309 oestrogens provoke endometrial bleeding which leads to the detection of symptomless tumours which would otherwise escape detection. Their argument depends on the existence of a large number of undetected tumours in the population, and in support of this they present a study of the frequency of undetected cancers in necropsy series from two New England hospitals. From these they estimated rates of 22 and 31 undetected endometrial tumours per 10 000 deaths, among women aged 45 or over with intact uteri. These rates are compared with the average annual rate of endometrial cancer reported by the Connecticut State Tumor Registry of 5-55 per 10 000 women aged over 45. Horwitz et al. conclude that the rate of detection at necropsy is four to six times greater than the rate of detection during life. However, the comparison they make is misleading and invalid since the annual rate of occurrence is quite different from the rate of detection during life. This can be seen by considering the occurrence of tumours among a group of 10 000 women aged 45 who are followed up until death. These women will live on average to 78’55 years, during which time 184 tumours will be diagnosed (555 per year for 3355 years), and a further 31 will be symptomless and remain undiagnosed at death. This gives a-ratio of 1 asymptomatic case to every 6 diagnosed cases, which is considerably different from the claim of Horwitz et al. of 3 asymptomatic to 1 diagnosed case. In fact, to obtain such a ratio they would need to have obtained a frequency of undiagnosed tumours in their necropsy series of 552 per 10 000. Horwitz et al. have thus presented data which demonstrate that detection bias is unlikely to produce any significant association of oestrogen use with endometrial cancer: it certainly cannot produce relative risks raised by four to five times their true value, as they claim. A.R.C
Epidemiology Research Unit, Stopford Building (University of Manchester),
IAIN K. CROMBIE
Manchester M13 9PT
JOHN TOMENSON
DEPRESSING EFFECT OF BURIAL CHARGES ON VITAL STATISTICS
SiR,-On June 9, at question time in the House of Commons, the Government accepted that the death grant, introduced in 1949 and not much increased since then, was well short of the cost of a funeral, and the Minister for Social Security promised an early decision on an increase. One effect of this shortfall could be inaccuracies in birth, fertility, and childhood mortality rates that depend on the accurate recording of the number of live births. If, after delivery, an infant "breathes or shows any other evidence of life, such as beating of the heart, pulsation of the umbilical cord, or definite movement of voluntary muscles, whether or not the umbilical cord has been cut or the placenta is attached" a live birth must be registered.2 Despite such clear definitions to distinguish live and stillbirth, "a proportion of the deaths which occur within a few days after birth are incorrectly registered as stillbirths, thereby inflating the stillbirth rate and lowering the neonatal mortality
rate."3 In Britain, a doctor who knowingly registers a birth or death incorrectly risks prosecution for perjury, so the scale of the practice of registering early neonatal deaths as stillbirths would be difficult to measure. If the practice were common, early childhood mortality rates would be depressed but deaths after the first day would be unlikely to be affected. The decline in first day mortality in England ---
and Wales since 1963 was attributed to the introduction of intensive care nurseries4 but the rate of decline in deaths in the 1-27 day age 1 U S. Department of Health Education and Welfare. Vital statistics in the United States 1973- vol II, mortality part A. Rockville, Maryland, 1977. 2 World Health
Organisation. Manual of the International Classification of Diseases, Injuries and Causes of Death, ninth revision; Vol I. Geneva: WHO, 1977: 763.
3 Barker
DJP, Rose G. Epidemiology in medical practice. Edinburgh: Churchill Livingstone, 1976: 135. 4 Pharoah POD, Alberman ED. Mortality of low birth weight infants in England and Wales,
1953 to 1979. Arch Dis Child
1981; 56: 86-89.
group showed no improvement, which might favour some artifactual change in the first day deaths. It is possible, though unlikely, that doctors might use this device artificially to lower neonatal and infant mortality rates. Alternatively, doctors might believe that distressed parents, would find it easier to accept a stillbirth than the death of a liveborn baby. Another possible explanation is an awareness of the great difference in cost between burying a stillborn fetus and an infant. Stillborn babies can be cremated or buried in unmarked graves at the expense of the health authority.5 Stillbirths or neonates can be buried privately in a marginal grave for about f25, but this means no mourners, no cars, no church service, and a rough wooden coffin. Even a simple burial with two cars, a church service, a marked grave, and a small gravestone will cost 280: the current death grant for a child under 3 years is f9. Neonatal deaths are more common among young parents of low social class.Such parents often have considerable financial problems in preparation for the arrival of a baby, yet they may be unwilling to have a common grave for their child. An awareness of these problems may have encouraged some doctors to register some early neonatal deaths as stillborn, unaware that this is perjurious. Thus burial charges might have had an indirect depressant effect on many vital statistics, and this potential artifact should be considered when mortality rates are compared. Dividing mortality into perinatal and postperinatal periods (i.e., putting the major division after the seventh day rather than the birth) might produce more consistent statistics. Children’s Hospital, Sheffield 10
R. SUNDERLAND
RELEASE OF CHILEAN DOCTORS
SIR,-Further to my letter in your issue of June 27 on the detention of three Chilean doctors, I am happy to be able to say that international pressure appears to have been effective, for these doctors have now been released and the charges dropped. They are, Iam reliably informed, back in their homes. This is yet one more demonstration that the Chilean Government has had to ’be responsive to some extent to international pressure. Department of Sociology, Bedford College, London NW1 4NS
MARGOT JEFFERYS
CIVIL DEFENCE AND NUCLEAR WAR
SIR,-The publisher of Protect Survive Monthly, Mr Watts, asks (July 18, p. 154) "if civil defence increases the risk of nuclear attack, as claimed by Dr Holdstock (July 4, p. 44), why does the Soviet Union have the most comprehensive civil defence programme in the world and the Swiss and the Swedes have highly comprehensive civil defence programmes which protect the
vast
majority of their
population?". Apart from the disingenuousness
of asking about the Swiss and Swedes when the answer was given in Holdstock’s letter and apart from comprehensively begging the question by claiming that such programmes "protect" the vast majority of their population, the answer to his question does not follow the tendentious lines that Watts hopes. If the Soviet Union were suddenly to put its programme into action and the entire population disappeared underground, it is hard to accept that NATO generals would believe that the risk of nuclear attack had also suddenly disappeared and mothball their own weapons. They might suspect imminent attack and adopt the strategy called counterforce which means using very accurate large surface-burst nuclear weapons to destroy some land-based missiles What to do after a death. Available from DHSS Leaflets Unit, PO Box 21, Stanmore, Middlesex HA7 1AY. 6. Office of Population, Censuses and Surveys. Infant and perinatal mortality (OPCS monitor DH1 79/2). London: HMSO, 1979.
5.
Department of Health and Social Security.
310
before launch. Cruise missiles have been developed over the past decade to fill this role. Lesser degrees of preparation for war provoke lesser degrees of military response, and only progressive disarmament can reduce the risks. The symmetry of the balance of terror dictates that the same argument applies to British as to Soviet civil defence. Indeed it has been said that civil defence is necessary to make deterrence credible because without it "they could call our bluff". This implies that a risk which would be unacceptable without civil defence would become acceptable. This reveals a gross ignorance of the effects of nuclear war. Just as there is no threshold for the biological effects of radiation, there is no threshold for the unacceptability of nuclear war. Finally, it would have been interesting to check Watts’ sources of information about the nature of the most comprehensive civil defence programme in the world, but they are not given. In common with claims for cancer cures, civil defence products are immune from prosecution under the Trade Descriptions Act. Doctors should examine their claims with equal care and likewise insist on
primary prevention. N. W. Regional Group, Medical Campaign against Nuclear Weapons, 5 Lorne Street,
NEILL SIMPSON
Mossley OL5 0HQ
RABIES VIRUS AND LABORATORY REGULATIONS
SIR,-Professor Kaplan raises
issue in his letter (July 4) on rabies virus which should be discussed again by the Dangerous Pathogens Advisory Group. After the smallpox incidents at the London School of Hygiene and Tropical Medicine in 1973 and the University of Birmingham in 1978, there was justifiable concern among microbiologists and the public about the safety of personnel in laboratories handling pathogens. One of the results of the deliberations of Sir George Godber’s Working Party on the Laboratory Use of Dangerous Pathogens was that infectious agents were placed in different categories according to risk. Rabies virus was placed in the same category as smallpox virus but this was illconceived because smallpox is contagious whereas rabies is not. In assessing the risks associated with working with a particular virus, the only real guide comes from our experience with it over the years. Kaplan cites two laboratory accidents when infection with rabies virus was caused by aerosols generated in the laboratory. What he does not mention is that in one of the accidents the person infected had an acute upper-respiratory-tract infection which almost certainly made him more susceptible to infection. Without wishing to make light of the risks involved in working with rabies virus, we note that the opinion of those involved seems to be that, provided the personnel at risk are protected by vaccination, there is little chance of infection. At Pirbright staff handling the virus are quite prepared to be vaccinated with the very effective rabies vaccine now available. The controls on handling the virus were included in the Rabies Order (March, 1979). Those ofus who were consulted when these restrictions were being drawn up had difficulty in reconciling our concepts of laboratory safety with those whose minds were already made up and who were no doubt reacting, not only to the smallpox incidents of 1973 and 1978 but also to the separate issue of the two imported dogs which, in 1969 and 1970, developed the disease after leaving quarantine. Nowadays, when all experts have their own jargon, it is surprisingly difficult for a virologist to convince a non-virologist that even inactivated or non-infectious virus really means safe and innocuous. How, then, can we convince a non-virologist that live virus may be handled safely with the minimum of containment by a skilled an
operator. Accidents are much less likely when movements can be simple, and precise and when the operator not handicapped by unnecessary manipulations in the cause of so-called protection. As our regulations stand, few young scientists are likely to become rabies virologists when they can work on other topics with greater comfort and less hindrance. The result will be that in a few years, few if any virologists in the U.K. will have experience of handling
quick,
rabies virus. When the application of genetic engineering to the production of rabies vaccines and the important study of rabies pathogenesis are being pursued elsewhere where the virus is being handled under much less stringent conditions, work in the U.K. is being slowed down to a snail’s pace by the restrictions imposed. We do not want rabies in Britain as an isolated incident. Nor do we want it to become enzootic. However, in this Institute we already handle a number of exotic viruses which could infect domestic or wildlife species far more rapidly than could rabies. If our containment facilities are adequate for these, why then do we need class A facilities in’addition for rabies virus. The situation borders on the ridiculous. Animal Virus Research Institute, Pirbright, Surrey GU24 0NF
F. BROWN
J. CRICK
AXILLARY OR RECTAL TEMPERATURES IN CHILDREN?
SIR,—Dr Pituch and Dr Klein (July 4, p. 43) recommended that axillary temperatures no longer be used to determine the presence or absence of fever. Their conclusions are suspect, since they are based on an analysis of repeated observations of rectal and axillary temperatures in only five children. "Data analysis that fails to distinguish between observations on the same individual and observations on different individuals is entirely meaningless. If the goal pertains to the distribution of (a variable) in the general population, the most important item regarding sample size is the number of different individuals and not the total number of observations."1 With n=5, Pituch and Klein’s data would almost certainly show no significant difference between axillary and rectal temperature. We have recently compared the rectal and axillary temperatures of 75 patients. The patients were aged 0-5 months, 6-11 months, 12-23 months, 2-12 years and over 12 years, with 15 patients in each age group. Each patient was studied on only one occasion. Temperatures were taken with a ’Clinic I’ electronic thermometer (AMI Medical Electronics, Ronkonkoma, U.S.A.), left in situ for one minute. The mean rectal temperature was 37-80±0-95°C (±SD) and the mean axillary temperature was 36 -73±0’97 °C. Axillary temeratures were therefore adjusted by adding 1-1°C. The mean difference between rectal temperature and adjusted axillary temperature was 0.33±0.24°C. Of the 38 patients with a rectal temperature of 37’5°C or less, 31 had an adjusted axillary temperature of 375’C or less (and all 38 had an adjusted axillary temperature of 38’0°C or less). Of the 37 patients with a rectal temperature over 37 -5°C, 30 had an adjusted axillary temperature over 37’5°C (and 36 had an adjusted axillary temperature over 37’0°C). In the 38 afebrile patients the rectal-adjusted axillary temperature difference was 0.29±0.24°C, and in the 37 febrile patients it was 0.41±0.27°C; the difference is not significant (t 1-957; 73 df; p>0. 05). The correlation coefficient between axillary and rectal temperature was 0-90 (p<0-001). The findings were similar in all age groups. Great precision in the measurement of temperature is unnecessary, since the definition of fever is arbitrary and "normal" temperature varies from person to person and depends on the time of day, state of activity and emotional state. On average, rectal temperature is 1-0°C higher at noon than at midnight.2 Rectal temperature depends on the depth to which the thermometer is inserted, and it rises after a bowel motion. In view of this, a difference of 0.33±0.24°C between rectal temperature and adjusted axillary temperature is of little importance. The measurement of axillary temperature is easier, less distressing to the child and cannot cause rectal perforation. Axillary temperature plus 1.1°C corresponds closely to rectal temperature, and is an acceptable way of measuring temperature in children. Goroka Base Hospital, P.O. Box 392, Goroka, Eastern Highlands Province, Papua New Guinea
FRANK SHANN ANGELA MACKENZIE
1. Colton T. Statistics in medicine. Boston: Little, Brown, 1974, 41-43. 2. Mills JN. Development ofcircadian rhythms in infancy. In: Davis JA, Dobbing J eds. Scientific Foundations of paediatrics. London: William Heinemann, 1974: 764