792 SIR-Goodeve and colleagues analysed 22 patients with severe haemophilia A for inversions in intron 22 of the factor VIII gene. Factor VIII inhibitor did not develop in 10 patients who showed the rearrangement but did so in 7 of 12 patients without the inversion. We did inversion analyses by Southern blot in 92 severely haemophilic patients whose inhibitor status was well known. 36 had an inversion and 13 of these (36%) had inhibitors. In two families more than one affected brother was analysed. As expected, in each family, the brothers presented the same inversion but they were not concordant in respect of the inhibitor development. Of the 56 non-inversion cases 9 (16%) had inhibitors (p < 0 02). In his commentary Tuddenham points out that the distribution of no-inhibitor-development in Goodeve’s patients with the inversion was probably atypical. Moreover, in our results there is a significant tendency for inhibitors to develop in patients who show the inversion. No clear prediction of the development of inhibitors can be made when an inversion of intron 22 is detected in a patient with severe haemophilia. E F Tizzano, C Altisent, J Tusell, M Dom&egrave;nech, M Baiget Molecular Genetics Unit, Hospital de la Santa Creu i, Sant Pau, Barcelona 08025, Spain; and Haemophilia Unit, Hospital Vall d’Hebr&oacute;n, Barcelona Risk of gastrointestinal bleeding from dexamethasone in children with bacterial meningitis SIR&mdash;The occurrence of duodenal perforation in a child with Haemophilus influenzae meningitis treated with adjunctive dexamethasone prompted us to do a meta-analysis on gastrointestinal complications of steroids in meningitis. A previously healthy 9-year-old girl presented with fever and meningism. The cerebrospinal fluid had 2-4 x 109 jL white blood cells (92% neutrophils), glucose less than 0-5 mmol/dL, and protein 244 mg/dL. Intravenous ceftriaxone (50 mg/kg every 12 h) and dexamethasone (0-15 mg/kg every 6 h for 4 days) were started simultaneously. Great improvement was seen within 2 days. Cerebrospinal fluid and blood cultures grew beta-lactamase positive H influenzae type b. Melaena was noted 4 days after admission. Coagulation indices and platelet count were normal. The patient had not taken non-steroidal anti-inflammatory drugs (NSAIDs). Cimetidine was started but within 24 h the packed cell volume dropped from 37 to 23%. Endoscopy showed a large duodenal ulcer with an actively bleeding vessel and perforation. After emergency coeliotomy the ulcer and the bleeding gastroduodenal artery were oversewn. Four units of packed red blood cells were transfused. The hospital stay was extended to 14 days. Although controversial, the association of corticosteroids with peptic ulceration and bleeding has been suggested even for short courses of treatment in settings other than meningitis.1 Concomitant use of NSAIDs may be a confounding factor, but steroids alone may lead to larger ulcers and further complications.2 Shorter courses (2 rather than 4 days) may have fewer gastrointestinal side-effects.3 The meta-analysis included 8 dexamethasone trials for bacterial meningitis in children3-9 (6 identified through MEDLINE; conference data for the two recent US and Canadian trials). All the prospective double-blind placebo- controlled trials that comment on bleeding complications were included. Retrospective studies were excluded as potentially biased in avoiding steroids in patients prone to bleeding, except for a study where consecutive selection of patients in the two groups eliminated this bias. Several studies monitored occult blood loss in the stool. However, occult blood loss has no clinical consequences. Thus we counted only cases with clinically significant bleeding (perforation or ulcer with haematemesis or bleeding requiring transfusion) excluding haemorrhage due to disseminated intravascular coagulation. The overall incidence of significant bleeding in the absence of coagulopathy was 4 of 803 (0-5%). There were no cases in placebo-treated (n=397) compared with 4 among steroid- treated (n=406) patients. The risk difference is 0-71% with 95% confidence intervals (CI) - 0 65 to 2-08% (DerSimonian and Laird model, 2p = 0 31) and the odds ratio is 3-96 with 95% CI 085-1848 (Mantel-Haenszel model, 2p=008). Larger studies are needed to verify this trend for increased bleeding in steroid-treated children. Overall, clinically important gastrointestinal bleeding from stress ulcers is rare in children with bacterial meningitis. The excess incidence related to steroids is probably less than 1%. However, this effect can be serious, as our case illustrates. A small increase in acute complications from steroids could be clinically relevant and might have to be weighted against the neurological benefits. John P A Ioannidis, Matthew D Samarel, Joseph Lau, Mark S Drapkin Department of Medicine, Divison of Geographic Medicine and Infectious Diseases, New England Medical Center Hospitals, Boston, MA 02111, USA; and Newton-Wellesley Hospital, Boston 1 Messer J, Reitman D, Sacks HS, Smith H, Chalmers TC. Association of adrenocorticoid therapy and peptic-ulcer disease. N Engl J Med 1983; 309: 21-94. 2 Carpani de Kaski M, Levi S, Rentsch R, Hodgson HJF. Steroid ulcers. BMJ 1992; 304: 1443-44. 3 Schaad UB, Lips U, Gnehm HE, Blumberg A, Heinzer I, Wedgewood J. Dexamethasone therapy for bacterial meningitis in children. Lancet 1993; 342: 457-61. 4 Marguet C, Mallet E. Interet de la dexamethasone au cours des menningites purulentes de l’enfant. Arch Fr Pediatr 1993; 50: 111-17. 5 King SM, Law B, Langley J, et al. A randomized controlled trial of dexamethasone vs placbo in children with bacterial meningitis. 32nd Interscience Conference on Antimicrobial Agents and Chemotherapy. Anaheim, California, Oct 12, 1992 (abstr 72, 122). 6 Wald E, and US Meningitis Study Group. Dexamethasone for children with meningitis. 32nd Interscience Conference on Antimicrobial Agents and Chemotherapy. Anaheim, California, Oct 12, 1992 (abstr 72, 123). 7 Odio CM, Faingezicht I, Paris M, et al. The beneficial effects of early dexamethasone administraton in infants and children with bacterial meningitis. N Engl J Med 1991; 324: 1525-31. 8 Lebel MH, Hoyt J, Waagner DC, Rollins NK, Finitzo T, McCracken GH Jr. Magnetic resonance imaging and dexamethasone therapy for bacterial meningitis. Am J Dis Child 1989; 143: 301-06. 9 Lebel MH, Freij BJ, Syrogiannopoulos GA, et al. Dexamethasone therapy for bacterial meningitis: results of two double-blind, placebo- controlled trials. N Engl J Med 1988; 319: 964-71. Immunoadsorption with protein A sepharose or silica SIR-Kabisch and colleagues (Jan 8, p 116) report adverse effects in the treatment of autoimmune thrombocytopenia (AITP) when they used prosorba columns (protein A silica; IMRE, Seattle) and that they have abandoned this treatment. We have treated 27 patients with various autoimmune diseases on 97 occasions, using Immunosorba columns (protein A sepharose; Excorim, Lund) without severe side-effects. Anticoagulation-dependent citrate reactions, controlled by oral calcium, developed 6 times; 5 transient hypotonic episodes did not require medication; and no patient experienced pain, chills, fever, rash, musculoskeletal pain, arthritis, or vasculitis. The immunoglobulin depletion was effective. 5 patients with AITP and 1 with Evans syndrome, resistant to conventional drug therapy, were treated by sequential