CHAPTER 3: LITERATURE REVIEW
3.7 Conclusions and Implications for Current Study
The literature review identified successfully, a large number of studies in depression after TBI. This should not be surprising in itself as both depression and TBI are among the most common of medical conditions and therefore the overlap is likely to be substantial.
However one may express some surprise at the lack of consistency in findings within this body of literature. There is no clear agreement around prevalence of depression post-TBI although it is
clearly raised. Furthermore there is no clear idea of the risk factors that are associated with
depression after TBI. This is due to the large variation between studies in terms of inclusion criteria and study quality including sampling biases, variable success in follow-up and difference in
measurement tools for depression.
There is therefore considerable scope for new studies to look at this prevalence, preferably in a large, well-designed, prospective study that includes all severity of TBI and the full spectrum of ages. There are some pointers that may help in this design from the body of previous work.
A major source of variation in depression levels is the difference in TBI severity that studies have looked at. Many took only severe or moderate injuries and those that looked at mild TBI often only considered those with abnormal CT scans or “complicated MTBI”. Some studies exclude individuals above a certain age e.g. 60 and many seem to have very young average ages within the group. For a study to truly reflect the general population with TBI, it is important to consider a population across all severities including normal CT scans and in all adult age groups. In this way a study may be
considered to be a “real-life” study. There will still be many individuals who do not present with their TBI but these will be the mildest of cases and it is difficult to locate such cases unless they present to a health service.
Selection of patients within studies also varied considerably. The best studies do not use selected referrals or a “convenience sample” but rather use consecutive, unselected admissions to a service ideally via ED departments. Exclusion criteria need to be minimal to avoid selection bias and the best studies explain clearly the number of cases that are either screened out or are lost at follow-up. In this way it is possible to directly compare the “included” and “excluded” populations to determine any clear differences that affect conclusions from the study. Some studies excluded any individuals with previous psychiatric problems or any medical comorbidity.
Most studies are single centre. The use of multi-centre studies is often considered a better design as it will reduce the bias of a single centre that may have a skewed population for one reason or another. The TBI Model Systems in the USA is an excellent example of such studies and two high
quality publications used data from 17 and 19 centres. These projects benefit from considerable resources especially staff. In an ideal world, it would be possible to extend the number of cases, centres and assessments that could be done but there is always a balance to the funding and resource available for a piece of work. In this study there is one researcher responsible for all of the assessments along with some support to chase up clinic attendance. It is not possible therefore to have multiple centres but one may aspire to such resources in future.
The source of TBI cases is important to consider. Again there are studies that draw upon referrals to a neuropsychology service or only consider patients that spend considerable time in a
neurorehabilitation service. One study contained 96% of cases caused by RTC and another 100% [Van Reekum 1996, Draper 2007]. This will limit the types of cases seen and not truly reflect the population of individuals that sustain TBI. Those studies that draw from the general population attending e.g. Trauma centres or ED, are much more likely to represent the population with TBI who present with the problem. A large ED with a Trauma Centre and serving a large population would seem ideal to draw a representative sample of adult TBI cases.
There is also considerable variation between studies in the timing of the assessments of outcomes such as depression. Some studies have looked at very long time spans since injury and are very revealing in their conclusions that depression is still elevated up to 50 years later. Some studies have taken repeated measures at time points e.g. over 2 years and they provide valuable insight as to how the levels of depression remain high for many years after TBI and in most cases, for life [Bombardier 2010]. However within individual studies, it is easier to draw conclusions if the majority of cases are at a similar time course in the evolution of their condition e.g. within the first year. In some studies, cases were between 3 months and 30 years elapsed since injury which makes it more difficult to make sense of results given the patient heterogeneity.
As we know that the history of TBI symptoms changes with time, it would be useful to try and design a study that takes cases along a set point in their evolution. This is most easily done in the first year since injury when cases can be picked up and then followed through. A longer elapsed time period
also affects the ability of an individual to recall events and medical records, if available, may be patchy as is described by some of the studies following up patients many years after the TBI. It is well known that TBI studies in particular, suffer from high attrition rates when it comes to follow-up [Corrigan 2003]. In some of these studies, more than 70% of cases were lost by time of follow-up [Macniven 1993, Hawley 2008]. Even short term studies lost 45% of cases within 12 weeks [Levin 2005].
This presents us with a challenge to ensure high level of follow-up in any study and to organise a system to chase follow-up; in particular, those who miss an assessment would have to be quickly chased up to try and facilitate a replacement appointment quickly. Despite the high loss of cases in many prospective studies, it is refreshing to see that one study managed 100% follow up at 1 year and others of over 80% after similar times [Gould 2011, Sigurdardottir 2013]. This suggests that with enough time and persistence, it should be possible to organise high level of follow-up.
It still remains unclear from the literature, which associated demographic and injury characteristics are most linked to depression. Most studies have looked at a few features but it is difficult to draw any firm conclusions as there is little agreement e.g. there are studies that find no effect of TBI severity, some that find more depression in severe injury and some find more in mild injury. This is reflected in most other features too e.g. some studies find an association between previous psychiatric history and depression and others find an inverse relationship. For the purposes of this study, it would seem appropriate to measure those features that are more readily identifiable such as individual demographics as well as key injury features such as TBI severity, mechanism of injury, associated injuries or influence of comorbidities or alcohol intoxication. Most studies seem to have one key extra feature that they measure which can be a complex composite function requiring a detailed assessment of its own e.g. functional outcome score, social isolation, fatigue or any number of psychological and cognitive scores. Most of these take considerable time to assess and it is clear that incorporating any such feature would require considerable extra time to assess and document. Therefore care is needed in choosing the items to study. As discussed in methods, the study
recorded a key global outcome measure (the Extended Glasgow Outcome Score), a measure of participation restriction (Rivermead Head Injury Follow-up Questionnaire) and the Rivermead Postconcussion Score to evaluate symptom severity. Along with a number of key injury and individual characteristics, these formed the basis of the features studied for association with depression.
In summary, it is apparent that there is no consistency across studies with regards to the prevalence of depression and no agreement on the injury and population features that may be associated with risk of depression after TBI. There is therefore a clear niche for a large, well-designed, prospective study that is representative of all adult ages and severities of injury, ideally within the same time span since injury.