SETTING: Many epidemiologic studies of tuberculosis are being conducted wor
ldwide. Fingerprinting with a secondary marker in strains with fewer than s
ix IS6110-hybridizing bands enhances the tracking of strains, but its impac
t on population-level inferences has not been well studied.
OBJECTIVE: To investigate the effects of secondary genotyping for low-copy
Mycobacterium tuberculosis isolates with polymorphic guanine-cytosine-rich
repetitive sequence (PGRS) on epidemiologic inferences in population-based
research settings.
DESIGN: For San Francisco tuberculosis cases (1991-1996), clusters were def
ined by IS6110 alone and by PGRSlIS6110 to 1) estimate recent transmission,
2) evaluate the theoretical influence of bacterial population parameters o
n these estimates, and 3) assess risk factors for recent transmission.
RESULTS: Secondary typing on low-copy strains (20.3% of all isolates) decre
ased the estimate of recent transmission from 29.1% to 25.3% (P = 0.03). Th
e most influential parameters in determining whether supplemental genotypin
g results in different estimates were the proportion of low-copy strains an
d the amount of clustering. Risk factors for recent transmission were ident
ical for both definitions of clustering.
CONCLUSION: The statistical and inferred effects of sec ondary genotyping o
f M. tuberculosis seem to depend on the proportion of low-copy strains in t
he population, When this proportion is low or when few secondary patterns m
atch, supplemental genotyping may yield minimal insight into population-lev
el investigations.