The present study addressed whether the positive effects on cognition and the ability to carry out ADLs of a 12-month multimodal therapy of dementia patients are sustainable 10 months after the end of therapy. This study is thus one of the first RCTs to systematically examine the sustainability of the effects of a non-drug therapy in dementia patients. Our results using a confound-adjusted multiple regression model indicated that the positive effects of the group therapy (i.e. therapy group vs. control group) on the ability to carry out ADLs were indeed sustainable. This effect was still existent but no longer statistically significant when controlled for a random subject-specific slope. Such a finding is unique in the literature. To date, only two randomised studies have addressed the sustainability of non-drug therapy procedures on cognition and the ability to carry out ADLs, but in combination with cholinesterase inhibitors.
In a recent study on dementia patients living at home, Giordano et al.
 showed that the success of therapy combining cholinesterase inhibitors and a 3-week Reality Orientation Training (ROT) could still be demonstrated 2 months after the active intervention phase as long as family members continued the ROT. Control patients were treated only with cholinesterase inhibitors. The patients of the therapy group profited, however, only in the cognitive area; effects on their ability to carry out activities of daily living were never observed. Moreover, therapy was never withdrawn as medication was maintained in both the ROT and control groups, and caregivers of the test group were asked to continue ROT at home, even if this potentially occurred in a less systematic form. One problem that this study had is the short time frame, as even placebo effects can last up to 9 months as Ito et al.
 demonstrated in a meta-analysis.
Chapman et al.
 studied the efficacy of a combination therapy consisting of donepezil and cognitive stimulation therapy in a controlled randomised study of 54 dementia patients. The control group received donepezil alone. The parameters were cognition, ability to carry out ADLs, and neuropsychiatric symptoms. The therapy was conducted once a week for 2 months. With the two scales used (ADAS-cog and Texas Functional Living Scale TFLS), which are comparable to those of our study, the authors found a significant decline in abilities in both the test and control groups over the course of a year. No effects on cognition or the ability to perform ADLs could be demonstrated at the end of therapy or 10 months after ending therapy. However, an advantage of the combined therapy was found in the MMSE and some subscales of the Neuropsychiatric Inventory (NPI) 10 months after the end of therapy. As effects of pharmacological treatment have been shown elsewhere to last less than 12 months
, the differences between the treatment and control groups are more likely related to the cognitive stimulation. The lack of effects on the ADAS-cog and the TFLS of this therapy may be attributable to its low intensity (8 sessions over 2 months), which is considerably less intensive than our MAKS therapy (about 300 sessions over 12 months). Other studies examining the sustainability of non-drug therapy procedures have usually addressed the neuropsychological symptoms of dementia, such as mood and behaviour e.g.
[26, 27] or are of limited reliability due to the small number of patients
In the main analysis, we found a long-term effect of the MAKS therapy on the ability to carry out ADLs but not on cognition. Future studies should test whether ADL training will preserve the self-reliance of the participating home residents more than that of the control group. This in turn would mean that ADLs are performed independently to a greater extent and are thus automatically trained further. By contrast, the long-term effect of MAKS therapy on cognitive functions is much lower. Here, the difference between the control group and therapy group was smaller 10 months after the end of therapy, and the effect of the therapy was no longer demonstrable in multivariate analyses. Compared to a strictly pharmaceutical therapy, a non-drug therapy thus appears to offer the potential of an effect that promotes independence in everyday living that lasts beyond the term of therapy, even if the positive effect on cognitive functions did not continue after therapy ended. Because the cognitive abilities as well as ADLs abilities were preserved in the MAKS group during therapy, a continuous therapy could perhaps stabilise both abilities even longer.
A limitation of the present study is the number of patients. After 22 months, only a relatively small sample number of patients remained (n = 52), owing in part to the high mortality rate in the age spectrum examined. The small number of cases is probably also the reason why the model in the sensitivity analysis with a random slope did not attain significance for the ability to carry out ADLs, especially since the variation of the measured E-ADL test scores increased over time (see Figure
2). Also, results are valid only for patients with degenerative dementias such as Alzheimer’s dementia or dementia of the mixed form. Effects on persons with vascular dementia were not examined. Another limitation might be the use of the ADAS-cog as the outcome instrument for cognition. The ADAS-cog does not have parallel versions for test-retest settings; stabilisation might therefore be due to learning effects rather than due to therapy. However, this effect would be the same in the control and therapy groups. Thus, differences between the groups cannot be explained by learning effects. Further studies with a considerably larger number of patients, an extended follow-up period, and more sensitive test items are thus necessary to verify the results shown here.
One strength of the study is the strict RCT design used during the therapy phase. To our knowledge, our study is thus the first to examine the sustainability of an exclusively non-drug therapy of dementia after completion of the active therapy phase. Our method is also notable in that all dependent variables were recorded blindly by testing and not, as is often the case, by using outside assessment.