Published ahead of print on May 25, 2006, doi:10.1164/rccm.200512-1977OC
© 2006 American Thoracic Society doi: 10.1164/rccm.200512-1977OC
What Is the Outcome of Targeted Tuberculosis Screening Based on Universal Genotyping and Location?Schools of Medicine and Public Health, University of North Texas Health Science Center at Fort Worth; Tarrant County Public Health Department, Fort Worth; Department of Geography, University of North Texas, Denton; Department of Microbiology, University of Texas Health Science Center at San Antonio, San Antonio; and the Bureau of Tuberculosis Elimination, Texas Department of Health and Human Services, Austin, Texas Correspondence and requests for reprints should be addressed to Stephen E. Weis, D.O., University of North Texas Health Science Center at Fort Worth, Patient Care Center, 855 Montgomery, Fort Worth, TX 76107. E-mail: sweis{at}hsc.unt.edu
Rationale and Objectives: Identifying and treating persons with latent tuberculosis (TB) infection (LTBI) at high risk for developing TB is part of the current TB elimination strategy. There are no specific criteria, other than medical risks, to designate groups as high risk for developing TB. We hypothesized that, if location-based screenings were done in communities where persons with genotypically clustered Mycobacterium tuberculosis resided, then persons with LTBI from recent transmission and with undiagnosed TB could be identified. Methods: Location-based TB screenings were done in partnership with multiple community-based organizations using resources previously used for other types of screening. Main Results: Location-based screenings identified one person with TB for every 83 screened, and one person with LTBI for every five screened. The yield of this targeted screening program for discovering persons with TB and LTBI exceeded what would be expected from nontargeted screening in a county with a TB incidence of 5.7 per 100,000 population. Conclusions: Genotyping combined with geographic information systems analysis can potentially be used to define high-risk status and to define areas for location-based TB screenings.
Key Words: genotyping • location-based screening • Mycobacterium tuberculosis For over 40 yr, screening and treating persons with latent tuberculosis (TB) infection (LTBI) have been important parts of the TB control strategy in the United States (1). Reduction of reported cases of TB has resulted in the Centers for Disease Control and Prevention altering its national TB elimination strategy from universal screening to targeted screening of specific high-risk populations (2). Consistent with this change in strategy, the Institute of Medicine recommended developing more effective methodologies in targeting populations with recently acquired infection for treatment in order to limit the spread of TB (3). Targeted TB screening allows for optimum performance of the tuberculin skin test (TST) used to diagnosis LTBI, and is economically more efficient (3, 4). Developing targeted testing strategies to identify the estimated 15 million individuals with LTBI in the United States is one of the current challenges of the TB elimination policy (5). Assessing Mycobacterium tuberculosis strains within a community by molecular genotyping may be useful to design intervention strategies (6). Studies from Europe and the United States, combining molecular surveillance with epidemiologic investigations, have shown that unrecognized transmission of TB can result in additional TB cases (7–14). Some of these studies have suggested that greater than 40% of persons with TB had a recently acquired infection (8–10). Population-based molecular epidemiologic surveillance combined with geographic information systems (GIS) analysis can identify discrete geographic areas in which on-going TB transmission is likely occurring (15). This suggests genotyping–GIS–based analyses can determine locations where traditional, person-orientated contact investigations are not successfully controlling transmission. If genotyping–GIS analysis can identify locations where traditional methods of surveillance are unsuccessful at identifying transmitted M. tuberculosis, then geographically based screening in those areas could identify persons with previously undiagnosed TB and latent TB infection (LTBI). This article presents the results of a targeted geographic TB screening program with genotyping–GIS analysis defined boundaries for identifying TB and LTBI.
The Tarrant County Health Department (TCHD) coordinated and implemented the TB screening program. The TCHD serves the western portion of the Fort Worth–Dallas metropolitan area in north central Texas and includes a population of approximately 1.5 million (16). The TCHD conducts screening of high-risk populations as part of the required TB control activities specified by its contract with the Bureau of TB Elimination of the Texas Department of Health and Human Services. Previously, the screening efforts were based on epidemiologic risk factors and not on geographic risk. Between January 1, 1993, and December 31, 2000, these screening efforts identified 2 persons with TB out of 2,911 persons screened (17). The epidemiologic risks of the screened population were unstable housing, which included living in congregate living centers, substance abuse, and low socioeconomic status (17). In the current study, TB screening resources were focused on a location-based, targeted screening program implemented in collaboration with local community-based organizations. These screenings were conducted within zip codes previously demonstrated by GIS to have genotypically clustered TB isolates consistent with uninterrupted transmission (15). The methods used to define geographic and genotypic clustering of M. tuberculosis have been previously described (15, 18, 19). Briefly, prospectively collected isolates had IS6110-based restriction fragment length polymorphism (RFLP) and spoligotype genotyping analysis performed to identify patients infected with the same strain (18, 19). Isolates were defined as genotypically clustered if they had seven or more identical IS6110 band patterns or if they had identical IS6110 patterns containing six or fewer bands and identical spoligotypes (15). GIS analysis demonstrated that spatial distribution of TB and of molecular clustering within the county was not homogenous, and displayed a distinct geographic distribution (15). The average annual TB incidence for the county from 1993 to 2000 was 5.9 cases per 100,000. When the incidence was divided using zip code boundaries, the average annual incidence in the three zip codes screened was 94, 55, and 32 cases per 100,000 population. Countywide, 55% of the isolates collected during the previous study period were molecularly clustered. In the three screened zip codes, 95 of 117 (81%) isolates collected were molecularly clustered (15). In the current study, we conducted screening in the three zip codes that had both the highest incidence and molecular clustering of TB and were contiguous. The area of these zip codes was 4.62, 5.74, and 5.88 square miles. The analysis was performed on data collected on persons screened at multiple locations between September 1, 2002, and December 31, 2004. Screening and treatment were done at the community-based organizations for patients' convenience, to take advantage of their client volume, and because they were considered to be locations where transmission was occurring. The community-based organizations participating in implementation of this program included the following: mental health facilities; temporary labor services; job training and living facilities sponsored by the Veterans Administration; dialysis centers; churches; community services center; HIV congregate living facility; and several congregate living facilities sponsored by different faith-based organizations. Persons screened included those using the services and living within the community. Screened individuals were identifiable by a county-issued screening identification card. To receive service at these community-based organizations, the clients needed their county-issued screening identification card. Those without a card were allowed to use services for up to 2 wk, and were encouraged to attend the next scheduled screening or to go to the health department. Each community-based organization had a screening schedule. The same locations were used for screening throughout the study, were accessible for persons with limited transportation, and four of the locations were less than 1.5 miles apart. After 2 wk without presenting a screening card, clients were no longer allowed access to services. Most, but not all, of the faith-based organizations participating in the screening required an identification card for entry to their services. The need for multiple services provided by a combination of community-based organizations provided a strong incentive for individuals to be screened. For example, many of the persons screened used multiple services (e.g., employed through the temporary labor center, received mental health care from mental health facility, washed clothing and used telephone at community center, ate meals at church, and stayed in community-based organization–supported housing). Each member of the screening team had experience in meeting the needs of vulnerable populations, and had successfully completed the Centers for Disease Control and Prevention core TB training modules (20). This training includes standardized placement and interpretation of Mantoux testing (TST), identifying common TB symptoms, principles of direct observation of therapy, recognition of medication side effects, and patient follow-up. The team included public health nurses and physicians; however, the majority of the staff involved did not have medical licenses. Licensed and bilingual staff members were onsite during each screening. The screening program included active TB case finding and testing for LTBI. Screenings were conducted on-site, and included a brief history with demographics, local employment and residential history, length of time using services in the target area, the number of facilities used in the target area, number of days per week using services in the targeted area, symptom assessment, TST, chest radiograph, issuing proof of screening, and creation of a photo identification card. TST screening was done with 5-tuberculin unit–strength purified protein derivative. Based on results of the prior GIS analysis, all area residents were presumed to have TB exposure, and TST reaction greater than or equal to 5 mm induration was defined as positive (21). Persons reporting a history of a prior positive TST were not retested but underwent all other aspects of the evaluation. TCHD personnel were available for 5 evening hours, Monday through Friday, to evaluate TST results and conduct directly observed therapy and directly observed preventive therapy. Persons with symptoms, positive TST, or abnormal radiograph had an additional medical evaluation. The photo identification card expired after 12 mo, and renewal of the card required another complete evaluation. At retest, an increase in TST induration of 5 mm or greater was defined as positive. Any person who developed symptoms consistent with TB before the card expiration date had another complete evaluation. All data were stored in a single database. Statistical analysis was performed using SPSS 13.0 (SPSS, Inc., Chicago, IL). Logistic regression analyses were performed, and the method of maximum likelihood was used to estimate odds ratios (ORs) and 95% confidence intervals (CIs) for all variables. Age was the only continuous variable. The age variable statistics were analyzed by comparing the group means. A 95% CI for the mean age difference was calculated by using the normal approximation, and an independent sample, two-tailed Student t test was used to assess the statistical significance of the mean age difference. Each eligible patient with TB prospectively enrolled and participated in a structured interview by S.E.W. as part of their routine initial medical evaluation; the interview tool was approved by the University of North Texas Health Science Center institutional review board.
Between September 1, 2002, and December 31, 2004, we conducted 5,672 complete evaluations for 3,645 individuals evaluated at 48 screenings (2,770 [76%] evaluated at least once; 1,987 [55%] evaluated at least twice; 756 [21%] evaluated at least three times; 159 [4%], evaluated at least four times). A total of 875 (24%) had a partial evaluation; consisting of epidemiologic and symptom assessment, chest radiograph, and placement of TST, but did not return to have the skin test read. Those with a partial evaluation did not differ statistically from those with a completed evaluation based on race (p = 0.61), age (p = 0.55), or sex (p = 0.23). During the study period, 44 persons (1.2%) were diagnosed with pulmonary TB and 681 individuals (18.6%) were diagnosed with LTBI. A total of 1,987 individuals (54.5%) had repeat evaluations, and 104 of them (5.2%) demonstrated TST conversion (range of increase, 13–37 mm). Over this period, the total number of persons identified with active TB, LTBI, and TST conversions decreased significantly. The number of persons identified with TB decreased from 28.5 cases per 1,000 to 2.4 cases per 1,000 during the 28 mo of screening. The number of persons identified with LTBI decreased during the same period, from 300.6 per 1,000 to 18.9 per 1,000 (Table 1).
As screening was mandatory if individuals wished to use the services available at the community-based organizations, we believe the screened population reflects the demographics of the target population. The screened population was significantly different by age (p = 0.001; data not shown), sex (p < 0.001), and race (p < 0.001) than the overall Tarrant County population (Table 2). When compared with the general Tarrant County population, the screened population was 2.9 times more likely to be male (OR, 2.9; 95% CI, 2.7–3.1) and 5.9 times more likely to be African-American (OR, 5.9; 95% CI, 5.6–6.3). The majority of the population screened was between the ages of 35 and 54 yr (64.9%; Figure 1). The mean age was 43.7 ± 12.9 (SD) yr for the overall screened population; 46.2 ± 10.3 (SD) yr for persons with LTBI; 46.9 ± 10.3 (SD) yr for persons with TST converting to positive on subsequent TST; and 51.2 ± 10.5 (SD) yr for persons with TB. Individuals with LTBI differed significantly by age (p < 0.001) from those without LTBI. On average, those with LTBI were 5.9 yr older than those without identified infection (mean difference, 5.9 yr; 95% CI, 4.8–7.1 yr). Individuals with TST conversion also differed significantly by age from those who did not convert (p < 0.001). On average, those with TST converted to positive were 5.4 yr older than those who did not (mean difference, 5.4 yr; 95% CI, 2.6–7.9 yr). Finally, persons with TB differed significantly by age from those without (p < 0.001). On average, identified individuals with TB were 9.4 yr older than the study population (mean difference, 9.4 yr; 95% CI, 6.1–12.3 yr; data not shown).
Of the 44 individuals identified as having TB, 29 (65.9%) had culture confirmation, 15 (51.7%) of these were smear-positive, and, in 15 (34.1%), the diagnosis was established on clinical characteristics. During the 28 mo after the initiation of the screening program, 9 (20%) developed TB after a normal chest radiograph. HIV status was available in 43 (97.7%) of the cases; 3 (7.0%) were HIV-seropositive. The majority of the individuals were white (56.8%), followed by black (34.1%). Male sex (OR, 4.8; 95% CI, 1.4–24.9) and white race (OR, 1.9; 95% CI, 1.0–3.6) were the only demographic variables significantly associated with TB. Self-identifying as being American Indian appeared to be an important factor for TB; however, the association was not statistically significant (OR, 13.9; 95% CI, 0.3–118.3; Table 3). Three (7%) of the diagnosed cases self-identified as born outside of the United States.
The screening identified 681 (18.6%) individuals with LTBI. Male sex (OR, 2.8; 95% CI, 2.2–3.6), black race (OR, 1.7; 95% CI, 1.4–2.0), and Hispanic (OR, 1.8; 95% CI, 1.4–2.3) were significantly associated with LTBI. Self-identifying as white race (OR, 0.4; 95% CI, 0.3–0.5) was associated with reduced risk (Table 4).
A total of 1,987 (54.5%) individuals received at least 1 follow-up evaluation. Of these, 104 (5.2%) developed a positive TST on retest. There were 1,326 person-years of observations. The overall risk of TST conversion in this targeted community was 8.9 per 100 person-years of exposure. TST conversion rates decreased after the first 14 mo, from 14.3 to 2.2 per 100 person-years of exposure over the last 14 mo. Being male (OR, 2.5; 95% CI, 1.4–4.7) was the only demographic variable significantly associated with developing a positive TST on retest. Of those with positive TST, blacks (48.1%) and whites (39.4%) had the highest incidence of conversion; however, stratified risk estimates remained nonsignificant (blacks: p = 0.724; whites: p = 0.488; Table 5).
Although the currently recommended strategy for TB elimination based on the concept of targeted testing and treatment of persons most likely to develop TB is ideal, how best to expand this screening beyond groups traditionally considered at high risk is uncertain. We have used, for the first time, genotyping–GIS analysis to design and implement a geographical screening program. This geographically targeted screening identified one person with TB for every 83 screened, and one person with latent TB infection for every five screened. The yield of this targeted screening program for discovering persons with TB, LTBI, and TST converters far exceeded what would be expected from nontargeted screening in a county with a TB incidence of 5.7 per 100,000 population year. These data suggest that genotyping–GIS analysis can identify high-risk groups for targeted TB testing programs. There is the potential to identify TB during the evaluation of each TST-positive individual (2, 21). Persons with TB identified through active case finding are more likely to be asymptomatic than persons passively identified (22). There is then, although not the primary purpose, a component of active case finding in every targeted testing program. Our data on screening based on genotyping–GIS analysis indicated a greater yield than many other targeted TB testing programs conducted with other methods and circumstances. A study of the most common form of targeted testing, contact investigation, conducted in five geographically diverse TB programs across the United States found one person with TB for every 319 screened, and one person with LTBI for every eight screened (23). A previous study of TB contacts from Tarrant County identified one person with TB for every 222 persons screened, and one person with LTBI for every five screened (24). A screening program of persons incarcerated in Tarrant County Jail found that screening 3,274 persons was needed to identify one person with TB, and 108 persons to identify one individual with LTBI (25). Other novel efforts targeting screening persons at increased risk who would not normally be evaluated for TB were all less successful at identifying TB (26–28). An advantage of screening based on genotyping–GIS analysis is that the entire community can be surveyed for TB risk and then screening can be targeted to the highest risk locations within the community. TST conversion is an important indicator for detecting recent transmission. We found a significant reduction in TST conversion from 14.3 per 100 person-years of exposure in the first 14 mo to 2.2 per 100 person-years of exposure in the last 14 mo. This magnitude of reduction is similar to the risk reduction that occurred in a high-incidence inner city medical center after full implementation of expanded infection control measures. TST conversion rates decreased from 5.9 to 1.1 per 100 person-years over a 5-yr period (29). It is likely that decreases in transmission are a result of reduced transmission from active case finding as well as the treatment of TB and LTBI. These data demonstrate that screening and treatment of TB based on genotyping–GIS analysis can reduce TB transmission in a community. Political and managerial challenges are as important impediments to TB control as medical challenges (30). Public concern and political will are critical to overcoming these barriers (3, 30). The genotyping–GIS analysis proved to be important in mobilizing community support for this screening. Several community-based organizations that previously would not fully cooperate with TB screening agreed to make it mandatory for individuals who wished to use their services after reviewing the molecular surveillance–GIS analysis indicating community transmission. The community support was crucial to the success of the screening and, subsequently, to the treatment program. Universal genotyping–GIS analysis has the potential to be a useful tool to increase community support and political involvement for TB control in other communities. Persons recently infected with TB are one of the groups at the highest risk for developing TB (31). In most of the developed world, where relatively few persons in the community at any one time have TB, the strategy for identifying persons most likely to have been recently infected is through investigation of the contacts of persons with TB. Currently, contact evaluation uses a concentric circle model that was developed at a time when most contacts occurred within the home and family (32–34). Molecular cluster analysis of M. tuberculosis isolates conducted in developed countries have demonstrated transmission links at times and in places not identified through conventional, concentric circle, person-based contact investigation (8, 9, 11–14, 35–38). These data have led to the suggestion that contact investigations, as currently designed, are insufficient for identifying transmission pathways in multicultural urban areas of the United States, and that they need to be updated to be not only person-orientated, but also place-orientated in design (34). In view of the large number of persons identified through this targeted screening with LTBI and active TB, this targeted screening approach demonstrates that genotyping–GIS analysis can help to delineate transmission pathways. There are no specific criteria to designate a group as high risk for developing TB (38). Identifying groups that are at increased risk at a local level requires public health authorities to develop an epidemiologic profile of persons through systematic screening for LTBI in their community. With currently available methods for diagnosing LTBI, this would be difficult and expensive to do in each community. Since 2004, the Centers for Disease Control and Prevention has sponsored a universal genotyping program for TB. This program expands the availability of genotypes for TB diagnosed in United States. Isolates are characterized by polymerase chain reaction–based molecular techniques, using spoligotyping and mycobacterial interspersed repetitive unit typing (39, 40). These typing methods can now provide local programs with access to genotypes of isolates that are relatively easy to interpret and rapid enough that the information can be used for epidemiologic and clinical decision making. The widespread availability of genotyping could allow for the identification of high-risk TB status defined by geography. The genotyping–GIS analysis could be performed at a local, state, or national level. If our results are confirmed, then genotyping–GIS analysis could become a method to rapidly develop and adjust community screening programs based on location risk. An important limitation to potential widespread implementation of these methods is the need for community-based organizations willing to partner with health authorities to perform screening. This program was successful because it met the needs of persons being screened to obtain services and of the community-based organizations that did not want their members exposed to TB. In addition, TB programs in communities with smaller populations or lower TB morbidity may not have patient volume or expertise to do this level of analysis locally, and therefore this analysis may need to be conducted at a state or national level. Universal submission of at least one isolate per culture-positive case is also needed, as selective submission may affect cluster size and the determination of location-based risk. Additionally, clustering can represent prevalent community strains and not recently transmitted TB. It is also possible that circumstances may exist where geographic clustering could represent recently transmitted strains that were acquired from a different location. Treatment of LTBI, not screening, prevents TB, and for this method to be useful, it must be combined with successful treatment for LTBI. It will be important for these findings to be duplicated in other communities to ensure that they are generalizable. Dating back to the pioneering work of John Snow on cholera, disease mapping has provided important insights into geographic patterns of disease, risk, and mode of spread (41). These insights subsequently were used to provide the first practical epidemic control. We have similarly, through a targeted screening program based on evaluating the geographic pattern of disease, demonstrated that genotyping–GIS analysis can identify areas with prevalent undiagnosed TB and LTBI. Although there are many other associations with the risk of developing TB (e.g., socioeconomic status, country of birth, immunologic status, and exposure), genotyping–GIS analysis is an integrated view of the expression of these risks from the perspective of location, and is a novel approach for defining high-risk status.
The team of public health professionals at the Tarrant County Public Health Department that did the screening consisted of Lisa Adame, John Biboa, Guadalupe Munguia-Bayona, Catalina Blanco, Stephen Fulmer, Debbie Greever, Nicole Hines, Karen Kardaras, Barbara King, Douglas Melendy, Coy Mickens, Mack Neal, Nho Nguyen, Thieu Nguyen, Debbi Roulston, Lee Sewell, Norma Shafer, Louise Slade, Gloria Stevenson, Kelly Taylor, Le Turk, Rose Young, and Catherine Zimmerman. Also, the leadership that made the screening possible included: Elvin Adams, Lou Brewer, Alex Hathaway, and John Suggs. Manuel Bayona provided direction. Francesca Sanchez provided several critical reviews of the manuscript.
Originally Published in Press as DOI: 10.1164/rccm.200512-1977OC on May 25, 2006 Conflict of Interest Statement: None of the authors has a financial relationship with a commercial entity that has an interest in the subject of this manuscript. Received in original form December 29, 2005; accepted in final form May 19, 2006
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
ebek M. DNA fingerprinting and contact investigation. Int J Tuberc Lung Dis 2000;4:S45–S48.