The Harris County Public Health Refugee Health Screening Program performs intake screening for TB, HIV, and pathogenic parasites and performs other routine laboratory screenings for all refugees, asylees, identified VHTs, parolees, and SIV holders resettling in the county. We performed a cross-sectional study of children 0–18 years of age who were evaluated by this program during January 1, 2010–December 31, 2015. We obtained demographic information, TB exposure history, symptom screening, and testing results from the Harris County Public Health Refugee Health Screening Program and the US Committee for Refugees and Immigrants (although the data originated from the Texas Department of State Health Services when the Texas Refugee Health Program resided there). All children were seen in a clinic run by the Harris County Public Health Refugee Health Screening Program. At this visit, demographic information and testing before immigration were obtained from the family and overseas records. Children not tested before immigration or who had no overseas records were tested during the clinic visit. Most of the children with a positive test for TB infection were evaluated at the Texas Children’s Hospital TB Clinic where additional testing and, if indicated, treatment were provided by 3 of the authors of this article (G.S.L., A.T.C., and J.R.S.).

Immigration classification was determined by the US Committee for Refugees and Immigrants. Immigration classifications are defined by US Citizenship and Immigration Services (Table 1) (18–22). We used WHO definitions for regions of origin (23). We predicted that refugees, asylees, and VHTs would be higher-risk groups for TB infection because they were more likely to have lived in congregate settings. We predicted the lower-risk groups to be parolees and SIVs because of their residence in lower TB-incidence nations and not living in congregate settings. Although overseas vaccination data were unavailable, we assumed that children were BCG immunized because most children emigrated from countries where universal BCG vaccination is practiced.

Clinicians in the Harris County Public Health Refugee Health Screening Program performed initial TB testing. Providers were able to choose the type of TB testing used; they typically used the TST in children <5 years of age and an IGRA in children >5 years of age (24). TST results were considered positive if there was >10 mm of induration, unless the child was living with HIV or had contact with a person with pulmonary TB, in which case the threshold was >5 mm of induration (25). The main IGRA used was the T-SPOT.TB (Oxford Immunotec, https://www.tspot.com), for which a result was defined as positive if >8 spots were noted in either well (24). A positive QuantiFERON Gold-In Tube (QIAGEN, https://www.qiagen.com) result was defined as an antigen-nil value of >0.35 IU/mL (24).

We classified children with positive test(s) for infection as TB infected, likely TB uninfected, or having TB disease. These classifications were determined by 2 authors (ATS and JRS) at the time each child was seen. In the first 2 categories, children had normal physical examinations and 2-view chest radiographs. We classified children as having TB infection if they had a positive IGRA result (IGRA+), if they had a positive TST result (TST+) and no IGRA was done (not all TST+ children had IGRAs performed), or if the IGRA result was indeterminate/invalid. We typically defined children as being likely uninfected if they were TST+/IGRA–, had normal physical examination findings and normal chest radiographs, and had no known contacts with TB disease or if they had negative tests with normal physical examination findings and chest radiographs. However, we did not classify all children who were TST+/IGRA– as being uninfected; some of these children were classified as having TB infection, most commonly because of young age (<2 years). TB disease was diagnosed in children who had clinical, physical examination, or radiographic findings consistent with TB disease (26).

We created 3 models to determine which factors were independently associated with positive tests of infection: a positive TST result, a positive IGRA result, and any positive TB test result (TST or IGRA). We compared demographic characteristics and other categorical variables among the higher-risk and lower-risk groups for statistically significant differences by the χ² test or Fisher exact test for dichotomous variables and Wilcoxon rank-sum or Kruskal-Wallis test for continuous variables. We included any factor with a p value <0.25 in the binary regression model. We created the final model using a backward-step approach. To assess secular trends in usage and positivity, we analyzed monthly totals using linear regression and the Wilcoxon signed-rank test. A p value <0.05 was considered significant.

We conducted all analyses using the SPSS Statistics 25 (IBM, https://www.ibm.com). We obtained institutional review board approval from the Harris County Public Health Department and Baylor College of Medicine (Houston, TX, USA).

Figure 1

Figure 1. Consort diagram of TST and IGRA results in internationally displaced children over a 6-year period, Harris County, Texas, USA, 2010–2015. The percentage reported for TB disease, infection, and uninfected are the...

During the study period, the Harris County Public Health Refugee Health Screening Program evaluated 5,990 children (Table 2), 98% (5,870) of whom received >1 test of TB infection (Table 3): IGRA, 3,730 (63.5%); TST, 1,842 (31.4%); both TST and IGRA, 298 (5.1%). In the TST and IGRA group, 206 (69.1%) were TST–/IGRA–, 29 (9.7%) TST+/IGRA+, 57 (19.1%) TST +/IGRA–, 3 (1.0%) TST+/IGRA indeterminate/invalid, and 3 (1.0%) TST–/IGRA+ (Figure 1). Discrepant test results occurred in 60 (20.1%) children tested with both TST and IGRA.

Overall, 364 children (6.2%) had >1 positive TB test. Among these children, 325 (89.3%) received diagnoses of TB infection, 35 (9.6%) were considered likely uninfected, and 4 (1.1%) received diagnoses of TB disease (Figure 1). The 35 children with a positive test who were considered likely uninfected were all TST+/IGRA–. In addition, 22 (38.6%) children who were TST+/IGRA– were classified as having TB infection, typically earlier in the study period because of young age, variability in provider practice, or both (Figure 1).

The Texas Children’s Hospital Tuberculosis Clinic in Houston cares for most of children in Harris County with TB disease. According to a chart review, none of the children who had TB infection or who were considered likely uninfected had TB disease developed during the following 4–9 years (2,427 person-years of follow-up). Furthermore, we cross-referenced the public health records for Harris County and found that none of these children had been reported to have TB disease develop.

We found 3 factors to be associated with a either a positive TST or IGRA: region of origin, age group, and HIV status. Immigration classification was associated with a positive TB test result on univariate analysis, but this association did not hold true on multivariate analysis. Children, irrespective of immigration classification and from all regions and age groups, had greater odds of having a positive TST result than a positive IGRA result (OR 2.92, 95% CI 2.79–3.59).

Region of Origin

Figure 2

Figure 2. Comparison of TST and IGRA results (using IGRAs as reference) in internationally displaced children over a 6-year period, Harris County, Texas, USA, 2010–2015. A) By location; B) by age. Brackets indicate...

On multivariate analysis, region of origin was a notable correlate for a positive test of TB infection. TST and IGRA positivity varied by region of origin. Using children from Southeast Asia as a reference group, we found that children from eastern Mediterranean countries (adjusted odds ratio [aOR] 0.48, 95% CI 0.33−0.70) and the Americas (aOR 0.19, 95% CI 0.09–0.39) had reduced odds for a positive TST result compared with children from Southeast Asia. Similarly, children from eastern Mediterranean countries (aOR 0.34, 95% CI 0.21−0.53) and the Americas (aOR 0.12, 95% CI 0.06–0.25) had reduced odds for a positive IGRA result compared with children from Southeast Asia (Table 4). Using the IGRA result as the reference, we found the odds of having a positive TST result to be greater in children from eastern Mediterranean countries (OR 3.99, 95% CI 2.59–6.16) and the Americas (OR 4.15, 95% CI 1.62–10.62) (Figure 2, panel A).

In the United States, 66% of reported TB cases occur among foreign-born persons, a rate 13 times higher than for persons born in the United States (27). Previous studies, mostly using the TST, found a prevalence of TB infection of 9%–35% among refugee children (10–15). However, positive results were less common in our cohort, likely because of the variety of immigration classifications included and expanded IGRA use.

We found that the prevalence of positive results for tests of TB infection varied by region of origin and age and that all children in the study had greater odds of a positive result from TST than from IGRA. Immigration classification was not associated with positive results for TB infection. We also found discordance between TST and IGRA results across the pediatric age spectrum, suggesting that the effect of BCG vaccination on TST positivity may be more prolonged than typically expected (28). In addition, potentially confounding our results, children who received TSTs before and after immigration may have had immunologic boosting, resulting in falsely positive TSTs and IGRAs (29).

Region of origin was a notable risk factor for positive TB test results. Children from Southeast Asia had greater odds of having a positive result for TST, IGRA, or both than did children from eastern Mediterranean countries and the Americas, but we found no difference for these children compared with those from Africa. This finding is consistent with known epidemiologic risk factors: the prevalence of TB disease is high in countries in Africa and Southeast Asia, and Southeast Asia has a higher prevalence of TB disease than eastern Mediterranean countries and Cuba (where most children from the Americas region originated) (30).

Age was another noteworthy risk factor for positive TB test results. Children <2 years and >14 years of age had a higher prevalence of positive results for TST, IGRA, or both compared with children 2–14 years of age. Older children (>14 years of age) in our cohort had increased prevalence of positive TB test results by both the TST and IGRA, which more likely represented true TB infection because of the children’s greater time outside the home and cumulative exposure to adults with infectious pulmonary TB (30–34). The higher frequency of positive tests in children <2 years of age, on the other hand, is more difficult to explain. Higher TST positivity in children <2 years of age likely represented greater cross-reactivity with BCG, given the temporal proximity to vaccination or potential boosting if children had serial TSTs performed (before and after immigration). However, BCG immunization cannot explain the increased prevalence of positive IGRA results in these young children. Furthermore, these data contradict previous studies that demonstrate that older children are more likely to have TB infection (30–34). One possible explanation is that IGRA-positive children <2 years of age in our cohort had more prolonged exposure to an adult family member with infectious pulmonary TB in the home, because very young children spend more time in the home than their school-aged counterparts. To date, IGRAs have not been used routinely for the diagnosis of TB infection in children <2 years of age because of a paucity of data on test performance (35).

Most previous estimates of the prevalence of TB infection in immigrant children used the TST as the test of choice, given an initial paucity of pediatric data and the scarcity and cost of IGRAs. Our findings paralleled the results of a recent study using IGRA testing, which estimated that 5.6% of immigrant children had TB infection, compared with previous estimates of 22% based on TST testing (36). American Academy of Pediatrics guidelines currently recommend IGRA use down to 2 years of age (35); some experts recommend using IGRAs in children as young as 1 year of age. Use of IGRAs rather than the TST would likely reduce false positive tests and allow for TB infection therapy to be targeted to those who would most benefit.

The CDC does not recommend tiered testing (that is, obtaining an IGRA if a TST result is positive) for TB infection. However, at times, the initial test of infection, selected either by choice or by necessity, is not the optimal test, particularly for a BCG-immunized child. In our study, all children had greater odds of having a positive TST result compared with a positive IGRA test regardless of immigration classification, region of origin, or age. In addition, we had almost 300 children in whom both TSTs and IGRAs were obtained, of whom 20% had discordant results, mostly TST+/IGRA–. Our findings are consistent with a prior study that demonstrated that for BCG-immunized children who have immigration-related testing, false positive TST results are common, and IGRAs should be the tests of choice for this patient population (10).

Discordance between TST and IGRA results was seen in all age groups. Although false-positive results are an expected limitation of TST use among BCG-vaccinated children, the impact of BCG vaccine in causing falsely positive TST results has typically been thought to be temporally related to BCG vaccine receipt, which in most countries is a single dose immediately after birth. However, discordant TST and IGRA results in older children suggest either that the effect of some strains of BCG on the TST result lasts longer than previously recognized (28) or that the effect of nontuberculous mycobacteria infections that also can cause a falsely positive TST result may be underestimated in children from developing nations. A third possibility is that immunologic boosting as a result of repeat skin testing before and after immigration may have resulted in false positive TSTs (29). Concerns about false positive TST results are considered in updated 2018 guidelines (9) for TB testing before immigration, thereby suggesting use of IGRAs for all persons >2 years of age.

Our study has limitations. Although we have follow-up data from the Texas Children’s Hospital TB clinic for most children evaluated for a positive test of TB infection, we do not have data for some children evaluated at other clinics or for children whose test for TB infection was negative. BCG vaccination status was not routinely documented; thus, the presumption of false-positive TSTs secondary to cross-reactivity with BCG assumes that most children were BCG-immunized. We assumed BCG vaccination because this vaccination is recommended in the national immunization programs of 95% of countries from which these children emigrated (37). HIV infection was rare in our cohort, precluding drawing meaningful conclusions regarding positive results for TST, IGRA, or both in HIV-infected children from our data. Finally, these data may not be generalizable to all immigrant children relocating to the United States because this study included predominantly children from TB high-burden countries (9).

In summary, the prevalence of positive TB test results in this cohort of children was lower than previously reported, and TB disease was rare. The lower prevalence of positive tests of TB infection in this childhood population likely stems from the predominant use of IGRA testing. The TST and IGRA results are frequently discrepant, particularly among those with lower-risk immigration classification, younger children who have received a BCG vaccine, and those from lower-burden countries.

As a result of these data and our experience, we advocate for other health jurisdictions to implement the routine use of IGRA testing for all children, regardless of immigration classification, region of origin, or age, who are evaluated as part of the immigration process to the United States. We also advocate for use of confirmatory IGRA testing in BCG-immunized children with no known TB contacts who have positive TST results. Use of IGRA as opposed to TST in BCG-immunized children would reduce false-positive test results and enable TB infection therapy to be targeted to those who would most benefit.