Journal of Clinical Epidemiology 55 (2002) 235–238 0895-4356/02/$ – see front matter © 2002 Elsevier Science Inc. All rights reserved. PII: S0895-4356(01)00470-X Urine testing to monitor adherence to TB preventive therapy Sharon Perry a , Melbourne F. Hovell b, *, Elaine Blumberg b , Jill Berg c , Alicia Vera b , Carol Sipan b , Norma Kelley b , Kathleen Moser d , Antonino Catanzaro e , Larry Friedman e a University of California San Francisco, San Francisco, CA, USA b Center for Behavioral Epidemiology and Community Health, San Diego State University, 9245 Sky Park Court, Suite 230, San Diego, CA 92123, USA c University of California Los Angeles, Los Angeles, CA, USA d Tuberculosis Control Unit of the San Diego County Health and Human Services Agency, San Diego, CA, USA e University of California San Diego Medical Center, San Diego, CA, USA Received 13 March 2001; received in revised form 14 August 2001; accepted 30 August 2001 Abstract This study examined the validity of the Arkansas urine test. One hundred ninety-four adolescents submitted an unannounced urine specimen monthly (for 6 to 8 months). Duplicate specimens were blindly tested with high agreement (kappa 90%). Sensitivity and specificity were esti- mated. In 68% of test runs, adolescents recalled taking INH within 24 hr of specimen collection. For recall intervals of 24, 48, and 72 hr, sensitivity was 87, 85, and 83%, respectively. Females were less likely to test positive when INH was taken within the previous 24 hr (sensitivity 84 versus 92% males). Specificity was 57, 91, and 95% at 24, 48, and 72 hr, respectively. The Arkansas urine test was practical to use, and results correlated well with self-reported adherence to INH for treatment of latent tuberculosis infection (LTBI), over several months of follow-up. The test may be useful as part of an adherence-monitoring program when used in conjunction with self-reported measures. © 2002 Elsevier Science Inc. All rights reserved. Keywords: Adherence; Latent tuberculosis infection (LTBI); Youth; Validity of Arkansas urine test; INH; Sensitivity/specificity 1. Introduction Although the incidence of tuberculosis (TB) has declined in many U.S. communities, rates continue to rise among im- migrants from high prevalence countries [1]. During 1997, the state of California accounted for one out of five of all new cases of TB reported in the United States, and two out of three of these cases occurred among persons of foreign birth. The majority of these new cases occurred in individu- als from Latin America or Southeast Asia [2,3]. In the bor- der community of San Diego, CA, prevalence of latent tu- berculosis infection (LTBI) in high-risk adolescents has been estimated to range between 10 and 20% [4,5]. Preven- tion programs targeted to these high-risk groups are consid- ered a key element of national and local TB control strate- gies in the United States [3]. Older U.S. studies have estimated that between 60 and 70% of individuals with LTBI complete treatment in a year [6,7]. However, the effectiveness of programs targeted to contemporary high-risk groups is less well understood. In part, these efforts have been hampered by a lack of informa- tion on the validity of different adherence measures for field research and clinical practice [3]. Self-reported adherence, pill counts, electronic monitoring devices, and more author- itarian techniques such as metabolite testing or directly ob- served therapy each have important disadvantages. As part of a behavioral intervention trial carried out in collaboration with San Diego schools and clinics [8], we have used the Potts-Cozart (Arkansas) urine test [9] to mea- sure adherence to isoniazid (INH) prophylaxis in Latino ad- olescents. The assay, a modification of the Belles-Littleman filter paper spot test [10,11], measures INH and the metabo- lite acetylisoniazid colorimetrically. The test is simple to perform, quick, robust, and in controlled studies has a sensi- tivity and specificity approaching 100% when correlated with clinical history [12,13]. Although urine testing is a po- tentially useful objective measure of adherence, experience in targeted prevention trials is limited [14]. The purpose of the present study was to examine the reli- ability and validity of the Arkansas test when compared to self-reported adherence in adolescents enrolled in a pre- scribed program of INH treatment for 6 to 9 months. To our knowledge, the Arkansas test has not been validated for high-risk Latino adolescents receiving treatment for LTBI. * Corresponding author. Tel.: 858-505-4772; fax: 858-505-8614. E-mail address:mhovell@projects.sdsu.edu (M.F. Hovell)