Predictors of Youths’ Posttraumatic Stress Symptoms Following a Natural Disaster: The 2010 Nashville, Tennessee, Flood

Abstract

Framed by a previously established conceptual model of youths’ posttraumatic stress (PTS) responses following a disaster, the current longitudinal study examined the relation of predisaster child characteristics (age, gender, depressive symptoms, ruminative coping), predisaster environmental characteristics (negative life events and supportive and negative friendship interactions), and level of disaster exposure to youths’ PTS symptoms in the wake of a natural disaster. Prior to the 2010 Nashville, Tennessee, flood, 239 predominantly Caucasian youth from four elementary and middle schools (ages = 10–15, 56% girls) completed measures of depressive symptoms, rumination, negative life events, and social support in the form of both supportive and negative friendship interactions. Approximately 10 days after returning to school, 125 completed measures of disaster exposure and postflood PTS symptoms. Bivariate correlations revealed that disaster-related PTS symptoms were unrelated to age, gender, or predisaster supportive friendship interactions and significantly positively related to level of disaster exposure and predisaster levels of negative life events, depressive symptoms, rumination, and negative friendship interactions. After controlling for level of disaster exposure and other predisaster child and environmental characteristics, depressive symptoms and negative friendship interactions predicted postdisaster PTS symptoms. The effect of child's flood-related experiences on PTS symptoms was not moderated by any of the preexisting child characteristics or environmental indicators. Faced with limited resources after a natural disaster, school counselors and other health professionals should focus special attention on youths who experienced high levels of disaster-related losses and whose predisaster emotional and interpersonal lives were problematic.

Notes

¹We conducted all analyses twice, once with and once without these two outliers. When included, no variables emerged as significant that had not been significant using the reduced analytic sample (without the two boys). In addition, youths’ flood experience remained a significant predictor and of similar magnitude (β = .40, p < .0001). The effect of depressive symptoms, however, although remaining significant, increased in magnitude (β = .45, p < .0001), reflecting the fact the both boys were also outliers on the CDI, one whose score was almost 2 standard deviations above the mean CDI score for the sample and the other whose CDI score was more than 5 standard deviations above the mean. Negative social interactions, however, was only marginally significant when including these two cases (β = .14, p = .085). Results reported here are without these two boys, given their unusually extreme characteristics.

²Regression models were fit using SAS, v.9.3. To use Full Information Maximum Likelihood estimation to handle missing data, models were also fit using AMOS, v.20. Results were similar to those in SAS. SAS results are reported here.