INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The psychosocial consequences of accidental injuries have long been insufficiently recognized. Over the past few years, however, a number of studies have been published, reporting prevalence rates of posttraumatic stress disorder (PTSD) in accident survivors of up to 39% (1). Apparently, accident victims not only develop PTSD, the most specific trauma-related psychiatric disorder, but also psychiatric symptoms associated with depression and anxiety (5, 6).

Injury severity in general does not predict posttraumatic psychiatric morbidity (4, 7), while the presence of a head injury has been reported to be a protective factor against (8), as well as a risk factor for, the development of PTSD (9). Pretraumatic variables such as a low educational level (4) or preexisting financial problems and stressful life events (7) seem to predict posttraumatic psychological problems, whereas data on the role of pretraumatic psychopathology including personality disorders and alcohol abuse are still controversial (1, 8). The subjectively perceived threat (10) and early occurrence of posttraumatic psychopathological symptoms (11), however, apparently are important risk factors for the development of PTSD in accident victims.

Variability of psychiatric outcomes following motor vehicle accidents may be due, at least in part, to methodological inadequacies, particularly the use of biased population samples. Moreover, most of the samples investigated to date have consisted of a mixture of slightly to severely injured accident victims. Finally, in many studies, the objective somatic accident and injury characteristics were relatively imprecisely assessed (12). A homogeneous sample of exclusively severely injured patients has, to our knowledge, never been explored. We therefore conducted a longitudinal, 1-yr follow-up study of critically ill accident victims who had sustained severe, mostly life-threatening accidental injuries. In an earlier publication, we presented the acute psychological reactions in this sample shortly after the accident (13). The aim of the present study is to report on the prevalence of psychiatric morbidity (posttraumatic stress disorder, depression, anxiety) 1 yr posttrauma in the same sample of severely injured accident victims. In addition, we were interested in identifying variables that predict psychiatric morbidity at 12 mo follow-up.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

To be included, patients had to have sustained accidental injuries requiring their referral to the intensive care unit (ICU) of the traumatology department at the University Hospital of Zurich. Patients who sustained their injuries as a result of a suicide attempt or from a physical attack were not included in the screening. An Injury Severity Score (ISS) (14) of 10 or more and a Glasgow Coma Scale (GCS) (15) score of 9 or more were required, thus excluding all patients with severe head injuries. Patients had to be 18-70 yr of age and capable with regard to both their clinical condition and fluency in German to take part in an extensive interview within 1 mo of the accident. Further details of the study design are described elsewhere (13, 16).

During a recruitment period of 18 mo, all ICU patients were consecutively screened: 135 subjects were eligible for the study. After the study was completely described, written informed consent was obtained from 121 patients; 14 (10.4%) refused to participate. The mean length of stay at the ICU was 5.7 d (SD 5.1, range 1-26). The mean number of days between accident and interview was 13.7 (SD 6.8, range 3-29 d). Follow-up interviews were performed 12 mo ± 3 wk after the trauma. Of 121 patients, 106 participated in the follow-up. Of the 15 dropouts (12.4%), one patient committed suicide, one had returned to her country of origin, and 13 refused to participate in the follow-up.

The patients who refused to participate in the study did not differ significantly from the final sample with regard to sex, age, ISS, and GCS scores. However, significantly more work-related accidents were found among the refusers (refusers: 7; 50%, sample: 13; 12.3%; Fisher exact test, p < 0.01). The 15 dropouts did not differ significantly from the final sample with regard to sociodemographic characteristics, accident-related variables, or measures of postaccident psychopathology.

In the initial semistructured interview, sociodemographic data and information about the accidental event were collected. Biographical protective and risk factors for the development of psychological and psychosomatic disorders were determined by using a compilation of scientifically established items, namely seven protective and 17 risk factors (17). Psychometric instruments included the Impact of Event Scale (IES) (18), the Social Network Index (19), an adapted version of the Social Support Questionnaire (20), and the Inventory for Determining Life-Changing Events (ILE) (21), as well as the Sense of Coherence Questionnaire (SOC) (22) and the Freiburg Questionnaire of Coping with Illness (FQCI) (23). At the end of the interview, patients were asked to make a subjective appraisal of the severity of the accident using a Likert scale ranging from "1 = very slight" to "5 = very severe." To assess psychiatric morbidity at follow-up, the Clinician-Administered PTSD Scale (CAPS-2) (24) and the Hospital Anxiety and Depression Scale (HADS) (25) were administered. (See online data supplement.)

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Sample Characteristics (N = 106)

The mean age of the sample was 37.9 yr (standard deviation 13.1). Seventy-nine patients (74.5%) were males. Forty-four patients (41.5%) were single, 48 (45.3%) were married, and 14 (13.2%) were divorced. Twenty-one patients (19.8%) were living alone, and 83 (80.2%) lived with family members, partners, or friends. One hundred patients (94.3%) were integrated in the occupational life (i.e., working full or part time, studying), whereas only six (5.7%) had no paid work (homemaker, retired, unemployed). Road traffic accidents were most frequent (64 patients; 60.4%), followed by sports and leisure-time accidents (23; 21.7%), accidents in the workplace (13; 12.3%), and household accidents (6; 5.7%). Mean ISS was 21.9 (SD 9.9, range 10-51). Mean GCS score was 14.4 (SD 1.4, range 9-15).

No significant differences in injury severity (ISS) were found between the four types of accident (F = 0.19, df = 3,102, p = 0.90). According to the surgeons' files, 40 patients (37.7%) suffered from retrograde amnesia; 44 patients (41.5%) sustained a traumatic brain injury; that is, they had objectively reported loss of consciousness and/or pathological findings in the cranial CT. A significant association was found between retrograde amnesia and traumatic brain injury (Pearson chi-square = 21.5, df = 1, p < 0.001). Twenty-six patients (24.5%) reported they had experienced the accident as life threatening.

Psychiatric Morbidity

One year after the accident, two patients (1.9%) met all criteria for posttraumatic stress disorder. Patients were diagnosed with "subsyndromal PTSD" if they met the symptomatic criteria for criterion B (reexperiencing cluster) plus either C (avoidance cluster) or D (hyperarousal cluster), but not C and D. Thirteen patients (12.3%) had subsyndromal PTSD 12 mo postaccident.

Psychiatric diagnoses cannot be made based on a self-rating questionnaire such as the Hospital Anxiety and Depression Scale HADS alone. However, the original literature of the HADS (25) offers cut-off values for the two subscales, allowing for a "possible" (> 7) and "probable" (> 10) diagnosis of a depressive or anxiety disorder. According to this definition, nine patients (8.5%) had a possible diagnosis, and a further nine (8.5%) had a probable diagnosis of anxiety disorder. Three patients (2.8%) had a possible diagnosis and six (5.7%) had a probable diagnosis of depressive disorder at 1-yr follow-up.

In summary, at 1-yr follow-up, fifteen patients (14.2%) had subsyndromal or full-blown PTSD, 18 (17.0%) had a possible or probable diagnosis of anxiety disorder, and 9 (8.5%) had a possible or probable diagnosis of depression. A considerable overlap of these three diagnostic categories was found (see Figure 1): 21 patients (19.8%) had a possible or probable diagnosis of depression and/or anxiety disorder; nine patients (8.5%) had subsyndromal or full-blown PTSD plus clinically relevant symptoms of anxiety and/or depression. Accordingly, the CAPS-2 total score correlated significantly with the HADS Anxiety subscale (Spearman's correlation coefficient, r = 0.61, p < 0.001) and with the HADS Depression subscale (r = 0.51, p < 0.001). Overall, a total of 27 patients (25.5%) suffered from clinically relevant psychopathological symptoms.

View larger version (18K): [in this window] [in a new window]   Figure 1.   Psychiatric comorbidity (subsyndromal or full posttraumatic stress disorder and/or clinically relevant symptoms of anxiety and/or depression) at 1-yr follow-up in 27 of 106 severely injured accident victims. Figures are numbers of patients. PTSD (CAPS-2), subsyndromal or full posttraumatic stress disorder (Clinician-Administered PTSD Scale). Anxiety (HADS), possible or probable anxiety disorder (Hospital Anxiety and Depression Scale). Depression (HADS), possible or probable depressive disorder (Hospital Anxiety and Depression Scale). *HADS data missing.

Prediction of Psychiatric Morbidity at 1-yr Follow-up

For the establishment of a predictive model, the following definition of psychiatric morbidity was used (Tables 1 and 2): Fulfilling the criteria for subsyndromal or full PTSD according to the CAPS-2, and/or a score of > 7 in the HADS Depression and/or Anxiety subscales. According to this definition, 27 patients (25.5%) showed psychiatric morbidity at 1-yr follow-up. A selection of 10 potential predictor variables, all assessed at the initial measurement shortly after the accident, was made based on both "pathogenic" and "salutogenic" considerations (22). ISS was chosen as the only objective accident-related variable. Sex was included because in general PTSD is more likely to develop in females than in males after exposure to a traumatic event (26). Biographical risk factors and stress due to life events were selected as potential pretraumatic risk factors. Furthermore, the patients' subjective view was represented in the model by their appraisals of the severity and threat of the accident. Early posttraumatic psychopathology was entered in the equation using the IES Intrusion subscale; salutogenic aspects were represented by the SOC and the patients' social network. Finally, the FQCI subscale "active, problem-oriented coping" was included because such coping strategies were most frequently used in our sample and also because the literature on the adaptivity of active coping strategies is still controversial (27, 28).

Due to missing data (incomplete HADS: five patients; missing data in the predictor variables: two patients, one of whom belonging to the subgroup showing psychiatric morbidity), only 99 patients could be used for regression analysis. Examination of stochastic independence showed that patients with psychiatric morbidity were not excluded in an unduely high number, as no significant association between psychiatric morbidity and exclusion from regression analysis was found (Fisher exact test, p = 0.47). Logistic regression analysis was carried out, using 1-yr psychiatric morbidity status as the dependent variable, and the predictor variables mentioned above as independent variables. The model allowed an overall correct classification of 83.8% of cases. Of patients with psychiatric morbidity 61.5% were correctly classified (sensitivity), whereas 91.8% of "psychologically healthy" patients were correctly classified (specificity, see Table 2). Two predictor variables, namely biographical risk factors and a sense of death threat, contributed significantly to the predictive model (see Table 1).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The aim of this study was to collect a homogeneous sample of patients with severe, life-threatening injuries, free from psychiatric morbidity attributable to head injuries. The mean ISS (21.9) and GCS (14.4) scores in our sample indicate that this goal was achieved. Furthermore, patients were excluded if they showed any signs of prior psychological problems. The exclusion of patients who had attempted suicide or who had been exposed to a physical assault further contributed to the homogeneity of the sample.

Contrary to other samples that were drawn from accident victims seeking treatment for their posttraumatic psychological problems (1), our sample was collected consecutively, with the ICU of the traumatology department at the University Hospital of Zurich as the single source. In traumatic stress research, it is particularly important to achieve high response rates as reluctance to participate in an interview focusing on the trauma might be a symptom of avoidance and thus indicate the possible presence of PTSD or other trauma-related mental disorders (29). In our study, 10% refused to participate. Compared with the literature, this is an unusually low rate. In all studies on accident victims, if mentioned at all, the refusal rate was substantially higher (1, 3). Moreover, we were able to keep the dropout rate reasonably low, and it can be assumed that the 15 dropouts did not represent a high-risk group with regard to posttraumatic psychiatric morbidity. With regard to sociodemographic variables, our sample can be seen as typical and representative for a trauma surgeons' ICU unit and for accident victims in general. However, generalizability of our data to trauma populations with different socioeconomic, ethnic, and racial components remains unclear.

Taking into account the seriousness of the accidents and related injuries in this sample, the number of patients showing psychiatric morbidity was substantially lower than could have been expected from the current literature. For instance, Blanchard and coworkers (30) found 24 (18.2%) of 132 accident victims suffering from full PTSD, plus 16 patients (12.1%) with subsyndromal PTSD, 12 mo after the trauma. The sample studied by this research group was recruited by referrals through medical practitioners and by "local media coverage and advertising" and comprised 67% females, whereas typical samples of accident victims comprise no more than 20-30% females. More recently, Koren and coworkers (2) reported on 32% of accident victims suffering from PTSD at 1-yr follow-up; unfortunately, the authors did not specify their sampling method. Ursano and coworkers (3) found 14% PTSD in a sample of serious motor vehicle accident victims 1 yr postaccident. However, this sample was recruited from a trauma center and through local police reports with 25-50% refusers and 48% females in the final sample; it is unclear for which population this sample can be regarded as representative.

We think that the relatively low prevalence of psychiatric morbidity in our sample can mainly be explained by the fact that we have investigated an unselected sample recruited consecutively from one single source of referrals. Given the fact that women in general have approximately double the risk of suffering from PTSD (26), the overrepresentation of males (75%) that is typical for accident patient populations certainly helped to keep the level of psychiatric morbidity low. Our findings are in accordance with those of Malt (31) who investigated the only truly randomized sample of accident victims published so far. Twenty-eight months posttrauma, Malt found only one of 107 patients suffering from PTSD, and a total of 9.3% of patients suffering from nonorganic psychiatric disorders caused by the accident (31).

Biographical risk factors for the development of psychological and psychosomatic disorders were found to predict psychiatric morbidity. This is an important finding, particularly because we had excluded all subjects who had shown any signs of pretraumatic pathology. To our knowledge, no comparison data are available in the literature. At least, Ursano and coworkers (3) examined the family history of psychiatric disorder in a sample of motor vehicle accident victims; the authors did not find any family history variables associated with a greater risk of PTSD at follow-up. To our best knowledge, there is currently no standardized research instrument available for the assessment of biographical protective and risk factors. As a consequence, the methodological level of our assessment (sum score of a number of dichotomized variables) was quite low. Therefore, the importance of this variable should not be overestimated. Nevertheless, our results indicate that pretraumatic characteristics may have an influence on the development of PTSD symptoms after a serious accident.

The significant predictive value of patients' sense of threat to their life experienced during the incident lends support to the findings of other authors who found that the subjective appraisal of the trauma was highly predictive for later development of psychological problems including PTSD (7, 32). It should be pointed out that in our study the subjective appraisal variables were unrelated to the ISS. This underlines the importance of the patients' subjective appraisals in the development of posttraumatic psychological problems. Moreover, this finding is in accordance with Lazarus and Folkman's (33) transactional theory of stress and coping that emphasizes the importance of the process of appraisal of a stressful event in the context of the patients' currently available coping strategies. The patients' cognitive and emotional appraisals with regard to the accident they had recently survived are developed through their own perceptions, but also through what they are being told by their relatives, the media, police reports, and health professionals. As a consequence, when talking to their patients, ICU personnel and trauma surgeons should consider not overemphasizing the severity and dangerousness of the accident: this might help prevent patients from developing long-term psychological problems.

Logistic regression analysis allowed a correct classification of a great majority (83.8%) of patients in this sample. However, only two-thirds of patients with psychiatric morbidity were correctly classified (sensitivity 61.5%), thus qualifying the predictive power of this model. Other authors did not find substantially stronger models, though. Shalev and coworkers (4), using PTSD status at 6-mo follow-up as the dependent variable, were able to classify 78.4% of injured trauma survivors correctly, with a sensitivity of 30.8% and a specificity of 94.7%. In the study of Shalev and coworkers, peritraumatic dissociation emerged as the only significant predictor of PTSD status. Koren and coworkers (2) demonstrated in a sample of traffic accident victims that early posttraumatic psychopathology, assessed 1 wk after the trauma, predicted PTSD status at 1-yr follow-up with a sensitivity of 70.8% and a specificity of 71.4%. Finally, Freedman and coworkers (11) recently reported on a logistic regression analysis in accident victims and found early depressive symptoms to be the most powerful predictor of PTSD status at 1-yr follow-up.

This study has a number of limitations that need to be addressed. First, patients had to be excluded from the study if they did not speak German sufficiently. Proficiency in the official language of a language area is a strong determinant of social integration; according to our clinical experience, patients with poor social integration have greater than average difficulties in dealing with the consequences of their accident. Therefore, in future studies, patients whose mother tongue is other than the language area's official language should be included using interpreters. Furthermore, the missing correlation between injury severity and psychiatric morbidity may be at least in part due to a statistically restricted range phenomenon: it is possible that in a study including mild, moderate, and severe injuries, thus covering the full range of ISS values (1-75), the ISS would become a significant predictor of psychiatric morbidity. Finally, as patients with work-related accidents more frequently refused participation and tended to show more posttraumatic stress symptoms, they should be studied more thoroughly. The small number of patients with work-related accidents did not allow us to draw any firm conclusions.

In summary, the incidence of psychiatric morbidity in this sample of severely injured accident victims during the first year posttrauma was somewhat lower than previously reported in the literature. Yet a substantial proportion of patients developed some form of psychiatric morbidity. A combination of mainly psychosocial variables (biographical characteristics and the patient's early cognitive appraisal of the accident in particular) allowed for a significant prediction of psychiatric morbidity at 1-yr follow-up.