Association Between Birth Region and Time to Tuberculosis Diagnosis After US Entry Among Non–US-Born Persons

– 143 words Approximately 90% of US tuberculosis (TB) cases among non–US-born persons are attributable 8 to progression of latent TB infection to TB disease. Using survival analysis, we investigated if 9 birthplace is associated with time of progression to TB disease among non–US-born persons. We 10 derived a Cox regression model comparing differences in time to TB diagnosis after US entry 11 among 19 global birth regions, adjusting for sex, birth year, and age at diagnosis. Compared with 12 persons from Western Europe, the adjusted hazard rate of developing TB was significantly 13 higher (p ≤ 0.05) for persons from all other regions, except North America and Northern Europe, 14 and highest among persons from Middle Africa (adjusted hazard ratio = 7.0; 95% confidence 15 interval: 6.5–7.4). Time to TB diagnosis among non–US-born persons therefore varied by birth 16 region, which represents an important prognostic indicator for progression to TB disease.


Introduction
We performed a bivariate analysis and a multivariate analysis examining the association 85 between birth country and time from initial US arrival to TB disease diagnosis, which was our 86 main outcome variable. In addition to birth country, we examined the following NTSS case 87 demographic variables as additional covariates for our analyses: age at TB disease diagnosis, 88 sex, and birth year. We performed the bivariate analysis to assess the association between time to 89 TB disease diagnosis and those variables individually, and we performed the multivariate 90 analysis to account for the effects of the additional demographic variables on the association 91 between birth country and time to TB disease diagnosis. For our analyses, time to TB disease 92 diagnosis was the number of months spent in the United States before TB disease diagnosis, 93 which we derived by subtracting the date of initial US entry from the date that the TB disease 94 case was reported. We used the case report date because information regarding actual disease 95 diagnosis date was unavailable through NTSS; the report date represents the earliest notification 96 to a local public health agency that the patient might have TB disease. Because >200 non-US 97 countries and territories of birth were reported to NTSS during the study timeframe, we used the 98 United Nations M49 standard, which categorizes countries and territories according to 99 geographic location and level of development, to divide these countries into 19 regions for ease 100 of statistical analysis (10,11) (see "Birth Country Categorization" in Appendix B). For our 101 bivariate analysis, we categorized age into 6 categories, but we kept age as a continuous variable because of multicollinearity with birth region (see "Exclusion of race/ethnicity variable" in 107 Appendix B).

109
We began by determining the number of patients who had TB disease not attributed to 110 recent transmission in each demographic category and determined the median number of months 111 (with interquartile range [IQR]) that these patients spent in the United States before receiving a 112 TB disease diagnosis. We also determined the median number of months according to birth 113 region and median age. We constructed a crude bivariate Kaplan-Meier survival curve to visually 114 demonstrate the overall distribution of time to TB diagnosis since US entry. We then constructed 115 bivariate Kaplan-Meier curves stratified by each demographic covariate. We tested for 116 statistically significant differences between curves at p ≤0.05 by using the log rank test and 117 reported global p values. 118 We then used Cox regression to examine the association of time to TB disease diagnosis 119 and birth region, adjusting for the effects of age at diagnosis, sex, and birth year. We tested the 120 proportional hazards assumption (see "Cox regression assumptions" in Appendix B), and 121 because our data are likely double-truncated, we applied a correction for double-truncation to our 122 sample set (13) (see "Double-truncation" in Appendix B). We reported adjusted hazard ratios 123 (aHRs) with associated 95% confidence intervals (CIs) and p values from our Cox regression 124 analysis.

125
Because the time after US entry at which persons who enter the United States with TB 126 disease typically receive a diagnosis is unclear, previous studies have used a range of post-entry 127 times to define cases as disease missed on entry as opposed to LTBI reactivation (14-16). This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted August 4, 2020. ; https://doi.org/10.1101/2020.08.02.20160135 doi: medRxiv preprint Therefore, we performed a sensitivity analysis to assess the effect of extending the exclusion 129 window from 3 months to 6 months for TB disease missed at time of entry for our Cox 130 regression analysis. Additionally, because TB in a child can be considered a sentinel event that 131 indicates recent transmission (17), we performed a sensitivity analysis to assess the effect of 132 excluding children aged <5 years on the basis of previous research demonstrating that recent 133 transmission is most commonly observed among children in that age group (3,4). 134 We performed all statistical analyses using R version 3.6.1, and we used the R code 135 developed by Rennert and Xie to correct for double truncation (13). All data were collected as 136 part of routine disease surveillance and were not part of human subjects research requiring 137 institutional review board approval.

139
During 2011-2018, 79,159 TB cases were reported to NTSS ( Figure 1). Of those cases, 140 28,745 cases were among US-born persons, were not reported from one of the 50 states or the 141 District of Columbia, or did not have a known origin of birth. We also excluded 15,655 cases 142 that were either attributed to recent transmission or had no information regarding recent 143 transmission available; we excluded 2,906 cases for which the time to TB diagnosis was ≤3 144 months. Finally, we excluded 3,403 cases with missing information for ≥1 of the variables 145 considered; because only approximately 10% of the included cases had missing data for ≥1 146 variable, we performed listwise deletion of those cases for our analyses. As a result, we included 147 28,450 cases for our analyses.

148
The median number of months that a non-US-born person spent in the United States 149 before receiving a diagnosis of TB disease not attributed to recent transmission was 144 months 150 for use under a CC0 license.
This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  198 The association between birth region and time to TB diagnosis may also be a 199 consequence of differences in risk factors for progression to TB disease among different regions.  This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.   This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted August 4, 2020. ;https://doi.org/10.1101https://doi.org/10. /2020   This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted August 4, 2020. ; https://doi.org/10.1101/2020.08.02.20160135 doi: medRxiv preprint This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted August 4, 2020. ; https://doi.org/10.1101/2020.08.02.20160135 doi: medRxiv preprint This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted August 4, 2020. ; https://doi.org/10.1101/2020.08.02.20160135 doi: medRxiv preprint  This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.