Table 4– Performance outcomes of smear microscopy, lipoarabinomannan (LAM) ELISA and LAM strip-test (using grade 1 and 2 cut-off points) in HIV-infected patients stratified by CD4 count (>200 and ≤200 cells per mL)
Diagnostic test(s)HIV-infected patients with CD4 count >200 cells per mLHIV-infected patients with CD4 count ≤200 cells per mL
Group 1 patientsGroup 2 patientsGroup 1 patientsGroup 2 patients
Microbiological reference (M. tuberculosis culture positive versus negative)Composite reference standard (definite- and probable-TB versus non-TB)Microbiological reference (M. tuberculosis culture positive versus negative)Composite reference standard (definite- and probable-TB versus non-TB)
SensitivitySpecificitySensitivitySpecificitySpecificitySensitivitySpecificitySensitivitySpecificitySpecificity
Subjects n55558417314784
Rapid smear microscopy58 (39–75) 15/2690 (79–100) 26/2938 (25–52) 17/4595 (55–100) 9/10No results56++ (45–66) 45/8191##,§§ (86–97) 84/9244## (36–53) 58/131100 (79–100) 16/16No results
LAM ELISA27*** (14–46) 7/2689** (77–100) 24/2728*** (17–43) 12/43100 (69–100) 10/10100** (93–100) 53/5370** (59–79) 56/8076§§,** (67–85) 69/9159*** (50–67) 76/12994 (70–100) 15/1697** (90–100) 28/29
LAM strip-test (grade 1 cut-off point)54 (36–71) 14/2676 (60–91) 22/2949 (35–63) 22/45100 (69–100) 10/1091¶¶ (83–99) 49/5472 (61–80) 58/8161# (51–71) 56/9264 (56–72) 84/13194 (70–100) 15/1687# (75–99) 26/30
LAM strip-test (grade 2 cut-off point)27 (14–46) 7/2683§ (69–97) 24/2929+ (18–43) 13/45100 (69–100) 10/10100§,¶¶ (93–100) 54/5458¶,¶¶ (47–68) 47/8172ƒ (63–81) 66/9252+,++ (43–60) 68/13194 (70–100) 15/1697ƒ (90–100) 29/30
LAM strip-test (grade 2 cut-off point) in smear-negative/sputum scarce M. tuberculosis culture-positive patients0 (0–26) 0/11N/A14 (6–32)N/AN/A51 (36–66) 21/41N/A43 (32–54) 31/73N/AN/A
Combined sputum smear microscopy and LAM strip-test (grade 2 cut-off point)58 (39–75) 15/2679 (65–94) 23/2947** (33–61) 21/4595 (55–100) 9/10No results75¶¶,++ (65–83) 61/8170**,## (60–79) 64/9268**,##,++ (60–75) 89/131100** (79–100) 16/16No results

  • Data are presented as sensitivity/specificity (95% CI) n/ntotal, unless otherwise stated. N/A: not applicable. Using the microbiological reference, 14 out of the 242 HIV-infected group 1 patients had no CD4 cell count data, leaving 228 patients for the above analysis. 107 Mycobacterium tuberculosis (M. tuberculosis) culture-positive group 1 patients were used for the sensitivity calculations and 121 M. tuberculosis culture-negative group 1 for specificity calculations. Using the composite reference, 12 out of of the 218 HIV-infected group 1 patients had no CD4 cell count data, leaving 202 patients for the above analysis. 176 definite- and probable-tuberculosis (TB) were used for the sensitivity calculations and 26 non-TB patients for specificity calculations. Additionally, 84 non-TB group 2 patients were used for a second specificity calculation. Rapid (within 24 h of hospitalisation) smear microscopy includes both sputum and non-sputum samples (e.g. fine-needle aspirate and cerebrospinal fluid samples). Two CD4 >200 cells per mL and CD4 ≤200 cells per mL group 1 patients (analysis 1) and three CD4 >200 and CD4 ≤200 (analysis 2) had a missing LAM ELISA result while one CD4 >200 (analysis 2) had a missing LAM strip-test result. For a comparison of differences in diagnostic accuracy measures between analyses and patient groups for a specific test (LAM ELISA or LAM strip-test) p-values were p<0.05; specific p-values: **: p =0.01; ***: p <0.001; #: p =0.009; : p =0.006; +: p =0.008; §: p =0.002; ƒ: p =0.004; nonsignificant p-values are not shown. For a comparison of the specific measures of diagnostic accuracy (sensitivity or specificity) between different tests (LAM ELISA versus LAM strip-test) or combinations thereof, or between different cut-off points of the urine LAM strip-test (grade 1 versus grade 2 cut-off point) p-values were p<0.05; specific p-values: ##: p<0.001; ¶¶: p=0.02; ++: p=0.008; §§: p=0.005; nonsignificant p-values not shown.