Table 3.

Predictors of Confirmed Tuberculosis (n = 110 vs 56): Multivariate Logistic Regression

Predictor	Odds Ratios	95% CI	P Value
Age, per 10 years older	0.72	0.51–1.00	.054
Female versus male	0.75	0.38–2.67	.971
Body mass index, per 5 kg/m2 increase	0.87	0.74–1.03	.087
HIV infection	0.61	0.21–1.75	.361
Cough	2.63	0.50–13.7	.251
Night sweats	0.68	0.25–1.84	.444
Weight loss	0.77	0.22–2.74	.688
Temperature, per each °C increase	2.48	1.52–5.03	.001
Pulmonary signsa	1.10	0.39–3.05	.859
Abdominal signsb	0.71	0.26–1.90	.493
Lymphadenopathyc	1.68	0.58–4.83	.338
Abnormal chest x-rayd	6.19	1.96–19.6	.002
≥1 FASH sign	3.33	1.21–9.12	.019

Odds ratios were adjusted for all variables listed; ≥1 FASH sign, presence of at least 1 original FASH sign. P values in bold numbers indicate that the association is statistically significant.

Abbreviations: CI, confidence interval; FASH, focused assessment with sonography for HIV-associated tuberculosis; HIV, human immunodeficiency virus.

^aPulmonary signs included crackles, wheezing, pleural friction in lung auscultation, or dullness in lung percussion.

^bAbdominal signs included organomegaly, ascites, and abnormal bowel sound.

^cLymphadenopathy was diagnosed if palpable enlarged axillary, cervical, or nuchal lymph nodes were present on physical examination.

^dAbnormal chest radiogram included any infiltrate, cavernous lesions, miliary pattern, or pleural effusion.