We identified patients reported to the Oregon Health Division with pulmonary TB during 2005–2006 who lived within the Portland metropolitan region (Clackamas, Multnomah, and Washington Counties). This region had a combined population of ≈1.55 million in 2005–2006 (7
). In 2000, the predominant ethnicity in this region was white (75.8%), followed by Hispanic or Latino (11.4%), Asian (6.3%), and black (3.6%), and 11.9% of the population had been born outside the United States (7
). From a statewide surveillance project, we identified all tri-county residents with NTM respiratory isolates obtained during the same period and then used pulmonary NTM disease criteria of the American Thoracic Society/Infectious Diseases Society of America to define cases of pulmonary NTM disease (3,8
). For each pulmonary TB and NTM case-patient within the tri-county region, we collected demographic information. From physician records, we collected clinical data. We conducted this project under the authority of the Oregon Administrative Rules for special studies to control a public health problem.
We used SAS version 9.1 (SAS Institute Inc., Cary, NC, USA) to compare categorical variables in univariate fashion by the χ2
or Fisher exact tests. We calculated the relative proportion (RP) of TB patients with each risk factor compared to the proportion of NTM patients with the risk factor. We used the Student t
test to evaluate continuous variables. We considered factors with a p value <0.2 for multivariate logistic regression and performed stepwise backward elimination of variables not reaching levels of statistical significance (p<0.05). Using significant variables from our multivariate model, we calculated the positive predictive value (PPV) and 95% exact binomial confidence intervals (CIs) of variables, alone and in combination, for distinguishing TB from NTM disease. Age was dichotomized (<
50 and >50 years) based on the age of NTM case-patients to simplify calculation of PPV (9
Eighty-two pulmonary TB patients were reported; all but 2 had complete clinical records for review. We identified 407 patients with respiratory NTM isolates. Clinical records were present for 283 (69.5%) of these patients, of whom 127 (44.9%) met clinical criteria of the American Thoracic Society/Infectious Diseases Society of America for pulmonary NTM disease (8
). Fifty-four patients lacked information on country of birth. In patients for whom smear data was available, no important difference was found in proportion of case-patients with smear-positive results (38/79 [46%] of TB case-patients vs. 28/47 [60%] of NTM case-patients). In comparison to NTM case-patients, TB patients were younger (median age 44 years, range 5–86 years vs. 67 years, range 12–92 years; p<0.01), more likely to be male (RP 1.6 , 95% CI 1.2–2.2, p<0.01), and more likely to have been born outside the United States (RP 4.0, 95% CI 2.5–6.3, p<0.01) (). Mycobacterium avium-intracellulare
complex was the most common etiologic agent of NTM disease in our cohort (114 [90%]).
Demographic, clinical, and radiographic features of TB patients compared with NTM patients, Oregon, USA, 2005–2006*
Clinically, TB patients were more likely to report constitutional symptoms (56 [70%] vs. 61 [48%], RP 1.5, 95% CI 1.2–1.8, p<0.01), less likely to have chronic obstructive pulmonary disease (COPD) (2 [3%] vs. 29 [23%], RP 0.1, 95% CI 0.0–0.4, p<0.01]), and less likely to be using immunosuppressive medications than NTM patients (8 [10%] vs. 34 [27%], RP 0.4, 95% CI 0.2–0.8, p<0.01) (). The most common immunosuppressive medications were systemic corticosteroids (30 patients [14%]). Patients with TB were more likely to have cavitation (18 [23%] vs. 11 [9%], RP 2.7, 95% CI 1.3–5.3, p<0.01) and infiltrate reported (68 [87%] vs. 69 [54%], RP 1.6, 95% CI 1.3–1.9, p<0.01) on chest radiograph ().
Birth outside the United States (odds ratio [OR] 26.3, 95% CI 9.9–69.6, p<0.01), constitutional symptoms (OR 3.0, 95% CI 1.1–8.0, p = 0.03), and infiltrate on chest radiograph (OR 7.8, 95% CI 2.6–23.9, p<0.01) were significantly associated with TB in multivariate analysis. Age was inversely related to the likelihood of having TB with an OR of 0.95 (95% CI 0.93–0.98, p<0.01) for each year increase in age. Because of its clinical significance, COPD (OR 0.3, 95% CI 0.1–1.7, p = 0.19) was maintained in the multivariate model. Four patients with missing covariate data were excluded ().
In our predictive model, age <50 years and birth outside the United States together were highly predictive for TB (PPV 0.98, 95% CI 0.88–1.0). COPD was poorly predictive of TB (PPV 0.06, 95% CI 0.01–0.21). Age >50, US-born status, and COPD together had a PPV for TB of 0.08, 95% CI 0.00–0.38 (; ).
PPVs of patient characteristics for tuberculosis in Oregon, USA, an area of low tuberculosis incidence, 2005–2006*
Positive predictive values (PPV) for tuberculosis of demographic and clinical factors in combination. TB, tuberculosis; COPD, chronic obstructive pulmonary disease; *9 patients missing birthplace; †45 patients missing birthplace.