Quotes from Doctors & Researchers:

It is often claimed that scientists and doctors are united in their belief in the value of, and necessity for, animal experiments to protect human health. This stream of quotes from scientists and doctors, stretching from as far back as the early 20th Century right up to today, shows that there has been a long tradition of scepticism about this issue. Many of the quotes are from scientists who support or conduct animal research; far from diminishing the impact of their words, this fact ought to give them extra weight.

Note: FDA stands for Food and Drug Administration, the US drug regulatory body.

Cancer | Neurological diseases/ conditions | Teratogenicity and thalidomide | Immunology & AIDS & TGN1412 | Miscellaneous

Cancer

2008

Mouse xenograft models of cancer, understandably, have a terrible reputation. Although researchers and companies routinely use these human tumors in mice for preclinical drug testing, individual models poorly predict how drugs will act in the clinic. Retrospective reviews published by the National Cancer Institute in 2001 and the National Cancer Institute of Canada in 2003 came to the same conclusion: Drugs that work againstcancer in xenograft mice rarely work in people with the same tumor, with the exception of lung and possibly ovarian cancer. "There's this mantra: 'Xenografts don't predict for human effects,'" said Peter Houghton, Ph.D., a cancer researcher at the St. Jude Children's Research Hospital in Memphis, Tennessee. Ken Garber, Journal of the National Cancer Institute, 30^th December.

2007

We have learned well how to treat cancer in mice and rats but we still can’t cure people. Professor Colin Garner, quoted in Accelerator MS Is a Powerful New Tool, Genetic Engineering & Biotechnology News, Vol. 27, No. 15.

2006

We do trials in people because animal models do not predict what will happen in humans. Dr Sally Burtles, Cancer Research UK, Report of the Expert Scientific Group on phase one clinical trials, following the TGN1412 clinical trial disaster.

You really have to design the medicine for the species of interest…You'll find it very rare to find a medicine that will work in both… Patrick M. O'Connor, head of oncology research for Pfizer, quoted in The New York Times, 24 November.

In summary, mouse xenograft models should not be viewed as ideal models for cancer drug development. Altered, nonhuman host stroma, poor predictive value when applied in an empirical sense, and questionable relation to the naturally occurring human disease are but a few features, which temper enthusiasm for their use. Sausville & Burger, Cancer Research, 66, 3351-3354, April 1.

Even when drugs with evidence of anticancer activity in preclinical in vivo models are given at their maximum tolerated doses, they frequently fail to produce useful activity in humans. Sausville & Burger, Cancer Research, 66, 3351-3354, April 1.

In the U.S. NCI retrospective (2001), activity in at least 33% of models of a variety of histologies predicted for clinical activity in some disease. In the NCI of Canada retrospective (2003), generally similar conclusions were reached. It should be cautioned, however, that the drugs used in these studies were for the most part "classic" cytotoxics. Whether "targeted" therapeutics, such as signal transduction inhibitors, antiangiogenic, or stroma-modifying agents, would perform better or worse remains to be defined. Sausville & Burger, Cancer Research, 66, 3351-3354, April 1.

Change is needed. Thirty years of experience with subcutaneous xenografts, human tumors implanted under the skin of the mouse, have satisfied few because so many drugs that cure cancer in these mice fail to help humans. A 2004 analysis in the Journal of the American Medical Association showed that only 3.8% of patients in phase I cancer drug trials between 1991 and 2002 achieved an objective clinical response — and the response rate is declining. Almost all drugs tried in humans work against subcutaneous xenografts in mice. “How many more negative data do you want? It’s very depressing.” said Isaiah Fidler , Ph.D., of the University of Texas M. D. Anderson Cancer Research Center in Houston. Ken Garber, Journal of the National Cancer Institute, Vol. 98, No. 17, September 6.

2005

Given that many of these investigational anticancer drugs eventually fail, the animal models on which clinical trials are predicated must at best be limited in power, and at worst wildly inaccurate. Dr Alexander Kamb, Global Head of the Oncology Disease Area at the Novartis Institutes for Biomedical Research, Nature Reviews Drug Discovery, 4, 161 - 165.

The problem with animal carcinogenicity tests is not their lack of sensitivity for human carcinogens, but rather their lack of human specificity. A positive result has poor predictive value for humans. Knight, Bailey & Balcombe, British Medical Journal USA, Vol. 5, p477.

2004

It’s been well known for maybe two decades that many of these preclinical human cancer models have very little predictive power in terms of how actual human beings – actual human tumours inside patients – will respond…Preclinical models of human cancer, in large part, stink…Hundreds of millions of dollars are being wasted every year by drug companies using these [animal] models…Prof. Robert Weinberg, Massachusetts Institute of Technology, Fortune, 9th March.

[mouse models are] woefully inadequate…if you look at the millions and millions and millions of mice that have been cured, and you compare that to the relative success, or lack thereof, that we've achieved in the treatment of metastatic disease clinically, you realize that there just has to be something wrong with those models. Homer Pearce, research fellow at Eli Lilly. Fortune, 9th March.

2003

In Tamoxifen’s case, a drug first developed as a potential contraceptive languished for many years before its present application was found. Furthermore, its propensity to cause liver tumours in rats, a toxicity problem that thankfully does not carry over into humans, was not detected until after the drug had been on the market for many years. If it had been found in preclinical testing, the drug would almost certainly have been withdrawn from the pipeline. Nature Reviews Drug Discovery 2003; 2:167.

The in vitro cell line model was predictive for non-small cell lung cancer under the disease-oriented approach, for breast and ovarian cancers under the compound-oriented approach, and for all four tumor types together. The mouse allograft model was not predictive. The human xenograft model was not predictive for breast or colon cancers, but was predictive for non-small cell lung and ovarian cancers when panels of xenografts were used. Voskoglou et al. Clinical Cancer Research Vol. 9, 4227-4239.

2001

…some findings in colon cancer mice, which were very good models, actually led to clinical trials in humans which resulted in an increase in cancer. Dr Jeffrey E. Green of the National Cancer Institute’s Laboratory of Cell Regulation and Carcinogenesis. Journal of the National Cancer Institute, 93:976.

For 39 agents with both xenograft data and Phase II clinical trials results available, in vivo activity in a particular histology in a tumour model did not closely correlate with activity in the same human cancer histology, casting doubt on the correspondence of the pre-clinical models to clinical results. Johnson and colleagues, British Journal of Cancer, 84(10):1289-90.

1999

People are very complacent with their animal models. But this begs the question of whether there exists a good model of cancer. Dr Andy Maniotis, The American Journal of Pathology, 155: 739.

1998

My own medical perspective is that animal cancer research should be regarded as the scientific equivalent of gossip – with about the same chance of turning out to be true, i.e. truly effective in humans. Some gossip turns out to be true, but most of it does not…and gossip can cause great anguish for those affected, in this case millions of desperate cancer patients worldwide. G. Timothy Johnson MD, Boston Globe, May 22.

The history of cancer research has been a history of curing cancer in the mouse. We have cured mice of cancer for decades, and it simply didn’t work in humans. Dr Richard Klausner, Director, National Cancer Institute, LA Times, May 6.

God knows we’ve cured mice of all sorts of tumours. But that isn’t medical research. Thomas E Wagner, senior scientist at Ohio University’s Edison Biotechnology Institute, the Columbus Dispatch, March 20.

1997

The fundamental problem in drug discovery for cancer is that the model systems are not predictive at all," says Alan Oliff, executive director for cancer research at Merck Research Laboratories in West Point, Pennsylvania…Researchers blamed the failures on the fact that the drugs were being tested against mouse, not human, tumors… the xenograft tumors don't behave like naturally occurring tumors in humans – they don't spread to other tissues, for example. Thus, drugs tested in the xenografts appeared effective but worked poorly in humans. "We had basically discovered compounds that were good mouse drugs rather than good human drugs," says Sausville (associate director of the division of cancer treatment and diagnosis, NCI). Science, 278; 1041-1042.

One might expect that these animals would mimic human symptoms, not just the genetic mutations. In fact, that is usually the exception, not the rule. Dr Tyler Jacks, regarding genetically modified mice in cancer research. Science, 287: 1041.

1995

Since its inception 25 years ago, EPA [the US Environmental Protection Agency] has applied the same logic to hundreds of other substances, extrapolating from high levels in animal studies to arrive at acceptable levels for humans. But that approach, say scientists both inside and outside the federal government, may no longer be the best way to safeguard public health...EPA's new emphasis on molecular data is based on a growing body of evidence that extrapolations from megadoses can provide a misleading picture of the effects of low-level exposure. Chloroform is a good example. EPA's current strict standards were derived from a study in which mice developed liver tumors after exposure to massive daily doses of chloroform pumped into their stomachs over several months. However, those findings may not be relevant to human exposures, according to a paper picked by the Society of Toxicology as the best published last year in its journal. Richard Stone, Science, vol 268, p 356-357.

1993

1-3 Butadiene, an important industrial chemical and a common environmental air pollutant, has been shown to be a weak carcinogen in the rat, but a potent carcinogen in the B6C3F1 mouse. This species difference makes risk extrapolation to humans difficult, and the underlying mechanism must be clarified before meaningful risk extrapolation to humans can be made. Dr Gunnar Johanson, of the National Institute of Occupational health, Sweden, Alternatives to Laboratory Animals, vol 21, p 173180.

Why the dog was ever considered as an appropriate animal for carcinogenicity testing is also not entirely clear... Despite the obvious problems of study design and interpretation, carcinogenicity tests in the dog, lasting 7 years, were requested by regulatory authorities from the late 1960s...One of the best known examples of the inappropriate use of the dog was the carcinogenicity testing of hormonal contraceptives. It is now understood that mammogenesis in the dog is very different from that in primates; quantitative and qualitative differences exist in the feedback control mechanisms, receptor content and behaviour, and target sensitivity and responsivity. As a result of this biological difference there was a high incidence of mammary tumours in long-term studies in dogs treated with progestagens/contraceptive steroids such as lynestrol. Ultimately pressure from the scientific community led, relatively recently, to the requirement for carcinogenicity studies in dogs being dropped. Parkinson and Grasso, Human and Experimental Toxicology, vol 12, p 99-109.

1992

In the course of tumour progression, it has been known for many years that mice and men are totally different. Nature, Nov 26.

The following quotes are all from Philip H Abelson, Science, vol 255, p 141.

- It has been conventional practice to test potential carcinogens using highly inbred strains of rodents. The rationale was the supposed superior reproducibility of results compared with those obtained from wild-type animals. However, that assumption can be questioned. At least three examples of genetic drift of inbred strains can be cited... Lifetime expectation [in an inbred strain of mice] of developing one or another form of neoplasm ['spontaneous tumour'] had risen from 10 to 80 percent.

- The use of inbred strains as test animals can be further questioned on the basis that they often develop spontaneous tumors in organs where cancers are not frequent in humans. For example, incidences of mouse liver tumors in 2-yearold B6C3F1 mice has ranged from 17.8 to 46.9 percent. In contrast, the death rate from liver cancer in the United States is about 0.005 percent.

- Results of the animal studies raise questions about the validity of federal regulations that are based on ad lib-fed inbred strains of rodents. Are humans to be regarded as behaving biochemically like huge, obese, inbred cancer-prone rodents?

- The estimate made by NIOSH was 597 excess cancers per 10,000 workers having that same exposure... Instead of extra cancers predicted by NIOSH, workers had fewer cancers than expected...With trillions of dollars, loss of competitiveness, and jobs at stake, a searching review of the risk assessment methodology of the regulatory agencies is overdue.

[Regarding tamoxifen, an anti-cancer drug] "Experimentally, tamoxifen has carcinogenic potential. In some strains of rat, but not mouse or hamster, tamoxifen can cause liver cancers at doses as low as 5mg/kg per day... However, there are doubts about the correlation of [the results] with the risk of malignant disease even in rats, let alone in other rodents, mammals, or human beings. There are many uncertainties in extrapolating these experimental data from rats to women. The effect depends on bioavailability, hepatic [liver] blood flow, and hepatic [liver] metabolism to active genotoxic carcinogens, all of which differ enormously between rat and man. Dr Trevor Powles, The Lancet, vol 340, p 1145-1147.

1991

For example, the control incidence of mouse liver tumour varies between 0 and 58% in 41 NTP [National Toxicology Program] bioassays where these tumours are induced by a test chemical. Ashby and Morrod, Nature, vol 352, p 185-186.

Yes, I think it is very clear to all of us who are engaged in the business of assessing toxicity data that, when volumes of data are proudly presented to us after a carcinogenicity study, showing that there was a tumour in this organ or that, we look at it and we scratch our heads, and we wonder what on earth we can make of it. This is especially true when huge doses are given, with nothing to suggest what would be expected at low doses. I think very often the carcinogenicity studies are a waste of everybody's time and a fearful waste of animals. They are conducted partly because we are not sure what to do instead, and partly because they are a political gesture and a very miserable one at that. Professor Andre McLean, speaking at a conference reported in Animals and Alternatives in Toxicology, p86, ed. Balls, Bridges and Southee (publ. Macmillan).

The [pharmaceutical] industry is left with an expensive and time consuming test which uses large numbers of animals and whose very basis is questioned by scientists. Professor D Davies, quoted in the pharmaceutical magazine Scrip, 2nd October, p 23.

[Animal carcinogenicity tests on new drugs are] inaccurate, often insensitive and generally misleading. Dr John Griffin, Director of the Association of British Pharmaceutical Industry, quoted in the pharmaceutical magazine Scrip, 2nd October, p 23.

It seems sometimes that almost everything we eat, drink or take can cause cancer in rats. That does not necessarily relate to tumours in humans. Tony Watson, President of the British Association of Plastic Surgeons, quoted in The Times, 9^th May.

1990

Our risk models are based on at least 50 assumptions, none of which has been scientifically demonstrated. For example, we assume that there is no difference between continuous (as in animal tests) or intermittent (as in human experience) dosages. But that ignores our growing knowledge of the way in which DNA repairs the human system . . .We feed rodents `all-you-can-eat' buffets every day, yet we know that caloric intake is the single greatest contributing cause of cancer [in rodents]. In fact, we found you can modify the cancer causing impact of one of the most potent carcinogens from 90% down to less than 3%, just by cutting caloric intake 20%. Dr. Ronald Hart, Director of the Center for Toxicological Research in Arkansas, quoted in Business Review Weekly, 27th April.

… Predictions of carcinogenicity from laboratory animals are without meaning for there is no evidence that the studies were conducted in a way that took into consideration the pharmacodynamics in the species investigated, or with any appreciation of end organ sensitivity (with respect to contraceptive steroids). Ralph Heywood, Chapter 7: Clinical Toxicity- could it have been predicted? Post-marketing experience, Animal Toxicity Studies: Their Relevance for Man (publ. Quay).

Risk assessment policy that relies solely on screening bio-assay results from the most sensitive species is not based on scientific principles. Neither is it credible or reliable. Dr Vernon Houk, director of environmental health at the American Center for Diseases Control, addressing a conference in the USA, 1989, and quoted in Business Review Weekly, 27th April.

The principal method of determining potential carcinogenicity of substances is based on studies of daily administration of huge doses of chemicals to inbred rodents for a lifetime. Then by questionable models, which include large safety factors, the results are extrapolated to effects of minuscule doses in humans... The rodent MTD test that labels plant chemicals as cancer-causing in humans is misleading. The test is likewise of limited value for synthetic chemicals. The standard carcinogen tests that use rodents are an obsolescent relic of the ignorance of past decades. Philip H Abelson, Science, vol 249, p 1357.

... numerous chemicals have been found to have potential toxicity/carcinogenicity in rat or mouse, which, we are reasonably certain, have little or no potential hazard in man. The reason for this is again species differences, for the biological defence mechanism which protects against toxic chemicals is most highly evolved in man, who therefore generally has a higher resistance to chemical toxicity and carcinogenicity than have rodents and other species… Parke, Ioannides and Lewis, Alternatives To Laboratory Animals, vol 18, p 91-102.

1989

Elsewhere, I have pointed out that overfeeding of rats profoundly influences the incidence particularly of endocrine tumours. Furthermore, endocrine tumours and tumours of tissues, such as the breast and uterus, which are very directly under sex-hormone control, constitute a very high proportion of the tumours observed in most carcinogenicity studies in rats. [Comparing] the incidences of tumours in these categories in men and women with those in male and female rats... [shows] the differences between the two species are sufficiently striking to make one wonder how appropriate the laboratory rat is as a model for man in terms of the spectra of tumours to which they are prone. Dr Francis Roe, Advances in Applied Toxicology, p 10, ed. A D Dayan & A J Paine (publ. Taylor & Francis).

1988

There are marked differences in carcinogenicity across sexes, strains and species. Often, the-same chemical will cause one kind of cancer in one experiment and another kind in another experiment Indeed, the most hard-bitten advocates of animal experiments do not claim to be able to predict which organ will be affected in humans by a chemical that is carcinogenic in animals. Freedman and Zeisel, Statistical Science, vol 3, p 3-28.

[Regarding animal carcinogenicity tests on saccharin] Published risk estimates, starting from the same animal data but using various [statistical] models, differ by factors of over 5,000,000. Freedman and Zeisel, Statistical Science, vol 3, p 3-28.

[Regarding the difficulties of applying animal results to humans] "There turn out to be many different ways to measure this difference [in size]. For example, a man weighs 2800 times as much as a mouse, eats 300 times as much per day and lives 40 times as long. Which factor should be used to rescale the dose? Freedman and Zeisel, Statistical Science, vol 3, p 3-28.

There are many unresolved problems in the standard bio-assay, including... how to account for inter-species and intra-species differences in metabolism and pharmacokinetics. Omenn & Lave, Mutation Research, vol 205, p 41.

Data from dose-response relationships are sometimes employed to estimate the expected tumour incidence at dose levels very much lower than those which could possibly be employed in conventional [animal] experiments. A number of mathematical models have been employed in such estimates and the results obtained vary considerably. Despite these uncertainties - and the possibly greater ones in extrapolating from animals to man - such models are often employed in some quarters to estimate the likely risk to man. Dr Paul Grasso, Perspectives in Basic and Applied Toxicology, p268-284, ed. Ballantyne (publ. Butterworth).

Extrapolating from one species to another is fraught with uncertainty... For almost all of- the chemicals tested to date, rodent bio-assays have not been cost-effective. They give limited and uncertain information on carcinogenicity, generally give no indication of mechanism of action, and require years to complete." [They are] "rarely the best approach for deciding whether to classify a chemical as a human carcinogen. Lave, Ennever, Rosenkrantz and Omenn, Nature, vol 336, p 631.

1987

...of the 20 probable human non-carcinogens with conclusive animal bioassay results, only one, methotrexate, is negative, and the other 19 are positive... Thus, the standard interpretation of animal bioassay results provides essentially no differentiation between definite human carcinogens and probable human non-carcinogens. Drs Ennever, Noonan and Herbert, Mutagenesis, vol 2, p 73-78.

The standard carcinogenicity bioassay, which involves treating two rodent species for a minimum of 2 years, at a range of doses, is acknowledged to be an insensitive tool because of the background `noise' of spontaneous disease. Most strains of rat used in such studies have high incidence of pituitary and mammary tumours; some inbred rat strains frequently develop leukaemia or testicular tumours; mice strains show high incidence of malignant lymphomas and liver tumours. Dr Mary Tucker, Human Toxicology, vol 6, p107-109.

1986

If we wish to understand human cancer, the [research] effort should be made in humans because the genetic control seems to be different in different species. Renato Dulbecco (Nobel Laureate). Science, 231: 1055-1056.

Let us look at some animal carcinogens – gold, DDT, clofibrate and bromocriptine. There is no doubt that all of these can rightly be regarded as carcinogenic for rodents, and yet there is really quite good evidence that they are not carcinogenic to man. Prof Andre McLean, Long Term Animal Studies-Their Predictive Value for Man (publ. MTP Press).

1985

During a chronic study the problem may be complicated by the presence of concomitant diseases; it is in fact known that acute inflammation or the presence of a tumor may affect the kinetics of chemical, thus altering their potential toxicity. Silvio Garattini, Toxic Effects of Chemicals: Difficulties in extrapolating data from animals to man; Critical Reviews in Toxicology, vol 16, issue 1, p1-29.

1984

It is painfully clear that carcinogenesis in the mouse cannot now be predicted from positive data obtained from the rat and vice versa. Dr F J Di Carlo, Drug Metabolism Reviews, vol 15, p 409-413.

1983

The major problems of animal studies are the validity of cross-species comparisons and relevance to human disease. Johanna Dwyer, Fundamental & Applied Toxicology 3: 63-67.

The lifetime feeding study of mice and rats appears to have less than a 50% probability of finding known human carcinogens. On the basis of probability theory, we would have been better off to toss a coin...The `definitive bioassay for carcinogenesis' as now designed has never been subjected to proper validation as an assay for human carcinogens. At attempt made in this paper to examine the literature suggests that it may have an unacceptably high false negative rate and that it produces so many contradictory answers as to suggest a very poor specificity. Dr. David Salsburg, Fundamental and Applied Toxicology, vol, 3, pp. 63-67.

1982

Animal model systems in cancer research have been a total failure…not a single essential drug for the treatment of human cancer was first picked up by an animal model system. All of the drugs in wide current clinical use were only put into animal model systems after finding clinical clues to their therapeutic possibility. The money was spent…for two main reasons. First, it was a highly profitable undertaking for certain medical schools and research institutions that were incapable of doing any genuine cancer research. Second, it was sustained by a superstitious belief in a grossly unscientific notion: mice are miniature men…in sum, from the standpoint of current scientific theory of cancer, the whole mystique of the animal model systems is hardly more than superstitious nonsense…the moral is that animal model systems not only kill animals, they also kill humans. There is no good factual evidence to show the use of animals in cancer research has led to the prevention or cure of a single human cancer. Dr Irwin Bross, Dr Irwin Bross (formerly Director of the Roswell Park Memorial Institute for Cancer Research) November issue, Fundamental and Applied Toxicology.

1981

The rodent’s 6 pairs of mammary glands suggest an increased likelihood for tumor development compared to the human. Bernard L. Oser, Journal of Toxicology & Environmental Health; vol. 8, p521-642.

The discovery of chemotherapeutic agents for the treatment of human cancer is widely heralded as a triumph due to the use of animal models… However, there is little, if any, factual evidence that would support these claims… Indeed, while conflicting animal results have often delayed and hampered advances in the war on cancer, they have never produced a single substantial advance in either the prevention or treatment of human cancer. Dr Irwin Bross (formerly Director of the Roswell Park Memorial Institute for Cancer Research) testifying to US Congress.

1980

Even when there are common target sites for a given carcinogen, there are usually important differences, between man and animals, and between different species and strains of animals. These 'spontaneous' tumours in rats and mice... [vary] widely according to sex, strain, diet, conditions of maintenance, hormonal status, immunological status and latent virus infections. Dr R L Carter, British Journal of Cancer, vol 41, p 494.

…one of the great fallacies in this calculation is that they are assuming that the mouse or rat or the hamster predicts for man, and we have no basis for this prediction…So it’s again a half-baked guess… Does the animal model have any relevance to human disease? If not we’re wasting a lot of time, a lot of money, a lot of good scientists, and a lot of good space at NIH… I completely agree with Dr. Clayton that extrapolation is unscientific… the chief objective here is to keep us all employed and to make sure we do interesting experiments so we can keep coming back to nice places like this. Coulston and Shubick (Eds) Human Epidemiology and Animal Laboratory Correlations in Chemical Carcinogenesis, p391-3 and p309 (publ. Ablex).

It is in fact hard to find a single, common solid neoplasm [cancer] where management and expectation of cure has been markedly affected by animal research. Most human cancers differ from the artificially produced animal model… Harrison, Clinical Oncology, 15: 1-2.

We have a discrepancy between animal data and human data…clearly, right now our animal models are totally and absolutely inadequate to answer all the obvious questions before us. Human Epidemiology and Animal Laboratory Correlations in Chemical Carcinogenesis, Coulston and Shubick (Eds), p13 (Ablex Publ.).

1979

As a cancer specialist engaged in clinical practice, I can’t agree with the researchers who believe that results obtained with laboratory animals are applicable to human beings. Dr Heinz Oeser, Quick, 15th March.

1976

Unfortunately, extrapolations from animal results to man remains largely problematic and no amount of mathematical sophistication can render such extrapolation more certain. Higginson and Muir, Cancer Detection and Prevention, 1(1) p79-105.

1952

Warning is given not to carry over, without reservation, to man, the conclusions based on animal experiments. In monkeys none of the powerful carcinogens [of man] has been shown to produce cancers. The Lancet (1952) Aug 9, p 274.

1950

The characteristic effects in leukaemia were detected solely as a result of clinical observation. The various leukaemias in the mouse and rat were relatively refractory to the influence of urethane, and the remarkable effect in the human might have eluded discovery if attention had been directed to the animal alone. That illustrates the hazards of such work. Dr Alexander Haddow, British Medical Journal, Dec 2, p1272.