Discussion
The aim of this study is to describe the psychological long-term effects of major physical trauma and to detect factors associated with mental impairment after 14 years of follow-up. Several authors report a high prevalence of concurrent PTSD and MDD [
29] at 8 months and 4–6 years after trauma [
30,
31]. There are different hypotheses about the relation between both affections. Bleich et al. assume a shared predisposition to PTSD and MDD with the traumatic experience serving as a trigger. They base this theory on the chronological relationship of both disorders and the reports of their patients. Other theories explain the link between PTSD and MDD with the development of MDD as a secondary consequence of PTSD [
32]. Because of the symptom overlap and concomitance, a clear distinction between both disorders can be difficult. In consideration to this, we combined all patients with either a PTSD or a depressive disorder or both into the one group with symptoms. The “healthy” control group consists of individuals without these diagnoses. In the search for recent literature, we mostly find studies referring to either PTSD or depressive disorders with a definitely higher proportion of PTSD studies. As a result, a direct comparison to other studies is not always possible. Also, the time period of observation after physical trauma is much shorter in the majority of the studies. However, we apply literature with similar issues to place our findings into context.
Our main findings can be summarised as follows:
1.
Polytraumatised patients still show symptoms of mental impairment more than 14 years after the initial physical trauma.
2.
Gender, age or injury severity does not affect the psychological outcome in our population.
3.
Only the injury-related variables “GOS < 5” and “need for physical rehabilitation” are associated with mental impairment.
4.
Individuals without mental impairment achieve higher scores in overall rehabilitation and health-related QoL and show more satisfaction with their rehabilitation results.
Contrary to our results and recent literature, women are found to be more vulnerable for psychological impairment such as PTSD after trauma when compared to men [
12,
33,
34]. When comparing different studies with regard to PTSD or depressive disorders, study populations and trauma types have to be considered as relevant and determining parameters [
35,
36].
In general, men are more often confronted with accidents, non-sexual assault, combat and war, while women experience more assaultive violence in every day life [
37]. As a result, researchers observe a greater risk for the development of PTSD in the female gender in the general US [
38]. We assume that women in general tend to seek more psychological support compared to male. As a result, women have a better chance to recover from PTSD or MDD over the years. This may also be a further explanation for the similar gender distribution in our study groups.
As far as age, previous studies indicate opposite results. Soberg et al. described younger age as a predictor for PTSD symptoms in probands 2 years after severe multiple trauma. Others came to similar results in their studies about flood victims and victims of burn injuries [
11,
39]. However, Zhang and Ho demonstrated that older age was a risk factor for PTSD in survivors of an earthquake after 2 months. An epidemiological study in Germany showed a higher prevalence of PTSD among older age groups, which is assumed to be a consequence of World War II (WWII) [
40]. As results may vary from one trauma type and study population to another, a direct comparison between studies is not always possible.
As far as our study, the participants are involved into traffic accidents and non-assaultive events, which may be an explanation for different results compared to other studies. In the review of the available literature, researchers come to the conclusion that women are not at greater risk for the development of PTSD after a motor vehicle accident (MVA) compared to men [
41].
Besides trauma type and study population, the observation period has to be taken into account. The long-term aspect of our study may be another reason for converse findings in comparison to short-term observation studies. Other study groups find that female gender is associated with PTSD after 6 months observation, but showed no difference after 1 year [
10,
42]. Beck et al. confirm these results by finding no sex differences in the rate of chronic PTSD in MVA victims at 6 months or longer after the physical injury. The hypothesis is that there are different mediators of acute and chronic symptoms of PTSD. Factors such as the individual coping strategy and psychosocial resources may be responsible for either maintaining or recovering from chronic PTSD [
12]. Soberg et al. prove that persons using avoidant coping strategies after severe multiple trauma are more vulnerable to PTSD. In contrast to other studies, we excluded patients under the age of 10 from the study population to avoid a possible influence through lack of awareness. We find that there were no significant differences of the mean age in both of our groups.
We find several studies examining the correlation between the severity of injury and the onset of PTSD. Our results are confirmed by several authors [
34,
43‐
48] who found no positive correlation between the incidence of PTSD and a high ISS or NISS. In contrast, the study of Han et al. reported ISS as a strong predictor for the onset of postinjury depression in trauma survivors. Patients with an ISS ≥ 16 demonstrated twice as much depressive symptoms than patients with moderate injuries. Frommberger et al. reported similar findings—higher ISS scores in patients with psychiatric disorders compared to patients without psychiatric disorders following a MVA [
47]. The differences in the results of the studies may be due to different methodologies. The studies vary in approaches for the assessment of diagnoses, recruitment of subjects or the observation period. For example, the study of Han et al. included only children and young adolescents aged 12–19 years, while the follow-up period of 6 months was very short in the study of Frommberger et al. [
47].
Similar to our results, no significant differences in the appearance of PTSD between patients with or without a head injury have been found by other authors [
49,
50]. Other researchers reported that head injury with following amnesia can be protective against the development of mental disorders [
51,
52]. Similar results were obtained by other researchers after an extended period of unconsciousness in study probands [
53]. They observed far more frequent intrusive memories in accident victims who had been conscious. This could also explain the results of Zatzick et al. investigated the influence of severe, moderate and mild TBIs on the prevalence of PTSD [
54]. The findings state that more severe TBIs are associated with a diminished risk of PTSD. In contrast, Chossegros et al. provided a positive correlation between a head injury and subsequent PTSD [
55]. Others find a link between posttraumatic amnesia and PTSD [
56]. We do not have information on the presence of amnesia or the duration of unconsciousness accompanying the TBI in our population or the influence of long-term sedatives during the ICU stay. As a result, further investigations referring to this information are not possible in our study.
With regard to the GOS, the patients of the control group achieved significantly higher values and, therefore, less severe cerebral dysfunction. Levin et al. [
57] reported similar results with more serious cognitive impairment in patients with MDD. A GOS of 3 correlates with severe disability, possibly with no improvement over the years that result in a higher risk for the development of a depressive disorder.
Our findings about the functional outcome and health-related QoL were as anticipated. The control group achieved higher values in the SF-12, which correlates with better health-related QoL. Regarding the HASPOC, higher values correspond with a poorer outcome (e.g. lower values with excellent overall rehabilitation outcome). Our results show a significantly poorer rehabilitation outcome among the mentally impaired probands, regarding both subjective and objective evaluations. Regardless of the method used, several studies observe functional impairments and decreases in health-related QoL in association with PTSD or MDD. A significant association exists between lower scores on the SF-36 physical component summary and the incidence of PTSD and MDD [
58]. Others have reported unfavourable outcomes in QoL with correlations to PTSD, which demonstrates an association between partial PTSD and poor mental health functioning in veterans from Iraq and Afghanistan [
59]. Also, other authors have seen decreases in health-related QoL in PTSD patients 2 years after injury [
60]. Comparable to previous findings, the patient’s subjective opinion toward their rehabilitation result is statistically significant and more dissatisfied among patients in the “impaired” group.
Different conclusions can be drawn about the relationship between the overall rehabilitation outcome (objective and subjective) and potential psychopathological sequelae. One hypothesis is that an unsatisfactory rehabilitation status is a risk factor for the development of a mental disorder, because the patients get reminded daily of the traumatic event causing their disability or dysfunction. These reminders have a negative effect on the emotional state and the potential for immobility may bar the person from taking part in social activities. On the other hand, a mental disorder can have a negative effect on physical functioning leading to psychosomatic disease and a low functional outcome. We think that there is a close connection between the physical and psychological state of health with inevitable mutual interaction.
Both functional rehabilitation and psychosocial support play important roles while recovering from severe trauma. Several authors report positive correlations between low social support and the development or maintenance of PTSD [
61‐
63]. These results are comparable to our findings with the HASPOC—a loss of the partner or friends as well as financial loss after the trauma occurs more often in the sample of patients with a diagnosed mental disorder.
Recruitment and retention are common problems in studies performed over a long-term period. As the study is based on volunteer participation, some kind of a pre-selection of only motivated probands must be taken into consideration. Concerning the subject of this study, several causes might be the reason for non-attendance (e.g. the probands do not want to be reminded of the accident or they are restrained to show their emotions). Probands of the older generation, especially, may have problems admitting and expressing their feelings. Another reason may be the fear of becoming stigmatised with a psychological diagnosis. The questionnaires are marked only with a number for allocation so that there is no direct link to the data of the probands. As a result, a lost questionnaire cannot be used by outsiders to draw any conclusions. Due to the long-term aspect of clinical studies, the possibility of blurred memories (recall bias) or extrusion mechanisms are increased and must be taken into consideration. Another limitation of our study is its successive design. Thus, we cannot provide information about the onset of the examined mental disorders. The only information we can provide is about the present status to the time of the succession study. Finally, we present single centre data that may be biased due to the therapeutic regimen of the institution.
The period of trauma, collecting the data and their interpretation is a potential drawback of this study. Since we report data of patients with treatment algorithms from 1973 to 1990 one might assume that there is a potential bias due to improved therapeutic strategies and concepts of today’s management. However, this is a large follow-up study investigating patients more than 10 years after injury and we thus will always be dealing with treatment algorithms at least on century or—more likely—2 centuries ago. One potential strategy would be a prospectively collected database of patients, e.g. 1, 2, 5 and 10 years after the injury. This would also be interesting in terms of a longitudinal investigation of the outcome.
These facts come also true for the latest PTSD treatment concepts since there have been some reasonable changes over the last decades.
Modern multimodal concepts consist of but are not limited to: treatment on ICU wards, out-patient follow-up and other concepts such as ICU diaries [
64,
65]. In addition, there is evidence for trauma-focused cognitive-behavioural therapy in patients after critical illness or major trauma and ICU treatment [
66]. However, only a limited fraction of patients are seeking psychological support due to various reasons including stigmatisation and embarrassment. Thus, internet-based interventions and protocols were developed over the past years with more than promising results. Internet-based strategies are able to overcome the trouble with stigmatisation and also availability of health care professionals which makes it a valuable option in patients suffering from PTSD [
67]. Upcoming studies also deal with specifications such as internet-based cognitive-behavioural writing therapy [
68].
Some authors and guidelines recommend a phase-based treatment for individuals—although there is promising literature, the phase-based strategy should not delay the onset of therapy [
69].
Altogether, it seems more than likely that an early onset of the specific therapy after ICU treatment or even on the ICU ward is beneficial for the clinical course of the affected patients.
Despite its limitations, several strengths distinguish our work from previous research. For instance, it is characterised by its long-term period (a minimum of 14 years) and sample size (n = 324 probands). To our knowledge, long-term studies of this extent are sparse. Regarding the patient recruitment, we believe that a response rate of 53% is a respectable result and is appropriate for an outcome study with a minimum follow-up of 14 years. We cannot examine the effect of non-attendant patients on the outcomes of this study, as the attendant and non-attendant patients are comparable in demographic data and showed no considerable deviation. We, therefore, believe that the non-attendant patients do not bias the overall outcomes of this study, and the results obtained in this study are reliable.
Authors’ contributions
LF interpretation of data, statistics, writing and revision of the manuscript. CZ, CP analysis and interpretation of the data, statistics, writing and revision of the manuscript. PM interpretation of the data, revision of the manuscript. MW interpretation of the data, revision of the manuscript. BZ, MP, PM acquisition and interpretation of the data, revision of the manuscript. HCP study design, interpretation of the data, revision of the manuscript. CK interpretation of the data, revision of the manuscript. All authors read and approved the final manuscript.