Global epidemiology of paediatric tuberculosis

Abstract

Tuberculosis is one of the major infections affecting children worldwide. It causes significant morbidity and mortality, especially in infants and young children. Factors such as overcrowding, poverty and the HIV epidemic have all contributed to the resurgence of tuberculosis globally. The highest rates of tuberculosis occur in resource-poor countries and over the last decade case notifications in children have been increasing steadily, particularly in Sub-Saharan Africa. Mycobacterium tuberculosis infects millions of children worldwide every year, yet accurate information on the extent and distribution of disease in children is not available for most of the world.

We describe some of the unique aspects of tuberculosis infection in children and review the epidemiology of disease in children worldwide.

Introduction

Tuberculosis (TB) continues to be one of the most devastating and widespread infections in the world. It is an important cause of morbidity and mortality in children worldwide, especially in resource poor countries. In 1990, the World Health Organisation (WHO) estimated that 7.5 cases of TB occurred worldwide, with approximately 650,000 cases occurring in children.¹ This estimate was revised in 1994, and a prediction made that by 2000 this figure would have increased to more than one million childhood cases, making up 10% of all cases.² Many western nations have been experiencing a resurgence of TB with most of the new cases imported from countries with high rates of the disease. Young children carry the greatest burden of disease. They are the most likely to develop disease after infection and are significantly more likely to develop extrapulmonary and severe disseminated disease than adults. Infected children represent the pool from which a large proportion of future cases of adult TB will arise. In addition, childhood TB is a sentinel event, indicating ongoing transmission of TB within communities.

Despite the importance of paediatric TB there is very little information about the epidemiology of TB in children worldwide. WHO data for children with TB only include the small percentage of children who are sputum smear positive. In much of Sub-Saharan Africa where the concurrent HIV epidemic is speeding the TB epidemic and the number of adult TB cases is rapidly increasing, the rates of TB in children are unknown. This is an epidemic that in most of the world is spreading unnoticed.

We describe the unique aspects of paediatric TB and review the literature on the epidemiology of paediatric TB worldwide.

Infection with Mycobacterium tuberculosis can lead to a variety of outcomes. A child is said to have been infected with the bacteria if they have a positive tuberculin skin test (TST) without signs or symptoms of disease, and have a normal chest radiograph. Tuberculosis disease occurs when clinical manifestations of tuberculosis become apparent either by clinical signs and symptoms or by chest radiograph. The term TB is generally used to refer to those who have disease and not those with infection only.³

The presentation and clinical features of tuberculosis in the infants and children are generally different to that in the adult population. Firstly, children are significantly more likely than adults to develop disease following infection. The risk of developing disease after being infected with M. tuberculosis has been estimated to be 5–10% in adults, 15% in adolescents, and 24% in children 1–5 years of age. In infants less than a year of age this figure is estimated to be as high as 43%.⁴ In addition, disease in children is likely to occur early after infection, termed progressive primary infection. The time between asymptomatic infection and development of symptomatic disease in children is generally between 1 and 6 months. Thus the clinical distinction between primary infection and disease is not always clear in children, as it usually is in adults.⁵

In addition to developing disease earlier than adults, children are also as much greater risk of developing extrapulmonary disease. Superficial lymphadenitis is the most common manifestation of extrapulmonary TB in children, but disease also can occur at other locations such as the central nervous system, bones, joints, and skin. Ussery et al.⁶ found that of 14,414 paediatric (<15 years of age) cases of tuberculosis reported in the US between 1985 and 1994, 71% were reported to have pulmonary disease only, while 22% had only extrapulmonary disease. If only children under the age of four are considered the rate of extrapulmonary tuberculosis is as high as 25%. This compares with a rate of extrapulmonary disease for all tuberculosis cases of 16% reported by the Centre for Disease Control (CDC) between 1991 and 1994.

Two of the most severe manifestations of tuberculosis, miliary disease and tuberculous meningitis, are significantly more common in children. Miliary disease is an early complication of primary infection, which occurs following haemotogenous dissemination and spread to two or more organs. The clinical presentation can be highly variable, depending on the numbers of organisms released into the circulation.⁵ In a study of paediatric tuberculosis in California between 1985 and 1995, Lobato et al.⁷ found that children 0–4 years accounted for 71% of miliary disease. Meningitis is the most feared complication of tuberculosis in children, particularly due to neurological sequelae including blindness, deafness, intracranial calcification, diabetes incipidus and mental retardation. It usually occurs within 3–6 months of initial infection and again is most common in the younger children, with those 0–4 years of age accounting for 80% of meningitis cases.⁷

Tuberculosis is often difficult to diagnose in children, even in those with advanced or severe disease. Young children seldom cough or produce sputum and hence few children are sputum smear-positive. Early morning gastric aspirates and induced sputum production are techniques used to enhance detection of M. tuberculosis but even with these, microbiological confirmation is lower than for adults. Less than 20% of children with proven TB will have a sputum or gastric aspirate sample that is positive on acid-fast bacilli stain, compared to 75% in adults.8., 9. The diagnosis of TB is therefore dependant in many cases on the clinical history and presentation, a history of an infectious contact and investigations such as chest radiograph and the TST. Furthermore, the TST can be negative in up to 10% of culture proven TB, and in up to 40% of children with tuberculous meningitis.⁸ This has been attributed to delayed skin test conversion, or weaker immune responses in those with severe disease. Other factors that can contribute to false-negative TST results include associated immunosuppressive illnesses such as HIV, malnutrition, or other bacterial or viral infections. Some of these children will become reactive during treatment suggesting that TB disease itself may contribute to immunosuppression. More than 50% of childhood TB cases will be asymptomatic initially and require chest radiographs to confirm the diagnosis.¹⁰ In countries that do not have such diagnostic facilities universally available, many children with TB will not be diagnosed. A recent autopsy study in children dying of respiratory disease in Zambia found that 20% of cases had evidence of TB, which in many cases was only diagnosed post-mortem.¹¹

An accurate description of the burden of TB on children worldwide is very difficult to obtain. In many countries surveillance data are often unreliable because of poor diagnostic facilities and reporting systems. Because of these issues the International Union Against Tuberculosis and Lung Disease has stated that reliable information on the incidence of TB in childhood can only be obtained in developed countries.¹²

The best information available on global trends in TB comes from the WHO.¹³ They collect data submitted from countries throughout the world annually to monitor disease incidence and the effectiveness of local TB control programmes. The case definition for TB is based on the presence of M. tuberculosis on sputum smears. This means that more than 80% of children with TB will not be represented in WHO figures as they will not be smear positive. The proportion of TB cases in individual countries that occur in children is highly variable. In low-prevalence countries this may be <5% whereas in some high-prevalence countries it is estimated to be four times this figure. Thus, WHO figures cannot be used to accurately estimate rates of paediatric TB within a country, and epidemiological studies in children are required to obtain a more accurate estimate of the incidence of TB in childhood. Unfortunately for most of the world this type of information is not readily available, and when it is available differences in methodology and case definitions make comparisons between countries difficult. A few countries do have surveillance systems in place with case definitions that take into account the different paediatric presentations, thus giving more accurate data for TB in children. Where possible this review uses information based on such studies, but for much of the world estimates of childhood TB are still based on trends in the prevalence of adult disease and information from small, localised studies.

Tuberculosis case notifications in Europe make up less than 10% of worldwide notifications to the WHO. Case notifications reveal alarming disparities between countries in western Europe, and those in eastern Europe, comprising mostly states of the former Soviet Union. For example, WHO figures for 2001 show both France and the UK had notification rates for all cases of 10 per 100,000 population. This is compared to Romania and the Russian Federation whose rates are greater than 130/100,000 and Kazakstan with rate of 181/100,000. Case notification rates in France and the UK have fallen since 1980 and in the last 5 years remained relatively unchanged. In contrast many eastern European countries have steadily increasing rates of TB notifications which shown no sign of abating. In 1995 the Russian Federation had 84,980 notifications, increasing to 132,477 by 2001. In the same time period notifications in Kazakstan went from 11,310 to 26,224.¹³

There is very little paediatric data on tuberculosis available from eastern European countries. The WHO had 601 notifications of smear positive children aged 0–14 years in the European region for 2001, with one in five of these children living in Kazakstan.¹³ These figures are likely to significantly underestimate the true rate of smear-positive TB in children for several reasons. Firstly, in Kazakstan the total numbers of paediatric notifications makes up just over 1% of all notifications, with even lower reported rates in children for the Russian Federation. This is substantially lower than in the UK where children make up 2.5% of all notifications and suggesting that these figures may reflect poor detection or reporting of paediatric cases.

Another reason that higher rates of paediatric TB notifications might be expected in these high-burden countries is the age distribution of adult TB. In many eastern European countries particularly high rates of adult disease occur in younger adults in the 25–44 years age bracket. In the Russian federation case notifications are three times greater amongst those 25–34 compared to those over 65 years. In the UK notifications in those over 65 are almost equal to those in the younger age group (Fig. 1). More younger adults in eastern Europe having active TB is likely to mean that more children are exposed to TB from their parents or carers, and hence higher rates of disease can be expected in children.

In most of the low-prevalence countries in Europe, such as the UK, notification rates for TB have declined over the last 20 years.¹³ National survey data from the UK show a decrease in TB rates in all age groups from 1978/9 onwards, reaching their lowest levels in the mid-80s before beginning to rise again. Rates in children and young people overall have remained relatively constant over the last 5 years.¹⁴

However there have been specific changes in the pattern and distribution of paediatric tuberculosis in the UK that are not reflected in national data. Overall, TB notifications have increased substantially in London, which accounts for 40% of national cases. Several boroughs in London now have TB notification rates greater than 60/100,000, with some having rates >120/100,000. Cases in children under 16 in London have risen almost every year since 1988, with increases across all age groups. Some areas have recorded rates of >40/100,000 in children.¹⁵ The proportion of Black African children with TB in 1998 (44%) had increased substantially from 1993 (23%), whereas the proportion paediatric cases from the Indian Subcontinent (ISC) had fallen (21% in 1998 compared to 50% in 1993). The data from 1998 show that 66% of the African children with TB were born abroad and that the majority of those developed disease within 5 years of entering the UK.

Other urban areas in the UK have shown similar trends with an overall decrease in TB notification rates but a greater proportion of cases coming from ethnic minorities. In Birmingham the overall number of TB cases decreased steadily between 1989 and 1994.¹⁶ A substantial proportion of all paediatric cases in the city were either ISC or West Indian. Data from Leeds, where TB notification rates are low, also showed a high proportion of ISC children.¹⁷

Similar patterns have occurred in other low-prevalence countries in Europe. Eriksson et al.¹⁸ reviewed the epidemiology of paediatric TB in Stockholm between 1976 and 1995. They found the overall rates of TB in children increased from <1/100,000 early in the study to 5.8/100,000 between 1991 and 1995. The increase in case notifications was mostly due to children born outside the country (50% of children with TB were from Africa) or who had at least one parent who was born abroad. There were no cases of TB in children whose parents were both Swedish born in the years 1991–1995. A similar study in Copenhagen between 1984 and 1993 found that 70% of children with TB had immigrant parents.¹⁹

WHO case notification rates for TB show alarming numbers of TB cases in Sub-Saharan Africa. Countries of Sub-Saharan Africa make up a large proportion of high-burden countries listed by the WHO, with the highest numbers of case notifications occurring in Nigeria and South Africa (Table 1). Although the estimated number of cases overall are substantially lower than in some Asian and Western Pacific countries the estimated rates per 100,000 population are the highest anywhere in the world. Both South Africa and Zimbabwe have estimated case rates three times greater than those of the Russian Federation and double that of most of the other high-burden countries. In addition most Sub-Saharan African countries have had substantial increases in the TB case notification rates over the last 10 years.¹³

Many of the countries in this region are resource-poor, making reliable estimates of TB incidence difficult, especially in children. Lack of resources makes accurate diagnosis of TB cases more difficult, and in many counties TB control programs rely almost exclusively on sputum microscopy, as part of WHO TB control strategies, for the diagnosis of TB. This means that TB in children is both underreported and undiagnosed. Inadequate health information and surveillance systems in many countries may also contribute to inaccuracies in the available data. One study in the Western Cape of South Africa showed that of 689 cases of TB meningitis in children aged <14 years, only 55% had been notified.²⁰

It is clear that in Sub-Saharan Africa, paediatric cases make up a substantially higher proportion of overall TB cases than in other countries. While in most resource-rich countries children make up 5% or less of the TB caseload, one recent paper estimated that in African countries this figure is likely to be closer to 20%.²¹ Although these estimates are open to dispute, several studies in resource-poor countries would agree with this figure.22., 23. A study in an urban community in the Western Cape Province of South Africa (overall TB incidence estimated 1149/100,000) found that 39% of the caseload was due to children younger than 14 years of age.²⁴

The annual risk of TB infection (ARTI) has been used to estimate TB infection in children. This is obtained from TST surveys in similar populations of children, in the same places in different years. In Tanzania, where adult TB notifications have doubled in the last decade, the Tanzanian Tuberculin Survey Collaboration estimated the ARTI in grade 1 and 2 school children.²⁵ They showed that there was a reduction in the ARTI from 1% in the period 1988–1992 to 0.9% for 1993–1998. However, in areas where TB notification rates were the highest the ARTI had increased. They concluded that HIV prevalence among school children was likely to be low and that HIV-induced anergy was unlikely to have biased their findings. A Kenyan study found an overall increase in the ARTI from the period between 1986 and 1990 to 1.1% in the period 1994–1996.²⁶ Similarly they found that areas with the highest incidence of TB, such as Nairobi, had the highest increase in ARTI (2.9% Nairobi). The districts with high prevalence rates for TB were also found to have high prevalence rates for HIV infection. Increasing rates of HIV infection will make the interpretation of future ARTI studies much more difficult.

The South-East Asia region has the largest number of people with TB in the world, with almost four times as many case notifications as Europe in 2001. Three countries, India, Indonesia and Bangladesh are in the top four high-burden countries listed by the WHO, and India alone had an estimated 1.8 million cases of TB in 2001. The two countries with the lowest rates in this region are the Maldives (22/100,000) and Sri Lanka (26/100,000), whose rates are twice that for western Europe. Annual WHO case notifications have been rising steadily over the last 5 years in DPR Korea, Indonesia, Myanmar, Nepal, and Thailand. In India, after peaking in 1999, case notifications have fallen and in 2001 were the lowest since 1993, possibly due to widespread implementation of DOTS programs. This has meant the overall notification rates for the region have remained essentially unchanged for the region between 1999 and 2001.¹³

There is little information on rates of paediatric TB in this region. In the Chingleput district of Tamil Nadu State, South India, it has been estimated that the ARTI is 2%.²⁷ Although this is high, the study found that this had remained unchanged over 30 years suggesting the TB risk for children is not increasing. However, it also highlights how ineffective the local TB control measures have been, with ongoing TB transmission within the community.

In 2001, 3188 smear-positive children were notified to the WHO in India. However, these figures are unlikely to be indicative of the full burden of disease in children. Chakraborty estimated that approximately 10 million children in India each year are at risk of being infected because of contact with a smear-positive adult.²⁸ Given that 5% could be expected to develop TB in the first 2 years following infection and another 2–3% over the next 2 years, he estimated that 7 of 800,000 children would develop TB over the subsequent 5 years. If this was the case over 7–800,000 children could be expected to die or suffer serious disability over the same 5-year period.

WHO TB case notification figures for the Western Pacific region are dominated by China, which had an estimated 1.4 million cases in 2001. It has the second largest number of new TB cases in the world, mostly due to its vast population, with an overall estimated notification rate of 51/100,000. Apart from China, the Western Pacific region is mostly made up of small nations, generally with estimated TB rates of less than 50/100,000. While the rates of TB have increased in the last 5 years in some countries, only Vietnam shows increases in rates similar to those in high-burden Sub-Saharan African countries. Case notifications in Vietnam increased from, approximately, 50,000 in 1990 to over 90,000 TB cases notified in 2001. One of the success stories in the region has been the Phillipines, which has implemented DOTS programs widely. It has had high rates of TB for many years, but estimated rates of disease have been falling steadily since 1995 (139/100,000) to a low of 77/100,000 in 2001. Excluding China, most countries have very low numbers of smear positive cases in children reported to the WHO.¹³

One study in Australia demonstrates clearly how immigration from high prevalence countries can contribute significantly to the disease pattern within a low prevalence population. Heath et al.²⁹ studied TB surveillance data for New South Wales between 1975 and 1995. During this period the overall TB notification rates for Australian-born population fell exponentially to around 2/100,000, while the percentage of notifications in overseas born people rose from 30 to 79%. In children <15 years of age over this time period, rates initially fell, but rose again from 1992 to 1995. Notification rates were highest in children born overseas, while 51% of Australian-born children with TB were from non-English speaking households.

In general, TB notifications to the WHO in this region have remained unchanged, or have fallen in countries in this region since the later part of the last decade. This region does not contribute substantially to the worldwide burden of TB, and has the lowest rates of disease in all the WHO regions. The highest TB notification rate for 2001 was in the small nation of Djibouti (652/100,000). Other nations with high rates include Morocco (93/100,000), Somalia, Sudan (both 75/100,000) and Yemen (68/100,000).¹³

The country with the largest population in the region, Pakistan, has had a significant reduction in TB notifications between 1980 and 2001. However, case detection rates in Pakistan are very low, with the WHO estimating that only 10% of sputum smear-positive cases are being detected.¹³ While the case notification rate for Pakistan was 23/100,000 in 2001, the WHO report estimated the rate for all cases to be 171/100,000. Apart from WHO reports, no data is available for childhood TB in this region.

The total number of TB case notifications for this region has remained remarkably constant since 1980. This is despite the two countries with the greatest number of TB cases, Brazil and USA, having both had steady decreases in case notifications since 1998. Overall TB cases in the USA peaked in 1992, and have been steadily falling since that time. In 2002 the CDC reported an estimated overall TB rate of 5.6/100,000, the lowest since reporting began in 1953.³⁰ California and New York have the highest numbers and rates of TB of all states, but both have had substantial reductions in TB rates since 1992. In New York the rate of case notification fell by over 70% during this time period. The countries in this region with the highest estimated rates of TB per 100,000 population are Bolivia (216), Dominican Republic (133), Ecuador (141), Guyana (103), Haiti (328) and Peru (196).¹³

Most of the paediatric data in this region comes from the USA, where there is a comprehensive system of notifications to the CDC in Atlanta. The CDC uses a clinical case definition that includes a positive TST, a completed set of diagnostic investigations, signs and symptoms compatible with TB and treatment with two or more anti-tuberculous drugs. National rates in children 0–14 years of age have followed a similar pattern to that seen in adult TB, falling from 3.1/100,000 in 1992 to 1.5/100,000 in 2001. Over this time period the percentage of overall cases that occurred in this age group has remained constant at 6%. For children under 5 years, and in the 5–14 years category TB case rates are the highest for ethnic minority groups. For example, in 2001 the rates for white children under 5 years was 0.5 compared to Hispanic, Black, Non-Hispanic and Asian/Pacific Island children whose rates were 7.0, 6.2, and 6.5 per 100,000, respectively. In 2001, approximately half of cases reported to the CDC occurred in people born outside the USA, with nearly a quarter of these from Mexico.³⁰ In California the overall rates of paediatric TB have been falling and since 1992, this decline has been greatest in children born overseas.⁷ Between 1985 and 1995 in California, foreign-born children accounted for 42% of all TB cases in children. However, one study in San Diego found that over 90% of US-born children under 5 with TB had a parent or source case from a country where TB is highly endemic or a primary household language other than English.³¹

One population-based study conducted in seven states, looked at the sources of TB in children.³² They analysed data from CDC and health department records found 111 culture confirmed cases in children <5 years of age, 59% had at least one epidemiologically related TB case from public health investigations. Of particular concern is that 14% of these children had more than one possible related source suggesting that multiple exposures may need to be considered when assessing sources. In 72% of suspected child–adult pairs the source patient lived in the child's household and in 15% of the source patients their infection resulted in more than one young child being infected. This underlines the importance of case detection and completion of treatment within communities for TB control.

Childhood TB is closely linked to adult disease, and in most cases is a sentinel event in the community, representing recent transmission. Many of the factors contributing to the spread of adult disease will indirectly affect disease in children. A complex interplay of factors such as immigration, poverty, social deprivation and other diseases affecting children and adults affect the epidemiology of paediatric TB within a community. The effectiveness of TB control programs worldwide is also highly variable, meaning that in some places the disease continues to spread unchecked. There are significant differences in TB epidemiology between many western nations where TB is generally well controlled, and resource-poor nations where rates of TB in children continue to climb.

During the latter half of last century there has been an unprecedented movement of people between nations. Few studies have looked at the effect of migration in rates of TB in resource-poor nations, whereas in western countries with active TB surveillance programs this has been well studied. The movement of people from countries with high rates of TB to resource-rich nations has had a major impact on the epidemiology of TB in those nations. This is particularly evident in the USA where a significant proportion of new TB cases occur in recent immigrants.³⁰ In Canada, the overall risk of TB was found to be 12 times higher in immigrants compared to those people born in Canada.³³ In Europe, studies from Norway, Denmark and the UK have all demonstrated high rates of adult disease in those born overseas while in non-immigrant population rates of TB are low.16., 34., 35., 36., 37. Molecular epidemiological studies undertaken in Norway and the UK both suggest that many of the new TB cases in immigrants are due to reactivation of infections acquired abroad.37., 38., 39.

Immigrant children born in countries with high rates of TB are at much higher risk of developing TB for several reasons. Firstly, those who are recent arrivals may have been exposed to TB in their home country. One study of internationally adopted children in the USA found that 19% were TST positive.⁴⁰ In addition, these children are more likely to be exposed to TB either because of travel back to their country of origin, or exposure to infectious adults within their home. The long time between infection and reactivation of TB in adults means that this increased risk of disease in children extends to those whose parents were born overseas even if they themselves are born in a low-risk country.

TB has long been thought to be associated with overcrowding and poverty in communities, and several studies worldwide in children support this. Indeed, the decline on TB in the 20th Century occurred even before the introduction of antibiotics and BCG vaccination. Drucker et al.⁴¹ studied childhood TB in the Bronx, using Department of Health data from 1970–1990 and following all TB cases in children between 1986 and 1992 during a resurgence of TB in New York. There was a striking temporal relationship between overall TB rates and residential overcrowding in the period 1970–1990. The overall TB case rate in children increased substantially with the proportion of crowded households in neighbourhoods between 1986 and 1992. In Cape Town South Africa, van Rie et al.²⁴ found a significant correlation between TB case notification rates in childhood and crowding, economic status and parental education. Overcrowding showed the weakest association, but when the case definitions for overcrowding used in the New York study were applied to that in Cape Town, 100% of people would be classified as living in crowded conditions. It has been suggested that the weak association in such a crowded community may be due to a saturation effect, whereby a certain level of crowding will influence transmission of disease but will not change as crowding increases.

Elsewhere in the world in populations with low TB rates similar results have been found. In Leeds, in the UK (TB incidence rates were 3.9/100,000 1982–1997), Parslow et al.¹⁷ found that overall risk of TB was linked to deprivation, population density and ethnicity.

HIV infection has had a profound effect on the incidence of TB particularly in Sub-Saharan Africa where up to a quarter of the population are HIV infected. Increasing rates of TB in these countries has occurred at the same time as the emergence of the HIV pandemic. HIV is known to greatly increase the annual risk of progression from TB infection to active TB and is thought to be one of the principle causes of the resurgence of TB in this region. Cantwell and Binkin⁴² found that the average annual case rates for TB after 1985 increased approximately, twice as fast in countries with high versus low or intermediate HIV seropositivity ratings.

Information on how the HIV epidemic has impacted on the rates of paediatric TB are limited, but it seems likely that it has contributed to the increasing rate of disease in children in high prevalence countries. In Brazil, the numbers of children with TB in Rio de Janeiro City who were co-infected with HIV have been steadily rising. In 1995, 23.5% of these children were HIV co-infected, whereas, this figure had climbed to 31.4% by 1999.⁴³ Similarly high rates have been noted in South Africa, where in one study 48% of children under 12 years of age with culture proven pulmonary TB were also HIV infected.⁴⁴ Despite the high rates of co-infection with HIV and TB in these children, it is still unclear whether HIV-infected children are more vulnerable to TB infection or more likely to progress to disease than HIV-negative children.

Increased rates of childhood TB have been associated with increased rates of disease among HIV-infected adults in the community. In Sub-Saharan Africa it has been suggested that the effect of HIV infection as a risk factor for TB in children may be indirect.45., 46. An important risk factor for TB in HIV-infected children is that they are more likely to be a close contact of a smear-positive adult than HIV uninfected children. Because TB is the most common opportunistic infection in HIV infected adults in the developing world, children in this setting are more likely to be exposed to TB irrespective of their HIV status. In New York, Thomas et al.⁴⁷ found that HIV-infected children were at higher risk of TB than uninfected children born to HIV-infected mothers. Both groups had higher rates than other children in New York suggesting that increased exposure to TB may also be a contributing factor irrespective of the child's HIV status.

In children in resource-poor countries who are HIV infected, TB is likely to be underdiagnosed, and treatment outcomes may be poor. Not only is the diagnosis difficult because of HIV associated anergy on TST, but also the clinical presentation of pulmonary TB overlaps with other opportunistic pulmonary infections and HIV related illnesses, such as lymphoid interstitial pneumonitis. Children who are HIV infected have been shown to have higher mortality than those who are HIV negative when they develop TB. Palme et al.⁴⁸ found that Ethiopian children who were HIV positive had a 6-fold increased risk of death during an episode of TB compared with an HIV negative child. In a study in Cote d'Ivoire of childhood TB, Mukadi et al.⁴⁶ noted a 50% mortality over the 6 months treatment period for HIV co-infected children who had CD4 percentage of <10%. In HIV negative children the mortality rate was 4%.

Drug resistant TB is of great importance worldwide as it reflects on TB control in the population, but also makes therapy more expensive and more difficult to implement. When it occurs in children, it will usually be primary resistance, as children are less likely to have been treated previously. In most children, definite confirmation of drug resistant TB will not be possible, due to lower rates of microbiological identification from paediatric samples. Often the diagnosis of drug resistant TB will be made on the basis of confirmation of drug resistance in the index case. In high prevalence communities, however, children may have multiple exposures within the same household, so even if close contacts have drug sensitive disease the possibility of resistant disease in the child still exists.

At present there is little evidence to suggest that drug resistant TB is more infectious or more likely to cause disease than drug sensitive TB.⁴⁹ In general, the incidence and types of resistant organisms encountered in children will reflect the organisms circulating in the community. A large South African study, conducted between 1994 and 1998 in TB culture-positive children, found 5.6% had isoniazid (INH) resistance and 1% with multi-drug resistance (MDR—defined as resistance to both INH and rifampicin).⁵⁰ These results were essentially the same as surveillance data from adults with TB during the same time period, of 3.9% INH resistance and 1.1% MDR. In the UK the rate of INH resistance in paediatric cases is 6.5%, and that for MDR TB is 0.5%, again similar to the rates in adults of 6.4 and 1.2%, respectively.¹⁴