Overlooking the obvious: a meta-analytic comparison of digit symbol coding tasks and other cognitive measures in schizophrenia
Dickinson D, Ramsey ME, Gold JM;
Commented by , 31 May 2007
Aim of the study
Cognitive dysfunctions are very disabilitating symptoms of schizophrenia that are considered to be a core part of the disorder. There is an extensive literature on the assessment of this problem, but its interpretation is difficult in part due to the high number of different paradigms and tests available.
Furthermore, many of the available tests are not useful for daily routine, because they are too time consuming. In this light the authors compare the effects of a rather simple tests – the digit symbol coding tasks – and compared them with other neuropsychological instruments.
Method
The literature search was mainly based on the electronic databases MEDLINE and PsycINFO (1990 – April 2006). Included were English language articles assessing the cognitive performance of participants with schizophrenia with that of healthy controls. Studies that used digit symbol coding tasks and at least two other cognitive tests were included.
Random effects meta-analysis was used to compare the effect sizes obtained in the coding task with those in the other tests. A number of potential moderator variables relating to clinical characteristics (chronicity, severity, age, education, medication use) and to design features (publication date, coding task used, matching) were also assessed.
Results
37 studies with 1961 participants with schizophrenia and 1444 healthy controls were included. The mean effect size (Hedges's g) obtained by digit symbol coding tests (-1.57) was significantly larger than the mean effect size for all other cognitive tests combined (Hedges's g = -0.98).
Analysing comparisons of coding tasks with specific cognitive tests yielded higher effect sizes of the former compared to 33 out of 36 common measures of executive functioning, episodic memory, and working memory. Although as expected younger patients were less impaired than older patients, the results were robust towards the effects of most of the moderator variables mentioned above.
The coding tasks were also more sensitive than other measures towards cognitive impairments in studies on relatives of people with schizophrenia.
Dr Leucht's comments
The intriguing finding of the meta-analysis is that simple 5-minute digit symbol coding tasks that are easy to administer prove to be more sensitive than other measures to detect cognitive impairment in people with schizophrenia or their relatives. The authors, thus, argue that this measure should receive more weight in research than they currently have.
These tasks measure the processing speed with which different cognitive operations can be executed. It might be a central part of the cognitive dysfunction of people with schizophrenia.