Risk Factors for Bereavement Outcome: A Multivariate Approach

Abstract

Bereavement increases the risk of ill health, but only a minority of bereaved suffers lasting health impairment. Because only this group is likely to profit from bereavement intervention, early identification is important. Previous research is limited, because of cross sectional designs, small numbers of risk factors, and use of a single measure of bereavement outcome. Our longitudinal study avoids these pitfalls by examining the impact of a large set of potential risk factors on grief, depressive symptoms, emotional loneliness, and positive mood following recent bereavement (3 years maximum). Participants provided information 3 times over 6 months. A multivariate approach was chosen to avoid reporting spurious results due to confounding. As expected, risk factors were differentially related to different outcome measures. For example, being high in anxious attachment and having lost a partner were related to more intense feelings of emotional loneliness, whereas these variables did not predict any of the other outcome variables. By contrast, social support did not influence emotional loneliness but did predict grief, depressive symptoms and positive mood. Implications of these findings are discussed.

Karolijne van der Houwen is now affiliated with Statistics Netherlands.

Notes

¹Participants assigned to the control condition were offered the opportunity to participate in the intervention after answering the last set of questionnaires following the end of their participation in the study.

²Social and environmental risk factors were combined into one predictor group to be in line with the risk factor framework developed by M. Stroebe et al. (2006) and because of overlapping variance between the two categories (e.g., between “being a practicing member of an organized religion” and “social support” and between “current living arrangements” and “current financial situation”).

³Time × Time was added as a quadratic predictor to capture any non-linear relationship that might exist between time and the dependent variables and thereby improve on the model.

^a (0 = partner).

^b (0 = natural causes).

∗p < .05. ∗∗p < .01. ∗∗∗p < .001.

∗p < .05. ∗∗p < .01. ∗∗∗p < .001.

∗p < .05. ∗∗p < .01. ∗∗∗p < .001.

†p < .10. ∗p < .05. ∗∗p < .01. ∗∗∗p < .001.

⁴The explained variance of the three predictor groups combined (shown in the last column of Table 2) is less than the sum of the explained variance of the three predictor groups separately (shown in the first three columns of Table 2). This is due to dependencies between predictors (i.e., variance that is shared by the predictor groups).

⁵The relationship between grief and financial situation deterioration almost reached significance (p = .051).

⁶In considering the intrapersonal and social and environmental predictors, it is important to keep in mind that this study focused on bereaved persons only and the differences between subgroups among them. Given this focus, we cannot be sure whether the variables that turned out to be important significant predictors of the general (i.e., non-grief specific) outcome measures would also be significant predictors among non-bereaved samples of people. Clearly, this point does not apply to our grief measure.