in 20 000 live births and this includes both familial and non-familial cases. There were 1068 children born in Seascale between 1950 and 1983.2 There would have to be a 20-fold increase in incidence before a single case was expected in this population. It is nonsense to state that retinoblastoma should be "generally more common in Seascale." Similar arguments apply to Millom rural district. It is the extreme rarity of retinoblastoma that makes the observation of five cases linked to Seascale so extraordinary and noteworthy. The ideas in our paper are neither confusing nor contradictory. The data, in fact, strongly suggest a link between retinoblastoma and residence in Seascale. The nature of the association remains to be determined, but it is wrong to imply that radiation can be excluded as the causative agent. J A MORRIS Lancaster Moor Hospital, Lancaster LAl 3JR
JENNIFER BUCKLER The Surgery, Kendal LA9 4BD 1 Gardner MJ, Hall AJ, Downes S, Terrell JD. Follow up study of children born elsewhere but attending schools in Seascale, West Cumbria (schools cohort). BMJ 1987;295:819-22. 2 Gardner MJ, Hall AJ, Downes S, Terrell JD. Follow up study of children born to mothers resident in Seascale, West Cumbria (birth cohort). BMJ 1987;295:822-7. 3 Gardner MJ, Snee MP, Hall AJ, Powell CA, Downes S, Terrell D. Results of a case-control study of leukaemia and lymphoma among young people near Sellafield nuclear plant in West Cumbria. BMJ 1990;300:423-9.
Sexual expression in paraplegia SIR,-Although we welcome the prominence given to an important subject, Dr J M Kellett's editorial on sexual expression in paraplegia' contains misleading elements. Lesions of the cauda equina do not "remove all except seminal emission"; men with complete lesions of the cauda equina often have fully stiff psychogenic erections.2 The efferent pathway for these erections is presumably the sympathetic erectile pathway demonstrable by stimulation of the hypogastric plexus.3 The statement that "lesions above Ti 1 allow reflex erections" is misleading, because lesions as high as T6 may abolish erections if there is descending damage below the primary lesion (as is common in spinal injuries), but complete transections as low as the L5 segment of the cord sometimes permit reflex erections if the sacral segments survive. It is unfortunate that no mention was made of fertility and parenthood. Reduced fertility in men45 and difficulties with pregnancy and delivery in women6 are vitally important sexual issues to patients with spinal cord injuries. Important recent developments in both of these have greatly improved this aspect of sexuality. G S BRINDLEY
B P GARDNER
Stoke Mandeville Hospital,
Aylesbury HP21 8AL I Kellett JM. Sexual expression in paraplegia. BMJ 1990;301: 1007-8. (3 November.) 2 Bors E, Comarr AE. Neurological disturbance of sexual function with special reference to 529 patients with spinal cord injury. Urological Survey 1960;10: 191-222. 3 Brindley GS, Sauerwein D, Hendry WF. Hypogastric plexus stimulators for obtaining semen from paraplegic men. BrJ Urol 1989;64:72-7. 4 Brindley GS. The fertility of men with spinal injuries. Paraplegia 1984;22:337-48. 5 Bennett CJ, Seager SW, Vasher EA, et al. Sexual dysfunction and ejaculation in men with spinal cord injury: review. 7 Urol 1987;139:453-7. 6 Verduyn WH. Spinal cord injured women, pregnancy and delivery. Paraplegia 1986;24:231-40.
SIR, -Dr G A Kaplan and Professor J T Salonen found that adults who reported a lower socioeconomic status during childhood had an increased rate of ischaemic heart disease. ' However, no explanation was proffered for the findings, and data that might provide an explanation, such as the relative weights of subjects in childhood and adulthood, were not reported. In many communities the poor tend to be underweight infants but overweight adults. Abraham et al found that such a change substantially increased the risk of cardiovascular disease.2 A partial explanation may have been provided by Williams, who found that plasma concentrations of high density lipoprotein cholesterol were related more to weight change from the set point weight than to actual weight.' Nevertheless, despite the absence of data on relative weights Professor D J P Barker argued that this study is further evidence that fetal and infant malnutrition manifesting as low birth weight and low infant weight predispose to cardiovascular disease in adult life.4 This hypothesis has at least three problems. Firstly, in developing countries infant weight is often considerably lower than it is in developed countries, yet cardiovascular disease is less prevalent. For example, in rural Bangladesh the 50th centiles for weight at birth and at 1 year are 2 5 kg and 6-5 kg; by contrast, the Harvard standard is 3-5 kg at birth and 10 kg at 1 year.5 Secondly, there have been no reports of increased cardiovascular disease in people subjected to famine prenatally or during infancy. During the Dutch famine of 1944 birth weight and placental weight declined, but no subsequent cardiovascular effects have been detected. In fact, the adult survivors of this intrauterine exposure to famine seem to be indistinguishable in most respects from control cohorts.6 Similar results were found after food rationing in Great Britain.7 Thirdly, there is evidence that, except during famine, genetic factors play an important and often decisive part in determining weight at birth and during infancy.89 I believe that it is only infants who are overweight for genetic reasons who are relatively protected from cardiovascular disease8"' and that Professor Barker is wrong when he advocates nutritional intervention to promote a heavier infant weight." Court and Dunlop found that infants who were overweight because of overnutrition had raised plasma lipid concentrations whereas infants who were overweight for genetic reasons had normal plasma lipid concentrations. " It seems overwhelmingly clear that overfeeding infants whose "underweight" is genetically determined will increase, not decrease, the risk of cardiovascular disease. PATRICK J BRADLEY PO Box 892, Bondi Junction,
Sydney, Australia 2022
MRC Neurological Prostheses Unit, Institute of Psychiatry, London SE5 8AF
Fetal and infant origins of adult disease
12 JANUARY 1991
1 Kaplan GA, Salonen JT. Socioeconomic conditions in childhood and ischaemic heart disease during middle age. BMJ 1990; 301:1121-3. (17 November.) 2 Abraham S, Collins G, Nordsieck M. Relationship of childhood weight status to morbidity in adults. Health Services and Mental Health Administration Health Reports 1971;86:273-84. 3 Williams PT. Weight set-point theory predicts HDL-cholesterol levels in previously obese long-distance runners. Int Obes 1990;14:42 1-7. 4 Barker DJP. The fetal and infant origins of adult disease. BMJ 1990;301:1111. (17 November.) 5 Khan M. Infant feeding practices in rural Meheran, Comilla, Bangladesh. AmJ Clin Nutr 1980;33:2356-64. 6 Stein Z, Susser M, Saenger G, et al. Famine and human development: the Dutch hunger winter of 1944/45. New York: Oxford University Press, 1975. 7 Strachan DP, Hart 3T. Fetal and placental size and risk of hypertension in adult life. BMJ 1990;301:552. (15 September.) 8 Bradley PJ. Weight in infancy and death from ischaemic heart disease. Lancet 1989;ii:985.
9 Bradley PJ. Fetal and placental size and risk of hypertension in adult life. BMJ 1990;301:551. (15 September.) 10 Bradley PJ. Obesity and risk of coronary heart disease in women. N Engl1j4Med 1990;323:1144. 11 Bradley PJ. Is obesity an advantageous adaptation? IntJ Obes 1982;6:43-52. 12 Barker DJP, Winter PD, Osmond C, Margetts B, Simmonds SJ. Weight in infancy and death from ischaemic heart disease. Lancet 1989;ii:577-80. 13 Court JM, Dunlop M. Obese from infancy: a clinical entity. In: Howard A, ed. Recent advances in obesity research. I. London: Newman Publishing, 1975:34-6.
Socioeconomic conditions and ischaemic heart disease SIR, - Drs George A Kaplan and Jukka T Salonen attempt to deal with the possibility that poor socioeconomic state in childhood is followed by worse conditions in later life.' They follow the usual path of controlling for confounding factors by adjusting the odds ratio in a multiple logistic model and conclude that socioeconomic conditions in childhood and adulthood were "independent predictors" of ischaemic heart disease. We suggest that the analyses presented, and the methods used, do not allow this conclusion to be drawn. The magnitude of the association between a risk factor (exposure) and a disease outcome depends to a large extent on the degree of imprecision in measuring the exposure. In cases in which there is only one risk factor random misclassification of exposure will lead to underestimation of the magnitude of the association between exposure and disease.2 When two or more potential exposures are strongly correlated, such as childhood and adulthood socioeconomic conditions, the case is more complex. If the exposures are considered together in, for example, a multiple logistic model differential degrees of imprecision of measurement can lead to associations being overestimated, underestimated, or present when in reality they are non-existent.3 In particular, even small degrees of imprecision can lead to considerable residual confounding even after other exposures have supposedly been adjusted for.4 In the study by Drs Kaplan and Salonen an odds ratio of 144 associated with markers of poor childhood socioeconomic conditions was reduced to 1 21 after adjustment for a summary index of adult socioeconomic conditions (this is not a 16% reduction as they claim). This gives the impression that the association between childhood socioeconomic state and ischaemic heart disease is largely independent of conditions in later life. Socioeconomic state, however, serves as a proxy measure for factors that directly increase the risk of disease. Improving the categorisation of socioeconomic state-by combining factors such as social class, car ownership, housing tenure, occupational level, etc-greatly enhances the differences seen in mortality risk.56 Thus it is likely that a considerable degree of residual confounding remains after adulthood socioeconomic state has ostensibly been controlled for. In their study after adjustment for adult socioeconomic state the 95% confidence intervals for the odds ratio for childhood socioeconomic state include 10, though the authors do not directly refer to this. But when they found that adjusting for prevalent disease led to a similar change in odds ratio they said that the association was substantially reduced and non-significant. Significance, however, is not a crucial issue in these cases: an association could remain strongly significant and still be entirely due to residual
confounding. Many investigations depend on isolating independent effects from an intercorrelated web of potential risk factors. The identification of these apparently independent effects often better reflects the measurement precision and adequacy of proxy measures of exposures than the underlying causal mechanismns. A more sophisticated analysis is 113
required to separate the effects of childhood socioeconomic conditions from the whole life course of differential exposures with which they will be associated. Most importantly, we need to maintain a much more critical and cautious attitude to declarations that effects are independent.7 GEORGE DAVEY SMITH ANDREW PHILLIPS London School of Hygiene and Tropical Medicine, London WC I E 7HT 1 Kaplan GA, Salonen JT. Socioeconomic conditions in childhood and ischaemic heart disease during middle age. BMJ 1990;301: 1121-3. (17 November.) 2 MacMahon S, Peto R, Cutler J, et al. Blood pressure, stroke and coronary heart disease. 1. Prolonged differences in blood pressure: prospective observational studies corrected for the regression dilution bias. Lancet 1990;335:765-74. 3 Rosner B, Spiegelman D, Willett W. Correction of logistic regression relative risk estimates and confidence intervals for measurement error: the case of multiple covariates measured with error. AmJ Epidemiol 1990;132:734-45. 4 Savitz DA, Baron AE. Estimating and correcting for confounder misclassification. AmJ7 Epidemiol 1989;129:1062-71. 5 Goldblatt P. Mortality and alternative social classifications. In: Goldblatt P, ed. Longitudinal study: mortality and social organisation. London: HMSO, 1990:163-92. 6 Davey Smith G, Bartley M, Blane D. The Black report on socioeconomic inequalities in health 10 years on. BMJ 1990;301:373-7. (18-25 August.) 7 Smnith GD, Phillips AN. Declaring independence: why we should be cautious. J Epidemiol Community Health (in press).
SIR,-Drs George A Kaplan and Jukka T Salonen have inadvertently underestimated the effect of adjusting for adult social position on the relationship between childhood socioeconomic factors and ischaemic heart disease.' After adjusting for age and adult social circumstances the odds ratio of myocardial ischaemia among men from the lowest childhood socioeconomic group was reduced from 1 44 to 1-21 compared with that among men from the highest childhood socioeconomic group. Thus the excess risk associated with childhood disadvantage fell from 0 44 to 0 21 as a result of this adjustment: a reduction of 52%. The authors' estimate of a 16% reduction seems to have been derived by relating the absolute reduction in risk (1 44-1 21=0 23) to the unadjusted odds ratio (1 44) rather than to the excess risk (0 44). After adjusting for age and adult social circumstances the 95% confidence interval associated with the odds ratio for the "low" childhood socioeconomic group was 0-97 to 1 51 and that for the "medium" group 0-98 to 145. These confidence intervals are consistent with a small independent effect of childhood socioeconomic conditions on adult ischaemic heart disease. As the 95% confidence intervals for both these adjusted risk estimates included 1 0, however, these results do not provide strong evidence that the association between childhood socioeconomic conditions and ischaemic heart disease in middle age is independent of adult social position. The correct interpretation of their data is that after adjusting for age and adult social position socioeconomic state in childhood was not significantly associated with ischaemic heart disease in middle aged men. JONATHAN ELFORD
PETER WHINCUP A G SHAPER Royal Free Hospital School of Medicine, London NW3 1 Kaplan GA, Salonen JT. Socioeconomic conditions in childhood and ischaemic heart disease during middle age. BMJ7 1990;301: 1121-3. (17 November.)
AUTHOR'S REPLY,-We agree with Drs Davey Smith and Phillips that issues of measurement imprecision are important to consider in the interpretation of the results ofmultivariate models. As is often the case, however, there was insufficient information to consider the impact of measurement error adequately. In such cases the nature of the
measures and where imprecision might be introduced have to be considered. The decades that have elapsed since childhood will probably result in more error in the recall of answers to the six questions that measure childhood socioeconomic conditions than in the assessment of current state. This would be likely to lead to an underestimate of the association between childhood socioeconomic conditions and ischaemia. The impact of imprecision in the measurement of adult socioeconomic position will have to be assessed relative to this underestimation. The measure of adult socioeconomic state used in our analyses did combine a wide variety of relevant information (educational state, income, occupation, occupational prestige, standard of living, and housing conditions), contrary to the assertions of Drs Davey Smith and Phillips. Dr Elford and colleagues' concerns revolve around the conceptualisation of the role of adult socioeconomic position in our analyses. How should adult socioeconomic state be incorporated into analyses that also examine childhood factors? Certainly, adult state does not develop independently of childhood circumstances, and constraints on a person's life chances are, to some extent, set out early in life. To the extent that this is true then adult socioeconomic state represents part of the causal web linking childhood state and adult risks of disease. For these reasons we regret our unwise characterisation of childhood and adult socioeconomic states as independent predictors. For the same reasons, we cannot accept the suggestion by Dr Elford and colleagues that our results argue that childhood socioeconomic conditions are not importantly related to ischaemia because the association was reduced with adjustment for adult state. The assessment of confounding must consider as a special case those variables that seem to be confounders because they are, in part, consequences of the predictor. We agree, however, that computation of the' percentage decrease in relative risk is best based on excess risk. GEORGE A KAPLAN California Department of Health Services, Berkeley, California 94704-9980, United States
Lassa fever SIR,-Minerva comments about an editorial in the New England journal of Medicine on Lassa fever.' The article accompanying that editorial broadened the discussion about the nosocomial risks of the disease,2 but neither mentioned the two laboratory acquired infections in the United States that were important in stimulating strict precautions. Casual contact carries no risk,' but opinions vary about nosocomial contact. Some reports describe "hospital associated" cases.4 Others compare small numbers of hospital staff with large numbers of residents (both from endemic areas with high seroconversion rates), making it difficult to detect low level nosocomial transmission.5 Tribavirin can modify the course of severe Lasssa fever, though it is our impression that patients from Nigeria are more severely affected than those from Sierra Leone, where trials of tribavirin have so far been carried out. Early diagnosis is essential for the best chance of improvement, so a considerable level of alert is important in ensuring prompt admission. This has proved possible in the United Kingdom and no death from Lassa fever has occurred since 1971. In animal studies tribavirin prophylaxis works best if given very early. It reduces, but does not abolish, an established viraemia. Even in patients who have recovered from Lassa fever, viraemia persists during convalescence, whether recovery has been spontaneous6 or has followed treatment with tribavirin.7 In some arenavirus infections
prophylaxis during viraemia can result in delayed manifestations of disease.8 Reports of prophylaxis in humans are few and uncontrolled. Although strict isolation for patients with HIV and hepatitis B virus is pointless because staff are as likely to be exposed unknowingly as knowingly to such endemic diseases, in Western countries isolation is necessary for patients with viral haemorrhagic fevers which are non-endemic. Ebola, Marburg, and Crimean haemorrhagic fevers have caused outbreaks and are undoubtedly transmissible from person to person.90 They are less sensitive to antiviral drugs than is Lassa fever, and all present non-specifically, requiring expert assessment to avoid missing cases. Infection control methods used in this country are convenient and acceptable in trained hands. Continued vigilance ensures that patients with haemorrhagic fever are recognised early and screened promptly and expertly before a decision is made on high security isolation procedures. High quality pathology services are available on site; tribavirin is immediately available. This is particularly important in Britain, which is on established air routes from west Africa and receives more imported cases than does the United States. The United Kingdom High Security Infectious Diseases Units should therefore be maintained and, in particular, the skilled patient management services available on a 24 hour, year round basis. BARBARA BANNISTER
Royal Free Hospital, London N IO 1JN 1 Minerva. Views. BMJ 1990;301:1056. (3 November.) 2 Holmes GP, McCormick JB, Trock SC, et al. Lassa fever in the United States. NEnglJ3Med 1990;323:1120-3. 3 Galbraith NS, Berrie JRH, Forbes P, Young S. Public health aspects of viral haemorrhagic fevers in Britain. Royal Society Health Journal 1978;8:152-60. 4 Banatvala JE. Lassa fever. BM3' 1986;293:1256-7. 5 Monson MH, Frame D, Jahrling PB, Alexander K. Endemic Lassa fever in Liberia. 1. Clinical and epidemiological aspects at Curran Lutheran Hospital, Zorzar, Liberia. Trans R Soc Trop Med Hyg 1984;78:549-53. 6 Emond RTD, Weir WRC, Bowen ETW, Lloyd G, Southee T. Managing Lassa fever. Lancet 1984;ii:926. 7 Bannister B. Management of Lassa fever. Lancet 1986;i:35. 8 McKee K, Huggins JW, Trahan GJ, Mahlandt BG. Ribavirin prophylaxis and therapy for experimental Argentine haemorrhagic fever. Antimicrob Agents Chemother 1988;32:1304-9. 9 Suleiman MNEH, Muscat-Baron JM, Harries JR, et al. Congo/Crimean haemorrhagic fever in Dubai. Lancet 1980;ii: 939-41. 10 Minerva. Views. BMJ7 1985;291:1730.
Acute intestinal ischaemia SIR,-Mr M Welch' referred to Canadian work on hyperphosphataemia that was published nine years ago.2 As was generously admitted by the paper's principal author, Dr Jamieson, at the recent Vascular Society meeting, the results have not been reproduced everywhere and the work has not gained general acceptance. I therefore stand by my statement that at present there is no specific test that will distinguish intestinal ischaemia from other acute abdominal conditions.' ADRIAN MARSTON Middlesex Hospital, London WIN 8AA 1 Welch M. Acute intestinal ischaemia. BMJ 1990;301:1334. (8 December.) 2 Jamieson WG, Marchik S, Rowson J, Durand J. The early diagnosis of massive acute intestinal ischaemia. Br J Surg 1982;69:552-3. 3 Marston A. Acute intestinal ischaemia. BMJ 1990;301:1174-6. (24 November.)
SIR,-We agree with the points made by Mr Adrian Marston' and have adopted a similar policy of resection with second look laparotomy in most patients with acute intestinal ischaemia.2 Nevertheless, the role of aortography or selective
12 JANUARY 1991