meningococcal infection.7,8 It is likely that strains with the P1.16b epitope will show differences in their recognition by antibodies which result from
immune response to the typical PI. 16 epitope which, in turn, could affect individual susceptibility to meningococcal disease and the level of protection conferred by novel 1meningococcal vaccines based on outer membrane proteins. an
Supported by the National Meningitis Trust and by a Medical Research Council project grant. B. T. McG. had an MRC Research Studentship. REFERENCES Cartwright KAV, Stuart JM, Jones DM, Noah ND. The Stonehouse survey: nasopharyngeal carriage of meningococci and Neisseria lactamica. Epidem Inf 1987; 99: 591-601. 2. Poolman JT, Lind I, Jónsdóttir K, Frøholm LO, Jones DM, Zanen HC. Meningococcal serotypes and serogroup B disease in north-west Europe. Lancet 1986; ii: 555-58. 3. Knight AI, Cartwright KAV, McFadden J. Identification of a UK outbreak strain of Neisseria meningitidis with a DNA probe. Lancet 1.
1990, 335: 1182-84. 4. McGuinness B, Barlow AK, Clarke IN, et al. Deduced amino acid sequences of class 1 protein (Por A) from three strains of Neisseria meningitidis: synthetic peptides define the epitopes responsible for serosubtype specificity. J Exp Med 1990; 171: 1871-78. 5. Frasch CE, Zollinger WD, Poolman JT. Serotype antigens of Neisseria meningitidis and a proposed scheme for designation of serotypes. Rev Infect Dis 1985; 7: 504-10. 6. Poolman JT, Timmermans HAM, Hopman CTP, et al. Comparison of meningococcal outer membrane protein vaccines solubilized with
Leeuwenhoek 1987; 53:
7. Wedege E, Froholm LO. Human antibody response to a group B serotype 2a meningococcal vaccine determined by immunoblotting. Infect Immun 1986; 51: 571-78. 8. Saukkonen K, Leinonen M, Abdillahi H, Poolman JT. Comparative evaluation of potential components for group B meningococcal vaccine by passive protection in the infant rat and in vitro bactericidal assay. Vaccine 1989; 7: 325-28. 9. Frasch CE, Tsai C-M, Mocca LF. Outer membrane proteins of Neisseria meningitidis: structure and importance in meningococcal disease. Clin Invest Med 1986; 9: 101-07. 10. Mandrell RE, Zollinger WD. Human immune response to meningococcal outer membrane protein epitopes after natural infection or vaccination. Infect Immun 1989; 57: 1590-98. 11. Barlow AK, Heckels JE, Clarke IN. Molecular cloning and expression of Neisseria meningitidis class 1 outer membrane protein in Escherichia coli K-12. Infect Immun 1987; 55: 2734-40. 12. Barlow AK, Heckels JE, Clarke IN. The class 1 outer membrane of Neisseria meinngindis: gene sequence, structural and immunological similarities to gonococcal porins. Mol Microbiol 1989; 3: 131-39. 13. Abdillahi H, Poolman JT. Neisseria meningitidis group B serosubtyping using monoclonal antibodies in whole cell ELISA. Microb Pathogen
1988; 4: 27-32. 14. Abdillahi H, Poolman JT. Definition of meningococcal class 1 outer membrane protein subtyping antigens by monoclonal antibodies. Fed Eur Microbiol Soc Microbiol Immun 1988; 47: 139-44. 15. Virji M, Zak K, Heckels JE. Outer membrane protein III of Neisseria gonorrhoeae: variations in biological properties of antibodies directed against different epitopes. J Gen Microbiol 1987; 133: 3393-401. 16. Jones DM, Foz AJ, Gray SJ, Saunders NA. DNA probes for typing Neisseria meningitidis. Lancet 1990; 336: 53-54.
SHORT REPORTS No confirmation of visual evoked potential diagnostic test for
migraine We have attempted to replicate the results of studies on a diagnostic test reported to have 90% sensitivity and 89-96% specificity for migraine. The technique is based on peak-to-peak measurements of fast
background electroencephalographic activity during visual evoked potential (VEP) study. VEP latencies and amplitudes did not differ significantly, and showed substantial overlap, between a group of eight migraine patients and ten age-matched healthy controls. We could not recognise previously described fast activity or measure it objectively by a
measurements. We cannot confirm
Mortimer and colleagues measured "fast activity in the background of the VEP", which corresponds to EEG beta activity rather than to the VEP itself. This activity should be investigated by techniques allowing single response analysis;6,7however, Mortimer et al based their conclusions on peak-to-peak measurements in averages of five We have responses.8 attempted to replicate this technique to assess its diagnostic value. Results of single-stimulus analysis methods will be published elsewhere. Because of the high reported sensitivity and specificity the number of subjects in our study could be small. Eight patients with migraine, based on the diagnostic criteria of the International Headache Society9 (two male, six female; mean age 33 [25-62] years), and ten healthy controls (five male, five female; 33 [23—52] years) were studied. Exclusion criteria were low visual acuity (Snellen cards), use of prophylactic antimigraine or other long-term medication, and a migraine attack in the 5 days before the study. Three of the eight patients had aura with migraine. The mean duration of migraine was 19 (5-40) years, the mean attack frequency 1-4 (0 25-6 0) per month, and the mean time since last attack 47
Mortimer and colleagues1-3 have advocated the use of visual evoked potentials (VEP) for the diagnosis of migraine. The sensitivity (90%) and specificity (89-96%) they reported are surprisingly high ’in view of earlier VEP studies in migraine.4,5 Electroencephalographic (EEG) abnormalities after photic stimulation have been described in migraine patients,4 but no method based on photic stimulation (including the VEP) has been accepted as a
A Nicolet ’Pathfinder II’ was used for the VEP study. A check pattern with a check size subtending 1°10’ of arc was used. We used three EEG leads-Fz-Oz, Fz-Cz, and Cz-Oz, with Fpz as ground-and a 1-40 Hz bandpass filter was used. Artifactcontaminated responses were rejected when the voltage passed a 100 µV threshold. The analysis period was sampled in 512 data points. Approximately sixty stimuli were given with a frequency of 1 9Hz; 20 Hz1,2 was not used to avoid synchronisation with power sources. Twenty randomly chosen responses were separately stored on hard disc. The remaining responses were averaged on acquisition. N70 and PI 00 latencies and their interpeak amplitude were measured from the Fz-Oz lead of these forty-stimulus averages by a scorer unaware of the clinical diagnosis. The twenty separately stored EEG responses were used to form four averages for each person-five-stimulus averages (each of five consecutive
diagnostic test in migraine.
of fast wave activity in the VEP is useful in diagnosis of migraine.
The mean latencies and amplitudes of the groups were compared by Wilcoxon’s two-sample test. A p value below 0-05 was considered significant.
There was no significant difference between migraine patients and controls in VEP N70 latency (mean 70-2 [SD 9-3] vs 72-7 [4’6] ms), P100 latency (105-7 [5’2] vs 105-7  ms), or N70 -P100 amplitude (10-2 [4’6] vs 9-6 [4’5] V). We attempted to measure the amount of fast wave activity in the 250-500 ms post-stimulus period with peak-to-peak measurements in the five-stimulus averages.1,8 Mortimer et al did not explain the latency and amplitude criteria they used to define fast wave activity. The averages showed no suggestion of any recognisable activity on which such criteria might be based. The figure shows the first of the four five-stimulus averages obtained for each subject. We found large differences in the pattern of 250-500 ms activity among each subject’s four averages, without any suggestion of a consistent tendency towards beta frequencies in either group.
We found no significant difference in VEP latencies or amplitudes between migraine patients and controls, as we expected in view of previous reports.4 We were unable to replicate Mortimer and Good’s measurements of fast wave activity, possibly because we did not have a clear and full description of the measurement technique. More importantly, our five-stimulus averages showed too much variability to allow recognition of fast activity or definition of any such criteria. Our traces do not resemble those of Mortimer and Good,1,3in which the amount of activity in the control group is surprisingly low since these averages are based on only five responses. It is important to understand the theory underlying the averaging process:7 the EEG after a stimulus consists of waves linked in polarity and latency to the stimulus and background waves, many of larger amplitude, without this relation to the stimulus. Averaging cancels the background waves, leaving the replicable stimulus-related waves visible. The term "evoked potential" is usually reserved for these recognisable waveforms, which are described with latencies and amplitudes, and it is confusing to use this term for background EEG activity. Averaging of a larger number of responses will improve the evoked potential, because background EEG waves will become less apparent. Five
Five-stimulus averages of migraine patients and controls. Each trace=first average of five responses for one subject. 250-500 ms post-stimulus period used to search for fast wave
are not enough to suppress background EEG sufficiently for a VEP study: thirty to one hundred stimuli are used in clinical practice. If changes in the background EEG are of interest, single EEG responses have to be studied with appropriate techniques, such as spectral analysis. Averaging is designed to suppress the background EEG, not to study it. Averaging theory does not explain how an average of five responses can result in optimum recognition of background EEG patterns.8 The number of patients in our study was small. However, the reported sensitivity and specificity values1.2 are so high
that differences between the groups should have been apparent in our study. We cannot confirm the claim that this technique can be of any help in diagnosing migraine. REFERENCES 1. Mortimer
MJ, Good PA. Visual evoked respoinses in children with migraine. Lancet 1990; 335: 75-77. 2. Marsters JB, Good PA, Mortimer MJ. A diagnostic test for migraine using the visual evoked potential. Headache 1988; 28: 526-30. 3. Mortimer MJ, Good PA, Marsters JB. The VEP in acephalic migraine. Headache 1990; 30: 285-88. 4. Winter AL. Neurophysiology and migraine. In: Blau JN, ed. Migraine: clinical, therapeutic, conceptual and research aspects. London: Chapman and Hall, 1987: 485-521. 5. Dijk JG van, Kamphuisen HAC. Migraine, epilepsy and clinical neurophysiology. In: Ferrari MD, Lataste X, eds. New trends in clinical neurology: migraine and other headaches. Carnforth, New Jersey: Parthenon, 1989:107-18. 6. Dijk JG van, Ferrari MD, Peters ACB. Visual evoked responses in children with migraine. Lancet 1990; 335: 480. 7. Regan D. Human brain electrophysiology. Amsterdam: Elsevier, 1989. 8. Mortimer MJ, Good PA. Migraine and visual evoked potentials. Lancet 1990; 335: 789. 9. Headache Classification Committee of the International Headache Society. Classification and diagnostic criteria for headache disorders, cranial neuralgias and facial pain. Cephalalgia 1988 (suppl 7): 1-96.
Department of Neurology/Clinical Neurophysiology, Leiden University Hospital, PO Box 9600, 2300 RC Leiden, Netherlands (J. G van Dijk, MD, M Dorresteijn, J. Haan, MD, M. D Ferrari, MD) Correspondence to Dr J. G van Dijk.
Co-trimoxazole for childhood febrile illness in malaria-endemic
The efficacy of co-trimoxazole for the treatment of Plasmodium falciparum parasitaemia in children younger than 5 years of age was evaluated in Malawi. 46 children with P falciparum parasitaemia, 37% of whom also met clinical criteria for a diagnosis of acute lower respiratory tract infection, were treated with 20 mg/kg co-trimoxazole twice daily for five days. Parasitaemia (mean clearance time 2·7 days) and symptoms were rapidly abolished and improvement was maintained during follow-up for 14 days. Co-trimoxazole may be an effective single treatment for febrile illness in young children in areas where malaria is endemic, resources are few, and diagnosis must rely on clinical findings alone.