Breaking a law: tuberculosis disobeys Styblo’s rule Christopher Dye a In 1985, Karel Styblo derived one of of progressing to disease depends on transmission rate. This seems unlikely, the guiding rules of tuberculosis (TB) various personal characteristics, such however, because drug treatment is epidemiology. Bringing together data as age, sex, genes, physiology, behav- expected to reduce the average duration from 16 countries, he proposed that an iour, and whether or not the exposure of infectivity, not to increase it. annual incidence of 50 sputum-smear- is a new infection or reinfection.3 But Adding this new evidence to the positive TB cases in a population of if this risk of 10% is roughly cor- similar results of an earlier study,4 we 100 000 generates an annual risk of rect and noting, with the postulates clearly cannot assume that Styblo’s infection of 1%.1 This rule of thumb above, that around half of all new rule still holds. How then should we quickly became an established and cases will be smear-positive, then each evaluate TB burden and trends? If cherished part of the epidemiological new smear-positive case will generate measures of the risk of infection cannot canon because it provided a way to 10 × 2 × 0.1 × 0.5 = 1 new smear-positive reliably estimate incidence, they can estimate, albeit indirectly, an important case over one transmission cycle. Un- at least measure geographical variation but elusive quantity (disease incidence) der these conditions, TB is stably en- and trends in transmission. Disease from a comparatively simple measure- demic in its host population. Styblo’s prevalence surveys will also make an ment procedure (risk of infection via tu- rule is therefore not simply a method important contribution;2 despite being berculin surveys). Though the rule has for estimating incidence; it is part of costly and logistically demanding, never formally been viewed as anything the foundations of TB epidemiology. surveys will be carried out in several better than approximate, its theoretical The rule is vulnerable on its key high-burden countries over the next and practical underpinnings have been premises: the two-year duration of 5 years. But the ultimate method of steadily eroded by the last two decades infectiousness and the transmission of assessment is routine surveillance: of epidemiological change. In this issue 10 infections each year. First, drugs are comprehensively counting cases and of the Bulletin, van Leth et al. present now widely available and are expected deaths in all countries.5 yet another analysis suggesting that TB to shorten the duration of infectiv- The legacy of Karel Styblo is im- disobeys Styblo’s rule.2 The rule can ity, though poor treatment practices mense and undiminished by these new no longer be trusted as a method for can generate chronic illness. Second, findings. His bold attempt to distil estimating TB incidence. improved living conditions with simplicity out of apparent complexity To understand why the rule has better aeration and a lower density remains an object lesson. It is a matter of become inapplicable, we need to go of inhabitants are expected to reduce regret, therefore, that his name may ap- back to first principles. Working with the number of infective contacts per pear less frequently in conversation, now the limited data available at the time, unit time. Third, when TB is not in that the eponymous rule is defunct. ■ Styblo deduced that deaths per year, a steady-state, incidence will change incidence per year and prevalence of more slowly than the risk of infection. References smear-positive TB were held in the ratio So as incidence begins to fall, there 1. Styblo K. The relationship between the risk of 1:2:4. That is, in the pre-drug era, a will appear to be, at any point in time, tuberculous infection and the risk of developing smear-positive TB case remained infec- more cases per infective contact. All infectious tuberculosis. Bull Int Union Tuberc tious for an average of 2 years and the three processes would generate ratios Lung Dis 1985;60:117-9. case fatality rate was 50%. These obser- of incidence to risk of infection that 2. van Leth F, Van der Werf MJ, Borgdorff MW. Prevalence of tuberculous infection and incidence vations allowed him to estimate smear- are greater than 50:1. of tuberculosis: a re-assessment of the Styblo positive incidence from prevalence In their paper, van Leth et al. have rule. Bull World Health Organ 2008;86:20-26. and mortality data, and compare these tested just one of the two premises – 3. Vynnycky E, Fine PEM. Life time risks, incubation incidence estimates with the measured that each smear-positive TB case trans- period, and serial interval of tuberculosis. Am risk of infection. The canonical ratio of mits about 10 infections per year. J Epidemiol 2000;152:247-63. doi:10.1093/ smear-positive incidence per 100 000 Using the ratio of smear-positive aje/152.3.247 PMID:10933272 (I) to per cent risk of infection per year prevalence to risk of infection measured 4. Trunz Bourdin B, Fine P, Dye C. Effect of BCG vaccination on childhood tuberculous meningitis (l) then comes from I/l = (I × 10⁵)/ in China, the Republic of Korea and and miliary tuberculosis worldwide: a meta- (2 × 10 × I × 10²) = 50, assuming that the Philippines, they found that the analysis and assessment of cost-effectiveness. each smear-positive case generates an actual number of infections per year lay Lancet 2006;367:1173-80. doi:10.1016/ average of 10 new tuberculin-detectable between 2 and 6. Formally, this result S0140-6736(06)68507-3 PMID:16616560 infections each year. does not disprove Styblo’s rule because 5. Dye C, Bassili A, Bierrenbach AL, Broekmans JF, Conventional wisdom also holds it examines the relation between risk of Chadha VK, Glaziou P, et al. Measuring that the lifetime risk of progressing infection and prevalence, not incidence. tuberculosis burden, trends and the impact of control programmes. Lancet Infect Dis. In press. from infection to active TB is about The 50:1 ratio could be preserved if, 10% on average (5% within 5 years, for example, the duration of illness 5% thereafter). In fact, the chance has increased to offset the reduction in a Stop TB Department, World Health Organization, 20 avenue Appia, 1211 Geneva 27, Switzerland. Correspondence to Christopher Dye (e-mail: [email protected]). doi:10.2471/BLT07.049510 4 Bulletin of the World Health Organization | January 2008, 86 (1).