Elsevier

Journal of Psychiatric Research

Volume 69, October 2015, Pages 72-79
Journal of Psychiatric Research

Trajectories of depression following spousal and child bereavement: A comparison of the heterogeneity in outcomes

https://doi.org/10.1016/j.jpsychires.2015.07.017Get rights and content

Highlights

  • Most of our understanding of bereavement outcomes comes studies of spousal loss.

  • The extent to which findings are applicable to child bereavement is unknown.

  • The heterogeneity of outcomes following spousal and child bereavement is examined.

  • Latent growth mixture modelling identified similar trajectories for both loss types.

  • Findings underscore the importance of using prospective population based data.

Abstract

Our understanding of how individuals react to the loss of a close loved one comes largely from studies of spousal bereavement. The extent to which findings are relevant to other bereavements is uncertain. A major methodological limitation of current studies has been a reliance on retrospective reporting of functioning and use of samples of individuals who have self-selected for participant in grief research. To address these limitations, in the current study we applied Latent Growth Mixture Modelling (LGMM) in a prospective population-based sample to identify trajectories of depression following spousal and child bereavement in later life. The sample consisted of 2512 individual bereaved adults who were assessed once before and three times after their loss. Four discrete trajectories were identified: Resilience (little or no depression; 68.2%), Chronic Grief (an onset of depression following loss; 13.2%), Depressed-Improved (high pre-loss depression that decreased following loss; 11.2%), and Pre-existing Chronic Depression (high depression at all assessments; 7.4%). These trajectories were present for both child and spousal loss. There was some evidence that child loss in later life was associated more strongly with the Chronic Grief trajectory and less strongly with the Resilience trajectory. However these differences disappeared when covariates were included in the model. Limitations of the analyses are discussed. These findings increase our understanding of the variety of outcomes following bereavement and underscore the importance of using prospective designs to map heterogeneity of response outcomes.

Introduction

In recent years there has been a transformation in our understanding of how people respond to the loss of close loved ones. Emerging longitudinal studies have highlighted the significant heterogeneity in bereavement outcomes: Some people experience ongoing, disabling levels of distress, some people experience symptoms that gradually decline, and others experience little or no disruption in functioning (e.g., Bonanno et al., 2004, Bonanno et al., 2002, Galatzer-Levy and Bonanno, 2012, Mancini et al., 2011, Thomas et al., 2013). Latent growth mixture modelling (LGMM) has emerged as a particularly strong methodology to explore such population heterogeneity. LGMM tests whether the population under investigation is best represented by a single response trajectory or is composed of several discrete populations each characterized by a different growth curve (Curran and Hussong, 2003). LGMM has been used to investigate heterogeneity in a wide variety of potentially stressful life events, including injury, cancer, combat, and bereavement (Bonanno et al., 2012, Bryant et al., 2015, Burton et al., 2014, Galatzer-Levy and Bonanno, 2012, Karstoft et al., 2013, Pietrzak et al., 2014).

In the context of bereavement, Bonanno and colleagues (Boerner et al., 2005, Bonanno et al., 2004, Bonanno et al., 2002) identified trajectories of depression following spousal loss in a prospective sample of participants drawn from the changing Lives of Older Couples (CLOC) study, who were followed up to 48 months post-bereavement. LGMM analysis identified a robust model solution characterized by four distinct trajectories (Galatzer-Levy and Bonanno, 2012): Participants who reported little or no depression at any time point (Resilience; 66.3%); participants who reported a high level of depression at all points (Pre-existing Chronic; 14.5%); participants who reported high levels of depression at the pre-bereavement assessment followed by lower levels of post-bereavement depression (Improved; 10.1%); and participants who reported low pre-bereavement levels of depression followed by high post-bereavement levels (Chronic grief; 9.1%; see also Lotterman et al., 2014). Underscoring the utility of identifying distinct trajectories, financial stress was associated with elevated depression for all classes, whereas poor health was associated with higher depression in the resilient and remitting classes only. That is, financial stress appeared to be a general stressor, whereas, health impacted depression only in populations that were otherwise doing well. Age did not predict response trajectories for any class.

To date, our understanding of the distinct trajectories of bereavement comes largely from studies examining spousal bereavement (e.g., Galatzer-Levy and Bonanno, 2012, Holland et al., 2014, Utz et al., 2012). The extent to which these trajectories are present following other bereavements has yet to be investigated. It is possible that the number, shape, and predictors of response trajectories may vary across loss type, indicating that bereavement response differs by loss type. This may be the case as studies have suggested that individuals who experience the death of a child are more likely to show chronically high levels of distress (Dyregrov et al., 2003, Meert et al., 2011). However this research has relied on samples of individuals who have self-selected for participation in bereavement studies and most involve the loss of young children (e.g., Arnold et al., 2005, Bonanno et al., 2005, Dyregrov et al., 2003, Meert et al., 2011). The extent to which such findings are representative of the population of individuals who have experienced the death of a child is unknown (Levav et al., 1988). Moreover, existing studies have typically relied on retrospective self-report or longitudinal assessments lacking pre-loss assessments, which may also introduce bias associated with self-report of past emotional functioning (Watkins, 2002). True prospective studies are important, however, not only for overcoming memory biases, but also for identifying the full set of population trajectories. In the CLOC study, more than half of participants with high levels of post-bereavement depression were in fact depressed prior to their spouse's death. However, not all individuals who were depressed prior to the death remained depressed (Galatzer-Levy and Bonanno, 2012). Thus, while existing investigations of the impact of child loss have been helpful in highlighting potential differences between child and spousal loss, there is a critical need for the issue to be examined using a prospective design in a sample of individuals not self-selected for bereavement.

We addressed this issue in the current study using data from the Health and Retirement Study, a large nationally representative longitudinal survey that collects data from individuals on an array of family, health, and economic variables every 2 years. Of particular value for addressing questions relating to bereavement, loss information has been recorded but individuals have not been included on the basis of their losses, thus minimizing the possible impact of selection biases. The database also allows for ongoing assessment of pre- and post-loss functioning. It has previously been used to map the heterogeneity of responses to such events as cancer and chronic pain (Burton et al., 2014, Zhu et al., 2014). Accordingly, in the current study, we applied LGMM to the HRS to identify prospective trajectories of depression following spousal and child bereavement. We expected that the trajectories associated with spousal bereavement would be similar to those identified in Galatzer-Levy and Bonanno (2012; see also Bonanno et al., 2004). We hypothesized that we would observe a large class of resilient individuals and three smaller classes comprising pre-existing chronic depression, chronic grief, and improved depression. In the case of child bereavement, we expected that similar trajectories would be identified; however, based on existing literature we anticipated that the proportion of individuals classified into the chronic grief trajectory would be greater (Dyregrov et al., 2003, Meert et al., 2011). As the HRS comprises mainly older respondents, and the majority of the children in the sample were adults at the time of their death, we included age of the deceased among the potential predictors of trajectory membership.

Section snippets

Data

Data were taken from the Health and Retirement Study (HRS), a nationally representative longitudinal study designed to explore numerous aspects of aging in American adults. The study commenced in 1992 and collected data every 2 years. Data for the current analysis were drawn from 9 waves (1994–2010) from the RAND HRS vM and RAND family database. Data from Wave 1 (1992) were excluded as that wave used an alternate measure of depression. (For details of the survey and sampling methods see Sonnega

Unconditional model

Table 2 presents the information indices and likelihood tests for the one to five class models. These indices showed improved fit as the number of classes increased, with the exception of a negligible reduction in entropy and a non-significant result for Lo-Mendell test for the five-class model. This indicated that the five-class model failed to fit the data better than the four-class model. Overall consideration of the indices suggested that the four-class model provided a better fit than

Discussion

In this study we used a large population-based sample and a prospective design to examine trajectories of depression following spousal and child bereavement in later life. While a number of population-based studies have examined depression trajectories in the context of spousal loss (Galatzer-Levy and Bonanno, 2012) this is the first study to use a population-based prospective design to explore trajectories following spousal and child loss. In doing so, this study enables direct comparison of

Role of the funding source

Funding bodies had no involvement in any aspect of this study.

Author contributions

F. Maccallum, I. R. Galatzer-Levy and G. A. Bonanno designed the study and wrote the manuscript. F Maccallum analyzed the data. All authors have approved the final manuscript.

Declaration of conflicting interests

Conflicts of interest: none.

Acknowledgments

Fiona Maccallum is supported by a National Health and Medical Research Council Early Career Research Fellowship (NHMRC GNT1053997). I R Galatzer-Levy is supported by an NIMH grant K01MH102415. G. A. Bonanno is the recipient of an NIMH grant R01MH091034.

References (46)

  • G.A. Bonanno et al.

    Trajectories of resilience, depression, and anxiety following spinal cord injury

    Rehabil. Psychol.

    (2012)
  • G.A. Bonanno et al.

    Prospective patterns of resilience and maladjustment during widowhood

    Psychol. Aging

    (2004)
  • George A. Bonanno

    Loss, trauma, and human resilience: have we underestimated the human capacity to thrive after extremely aversive events?

    Am. Psychol.

    (2004)
  • George A. Bonanno et al.

    Toward an integrative perspective on bereavement

    Psychol. Bull.

    (1999)
  • George A. Bonanno et al.

    Resilience to loss in bereaved spouses, bereaved parents, and bereaved gay men

    J. Personal. Soc. Psychol.

    (2005)
  • George A. Bonanno et al.

    Resilience to loss and chronic grief: a prospective study from preloss to 18-months postloss

    J. Personal. Soc. Psychol.

    (2002)
  • R.A. Bryant et al.

    Trajectory of post-traumatic stress following traumatic injury: 6-year follow-up

    Br. J. Psychiatry

    (2015)
  • C.L. Burton et al.

    Treatment type and demographic characteristics as predictors for cancer adjustment: prospective trajectories of depressive symptoms in a population sample

    Health Psychol.

    (2014)
  • Arin M. Connell et al.

    Family check up effects on adolescent arrest trajectories: variation by developmental subtype

    J. Res. Adolesc.

    (2012)
  • P.J. Curran et al.

    The use of latent trajectory models in psychopathology

    J. Abnorm. Psychol.

    (2003)
  • Kari Dyregrov et al.

    Predictors of psychosocial distress after suicide, SIDS and accidents

    Death Stud.

    (2003)
  • C.K. Enders

    The impact of nonnormality on full information maximum likelihood estimation for structural equation models with missing data

    Psychol. Methods

    (2001)
  • I.R. Galatzer-Levy et al.

    Optimism and death: predicting the course and consequences of depression trajectories in response to heart attack

    Psychol. Sci.

    (2014)
  • Cited by (65)

    • Predictors of prolonged grief in an internet-based grief therapy for people bereaved by suicide

      2022, Journal of Psychiatric Research
      Citation Excerpt :

      Furthermore, analyses did not differ between different kinships to the deceased, which might be due to the low number of participants. As trajectories of depression between child and spousal bereavement have been compared in previous studies (Maccallum et al., 2015), further research on grief trajectories for different kinships is needed. Moreover, variables were investigated in an exploratory manner.

    • Depressive-Symptom Trajectories From End-of-Life Caregiving Through the First 2 Bereavement Years for Family Caregivers of Advanced Cancer Patients

      2021, Journal of Pain and Symptom Management
      Citation Excerpt :

      However, we identified two uncommon trajectories, and the commonly identified trajectories’ prevalences were not totally consistent with the literature in our caregiving-to-bereavement longitudinal study. These two uncommon trajectories (i.e., preloss-grief only and delayed symptomatic) were not observed in studies of bereaved caregivers25,26 or prospective cohort studies of bereavement30−32 with preloss data. The preloss-grief-only trajectory was characterized by slight-to-moderate depressive symptoms only during EOL caregiving, subsiding quickly to near threshold 1-month postloss.

    View all citing articles on Scopus
    View full text