We found significant differences in the distribution of tuberculosis cases by sex and age groups in all three countries. Cases were more frequently male in all tuberculosis categories, and in all three countries. A previous study shows that the female-to-male prevalence ratios were less than 0.5 in surveys in the South-East Asia and Western Pacific region, and approximately 1 in the African Region[2]. The information on sex and age groups in all tuberculosis categories is not available in standard WHO routine reports. It is however known to vary importantly by age, sex, and country of origin from several analyses that had examined it more closely[2, 8–13].
We also found that the proportion with extrapulmonary tuberculosis among all cases was decreasing in frequency with increasing age.
In all three countries, three quarters of patients occurred among adolescents and adults aged less than 65 years, and one fifth among patients aged 65 years or more. The distribution by age in tuberculosis patients seems to be similar with this figure in 32 European countries reported in the year 1995[14]. A rising age in tuberculosis is regarded as an epidemiologically encouraging sign as it reflects a decrease in infection risk over calendar time with a resulting shift of the bulk of the prevalence of prior acquired infection to older age groups from which cases continue to emerge[3]. That the proportion of cases in the age group 65 years and above was lowest in Cambodia compared to China and Viet Nam thus might indicate that the tuberculosis epidemiology in Cambodia is still larger than in the two other countries. This information is also consistent with the results of the three prevalence surveys which showed that the prevalence of smear-positive tuberculosis was higher in Cambodia (269 per 100,000 population) than in Viet Nam (145 per 100,000) and China (122 per 100,000)[4–6]. The incidence estimates for Cambodia are also higher than those for Viet Nam and China[15]. The relationship between the proportion of tuberculosis cases in the oldest age group of the 65-year-old and older population and the findings from prevalence surveys would deserve a more thorough investigation to ascertain to what extent such an indicator might be used to assess the epidemiology of tuberculosis.
The male-to-female ratio in China and Viet Nam was more than 2, a common observation in many countries[4, 11]. In contrast, in Cambodia there was roughly a balance of cases among the two sexes. Differences in the sex ratio can be attributed to the risk of becoming and being infected, risk differences in progression from latent infection to overt clinical tuberculosis, and differences in accessibility to health services[3]. Male-female differences in the prevalence of infection with Mycobacterium tuberculosis are frequently small up to adolescence while subsequently the risk of becoming infected with M. tuberculosis and hence the resulting prevalence of infection is frequently higher among males than among females[3]. However, the risk of progression from latent infection to tuberculosis varies greatly by age and has commonly been found to be larger among females than males in young adults while this changes to the opposite with increasing age[16–19]. As these two findings run in the opposite direction, one thus might expect that the contribution of females to the total number of cases to be larger where tuberculosis occurs among the young. That this is indeed the case has been shown in Denmark for example where the male-to-female ratio increased steadily over time in parallel with an improvement of the epidemiologic situation and a shift of the age structure of tuberculosis patients to older age groups[16].
The low proportion of extrapulmonary tuberculosis reported in China finds perhaps largely an explanation in the fact that the treatment for extrapulmonary tuberculosis, in contrast to sputum smear-positive tuberculosis, is not offered free of charge and patients thus may seek care outside the governmental clinics captured in this sample.
Extrapulmonary tuberculosis is virtually always and only based on a clinical decision and rarely supported by bacteriological findings. In all our settings the diagnosis of extrapulmonary tuberculosis must be made by a qualified clinician.
In this study, we observed a substantial decrease in the proportion of extrapulmonary among all cases with age, clearly among both male and female patients in Cambodia and clearly among female patients in Viet Nam. That the proportion of extrapulmonary tuberculosis among all patients decreased with increasing age had also been observed in other studies[8, 20].
We observed the proportion of extrapulmonary among all cases to be higher among females than among males, a finding similar to that reported in other studies[21–23]. That extrapulmonary tuberculosis disproportionately affects females is a well-known observation [8, 22, 23, 23]. Extrapulmonary tuberculosis is also associated with young age[8, 22]. This can explain, at least partially, the larger female proportion in Cambodia compared to Viet Nam. Nevertheless, failure of diagnosing extrapulmonary tuberculosis, particularly unusual manifestations, might be of concern in many national programs and needs to be better investigated. That the proportion of extrapulmonary tuberculosis was higher in Cambodia than in Viet Nam might also find a contributory explanation in the higher prevalence of HIV infection in Cambodia than Viet Nam[15].
A comparative analysis of ratios between notification data and the prevalence survey data has the strength that it can potentially unravel relative differences resulting from other factors as the comparative measure does not depend on population figures.
In comparison with the prevalence surveys we found that the ratios of being notified as a new female smear-positive case compared to being a female smear-positive case in the prevalence survey were 2.1, 0.9 and 1.8 in Cambodia, China and Viet Nam. This suggests that males, compared to females, were under-diagnosed in Cambodia and Viet Nam. If this hypothesis is correct, then the national tuberculosis programs in Cambodia and Viet Nam need to pay more attention to case finding among males. This observation is also consistent with the conclusion from the prevalence survey in Cambodia in the year 2000 which noted that case detection among males must be much lower than that among females, [4] an observation also reported from the prevalence survey conducted in Viet Nam in the years 2006-2007. The patient diagnostic rate, a program performance measure proposed by Borgdorff, [24] is the rate at which prevalent cases are detected by a control program, and it was higher in women compared to men in the survey in Viet Nam[6].
An important limitation of our study is that it did not cover patients treated in the private sector and those parts of the public sector failing to report to the NTP. It is conceivable that there may be differences in sex and age among such patients compared to the public sector reporting to NTPs. Another limitation lies with the impossibility to synchronize the study period with the time period covered by the prevalence survey. To the extent the tuberculosis prevalence had changed unequally for age and sex over time our analysis would have been biased. However, the time difference between the prevalence surveys and the notification data captured in our analysis is rather small and such a bias might thus be of minor importance.