Skip Navigation


American Journal of Epidemiology Advance Access originally published online on April 26, 2006
American Journal of Epidemiology 2006 164(1):21-31; doi:10.1093/aje/kwj153
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
164/1/21    most recent
kwj153v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (2)
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Driver, C. R.
Right arrow Articles by Zhao, B.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Driver, C. R.
Right arrow Articles by Zhao, B.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

American Journal of Epidemiology Copyright © 2006 by the Johns Hopkins Bloomberg School of Public Health All rights reserved; printed in U.S.A.

Original Contribution

Which Patients' Factors Predict the Rate of Growth of Mycobacterium tuberculosis Clusters in an Urban Community?

Cynthia R. Driver1, Michelle Macaraig1, Peter D. McElroy2, Carla Clark1, Sonal S. Munsiff1,2, Barry Kreiswirth3, Jeffrey Driscoll4 and Benyang Zhao1

1 New York City Department of Health and Mental Hygiene, New York, NY
2 Centers for Disease Control and Prevention, Atlanta, GA
3 Public Health Research Institute, Newark, NJ
4 Wadsworth Center, New York State Department of Health, Albany, NY

Correspondence to Dr. Cynthia R. Driver, Epidemiology Office, Bureau of Tuberculosis Control, New York City Department of Health and Mental Hygiene, 225 Broadway, 22nd Floor, New York, NY 10007 (e-mail: cdriver{at}health.nyc.gov).

Received for publication December 8, 2005. Accepted for publication January 19, 2006.


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 References
 
Factors influencing tuberculosis cluster growth are poorly understood. The authors examined clusters of two or more culture-confirmed Mycobacterium tuberculosis cases between January 1, 2001, and December 31, 2003, that had insertion sequence 6110 (IS6110) restriction fragment length polymorphism and spoligotype patterns identical to those of another study case. Genotypes first seen in New York, New York, before or during 1993 were considered historical; recent strains were those first seen after 1993. The authors examined the effect of the combined characteristics of infectiousness of the first two cases in a cluster on the rate of cluster growth. Genotyping was performed for 2,408 (91.8%) of the 2,623 tuberculosis cases diagnosed; 873 cases were in 212 clusters. Thirty-one clusters had historical strains, 153 were recent, and 28 were of unknown period. Patients' infectiousness was not associated with the rate of cluster growth among historical strain clusters. Among recent strain clusters, infectiousness of both of the initial cases was associated with a higher rate of cluster growth compared with clusters in which neither initial case was infectious, upon adjustment for male sex (rate ratio = 2.62, 95% confidence interval: 1.19, 5.78). The rate of genotype cluster growth should be monitored regardless of how long the strain has been present in the community. However, infectiousness in the first two cases may be useful to prioritize genotype cluster investigations.

cluster analysis; genotype; tuberculosis

Abbreviations: CI, confidence interval; HIV, human immunodeficiency virus; IS6110, insertion sequence 6110; RFLP, restriction fragment length polymorphism; RR, rate ratio


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 References
 
Mycobacterium tuberculosis genotyping allows differentiation of tuberculosis strains. "Clustered cases" are defined as isolates with matching strains and are considered to reflect recent tuberculosis transmission events. Clusters are then investigated for previously unrecognized epidemiologic links between cases and, if detected, additional control measures can be applied to interrupt further spread. M. tuberculosis genotyping is increasingly used to assist tuberculosis control efforts. Recently, the Centers for Disease Control and Prevention began offering genotyping services to all tuberculosis control programs in the United States (1Go, 2Go). The majority of clusters, however, do not increase in size beyond the first two cases identified (3Go, 4Go). Genotype cluster investigations are resource intensive and complex, thus precipitating the question of which clusters to investigate. Knowing the determinants of growth in cluster size beyond the initial two cases could assist health departments to prioritize particular clusters for investigation and to direct limited resources where additional intervention measures are warranted.

Studies of the molecular epidemiology of tuberculosis have used primarily patient-level data to identify risk factors for genotype clustering of tuberculosis cases. In population-based studies, tuberculosis patients' factors associated with genotype clustering include younger age, nationality, Black race, male sex, respiratory acid-fast bacilli smear positivity, acquired immunodeficiency syndrome (AIDS) and human immunodeficiency virus (HIV) infection, drug and alcohol abuse, and failure to identify contacts (3Go–15Go). Concurrent infection with HIV does not make tuberculosis patients more infectious; however, among persons with latent tuberculosis infection, it is the single most important risk factor for progression to active disease. Persons who had concurrent HIV infection have a 7–10 percent annual risk of tuberculosis disease compared with a 10 percent lifetime risk among persons infected with tuberculosis but not with HIV (16Go–18Go). Whether these factors are associated with the rate of cluster growth has not been described.

Since January 1, 2001, genotyping of isolates from all incident tuberculosis cases in New York, New York (New York City), has been performed using both insertion sequence 6110 (IS6110) restriction fragment length polymorphism (RFLP) and spoligotyping techniques. In this paper, we examine the influence of demographic, clinical, social, and behavioral characteristics of the first two tuberculosis cases in a molecular cluster on the rate of additional cases in the cluster.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 References
 
We used surveillance data from the New York City Department of Health and Mental Hygiene tuberculosis case registry of incident tuberculosis cases and from the tuberculosis genotyping database. M. tuberculosis isolates from all culture-confirmed cases diagnosed from January 1, 2001, to December 31, 2003, were eligible for genotype analysis. Genotyping was performed following standard procedures (19Go–22Go). A "genotype cluster" was defined as two or more cases within the study period that had matching IS6110 RFLP and spoligotype patterns. Before 2001, genotyping was performed on selected isolates in New York City for investigation of tuberculosis outbreaks and clusters, investigation of potential false positive cultures, and surveillance for tuberculosis drug resistance. On the basis of genotyping results of selected isolates from 1990 to 2000, isolates from 2001 to 2003 were classified as historical strains if the genotype was first seen in New York City before or during 1993; recent strains were those first seen in New York City after 1993.

Data collection
Routine data.
Demographic data and clinical information for each patient were obtained by interview and medical-record review by trained Bureau of Tuberculosis Control staff, using the standard data collection forms of the program. Patients were offered HIV testing by their medical providers as part of routine care; however, not all patients accepted HIV testing. HIV test results were obtained from patients' interviews and medical-record reviews as part of routine tuberculosis control program activities and used for this study. HIV status was coded as negative if there was a negative HIV test result within the year before the diagnostic evaluation for tuberculosis documented by laboratory or physician report. HIV status was coded as positive if the patient gave a history of a positive HIV test or acquired immunodeficiency syndrome, or if there was documentation in the medical history or a laboratory result of a positive HIV test result. If neither of these criteria was met, patients were classified as having unknown HIV status for these analyses. We examined whether the presence of HIV infection in one or both of the initial cases in a cluster was associated with the rate of cluster growth in two separate analyses. In the first analysis, we included all clusters; in the second analysis, we excluded clusters in which either of the first two cases had unknown HIV status.

Contacts identified by the patient at the time of initial contact investigation or subsequently during treatment were also interviewed and evaluated for latent tuberculosis infection or tuberculosis and offered treatment for either if indicated. Information on contacts named by the patient and the tuberculin skin test results were also entered into the tuberculosis case registry and linked to the source case record. The first M. tuberculosis isolate from each patient was sent to the public health laboratory by the clinical laboratory where M. tuberculosis was cultured. The tuberculosis case registry was searched twice each month to identify new culture-positive tuberculosis cases eligible for genotyping, and a list of these cases was sent electronically to the public health and genotyping laboratories. If eligible isolates were not submitted, the clinical laboratory was contacted to obtain the isolate. Case-patients' data from the tuberculosis registry and patients' interview forms were merged with the molecular genotyping database.

Genotype data.
Before January 1, 2001, genotyping of selected M. tuberculosis isolates was performed in New York City as part of cross-sectional surveys, surveillance of tuberculosis drug resistance, and investigation of outbreaks and potential false positive cultures. In 1992 and 1993, five hospitals (representing 7–8 percent of culture-positive cases reported in those years) performed genotyping of all tuberculosis cases diagnosed at these hospitals (23Go). Genotyping for all studies was done at the Public Health Research Institute, Newark, New Jersey, by use of IS6110 RFLP, except for false positive culture investigations, which were performed at the genotyping laboratory of the Wadsworth Center, Albany, New York. The year 1993 was used as a cutpoint for classifying the time period when genotypes were first seen in New York City, since the number of incident tuberculosis cases decreased beginning in 1993 (24Go–26Go). The date when the genotype was first identified in New York City was obtained from the genotyping laboratories. "Recent strains" were defined as those first seen in New York City after 1993. "Strains of unknown period" were defined at those with IS6110 RFLP patterns seen before 1993 that were associated with more than one spoligotype in our database. The effect of historical strains on genotype clustering was examined in a subanalysis of clusters whose genotype was first seen after 1993.

Genotype cluster investigations.
Upon detection of each genotype cluster, investigations were performed to determine links between cases. An "epidemiologic link" was defined as a patient's having named the other as a contact, having contacts in common, or reporting having been in the same physical location prior to diagnosis. Subsequent review included information from the patient's and contact's interviews by use of a standard guide used in the program. If an epidemiologic link was not found, the patient was reinterviewed to elicit places of social aggregation by use of an interview guide. Information collected in the cluster investigation, such as mutual contacts, potential sources of tuberculosis, potential locations of transmission, prior history of tuberculosis infection and disease, shared characteristics of clustered case-patients, and epidemiologic links, was entered into the genotyping database.

Variables
The unit of analysis was a genotype cluster. The outcome variable of interest was the rate of additional cases in a cluster, which was calculated as the number of additional cases in the cluster at the end of the study period minus the first two cases, divided by the cluster-months of follow-up. The number of cluster-months of follow-up for cluster growth was the interval from the date of collection of the first positive culture of the second case in the cluster until the end of the study period, December 31, 2003.

The primary exposure variables of interest were infectiousness as a dichotomous variable and the HIV serostatus of the two intial cases in the cluster. Pulmonary tuberculosis was characterized with acid-fast bacilli smear-positive sputum by microscopy and the presence of cavitary lesions (known markers of infectiousness), and persons with such disease are known to be more likely to transmit infection to their contacts (27Go–29Go). We hypothesized that infectiousness would be associated with a higher rate of cluster growth due to increased transmission.

We hypothesized that HIV infection would be associated with a higher rate of cluster growth due to the increased rate of progression to disease, not increased infectiousness. The exposure variables of interest and potential confounders and effect modifiers were coded as the combined characteristics of the first two cases in each cluster. For example, male sex was classified as follows: 0 (neither case was a male), 1 (either case was a male), and 3 (both cases were males). Characteristics of infectiousness were examined as a combined variable on the bases of the presence of pulmonary, acid-fast bacilli smear positivity and cavitary disease (i.e., both, either, or neither case's having pulmonary, acid-fast bacilli smear positivity or cavitary disease). Although having either acid-fast bacilli smear-positive disease or cavitary lesions alone may be an indication of infectiousness, we used the combined variable with the presence of both characteristics in order to increase the precision of this measure. The HIV serostatus of the first two cases was coded as both being HIV infected, either being HIV infected, or neither being HIV infected. The time between the diagnoses of the first and second cases was the number of months between the initial positive culture dates for the first and second cases in the cluster. A subanalysis that excluded clusters in which either of the first two cases had unknown HIV serostatus was performed. We also examined the effect of HIV on cluster growth as a binary outcome variable.

"Homelessness" was defined as lacking fixed, regular housing or living in a shelter/single-room-occupancy hotel before diagnosis or during tuberculosis treatment. Substance abuse included using injection and noninjection illicit drugs during the 12 months before diagnosis. A positive tuberculin skin test (≥5 mm) or treatment for latent tuberculosis infection 6 or more months before diagnosis of the current case was considered prior latent tuberculosis infection.

Statistical analysis
Statistical analyses were performed with SAS, version 8.02, software (SAS Institute, Inc., Cary, North Carolina). The first two cases were compared according to primary exposure variables and covariates. Pearson's chi-squared test or Fisher's exact test was used to test the differences for categorical variables. The influence of the primary exposure variables and covariates on the rate of cluster growth was examined first in all clusters and subsequently among recent strains, with a log-linear model that used a Poisson distribution in which the outcome was the number of cases in the cluster at the end of the study period minus the first two cases. The log months of follow-up from diagnosis of the second case to the end of the study period were used as an offset term to account for the different observation time for each cluster. Covariates that were significantly associated with the rate of cluster size increase and with the primary exposure variables in bivariate analyses were considered potential confounders and were included in multivariate analyses. Crude and adjusted rate ratios for cluster growth and 95 percent confidence intervals were derived from log-linear regression analysis. Effect modification, a difference in the magnitude of the association of the primary exposure in different subgroups, was also explored.

Genotyping of M. tuberculosis isolates and molecular epidemiology activities, as well as the present analyses, received ethical oversight and approval by the Institutional Review Board of the New York City Department of Health and Mental Hygiene. In addition, this project was reviewed by the Associate Director for Science of the National Center for HIV, STD [sexually transmitted disease], and TB [tuberculosis] Prevention at the Centers for Disease Control and Prevention and determined not to be research on human subjects that would have required review.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 References
 
Cluster enumeration
From January 1, 2001, through December 31, 2003, there were 3,434 tuberculosis cases reported in New York City; 2,623 (76.3 percent) had M. tuberculosis cultured from one or more clinical specimens, and 2,408 (91.8 percent) patients' isolates were genotyped. Among the cases with genotyped isolates, 1,535 had genotypes that did not match those of another case in the sample, and 873 (36.2 percent) had genotypes matching one or more other cases. This corresponded to 212 clusters that ranged in size from two to 85 cases (table 1). Thirty-one clusters had genotypes that were seen first in or before 1994, 153 were recent strain clusters, and for 28 the period could not be determined. The median number of years since the genotype was first seen increased with cluster size, from 0.8 years for clusters of size two to 9.9 years for the largest clusters (p < 0.001).


View this table:
[in this window]
[in a new window]
  		TABLE 1. Frequency of clusters (n = 212) by cluster size, New York, New York, 2001–2003

 
The median time of diagnosis between the first and second cases in a cluster was 5.3 months (range: 0–35.1 months). In 89 (41.9 percent) clusters, the second case was diagnosed within 3 months of the first. The median duration of cluster follow-up after diagnosis of the second case was 18.2 months (range: 0.6–36.1 months); 185 (87 percent) clusters were followed for 6 or more months after the second case in the cluster was diagnosed. The median duration of follow-up increased with increasing cluster size, from 12.1 months among clusters of size two to 34.1 months for the largest cluster (p < 0.001).

In six clusters, the first case was younger than 10 years of age and, in an additional eight clusters, the second case was younger than 10 years of age; 11 were less than or equal to 4 years of age. Cluster size was not associated with either one or both of the first two cases being younger than the age of 10 years nor with the order of the case younger than 10 years. There was a great deal of similarity in the characteristics of the first two cases in each cluster. No significant differences according to demographic (figure 1) or clinical (figure 2) characteristics were detected.


Figure 1
View larger version (11K):
[in this window]
[in a new window]
  		FIGURE 1. Demographic characteristics of the first two cases per cluster, New York, New York, 2001–2003 (p > 0.05 for all comparisons).

 

Figure 2
View larger version (9K):
[in this window]
[in a new window]
  		FIGURE 2. Clinical characteristics of the first two cases per cluster, New York, New York, 2001–2003 (p > 0.05 for all comparisons). AFB, acid-fast bacilli; HIV, human immunodeficiency virus; TB, tuberculosis.

 
In 31 (14.6 percent) clusters, one or both of the initial cases had an epidemiologic link detected to another case in the cluster. The proportion of clusters with epidemiologic links detected was higher in clusters of recent strains compared with historical strains, but the difference was not statistically significant (17.6 percent vs. 9.6 percent; p = 0.27). Overall, 176 (20.1 percent) of the 873 clustered cases had epidemiologic links detected; 100 (56.8 percent) of these links were made during the initial contact investigation, 50 (28.4 percent) had links made as part of the genotype cluster investigation, and 26 (14.7 percent) had both contact and genotype cluster investigation links. In 30 (14.1 percent) of the 212 clusters, an epidemiologic link was detected between the first and second case; 28 of these were recent, and two clusters (with two and four cases, respectively) were historical strain clusters. Most links (26 (86.6 percent) of 30) were made in the traditional contact investigation; 21 of these were in the household or among family members, and five were friends or leisure contacts. Four links were identified as a result of the genotype cluster investigation; these links were a shelter for homeless persons, a local grocery store, or the same or neighboring address. The presence of a link between the first and second cases was not associated with the rate of cluster growth among recent strain clusters but was significantly associated with the lower rate of cluster growth in historical strain clusters (0.47 vs. 2.91 cases per 10 cluster-months of follow-up) (p < 0.001).

Rates of cluster growth and crude associations
The combined characteristics of the first two cases by the rate of additional cases in the cluster are shown in table 2. For all clusters combined, the total number of additional cases after subtracting the first two cases in each cluster was 449. The total cluster-months of follow-up were 3,915 (average rate of additional cases: 1.1 per 10 cluster-months of follow-up). The rates of new cases were highest for clusters where at least one case was homeless, a substance abuser, or HIV infected. Crude rate ratios for increase in cluster size are shown in table 3. The rate of cluster increase was not associated with infectiousness of both cases, the primary exposure variable, among all strains, but it was associated with the higher rate of clustering in subanalysis of recent strains (rate ratio (RR) = 2.18, 95 percent confidence interval (CI): 1.01, 4.70). Among historical strain clusters, infectiousness was not associated with a higher rate of cluster growth (RR = 0.36, 95 percent CI: 0.24, 0.55). The rate of cluster growth was not associated with an interval of less than the median number of months between diagnoses of the first and second cases in the cluster (5.3 months). The median number of months from the second to the third cases in a cluster was 4 (range: 0–27) months.


View this table:
[in this window]
[in a new window]
  		TABLE 2. Rate of increase in tuberculosis cluster size by the combined characteristics of the first two cases per cluster, New York, New York, 2001–2003

 

View this table:
[in this window]
[in a new window]
  		TABLE 3. Rate ratio for the increase in size of tuberculosis cluster by the combined characteristics of the first two cases per cluster, New York, New York, 2001–2003

 
Among historical strain clusters, a higher rate of cluster growth was associated with HIV infection in either case (RR = 4.2, 95 percent CI: 3.0, 6.0); the rate of cluster growth was not associated with HIV infection in both cases (RR = 1.0, 95 percent CI: 0.5, 2.2). The higher rate of cluster growth was not associated with HIV infection in both or either of the initial cases, among recent strain cases. In a logistic regression model with cluster growth as a binary outcome variable, cluster growth was not associated with HIV infection in either or both initial cases (odds ratio = 1.20, 95 percent CI: 0.64, 2.27; odds ratio = 2.62, 95 percent CI: 0.83, 8.22, respectively). These estimates did not change after adjustment for the period when the strain was first seen.

The number of cases by HIV serostatus is shown in figure 2. There were 91 clusters in which at least one of the first two cases in the cluster had unknown serostatus and 121 clusters in which the HIV test results were known for both of the first two cases. The association of the rate of cluster growth with HIV serostatus remained the same after we excluded clusters in which either case had unknown HIV serostatus. The results of this subanalysis of 121 clusters, in which both of the initial cases were tested, found that higher rates of cluster growth were associated with HIV infection in both or either case (RR = 2.96, 95 percent CI: 2.12, 4.11; RR = 3.30, 95 percent CI: 2.61, 4.18, respectively) compared with HIV infected in neither case. Among recent strain clusters in which both of the initial cases were tested for HIV (n = 86), the rate of cluster growth was not associated with HIV infection in both or either of the initial cases (RR = 0.95, 95 percent CI: 0.38, 2.36; RR = 0.65, 95 percent CI: 0.36, 1.18, respectively).

Confounding and effect modification
We examined the role of potential confounding and effect modification on the association between infectiousness and HIV infection with the rate of cluster growth among all clusters and for recent strain clusters separately. In the analysis of all clusters, the following variables met criteria for potential confounders of the association of HIV and clustering in all strains: Black or Asian race, being born in the United States, homelessness, substance abuse, and prior tuberculosis disease. Effect modification of the association between HIV and clustering was present for the following variables: male sex, Hispanic ethnicity, Black and Asian race, being born in the United States, substance abuse, infectiousness, and acid-fast bacilli smear grade among all strains (table 4). Among clusters in which both of the initial cases were HIV infected, the rate of increase in cluster size was greater when either was Asian, born in the United States, homeless, infectious, or had an acid-fast bacilli smear grade of 4 compared with those with neither having these characteristics. Among clusters in which either or both of the initial cases were HIV infected, the rate of increase in cluster size was highest if either or both were males, Black, Hispanic, born in the United States, or substance abusers compared with those in which neither had these characteristics.


View this table:
[in this window]
[in a new window]
  		TABLE 4. Rate ratio for the increase in tuberculosis cluster size by the selected characteristics of the first two cases stratified by human immunodeficiency virus serostatus, among all 212 clusters, New York, New York, 2001–2003

 
Among recent strain clusters, none of the covariates was associated with infectiousness. Clusters in which both or either case was a male were also more likely to be infectious, although the association was not statistically significant (p = 0.055). Infectiousness in both cases and male sex in either case remained independently associated with clustering in a model that included both variables (RR = 2.62, 95 percent CI: 1.19, 5.78; RR = 2.44, 95 percent CI: 1.29, 4.61, respectively). We observed differences in the magnitude of the association between infectiousness and clustering (i.e., effect modification) among Hispanics and homeless persons (table 5). Among clusters in which both of the initial cases were infectious, the rate of increase in cluster size was greater when either was Hispanic or when both were homeless compared with those where neither was Hispanic or homeless.


View this table:
[in this window]
[in a new window]
  		TABLE 5. Rate ratio for the increase in tuberculosis cluster size by selected characteristics of the first two cases stratified by infectiousness, among 153 recent clusters, New York, New York, 2001–2003

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
References
 
Since most clusters do not grow beyond two cases, it would be useful to know which clusters are more likely to expand over time and to prioritize these for control activities. Our analysis showed that in New York City the presence of characteristics of infectiousness of the first two cases in a cluster was associated with a higher rate of cluster growth in clusters of recent genotypes. Furthermore, the presence of this association among recent strains but not among all strains provides further evidence that historical strains dilute the effect on genotype clustering of factors associated with recent transmission. By definition, transmission of historical strains would have occurred over a longer period of time than transmission of more recent strains. Therefore, indirect or second-generation transmission is likely to be greater among historical strains. In clusters of historical strains, the first two cases are less likely to be the true first cases in a cluster; rather, these cases are more likely to represent progression to disease from latent tuberculosis infection that occurred in previous generations of transmission of the strain. Although not statistically significant, the greater proportion of clusters with recent strains that had detected epidemiologic links compared with historical strain clusters is also consistent with more ongoing transmission among recent strain clusters. The overall proportion of 20.1 percent genotype-clustered cases with epidemiologic links in our study was slightly lower than the proportions seen in a population-based study of seven sentinel sites in the United States, which ranged from 25 percent to 42 percent (10Go). Among clusters in which either case was infectious, a higher clustering rate was associated with being homeless or Black. The association of Black race and being homeless with clustering has been reported previously in the tuberculosis literature (3Go, 10Go). In New York City, continuing tuberculosis control efforts are needed to interrupt transmission in these groups.

The effect of HIV serostatus on the rate of clustering in our analysis was consistent, with greater likelihood of progression to disease from latent tuberculosis infection but not transmission, since the effect of HIV was seen among historical but not among recent strain clusters. The association appears to be robust, as it was also present in subanalysis of clusters in which both of the initial cases in the cluster were tested for HIV. This is consistent with prior knowledge of tuberculosis transmission and pathogenesis. The presence of an effect of HIV among all strains persisted when we examined only historical strains (RR = 4.21, 95 percent CI: 2.98, 5.96 for HIV infection of either case) and also when we included strains of an unknown period when both cases were HIV infected (RR = 2.71, 95 percent CI: 1.90, 3.88) or either was HIV infected (RR = 2.12, 95 percent CI: 2.12, 3.54). However, the absence of an effect in the subanalysis of recent cases should be interpreted with caution. An alternative explanation for the absence of an association of HIV infection and clustering in recent strain clusters may be that the sample size provided insufficient power to detect an association. The lowest detectable effect estimate of the association of HIV serostatus and the rate of clustering with 212 clusters, 80 percent power, and one-sided alpha of 0.05 was a rate ratio of 1.44.

These results cannot be generalized to other areas since the epidemiologic profiles vary greatly by location, which may affect clustering and cluster growth rates. Our results should be confirmed in other settings. These analyses are also limited by a number of other factors. We did not study other factors that are known to affect the likelihood of transmission. Characteristics of the exposure to an infectious tuberculosis case, such as the proximity and duration of the exposure and the environment where the exposure occurred, are associated with the likelihood of transmission. In addition, characteristics of the contacts and the contact investigation, such as clinical conditions (other than HIV) that weaken the immune status, which can contribute to the likelihood of progression of disease if infected, as well as the quality of the contact investigation, were not available for analysis. Second, our analytical approach does not distinguish different rates of growth of the cluster over the study period. Instead, the outcome variable represents an average rate over the observation period, which may not capture important fluctuations in rates that are better indications of transmission. This model also does not measure the interactions among individuals that are likely to influence transmission patterns and clustering, such as are addressed in social network models. Such models require information on the interactions between all case pairs, which was not available in the data. Finally, loss to follow-up due to movement of potential cluster members out of the study area may have reduced the number of cases in a cluster. However, loss to follow-up is not expected to differ by genotype.

In summary, although these findings suggest that cluster investigations might be prioritized according to a strategy that takes into account the specific characteristics of the first two cases in a genotype cluster, they should be validated in other settings. The fact that the effect of infectiousness and HIV infection on the rate of cluster growth is consistent with prior knowledge of tuberculosis transmission and pathogenesis lends further support to the idea that historical genotypes are more likely to represent progression to disease from remote tuberculosis infection than recent transmission. However, tuberculosis control programs should not be lulled by historical strain clusters since transmission of these strains is also possible. The rate of genotype cluster growth should be monitored regardless of how long the strain has been present in the community.


   ACKNOWLEDGMENTS
 
The authors thank Dr. Thomas R. Navin and Dr. Timothy J. Dondero for critical review of the manuscript and Midelyn Montilla for assistance with manuscript preparation.

Conflict of interest: none declared.


   References
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 References
 

  1. McNabb SJ, Braden CR, Navin TR. DNA fngerprinting of Mycobacterium tuberculosis: lessons learned and implications for the future. Emerg Infect Dis 2002;8:1314–19.[Web of Science][Medline]
  2. Rosenblum LS, Navin TR, Crawford JT. Molecular epidemiology of tuberculosis. (Letter). N Engl J Med 2003;349:2364.[Free Full Text]
  3. Small PM, Hopewell PC, Singh SP, et al. The epidemiology of tuberculosis in San Francisco. A population-based study using conventional and molecular methods. N Engl J Med 1994;330:1703–9.[Abstract/Free Full Text]
  4. van Soolingen D, Borgdorff MW, de Haas PE, et al. Molecular epidemiology of tuberculosis in the Netherlands: a nationwide study from 1993 through 1997. J Infect Dis 1999;180:726–36.[CrossRef][Web of Science][Medline]
  5. Bishai WR, Graham NM, Harrington S, et al. Molecular and geographic patterns of tuberculosis transmission after 15 years of directly observed therapy. JAMA 1998;280:1679–84.[Abstract/Free Full Text]
  6. Borgdorff MW, Nagelkerke N, van Soolingen D, et al. Analysis of tuberculosis transmission between nationalities in the Netherlands in the period 1993–1995 using DNA fingerprinting. Am J Epidemiol 1998;147:187–95.[Abstract/Free Full Text]
  7. Braden CR, Templeton GL, Cave MD, et al. Interpretation of restriction fragment length polymorphism analysis of Mycobacterium tuberculosis isolates from a state with a large rural population. J Infect Dis 1997;175:1446–52.[Web of Science][Medline]
  8. Chin DP, DeRiemer K, Small PM, et al. Differences in contributing factors to tuberculosis incidence in U.S.-born and foreign-born persons. Am J Respir Crit Care Med 1998;158:1797–803.[Abstract/Free Full Text]
  9. Diaz R, Kremer K, de Haas PE, et al. Molecular epidemiology of tuberculosis in Cuba outside of Havana, July 1994–June 1995: utility of spoligotyping versus IS6110 restriction fragment length polymorphism. Int J Tuberc Lung Dis 1998;2:743–50.[Web of Science][Medline]
  10. Ellis BA, Crawford JT, Braden CR, et al. Molecular epidemiology of tuberculosis in a sentinel surveillance population. Emerg Infect Dis 2002;8:1197–209.[Web of Science][Medline]
  11. FitzGerald JM, Fanning A, Hoepnner V, et al. The molecular epidemiology of tuberculosis in western Canada. Int J Tuberc Lung Dis 2003;7:132–8.[Web of Science][Medline]
  12. Garcia-Garcia ML, Jimenez-Corona ME, Ponce-De-Leon A, et al. Mycobacterium tuberculosis drug resistance in a suburban community in southern Mexico. Int J Tuberc Lung Dis 2000;4(suppl 2):S168–70.[Medline]
  13. McConkey SJ, Williams M, Weiss D, et al. Prospective use of molecular typing of Mycobacterium tuberculosis by use of restriction fragment-length polymorphism in a public tuberculosis-control program. Clin Infect Dis 2002;34:612–19.[CrossRef][Web of Science][Medline]
  14. Nguyen D, Proulx JF, Westley J, et al. Tuberculosis in the Inuit community of Quebec, Canada. Am J Respir Crit Care Med 2003;168:1353–7.[Abstract/Free Full Text]
  15. Weis SE, Pogoda JM, Yang Z, et al. Transmission dynamics of tuberculosis in Tarrant County, Texas. Am J Respir Crit Care Med 2002;166:36–42.[Abstract/Free Full Text]
  16. Cruciani M, Malena M, Bosco O, et al. The impact of human immunodeficiency virus type 1 on infectiousness of tuberculosis: a meta-analysis. Clin Infect Dis 2001;33:1922–30.[CrossRef][Web of Science][Medline]
  17. Guelar A, Gatell JM, Verdejo J, et al. A prospective study of the risk of tuberculosis among HIV-infected patients. AIDS 1993;7:1345–9.[Web of Science][Medline]
  18. Selwyn PA, Hartel D, Lewis VA, et al. A prospective study of the risk of tuberculosis among intravenous drug users with human immunodeficiency virus infection. N Engl J Med 1989;320:545–50.[Abstract]
  19. Driscoll JR, Bifani PJ, Mathema B, et al. Spoligologos: a bioinformatic approach to displaying and analyzing Mycobacterium tuberculosis data. Emerg Infect Dis 2002;8:1306–9.[Medline]
  20. Kamerbeek J, Schouls L, Kolk A, et al. Simultaneous detection and strain differentiation of Mycobacterium tuberculosis for diagnosis and epidemiology. J Clin Microbiol 1997;35:907–14.[Abstract/Free Full Text]
  21. Kreiswirth B, Moss AR. Genotyping multidrug-resistant M. tuberculosis in New York. In: Rom WN, Garay S, eds. Tuberculosis. New York, NY: Little, Brown and Company, 1996:199–209.
  22. van Embden JD, Cave MD, Crawford JT, et al. Strain identification of Mycobacterium tuberculosis by DNA fingerprinting: recommendations for a standardized methodology. J Clin Microbiol 1993;31:406–9.[Abstract/Free Full Text]
  23. Tornieporth NG, Ptachewich Y, Poltoratskaia N, et al. Tuberculosis among foreign-born persons in New York City, 1992–1994: implications for tuberculosis control. Int J Tuberc Lung Dis 1997;1:528–35.[Web of Science][Medline]
  24. Frieden TR, Sterling T, Pablos-Mendez A, et al. The emergence of drug-resistant tuberculosis in New York City. N Engl J Med 1993;328:521–6.[Abstract/Free Full Text]
  25. Frieden TR, Fujiwara PI, Washko RM, et al. Tuberculosis in New York City—turning the tide. N Engl J Med 1995;333:229–33.[Abstract/Free Full Text]
  26. Tuberculosis in New York City, 2002: information summary. New York, NY: New York City Department of Health and Mental Hygiene, 2003.
  27. Canetti G. Present aspects of bacterial resistance in tuberculosis. Am Rev Respir Dis 1965;92:687–703.[Web of Science][Medline]
  28. Grzybowski S, Barnett GD, Styblo K. Contacts of cases of active pulmonary tuberculosis. Bull Int Union Tuberc 1975;50:90–106.[Medline]
  29. Shaw JB, Wynn-Williams N. Infectivity of pulmonary tuberculosis in relation to sputum status. Am Rev Tuberc 1954;69:724–32.[Web of Science][Medline]

Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?


This article has been cited by other articles:


Home page
 J. Clin. Microbiol.Home page
L. C. O. Lazzarini, S. M. Spindola, H. Bang, A. L. Gibson, S. Weisenberg, W. da Silva Carvalho, C. J. Augusto, R. C. Huard, A. L. Kritski, and J. L. Ho
RDRio Mycobacterium tuberculosis Infection Is Associated with a Higher Frequency of Cavitary Pulmonary Disease
J. Clin. Microbiol., July 1, 2008; 46(7): 2175 - 2183.
[Abstract] [Full Text] [PDF]

Home page
 Am. J. Respir. Crit. Care Med.Home page
S. V. Kik, S. Verver, D. van Soolingen, P. E. W. de Haas, F. G. Cobelens, K. Kremer, H. van Deutekom, and M. W. Borgdorff
Tuberculosis Outbreaks Predicted by Characteristics of First Patients in a DNA Fingerprint Cluster
Am. J. Respir. Crit. Care Med., July 1, 2008; 178(1): 96 - 104.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
164/1/21    most recent
kwj153v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (2)
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Driver, C. R.
Right arrow Articles by Zhao, B.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Driver, C. R.
Right arrow Articles by Zhao, B.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?