Main findings

This study found five different trajectories of PTSD response over the 6 years after injury. Consistent with previous studies, we observed a chronic trajectory that maintained elevated PTSD symptoms over 6 years, a resilient group that had very few symptoms at any point, a recovery trajectory that showed initial symptoms that eased over time, and a delayed or worsening trajectory. In addition, and unlike previous studies, we observed a group of patients who worsened over time but then recovered (worsening/recovery).

The most concerning finding of this study is that a small, but significant, proportion of trauma survivors display persistently chronic stress reactions. The chronic class showed no signs of remitting, even after 6 years, and suggests that they are embarking on a very long-term course of PTSD. Worryingly, this group had more overall psychological impairment, as measured by the WHOQOL, than any other group at each assessment. This pattern suggests that this subgroup of injury survivors is likely to experience long-term impairment. The other curious finding, in those with a recovery trajectory, was that they had comparably impaired functioning relative to those in the chronic class. This points to the possibility that having experienced more severe PTSD this left residual impairments despite the remission of symptoms; this interpretation is consistent with findings that comparable levels of impairment are found in people with full and subsyndromal PTSD. ^{Reference Stein, Walker, Hazen and Forde4,Reference Zlotnick, Franklin and Zimmerman23}

Comparison with findings from other studies

The most novel aspect of this study was it followed up trauma survivors at least 3 years longer than previous studies. By adopting this approach, we were able to demonstrate that nearly one-tenth of the cohort who appeared to be moving on a trajectory towards worsening PTSD 2 years after the injury, subsequently recovered by 6 years. This suggests that it will take some trauma survivors considerably longer to adapt to the injury than others, even though it appears in the initial years that they are on a worsening trajectory. This observation contrasts with previous studies that have ceased assessing patients at shorter time frames since the trauma. Whereas these studies conclude that a subgroup of trauma survivors will be on a course of worsening symptoms, ^{Reference deRoon-Cassini, Mancini, Rusch and Bonanno9,Reference Bonanno, Kennedy, Galatzer-Levy, Lude and Elfstrom24} our longer-term follow-up suggests that these short-term follow-ups portray an incomplete profile of adaptation because many of these patients who are initially worsening over time do eventually recover.

Identifying those most at risk of long-term mental health problems

One of the challenges raised by the current findings is how early detection can be achieved in the acute phase after traumatic injury to identify the subsequent trajectory that patients will follow. Figure 1 highlights that in hospital there were essentially two groups of patients: those with elevated PTSD levels (including those belonging to the chronic and recovery classes) and those with low PTSD levels (including those belonging to the resilient, worsening and worsening/recovery classes). The current findings underscore how mental health outcomes cannot necessarily be predicted on the basis of severity of injury, as injury severity score was not a consistent predictor of subsequent outcome. Prior attempts to identify longer-term adaptation after trauma have been modest. ^{Reference Bryant25} However, admission to ICU did differentiate the chronic from resilient class, suggesting that ICU admission may be one potential marker of those who are more at risk of long-term PTSD symptoms. The current findings highlight that a major reason for this difficulty in accurate prediction of outcome of psychological state is that there is not a linear relationship between acute and longer-term responses.

In terms of predictors of trajectory classes, mTBI may contribute to worse outcomes following trauma for several possible reasons. The prevailing models of PTSD postulate that the disorder is maintained because of impaired regulation of the amygdala by the ventral medial prefrontal cortex. ^{Reference Rauch, Shin and Phelps26} The prefrontal cortex is commonly impaired in mTBI and so this damage may compromise the capacity of patients to manage the emotional sequelae and cognitive demands of the traumatic experience. ^{Reference Bryant27} This finding suggests that patients with mTBI should be considered at higher risk for one of the trajectories that will lead to longer-term psychological problems following traumatic injury.

The finding that stressful life events since the trauma characterised poor adjustment at each assessment accords with evidence that post-trauma stressors are associated with PTSD. ^{Reference Smid, van der Velden, Lensvelt-Mulders, Knipscheer, Gersons and Kleber28} Fear-reinstatement models explain this finding in terms of subsequent stressors reactivating previously conditioned trauma responses. ^{Reference Norrholm, Jovanovic, Vervliet, Myers, Davis and Rothbaum29} Sensitisation models posit that trauma can sensitise neural circuitry, which then leads to excessive stress responses to aversive events. ^{Reference Post, Weiss, Smith, Friedman, Charney and Deutch30} Further, resource models account for this pattern in terms of additional stressors diminishing limited resources following trauma, and this compounds the PTSD reaction. ^{Reference Hobfoll, Tracy and Galea31} In any case, these findings highlight that the trajectory one follows after trauma will be influenced by ongoing stressors.

The pattern of female gender being associated with more distressed trajectory classes is consistent with evidence that females are markedly more at risk of PTSD than males. ^{Reference Brewin, Andrews and Valentine22} One curious finding was that female gender predicted membership of the resilient class relative to the worsening/recovery class. The worsening/recovery class, which has not been previously identified in trajectory analyses, reflects a course in which there is a time-limited period of worsening distress followed by good mental health. This trajectory is poorly understood and it is difficult to understand why females are more likely to be included in the resilient class than in the class that is worsening and then recovering. It is possible that worsening symptoms followed by recovery reflects a distinctive pathway that is less susceptible to the expected influences of gender. This course requires further study to understand the predictors of this long-term recovery. In this context, we also note that it is possible that some traumatic injury patients may have a worsening trajectory during the protracted period of physical injury, pain and rehabilitation, followed by remission of symptoms as they progress beyond this period.

Three of the trajectories were characterised by a worsening of symptoms over the initial year after traumatic injury: the chronic, worsening, and worsening/recovery classes. This observation is consistent with sensitisation models of PTSD, which propose a vulnerability to excessive responses to less stressful events following traumatic exposure because neural circuitry is sensitised by the initial traumatic experience. ^{Reference McFarlane32} This model accords with evidence that prior trauma is linked to more reactive responses to subsequent stressors. ^{Reference Breslau, Davis and Andreski33} There is also evidence that trauma survivors who subsequently develop PTSD do not display elevated startle responses immediately after the traumatic event but do so after several months have elapsed. ^{Reference Shalev, Peri, Brandes, Freedman, Orr and Pitman34} It is possible that the psychological effects of the traumatic injury may not be fully experienced by survivors in the initial period, however, the subsequent stressors that are often faced by injury survivors may compound initial reactions via sensitisation processes, thereby leading to more severe PTSD responses in the following year. This conclusion remains speculative because to address this issue directly, studies need to model ongoing stressors in terms of how people can transfer between classes over time (via latent transition analysis).

Limitations

We recognise several limitations. First, nearly half the sample was lost by the 6-year assessment, and these participants were younger and had more severe initial PTSD at baseline than those who were retained; this differential drop-out may have affected results. Second, the cohort comprised injury survivors who had been admitted to hospital, and so the findings need to be confirmed in samples exposed to different types of trauma. Third, we assessed PTSD at each follow-up by structured clinical interview via the telephone rather than face-to-face interview; we note, however, that there is very strong concordance between these two formats. ^{Reference Aziz and Kenford15} Fourth, this was a largely male sample and these results may not generalise to females. Fifth, future research could usefully index utilisation of general health services to map how people in different trajectories access services, including profiling of health economics associated with each class.

Implications

The major outcome of this study is that it highlights the long-term psychological dysfunction and associated impairment, experienced by a proportion of traumatically injured patients. One implication of the study is that by identifying the majority of those who are resilient, primary care facilities may be able to triage the remainder of patients to a category that may benefit from either early intervention or subsequent monitoring to index their mental health. Further work is required, however, to develop the means to discriminate between injured patients in the hospital setting who are most likely to enjoy a resilient trajectory and those who at some time may benefit from mental health assistance.