Interventions against Tuberculosis
TB can be controlled by preventing infection, by stopping progression from infection to active disease, and by treating active disease. The principal intervention is the DOTS strategy and its variations, centered on the diagnosis and treatment of the most severe and most infectious (smear-positive) forms of TB but including treatment for smear-negative and extrapulmonary cases as well. Anti-TB drugs can also be used to treat latent M. tuberculosis infection and active TB in patients with HIV coinfection, and the widely used bacillus Calmette-Guerin (BCG) vaccine prevents (mainly) severe forms of TB in childhood. These biomedical interventions directed specifically against TB can be implemented in a variety of ways through medical services and public action and can be supported by other efforts to reduce environmental risk factors (online annex 1).
Vaccination
Currently, the only means of immunizing against TB is with the live attenuated vaccine BCG, although other vaccines are under development (Fruth and Young 2004; Goonetilleke and others 2003; Horwitz and others 2000; Letvin, Bloom, and Hoffman 2001; Reed and others 2003; Young and Stewart 2002). Randomized controlled trials and case-control studies have shown consistently high protective efficacy of BCG against serious forms of disease in children (73 percent [95 percent confidence limits 67-79 percent] for meningitis and 77 percent [95 percent confidence limits 58-87 percent] for miliary TB) but highly variable—and often very low—efficacy against pulmonary TB in adults (Bourdin Trunz, Fine, and Dye, forthcoming; Fine 2001; Rieder 2003). Thus, even with the high coverage now achieved, BCG is unlikely to have any substantial effect on transmission. In parts of Europe and North America that did and did not use BCG, TB declined at rates that were not measurably different (Styblo 1991). In areas of high incidence, BCG vaccination is recommended for children at birth or at first contact with health services. Vaccination is being discontinued in many low-incidence countries because the risk of infection is low and because the response to BCG confounds the interpretation of tuberculin skin tests used to track persons infected during occasional outbreaks. BCG may have substantial nonspecific effects on child mortality—that is, in reducing deaths from causes other than TB—but this possibility is still controversial (Kristensen, Aaby, and Jensen 2000).
Reported BCG vaccination coverage has increased throughout the world during the past 25 years, reaching about 100 million infants, or 86 percent of all infants, in 2002. An estimated 92 percent of children were vaccinated in Europe and 62 percent in Africa in 2002 (WHO 2001). During the past 15 years, coverage has generally been most variable among African countries and least variable in Europe and the Americas. The most complete analysis of the effect of BCG vaccination suggests that BCG given to children born in 2002 prevents about 29,700 cases of childhood meningitis and 11,500 cases of miliary TB during the first five years of life, or one case for every 3,400 and 9,300 vaccinations, respectively (Bourdin Trunz, Fine, and Dye, forthcoming).
Treatment of Latent Infection
Individuals at high risk of TB who have a positive tuberculin skin test but not active disease (for example, associates of active cases, especially children and immigrants to low-incidence countries) can be offered treatment for latent TB infection (TLTI), most commonly with the relatively safe and inexpensive drug isoniazid. Studies among those who have contacts with active cases have demonstrated that 12 months of daily isoniazid gives 30 to 100 percent protection against the development of active TB (Cohn and El-Sadr 2000; Comstock 2000). For patients who may be carrying a strain resistant to isoniazid, rifampicin daily for 4 months is an acceptable alternative (or rifabutin, if used with protease inhibitors for HIV-infected people; Cohn 2003; Menzies and others 2004). Nevertheless, TLTI is not widely used. The main reason is that compliance with long-term daily treatment tends to be poor among healthy people—a relatively high risk of TB among those who are latently infected is usually still a low risk in absolute terms. An additional reason is that the tuberculin skin test tends to be less specific when applied to individuals who have been vaccinated with BCG. Although it is sometimes possible to make separate estimates of the number of individuals in a population who have been infected and who have received BCG (Neuenschwander and others 2002), distinguishing the responses to BCG and infection is harder in any given individual.
The exceptionally high risk of TB among persons coinfected with M. tuberculosis and HIV is a reason for encouraging wider use of TLTI, especially in Africa. However, there are significant barriers to making TLTI effective for coinfected individuals living in areas of high transmission (in addition to those listed earlier). Although trials of TLTI with individuals infected with HIV whose tuberculin skin test was positive have averaged about 60 percent protection for up to three years (with a good deal of variability), the effects have been lost soon afterward, and little or no effect has been seen on mortality (Bucher and others 1999; Johnson and others 2001; Mwinga and others 1998; Quigley and others 2001; Whalen and others 1997; Wilkinson, Squire, and Garner 1998). In addition, identifying M. tuberculosis infection is more difficult in HIV-positive individuals than in those who are HIV-negative because the former are often anergic and are, therefore, unresponsive to tuberculin. Early studies have also experienced problems with uptake and compliance. In a pilot project in Zambia, for example, only 35 percent of HIV-infected individuals identified through HIV testing and counseling services actually started TLTI, and, of those who started, only 23 percent completed at least six months of treatment (Terris-Prestholt and Kumaranayake 2003).
TLTI has been used as a component of intensive, local control campaigns, such as those carried out for North American and Greenland Eskimos, but probably had effects secondary to the prompt treatment of active disease (Comstock, Baum, and Snider 1979; Styblo 1991). At present, TLTI plays no more than an accessory role in TB control in any setting, although the number of recipients around the world has been neither directly quantified nor indirectly estimated.
Treatment of Active Disease: The DOTS Strategy
The cornerstone of TB control is the prompt treatment of active cases with SCC using first-line drugs, administered through the DOTS strategy (WHO 2002a) within targets framed by the MDGs. The DOTS strategy has five elements:
-
political commitment
-
diagnosis primarily by sputum-smear microscopy among patients attending health facilities
-
SCC with effective case management (including direct observation of treatment)
-
a regular drug supply
-
systematic monitoring to evaluate the outcomes of every patient started on treatment.
Standard SCC can cure more than 90 percent of new, drug-susceptible TB cases, and high cure rates are a prerequisite for expanding case finding. Although the DOTS strategy aims primarily to provide free treatment for smear-positive patients, most DOTS programs also treat smear-negative patients, usually without a fee. DOTS can be used as the basis for more complex TB control strategies where rates of drug resistance or HIV infection are high.
Mathematical modeling and practical experience suggest that the incidence of TB will decline at 5 to 10 percent per year when 70 percent of infectious cases are detected through passive case finding and 85 percent of these cases are cured, even though that level represents a treatment success rate among all infectious cases of only 60 percent (Dye 2000; Dye and others 1998). In principle, TB incidence could be forced down more quickly, by as much as 30 percent per year, if new cases could be found soon enough to eliminate transmission. In general, the decline will be faster where a larger fraction of cases arises from recent infection (that is, in areas where transmission rates have recently been high) and slower where there is a large backlog of asymptomatic infection. As TB transmission and incidence go down, a higher proportion of cases comes from the reactivation of latent infection and the rate of decline in incidence slows. These facts explain why it should be easier to control epidemic than endemic disease: during an outbreak in an area that previously had little TB, the reservoir of latent infection will be small, and most new cases will come from recent infection.
In the control of endemic TB, largely by chemotherapy, the best results have been achieved in communities of Alaskan, Canadian, and Greenland Eskimos, where incidence was reduced at 13 to 18 percent per year from the early 1950s (Styblo 1991). Over a much wider area in Western Europe, TB declined at 7 to 10 percent per year after drugs became widely available during the 1950s, although incidence was already falling at 4 to 5 percent per year before chemotherapy (Styblo 1991). More recently, between 1994 and 2000, the incidence of pulmonary TB among Moroccan children 0 to 4 years of age fell at more than 10 percent per year, suggesting that the risk of infection was falling at least as quickly (S. Ottmani, personal communication 2005). The overall reduction in pulmonary TB was only 4 percent per year, in part because of the large reservoir of infection in adults. DOTS was launched in Peru in 1991, and high rates of case detection and cure appear to have pushed down the incidence rate of pulmonary TB by 6 percent per year (Suarez and others 2001). For epidemic TB, as a result of aggressive intervention following an outbreak in New York City, the number of MDR-TB cases fell at a rate of more than 40 percent per year (Frieden and others 1995).
Although the long-term aim of TB control is to eliminate all new cases, cutting prevalence and death rates is arguably more important. About 86 percent of the burden of TB, as measured in terms of disability-adjusted life years (DALYs) lost, is attributable to premature death rather than illness, and prevalence and mortality can be reduced faster than incidence in chemotherapy programs. Thus, the TB death rate among Alaskan Eskimos dropped at an average of 30 percent per year in the period 1950-70 and at an average of 12 percent per year throughout the Netherlands from 1950 to 1990. Indirect assessments of the effect of DOTS suggest that 70 percent of the TB deaths expected in the absence of DOTS were averted in Peru between 1991 and 2000, and more than half the TB deaths expected in the absence of DOTS are prevented each year in DOTS provinces of China (Dye and others 2000; Suarez and others 2001). There have been few direct measures of the reduction in TB prevalence over time, but surveys done in China in 1990 and 2000 showed a 32 percent (95 percent confidence limits 9-51 percent) reduction in the prevalence rate of all forms of TB in DOTS areas, as compared with the change in the prevalence rate in other parts of the country (China Tuberculosis Control Collaboration 2004; PRC Ministry of Health 2000). These findings imply that the targets of halving prevalence and death rates between 1990 and 2015 are technically feasible, at least in countries that are not burdened by high rates of HIV infection or drug resistance.
Many of the 182 national DOTS programs in existence by the end of 2003 have shown that they can achieve high cure rates: the average treatment success rate was 82 percent (that is, the percentage that were sputum-smear negative at the end of treatment plus the percentage that had completed treatment but for whom cure was not confirmed by sputum smear), not far below the 85 percent international target (WHO 2005). The outstanding deviations below that average were in Africa (73 percent) and some former Soviet republics (for example, 67 percent in Russia). Although the completion of treatment was almost a guarantee of cure before the spread of HIV and drug resistance, "completed" is an unsatisfactory way to report the outcome of treatment if cure is in doubt.
Although most TB patients probably receive some form of treatment, only 45 percent of all estimated new smear-positive cases were reported by DOTS programs to WHO in 2003. The case-detection rate in DOTS programs has been accelerating globally since 2000, but the annual increment must be still greater if the 70 percent target is to be reached by the end of 2005. Observations on the way DOTS is presently implemented suggest that a ceiling on case detection might be reached at about 50 to 60 percent (Dye and others 2003; WHO 2005). This fraction is about the same as the percentage of all cases reported annually to WHO from all sources (that is, from DOTS and non-DOTS programs). The problem is that, as DOTS programs have expanded geographically, they have not yet reached far beyond existing public health reporting systems.
