difficulties in data collection, most studies on the symptomatic distress of rescue workers after a major disaster are conducted during the recovery stage, long.
Psychological Distress in Major-Disaster Rescue Workers
ORIGINAL ARTICLES
ASSOCIATION OF PSYCHOLOGICAL DISTRESS WITH PSYCHOLOGICAL FACTORS IN RESCUE WORKERS WITHIN TWO MONTHS AFTER A MAJOR EARTHQUAKE Shih-Cheng Liao,1 Ming-Been Lee,1,2 Yue-Joe Lee,1 Tei Weng,3 Fu-Yung Shih,3 and Matthew HM Ma3
Background and Purpose: Studies of the health of rescue workers after a major disaster have frequently focused on posttraumatic stress disorder. This study aimed to determine the characteristics of psychological distress and its psychosocial predictors in rescue workers within a 2-month period after an earthquake that struck central Taiwan on September 21, 1999. Methods: A total of 1,104 rescue workers serving in the earthquake were enrolled in the study. Psychological distress was measured using the Brief Symptom Rating Scale (BSRS), personality traits using the Maudsley Personality Inventory (MPI), and family function using APGAR (adaptability, partnership, growth, affection, and resolve) indexes. These measurements were performed within 2 months of the earthquake. Univariate and multivariate analyses were applied to examine the association between psychological distress and various psychosocial factors. Results: BSRS assessment revealed severe psychological distress in 137 (16.4%) subjects. The most common symptom dimension was phobic-anxiety (18.7%), followed by hostility (17.6%), obsessive-compulsive symptoms (16.2%), depression (14.9%), paranoid ideation (14.2%), interpersonal sensitivity (13.3%), psychoticism (11.9%), anxiety (10.8%), additional symptoms (8.5%), and somatization (6.2%). Pre-disaster major life events (R2 = 0.03) and most of the factor scores of the MPI (including moodiness, anxiety-prone, outgoing, conscientiousness, activity, and sociability factors; R2 = 0.25) predicted the severity of psychological distress. Time of arrival at the scene, previous exposure, age, and family function had no or trivial predictive power. Conclusion: The results of this study indicated that prevalence of general psychological distress is high among rescue workers in the first 2 months after a major earthquake. Personality traits and pre-disaster life adjustment had a dominant predictive power for psychological distress. Immediate psychosocial intervention should be considered to ameliorate the distress related to disaster rescue work.
In their efforts to help the victims, rescue workers in major disasters are placed at high risk of developing psychological symptoms. Two studies of emergency services personnel responding to the 1989 Loma Prieta earthquake suggest that approximately 9% of workers showed psychological symptoms at the level of those of
(J Formos Med Assoc 2002;101:169–76) Key words: earthquake rescue workers psychological distress psychosocial factors
a psychiatric outpatient population [1, 2]. Previous studies have mainly focused on posttraumatic stress disorder (PTSD) due to its high prevalence rate among rescue workers involved in major disasters [3–5]. A longitudinal study revealed that 21% of firefighters who were exposed to the Ash Wednesday bushfires in
Departments of 1Psychiatry and 3Emergency Medicine, National Taiwan University Hospital, and 2Department of Social Medicine, National Taiwan University Medical College, Taipei. Received: 20 January 2001. Revised: 2 May 2001. Accepted: 6 November 2001. Reprint requests and correspondence to: Professor Ming-Been Lee, Department of Psychiatry, National Taiwan University Hospital, 7 Chung-Shan South Road, Taipei, Taiwan. J Formos Med Assoc 2002 • Vol 101 • No 3
169
S.C. Liao, M.B. Lee, Y.J. Lee, et al
South Australia continued to experience disturbing imagery of the disaster between 4 and 29 months after the disaster [6]. Rescue workers exposed to traumatic death had increased intrusive and avoidant symptoms, hostility, and somatization that could persist for months [7]. Therefore, the identification of mental health problems and prompt intervention in rescue workers after exposure to a major disaster are an important part of effective relief efforts. The risk factors for development of psychosomatic distress following exposure to a major disaster have been extensively examined. McFarlane found that neuroticism and a history of treatment for psychological disorder were better predictors of posttraumatic morbidity than the degree of exposure to the disaster or the losses sustained [8]. These results raise doubts about the postulated central etiologic role a traumatic event plays in the onset of morbidity [6]. Other studies replicated the results that personality traits predicted post-disaster distress in rescue workers [9, 10]. In addition, several other predictors, mainly related to PTSD, in rescue workers were identified, such as social support [9–11], years of experience on the job [9], pathologic identification with disease [12], peritraumatic dissociation [2, 9], intensity of impact or exposure [2, 10], even identity of their profession [4]. In addition, identification of risk factors helps to predict which mental health problems are most likely to occur after disasters and enables the development of intervention programs. Owing to the methodologic limitations and difficulties in data collection, most studies on the symptomatic distress of rescue workers after a major disaster are conducted during the recovery stage, long after the immediate impact. Therefore, the requirements for immediate mental health intervention and the profiles of acute symptomatology are hard to assess. Previous studies have also focused more on PTSD and less on global symptomatology, which may have led to underestimation of the importance of some types of distress not caused by PTSD and the general mental health condition of rescue workers in major disasters. Anxiety, depression, and other psychosomatic symptoms may also be prevalent during the acute stage after a major disaster and necessitate immediate intervention [6]. The present study aimed to determine the profiles of psychological distress in rescue workers within a 2month period after a major earthquake using a series of standardized questionnaires to assess psychological symptoms, personality trait, and family function. The predictive effect of personality traits, family function, and other psychosocial variables on the severity of psychological distress was also examined. The latent dimensions of personality assessment using data reduc-
170
tion methods, which may reveal more detailed content of personality structure, were also examined.
Materials and Methods Subjects A total of 1,104 rescue workers involved in a major earthquake were enrolled in this study. The earthquake, measuring 7.2 on the Richter scale, struck central Taiwan at 1:47 am on September 21, 1999, causing severe damage to property in several counties near the epicenter and resulting in the death of more than 2,300 people. Study subjects were recruited from those enrolled in a collaborative program to survey mental health problems conducted by the Taipei Bureau of Fire Safety and National Taiwan University Hospital. All assessments were completed within 2 months after the earthquake. All subjects were male. Data from the Taipei Bureau of Fire Safety showed an average educational level for our sample. To explore the possible conditioning effect of trauma, experience with previous disaster rescue work (defined as more than 15 casualties) was recorded (yes or no). Major life events defined as death of, or separation from, close relatives, major systemic disease, and problems at work in the past 6 months were also documented. Time of arrival at the disaster site was noted for its possible effect on cognitive rehearsal of the disaster experience.
Psychological symptoms The frequencies and severity of psychological symptoms were measured using the Brief Symptom Rating Scale (BSRS) [13]. The BSRS is a self-reporting measure with 50 items rated from 0 to 4 on the basis of the degree of distress caused over the past week. Each item was scored as 0 = not at all, 1 = a little bit, 2 = moderately, 3 = quite a bit, or 4 = extremely. The BSRS has been shown to be a reliable and valid psychiatric self-rating scale for use in medical practice [14, 15]. It was modified from the widely used Derogatis’ Symptom Check List-90-R (SCL-90-R) [16] and designed to be used as a shorter form. Like SCL-90-R, the BSRS covers nine dimensions of psychopathology: somatization, obsessive-compulsive, interpersonal sensitivity, depression, anxiety, hostility, phobic-anxiety, paranoid ideation, and psychoticism. Additional symptoms included vegetative signs and suicidal ideation. The General Symptom Index (GSI) is essentially a mean score of all BSRS items. The BSRS has been shown in different populations to have an excellent split-half reliability as well as good internal structure J Formos Med Assoc 2002 • Vol 101 • No 3
Psychological Distress in Major-Disaster Rescue Workers
according to factor analysis [17–19]. In addition, BSRS scores are highly correlated with the parental form of SCL-90-R among the medical population for each symptom dimension [13].
Personality traits Personality traits were measured using a short-form of the Maudsley Personality Inventory (MPI), a 26-item self-reporting test to measure subscales of neuroticismstability and extroversion-introversion [20]. Each item was rated as 0 = no, 1 = uncertain, or 2 = yes. The MPI was originally developed under the personality construct proposed by Hans Eysenck, who hypothesized that individual differences in the dimension of introversion-extraversion, neuroticism-stability, or psychoticism determined the ease with which individuals could acquire conditioned responses or be susceptible to stress, which in turn determined the form of psychopathology to which they were prone [20]. Its psychometric properties and time stability have been demonstrated elsewhere [21], and it has been used widely in both community and medical settings in Taiwan since the 1970s [21, 22]. The short-form MPI used in this study included 14 items of neuroticismstability and 12 items of extroversion-introversion dimensions. The items of psychoticism and lie scale were excluded.
Family function The family APGAR (adaptability, partnership, growth, affection, and resolve) index, which contains five structured questions about family interaction, assessed family function. The scoring was 0 = seldom, 1 = sometimes, 2 = most of time, or 3 = always. A higher score indicates higher quality of family support [23].
Data analysis All data were analyzed using SPSS for Windows (Version 7.0, SSPS Inc, Chicago, IL, USA). Cases not included in the analysis owing to missing data in any measurement were examined using logistic regression with the forward procedure and Wald test. To examine the severity of psychological symptoms in our sample, an adjusted T score was determined from data obtained from medical inpatients without formal diagnosis of psychiatric disorder based on in-depth interview by senior psychiatrists. Control group data were obtained from a previous study of 1,638 subjects randomly selected from the Psychiatric Outpatient Clinic, the Family Medicine Clinic, and nonpsychiatric medical inpatients [24]. After adjustment, a T score of 50 was considered identical to the mean of the reference group and the standard deviation (SD) was set at 10. One-sample t-test was used to determine statistical J Formos Med Assoc 2002 • Vol 101 • No 3
significance. A p value of less than 0.05 was considered significant. Significant severity was defined as a symptom score greater than the mean score of the reference group by two SDs (> adjusted T score of 70). To explore the latent structure of personality assessment, explorative factor analysis was applied using the principle component method and orthogonal equamax rotation. The criterion of factor extraction was an eigenvalue greater than 1 and the factor scores were estimated by the regression method for further analysis. Since previous studies have found that concurrent psychological symptoms might influence cross-sectional measurement of personality traits [25, 26], we further examined this hypothesis in our samples using a multiple regression model that included 10 symptom dimensions as independent variables and each factor score as a dependent variable. A multiple regression model that employed the enter method was used in this test. The correlation between basic data, personality factor scores on the MPI, and dimensional scores of symptoms in the BSRS were tested using Pearson’s product moment correlation coefficient. The stepwise regression model was used to determine the predictors of psychological symptoms. The independent variables in each regression model included basic data (age, time of arrival at the site, previous exposure to a major disaster, and previous major life events) and factor scores of MPI. The dependent variables were the dimensional scores of the BSRS.
Results Due to scattered missing data from the BSRS and ambiguous responses on the MPI, 268 cases were excluded and 836 cases (75.72%) were included in the analysis. From a comparison of the basic data (Table 1) for the 268 excluded and the 836 included cases, age was the only significant predictor of exclusion in the final regression model (beta, .9571; 95% confidence interval, .9428 to .9716; Cox & Snell R square, .031). This suggested that younger subjects may have been over-represented in our initial sample. The most frequent items for which subjects were in ‘quite a bit of’ or extreme distress were sleep disturbance (9.8%), muscle ache (9.2%), loneliness (8.1%), depressive mood (7.9%), irritability (6.9%), inattentiveness (6.7%), tension (6.2%), and lack of interest (6.0%). ‘Quite a bit of’ or extreme suicidal ideation was noted in 3.0% of subjects. All T scores for the 10 symptom dimensions in BSRS were significantly greater in subjects than in the reference group (p < 0.01).
171
S.C. Liao, M.B. Lee, Y.J. Lee, et al Table 1. Basic data of rescue workers (N = 836) Age Time of arrival at the scene Never 3d Major life event in the past 6 months Yes No Previous rescue work in a major disaster Yes No
28.32 ± 8.06 67 333 123 73 86 154
(8.0%) (39.8%) (14.7%) (8.7%) (10.3%) (18.4%)
148 (17.7%) 688 (82.3%) 440 (52.6%) 396 (47.4%)
In order to determine the percentage of subjects with severe distress in BSRS symptom dimensions, we defined a T score greater than 70 as severely distressing (the probability of identical severity in the reference group was less than 2.5%). The most frequent distressing symptom dimension was phobic-anxiety (T score > 70 in 18.7% of subjects) followed by hostility (17.6%), obsessive-compulsive symptoms (16.2%), depression (14.9%), paranoid ideation (14.2%), interpersonal sensitivity (13.3%), psychoticism (11.9%), anxiety (10.8%), additional symptoms (8.5%), and somatization (6.2%) (Table 2). The Cronbach alpha of each dimension ranged from 0.76 to 0.90, showing acceptable reliability. The rotated solution of the explorative factor analysis for the short-form MPI showed a simple factorial structure (Table 3). For factor loading greater than 0. 4, no double loading or omission of each indicator variable was found. Six factors were extracted and their order of explained variance was as follows: factor 1 = moodiness factor; factor 2 = anxiety-proneness factor; factor 3 = outgoing factor; factor 4 = activity factor;
factor 5 = conscientiousness factor; and factor 6 = sociability factor. Regarding the influence of symptoms of the cross-sectional personality assessment, although the ANOVA test was significant for each model, the R 2 values were minimal: (moodiness factor R2 = 0.095; anxiety-proneness factor R2 = 0.096; outgoing factor R2 = 0.022; activity factor R2 = 0.035; conscientiousness factor R2 = 0.057; sociability factor R2 = 0.022). Pearson’s product moment correlation showed significant correlation between symptom dimensions and personality dimensions (Table 4), except for the outgoing factor, which was only significantly correlated with interpersonal sensitivity, psychoticism, and GSI. None of the symptom dimensions was correlated with time of arrival at the site and the correlations with family APGAR score only had marginal or no significance. Age was only correlated with anxiety, additional symptoms, and GSI. In the stepwise regression models (Table 5), the anxiety-proneness factor was always the first predictor entered into the model except in predicting the depressive dimension. It had the largest predictive power for the variance of symptom dimensions and reflected that the nature of distress after the rescue work came largely from anxiety-related traits. Other personality dimensions also had predictive power for symptom dimensions, except for the outgoing factor, which was removed from the models of phobic-anxiety reaction, hostility, paranoid ideation, and additional symptoms. Previous major life events had a surprisingly higher ranking than some personality predictors, suggesting that pre-disaster life adjustment affected the subjects’ psychological wellbeing after rescue work. Previous exposure to a major disaster only predicted the severity of the dimensions of hostility and obsession, and family APGAR score only predicted obsession and anxiety. Age positively predicted the severity of the symptom dimensions of anxiety, additional symptoms, and somatization. Time of arrival at the scene was removed from all models.
Table 2. Percentile of adjusted T score* and reliability test of Brief Symptom Rating Scale (BSRS) items (N = 836) Adjusted T scores of BSRS†
Percentile PHO Mean SD % with score > 70 Cronbach alpha
HOS
OBS
DEP
PAR
SEN
PSY
57.76 57.50 56.53 54.86 58.29 55.13 57.74 20.64 17.81 15.87 15.94 17.94 13.90 19.86 18.7% 17.6% 16.2% 14.9% 14.2% 13.3% 11.9% .89 .86 .89 .86 .85 .85 .87
ANX
ADD SOM
GSI
51.12 51.98 49.31 56.15 13.11 12.92 10.70 19.64 10.8% 8.5% 6.2% 16.4% .90 .76 .80 .84
*Raw BSRS score adjusted by the means and standard deviations (SDs) derived from physically ill patients [13]. †Sorted by 70% case numbers except general symptom index (GSI). PHO = phobic-anxiety; HOS = hostility; OBS = obsessive-compulsive symptoms; DEP = depressive symptoms; PAR = paranoid tendency; SEN = interpersonal sensitivity; PSY = psychoticism; ANX = anxiety symptoms; ADD = additional symptoms; SOM = somatic complaints.
172
J Formos Med Assoc 2002 • Vol 101 • No 3
Psychological Distress in Major-Disaster Rescue Workers Table 3. Factor loadings of explorative factor analysis* in selected Maudsley Personality Inventory (MPI) items (N = 836) Factors†
MPI Items
MPI02 Fluctuation of mood between depression and happiness MPI09 Frequent ups and downs MPI03 Poor concentration MPI08 Frequently getting depressed MPI06 Easily distracted when talking to others MPI12 Sometimes feeling sad without reason MPI17 Intermittently feeling lonely MPI23 Frequent insomnia resulting from excessive worrying MPI26 Intermittent restlessness MPI24 Bothered by inefficient overthinking MPI20 Frequent fatigue and lack of initiation MPI25 Frequently get angry MPI21 Not self-inhibited to talk in a group MPI10 Not passive at parties MPI18 Not retiring in front of the opposite sex MPI05 Preferring action to planning for actions MPI07 Self-rating as a lively individual MPI01 Happiest when involved in projects calling for rapid action MPI04 Take the initiative to make new friends MPI15 Not too serious towards work MPI14 Frequently bothered by guilty feelings MPI11 Unable to become fully relaxed at a happy social gathering MPI13 Easygoing in social circles MPI16 Enjoy numerous social activities MPI22 Enjoy parties a lot MPI19 Carefree and optimistic Eigenvalue % Of total variance Cumulative % of total variance
1
2
3
4
5
6
.68 .68 .58 .56 .47 .47 .46 .17 .25 .28 .32 .00 –.08 –.09 –.11 –.09 –.15 .33 .07 –.04 .19 .27 –.14 .25 –.15 –.17 5.89 22.65 22.65
.05 .28 .29 .21 .27 .23 .27 .68 .66 .64 .50 .48 -.18 .00 -.08 -.06 -.08 -.08 .08 .11 .35 .10 -.16 -.15 .15 .05 2.34 8.99 31.64
–.20 –.05 –.12 –.06 –.17 –.29 –.23 –.04 –.17 –.19 –.26 –.09 .75 .69 .64 .03 –.04 –.05 .33 –.07 –.08 –.37 .10 .06 .10 –.05 1.43 5.52 37.15
.08 –.06 -.14 -.02 .12 -.04 -.24 -.04 -.03 -.05 -.24 .02 .00 -.02 .16 .68 .60 .60 .54 -.03 .05 .01 .07 .01 .32 .38 1.29 4.98 42.13
.18 .22 .06 .33 .09 .37 .15 .17 .22 .14 .10 .17 –.06 –.16 –.03 –.10 –.17 .03 .03 .82 .55 .43 .38 –.02 –.17 –.14 1.16 4.45 46.57
–.01 –.13 –.03 –.20 –.05 –.10 .14 –.05 –.05 .00 .08 –.20 .15 .17 –.22 .12 .30 –.03 .10 .16 –.12 –.24 .68 .65 .56 .51 1.09 4.21 50.78
*Extraction method: principle component analysis; rotation method: equamax rotation; inclusion criteria: factor eigenvalue > 1; factor loading > 0.4 in bold type. †Factor 1 = moodiness factor; factor 2 = anxiety-proneness factor; factor 3 = outgoing factor; factor 4 = activity factor; factor 5 = conscientiousness factor; factor 6 = sociability factor.
Table 4. Pearson’s product moment correlation coefficients between dimensions of the Maudsley Personality Inventory, APGAR index, age, and adjusted subscores of the Brief Symptoms Rating Scales (BSRS) (N = 836) Dimensions of BSRS
Personality trait Moodiness factor Anxiety-proneness factor Outgoing factor Activity factor Conscientiousness factor Sociability factor Neuroticism scores Extroversion scores APGAR Age Time to arrival
PHO
HOS
OBS
DEP
PAR
SEN
PSY
.15** .26** –.06 -.11** .21** –.15** .36** –.15** –.07* .03 .04
.24** .27** –.06 –.11** .20** -.13** .43** –.15** –.05 .05 .03
.23** .30** –.09 –.16** .20** –.14** .45** –.19** –.06 .05 .03
.27** .21** .26** .28** -.07 -.05 –.17** –.11** .22** .22** –.15** –.16** .45** .42** –.20** –.15** –.12** –.08* .01 .00 .04 .05
.23** .27** –.12** –.17** .20** –.14** .44** –.21** –.08* –.03 .00
.20** .26** –.08* –.12** .24** –.18** .42** –.18** –.08* .01 .04
ANX
ADD
SOM
.22** .22** .19** .30** .31** .29** –.05 –.04 -.06 –.13** –.11** –.14** .23** .22** .17** –.17** –.13** –.12** .45** .43** .39** –.17** –.14** –.17** –.06 -.06 –.07* .09** .09* .10** .04 .04 .02
GSI
.24** .31** –.08* –.15** .23** –.16** .47** –.19** –.08* .06 .04
*p < 0.05; **p
