How to quantify exposure to traumatic stress? Reliability and predictive validity of measures for cumulative trauma exposure in a post-conflict population

Figures & data

Table 1 Demographic and clinical information by diagnostic group

	Current PTSD (N=19)		Remitted PTSD (N=18)		Never PTSD (N=19)	Statistica	p
N female (%)	9 (47)		9 (50)		10 (53)	Fisher's exact test	1.00
Mean age (SD)	34 (8.88)		35.56 (12.20)		33.95 (10.44)	F 2.53=0.14	0.872
Mean number of school years (SD)	5.37 (2.52)		6.28 (2.91)		5.87 (3.70)	F 2.53=0.40	0.670
Mean number of event types lifetime (SD)	37.58 (8.75)b	>	28.72 (6.05)	=	22.05 (8.13)	H 2=24.01	<0.001
Mean experienced events frequency score lifetime (SD)	79.00 (21.12)b	>	46.78 (14.12)	=	39.37 (21.74)	F 2.53=22.35	<0.001
Mean number of event types experienced under the age of 6 (SD)	0.05 (0.23)	=	0.28 (0.57)	=	0.00 (0.00)	H 2=6.05	0.049
Mean number of event types experienced under the age of 14 (SD)	4.26 (8.88)		6.22 (8.52)		3.32 (4.63)	H 2=0.54	0.762
Mean number of event types experienced as an adult (SD)	33.74 (9.91)b	>	23.83 (8.38)	=	19.16 (8.21)	F 2.53=13.35	<0.001
Mean PDS score (SD)	14.89 (4.99)b	>	1.67 (1.68)	=	1.21 (1.99)	H 2=38.67	<0.001

PDS, Posttraumatic Diagnostic Scale.

a ANOVA F-test for continuous data if test residuals were normally distributed, Kruskal–Wallis H-test for continuous data if residuals were not normally distributed, and Fisher's exact test for categorical data.

Global comparisons of the means of continuous variables for the three groups were followed by parametric or non-parametric post hoc tests, if the F-test or Kruskal–Wallis H-test was significant. The results of the post hoc tests are visualized by the symbols >, <, and =.

b Indicates a significant difference between the current PTSD and never PTSD group.

Fig. 1 Test–retest reliability of trauma exposure assessed as the number of different traumatic event types experienced (left) and assessed as the experienced traumatic events frequency score (right). The line represents the linear relationship (estimated by fitting an ordinary least square regression) between the two repeated measurements across the entire reliability sample.

Table 2 Goodness of fit statistics for models including different trauma measures as predictors of lifetime PTSD, current PTSD, and current PTSD symptom severity

	AIC	D	Nagelkerke's R ^2
Prediction of lifetime PTSD
Number of traumatic event types experienced	230.15	0.19	0.25
Experienced traumatic events frequency score	242.22	0.13	0.19
Number of experienced traumatic event types under the age of 14	270.42	0.01	0.02
Number of experienced traumatic event types as an adult	252.06	0.09	0.13
Prediction of current PTSD
Number of traumatic event types experienced	193.09	0.22	0.30
Experienced traumatic events frequency score	186.91	0.25	0.34
Number of experienced traumatic event types under the age of 14	238.92	0.02	0.03
Number of experienced traumatic event types as an adult	225.79	0.08	0.11
Prediction of current PTSD symptom severity
Number of traumatic event types experienced	1103.87	–	0.31
Experienced traumatic events frequency score	1100.28	–	0.33
Number of experienced traumatic event types under the age of 14	1148.89	–	0.04
Number of experienced traumatic event types as an adult	1141.75	–	0.08

AIC, Akaike's Information Criterion; D, Coefficient of discrimination.

Displayed is the goodness of fit statistic from regression models with the different trauma measurements as predictors. For the prediction of lifetime and current PTSD, logistic regression models were fitted, whereas a negative binomial regression with cubic splines was estimated for the prediction of current PTSD symptom severity.

Fig. 2 Left panel: The probability of lifetime PTSD is best predicted by a trauma assessment which measures the number of different traumatic event types experienced. Depicted are the observed and predicted proportions of lifetime PTSD against the number of traumatic event types, with 95% bootstrapped confidence intervals of the prediction. Right panel: By contrast, the probability of current PTSD is best predicted by a trauma assessment which considers the frequency of the traumatic events experienced. Depicted are the observed and predicted proportions of current PTSD against the experienced traumatic events frequency score, with 95% bootstrapped confidence intervals of the prediction. For this graphical illustration, data on trauma exposure was aggregated in groups of ≥9 individuals per group in order to be able to calculate meaningful proportions.