a sensitivity of 92% similar to the report of Mestitz et al., 7 but unlike the results of Onadeko. 8 It may, however, be the only way of establishing a diagnosis especially in the very ill patients. In none of the patients with tuberculous pleurisy was acid-fast bacilli identified by the Ziehl–Neelsen staining of the pleural aspirate. This may not be surprising, as positive cultures for the acid-fast bacilli are found in only 30% of patients with tuberculous pleurisy. 8 However, this should not justify empirical antiTB therapy, as our study revealed no statistical significance between pleural effusion of undetermined aetiology and TB. This finding is different from that of Onadeko. 8 It may be due to improvement in the drug treatment of TB. Pleural effusion of uncertain aetiology is malignant in 29.4%, and this is significant (P ¼ 0.022). Needle pleural biopsies, in our experience, established this in 91% of cases with 9% false negative (Table 1). Seventy-two percent of these malignant effusions will have the lung as the primary site. This finding of eight cases of lung malignancy within an 18-month period is at variance with five cases of Onadeko 8 over a 10-year period; it certainly shows that malignancy of the lung is becoming increasingly common amongst Nigerians. Conclusion The empirical use of antiTB drugs in pleural effusion of undetermined aetiology should be discouraged, as it tends to delay efforts at further investigations for causes of such effusion. References 1 Anyanwu CH, Egbue M. Management of pleural sepsis in Nigerian children. Thorax 1981;36:282–5 2 Anyanwu CH, Okeahialam TC, Okoroma EO. Postpneumo- nic pleural suppuration in children. Trop Doct 1983;13: 57–60 3 Osinowo O, Adebonojo SA, Adebo O, et al. Childhood empyema in Ibadan, Nigeria. Niger Med J 1982;12: 337–347 4 Fenton KN, Richardson JD. Diagnosis and management of pleural effusions. Am J Surg 1995;170:69–74 5 Abrams LD, Birm MB. A pleural biopsy punch. Lancet 1958;i:30 6 Baumgartner WA, Mark JBD. The use of thoracoscopy in the diagnosis of pleural disease. Arch Surg 1980;115:420–1 7 Mestitz P, Purves MJ, Pollard AC. Pleural biopsy in the diagnosis of pleural effusion. A report of 200 cases. Lancet 1958;ii:1349 8 Onadeko BO. A preliminary study of the pattern of pleural effusion in Africans. Niger Med J 1977;7:138–43 Does long-term chloramphenicol cause anaemia in Malawi? A C Peek MBBS C B D Lavy MCh FRCS M Thyoka MBBS A Pitani DipOrth Department of Orthopaedic surgery, Queen Elizabeth Central Hospital, PO Box 95, Blantyre, Malawi Correspondence to: Professor C B D Lavy Email: lavy@malawi.net TROPICAL DOCTOR 2006; 36: 114–115 SUMMARY Common clinical practice in many tropical paediatric departments is that chloramphenicol courses are limited to 2 weeks due to concerns about anaemia. However, this approach is not supported by current research and animal models. We used chloramphenicol for 6 weeks in 146 children with septic arthritis. All the children improved clinically. Most children were anaemic on presentation (mean haemoglobin [Hb] 8.43 SD 1.9), but the anaemia improved rapidly with clinical resolution of the infection and was maintained at 6 months after pre- sentation (mean Hb10.57 SD1.86). Introduction Chloramphenicol is associated with two well-documented haematological side effects. The first is a rare idiosyn- cratic, unpredictable and potentially fatal aplastic anae- mia, reported to occur in one in 30,000 treatment courses. 1 The second is a common, dose-related, reversible anaemia, due to mitochondrial synthesis inhibition. 1 However, despite this, it is still commonly used in tropical Africa and other areas, as a cheap and effective antibiotic, in serious infections that are sensitive to it. The reversible anaemia is not a precursor stage to the aplastic anaemia, as it seems committed progenitor cells are affected rather than stem cells, as seen in aplastic anaemia. 2 Animal models have now been developed to investigate these effects. 2,3 As a result of concerns regarding anaemia, there is a commonly held idea in tropical paediatrics that chloramphenicol should not be given to children for more than 2 weeks. We have heard this advice in several occasions but were unable to find any strong evidence to support it. Murine models have revealed a mild anaemia with reticulocytopenia which develops during treatment, but which rapidly reverses upon cessation of treatment, and also, interestingly, a ‘tolerance’ effect which develops during treatment, with blood parameters returning to normal. 4 The mechanism of this effect is unclear, but it may be that chloramphenicol may disrupt its own metabolism, as chloramphenicol blood levels have been observed to drop with prolonged treatment. 4 It seems that the accepted practice of limiting chloramphe- nicol courses in children to 2 weeks is therefore not evidence- based and we report our experience of using chloramphenicol for 6 weeks in 146 children. This report does not aim to investigate the rare incidence of aplastic anaemia. Short Reports 114 Tropical Doctor April 2006, 36