Guide to the Application of Genotyping to Tuberculosis Prevention and Control

Combining Genotyping and Epidemiologic Data to Improve Our Understanding of Tuberculosis Transmission

Clustering as a Surrogate Measure of Recent Transmission

TB programs that implement universal genotyping, either program-wide or restricted to a county or adjacent group of counties, will have a powerful tool to analyze the epidemiology of TB in their jurisdiction. Programs that implement universal genotyping will be able to monitor changes in the percentage of genotyping clustered cases. To the extent that clustering reflects recent transmission, declines in this percentage over time will reflect progress toward eliminating transmission.

There are several important caveats, however, in using genotype clustering as a surrogate measure of recent transmission. Some of these have already been described in the preceding discussion of matching genotypes. These include insufficient discriminatory power of genotyping methods, transmission of an endemic strain in a relatively closed population, false-positive cultures, and laboratory error. Each of these limitations leads to an overestimate of the rate of recent transmission.

One other factor also leads to an overestimate of recent transmission. Two cases with matching genotypes are counted as two clustered cases, but if one is the source case and the other is a secondary case, they represent only one episode of recent transmission. Similarly, three clustered cases, when one is the source case, represent only two episodes of recent transmission. Some epidemiologists have suggested that an adjustment should be made to account for this phenomenon (Small 1994). They argue that the most accurate way to apply genotyping results to make estimates of recent transmission is to exclude one case from the count of each cluster.

Other factors may lead to underestimates of recent transmission. For example, an isolate from the source case might not have been genotyped, either because no culture was available or the isolate was not sent for genotyping. This happens frequently at the start of a new genotyping program, when there are few genotypes in the database. It is common for the percentage of isolates that cluster to increase over the first 2 or 3 years of a TB program’s new genotyping effort. Underestimation of recent transmission may also occur if the source patient lived in a different jurisdiction from the secondary patient; unless the two TB programs compared genotyping results, the identical genotypes would not be recognized as matching. Emilia Vynnycky and colleagues have developed sophisticated computer models to accurately predict how these factors influence the accuracy of clustering data (Vynnycky 2003)

Because of the shortcomings associated with using genotype clustering as a surrogate measure of recent transmission, another alternative should be considered. As discussed in more detail in Chapter 6, Applying Genotyping Results to Tuberculosis Control Practices, by combining genotyping results with epidemiologic information, TB programs can obtain a more specific estimate of the amount of recent TB transmission that is occurring in their jurisdictions. We define epidemiologically confirmed recent transmission as a patient who belongs to a genotyping cluster and shares known epidemiologic links with another patient in that cluster. Monitoring the percentage and the rate of recent transmission provides useful information about the effectiveness of programs to interrupt transmission.