4. DISCUSSION
4.5 LIMITATIONS OF THE STUDY 1 Sample
The sample consisted o f subjects who had different precipitating traumas. This is in contrast to other studies of PTSD and memory that have investigated subjects according to their traumatic stressor. Thus, the results o f this study may not generalise well to specific populations of adults with PTSD, such as survivors of RTAs or combat exposure. However, the results may generalise well to clients with PTSD who present for treatment.
History of physical and sexual abuse was not thoroughly assessed. There may have been subjects included in the sample who had previous trauma of this nature. Childhood trauma has been associated with memory difficulties in adulthood. For example, Kuyken & Brewin (1995) found that subjects with a history of childhood sexual or physical abuse had more overgeneral memories than subjects without a history of abuse. However, Stein et al., (1999) did not find evidence o f memory difficulties in adult survivors of childhood sexual abuse on their neuropsychological measures o f memory functioning.
A further limitation of the current study was the small sample size. It is possible that the relationship between verbal memory and poor outcome would become weaker with a greater sample size. It is also possible that it would become stronger.
4.5.2 Design
Outcome studies o f PTSD have measured outcome at various intervals. For example, Ehlers (1998) measured outcome at session eight while Nisith et al. (1999) measured outcome at session six. In this study, outcome was measured at session eight irrespective o f severity or type o f PTSD symptomatolgy at intake.
This may not have been the most appropriate time to measure outcome, particularly as clients with different types of PTSD reactions were given different lengths of treatment. For example, a client with PTSD complicated by problems o f shame and guilt would likely have been offered between 16 to 20 sessions o f treatment, whereas a client with PTSD characterised by intense fear would likely have been offered between eight to ten sessions. Thus, subjects were offered different lengths of treatment that may have progressed at different rates. Some subjects may have experienced more active ingredients of recovery (e.g. imaginai exposure) in one length of treatment versus another, and this may have affected the rate of recovery.
This could be addressed in a future study by assessing outcome at a similar time depending on the overall number of sessions offered. Thus, for clients streamlined into a CBT protocol o f 10 sessions, outcome could be measured at session eight. For clients streamlined into a protocol of 16 sessions, outcome could be measured at session 14.
versa. Some therapists may have been more skilled than other therapists. In order to correct this sort o f shortcoming, an outcome study in which therapists are all trained to the same level and all trained in the same protocol would be necessary. Further, it would be necessary to monitor the protocol adherence of therapists. This could be done by taping sessions and assessing treatment adherence.
4.5.3 Measures
(i) Sensitivity of the Diagnostic Questionnaires
At intake, the CAPS (either clinician administered or self-report version), identified similar numbers o f clients with DSM-IV criteria for PTSD as the SRS-PTSD. However, at outcome, the CAPS identified 30 percent of clients as meeting a diagnosis of PTSD, whereas the SRS-PSTD identified 42 percent. This discrepancy suggests that perhaps the CAPS was sensitive to treatment changes or that the SRS- PTSD was insensitive.
The SRS-PTSD is a self-report questionnaire and Carlier et al. (1998) reported that clients tended to overreport PTSD symptoms on the SRS-PTSD compared to the Structured Clinical Interview for PTSD (SI-PTSD).
It is possible that the clinician-administered version of the CAPS is an under estimation o f the true prevalence of PTSD in clients who have attended treatment. The CAPS was administered by the client’s therapist at outcome and this may reflect experimenter bias. That is, the therapists may have been biased in observing improvement in their clients, particularly if they perceived clients’ outcome to reflect on their skills as a therapist.
These reasons may explain the discrepancies between the CAPS and the SRS-PTSD at outcome. In retrospect, it would have been helpful to have had a blind assessor administer the CAPS at outcome.
(ii) Convergent Validity of the CAPS Self-Report Version
Ideally, in order to provide a more accurate picture of the convergent validity o f the CAPS self-report version, it would have been best to administer the CAPS clinician version to a group of subjects who also had completed the self-report version. Correlations could then have been calculated between the self-report CAPS scores and the clinician-administered CAPS scores to assess the strength of association.
(iii) The Adult Memory and Information Processing Battery (AMIPB)
Although the AMIPB is becoming a more widely recognised measure of memory, learning and information processing in British samples, it only provides age-related percentile range scores as opposed to age-related standardised scores. Therefore statistical analyses must use range scores, which are not as accurate as standardised scores. Further, the delay period on the verbal and non-verbal task is just thirty minutes, whereas on other measures of memory, the delay period is 45 minutes or more, and often for verbal material 90 minutes (e.g. Wechsler Memory Scale).
(iv) The Alcohol and Drugs Questionnaire (ADQ)
The Alcohol and Drugs Questionnaire used in this study is a self-report questionnaire. Research has suggested that clients under-report alcohol and drug-
al., 1992). It is possible that clients attending treatment are even more likely to under-report this behaviour, as the information becomes part of their clinical file which then becomes property of the NHS.
It is difficult to assess alcohol and drug related behaviour and perhaps in retrospect, administering a separate alcohol and drug questionnaire as part of the research protocol may have reduced under-reporting, if there was indeed under-reporting. If an alcohol questionnaire was administered during the research session, and clients could be ensured o f confidentiality, this may have yielded more frequent and severe alcohol and drug use. However, as the treatment centre screened for alcohol and drug use before commencing treatment, it is also possible that the measure of alcohol and drug use is accurate in this sample.