CROI 2016 Abstract eBook

Abstract Listing

Poster Abstracts

739 Diagnostic Yield of Household-Level Active Case Finding in Urban Slums in Haiti Vanessa R. Rivera 1 ; JeanW. Pape 2 ; Serena P. Koenig 3 ; Marc Antoine Jean Juste 2 1 Weill Cornell Med Coll, New York, NY, USA; 2 GHESKIO, Port-au-Prince, Haiti; 3 Brigham and Women’s Hosp, Harvard Med Sch, Boston, MA, USA Background: Despite major advances in diagnostic and treatment services in settings where TB is endemic, many people with active TB remain undiagnosed. We evaluated the diagnostic yield of TB active case finding (ACF) at the household level; the programwas funded by TB REACH. Methods: From August 1, 2014 to July 31, 2015, community health workers (CHWs) screened residents in 8 slum communities for cough >2 weeks. Household GPS coordinates were recorded using smart-phone technology for TB mapping purposes. All TB suspects were referred to a TB clinic for HIV and TB screening. All patients with confirmed cough >2 weeks were evaluated for TB with CXR and sputum acid fast bacilli (AFB) smear and Xpert tests. Patients diagnosed with TB were started on treatment within 5 days of initial referral. Results: 103,000 individuals were screened for cough through ACF activities. 6,926 (7%) reported cough >2 weeks and were referred for physician evaluation. Cough was confirmed in 3,397 (49%) of the patients referred, and 3,147 (93%) received smear and Xpert testing and were offered HIV-testing. 302 (10%) were HIV-positive, 2,644 (84%) were HIV-negative, and 201 (6%) had indeterminate HIV test results or were not tested for HIV. 90 of 302 HIV-infected patients (30%) were diagnosed with TB; 47 were smear and Xpert positive, 25 were smear negative and Xpert positive, and 18 were smear and Xpert negative but diagnosed with clinical and radiographic criteria. 571 of 2,644 HIV-negative patients (22%) were diagnosed with TB; 408 were smear and Xpert positive, 18 were smear positive and Xpert negative, 104 were smear negative and Xpert positive, and 41 were smear and Xpert negative but diagnosed with clinical and radiographic criteria. 661 patients with cough who received HIV, smear and Xpert testing were diagnosed with TB, and 129 (20%) were smear-negative but Xpert-positive. TB mapping revealed clusters of higher prevalence of cough and diagnosed TB at the block-level within slum communities. Conclusions: Household-level screening for cough by CHWs was highly effective; 21% of those tested were diagnosed with active TB. A high rate of TB was found among HIV- positive and HIV-negative patients, and Xpert testing resulted in 27% additional diagnoses of bacteriologically-confirmed TB among both groups. Further studies are needed to determine the cost-effectiveness of these strategies. 740 Integration of HIV-TB Screening and Linkage Strengthens Community-Based HIV Care Ruanne V. Barnabas 1 ; Heidi van Rooyen 2 ; Stephen Asiimwe 3 ;Torin Schaafsma 1 ; Meighan Krows 1 ; Alastair van Heerden 4 ; BoscoTuryamureeba 3 ; James P. Hughes 1 ; Jared M. Baeten 1 ; Connie M. Celum 1 ; for the Linkages StudyTeam 1 Univ of Washington, Seattle, WA, USA; 2 Human Scis Rsr Council, Durban, South Africa; 3 Kabwohe Clinical Rsr Cntr, Bushenyi, Uganda; 4 Human Scis Rsr Council, Msunduzi, South Africa Background: In sub-Saharan Africa the burden of HIV-associated tuberculosis is high; early antiretroviral therapy (ART) combined with isoniazid preventive therapy (IPT) reduces HIV-associated morbidity and mortality. In generalized HIV epidemics in Africa, community-based HIV testing and counseling (HTC) links HIV-positive persons to care; integration of TB symptom screening into HTC can link HIV-positive persons to diagnostic testing for active TB or initiation of IPT and avert disability. Methods: We conducted a multisite program of community-based HIV testing and counseling, linkage to HIV care, and standardized WHO TB symptom screening in rural communities in KwaZulu-Natal, South Africa and Sheema district, Uganda. HIV testing was done at home or through mobile units. HIV-positive persons received the TB symptom screening and were referred to local clinics for care and diagnostic testing for active TB. At follow-up visits participant linkage to TB diagnostic testing, treatment for active TB, and IPT was assessed. Results: Between June 2013 and February 2015, 15,332 persons received HIV testing and counseling. Among 1,325 HIV-positive persons identified, the median CD4 count was high (486 cells/mL). At enrollment, 976 (74%) participants reported no symptoms of active TB and 157 (12%), 107 (8%), 64 (5%), and 21 (1%) reported 1, 2, 3 and 4 symptoms of TB. Linkage to HIV clinics was high (93%). After 9 months of follow-up, 113/346 (33%) persons with .≥1 TB symptom had sputum collected for TB testing vs. 266/957 (28%) among participants with no symptoms (RR=1.26, 95% CI 1.07-1.48). Of those tested for TB, 63% (240/379) reported they had received their sputum results. Among participants reporting symptoms at enrollment and tested, 15/345 (4%) were diagnosed with active TB vs. 11/957 (1%) of participants reporting no symptoms (RR=4.11, 95% CI 1.98-8.54). Only 54% (14/26) of HIV infected persons diagnosed with active TB initiated TB treatment. Ten percent (34/338) of participants who reported symptoms and 13% (119/951) of persons without TB symptoms initiated IPT. Conclusions: Among asymptomatic HIV-positive persons identified through community-based HIV testing, TB symptom screening increased the likelihood of TB diagnostic testing and TB diagnosis, but linkage to TB diagnostic testing, treatment, and IPT is low. Community-based HIV care requires efficient diagnostic strategies, such as rapid TB testing at HIV diagnosis, and effective linkage strategies to address low uptake of IPT and TB treatment. 741LB WITHDRAWN 742 A Clinical Prediction Rule for the Diagnosis of Tuberculosis in Seriously Ill Adults Rulan Griesel 1 ; Annemie Stewart 2 ; Helen van der Plas 3 ;Welile Sikhondze 4 ; Molebogeng Rangaka 5 ; Gary Maartens 1 ; Marc Mendelson 1 1 Univ of Cape Town, Cape Town, South Africa; 2 Clinical Rsr Cntr, Univ of Cape Town, Cape Town, South Africa; 3 Vincent Pallotti Hosp, Cape Town, South Africa; 4 FIND, Geneva, Switzerland; 5 Univ Coll London, London, UK Background: The World Health Organization’s (WHO) algorithm for the diagnosis of tuberculosis in seriously ill HIV-infected patients with danger signs (any one of respiratory rate >30/min; heart rate >120/min; temperature >39 o C; unable to walk unaided) and cough for ≥14 days uses chest x-ray and sputum smear results to start empiric antituberculosis therapy. The WHO algorithm preceded the availability of the rapid Xpert MTB/RIF assay. We aimed to develop a clinical prediction rule (CPR) for the diagnosis of tuberculosis by determining an evidence base for the duration of cough, the role of other tuberculosis symptoms, and simple laboratory tests (haemoglobin and white cell count). In addition we determined the diagnostic performance of Xpert MTB/RIF. Methods: A prospective cohort study was conducted at 2 secondary level hospitals in Cape Town, South Africa. Inclusion criteria were: HIV-infected, cough (any duration), WHO danger signs, age ≥18 years, and able to produce spontaneous/induced sputum. Chest x-rays were assessed by a specialist radiologist and categorised as unlikely, possibly or likely tuberculosis. Culture of M. tuberculosis from blood or sputum (2 samples sent) was the reference standard for the diagnosis of tuberculosis. In a multivariable model we assessed the ability of the following a priori chosen variables to predict the diagnosis of tuberculosis: WHO danger signs; duration of cough; tuberculosis symptoms (fever, night sweats, and weight loss); chest x-ray assessment; haemoglobin; and white cell count. The most predictive variables were used to establish a CPR for the diagnosis of tuberculosis. Results: 484 participants were enrolled into the study: median age 36 years; 317 female; median CD4 count was 89 cells/μL (IQR 34-210); and 171 on ART. 256 participants were culture positive for tuberculosis. Sputum smear had a sensitivity of 57.0% and a specificity of 98.7%. Xpert MTB/RIF had a sensitivity of 86.3% and a specificity of 96.1%. The final model included the following variables: cough ≥14 days, temperature >39 o C, being unable to walk unaided, chest x-ray assessment, haemoglobin, and white cell count. Chest x-ray assessment of “likely tuberculosis” and anaemia were the strongest predictors of tuberculosis. The ROC AUC for the CPR was 0.81 (95% CI 0.80-0.82). The CPR is depicted in the table. Conclusions: The CPR could facilitate rapid initiation of empiric tuberculosis therapy in seriously ill patients using simple measures. Xpert MTB/RIF performed well in this population.

Poster Abstracts

308

CROI 2016