1663 the Government of India has grudgingly adopted this strategy in selected districts. Motivation of patients during the course of treatment (yet another factor in default5) can also be maintained if treatment duration is shortened. Mangla is not explicit about the World Bank’s improved drug delivery system. It would be tragic if World Bank assistance is used to pursue the ineffective treatment regimen. Departments of Community Medicine and Surgery, All India Institute of Medical Sciences, New Delhi 110 029, India SHASHI KANT L. R. MURMU 1. Singh MM, Banerji D. A follow up study of patients of pulmonary tuberculosis treated in an urban clinic. Ind J Tuberc 1968; 15: 157-63. 2. Bhatia JL, Raj B. Twice weekly streptomycin and isoniazid regimen without supervision. Ind J Tuberc 1968; 15: 67-73. 3. Anderson S, Banerji D. A sociological inquiry into an urban Tuberculosis Control Programme in India. Bull World Health Organ 1963; 29: 685-700. 4. Griffeths ML, Makgothi MM, Nordesjo G. Tuberculosis management in rural community-factors in failure. S Afr Med J 1981; 59: 14-16. 5. Pande RV. Treatment default of tuberculosis patients in a domiciliary service scheme. Ind J Tuberc 1968; 15: 107-12. Heliotrope poisoning in Tadjikistan SiR,—The seeds and roots of Heliotropium lasocarpium contain a pyrrolizidine alkaloid that causes toxic liver injury and occlusion of centrolobular veins in the liver.’ In southern Tadjikistan, the Farkhar region was under blockade from May to November, 1992. This led to a famine and a delay in the wheat harvest of 2 months. At the same time, a drought affected the area, and heliotrope had time to grow and go to seed. The wheat was collected with heliotrope, ground, and made into bread. On Oct 15, 6 weeks after consumption of the contaminated bread started, the first case of liver toxicity was reported to the Farkhar hospital and an epidemiological surveillance system was established. At the request of the Tadjik government, the Association Europeenne de Developpement de Sante, and Mededns Sans Frontieres, we reviewed data collected in the hospitals of Farkhar and the ten surrounding kolkhozes (state-run farms serving 5000-10 000 people). Toxic liver injury was defined as such if recorded in the log book of the hospitals. Up to March, 1993, 3906 cases were recorded, an attack rate of 4%. The attack rates were respectively 0-4%, 54%, 40%, 28%, and 1 -5% for the age groups less than 1,1-14,15-30, 31-50, and older than 50 years. Within the same age groups, case fatality ratio increased with age: respectively 0%, 0-8%, 1 -3%, 2-3%, and 5-9%, and the overall case fatality ratio was 1 3%. 2 of the 10 kolkhozes contributed 83-3% of the cases (figure) with attack rates up to 16°9% and 23-6%, compared with less than 3% for each of the 8 others. Clinical signs and symptoms were recorded in 1102 cases. Four stages of illness were defined: I, abdominal pain, nausea or vomiting, and weakness; II, hepatomegaly; III, ascites; and IV, alteration of consciousness. Each stage was represented, respectively, by 55-5%, 29-9%, 13-7%, and 0-9% of the cases. The last case was reported on March 4,1993. At the time of the investigation, 300 cases were still in hospital and it is not known how many will later develop cirrhosis of the liver, to which veno-occlusive disease is known to be contributory factor.2 Toxic liver injury: hospital admissions 1992-93. The population was informed of the danger of eating contaminated bread early in the outbreak. However, because of the blockade and the related famine, they continued to eat the contaminated flour until the local authorities started an exchange for uncontaminated flour in mid-December, 1992. Similar outbreaks have been observed previously (India, 1977,3 Afghanistan, 1975,4 4 Iraq, 1970, South Africa and Jamaica in the 1950s, and USSR in the 1930s). Owing to the difficult political and economic situation faced by ex-USSR nations, such outbreaks may occur again. They should be detected early and recorded both to increase the awareness of the local population and to motivate prompt reaction from the international community. Unité de Recherche Biomathématiques et Biostatistiques, INSERM U263, Paris, France PIERRE CHAUVIN Institut National d’Agronomie, Départment de Nutrition Humaine, Paris-Grignon, France JEAN-CLAUDE DILLON Epicentre, Paris, France ALAIN MOREN General Hospital, Farkhar, Tadjikistan SAIDOV TALBAK Institute of Gastroenterology, Dushambe, Tadjikistan SALAM BARAKAEV 1. International Programme on Chemical Safety. Pyrrolizidine alkaloids and human health (Environmental Health Criteria 80). Geneva: WHO, 1988: 13-17. 2. Gupta PS, Gupta GD, Sharma ML. Veno-occlusive disease of the liver. BMJ 1963; i: 1184-86. 3. Tandon BN, Tandon HD, Koshy A, et al. Epidemiological, clinical, biochemical and hemodynamic study of veno-occlusive disease of the liver due to crotalaria alkaloids in India. J All Ind Inst Med Sci 1977; 3: 165-75. 4. Mohabbat O, Srivastava RN, Younos MS, et al. An outbreak of hepatic veno-occlusive disease in north-western Afghanistan. Lancet 1976; ii: 269-71. Adverse reactions and sumatriptan SIR,-I was interested in the two case-reports of Dr Luman and Dr Grey (April 24, p 1091). They correctly point out "that the half-life of sumatriptan is 2 hours" and "the intervals between drug administration and cerebral infarction of 1 week in the first case and 12 hours in the second, exonerate sumatriptan", but they then go on to state "no other risk factor for vascular disease was identified in our patients". However, migraine itself seems to be a risk factor. They also state that "the hemiparesis in the second case was on the opposite cortical side from the. migraine-associated "hypoaesthesia", but in their case-report both the numbness and the hemiparesis were right-sided. London Neurological Centre, 110 Harley Street, London W1N 1AF, UK F. CLIFFORD ROSE Author’s reply SIR,-We thank Dr Rose for his letter. Neither of our patients had typical migrainous headache before their hemiplegia. However, their headache could have been relieved by sumatriptan, though we are unable to explain the failure of sumatriptan to prevent migraine-induced hemiplegia (if this was indeed the case). With regard to the second patient, recent correspondence with her general practitioner reveals that she initially developed intermittent leftsided headache and right hypoaesthesia without motor signs 3 years ago when she was in her first pregnancy. The same symptom recurred a year later when she took the oral contraceptive pill. Before her mentioned admission she developed leftsided numbness on re-introduction of oral contraceptive and subsequently left hemiparesis 12 hours later after sumatriptan treatment. She was well on review a month later and did not have any residual motor or sensory signs. Her GP informed us that she has had a further recurrence of left hemiparesis and, after investigation, the provisional diagnosis is demyelination. This information was not available at the time of the report in April. The Committee on Safety of Medicines has received other reports of suspected neurological reactions, including hemiparesis, facial palsy, dysphasia, dysarthria, paraesthesia, and hemianopia