Few Effects of Far Transfer of Working Memory Training in ADHD: A Randomized Controlled Trial

Saunes B-K (2013) Few Effects of Far Transfer of Working Memory Training in ADHD: A Randomized Controlled Trial. PLoS ONE 8(10): e75660. doi:10.1371/journal.pone.0075660 Few Effects of Far Transfer of Working Memory Training in ADHD: A Randomized Controlled Trial Jens Egeland 0 Anne Kristine Aarlien 0 Brit-Kari Saunes 0 Marianna Mazza, Catholic University of Sacred Heart of Rome, Italy 0 1 Division of Mental Health & Addiction, Vestfold Hospital Trust, Tnsberg, Norway, 2 Department of Psychology, University of Oslo , Oslo , Norway , 3 Division of Child and Adolescent, Telemark Hospital Trust , Skien , Norway Objective: Studies have shown that children with ADHD profit from working memory training, although few studies have investigated transfer effects comprehensively. The current Randomized Controlled Trial analyzes transfer to other neuropsychological (NP) domains, academic performance and everyday functioning at home and school. Method: Sixty-seven children with ADHD were randomized into a control group or a training group. The training group underwent Cogmed's RoboMemo program. All participants were assessed pre-training, immediately after and eight months later with a battery of NP tests, measures of mathematical and reading skills, as well as rating scales filled out by parents and teachers. Results: There was a significant training effect in psychomotor speed, but not to any other NP measures. Reading and mathematics were improved. There were no training induced changes in symptom rating scales either at home or at school. The increased reading scores remained significant eight months later. Conclusion: The study is the most comprehensive study of transfer effects to date, and with mixed results compared to previous research. More research is needed regarding how to improve the training program and the conditions and thresholds for successful training. Trial Registration: Controlled-Trials.com ISRCTN19133620 PLOS ONE | www.plosone.org - Funding: The study has been funded with grants from the Centre for child and Adolescent Mental Health, Eastern & Southern Norway and from The Norwegian Resource Center for ADHD, Tourette and Narcolepsy. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. Impaired working memory (WM) is characteristic of a multitude, or even most, neuropsychiatric [1] and developmental disorders [2]. A meta-analysis of studies of WM performance in ADHD [3] showed impairments ranging from half to more than one standard deviation, depending on whether tasks demanded information manipulation or merely storage. There are several models of WM [4,5]. Common to them is that WM operates on a limited time period after information is being presented enabling the person to hold the information online for the time needed to process it. Simple storage is equivalent to short term memory or attention span, either within the visual of auditory modality, while manipulation involves higher level executive functions. Typical tasks measuring storage is visual or auditory span tasks, measuring the extent of information the person can grasp without rehearsing. Manipulation tests can differ from simple manipulation involved in reversing a sequence of numbers held in simple storage, to more complex manipulation involved for example in reorganizing both numbers and letter sequences [6]. In the multimodal model of Baddeley [4], simple storage is part of the WM model, although differentiated from the Central Executive. Other researchers have found evidence of the distinction of short term memory vs. working memory [7], and reserve the WM construct for manipulation only. The working memory training regime tested here is based on Baddeleys model incorporating both simple storage in the short-term memory sense as well as manipulation, into the model. Several studies have reported that WM capacity in ADHD [8,9] and other disorders [8,10] can be improved by training on specially designed computer-based training programs. The producer of Cogmed-program RoboMemo claims that training has been implemented in more than 800 schools in Sweden, where the program was developed [11]. Pearson Assessment is marketing the program in the USA, and promoting it as an effective treatment of WM deficit in several disorders in addition to ADHD. On the other hand, recent studies have harshly criticized these positive claims [12,13] initiating a debate about the theoretical and empirical basis for a possible treatment effect [14,15]. There are several reasons for being enthusiastic about WM training: Whereas medication effects depend on continued use, WM training can potentially cure the deficit in the sense that the effect lasts beyond the training period. How long it lasts is not clear, since no study to date has had follow up more than 6 months after completed training [16,17]. Some parents do not want their children to use stimulant medication. In some cases medication is terminated due to side effects or lack of treatment effect on everyday behavior. Although considered an effective drug in reducing behavioral symptoms of ADHD, another reason for exploring the possibility of training WM is that methylphenidate (MPH), may not have the equivalent effect on working memory as on other aspects of executive function (EF). Some studies have shown no effect on WM while attention in general is improved [18,19]. A metastudy by Pietrzak and colleagues [20] showed that only half of published studies of MPH-treatment showed effects on the manipulation element of WM. The authors discuss the possibility that the modulation of dopamine turnover affects WM less, possibly because WM requires simultaneous application of multiple cognitive processes exceeding those modulated by MPH. While WM training represents a longed-for non-pharmacological treatment of cognitive symptoms of ADHD, it is nevertheless important to assess the transfer effects critically. Reports from new and experimental treatments will typically suffer from a publication bias. Negative findings are not relevant since the treatment is not broadly applied. Shipstead, Redick and Engle [21] present a list of factors that can contaminate the validity of training experiments. They claim that the effects on measures of WM are sufficiently documented, while transfer effects are not. They point out that results on single tests cannot be interpreted as an increase in a function, but that multiple measures of the same construct must converge to conclude about a transfer effect. In addition, if several studies conclude on transfer effects within different domains, the overall picture may be more negative. The meta-analysis of Melby-Lervag and Hulme [12] offers an excellent overview of previous research. The authors conclude that WM training programs appear to produce short-term, specific tra (...truncated)