The study was conducted at Pham Ngoc Thach Tuberculosis and Lung Disease Hospital, Ho Chi Minh City, Vietnam, and Tien Giang Provincial Tuberculosis and Lung Disease Hospital, My Tho, Vietnam. The study area consisted of 3 adjacent rural districts in Tien Gang Province, situated in the Mekong River Delta in southern Vietnam. These 3 districts have implemented DOTS strategies since 1994. Each district has a district TB unit that examines sputum smears and treats ambulatory patients with smear-positive results according to the DOTS strategy. HIV testing of TB patients is performed only when HIV infection is suspected on the basis of clinical signs. Details of the study area have been described elsewhere (13).

Eligible for inclusion were all patients >15 years of age who were residents in the study area and who had registered for treatment of smear-positive pulmonary TB from January 1, 2003, through December 31, 2006, at the participating district TB units or at the provincial TB hospital, and had started treatment for TB <2 weeks earlier. Smear-positive TB was diagnosed by microscopic examination of >2 Ziehl-Neelsen–stained sputum smears following international guidelines (23,24). Eligible patients were included in the study after they provided written informed consent. Scientific and ethical clearance was obtained from the Ethical Health Committee of the Ho Chi Minh City Council. For technical reasons, in 1 of 3 districts, data collection did not start until 2004.

During 1997–2006, CNRs of new smear-positive TB in the study sites decreased by 5.1% per year (95% confidence interval [CI] 4.4%–5.9%), to a lesser extent for men (4.3%, 95% CI 3.4%–5.2%) than for women (7.1%, 95% CI 6.3%–9.1%). Except for patients 15–24 years of age, these decreasing trends were observed for both sexes (Figure 2) and age groups (Figure 3).

During 2003–2006, a total of 2,337 new smear-positive TB patients were registered for treatment in the 3 participating districts; 2,249 (96%) of these patients met inclusion criteria. We excluded 31 patients who did not have data collection forms and 8 other patients whose sputum samples were lost. Patients excluded for this reason did not differ by age or sex from the other patients. A total of 2,210 (94.9%) culture results were available. Of these results, 84 were negative and 29 grew nontuberculous mycobacteria. The remaining 2,097 isolates (89.7%) were genotyped (Table 1).

Of these 2,097 isolates, 682 (32.5%) were of the Beijing genotype, 1,063 (50.7%) were of the Vietnam EAI-VNM genotype, and 353 (16.8%) were of other genotypes. There were no differences in age and sex among the patients from whom the isolates were obtained and subjected to genotyping and all remaining patients registered during the study period. The proportion of isolates with a Beijing genotype was significantly higher for patients 15–24 years of age (52.5%) than for patients of other ages (30.6%; p<0.001). The proportion of isolates with the Vietnam genotype was higher among men and among patients >24 years of age; no differences were found for the other genotypes when combined into 1 group (Table 1). No time trends were evident after stratification of the data by age group or sex.

Because the proportion of genotypes other than Beijing and Vietnam was low and was composed of a heterogeneous group, we limited analyses of CNRs by genotype to the isolates of the Vietnam and Beijing genotypes. The age groups 25–34, 35–44, 45–54, and 55–64 years were grouped together because we found no variation in genotype distribution within these subgroups (Table 1).

During 2003–2006, the overall CNR for new smear-positive TB decreased by an average of –4.3% (95% CI –8.0% to –0.3%) per year. Decreasing trends were also seen for patients 25–64 years of age (–5.1% per year, 95% CI –9.7% to –0.2%) and >65 years of age (–8.0% per year, 95% CI –14.8% to –0.7%). For patients 15–24 years of age, CNRs showed an average annual increase of +5.2% (95% CI –8.4%–20.9%). Rates for women showed an average decrease of –7.9% per year (95% CI –15.2% to –0.2%) and for men, an average decrease of –3.0% per year (95% CI –7.3%–1.6%).

Overall trends in CNRs were similar for patients infected with the Beijing and with the Vietnam genotypes (decreases of –5.9% and –4.4% per year, respectively) (Figure 4). Decreasing trends were observed for middle-age and elderly persons (–6.3% and –5.0% per year for patients 24–64 years of age and –11.3% and –7.1% per year for patients >65 years of age).

However, for patients 15–24 years of age, CNRs increased for patients infected with either genotype. This increase was higher for those with the Beijing genotype (9.2%) than for those with the Vietnam genotype (2.0%), but this difference was not statistically significant (p = 0.860). This difference could be explained by a divergent trend among women: a 7.3% increase per year for those with the Beijing genotype (95% CI –20.2%–44.3%) versus a –13.6% decrease for those with the Vietnam genotype (95% CI –55.1% to –3.5%, p = 0.360) (Figure 4).

Closer inspection of CNRs during 2003–2006 showed an increase, followed by a decrease, which was consistent across age groups but with different patterns over time (Figure 5). The distribution of genotypes over time differed by age group. After correction for differences in absolute changes between years, the proportion of variation that was caused by the Vietnam genotype was ≈50% for patients >25 years of age but only 3% for patients 15–24 years of age (Figure 5, Table 2). The proportion of variation caused by the Beijing genotype was 85% for patients 15–24 years of age compared with 53% for patients 25–64 years of age and 23% for patients >65 years of age. Differences between the Beijing genotype and remaining genotypes within the youngest and oldest age groups and differences in genotype distribution between the youngest and the oldest age groups were statistically significant (Table 2).

During 1997–2006, CNRs for new smear-positive TB for the study site decreased by 4.3% per year likely due to introduction of the DOTS strategy in 1994. As in other studies in Vietnam (9,10), we found an underlying pattern of increasing CNRs for young adults, which partially compensated for strongly decreasing CNRs for other age groups.

In the study area, where the Beijing genotype is associated with young age and female sex (13), we observed no effect of variations in genotype distribution on the CNRs for all ages and sexes combined. For patients >25 years of age, decreasing trends were observed for the Beijing and Vietnam genotypes. There was also no overall difference in CNRs between men and women. However, for patients 15–24 years of age, an increasing trend for both major genotypes was observed. This trend was stronger for the Beijing genotype than for the Vietnam genotype, although the difference was not significant. This difference was most apparent among women in this age group.

Increases in CNRs among young adults are generally considered to reflect recent increases in transmission (29). Therefore, our findings may suggest that the Vietnam genotype is being replaced by the Beijing genotype at the population level in rural Vietnam. The lack of an association between the genotype and the trend in CNRs could then reflect random error or a too-short observation period (4 years). Larger studies of longer duration may be needed to determine whether such an association exists.

The increase in CNRs of new smear-positive TB for persons 15–24 years of age may also be explained by development of the HIV epidemic. HIV increases the risk for progression of M. tuberculosis infection to TB and probably increases susceptibility to infection (1). In sub-Saharan Africa, 9% of all new TB cases in persons 15–49 years of age were attributable to HIV infection (8). In Ho Chi Minh City, the HIV prevalence among TB patients during 1997–2002 increased exponentially from 1.5% to 9% (11). This increase in CNRs of new smear-positive TB in patients 15–34 years of age was attributable to HIV, although HIV could not explain the lack of an expected decrease. In our study, HIV data for TB patients were not obtained. The estimated prevalence of HIV infection among persons 15–49 years of age in Tien Giang, Vietnam, in 2005 was much lower than in Ho Chi Minh City (0.5%) (30), similar to the national average. The estimated HIV prevalence among TB patients in 30 sentinel provinces was 4.8% in 2004 (D.H. Thanh, Vietnam National Tuberculosis Programme, unpub. data). Thus, it is unlikely that HIV played a role in the observed associations. Furthermore, no association has been found between HIV infection status and isolation of Beijing genotype from pulmonary TB patients elsewhere (31).

However, we have alternative explanations for our results. We found that among young adults, nearly all of the variation in CNRs between the years of collection was attributable to variations in CNRs caused by the Beijing genotype; this variation was less apparent for persons in older age groups. This finding suggests that Beijing strains circulate more abundantly among young adults as a consequence of high transmission rates within this age group. Alternatively, the observed fluctuations in numbers of cases caused by Beijing strains may reflect importation from urban areas. Although we did not collect data from these areas, young adults in Tien Giang often travel (70 km) to Ho Chi Minh City for school or work. This explanation is supported by the following findings: the prevalence of infections with the Beijing genotype among persons with TB in this city was higher than the prevalence of comparable patients in rural districts (13,21) and our earlier result that infections with the Beijing genotype in Tien Giang were more common in patients living along the main road to Ho Chi Minh City (13).

High transmission rates among young adults, particularly in urban areas, may increase emergence of Beijing strain TB infections in Vietnam. Recent data from The Gambia suggest that Beijing strain infections do not show increased secondary attack rates but have shorter incubation periods than other genotypes (32). In settings with high transmission rates, such strains may be preferentially selected. Even if their risk for transmission as such is not increased, their faster progression to TB and infectiousness will give them a selective advantage. This hypothesis may also explain the association between the Beijing genotype and a history of imprisonment (e.g., in the former Soviet Union) (19,33).

To test this hypothesis, more studies of genotype-specific variations in incubation period and variations in genotype between high-transmission and low-transmission settings are needed. Different sublineages of the Beijing genotype may have different pathogenic characteristics (34,35). Data from Vietnam suggest that the more recently evolved typical Beijing strains have a higher propensity to evade immunity from M. bovis BCG vaccination (36). In addition, studies among immigrant populations in the United States, South Africa, and Canada have suggested that the transmission propensity of different genotypes and of different sublineages of the Beijing genotype is dependent on the host population (19,33,35,37–39). Such studies (19,33) should enable analyses of different sublineages of the Beijing genotype and involve non-East Asian populations.

The association of the Beijing genotype with young adults could threaten the effectiveness of the DOTS strategy in Vietnam and elsewhere. However, this possibility is unlikely because in our study, although 25% of the elderly patients were infected by Beijing strains, overall CNRs of TB caused by Beijing strains still decreased. Conversely, associations have been found in some studies, including one in the same study area, between the Beijing genotype and drug resistance, particularly for multidrug-resistant TB (MDR TB) (13,17,18,21,31). Although until 2001 the prevalence of MDR TB among new patients with smear-positive TB in the study area was relatively low and not increasing (40), the effects of genotype-associated risks for increased acquisition or transmission of MDR TB may threaten the effectiveness of long-term TB control.

Our study has several other limitations. First, our data were limited to new patients with smear-positive pulmonary TB and may not be representative of patients with other types of TB, including previously treated patients. In our study area, the Beijing genotype was more prevalent in previously treated patients than in new patients (13), which suggested that a stronger effect of genotype on TB trends might be observed if previously treated patients were included in the analyses. Second, we did not include patients who were treated in the private health sector. We collected no data on the proportion of patients treated in the private sector, but local health authorities estimate the proportion of these patients to be <10%. Third, we did not include children <15 years of age because of ethical constraints and because their number was expected to be small. In Vietnam, children account for no more than 0.2% of notified smear-positive TB cases (National Tuberculosis Program, unpub. data). However, pediatric TB may be underdiagnosed if the only test for diagnosis is microscopic examination of sputum smears.

Our data do not prove or refute that the increase of TB in young adults in rural Vietnam is related to emergence of the Beijing genotype. However, they do suggest that the association between the Beijing genotype and young age reflects importation of Beijing strain infections from urban areas into rural areas. This importation may be linked to or driven by high rates of transmission among young adults.