Reviews

Tuberculosis: story of medical failure

Western medicine's conceit is such that most doctors remember streptomycin only for giving birth to randomised controlled trials. They have a far clearer recall of the wider intellectual revolution it spawned than of the disease originally targeted.

The streptomycin trial was the first of many conducted by the British Medical Research Council (MRC) to engineer effective chemotherapy for tuberculosis. Failure to implement the resultant standardised regimens has led to an international resurgence of tuberculosis. Despite the effective treatment developed nearly 50 years ago, tuberculosis now kills more people than ever.

This should trigger some questions from the public, which funds all medical research. For example, why, after an estimated one million controlled trials on every treatment under the sun, have we failed to apply the results of the initial ones on tuberculosis chemotherapy? Why do so many countries still lack basic drugs and microscopes, and why is the principal curative distillation of all this effort still available to only one in six patients with tuberculosis throughout the world?

Tuberculosis certainly had a higher priority in 1948. The MRC owes its existence to public clamour over the disease. Tuberculosis was also responsible for Austin Bradford Hill's career switch from medicine to statistics.

We had the chance to control this disease and we blew it

The success of the streptomycin trial spawned a string of others by the MRC's tuberculosis research unit. The unit's overseas collaborations showed how rigorous science could be specifically modelled to meet the needs of individual developing countries in order to provide affordable and effective treatment. Its work remains a paragon which has yet to emulated.

A trial in India showed that people in the worst possible conditions could be cured just as well at home as in hospitals. At a stroke the entire purpose of sanatoriums disappeared, wiping billions of dollars off European and North American healthcare budgets. The Swiss resort of Davos reinvented itself as a meeting place for the finance ministers of the world's richest nations.

But in the colonial expansion of the last century European emigrants were primary exporters of tubercle bacilli. Tuberculosis killed more native Americans than the Seventh Cavalry. In return for gold and diamonds, native mine workers in Southern Africa were exposed to bacilli coughed up, if not by Cecil Rhodes then certainly by many of his friends. The MRC's extensive studies in East Africa formed the basis of later work in Tanzania which ultimately provided the current DOTS (directly observed treatment, short course) strategy.

Mainstream Western medicine seized on the idea of randomised trials while simultaneously expunging tuberculosis from its consciousness. It disappeared from medical school curriculums as quickly as it did from the research programmes of pharmaceutical companies and the political agenda.

By 1986 the MRC unit had been disbanded by Mrs Thatcher's government, a result of what some saw as her disdain for altruistic foreign aid. Such indifference was reflected at the World Health Organisation. Tuberculosis scarcely figured in its publications during the late 1980s and by the end of the decade its entire tuberculosis monitoring and control operation had been reduced to one person.

This complacency was ruthlessly exposed by the advent of HIV, which allowed dormant tuberculosis to reactivate. A string of extraordinary events followed in the 1990s when the affluent world woke up to the fact that it was still at risk of a disease which does not recognise any geographical, social, or racial boundaries.

New York City had to look to Tanzania to find out how to control its own explosion in tuberculosis. In 1993, a reinvigorated WHO took the unprecedented step of declaring tuberculosis a global emergency and has since made intensive efforts, backed by the World Bank, to tackle the epidemic.

We had the chance to control this disease and we blew it. The easy option is to blame politicians and the media and revert to standard pessimism that nothing can be done in the face of the complexities and cost of curing people of tuberculosis and the ever rising economic gulf between rich and poor nations.

But this should not be used as a fig leaf for medicine's failings. Ironically, the first half of this century saw unprecedented press and political interest in tuberculosis when all medicine came up with was quackery. When medicine was able to offer the prospect of cure honed on the anvil of science, the silence from the rest of us was deafening.

The MRC studies highlighted from the outset the dangers of emerging drug resistance and devoted considerable energy to developing treatments to minimise poor patient compliance. The failure of doctors to follow established protocols and proved regimens is well documented in both rich and poor countries.

Trials undoubtedly have a place in shaping medical practice, but patients do not get better on confidence intervals. Evidence based medicine can never work if a significant minority of doctors deem it inadmissible at the outset. Mainstream medicine in affluent countries chose to weave its entire epistemological framework around the concept of controlled trials. Doctors need them to further their careers and fill an estimated 22 000 biomedical journals around the world. In the process they have spawned a monster of uncontrolled research which is impossible to digest and apply in practice.

Franz Kafka accused medicine of "hunting a beast through endless forests" CREDIT: YACOV ECKEL

Something has got lost along the way in the quest for the holy grail of therapeutic paradigm: a wider sense of purpose and reality. Franz Kafka may have predicted it before succumbing to tuberculosis when he equated medicine blindly chasing cure as though "hunting a beast through endless forests."

Maybe he was right. It is something the leaders of the world's richest nations might chew over when they next pig out at Davos. For a disease caused by inspiration, there is plenty of room for inspirational leadership. But don't hold your breath.

Evaluating new treatments    

Fifty years ago medicine entered a new phase in evaluating new medical treatments in the form of randomised clinical trials, with control patients often given placebos. At the same time there was a widely accepted view that retrospective controls were "inherently fallacious." But 20 years ago, this argument was challenged in the BMJ, and retrospective controls were advanced as an alternative to randomisation.

It was recognised that the advent of modern computer technology made more practical the use of retrospective controls by facilitating the storage and retrieval of detailed medical data on past cases, and in comparing past with current medical experience. The BMJ endorsed this view. But the intervening 20 years have seen little change in practice. Today, randomised clinical trials with placebos are, with few exceptions, the prevailing method of evaluating new treatments. Yet alternatives surely warrant more attention than they have receivedin particular, what is now usefully called a "computrial."

Alternatives surely warrant more attention than they have received

A computrial is one which admits all suitable patients to the same experimental treatment, while controls are drawn from the records of patients who have received treatment in historically recognised, leading treatment centres, not just random treatments. Controls are matched individually by computer with appropriate software to the trial patients for all significant prognostic factors. A computrial generates not just statistical summary data on the collective response to the new treatment. It creates a record which can be cross analysed for the effects of particular prognostic factors, with implications for individual patient treatment, and it compares a new treatment with the best of the old. It eliminates placebos and the deterring uncertainty that faces every patient in a randomised trial as to whether he or she is receiving the newest treatment or is being relegated to the passive role of a control.

The computrial concept claims respectable origins. In 1965 the late Sir Austin Bradford Hill, in his famous Heberden Oration, Reflections on the Controlled Trial, expressed reservations about the indiscriminate use of randomised trials, and discussed in favourable terms the use of retrospective controls based on patient histories instead. On 19 November 1979 the BMJ published my paper, repeatedly rejected since 1972, "Do retrospective controls make clinical trials `inherently fallacious'?" which challenged the then accepted view.

Debate started with a concurrent editorial, "Randomised controlled trials?" that endorsed my views, citing the power of computers to match trial patients with historical controls, and concluded with these words: "The controlled trial has been placed on too high a pedestal and needs to be brought back to earth." Some readers wrote in support; others were opposed. The debate ended six months later with my rejoinder to critics.

With such a send off from the editors of BMJ, computer enabled clinical trials with retrospective controls should soon have had their day, at least on an experimental basis, particularly with the explosive growth in numbers and power of computers. This has not been the case, however, despite many strong criticisms of randomised trials over the years.

Heartened by the response to my 1979 publication, I have repeatedly encouraged use of what I now call computrials, but the effort has been unproductive even though the original concept has been expanded to include use of computers to particularise treatment to individual patients rather than provide statistical results only, and emphasis is placed on controls that have received the best treatment medicine had to offer before the new treatment appeared. The most recent experience with one of the world's leading pharmaceutical companies is worth recording because it has lessons for medical innovators generally. Its essential result was a refusal to engage in discussions among colleagues, and a summary rejection of the written, fully documented proposal, with no reasons given.

The ethical and management issues raised in the encounter just described with the technical staff of a world famous pharmaceutical company may be as important as the ethical issues involved in randomised clinical trials themselves. If there is no dialogue among colleagues on matters that affect the evaluation and use of a company's innovative products how can you assume that issues raised have been dealt with in good faith, and that serious concerns of cost, efficacy, and ethical propriety are not being buried to conceal excessive costs and ethical shortcomings of current practices, or that official favour is being curried at the expense of scientific advance?

A colleague takes the view that only legislation that relieves the United States Food and Drug Administration of responsibility for evaluating the efficacy of drugs will improve the present situation. My own view is that the FDA is a responsible agency, and that it will in due course accept a less theoretically rigid and more empirical policy towards evaluating new drugs and other treatments if the medical and scientific community says it should.

Chris Holme. 

Herald newspaper, Glasgow

Lawrence Cranberg. 

consulting physicist, Austin, Texas, USA

James Owen Drife. 

professor of obstetrics and gynaecology, Leeds

Acknowledgments

I would like to thank Professor Emil J Freireich for his useful comments.