ORIGINAL STUDIES Recurrent Culture-Confirmed Tuberculosis in Human Immunodeficiency Virus-Infected Children H. Simon Schaaf, MD Paed,* Sara Krook,‡ Dete W. Hollemans,‡ Robin M. Warren, PhD,† Peter R. Donald, MD Paed,* and Anneke C. Hesseling, MbChB, MSc Epidemiol* Introduction: Recurrent tuberculosis (TB) is more common among human immunodeficiency virus (HIV)-infected than HIV- uninfected adults. There are limited data regarding recurrence of TB in children. Objective: To determine the occurrence of recurrent TB in HIV- infected children with culture-confirmed tuberculosis. Methods: HIV-infected children with culture-confirmed TB, iden- tified from 1992 to 2000, were followed until February 2004 for further confirmed TB episodes 6 months or more after completion of previous antituberculosis therapy. Clinical data and results of special investigations were recorded. Restriction fragment length polymor- phism (RFLP) analysis of Mycobacterium tuberculosis isolates was done when possible. Results: Of 87 children, 9 had a second episode; 2 of these had a third episode of confirmed TB. Adherence to treatment was good in 8; 2 experienced hepatotoxicity, and regimens were changed. Chest radiographs were normal in only 2 children after first treatment completion. Bacteriologic cure was documented in 7 episodes before recurrence. RFLP analysis showed 3 children infected with the same strain (relapse) and 1 child with a different strain between episodes 1 and 3 (reinfection). Two further cases had reinfection based on epidemiologic data and drug susceptibility test results. Full compar- ison of strains by RFLP was not possible because of the unavail- ability of isolates of the first episode in 5 cases. Conclusion: Recurrent TB in HIV-infected children is common in a high burden TB setting. Both relapse and reinfection occur. Key Words: tuberculosis, recurrent, relapse, reinfection, children, human immunodeficiency virus (Pediatr Infect Dis J 2005;24: 685– 691) T uberculosis (TB) recurs in 2–7% of human immunodefi- ciency virus (HIV)-uninfected patients with drug-suscep- tible Mycobacterium tuberculosis strains with current short course chemotherapy. 1–3 Recurrence of TB has been well- described in HIV-uninfected and HIV-infected adults. 4–8 Endogenous reactivation involving the original strain (re- lapse) and exogenous reinfection with a new strain (reinfec- tion) both occur and have been confirmed by restriction fragment length polymorphism (RFLP) analysis. 9 Reinfec- tion is more common against a background of high TB incidence, whereas relapse is influenced by inadequate treat- ment regimens or poor patient adherence to therapy. HIV infection, as a cause of severe immunosuppression, may cause an increase in both. TB is a common opportunistic infection in HIV-in- fected children in high TB burden settings. Poor response to treatment and recurrence of TB after treatment completions have been reported. 10 –12 However, in children, as was pre- viously the case in adults, it is unknown whether recurrence in HIV-infected patients is as a result of relapse or reinfection or both. Further, little is known about the risk factors for recurrence. We previously described a case of relapse of TB in an HIV-infected child. 13 The aim of this study, which includes the previous case report, was to determine the occurrence of recurrent TB among a cohort of human immu- nodeficiency virus-infected children with culture-confirmed tuberculosis. METHODS Study Design and Setting. This is a retrospective descriptive study based in the Western Cape Province of South Africa, an area with a high incidence of tuberculosis (500/100,000 population in 2000). 14 The HIV prevalence in the Western Cape Province among women attending the public sector antenatal clinics increased from 0.7% in 1992 to 8.7% in 2000. 15 All children younger than 13 years of age diagnosed with both HIV infection and culture-confirmed tuberculosis at 2 referral hospitals, Tygerberg Children’s Hospital (January 1992–December 2000) and Brooklyn Hospital for Chest Dis- eases (January 1998 –December 2000), were eligible. All identified cases were treated for tuberculosis according to the National Tuberculosis Control Guidelines with 3 or 4 antitu- berculosis drugs including isoniazid and rifampin in both the intensive and continuation phases of treatment. Children were followed from the time of their first culture-confirmed epi- sode of tuberculosis until the end of the study period in February 2004. End points during the follow-up period in- cluded death during or subsequent to antituberculosis ther- apy, loss to follow-up (that is not returning to any regional health facility) or being alive at the end of the study period. Only children with a culture-confirmed further episode of tuberculosis were included in this study. Accepted for publication March 2, 2005. From the *Department of Paediatrics and Child Health and the †MRC Centre for Molecular and Cellular Biology, Department of Medical Biochemis- try, Stellenbosch University, Tygerberg, South Africa; and the ‡Univer- sity of Amsterdam, Amsterdam, the Netherlands Reprints not available. Copyright © 2005 by Lippincott Williams & Wilkins ISSN: 0891-3668/05/2408-0685 DOI: 10.1097/01.inf.0000172933.22481.36 The Pediatric Infectious Disease Journal • Volume 24, Number 8, August 2005 685