Professor Georges Canetti was, I think, the first person to write at length about the central position of the infector pool in the continuing epidemic of tuberculosis, and it may have been he who coined the term (1). He defined the infector pool of tuberculosis as constituted by the mass of tuberculous people with positive sputum who exist at any given moment in the world. Canetti’s report is the first reference in Spencer’s paper in which he sets out a proposal for an epidemiological instrument, designed to assist health professionals in understanding the dynamic of the epidemic of tuberculosis (2). Spencer, as a result of his extensive field work in Soweto, placed the infector pool in the centre of his diagrammatic representation of the epidemic, and showed how the strategy for tuberculosis control should be modified in the light of knowledge of the size of the infector pool. It was Spencer’s work which was the basis of my early work on the eradication of tuberculosis in Rhodesia in the 1960s (3, 4).
History teaches that pulmonary tuberculosis is essentially a relapsing disease. In epidemiological terms this indicates that a proportion of treated cases will be subject to recurrent disease and will re-enter the infector pool. Whether recurrence follows re-activation of residual disease or is due to super-infection with a distinct strain of M. tuberculosis is not important, but the speedy recognition of recurrent disease is vital. In a phase of successful control and declining incidence there may not be much recurrent disease, but when the epidemic is out of control – as in South Africa – and the public health service is overloaded, recurrent disease becomes a major factor in continuing transmission (5). It is among those with recurrent disease that drug resistance is found at first, and only when the epidemic is well and truly rampant, or complicated by high rates of co-infection with HIV, is initial infection of new cases with MDR or XDR organisms a significant problem.
In the light of a recent paper (6) discussing the failure of tuberculosis control in South Africa it is instructive to revisit a state of the art review from the USA written in the early 1970s (7). For those really intent upon mastering communicable disease problems in South Africa it might be instructive to read this alongside Canetti’s 1962 report. The scene in the USA is set using mortality figures – 78,880 deaths and 96,500 new active cases in 1932, and 4,000 deaths and 33,500 new active cases in 1972. The new case rate in 1932 was 76.7 per 100,000 compared with 16.1 in 1972. Like Canetti they address the character and the size of the infector pool. For example, ‘Forty-seven per cent of the new active cases in 1972 came from the 153 cities containing 100,000 or more citizens. Thirty-nine per cent of new active cases in 1972 came from the 59 cities with populations of 250,000 or more’. Data from the City of Cape Town, which is the subject of the paper by Wood and his colleagues, describes a massive urban reservoir currently reporting twice as many new cases annually from its population of 3 million as the USA does from a population of 300 million. Johnston and Wildrick (7) report the decline in the prevalence of positive skin sensitivity in some detail – thus in 1973 7% of the USA population was tuberculin positive and more than two thirds of those were more than 35 years of age. Of direct relevance to the consideration of the size and extent of the infector pool is their statement that ‘300,000 reactors died each year, while approximately 50,000 people acquired new infection’. This signals a very rapid (exponential) decline in the infector pool, over the previous several decades no doubt. In 1971 0.2 per cent of children in the first year of primary schooling and 0.7 per cent of youngsters in the first years of secondary school were tuberculin positive. The annual infection rate was estimated to be 3 new tuberculin conversions per 5-10,000 persons per annum.
Wood et al (6) gathered a significant new set of data on the tuberculin status of children in Cape Town. The picture is bleak and reflects a massive increase in the infector pool. This long quotation is lifted without apology from the paper by Wood et al because it puts the position regarding an unfamiliar subject so clearly – ungarbled by me! “In 2005, the prevalence of a positive (>10 mm induration) tuberculin skin test (TST) among 7,457 primary school children (median age 8.6 years) was 37.4%. A TST survey in a Cape Town school in 2005 reported [the tuberculin positive] prevalence to be 26.2% in 5-8 year olds, increasing to 52.5% in 14-17 year olds. The prevalence of tuberculosis infection increased from 20% at school entry to 52% at 15 years, and reached 75% at 25 years in another study including HIV-negative adolescents and adults. Note also that a further 8% of the population will have developed tuberculous disease by the age of 25 years and they were excluded from the prevalence surveys. In two neighbouring urban communities with a low HIV prevalence, a high prevalence of tuberculous infection in children (6-9 years) was reported. Transmission rates increased between 1998 and 2005 and remained among the highest (4.1-5.8% per annum) in the world”. Set against the position in the USA in 1972 this indicates that there is a hundredfold difference in the rate at which tuberculin conversion takes place in Cape Town now, and the rate in the USA about 40 years ago.
Whether we talk explicitly of the infector pool, or refer to the tuberculin conversion rate, the incidence of new or recurrent case, the force of the epidemic or the reproductive rate of the disease, or the prevalence of a particular feature of tuberculosis (slide positivity, culture positivity, radiological extent of disease, drug resistance or clustering) we are in fact stating something about the nature and/or the extent of the driving force behind the epidemic. When we first began to describe the infector pool in the Midlands Province of Zimbabwe in the early 1960s we had, inevitably, to make some informed guesses (3, 4). We thought that there might be three main divisions of the “active” infector pool, and for convenience assumed they were of equal size – a third contributing this year’s crop (of new or recurrent cases) largely as a result of felt need; a third in waiting to contribute next years crop (probably already culture positive but not yet slide positive) and amenable to active case finding or co-operative if asked; and a third which we described as inaccessible for any one of a number of reasons.
We were also aware that the “potential” or “expanded infector pool might be much larger and be cumulative over quite a long period of time – say 20 years. Johnston and Wildrick (7) made the point clear when they described the origin of newly reported cases in the USA in the 1970s. Only 8% of cases arose directly from new infections. The remaining new cases originated among persons who had been infected in the remote past. “Of this 92% it was estimated that 58% of the cases developed in persons who previously had normal [chest x-rays] and about 34% occurred in persons with pre-existing abnormal [x-rays]. Of those with abnormal [x-rays] approximately 19% had a previous diagnosis of tuberculosis and received treatment in most instances, while 15% had not had a previous diagnosis of tuberculosis and had not been treated”. What does this say about post-treatment surveillance? It certainly contributes new insights into the importance of what we christened back-registration of previously treated patients who formed an important part of the infector pool, and increased our sensitivity to re-entry into the infector pool of known (treated) cases. Our enthusiasm for life-long surveillance was reinforced. Recent DNA finger-printing studies have described the clustering of new cases on mines around an incompletely treated or recurrent case, and estimated that 50% of continuing transmission takes place through this mechanism. (5)
After the widespread introduction of directly observed short course chemotherapy it has become increasingly clear that several “politically correct” elements in the strategy were in fact serious mistakes: regardless of the extent of lung destruction at presentation six months treatment was the rule (one size fits all unlike shoes); rifampicin containing chemotherapeutic regimes eradicate mycobacteria from the lung; old fashioned adjuncts to chemotherapy – cough mixture, vitamin and mineral supplements; careful monitoring of weight for height gain – were wasteful expenditure; slide positivity as the criterion for treatment/passive case finding (for co-infected patients it certainly is not); post-treatment surveillance is not required; the virtual cessation of active case finding. Return to go, do not collect $200, and do not control tuberculosis.
Implicit in the concept of the centrally located infector pool is a feedback or re-entry loop, taking the patient with recurrent disease back into the pool. This group if numerous enough will keep the tuberculosis service going at full stretch indefinitely, and may lead to a massive expansion of the infector pool. The nature and extent of the infector pool underlies the failure of tuberculosis control described so graphically by Wood et al.
In the Western Cape, and elsewhere in South Africa, the infector pool is growing rapidly because neither the epidemic of tuberculosis nor the fundamentals of control are clearly understood. As a result of a cluster of facile assumptions (about tuberculosis, short course chemotherapy and HIV/AIDS) the infector pool is now very large, and growing. The central issue in the strategy to control tuberculosis in South Africa is an understanding of the nature and extent of the infector pool and the dynamics of its maintenance and expansion. The combination of the unrestricted spread of HIV and an ineffective unbalanced tuberculosis control strategy is deadly.
In order to control tuberculosis we must work towards cogent argument and evidence based strategy design – and must insist on quality control of bright new ideas. Tuberculosis registers should be regularly audited and provision made for dis-aggregation to define high-risk groups by occupation, place of residence, age, gender and so on. Simple card-based registers suitable for analysis by the staff of front-line units would help. Cherished notions asserted aggressively should be challenged – they are sometimes wrong.