Chapter 2; The Systematic Reviews
NAME OF STUDY
2.3.4. Conclusions
Results showed that RO had significant positive effects on both cognition and behaviour for people with dementia. Trials varied greatly in factors such as length of intervention, methodological quality and outcome measures used. There was no observed relationship between the total amount of intervention (in terms of time or length of sessions) and outcome. In fact, the most significant changes resulted from the study with the shortest duration of intervention, 600 minutes in total (Breuil et al, 1997). Additionally, there was variation in the alternative activities offered to control groups, with no treatment in some trials (Baines et al, 1987; Breuil et al, 1994; Ferrario et al, 1991), and an alternative ‘social therapy’ in others (Gerber et al, 1991; Wallis et al, 1983; and Woods, 1979). The results also showed no relationship between type of control activity and outcome, suggesting that the actual qualities of RO, rather than merely the therapeutic effect of social contact and attention, was effective. The largest study was that of Breuil et al (1994), with a 45.7% weight. It is possible that it slightly differed from the other studies in that its concepts were more theoretically advanced than those of the 1970’s (see table 2), and more akin to the sophisticated cognitive rehabilitation programmes used in brain injury.
It is important to look at the advantages and disadvantages of combining the results for meta-analysis. Firstly, the included studies were clearly heterogenous, with variations in the precise intervention used, and the design and conduct of the study. It could be argued that combining such results is not meaningful, and could result in an obscured meta-analysis. Secondly, it has been demonstrated that a reasonable level of bias can be expected in all the
included studies. In treatment trials, this bias usually tends to be in the direction of overestimating the effects of the intervention, and pooling data from different studies adds together these positive biases (Moncrieff, 1998), However, it could also be argued that the studies individually are too small to detect effects that are actually clinically significant, and only combining them achieves the power to detect such effects. Thirdly, one’s opinion about whether combining more or less trials is favourable might influence their interpretation of the cognitive and behavioural analyses, which differed in sample size.
A set of criteria for identifying empirically validated treatments was recently developed by the American Psychological Association (Gatz et al, 1998). Rigorous inclusion criteria (including the adherence to standardized treatment manuals) were set for “well established” treatments, and applied to disorders seen in practice. The authors concluded that RO for dementia is “probably efficacious in slowing cognitive decline”, lending support to RO as an intervention.
In summary, this review found that classroom RO had benefits for dementia sufferers in both cognitive and behavioural domains, suggesting that RO techniques could be considered as a standard part of dementia care. However, limitations such as heterogeneity and biases should be considered when interpreting the results. It is possible that the benefits of RO may only be short-lived, but a more longterm programme may help sustain improvements. As with all psychological interventions, the success of RO may be dependent on it being used at the appropriate time, by sensitive and experienced practitioners, to receptive patients.
2.4.
RT review
2.4.1. Selection of trials
From the information in the title and abstract, 15 publications were identified as possibly relevant following the literature search. 12 were discarded as 2 were not trials, 3 examined non-dementia patients, 1 was a case study, 2 were observational and 4 were controlled trials with neither randomisation nor appropriate outcome measures. This left 3 RCTs. Orten et al (1989) was later excluded due to a lack of clarity in the diagnosis of some subjects, and Goldwasser et al (1987) did not contain statistics needed for entry into metaview. Hence only one trial (Baines et al, 1987) could be entered. This trial was also used in the RO review, as the authors evaluated both RO and RT within the same trial.
2.4.2. Quality of included study
A description of the included study (Baines et al, 1987) can be found in section 2.3.2.
2.4.3. Analysis
As there was only one trial, data could not be combined for meta-analysis. Figures 3 and 4 show the results of the single trial. For the Information/Orientation subscale of the CAPE, WMD = 0.05, 95% Cl (-4.37, 4.77). For the behaviour subscale of the CAPE, WMD = -3.3, 95% Cl (-14.19, 7.59). Hence both scales showed insignificant results, with a positive trend in behaviour and a negative trend in cognition. No further statements could be generated from these results, as they were too limited.
Study Expt n Expt m ea n (sd ) Information/Orientation (CAPE) B a in es 1 9 8 7 5 - 5 .8 0 ( 2 .9 5 ) Subtotal (95%CI) 5 C hi-square 0 .0 0 (df=0) Z = 0 .0 9 Total (95%CI) 5 C hi-square 0 .0 0 (df=0) Z = 0 .0 9 Ctrl n Ctrl m ea n (sd ) WMD (95%CI R andom ) W eight WMD % (95%CI R andom ) -6 .0 0 (4.30) 1 0 0 .0 0 .2 0 0 [-4 .3 7 1 ,4 .7 7 1 ] 1 0 0 .0 0 .2 0 0 [-4 .3 7 1 ,4 .7 7 1 ] 1 0 0 .0 0 .2 0 0 [-4 .3 7 1 ,4 .7 7 1 ] -1 0 -5 10
Figure 5: RT; Behaviour. The length o f the lines represent the size o f the confidence intervals. Crossing the centre line indicates no change in behaviour. Expt Expt Study n m ea n (sd ) Ctrl n Ctrl m ea n (sd ) WMD (95%CI Fixed) W eight % WMD (95%CI Fixed) C A PE (Behaviour) B a in es 1 9 8 7 5 1 3 .7 0 (9 .0 7 ) 5 1 7 .0 0 (8 .4 9 ) ✓ 1 0 0 .0 1 nn n -3 .3 0 0 [-1 4 .1 9 0 ,7 .5 9 0 ] o n n \.-\A 1Qr> 7 qom 5 \ o u u i u ia i /oL/i/ V C hi-square 0 .0 0 (df=0) Z = 0 .5 9 T ntal q 5 y 1 nn n - o .o u u [ -14. 1 y u , / .o y o j q n n r.-tA i o n 7 q o n i o 1 Ulal ^yO/ov_»lj O C hi-square 0 .0 0 (df=0) Z = 0 .5 9 -o .o v ju [ -14. 1 y u , / .o y u j VO -1 0 -5 10
2.4.4. Conclusions
Only one trial (with ten participants) met the inclusion criteria of this review, and results were not statistically significant. The sample size was insufficient to reach any conclusions. It was also limited in that it only examined residents of local authority homes, who may differ from people with dementia living in the community. The two RCTs which were excluded also offered little insight into the effectiveness of RT as a treatment. Goldwasser et al (1987) found a slight but insignificant improvement in cognition in the RT group compared to the two others, no differences at all in behaviour, and a significant increase in depression for the RT group. Orten et al (1989) found that RT participants scored (insignificantly) higher in a “social behaviour scale”, and no correlation between social isolation and ability to participate in RT. Baines found that participants benefited more in both cognition and behaviour from RT following four weeks of RO, than from RT alone, suggesting that RT might be more beneficial for people with a higher level of orientation.
In summary, this review highlights the urgent need for more RCTs and generally more empirical research in the field. This should be interpreted as a positive outcome, indeed Williams (1998) stated that ‘*Ifwe confine systematic reviews to areas where there are lots o f RCTs, then work becomes data-driven rather than question-driven. Finding no R C V s is extremely important: this is our only chance o f influencing funding authorities to conduct the trials that should have been done years ago.”
Reported benefits of RT are mostly anecdotal, and research evidence is not strong enough to reach any firm conclusions. Research is needed to define when and how it should be used, and how it compares to other psychological therapies used in dementia care, such as RO and VT.