Measuring the Global Burden of Tuberculosis

  • I. Onozaki
  • N. Ishikawa
  • D. A. Enarson


There are several ways to measure the burden of TB. Estimated TB incidence and its rate have been utilized as the most popular indicators of TB burden since 1997. According to the WHO’s estimation of the global TB burden, there were 9.2 million new TB cases in 2006 with 1.7 million deaths due to TB. Although the estimated global incidence rate stabilized or began to fall slowly, the number of new cases was still increasing because population growth had a greater effect on the number of cases than did the declining rate of TB.

However, for each individual country, the uncertainty of the incidence estimation became a concern for monitoring and evaluating the progress of the TB situation at the country level. Methods for estimating current burden were reviewed. An important limitation is that the current estimation of the burden, incidence, is mainly based on the notification data from countries and assumption of case detection rate. Very few countries have scientific data to estimate the epidemiological situation and trend. It is essential to strengthen routine recording and reporting systems, and surveillance, to improve the estimates of burden and trend. However, as the current surveillance systems are not always reliable in many countries where TB is common, it is recommended that a prevalence survey be conducted every 5–10 years. Prevalence itself is one of the MDG’s indicators, and survey findings can be utilized to evaluate trends derived from routine surveillance. Combinations of prevalence surveys and notification data might be the best method to measure disease burden. Although a prevalence survey is costly and labour intensive, it may be a great help to a country in improving TB control policy. Surveys made in some countries might help neighbouring countries with similar situations to understand their epidemiological status of TB through interpretation of routine surveillance data. By adding delay analysis, much insight can be gained into the quality and use of TB services.


Chest Radiography Verbal Autopsy Prevalence Survey Delay Analysis Vital Registration 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

List of Abbreviations:


Anti- Tuberculosis vaccine containing the Bacille de Calmette et Guérin used by the expanded immunization programme worldwide


Case Detection Rate


Disabled Adjusted Life-Years


The brand name of a comprehensive TB control program recommended by the WHO and Stop TB Partnership originated from Directly Observed Treatment, Short Course


High-burden country where the absolute number of TB incident cases is high. 22 HBCs comprise 80% of the Global TB burden


Human Immuno-deficiency Virus


Interferon gamma release assays


 Millennium Development Goals


Multidrug-Resistant Tuberculosis. TB that is resistant, at a minimum, to the two major anti-TB drugs Rifampicin and Isoniazid


National TB Control Program




World Health Organization


Extensively Drug Resistant Tuberculosis, which is MDR but also resistant to any fluoroquinolone and any second-line anti-TB injectables, namely Amikacin, Kanamycin or Capreomycin

1 Introduction

The Burden of Tuberculosis (TB) usually refers to the epidemiological burden and is measured by disease incidence, prevalence and/or mortality, and infection prevalence, and incidence (annual risk). It may also be measured by economic indicators such as medical expenditure on TB and economic loss due to TB. Until the early 1990s, short-course chemotherapy, a rifampicin-containing regimen had been believed to be too expensive for developing countries. However, the cost effectiveness of TB control was demonstrated by estimating the number DALYs saved by its activities, and the World Bank started to invest in TB control (Murray et al., 1991; World Bank, 1993). This marked the turning of the tide in terms of international assistance to TB control with the adoption of DOTS as the international standard for TB services (WHO, 1994; WHO Tuberculosis programme, 1994).

Since 1997, the WHO publishes the Global TB report every year with estimates of disease incidence, prevalence and mortality at global, regional and national levels (WHO, 2008). The indicators are shown in absolute numbers and rates per 100,000 population. High burden countries (HBCs) for TB have been designated based on the absolute number of the estimated new TB cases per year, or incidence, in a country. Since then, the WHO’s estimation of TB incidence has been widely used to reflect the burden. According to the Global TB report of 2008, there were an estimated 9.2 million new cases in the world in 2006 (139 per 100,000 population) including 4.1 million smear positive cases. Twenty-two HBCs account for 80% of the global TB burden in terms of incidence (WHO, 2008).

For each country, smear positive incidence per 100,000 population has been the most popular indicator to express a country’s TB burden for a decade. Because one of the two pillars of international outcome targets was to detect 70% or more of existing smear positive cases (a case detection rate of 70% or more), along with a treatment success rate of 85% or more (WHO Tuberculosis programme, 1994), countries have become more interested in incidence. It is estimated that detecting 70% of incident cases and curing 85% of them will cause the TB burden to be halved in 8–12 years (in the absence of HIV) (Styblo and Bumgarner, 1991). The CDR target became a vital force to push countries to improve their case finding and their surveillance systems. It is also a visible indicator at the global level to demonstrate and monitor progress. In 2006, the global estimation of the case detection rate of smear positive TB was 61% and 77 countries met the 70% target (WHO, 2008).

However, as DOTS, and better TB services expanded in countries, several questions were raised concerning the extent to which these estimations are reliable especially at the country level. We observed several discrepancies even within the 22 HBCs: Myanmar achieved 70% detection in 2003, and efforts on further expansion of the service obtained 100% or more CDR consecutively in 2005, 2006 and 2007; Vietnam has been achieving both CDR and treatment success targets since 1997. However, it experienced no decline of case notification; despite extensive efforts, most African countries such as Nigeria, Ethiopia, Tanzania and Uganda fell far short of the 70% CDR target. Is the disease burden in Asian countries underestimated? Is it overestimated in African countries? How do countries reliably demonstrate impact of intervention in terms of decline of incidence? The uncertainty over the current approach to estimation of TB burden has created much discussion, and several countries are now showing interest in surveys to understand the epidemiological burden more accurately.

Current global targets and indicators for TB control have been developed within the framework of the MDGs as well as by the Stop TB Partnership and the WHO’s World Health Assembly (Stop TB Partnership, 2006). The impact targets are to halt and reverse TB incidence by 2015 and to halve the prevalence and death rates by 2015 compared with a baseline of 1990. Although it is questionable how accurately we can estimate a 1990 baseline, it is notable that a measurable indicator derived from community surveys, namely disease prevalence, became one of the indicators of the MDGs. Disease prevalence surveys began to be promoted to more accurately determine the TB situation in countries (Dye et al., 2008; WHO, Western Pacific Region, 2007), even as serious limitations of tuberculin surveys began to be recognized (Dye, 2008; Reider, 1995).

In this chapter, we first summarize the current estimation of the global TB burden by the WHO. Second, we try to provide an overview of current methods to measure burden, and, third, we discuss what might be the best way for a country to measure its TB burden especially in resource poor high burden settings.

2 Global TB Burden and its Trend

The WHO estimated that there were 9.2 million new cases of TB in 2006 (139 per 100,000 population) including 4.1 million smear positive cases (WHO, 2008). Although HIV is often thought to be the cause of the resurgence of TB (Corbett et al., 2003), 0.7 million HIV-positive cases make up only 8% of the total TB cases. However, the African Region has the highest incidence per capita (363 per 100,000 population), while the Western Pacific and the South East Asian Regions accounted for more than half of the incidence, 55%. Twenty-two countries designated as HBC accounted for 80% of the incidence, and 50% of those incident cases in the 22 HBC, 3.8 million, were in the top three countries: India, China and Indonesia (WHO, 2008; Table 67-1 ).
Table 67-1

Epidemiological burden of TB, 2006 estimated by WHO






HIV prev.* in


All forms

Smear positive

All forms

All forms

Incident TB cases














100,000 pop


100,000 pop


100,000 pop


100,000 pop






































South Africa








































































DR Congo












Russian Federation




































UR Tanzania












































































































High burden countries












African Region












American Region












Eastern Mediterranean Region












European Region












South East Asian Region












Western Pacific Region

























This table shows epidemiological burden of TB in 2006 by incidence, prevalence, and mortality

The magnitude of the TB burden in each country can also be expressed as the incidence rate (WHO, 2008; Figure 67-1 ). Higher incidence rates were observed mostly in countries in Africa, South East Asia and the Former Soviet Union. HIV co-infection, poverty and recent experiences of social instability seem to be major factors contributing to the high burden of TB in these countries. There were 29 countries that were estimated to have the highest incidence rate of all TB (300 or more per 100,000 population). All but three (Cambodia, East Timor and Kiribati) were countries in the African continent. The high incidence rates in African countries are partially explained by a high rate of HIV co-infection (Corbett, 2003). This may also explain the higher mortality in the African Region; while the African Region accounted for 31% of the incidence, it accounted for 39% of TB deaths.
Figure 67-1

Estimated TB incidence rate, 2006. Source: Figure 1.3 in page 20 in Global Tuberculosis Control 2008 (World Health organization 2008). This figure shows the estimated incidence rate over the world. Higher incidence rates were observed mostly in countries in Africa, South East Asia and the Former Soviet Union.

According to WHO estimates of global incidence of TB per capita, the incidence rate peaked around 2003 and appears to have stabilized or begun to decline slowly, while the mortality and disease prevalence rate began to decline earlier because these factors respond more quickly to the expansion of case detection and treatment (WHO, 2008; Figure 67-2 ). Among 134 countries that have a reliable series of case notification reports for the decade 1997–2006, data from 93 countries indicate that the incidence rate was falling. However, globally, the slow decline of the incidence rate was overwhelmed by population growth in actually increasing the numbers of cases (the estimated incidence in 2005 was 9.1 million, while that of 2006 was 9.2 million). Decline of HIV prevalence in the general population in high HIV prevalence countries in Africa was mainly contributing to the declining incidence rate of TB in the region as well as that in the world.
Figure 67-2

Estimated global prevalence, mortality and incidence rates, 1990–2006. Figure 1.20 in Global Tuberculosis Control 2008 (World Health Organization 2008). This figure shows estimated global prevalence, mortality and incidence rates. WHO estimates of global incidence of TB per capita, the incidence rate peaked around 2003 and appears to have stabilized or begun to decline slowly, while the mortality and disease prevalence rate began to decline earlier

Recently, the threat of MDR and XDR TB has started to attract significant attention. While most drug sensitive TB can be cured with a 6-month chemotherapy regimen costing 20$ per course, MDR and XDR TB are often fatal and require 2 years for treatment with a cost that is 100 times more expensive than that of drug sensitive TB. Therefore, MDR and XDR TB can be proposed as a significant part of the global burden; it is estimated that 490,000 MDR TB cases emerged in 2006, and 40,000 of those were XDR TB (The WHO/IUATLD Global Project on Anti-Tuberculosis Drug Resistance Surveillance, 2008).

3 How the Burden is Estimated

3.1 Incidence as an Indicator of Burden

As an indicator to decide HBC, the incidence of all TB is usually calculated from the estimation of smear positive incidence in most resource-limited countries, as diagnosis of smear positive TB is mostly standardized and available in most countries. Smear positive TB is said to account for 45% of all TB, even though it has been shown that this proportion varies widely in different population groups. High HIV prevalent countries are given a lower proportion with the assumption that the smear positive accounts for 35% of TB incidence among HIV positives (Corbett et al., 2003; Dye et al., 1999; WHO, 2008).

Although TB incidence is theoretically measurable by a prospective study, it is practically impossible as it is necessary to have a very large population to follow up because cases of TB disease occur only in hundreds per 100,000 a year in the general population even where TB is most frequent. Moreover, because there is no suitable biomarker to detect all TB cases accurately especially in resource poor settings. Thus it is no possible to measure TB incidence directly in the general population. Some may argue that it is easy to study the incidence of smear positive tuberculosis. However, because some cases may convert to negative without treatment prior to regular examination, and because other serious cases may die very quickly, it might be difficult to catch all smear positive incident cases as smear positive TB patients (Toman, 2004). Moreover, to accurately determine the incidence of smear positive tuberculosis must take account of that fact that some patients with smear negative TB might progress to smear positive without treatment, but carrying out such as study would be unethical. It is not feasible to estimate the incidence by direct measurement. We need alternative methods to measure burden and to estimate incidence.

The current estimations of the TB burden by the WHO are based principally on four different methods and their combinations: notification data with an assumption concerning the case detection rate (1); annual risk of TB infection using the Styblo ratio (2); prevalence from surveys with the assumption of an average disease duration (3); and TB mortality from vital registration data used together with case fatality rate (4). The method(s) selected for making estimates for a specific country depend(s) upon the availability of study results and quality of data. They use the following parameters:
  1. 1.

    Incidence = case notification/estimated proportion of cases detected.

  2. 2.

    Incidence (smear positive) =  annual risk of infection (ARI) (%) × 50/100,000.

  3. 3.

    Incidence = prevalence/average duration of condition.

  4. 4.

    Incidence = TB deaths/proportion of incident cases that die (case fatality rate).

We will discuss each method in the following four sections.

3.2 Estimation of Incidence Using Different Methods

Disease prevalence and tuberculin surveys have been conducted in a limited number of countries (WHO, Western Pacific Region, 2007), and most developing countries do not have reliable mortality data from vital registration that covers the whole population. Therefore, incidence for the majority of countries is estimated from notification data, routine surveillance, with assumptions of the case detection rate. Although the WHO’s estimates are the best guess of the country burden with careful review of all available information, it may cause confusion in countries as they calculate one of their key target indicators of CDR by dividing their notification by the incidence estimation that was estimated using an assumption of CDR for the country.

In countries where access to medical services is good for most of the population and where functional disease surveillance exists, notification is considered to be a reliable indicator of disease incidence. The private sector and laboratories are also included in the surveillance system in many of such countries in order to catch as many TB cases as possible. The burden of disease can be estimated correctly through the routine surveillance system in these countries, and this is ideal. However, it is obvious that this method cannot be applied in most TB endemic developing countries, where the surveillance system is weak. Not only does the TB service fail to reach every segment of the population, such as migrants, the very remote, and the urban poor, but the surveillance system often does not even cover public medical facilities outside the disease control services (such as teaching hospitals or central hospitals) nor does it cover the growing private sector.

In these settings, when data from other methods is not available, it is necessary to make a best assumption of the case detection rate. A baseline case detection rate was established in a series of consultation workshops in each WHO region in 1997, country by country (Dye et al., 1999).

If a country has some model or pilot areas of DOTS with improved access and quality of care, the case notification rate of those areas might be considered as a standard, assuming their case detection rate is high. However, such projects might be more likely to be implemented in very remote or hot spots where high incidence is expected or may attract clients from outside the area. Whether the results can be generalized to the entire country should be carefully discussed.

While continuous efforts are essential to expand services to reach those vulnerable patients without access, it is important to make efforts to improve the current disease surveillance system. However, since it may take years to establish it in most developing countries, it is necessary to have alternative ways to estimate the burden.

3.3 Estimation of Incidence from Annual Risk of TB Infection

Although the annual risk of TB infection (ARI) available from tuberculin surveys used to be utilized as a key indicator of TB epidemiology (Cauthen et al., 2002), using ARI to estimate TB incidence is no longer recommended (Dye, 2008; Dye et al., 2008). ARI available from a tuberculin survey has been utilized to estimate TB incidence through applying Styblo’s rule: a 1% ARI was considered to be equivalent to a smear positive TB incidence of 50/100,000 (Styblo, 1985). However, the results of recent tuberculin surveys are hardly interpretable without any anti-mode to distinguish infected from non-infected both in BCG vaccinated and non-vaccinated individuals. High BCG coverage also hampers the interpretation of studies. It has been very difficult to estimate ARI from most recently conducted tuberculin surveys (Dye et al., 2008; National Center for Tuberculosis and Leprosy Control, 2005; Reider, 1995). Moreover, it has been shown that the Styblo model does not fit the epidemiological situation in many countries where surveys were recently conducted (van Leth et al., 2008; National Center for Tuberculosis and Leprosy Control, 2005). One smear positive patient may not infect as many as was estimated; circulation of the infection may be occurring in populations far from children such as in factories in urban areas; schoolchildren might be better protected from TB infection than the general population. Disease incidence previously estimated by using ARI from tuberculin surveys seems to be underestimated. The WHO and its technical task force no longer recommend using a tuberculin survey to estimate TB disease incidence.

However, although the tuberculin survey is not recommended to measure the burden, it might help to estimate epidemiological trends in a country if a series of surveys can be conducted in comparable ways.

 Interferon-gamma release assays (IGRAs) introduce a new technology to diagnose latent infection of TB, and results are less affected by BCG vaccination and infections with mycobacteria other than tuberculosis (Lalvani, 2007; Pai et al., 2004). They may be expected to be utilized for surveys to determine the TB infection rate in the population. However, the technical stability of handling IGRAs with a large number of samples from a survey is still a concern in resource poor settings. Observations of conversion and reversion of results of IGRAs suggest that what is being measured by tuberculin test and by IGRAs may be different (Menzies et al., 2007). Moreover, there marked limitations because of the necessity of taking blood by veno-puncture where the majority of children (those who are negative) cannot obtain any benefit from the survey examination. Therefore, infection surveys with IGRAs among children are not recommended. Development of less invasive, simpler examinations to determine infection is essential.

3.4 Estimation of Incidence from Disease Prevalence Surveys

Like incidence, prevalence is a direct measure of illness in a population. Although measuring incidence is extremely difficult, prevalence can be measured in a single population study, and a series of studies can indicate a trend. Although disease prevalence surveys were conducted widely until 1970, the national level data was available only from Asian countries until recently (World Health Organization, Western Pacific Region, 2007). Since active case detection has not been encouraged for decades because of questions of cost and effectiveness, prevalence surveys that use the same methodology were also suspended. However, many countries are showing interest in disease prevalence surveys to identify the TB burden, incidence, and to determine the real CDR there. However, countries should be aware that to estimate the incidence even with a quality prevalence survey is extremely difficult and may be unreliable.

When smear positive disease prevalence from a survey was used to estimate incidence, prevalence was traditionally divided by two with the assumption that the average duration of a smear positive condition was 2 years. However, successful treatment greatly reduces average duration, drug resistant disease and inadequate treatment may prolong it and HIV infection may also change the average duration of disease, by rapid progression and death in advanced cases. It is necessary to provide country specific disease duration while also considering various factors such as service coverage, HIV prevalence and MDR TB prevalence. For most HBCs, it is necessary to consider six different scenarios: three kinds of patients with different treatment; patients who received DOTS, patients who received non-DOTS treatment, and those who received no treatment; and each of them with and without HIV. This process must always be applied when making such estimates as prevalence is a product of incidence and disease duration.

The proportion of patients who received non-DOTS or treatment in the private sector without notification might be available from a population-based prevalence survey. However, it is difficult to estimate treatment results, or average time to smear conversion. When treatment through non-DOTS accounts for a large proportion of the patients being treated in a country, it may affect the estimation of incidence.

For patients who cannot afford or do not have access to any modern treatment, a historical observation of 2 years is often given to the non-HIV positive. The WHO estimation gives 6 months to HIV positives, considering the rapid progress of disease without treatment (WHO, 2008). However, this assumption among the HIV negative patients is no longer reliable in the chemotherapy era. Characteristics of non-detected may be different from those detected; among those who do not have access to chemotherapy, some are too ill to visit the TB service and die earlier, while some stay in the community without taking action longer because they are not seriously ill. For HIV positive patients, disease duration might be longer than 6 months. Results of intensified case detection through screening cannot explain such a short duration of illness of 6 months (Cheng et al., 2008; Kimerling et al., 2002; Wood et al., 2007).

Therefore, even if a very high quality prevalence study is conducted in a country, the estimated incidence for a country is not easily calculated. For the 2006 estimates, disease duration (= prevalence/incidence ratios of smear positive cases) provided to countries by the WHO varies from 0.8 to 2.4 years, with an average of 1.46 years.

3.5 Estimation of Incidence from Vital Registrations of Deaths

Mortality itself has been a popular indicator of the TB burden, and it is one of the MDGs indicators. When we know the case fatality rate among TB patients in a specific population or in a whole country through a study, the case fatality rate can be used to estimate TB incidence from TB mortality.

The numbers of deaths during treatment are regularly reported to national TB programs and to the WHO. Increase of deaths during TB treatment alarmed TB programs in the 1990s so that they paid attention to HIV associated TB, before they were able to provide HIV testing to TB patients. MDR and X-DR TB also threaten TB patients with increasing treatment failure and deaths. However, in the era of short-course chemotherapy, the majority of deaths are not from TB/HIV or MDR/XDR TB; they are from those with delayed or no access to TB services, diagnosis and treatment. In countries with a working vital registration for deaths, any death should be reported with the cause(s) of death documented by a doctor’s certificate; and the police or local government even has the obligation to conduct an autopsy when the cause of death is uncertain. However, reliable vital registrations for deaths to cover sufficient populations are not available in most TB endemic countries, so the number of reported TB deaths is hardly available not only to measure the TB burden, but also to the estimate incidence.

The verbal autopsy is a method that can be used to clarify the proportion of TB deaths among all deaths. It is conducted to review registered deaths to improve the accuracy of the cause of death statistics (Gajalakshmi et al., 2002; Jha et al., 2006). Some proportion of deaths by “unknown cause” may be reclassified as TB deaths. However, sensitivity and specificity of verbal autopsy have not been accurately assessed, and a very large sample size is necessary to conduct a study for a rare event such as TB death (Yang et al., 2006). Incidence estimation from mortality is applied only in a few countries.

3.6 Estimation of Trends in Incidence

Once the estimation of incidence has been established with the consensus of experts, estimation of change gives incidence for a given year. Notification data, data from routine country surveillance, and regional trends (trends of other countries in the surrounding area) are usually used to estimate change in a country. When there are supportive studies such as prevalence surveys, tuberculin surveys and HIV prevalence studies among the general population as well as among TB patients, those are taken into account to calibrate surveillance data.

4 Prevalence as an Independent Indicator for Millennium Development Goals

Many countries are now showing interest in conducting TB disease prevalence surveys to determine the country’s epidemiological situation more accurately in order to combat disease, while funding agencies and donors also want to determine the impact of their efforts and investment on the burden of the disease. Countries’ primary interest seems to be to estimate incidence from a prevalence study, since a key target of CDR is based on incidence. However, although a series of disease prevalence surveys may provide evidence, countries should understand that the estimation of incidence from prevalence is difficult as we discussed in previous sections in this chapter.

MDGs impact targets are to halt and reverse TB incidence by 2015 and to halve prevalence and death rates by 2015 compared with the baseline of 1990 (STOP TB Partnership, 2006). Although, once again, it is questionable how accurately we can estimate a 1990 baseline, it is an important advance that a measurable indicator obtained from a community study, prevalence, has been adopted as one of the indicators of progress toward the MDGs. Prevalence, measurable by a prevalence survey, can be an independent indicator of TB burden, while other indicators are hardly measurable in most TB endemic countries due to the currently weak surveillance system. It also has the advantage that prevalence changes quickly in response to the expansion and/or improvement of the service that detects and treats patients more frequently and earlier; short-term impact can be observed as well. Moreover, it is the prevalence and not the incidence that determines the probability of transmission in a community.

Countries with an estimated TB prevalence of more than 100 per 100,000 are encouraged to conduct a TB prevalence survey or a series of such surveys, as these are likely to be beneficial in assessing disease burden and trends, and in optimising planning for TB control (China Tuberculosis Control Collaboration, 2004; World Health Organization, Western Pacific Region, 2007).

However, it is essential to standardize survey methods to have accurate estimation and to make international comparison possible. Two major challenges here are screening strategy for bacteriological examination and the availability of culture examinations. There are also major technical limitations to diagnose extra-pulmonary TB and TB in children through community surveys. Therefore, the TB prevalence survey is recommended to target pulmonary TB among adults aged 15 or more. A guidebook on TB prevalence surveys was published by the Western Pacific Regional Office of the WHO, and it is available through the Internet (World Health Organization, Western Pacific Region, 2007).

A prevalence survey is usually recommended only in countries with high prevalence, because a very large number of samples is necessary in a country in those with lower prevalence, and because countries with a high burden of TB often don’t have a notification or surveillance system with proven accuracy and completeness. Although the necessary sample size should be calculated according to established epidemiological methods, a rough idea of the necessary number of study participants for a single survey is available by answering the question: “How many participants are necessary to catch 100 target cases?”: If a country expects a smear positive prevalence of 200/100,000, we need 50,000 participants in a survey to detect 100.

However, if we would like to measure the impact of control efforts, decline of TB prevalence, by performing two surveys within some interval, at least 50–60% more participants are necessary in each survey to show a 30% decline with statistical significance.

4.1 How to Detect TB Cases in the Community

If we could provide smear, culture and chest radiography for all, theoretically, all pulmonary cases could be identified. Moreover, old TB cases with sequelae such as destroyed lung can be detected by chest radiography as a disease burden. However, it is frequently not feasible to conduct all examinations for all in a large-scale survey, considering limited resources. A typical survey in a high burden setting may require 100,000 culture examinations. High cost and the significant demand for laboratory capacity could be major constraints. Alternative approaches should be sought with current technologies depending on the availability of resources. Since a TB disease prevalence survey usually targets bacteriologically confirmed pulmonary TB cases in the community, different screening options could be considered (Table 67-2 ):
  • Screening method I: Sputum collections from all participants for smear.

  • Screening method II: Symptom screening by interview and sputum collection from those with suspected TB symptoms.

  • Screening method III: Screening by chest radiography and sputum collections from those with abnormal shadows.

  • Screening method IV: Screening by tuberculin test, and further examinations including sputum examinations from those with positive reaction.

The method of no screening (to take sputum from every participant for smear) is aimed at not missing any sputum positive patients in the community without screening (Method I) (Sebhatu et al., 2007). It might be the most economical and technically feasible way in a resource-limited setting to detect all smear positive cases in the community. Two or three sputum samples should be taken from every participant. Only smear positive cases can be detected by this method. Although the workload of the laboratory could be a limitation of this method, sputum samples can be screened by fluorescent microscope to cope with a large number of samples. Other limitations of this method are uncertainty of the quality of sputum sample collection from real TB patients and possible lower positive predictive value due to the low prevalence of bacteriologically positives. Additional examinations such as culture and/or chest radiography may be required to confirm a case. Further study is necessary prior to conducting a national prevalence survey only by Method I.
Table 67-2

How to detect TB cases in the community

Screening method

Smear positive

Smear negative Culture positive

TB suggestive by X-ray



No screening (sputum from all)


Theoretically all

Not feasible due to high lab workload


Possible low positive predictive value


Suspected TB symptoms



Less than 50%


Low sensitivity


Chest X ray abnormality

Chest radiography




Cost Capacity of X ray reading, Potential for overdiagnosis


TB infection

Tuberculin test

Probably most

Most, but limited in high HIV settings


Very high positive rate in most HBCs in adults, 3 days after injection


Gamma interferon assay




Cost, Time, Blood taking from vein


This table shows the different screening options to detect TB cases in the community and the weakness of each option

Sputum collection from study participants with symptoms compatible with TB such as chronic cough for 2 or 3 weeks or more identified by interviews (Method II) had been considered a reasonable approach, because it is a standard method of screening in routine case detection practice by NTP in resource-limited countries. And it has been used as a methodology of active case detection in the community. However, past prevalence surveys showed that the sensitivity of symptomatic screening of “TB suspects” by interview is very limited; only 30–70% of smear positive prevalent cases in the community can be identified by interviews (den Boon et al., 2006; Gopi et al., 2006; National Center for Tuberculosis and Leprosy Control, 2005). Characteristics of patients who show up at a clinic to seek a treatment and those of patients staying in the community are different. A survey with a screening methodology only by interviews is not recommended even in resource-limited settings because it underestimates the prevalence.

Chest radiography is usually used as a screening method for bacteriological examinations in prevalence surveys (Method III). Sputum samples are collected from participants with a chest radiography abnormality. Chest radiography is a very sensitive examination to screen TB, but the appearance of TB in chest radiography is not specific (Koppaka and Bock, 2004). Inter-reader discordance is not negligible. Sputum examinations should be performed from those with any abnormal findings, since a significant number of bacteriologically positive patients are detected from “TB healed” and “disease other than TB” categories. Therefore, sputum collection only from “TB suspects” by radiographic findings underestimates the prevalence. The limitation of chest radiography screening is highlighted especially in high HIV prevalent settings; according to the experiences in an intensified TB case detection program among the HIV positive, bacteriologically positive TB patients are detected among those without having any abnormality in a chest radiograph (Bakari et al., 2008; Day et al., 2006).

Method IV may be applied when a survey in children is conducted. As TB disease develops only in those infected, it is not necessary to examine those not infected. However, as already discussed, there are clear limitations to the tuberculin test both in sensitivity and specificity. Tuberculin negative often does not mean non-infected. Tuberculin negative TB is especially common among those infected with HIV (Cobelens et al., 2006). It is difficult to apply this screening method even for a survey among children in high TB burden settings. A waiting time of 3 days to read the result after tuberculin injection is also a constraint in a community survey. Recently, there has been a proposal to use new technology for screening instead of the tuberculin test; IGRAs have the potential to detect latent TB infection and TB disease with high sensitivity. However, the high cost and technical requirements of this method are constraints for a large-scale survey. Moreover, as it requires veno-puncture to collect blood, it is an invasive examination and it cannot be justified to introduce it in a prevalence survey from an ethical point of view.

Considering the advantages and disadvantages of the screening methods discussed above, currently, a combination of Methods II and III is recommended for the survey in TB endemic countries. The combination of interview to ask about symptoms and chest radiography is recommended as a screening strategy; sputum samples should be collected from those with any abnormality in chest radiography and those with symptoms that make them suspects for TB regardless of chest radiographic findings (WHO, Western Pacific Region, 2007). TB patients who are ill but without clear chest radiographic abnormality can be detected by this combined strategy. Having this combined screening strategy is also expected to contribute to quality sputum sample collection. Theoretically, there may be asymptomatic TB patients without any chest radiographic abnormality that cannot be identified by this strategy. But these are expected to be very limited in proportion, and there is no way to diagnose these cases by the medical service.

4.2 Culture Examinations are Recommended to Measure TB Burden and to Confirm TB

There is a debate over whether smear positive TB is enough to measure the TB burden or not. Experience shows that the proportion of smear positive cases among all those bacteriologically positive in the community detected by prevalence surveys has been declining (Hong, 1995; National Center for Tuberculosis and Leprosy Control, 2005; Tupasi et al., 1999; Table 67-3 ). Smear negative and culture positive cases are more than smear positive in most surveys where culture examinations are systemically conducted. This was observed not only in recent surveys in Asia, but also in national surveys in Japan in the 1950s–1960s (Omura et al., 1962; Yamaguchi, 1955) and in the Kolin Study in Czechoslovakia where only 29% of new bacteriologically positive cases were smear positive (Styblo et al., 1967). Quality TB service, DOTS, might have removed smear positive TB cases from the community efficiently enough to decrease smear positive TB prevalence. However, if we do not measure those underlying smear negative/culture positives, we may not be able to determine the real TB burden in the community. Especially under the Stop TB strategy, since all TB cases are targeted in detection and treatment, it is recommended that culture examination should be adopted in prevalence surveys as much as possible. There are other advantages of culture examination: it can identify mycobacterium other than TB that is frequently observed in immuno-compromised individuals and those with a history of TB treatment; the drug susceptibility pattern of patients in the community is available through culture, and it may help the NTP if it does not have national surveillance data.
Table 67-3

Whether is smear positive TB enough to measure the TB burden or not?


X-ray active



S(+)/B(+) (%)



































Republic of Korea

























(Prevalence /100,000 population)


This table shows that the proportion of smear positive cases among all those bacteriologically positive in the community detected by prevalence surveys has been declining

4.3 High Risk Groups and Their Role in Resurgence of TB

Countries may have a concentrated epidemic of TB in some high-risk populations such as migrants, prisoners and intravenous drug users (IVDUs). However, a prevalence survey in a country often targets only residents or registered populations. When TB among those high-risk populations is considered to account for a significant proportion of the country’s prevalence, it is necessary to review the survey design if those populations are properly included in a survey. Another feasible option might be to estimate the TB prevalence and number of those high-risk populations independently by other sources and studies, and add those when we estimate country prevalence from the prevalence survey.

4.4 Prevalence Survey Brings Other Benefits

As a community survey, prevalence surveys can be used to provide more information than simply of the burden of disease. For example, some risk factors such as tobacco use and socio-economic status, and patients’ behaviour and the utilization of the health system (for example, patient’s tendency to first seek TB treatment in the private sector) are often investigated. In addition, the sex ratio between notification and prevalence is also of interest. And, delay analysis (described immediately below) can also be conducted as part of a prevalence survey.

However, since it costs 1 million dollars to conduct a typical single survey to estimate country burden, and because a prevalence survey is very labour intensive, it should be discussed whether a country can afford to increase the sample size 50–60% to detect the impact in a country, or whether we should aim for this kind of impact analysis only at sub-regional, regional and global levels.

5 Delay Analysis

Delay analysis of individual patients can be conducted comparatively easily and provides the information on the service quality and its use. It measures the duration from onset of the disease to starting treatment against TB and analyzes the process of the health seeking behavior of TB patients and diagnosis. Longer duration (delay) before diagnosis and treatment shows the delayed health seeking behavior of the patients (= patient delay) or the poor quality of services (= health service delay). This leads to more severe TB disease in individual patients and increased transmission in the community. As a result, delay analysis can be a useful tool to measure the TB burden in terms of time duration (= delays) of the process before diagnosis and treatment, and also changes of the delays through various interventions, though it has a limitation due to recall bias (uncertain memory about the illness by patients). The Eastern Mediterranean Region of WHO recently conducted the comparative study among seven countries, using this method (WHO/EMRO, 2006).

6 New Technology for Screening and Diagnosis to Measure the Burden

In the near future, the development of molecular bacteriology may bring a new approach to screen study participants and to diagnose TB (STOP TB Partnership Retooling Task Force, 2007). Highly sensitive and non-invasive screening examination in combination with a new culture system such as liquid culture may have the potential to detect all targeted cases, smear positive, culture positive, culture negative, pulmonary, extra-pulmonary and childhood TB.

7 Conclusion

According to the WHO’s estimation of the global TB burden, there were 9.2 million new TB cases in 2006 with 1.7 million deaths due to TB. Although the estimated global incidence rate stabilized or began to fall slowly, the number of new cases was still increasing because population growth exceeded the declining rate of this disease.

However, for each individual country, the uncertainty of the incidence estimation became a concern to monitor and evaluate the progress of the TB situation at country level. There are several ways to measure the burden of TB. It is essential to strengthen the routine recording and reporting system and surveillance, in order to estimate the burden and its trend. However, as the current surveillance system is not always reliable in most TB endemic countries, conducting a prevalence survey is recommended every 5–10 years. Prevalence itself is one of the MDGs indicators, and survey findings can be utilized to calibrate routine surveillance. Combinations with prevalence survey and notification data might be the best method to measure the disease burden right now. Although a prevalence survey costs a lot and it is labor intensive, it may help a country considerably to improve TB control policy. Although a prevalence study cannot be implemented in every country, studies conducted in some countries will help neighbouring countries with similar situations to understand their epidemiological status of TB through the interpretation of routine surveillance data.

8 Case Study

One Country’s efforts to have more accurate measurement of TB burden – CAMBODIA.

In the National TB survey in Cambodia, 2002, BCG scar and Tuberculin surveys, presence of BCG scar and tuberculin test results were analyzed in children aged between 1 year and 15 years. 5,835 children without BCG scar and 5,886 with BCG scar were examined. The proportion of children with BCG scar is higher in the younger age group.

There was no obvious anti-mode to separate populations of infected and non-infected children in any age category (Figure 67-3 ), while tuberculin among TB patients provided a clear histogram (Figure 67-4 ).
Figure 67-3

Tuberculin reaction among children, age 5–9

Figure 67-4

Tuberculin reaction of TB Patients Kg Chlanang and Kg Thom Provinces, Cambodia, 2003

ARI was calculated by two different methods: One was with the conventional cut off point of 10 mm; the other was with a mirror image with 16 mm obtained by a survey of tuberculin reactions among TB patients. There were significant differences between the results using these two methods. It is consequently very difficult to determine estimated ARI (Table 67-4 ).
Table 67-4

Annual risk of TB infection by different methods


Cut-off 10 mm

16 mm mirror


Age group

Point estimate

95% Cl

Point estimate

95% Cl


























Source: National Center for Tuberculosis and Leprosy Control, 2005

If we applied Styblo’s rule to estimate the smear positive TB incidence rate with a point estimate for ARI of 2.8% obtained, the incidence rate for Cambodia would be estimated at 140 (2.8 × 50), and that was almost equivalent to the case notification rate in Cambodia, 141/100,000 in 2002, while the WHO’s estimated incidence at that time was 256/100,000. However, no one assumed because of this that Cambodia had achieved 100% CDR in 2002; the estimation of the TB burden by tuberculin must not have been valid. For the disease prevalence survey, 22,160 participants aged 10 or more underwent an interview and chest radiograph, and those with cough for 3 weeks or more and/or any abnormality in X-ray were asked to submit two sputum specimens for smear and culture.

Eighty-one smear positive and 190 smear negative/culture positive cases were detected. The prevalence of smear positive TB was 269/100,000 population, only half of the WHO’s estimation, 548/100,000, while the prevalence of bacteriologically positive TB was as high as 898/100,000. These results suggest that DOTS was working to decrease the smear positive prevalence in the community. Not only the number and rate, but also the pattern of age distribution of patients suggested that the program was working. Since 1997, the WHO had been attributing to Cambodia an increasing trend of TB incidence mostly due to the spread of HIV infection in the 1990s. However, according to the survey results, the trend was corrected from “increasing” to “declining,” following the Western Pacific Regional trend, although the estimation was still as high as 220/100,000 in 2006. The NTP learned several unexpected things from the survey: as smear positive TB made up only 30% of bacteriologically positive cases, the prevalence of all TB seemed to be higher than the WHO’s estimation; Only 62% of smear positive cases and 39% of bacteriologically positive cases complained of cough for 3 weeks or more; although females accounted for 50% of case notifications in Cambodia, the prevalence of smear positive in males was 2.6 times higher than that in females (Figure 67-5 ). In addition, the prevalence in older persons was extremely high. These findings helped the NTP and partners to develop further plans to expand and improve TB services (National Center for Tuberculosis and Leprosy Control, 2005).
Figure 67-5

Prevalence of bacteriologically positive TB by age and sex

Summary Points

  • The burden of TB can be described in various ways, such as epidemiological burden and financial burden.

  • Estimated disease incidence and its rate have been the most popular indicators of TB burden since 1997 when the WHO began to publish the estimates every year. There were an estimated 9.2 million new TB cases (139/100,000 population) and 1.7 million deaths globally in 2006, according the most recent estimation by WHO.

  • The estimated global incidence rate was reported to have stabilized or begun to fall slowly after a peak in 2003. However, as population growth exceeds the decline in incidence, the number of new cases is still increasing.

  • For an individual country, it is often difficult to have an accurate estimation of the incidence due to limited reliability and coverage of case notification through the routine disease surveillance system. As services expand, discrepancies have been noted between estimates of incidence and notification rates in several countries.

  • Tuberculin surveys are no longer recommended to measure TB burden, while disease prevalence surveys are promoted for countries with expected prevalence of smear positive TB of 100 or more per 100,000, in order to have more accurate figures with which to measure the TB situation.

  • Prevalence is not only an indicator that is measurable by a community survey, but also one of the indicators of the MDGs. A series of prevalence surveys can provide a measure of trend and may show the impact of control efforts. The results of successive surveys can also be utilized to evaluate the data on trend derived from routine surveillance.

  • It is essential for countries to make continuous efforts to improve their disease surveillance system, since it is cost efficient and an ideal way to measure the disease burden of TB as well as other diseases.


  1. Bakari M, Arbeit RD, Mtei L, Lyimo J, Waddell R, Matee M, Cole BF, Tvaroha S, Horsburgh CR, Soini H, Pallangyo K, von Reyn CF. (2008). BMC Infect Dis. 8: 32.PubMedCrossRefGoogle Scholar
  2. Cain KP, Kanara N, Laserson KF Vannarith C, Sameourn K, Samnang K, Qualls ML, Wells CD, Varma JK. (2007). Int J Tuberc Lung Dis. 11: 1008–1013.PubMedGoogle Scholar
  3. Cauthen GM, Pio A, ten Dam HG. (2002). Bull World Health Organ. 80: 503–511.PubMedGoogle Scholar
  4. Chheng P, Tamhane A, Natpratan C, Tan V, Lay V, Sar B, Kimerling ME. (2008). Int J Tuberc Lung Dis. 12 (Suppl 1): 54–62.PubMedGoogle Scholar
  5. China Tuberculosis Control Collaboration. (2004). Lancet. 364: 417–422.CrossRefGoogle Scholar
  6. Cobelens F, Egwaga SM, van Ginkel T, Muwinge H, Matee MI, Borgdorff MW. (2006). Clin Infect Dis. 43: 634–639.PubMedCrossRefGoogle Scholar
  7. Corbett EL, Watt CJ, Walker N, Maher D, Williams BG, Raviglione MC, Dye C. (2003). Arc Intern Med. 163: 1009–1021.CrossRefGoogle Scholar
  8. Day JH, Charalambous S, Fielding KL, Hayes RJ, Churchyard GJ, Grant AD. (2006). Int J Tuberc Lung Dis. 10: 523–529.PubMedGoogle Scholar
  9. den Boon S, White NW, van Lill SW, Borgdorff MW, Verver S, Lombard CJ, Bateman ED, Irusen E, Enarson DA, Beyers N. (2006). Int J Tuberc Lung Dis. 10: 876–882.PubMedGoogle Scholar
  10. Dye C, Scheele S, Dolin P, Pathania V, Raviglione MC. (1999). JAMA. 282: 677–686.PubMedCrossRefGoogle Scholar
  11. Dye C. (2008). Bull World Health Organ. 86: 4.PubMedCrossRefGoogle Scholar
  12. Dye C, Bassil A, Bierrenbach AL, Broekmans JF, Chadha VK, Glaziou P, Gopi PG, Hosseini M, Kim SJ, Manissero D, Onozaki I, Rieder HL, Scheele S, van Leth F, van der Werf M, Williams BG. (2008). Lancet Infect Dis. 8: 233–243.PubMedCrossRefGoogle Scholar
  13. Gajalakshmi V, Peto R, Kanaka S. Balasubramanian S. (2002). BMC Public Health 2: 7.PubMedCrossRefGoogle Scholar
  14. Gopi PG, Subramani R, Sadacharam K, Narayanan PR. (2006). Int J Tuberc Lung Dis. 10: 343–345.PubMedGoogle Scholar
  15. Hong YP, Kim SJ, Lew WJ, Lee EK, Han YC. (1995). Int J Tuberc Lung Dis. 2: 27–36.Google Scholar
  16. Jha P, Gajalakshmi V, Gupta PC, Kumar R, Mony P, Dhingra N, Peto R; RGI-CGHR Prospective Study Collaborators. (2006). PLoS Med. 3(2): e18.PubMedCrossRefGoogle Scholar
  17. Kimerling ME, Schuchter J, Chanthol E, Kunthy T, Stuer F, Glaziou P, Ee O. (2002). Int J Tuberc Lung Dis. 6: 988–994.PubMedGoogle Scholar
  18. Koppaka R, Bock N. (2004). In: Toman’s Tuberculosis Case Detection, Treatment and Monitoring: Questions and Answers, 2nd ed. World Health Organization, Geneva, pp. 51–60.Google Scholar
  19. Lalvani A. (2007). Chest. 131: 1898–1906.PubMedCrossRefGoogle Scholar
  20. Menzies D, Pai M, Comstock G. (2007). Ann Intern Med. 146: 340–354.PubMedGoogle Scholar
  21. Murray CJ, DeJonghe E, Chum HJ, Nyangulu DS, Salomao A, Styblo K. (1991). Lancet. 338(8778): 1305–1308.PubMedCrossRefGoogle Scholar
  22. National Center for Tuberculosis and Leprosy Control. (2005). National tuberculosis prevalence survey, 2002, Cambodia. Phnom Penh: Royal Government of Cambodia, Ministry of Health, Phnom Penh.Google Scholar
  23. Omura T, Oka H, Kumabe H, Kobayashi A. (1962). Bull World Health Organ. 26: 19–45.PubMedGoogle Scholar
  24. Pai M, Riley LW, Colford JM. Jr. (2004). Lancet Infect Dis. 4: 761–776.PubMedCrossRefGoogle Scholar
  25. Reider HL. (1995). Tubercle Lung Dis. 76: 114–121.CrossRefGoogle Scholar
  26. Sebhatu M, Kiflom B, Seyoum M, Kassim N, Negash T, Tesfazion A, Borgdorff MW, van der Werf MJ. (2007). Bull World Health Organ. 85: 593–599.PubMedCrossRefGoogle Scholar
  27. STOP TB Partnership Retooling Task Force. (2007). New Technology for Tuberculosis Control: A Framework for their Adoption, Introduction and Implementation. World Health Organization, Geneva.Google Scholar
  28. STOP TB Partnership. (2006). The Global Plan to Stop TB, 2006–15. World Health Organization, Geneva.Google Scholar
  29. Styblo K, Dankova D, Drapela J, Galliova J, Jezek Z, Krivanek J, Kubik A, Langerova M, Radkovsky J. (1967). Bull World Health Organ. 37: 819–874.PubMedGoogle Scholar
  30. Styblo K, Bumgarner JR. (1991). Tuberculosis Surveillance Research Unit Progress Report: 2: pp. 60–72.Google Scholar
  31. Styblo K. (1991). Epidemiology of Tuberculosis, 2nd ed. KNCV Tuberculosis Foundation, Hague, The Netherlands.Google Scholar
  32. Styblo K. (1985). Bull Int Union Tuberc Lung Dis. 60: 117–119.Google Scholar
  33. The WHO/IUATLD Global Project on Anti-Tuberculosis Drug Resistance Surveillance. (2008). Anti-Tuberculosis drug resistance in the world, Report No.4. World Health Organization, Geneva.Google Scholar
  34. Toman K. (2004). In: Toman’s Tuberculosis Case Detection, Treatment and Monitoring: Questions and Answers, 2nd ed. World Health Organization, Geneva, pp. 66–71.Google Scholar
  35. Tupasi TE, Radhakrishna S, Rivera AB, Pascual ML, Quelapio MI, Co VM, Villa ML, Beltran G, Legaspi JD, Mangubat NV, Sarol JN Jr., Reyes AC, Sarmiento A, Solon M, Solon FS, Mantala MJ. (1999). Int J Tuberc Lung Dis. 3: 471–477.PubMedGoogle Scholar
  36. van Leth F, Vander Werf MJ, Borbodorff MW. (2008). Bull World Health Organ. 86: 20–24.PubMedCrossRefGoogle Scholar
  37. WHO Tuberculosis Programme. (1994). Framework for Effective Tuberculosis C control. World Health Organization, Geneva.Google Scholar
  38. WHO, EMRO. (2006). Diagnostic and treatment delay in tuberculosis, an in-depth analysis of the health-seeking behaviour of patients and health system response in seven countries of the Eastern Mediterranean Region. WHO, EMRO, Cairo.Google Scholar
  39. Wood R, Middelkoop K, Myer L, Grant A, Whitelaw AD, Lawn SD, Kaplan G, Huebner R, McIntyre J, Bekker LG. (2007). Am J Respir Crit Care Med. 175: 87–93.PubMedCrossRefGoogle Scholar
  40. World Bank. (1993). World Development Report 1993: Investing in Health, Oxford University Press, New York, pp. 25–29.Google Scholar
  41. World Health Organization. (2008). WHO Report 2008, Global Tuberculosis Control: Surveillance, Planning, Financing. Geneva.Google Scholar
  42. World Health Organization, Western Pacific Region. (2007). Assessing Tuberculosis Prevalence Through Population-based Surveys. Manila.Google Scholar
  44. World Health Organization. (1994). TB: A Global Emergency, WHO Report on the TB Epidemic. World Health Organization, Geneva.Google Scholar
  45. Yamaguchi M. (1955). Bull World Health Organ. 13: 1041–1073.PubMedGoogle Scholar
  46. Yang G, Rao C, Ma J, Wang L, Wan X, Dubrovsky G, Lopez AD. (2006). Int J Epidemiol. 35: 741–748.PubMedCrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media LLC 2010

Authors and Affiliations

  • I. Onozaki
    • 1
  • N. Ishikawa
    • 1
  • D. A. Enarson
    • 2
  1. 1.The Research Institute of TuberculosisJapan Anti-Tuberculosis Association MatsuyamaKiyoseJapan
  2. 2.International Union against Tuberculosis and Lung DiseaseParisFrance

Personalised recommendations