Abstract
# Results were biased by self selection of cases {#article-title-2} E—Altmann et al's results on a group of litigants were published almost 10 years after the water pollution incident in Camelford.1 Although they acknowledge competing interests, they overlook the main problem—the bias inherent in self selection of cases. The cases may have already had unexplained symptoms and cognitive problems, the incident serving to focus attention on a possible cause. The results show significant impairment in neuropsychological and neurophysiological tests among the cases, which the authors argue must be the result of prolonged toxicity to acute exposure to aluminium in drinking water. Neuropsychological tests are assumed to be objective, automated, computerised, and quantitative, but they do require the conscious effort of subjects. Those complaining of poor memory and concentration are given a test that requires both, so performance cannot be taken at face value. Subjects are not carrying out a deliberate deception, but their performance like everyone else's is influenced by the context. Similarly, the choice of relatives as controls is unfortunate. Out of loyalty, they will tend to give more than their best, thus widening the gulf in performance. Sensory evoked potentials are less liable to such effects. The authors report an increased latency in “flash minus pattern” evoked potential (EP), an index of cognitive impairment in patients with Alzheimer's disease and aluminium related dialysis dementia. Their data are inadequate for evaluating this finding since neither the raw latencies for the controls nor the waveforms for the cases are given. The results hang on the difference between patterned visual evoked potentials (VEPs) and flash EPs. The VEP is also known as the P100 since it occurs reliably 100 ms after the onset of the stimulus. The response to flash is characteristically longer, by around 20 ms in normal subjects, more in those with brain disorders.2 3 The table …