predictors for the reduction of death anxiety in the final model, both predictors were significant in the univariate calculations. Contrary to results from previous studies (Yang et al., 2017; Wenzel et al., 1999), in our study older people in particular appeared to benefit from a reduction of death anxiety over time. This seems to be in line with the considerations of Neel et al. (2015), who assume that with increasing age, more confrontations with death take place and, accordingly, more thought is given to one’s own mortality. Taking these considerations further, it would also be plausible that older and attachment-anxious people have less contact and fewer opportunities to talk about their illness-related fears and worries, as well as their death anxiety, and could therefore particularly benefit from targeted psychotherapeutic interventions or appropriately focused short-term psychotherapy.

The secondary data analysis of the present study is based on the randomized and controlled CALM study (Mehnert et al., 2020), which is based on a large and representative sample. One limitation of the study is that there was no control group in the study design for ethical reasons. Thus, it cannot be completely ruled out that the time elapsed between the psychotherapeutic interventions had a greater effect on the study results than the interventions themselves, even if this is very unlikely.

In summary, the study results suggest that there are only a few predictors or indicators that clearly influence the experienced death anxiety in people with advanced cancer. It appears that it is not possible to prepare for existential fears such as death anxiety, but that targeted and specialized psychotherapeutic interventions are helpful in countering the fear. With its results, the current study contributes to expanding the relevance of psychotherapeutic interventions in relation to the treatment and reduction of death anxiety in people with advanced cancer, which is urgently needed in clinical practice.