Introduction
Although bereavement can be a severe stressor, normal grief does not typically require clinical intervention. On the other hand, bereavement can be complicated and trigger the onset of psychiatric disorders such as major depressive disorder (MDD) (Shear et al., Reference Shear2011). Many studies have shown that bereavement-related MDD does not differ from MDD that presents in nonbereavement contexts because they exhibit similar etiologies, patterns of comorbidity, disease course, and treatment responses (Zisook et al., Reference Zisook2012). Thus, the “bereavement exclusion” in the Diagnostic and Statistical Manual of Mental Disorders, fourth edition (DSM-IV), criteria for MDD was removed from the Statistical Manual of Mental Disorders, fifth edition (DSM-5) (American Psychiatric Association, 2013).
In palliative care, substantial numbers of family caregivers are at risk of depression following the death of the patient (Williams & McCorkle, Reference Williams and McCorkle2011). In fact, more than one-fifth of bereaved caregivers report depressive symptoms at 1 year after the death (Kuo et al., Reference Kuo2017). Because of the disease burden associated with MDD, the detection and management of MDD in bereaved caregivers is important; however, there are two problems related to this issue. First, after the patient with cancer dies, contacting the caregivers is difficult unless they are visiting the clinic; telephone monitoring of all bereaved caregivers would be inefficient. Thus, identifying preloss risk factors of postloss MDD in a palliative care setting would be helpful for detecting high-risk groups more efficiently. Recent large population-based studies (Ling et al., Reference Ling2013; Kuo et al., Reference Kuo2017; Nielsen et al., Reference Nielsen2017a; Rumpold et al., Reference Rumpold2017) have reported several predictors of depressive symptoms in bereaved caregivers, including preloss depressive symptoms, spousal relationship status, a younger age, being female, a lower educational level, preloss caregiving load, and so on (Kuo et al., Reference Kuo2017; Nielsen et al., Reference Nielsen2017a). However, there is another problem to consider. Depression scores peak at 6 months postloss and subsequently decrease during the bereavement period (Maciejewski et al., Reference Maciejewski2007); therefore, it is difficult to distinguish between an adaptive depressive mood and MDD using self-rating scales that are not intended to be diagnostic tools for MDD (Pettersson et al., Reference Pettersson2015). If researchers follow-up with their participants within 6 months of a patient's death, then mean depressive symptom scores could be elevated as in the normal grief trajectory (Maciejewski et al., Reference Maciejewski2007). Unfortunately, most previous studies assessed depression with self-rating scales for depression symptoms rather than with diagnostic criteria for MDD (Kim et al., Reference Kim, Lucette and Loscalzo2013; Kuo et al., Reference Kuo2017; Ling et al., Reference Ling2013; Nielsen et al., Reference Nielsen2017a; Rumpold et al., Reference Rumpold2017).
A literature review of electronic databases revealed just two articles that used diagnostic tools to evaluate preloss predictors of MDD in family caregivers of patients with cancer. One study assessed the participants preloss and then at 6 months postloss, and reported that caregiver preloss mental health was related to lower incidences of MDD and anxiety disorders (Garrido & Prigerson, Reference Garrido and Prigerson2014); however, the authors did not analyze the predictors of MDD alone. The other study reported that preloss religiousness prevented depression at 13 months postloss; however, the study was primarily designed to investigate religiousness rather than to identify risk factors of MDD (Fenix et al., Reference Fenix2006). Thus, the present study aimed to identify the prevalence and preloss predictors of MDD at 6 and 13 months postloss in caregivers of terminally ill patients with cancer.
Methods
Study overview
This analysis was performed as one component of a primary study designed to investigate the mental health status of patients and family caregivers of terminally ill patients with cancer in a palliative care unit (PCU) (Kim et al., Reference Kim2014). In the present study, the prevalence and predictors of MDD in caregivers were investigated using a prospective design. Family caregivers were assessed within 4–7 days of patient admission to the PCU. In a subsample of caregivers who agreed to follow-up assessments, additional investigations were performed via telephone at 6 and 13 months after the patient died. The dates of patient death were obtained from clinical records maintained in the database of the Jeonnam Regional Cancer Center, which is linked to the National Statistical Office for cancer deaths, or from bereaved families via telephone.
Participants
Family caregivers of patients with cancer who were admitted to the PCU of Chonnam National University Hwasun Hospital, Hwasun, South Korea, were consecutively enrolled in the main study between March 2009 and August 2011. Because the final patient of the enrolled caregivers died in October 2014, the last participant was followed until November 2015. The following inclusion criteria were applied: (1) caregivers of terminally ill cancer patients who had an estimated survival time of a few months and who were admitted to the PCU for more than 3 days and (2) family members (spouse, son, son-in-law, daughter, daughter-in-law, parent, brother, sister, or other relative) at least 18 years of age who served as the primary caregiver for the patient. The primary caregiver was defined as the family member who provided direct assistance to the patient for the majority of the time not covered by a professional caregiver. Caregivers who were unable to understand the informed consent or the study objectives were excluded. All caregivers received an explanation of the purpose and methodology of the study and subsequently provided informed consent. All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. The study was approved by the institutional review board of Chonnam National University Hwasun Hospital.
Measurements
Data regarding age, gender, years of formal education, marital status, relationship to the patient (spouse or not), whether the caregiver lived with the patient before admission, employment status, caregiving situation, and medical illness were collected at baseline in a face-to-face interview. The number of days from PCU admission to death was calculated. Coping strategies were assessed using the Stress Coping Questionnaire, which consists of 24 items graded on 5-point Likert scale that were selected from the Korean version of the Ways of Coping Questionnaire (Park, Reference Park1995). The Stress Coping Questionnaire, which was validated in a general Korean population, classifies coping strategies as active coping, which includes problem-focused coping and seeking social support, or passive coping, which includes tension reduction and wishful thinking. Scores range from 24 to 120, with higher scores indicating more frequent use of that particular coping strategy (Park, Reference Park1995).
Subjective caregiver burden was assessed with the 7-item short form of the Zarit Burden Interview (ZBI-7) instrument, which was specifically developed to assess the burden associated with palliative care (Gort et al., Reference Gort2005). Overall burden was indexed by total scores on the ZBI-7, with higher scores indicating greater caregiver burden. The ZBI-7 was validated in a Korean population (Kim et al., Reference Kim2006). Hopeful attitude was measured with the 7-item abbreviated version of the Beck Hopelessness Scale (BHS-7; Beck et al., Reference Beck1974), which was validated in a Korean population (Shin et al., Reference Shin, Park and Oh1990). The total BHS-7 score ranges from 0 to 7, with a higher score indicating increased hopelessness. In the present sample, the caregivers were categorized as either hopeful (BHS-7 score of 0) or nonhopeful (BHS-7 score ≥1) because the majority of caregivers (64%) reported a BHS-7 score of 0. This type of dichotomized analysis of BHS scores has been used in previous studies (Breitbart et al., Reference Breitbart2000; Jang et al., Reference Jang2013; Kim et al., Reference Kim2014).
Diagnosis of depression
Diagnoses of depression were determined at each examination per the DSM-IV diagnostic criteria for MDD using the Mini International Neuropsychiatric Interview, which is a structured diagnostic psychiatric interview for the DSM-IV that includes major depression categories (Sheehan et al., Reference Sheehan1988). To avoid interviewer bias regarding the diagnosis of postloss MDD, independent investigators interviewed each participant at the preloss and postloss time points. The Mini International Neuropsychiatric Interview was administered to the caregivers during a face-to-face interview with the clinical research coordinator at baseline to determine preloss MDD and then via telephone by a psychiatrist to determine postloss MDD at 6 and 13 months.
Statistical analysis
First, the prevalence of MDD was calculated. Then, a univariate analysis was performed on the independent variables to identify factors associated with caregiver depression at 6 and 13 months postloss. Categorical variables were compared using chi-square tests; continuous variables were compared using Mann-Whitney U tests. Factors potentially related to caregiver postloss MDD in the univariate analysis (p < 0.1) were entered simultaneously into multivariate analysis models to assess independence using binary logistic regression; p < 0.05 was considered to indicate statistical significance. All statistical tests were performed with the SPSS for Windows software package (v. 23.0; SPSS, Inc., Chicago, IL).
Results
Recruitment
Of the 305 caregivers who consented to participate in the parent study (Kim et al., Reference Kim2014), 198 refused to participate in the follow-up assessments after the patient died, 15 were lost to follow-up, and 92 (30%) were successfully followed at 6 months; these 92 participants formed the sample analyzed in this study. Caregivers who had a spousal relationship with the patient were significantly more likely to participate in the study (p = 0.004). At 13 months after the patient died, seven participants were lost to follow-up and 85 (92%) performed the follow-up assessment at this time point. Loss to follow-up was associated with greater likelihood of a spousal relationship, caregiving on a rotating basis, and a more hopeful attitude (p = 0.041, p = 0.010, and p = 0.047, respectively). Other sociodemographic and clinical characteristics, including MDD diagnosis, did not significantly differ between patient lost to follow-up and those who continued.
Prevalence of MDD
The prevalence of preloss MDD was 21.8% (n = 21). After the patient died, 34.8% of caregivers (n = 32) were diagnosed with MDD at 6 months and 24.7% (n = 21) were diagnosed with MDD at 13 months (Table 1). All 21 participants who were diagnosed with preloss MDD were followed to 13 months; 42.8% (n = 9) were persistently depressed (i.e., at both the 6- and 13-month follow-up; data not shown).
Prevalence rates of depression at preloss and during the bereavement period
MDD, major depressive disorder.
Univariate associations with MDD at 6 and 13 months postloss
The results of the univariate analysis are summarized in Table 2. Preloss MDD was associated with postloss MDD at both 6 and 13 months. Older age, less education, married/cohabiting, spousal relationship with the patient, solo caregiving situation, and number of survival days were associated with caregiver MDD at 6 months; being female, unemployment, and nonhopeful attitude at baseline were associated with persistent MDD at 13 months.
Sociodemographic and psychological characteristics and univariate associations of postloss major depressive disorder
* p value by chi-square and or Mann-Whitney U tests, as appropriate.
† Survival days from palliative care unit admission to death.
IQR, interquartile ratio; MDD, major depressive disorder; SCQ-AC, The Ways of Stress Coping Questionnaire-Active Coping; SCQ-PC, The Ways of Stress Coping Questionnaire-Passive Coping; ZBI-7, 7-item short version of the Zarit Burden Interview.
Subanalysis for persistent MDD from baseline to 13 months postloss
All baseline characteristics (except preloss MDD) and number of survival days were subanalyzed by chi-square or Mann-Whitney U tests to identify relationships with persistent depression among caregivers 13 months after the patient's death. Less education, a greater caregiver burden, a spousal relationship with the patient, and a nonhopeful attitude were significantly associated with persistent MDD (p < 0.05). These results are summarized in Supplementary Table 1.
Predictors of caregiver MDD at 6 months postloss
The multivariate logistic regression model revealed that only preloss MDD significantly predicted postloss MDD at 6 months (odds ratio [OR] = 5.38; 95% confidence interval [CI 95%] = 1.29, 22.43; Table 3).
Multivariate logistic regression models for predictors of MDD on bereavement
* Survival days from palliative care unit admission to death.
CI 95%, 95% confidence interval; MDD, major depressive disorder; OR, odds ratio.
Predictors of caregiver MDD at 13 months postloss
The multivariate logistic regression model revealed that preloss nonhopeful attitude predicted postloss MDD at 13 months (OR = 8.77; CI 95% = 1.87, 41.13). Additionally, unemployment status significantly predicted postloss MDD (OR = 7.10; CI 95% = 1.28, 39.36; Table 3).
Discussion
The present study found that the prevalence rates of MDD in caregivers of terminally ill patients with cancer in a PCU were approximately 35% and 25% at 6 and 13 months postloss, respectively. Preloss MDD in caregivers significantly predicted postloss MDD at 6 months, whereas a preloss nonhopeful attitude and unemployment status predicted postloss MDD in caregivers at 13 months.
The reported prevalence rates of postloss depressive symptoms range from 12% to 42% at 6 months (Kuo et al., Reference Kuo2017; Ling et al., Reference Ling2013; Nielsen et al., Reference Nielsen2017a) and from 18% to 39% at 13 months (Guldin et al., Reference Guldin2012; Kuo et al., Reference Kuo2017; Ling et al., Reference Ling2013). However, those studies used self-report measures of depressive symptoms rather than the diagnostic criteria for MDD. Several longitudinal studies have evaluated MDD diagnoses according to DSM criteria in bereaved families and found prevalence rates of 26–28% before death, 23% at 7 months postloss, and 7–16% at 13 months postloss (Cherlin et al., Reference Cherlin2007; Zisook & Shuchter, Reference Zisook and Shuchter1991). The prevalence of MDD was higher in the present study, but this may be explained by the higher prevalence of depression in Korea compared with Western countries (Cho et al., Reference Cho, Nam and Suh1998). A recent nationwide cross-sectional survey of depressive symptoms among bereaved families in Korea reported that the prevalence of depressive symptoms within 6 months of the death of a family member was 57.6%, which is much higher than that in Western samples (Jho et al., Reference Jho2016).
In the present study, preloss MDD was a predictor of postloss MDD at 6 months. Although the significance was attenuated at 13 months, it is important to note that about 43% of the participants who were diagnosed with preloss MDD were diagnosed with persistent postloss MDD at 6 and 13 months. The results suggest that preloss MDD was not temporary, instead persisting in many cases. Preloss depressive symptoms have consistently been identified as predictors of postloss depressive symptoms (Allen et al., Reference Allen2013; Kuo et al., Reference Kuo2017; Ling et al., Reference Ling2013; Nielsen et al., Reference Nielsen2017a). Of these studies, two investigated predictors at specific time points (Allen et al., Reference Allen2013; Nielsen et al., Reference Nielsen2017a). One reported that preloss depressive symptoms were related to postloss depressive symptoms at 6 months but did not assess the subjects at 12 or 13 months (Nielsen et al., Reference Nielsen2017a); the other reported that preloss depressive symptoms were related to postloss depressive symptoms at 12 months but did not assess the subjects at 6 months (Allen et al., Reference Allen2013). Thus, these findings cannot be directly compared with the present results in terms of the MDD diagnosis and follow-up period.
On the other hand, the present study found that postloss MDD at 13 months had different predictors than postloss MDD at 6 months. Postloss MDD at 13 months was predicted by a preloss nonhopeful attitude and unemployment status but not by preloss MDD. Wakefield et al. (Reference Wakefield, Schmitz and Baer2011) found that bereavement-related depression (BRD) lasting <9 weeks and BRD lasting 9–52 weeks had similar characteristics. However, BRD lasting >52 weeks (1 year) differed from BRD lasting <52 weeks: the former exhibited greater pathology in terms of severity and was associated with more melancholic features than the latter. These results may explain the different predictors of postloss MDD at 6 months versus those of postloss MDD at 13 months in the present study.
Hope is a future-oriented attitude and can be considered a prerequisite for attributing new meanings to one's own past. Because restoring meaning to one's past experiences is important for mental health recovery, hope is important for recovery (Schrank et al., Reference Schrank, Stanghellini and Slade2008). Low hope is a predictor of depressive symptoms in bereaved family members, and hope status exhibits a stable nature from preloss to the bereavement period (Rumpold et al., Reference Rumpold2017). Based on these findings, a nonhopeful attitude might have been related to postloss MDD at 13 months in the present study because a hopeful attitude is related to recovery from MDD in bereaved caregivers.
Preloss unemployment status also predicted postloss MDD at 13 months. It has been consistently reported that employment acts as a protective barrier during the bereavement period (Aber, Reference Aber1992; Roulston et al., Reference Roulston2017) and it is possible that this relationship is mediated by economic status; a poorer financial status increased the risk of depression in bereaved women (Corden et al., Reference Corden, Hirst and Nice2008). Employment may enhance meaning and purpose in life, improve sense of efficacy, increase feelings of control over life, and serve as a supportive resource during stressful life events (Lin et al., Reference Lin, Dean and Ensel1986; Aber, Reference Aber1992).
Meanwhile, more than one-half of caregivers with MDD at baseline did not persistently depress to 13 months postloss. According to the subanalysis, a greater burden during the preloss period was related to persistent MDD from baseline to 13 months after the patient's death. The preloss burden on a family caregiver creates high levels of stress and may influence the development of both preloss and postloss MDD (Nielsen et al., Reference Nielsen2017b). The relief of this burden after bereavement allows some caregivers to recover from MDD. For others, however, the burdens during the preloss period contributed to poor mental health (Große et al., Reference Große, Treml and Kersting2018) or persistent MDD among bereaved family caregivers. However, the relationships between preloss burden and postloss depression trajectory should be verified using multivariate analyses of data from larger samples of individuals with persistent MDD.
The present results have several important clinical and research implications because the prevalence and risk factors of MDD in bereaved caregivers were investigated using diagnostic criteria rather than self-reported depressive symptoms. During a period of grief, it is important to distinguish between normal depressive mood and MDD using accurate diagnostic tools because bereavement-related MDD has a similar prognosis to non–bereavement-related MDD (Zisook et al., Reference Zisook2012). Caregivers who exhibit preloss MDD, a nonhopeful attitude, and unemployment before their patient dying in a PCU should be cautiously monitored and supported. These preloss risk factors will also be useful in clinical settings because it is difficult to maintain contact with caregivers after the death of a patient.
The present study had several limitations that should be considered when interpreting the results. First, the diagnoses of MDD were made using DSM-IV rather than DSM-5 criteria. The DSM-5 includes important changes regarding the diagnosis of MDD and no longer contains the DSM-IV bereavement exclusion (American Psychiatric Association, 2013). However, the DSM-IV bereavement exclusion was applied within 2 months of the loss of a loved one (American Psychiatric Association, 2000) and was not considered at the 6- or 13-month assessments of MDD in the present study. Second, the present study did not differentiate between bereavement-related MDD and non–bereavement-related MDD. However, both should be managed because they have similar prognoses and disease burdens (Zisook et al., Reference Zisook2012). Third, only 30% of the participants from the parent study participated in the bereavement follow-up assessments, which could have resulted in selection bias. Similarly, the attrition rate at 13 months was related to a hopeful attitude, which also raises the risk of selection bias. Fourth, the small sample size of the present study may have limited power in terms of detecting the predictors of MDD. Fifth, we investigated the preloss risk factors for postloss MDD among caregivers of terminally ill patients with cancer, but our assessment of preloss MDD did not involve caregivers’ psychiatric history before the PCU admission of their family member. Further investigations that address the caregivers’ history of MDD before the end-of-life period of their family member are needed. Finally, information about the death of the patient (e.g., do not resuscitate status), may identify important influences on postloss MDD (Garrido & Prigerson, Reference Garrido and Prigerson2014); however, with the exception of survival days, we did not investigate this issue. Future large multicenter studies are warranted to identify the predictors of caregiver depression trajectory during the bereavement period more accurately.
Despite these limitations, the present study is the first to identify preloss risk factors of postloss MDD in caregivers from a PCU setting at 6 and 13 months using diagnostic criteria rather than self-reports. Thus, screening of preloss predictors of postloss MDD in caregivers of terminally ill patients in the PCU, including for preloss depression, a nonhopeful attitude, and unemployment status, is recommended.
Acknowledgments
This work was supported by a National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIP) (NRF-2015R1C1A2A01052445) and by a grant (HCRI16912-1) Chonnam National University Hwasun Hospital Institute for Biomedical Science. The funding sources(s) had no further role in the study design, collection, analysis and interpretation of data; writing if the report; or the decision to submit the manuscript for publication.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S1478951518000366.
Conflicts of interest
Prof. Seon-Young Kim is funded by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIP) and Chonnam National University Hwasun Hospital Institute for Biomedical Science. The remaining authors declare no conflicts of interest.
