Statistics from Altmetric.com
In their welcome systematic review of supratentorial cavernous malformations and epilepsy, Moran et al 1 illustrate the pitfalls of regarding the prognosis of a disease in selected case series as representative of its natural history.2 Studies of cavernous malformation prognosis have usually lacked clear inception cohorts with respect to mode of presentation and treatment. Referral filter bias has so often restricted ascertainment by tertiary referral centres, and further selection bias has made the prognosis seem worse than it really is, as demonstrated by the authors' own series of 33 patients in which temporal lobe lesion location and intractable seizures predominated. Conversely, by leaving community mortality unaccounted for, the prognosis can seem better than it actually is. Completeness of follow up has been variable and not always prospective. Furthermore, authors have varied in their choice of outcome, in particular their definition of haemorrhage (clinical or radiological), choice of period at risk (from birth, time of diagnosis, or start of observation) and calculation of outcomes for each patient or for each lesion. Any analyses of such heterogeneous case series should be ruthlessly systematic, but even so it is necessary to be wary about drawing firm conclusions from them.3
The only existing population based study of cavernous malformations,4 albeit with a denominator of merely 50 000, was retrospective. The study spanned fundamental developments in the non-invasive diagnosis of cavernous malformations during the 1980s with magnetic resonance imaging,5 which led to increasing detection rates with time.
There is, therefore, clearly a need for a large, population based, prospective, contemporary epidemiological survey of cavernous malformations to establish their frequency and prognosis. With a broad collaborative network, including the three other neuroscience centres in Scotland, the Scottish Intracranial Vascular Malformation Study (SIVMS) has been set up (http://www.dcn.ed.ac.uk/ivm/) to do just this for all types of intracranial vascular malformation (IVM). Using multiple, overlapping sources of case ascertainment we are building an inception cohort of all incident cases of any type of IVM diagnosed after 1 January 1999 in the population of Scotland (5.1 million). With prolonged follow up of this cohort we hope to settle some of the uncertainties highlighted by Moran et al.1 Moreover we agree that, with such poor data available, a randomised controlled trial of surgical versus conservative treatment for cavernous malformations is overdue.