TABLE 2—
Factors Associated With Being Part of a Tuberculosis Cluster Among New York City Housing Authority Patients With Tuberculosis vs Non–New York City Housing Authority New York City Patients With Tuberculosis: 2001–2009
| Covariatea | Patients Living in NYCHA (n = 362), AOR (95% CI) | Patients Living Outside of NYCHA (n = 6035), AOR (95% CI) |
| Birth in the United States (includes all US territories) | ||
| Yes | 2.31 (1.25, 4.29) | 2.81 (2.29, 3.46) |
| No (Ref) | 1 | 1 |
| Age group, y | ||
| 0–18 | 7.59 (4.11, 14.02) | 3.09 (2.20, 4.35) |
| 19–44 | 2.71 (1.65, 4.43) | 2.47 (2.07, 2.94) |
| 45–64 | 1.23 (0.73, 2.05) | 2.10 (1.75, 2.53) |
| ≥ 65 (Ref) | 1 | 1 |
| Gender | ||
| Male | 0.84 (0.58, 1.21) | 1.20 (1.08, 1.34) |
| Female (Ref) | 1 | 1 |
| Race | ||
| Non-Hispanic Black | 0.31 (0.09, 1.08) | 1.85 (1.45, 2.37) |
| Hispanic | 0.33 (0.09, 1.13) | 1.89 (1.47, 2.42) |
| Asian | 0.15 (0.03, 0.72) | 0.86 (0.65, 1.15) |
| Other | 0.15 (0.01, 1.52) | 0.79 (0.42, 1.49) |
| Non-Hispanic White (Ref) | 1 | 1 |
| History of substance use in past year | ||
| Yes | 2.93 (1.78, 4.81) | 1.35 (1.20, 1.53) |
| No (Ref) | 1 | 1 |
| High exposure to ≥ 4 sociogeographic measures | ||
| Yes | 5.26 (1.16, 23.91) | 1.31 (1.12, 1.54) |
| No (Ref) | 1 | 1 |
| High exposure to overcrowding | ||
| Yes | 0.50 (0.35, 0.72) | NA |
| No (Ref) | 1 | NA |
| History of mental illnessb | ||
| Yes | NA | 1.35 (0.97, 1.88) |
| No (Ref) | NA | 1 |
| History of homelessness | ||
| Yes | NA | 1.30 (1.03, 1.66) |
| No (Ref) | NA | 1 |
Note. AOR = adjusted odds ratio; CI = confidence interval; NA = not applicable; NYC = New York City; NYCHA = New York City Housing Authority; TB = tuberculosis.
Both models always included US birth, age, gender, and race. Other a priori covariates were tested, including any substance use in past year, HIV status, borough of residence, recent unemployment, history of mental illness, history of homelessness, high exposure to poverty, high exposure to severe overcrowding, high exposure to residential racial segregation, high exposure to residential racial concentration, high exposure to local foreign-born population, and high exposure in ≥ 4 of the aforementioned sociogeographic indices.
History of mental illness, despite being nonsignificant by our criterion, was retained in the final model for patients living outside of NYCHA because it provided significant predictive value during model construction.