From February 2003 through January 2004, 2,342 patients were registered for TB treatment in Ubon-ratchathani Province. Of these, 225 (10%) were known to be HIV infected before their TB diagnosis. Of the remaining 2,117 patients, 1,626 (77%) received HIV pretest counseling, 680/1,626 (42%) agreed to HIV testing, and 104/680 (15%) were found to be HIV infected. In all, 329 (14%) of the 2,342 total TB patients were either known to be HIV infected before TB diagnosis (225; 68% of all TB/HIV patients) or were identified as HIV infected after testing through the TB program (104; 32% of all TB/HIV patients).
The median age of the 329 HIV-infected TB patients was 32 years (range 10 months–68 years), 112 (34%) were female, and 307 (93%) had new TB cases (). TB was classified as sputum smear–positive in 120 (36%), sputum smear–negative in 107 (33%), and extrapulmonary in 102 (31%). CD4 count was unavailable or not performed in 134 (41%). Of the 195 patients with CD4 results available, the median CD4 count was 53 cells/mm3 (range 1–873); 93% had CD4 <200 cells/mm3.
Characteristics of HIV-infected patients with tuberculosis (TB), Ubon-ratchathani, February 2003 through January 2004
Sputum cultures were performed in 145 (64%) of 227 patients with pulmonary TB, including 93 (78%) of 120 with sputum smear–positive TB and 52 (49%) of 107 with sputum smear–negative TB (). Of the 93 patients whose sputum smears were positive and who had a culture performed, 65 (70%) grew Mycobacterium tuberculosis (MTB); of these, 4 (6%) isolates were resistant to at least isoniazid and rifampin, i.e., MDR TB. Of the 52 sputum smear–negative patients with a culture performed, only 3 (6%) were culture positive, and none exhibited MDR TB.
Results of culture and susceptibility testing performed on HIV-infected patients with pulmonary TB, stratified by sputum smear–positive versus –negative results, Ubon-ratchathani, February 2003 through January 2004*
Before TB treatment, 30 (9%) patients were receiving ART; an additional 45 (14%) patients began ART during TB treatment; and the remaining 254 (77%) patients did not receive ART before or during TB treatment. In 40 of the 45 patients who began ART during TB treatment and in whom a date of starting ART was available, the median time between TB diagnosis and ART initiation was 93 days (range 0–170 days). Among all patients receiving ART, 38 (51%) received a combination regimen of stavudine, lamivudine, and nevirapine; 35 (47%) received efavirenz instead of nevirapine; and 2 (2%) were on other regimens. During TB treatment, 225 (68%) received co-trimoxazole.
Of all 329 patients, 187 (57%) were cured or completed TB treatment; 99 (30%) died during TB treatment. In the remaining 43 patients, treatment failed (for 4 patients) or the patient defaulted (a WHO term defined as missing at least 2 continuous months of treatment) (31 patients), transferred out (4 patients), or received a final diagnosis other than TB (4 patients). Of the 4 patients with MDR TB, 3 died and 1 was recorded as having failed treatment with final outcome not recorded.
In univariate analysis restricted to the 290 patients with an outcome of cured, completed treatment, failed treatment, or died, we analyzed several factors associated with death during TB treatment. For all TB patients, having an unknown CD4 count was associated with increased likelihood of death, and receiving co-trimoxazole or ART was associated with reduced mortality (). For ART, 5 (7%) of 71 patients who received ART died compared with 94 (43%) of 219 patients who did not receive ART (RR 0.2; 95% confidence interval [CI] 0.1–0.4; absolute risk reduction 36; number-needed-to-treat 2.8). For sputum smear–positive TB patients, results were similar; additionally, male patients were at higher risk for death than female patients (RR 2.3, 95% CI 1.1–4.7).
Table 3 Univariate analysis of risk factors for death among HIV-infected TB patients with outcomes of cured, completed, failed, or died, stratified by all patients versus pulmonary, smear-positive patients, Ubon-ratchathani, February 2003 through January 2004* (more ...)
In multivariate analysis adjusted for CD4 count, smear status, hospital providing treatment, and co-trimoxazole use, ART remained strongly associated with reduced mortality during TB treatment (). The adjusted OR (aOR) for death in patients who received ART before or during TB treatment was 0.2 (95% CI 0.1–0.5) compared with that in patients who did not receive ART. Receiving co-trimoxazole was no longer significantly associated with reduced mortality. We found virtually identical results when we did the following: 1) restricted our analysis to only those patients who received ART during TB treatment compared with patients who did not receive ART during TB treatment; 2) restricted our analysis to previously untreated, non-MDR patients without nontuberculous mycobacteria; 3) coded patients with unknown CD4 as having CD4 >200 cells/mm3, as having CD4 <50 cells/mm3, or as missing (i.e., removed from the analysis). All analyses also produced essentially identical results when we reclassified cases of default as death.
Multivariate analysis of risk factors for death among HIV-infected TB patients with outcomes of cured, completed, failed, or died and adjusted for site of treating facility,* Ubon-ratchathani, February 2003 through January 2004†
When we restricted our analysis to sputum smear–positive patients, we found a similarly strong beneficial effect for ART. Because no deaths occurred in the group of smear-positive patients with CD4 >200 cells/mm3, we modeled CD4 as a continuous variable. The aOR was 0.1 (95% CI 0.0–0.9). Results were similar when we recoded patients with unknown CD4 count as having CD4 equal to 50 cells/mm3 (indicative of profound immunosuppression and imminent risk of death) or 250 cells/mm3 (not eligible for antiretroviral treatment in many country guidelines because they are relatively immune competent). Because of small sample size, we were only able to perform univariate, not multivariate, analysis for the 57 culture-positive patients who had an outcome of cured, completed, failed, or died. One (8%) of 13 culture-positive patients receiving ART died compared with 17 (36%) of 47 culture-positive patients not receiving ART (RR 0.2, 95% CI 0.0–1.5; absolute risk reduction 28; number-needed-to-treat 3.6).
Because patients who died soon after TB diagnosis were also unlikely to have begun ART, we modeled the effect of ART after excluding 32 patients who died in the first month of beginning TB treatment (aOR 0.1, 95% CI 0.0–0.8) and the additional 16 patients who died in the second month (aOR 0.1, 95% CI 0.0–1.2). Results were similar when we recoded patients with unknown CD4 count as having CD4 equal to 50 cells/mm3 or 250 cells/mm3, except that the effect of ART was now statistically significant for the analysis excluding deaths within the first and second months (aOR 0.1, 95% CI 0.0–0.7 for unknown CD4 re-coded as 50 cells/mm3; aOR 0.1; 95% CI 0.0–0.9 for unknown CD4 recoded as 250 cells/mm3).
We explored the effect of co-trimoxazole on mortality for the 218 patients who did not receive ART: 52 (38%) of 137 patients receiving co-trimoxazole died compared with 42 (51%) of 82 patients not receiving co-trimoxazole (RR 0.7, CI 0.6–1.0; absolute risk reduction 13; number-needed-to-treat 7.7). The association between co-trimoxazole and survival was not statistically significant when we excluded patients who died in the first month (RR 0.9, CI 0.5–1.5) or in the first 2 months (RR 1.2, CI 0.5–3.0) and when we limited the analysis to smear-positive patients (RR 0.8, CI 0.5–1.3). In multivariate analysis of patients who did not receive ART and adjusting for CD4 count, smear status, and hospital providing treatment, co-trimoxazole was not associated with survival (aOR 0.9, CI 0.5–1.9).