Tuberculosis - Epidemiology and Control Issues in Global Perspective

Although being one of the most thoroughly studied diseases in epidemiology, tuberculosis (TB) is now re-emerging with quite new epidemiological characteristics, such as TB combined with HIV infection, and multiple-drug resistance TB. These new aspects can add very serious impact on the existing health burden of TB, especially its serious inequality between developed and developing parts of the world in demographic, social and economic terms. These epidemiological aspects of TB problem will be illustrated in a stepwise way, using several mathematical models. The application of epidemetric model of TB is also shown for evaluation of some TB control measures. J Epidemiol, 1996 ; 6 : S57-S63.


Tuberculosis -Epidemiology and Control Issues in Global Perspective
Toru Mori Although being one of the most thoroughly studied diseases in epidemiology, tuberculosis (TB) is now re-emerging with quite new epidemiological characteristics, such as TB combined with HIV infection, and multiple-drug resistance TB. These new aspects can add very serious impact on the existing health burden of TB, especially its serious inequality between developed and developing parts of the world in demographic, social and economic terms. These epidemiological aspects of TB problem will be illustrated in a stepwise way, using several mathematical models. The application of epidemetric model of TB is also shown for evaluation of some TB control measures. J Epidemiol,1996 ; 6 : S57-S63.

epidemetric model, simulation, annual risk of infection, DALY, HIV epidemics
Tuberculosis is considered to be one of the most thoroughly studied problems in epidemiology. However, it is now reemerging with new epidemiological characteristics in addition to its classic features. Tuberculosis combined with HIV infection presents with a very unique natural history, from infection through clinical breakdown and progression. Multiple-drug resistant tuberculosis is another formidable and foreign problem. Thus, tuberculosis is challenging us in new ways, in both developed and developing countries toward the end of the 20th century in spite of the optimism of the 1960s and 1970s. In this brief lecture I would like to put the problems of tuberculosis in an epidemiological perspective, using mainly epidemiological models.

A Simple Epidemiological
Model and Simulation of the Secular Trend of Epidemics Figure 1 illustrates the mechanism of tuberculosis in a human population in a very simple way.  After the cycle we will have only 64 infections, and then 51, 41, 33, and so on. In this way, the epidemic's extent will diminish gradually given a reduced value of any of the parameters. However, the tuberculosis situation will deteriorate if something occurs which increases the value of any of the parameters. Apparent improvement or worsening of the situation, as seen in the history of tuberculosis in many countries during the progress of industrial revolution, wars, or economic development, may be the result of a readjustment of the parameter values, as illustrated above. It is easy to simulate using this model the changes in tuberculosis epidemics in terms of the number of new cases produced or the number of new infections, under conditions elevating or reducing the rate of reproduction due to a change in, for instance, the risk of clinical breakdown, duration of disease, or contagion. A change in these parameter values results in a continuous reduction or increase of tuberculosis epidemics. In the secular trend of tuberculosis mortality in England and Wales since the beginning of the century, it is observed that the rate of reduction during 1900 and 1945 was about 2.6% per year, and 13% per year after the war. It is supposed that before the war the gradual improvement in socioeconomic conditions and nutrition had the effect of reducing the parameter values to that extent, and that the post-war introduction of modem tuberculosis control measures was more effective than that. Although model I showed is a very simple example, it may be sufficient to simulate the basic trend of tuberculosis.

Annual Risk of Infection as a Parameter for Epidemiological Problems
Examining the amount of infections in a more realistic way, we can introduce the idea of risk of infection, i.e., the chance of an uninfected person becoming infected during a year. This parameter is intended to express the epidemiological size of the tuberculosis problem. Figure 2 shows the trend of annual risk of infection for the Netherlands 1), known to currently have the world's lowest risk, and that of Japan, where the risk is far higher than in the Netherlands 2). Many industrialized countries have a risk of infection at levels below 0.3%; Korea and other countries with newly industrialized economies are approaching this level or are below it. Many developing countries still remain at a level of 0.3% or higher, and their trend is only slowly downward, at a standstill, or even upward.
The annual risk of infection can be measured rather easily based on the tuberculin survey results. One useful element is that there is a rough relationship between the annual risk of infection and the incidence of active tuberculosis in high prevalence areas such as developing countries, i.e., 1% of the risk of infection corresponds to an incidence rate of 50 per 100,000 3). Based on this assumption, researchers of WHO and CDC of the USA estimated the burden of tuberculosis in developing areas, in terms of tuberculosis incidence (which is generally difficult to obtain) for the respective regions of the world 4). Obviously, tuberculosis presents a notable inequality of health which is unfavorable to developing countries . About 95% of new tuberculosis cases and 98 .5% of tuberculosis  In the model, the population is divided into 8 subgroups, each combined with epidemiological relations or flows. This set of subgroups and flows are assumed for every 5-year age group, with age-specific parameter values for corresponding age groups.
As an illustration of the output of this model, I produced an age-specific incidence rate for successive years for a fictitious low prevalence population ( Figure 4). As you see, the incidence peak shifts to the right, along with a downward trend of tuberculosis for all ages, reflecting the well-known cohort phenomenon. In this way, the model incorporates the age-specific aspects of tuberculosis epidemics. <i>

Quality of Tuberculosis Model in Developed vs. Developing Countries </i>
Here, apart from the model, I would like to examine this aspect of the global tuberculosis problem. The age-specific prevalence rates of tuberculosis infection in Japan for 1950 and 1990 were estimated from the trend of annual risk of infection as seen above. The tuberculosis situation in Japan in 1950 was that of the high prevalence area now . The comparison between these two situations is clear -in a high prevalence situation, the prevalence is already high for young persons, while in a low prevalence situation it is very low for the young, but considerably high for older persons, the latter reflecting the high infection risk in their past. By applying the age compositions of developed and developing countries to this infection curve, the age-composition of the infected persons in a population is obtained that is striking, as seen in the comparison between the USA and Tanzania. In a low prevalence situation such as the USA, about 93% of the infected persons are aged 40 years or more, while in a high prevalence situation only 35% are so.
<i> Disease Control Priorities Based on DALY </i> This difference in age-composition results in a remarkable difference in the age patterns of patients tuberculosis which developed from infected persons. The higher risk of clinical breakdown in adolescence also operates in producing this age pattern. In a low prevalence area. e.g. the USA 54 % of the cases are 55 years or older, while in a high prevalence area such as Tanzania only 18 % are of that age, which indicates that in a high prevalence area tuberculosis occurs in economically and socially active segments of the population, affecting society and households. The same may be true for the tuberculosis burden due to deaths. Thus, we should consider whether the tuberculosis problem is somewhat underestimated when it is evaluated only in terms of the number of deaths or new (Incidence, per 100,000) Age(yrs) Figure 4. Simulation of the Cohort Phenomenon (Fictitious population , Incidence Rate) cases. Recently, trials have been done to measure the disease burden more equitably, by giving weight to losses due to disease in the young. This approach is called DALY, or disability adjusted life years, and is used by WHO and the World Bank as one basis for disease control priorities 10). In a series of works, loss of DALYs due to death and illness or disability caused by various diseases, and also the DALYs to be saved with unit of money for the disease control program are compared. The results of these works are clamied to be a more relevant basis for priority in investment in the health sector. As a conclusion, tuberculosis is shown to be underestimated in the health policy in the developing world, by both the donor and recipient government, when the investment in the tuberculosis control program is more rewarding.

Evaluation of Treatment Program
Getting back to the Waaler model, the model is used to predict situations under various tuberculosis control programs, where the effect of each of the control measures is expressed as a modified parameter value of the relevant flows. The results of the simulation of case-finding and treatment at various intensities are shown in Figure 5 6). For example, when the case-finding and treatment program is introduced at time zero and maintained after that, then the number of new cases may take a course like this over time. This kind of prediction of fic-titious situations using a model enables the rational evaluation of a control program.
In the original Waaler model, the emergence of chronic bacillary cases and drug resistant tuberculosis cases resulting from treatment failure is not incorporated. In the face of this new problem, which is prevalent both in developing and developed countries, a refinement was made to the model in order to compare the outcomes of treatment programs with higher and lower treatment success rates 3>. The cost-effectiveness analysis was made to compare the treatment regimens of short-course chemotherapy that required the more expensive medications for 6-8 months, and the standard chemotherapy using cheaper medications for at least 12 months, both with and without hos-   In the above presentation I have attempted to show the usefulness of an epidemiological model to describe the global situation and problem of tuberculosis and the decision making of its control policy and program, not necessarily in a very sys-tematic way. Taking into account reservations regarding possible limitations and pitfalls, the models can be and should be utilized more widely, and recent advances in computer hardware, as well as software, will make it a more epidemiologistfriendly technology.