Several limitations need to be considered when interpreting the results of the study. Firstly, although common the sample size was small, possibly effecting statistical power and limiting generalisations. However, after reviewing the TBI caregiving literature, it appeared that small samples of less than 40 participants are common (Hickey et al., 1997; Katz et al., 2005; Leathem et al., 1996; Moules & Chandler, 1999; Nabors, Seacat, & Rosenthal, 2002; Norup, Siert, & Mortensen, 2010). Reasons for the small sample size possibly include that data were only collected through agencies supporting TBI families. Using other ways of recruiting participants, for example through the internet were explored but did not result in any viable options. In addition, case managers within the agencies contacted, even though very supportive, might not have had enough time available for the distribution of all information packs to potential participants. In addition, some potential participants declined taking part after deciding that the time it took to complete the questionnaire was more than they could spare. This could have also resulted in a sample of caregivers that was less stressed and not representative for this population.
A second limitation is a possible sample bias as caregivers were recruited via agencies. It is possible, that caregivers who were more distressed sought support from the organisations, than those caregivers that coped well. It could however also be argued that well adjusted caregivers sought assistance as the realisation that they required help was a sign of strength. In addition, some of the participants were attending support groups, while others were not. This could have had an impact on their ability to cope, levels of psychological distress, QoL and possibly posed a threat to the external validity. Therefore, the findings cannot be generalised to caregivers that are not connected to a support agency.
A third limitation is the use of a cross-sectional, retrospective design. Even though time since injury was found to be related to coping strategies used, these findings were not strictly applicable to one time period. As in the current sample injury occurred
117 between less than one year and 31 years earlier, the results can only be generalised across time. In addition, the retrospective design might have resulted in problems concerning recall of emotions or coping strategies that participants used at the time the stressful situation occurred. However, a benefit of a retrospective design is that it allows for data collection in a relatively short amount of time, as the collection of prospective data would take much longer.
Fourthly, the results reported are correlational and do not imply causality, requiring further investigation. For example, it could not be determined whether the use of emotion-focused coping led to higher levels of psychological distress or whether more distressed caregivers used this style of coping more frequently. There is evidence that depressed individuals have a tendency to use maladaptive coping strategies such as escape-avoidance (Billings & Moos, 1984)
A fifth limitation is the self-report nature of the data collection, which might have introduced a response bias. The emotional state of the caregivers might have influenced the levels of psychological distress that were reported. Underreporting of symptoms could have occurred as an attempt to come across as coping better than was actually the case. For example, the suicidal participant indicated having thoughts about suicide but would not carry them out. This underlines the importance of including interviews or other qualitative information to achieve a more accurate assessment of psychological distress. The use of more than one method within a study has been shown to give a more complete representation of a person’s experience and behaviour (Morse, 2003)
Previous research has shown that demographic variables can have an impact on coping, psychological distress, and QoL. However, the uneven distribution of the data in regards to race, gender, marital status, kinship, severity of injury, and age did not allow for an investigation of those variables and the impact they could possibly have had on the outcomes. The results can therefore only be generalised to female, married, European New Zealander TBI caregivers caring for a severely injured family member. As severity of injury has been found to have an impact on the caregiver, the fact that the majority of TBI individuals had a severe injury could have resulted in a more stressed sample of participants (Biegel, Sales, & Schulz, 1991; Livingston et al., 1985b)
In addition, no demographic characteristics were available for the potential participants that declined to take part in the study. The possibility that the responses of the potential participants could have differed from those that took part in the research
118 therefore existed. The absence of a control group did not allow for a comparison of coping strategies, psychological distress, and QoL to either caregivers in other populations or non-caregivers.
Comparison with previous research had to be made with caution as studies have used a wide variety of measures that were not always standardised. For example, QoL has been assessed using one or two questions or using constructs such as life satisfaction or HRQoL. The measurement of coping in previous research was also conducted using a wide variety of measures, with a range of factors. In addition, coping is subjective and therefore difficult to define and measure. Even though the coping measure used in this study was psychometrically adequate it did not evaluate the effectiveness of the coping strategies. The relationship between coping, depression, anxiety, and QoL might have been a matter of how effective the coping strategies were when the caregiver was reacting to a stressful situation.
Due to the lack of comparative information of coping, psychological distress, or QoL in the participants pre-injury, it was not possible to conclude based on the current results, that participants were negatively affected by the injury. There is some evidence in the literature that some caregivers that were depressed following TBI had a history of psychological difficulties pre-injury (Gillen et al., 1998; Kolakowsky-Hayner & Kishore, 1999; Sander et al., 2003). Any changes in the examined variables are therefore limited to post-injury.
Some comments need to be made about the mainly weak correlations between coping and the other variables. This does not seem to be unusual and could indicate a general problem with the coping measure or the conceptualisation of coping in general. Donnellan, Hevey, Hickey, and O'Neill (2006) found mainly weak correlations between coping and other variables in their review of 14 studies investigating coping in stroke victims. It could also be that coping does not play a significant role in the development of psychological distress and lowered QoL or that other variables act as mediators. It is also possible that a larger sample could have resulted in correlations that are more significant.
A final limitation was the absence of assessment of satisfaction and positive emotions in the TBI caregivers. Even though a large proportion of participants showed symptoms of depression and anxiety, the majority of them only had minimal or mild symptoms.
119 Despite these limitations, the present study represents an advance in understanding the impact TBI can have on the caregiver.