It's a very interesting question! But I'm not sure if I understand properly what you mean by "across languages". Does it mean that you want to assess language-specific cognitive abilities? If no, then I would say that "cognitive ability" here is a far too broad term. I think the methods would very much vary depending on what cognitive ability and hence what kind of empiric data you work with.

Thank you for your response. In this case, it is important for me to have an estimate of general cognitive ability, tapping hopefully into a generalized intelligence measure or assessing different areas and combining them into a composite. At least that is my first thought.

"Across languages" is meant in the sense that if studies or trials are done across cultures and languages, there will be different kinds of tests or perhaps methods that are traditional in each country (the Wortschatztest in Germany vs. the NART in the US are good examples) for estimating premorbid verbal ability or intelligence. Tthese methods only tap into one kind of intelligence, verbal ability, but do so in different ways: vocabulary test vs. reading ability. My concern is to be able to overcome differences in educational systems and cultures, even when the languages may be the same (USA vs. New Zealand - they share a language but have different educational systems). My idea would be to find ways to objectively assess general premorbid cognitive performance using a concrete assessment after disease onset, and a combination of educational and demographic variables.

My question is: Is there a way to converge on one type of method or set of methods or test batteries that are culturally-sensitive (I realize there will be tremendous limitations in the case of less industrialized areas and less literate populations).  I am interested in sparking a discussion about these issues since it is relevant to many kinds of acute-disease research in which a premorbid cognitive tesingt is usually impossible to obtain. Any ideas are greatly appreciated.

The only method I can think of that 'might' work across countries is a demographic prediction equation. John Crawford, Aberdeen published on one such back in the late 1980s/early 1990s. These equations were designed to predict IQ on the WAIS. The problem, as you so clearly articulate, is that these rely on predictors that may not be equally relevant across all countries. If your samples are large enough, perhaps you could use the equations to generate a prediction then calculate country-specific Z scores. That way, you'd be comparing someone with an average predicted IQ in one country with someone with an average predicted IQ in another.

http://homepages.abdn.ac.uk/j.crawford/pages/dept/pdfs/PAID_1990_Combined_Premorbid_Supplementary.pdf

http://homepages.abdn.ac.uk/j.crawford/pages/dept/pdfs/BJCP_1989_Demographic_Premorbid_IQ.pdf

And how about Mini Mental State Examination Scale by Folstein M.F.? If you compare MMSE with any scale for ADL assessment, you will get simpliest way to estimate critical illness burden.