261
SIR,-The confirmation by Dr Stein and his colleagues’ of earlier observations’’ of a small weight gain in the first few days post-partum is’ relevant to discussions concerning the setiology of post-partum hypertension (P.P.H.T.) and of peri-partum cardiac failure (P.P.C.F.).3-6 There are, however, reasons why these observations are unlikely to be of great importance in this context. The observed gains are small, and are reversed by one week after delivery. They presumably relate to the sodium retention necessary to raise the relatively low plasma-sodium concentrations of pregnancy to "normal". By contrast,’8 P.P.H.T. peaks, on average, in the second week post-partum, but it may not appear until several weeks later and after considerable weight loss (see figure), and it may also persist for months. In Zaria there were no differences between the weekly weight losses after delivery of patients with P.P.H.T. and of those without. P.P.H.T. certainly occurred in some who had no fluid retention, and others who had taken no kanwa, although both almost certainly further raised blood-pressure levels. P.P.C.F., likewise, most commonly dates from the second week post-partum, but usually presents a month or more after delivery. It, too, may
Ante-partum (sitting) and post-partum (lying) blood pressures and weights of a 15-year-old girl during her first pregnancy, on discharge one day after delivery, 5 days after delivery, and weekly thereafter. Kanwa was taken for the first 11 days of the puerperium only. without salt loading, and may present as acute left-ventricular failure in the absence of hypertension or of oedema. P.P.H.T. and P.P.C.F. appear to be entities in their own right, but there is little doubt, albeit on circumstantial grounds, that both are worsened in Zaria by taking kanwa after childbirth, and it is highly likely that P.P.H.T. is itself partly responsible for the very high incidence of P.P.C.F. among Hausa women in Zaria. Whether excessive sodium retention by otherwise normal women in the puerperium is fundamental to the genesis of
occur
P.P.C.F.,b
or
whether it is due
to an
increased
liability
to myo-
cardial slippage in response to modest hypertension,5 is open to question. Despite Sanderson’s argument6 and his unpublished data suggesting that cardiac output is normal in these patients, they usually have massive oedema, a markedly raised
1. 2.
Stein, G. S., Morton, J., Keating, P. Lancet, 1978, i, 1214. Dennis, K. J., Bytheway, W. R.J. Obstet. Gynæc. Br. Commonw. 1965, 72, 94.
Davidson, N. McD., Trevitt, L., Parry, E. H. O. Bull. Wld Hlth Org. 1974, 51, 203. 4. Sanderson, J. E. Lancet, 1977, ii, 1159. 5. Davidson, N. McD. ibid. 1978, i, 145. 6. Sanderson, J. ibid 7. Davidson, N. McD, Parry, E H. O. Q. Jl Med. (in the press). 8. Davidson, N. McD. D.M. thesis, University of Oxford, 1978. 3.
venous pressure, cardiomegaly which is often gross, and severe oliguria in the absence of renal damage and which is readily reversed by digoxin or diuretics. Most observers would, I suggest, agree to call such a syndrome cardiac failure (as in our descriptive term P.P.c.F.), although some pedants (like myself) would quarrel with calling it a cardiomyopathy until the pathophysiology has been more precisely defined.
jugular
PERI-PARTUM CARDIAC FAILURE
Department of Medicine,
University of Edinburgh, Western General Hospital, Edinburgh EH4 2XU
N. MCD. DAVIDSON
PROPHYLAXIS AGAINST POSTOPERATIVE PULMONARY EMBOLISM
SiR,—Dr Kiil and colleagues’ have confirmed that postoperative pulmonary emboli are reduced by low-dose heparin. Their
patients were randomly allocated to receive twice-daily injections of sodium heparin or saline for a week. Four died from pulmonary embolism in the placebo group and one in the heparin group. The preoperative and postoperative transfusion requirements were identical in the two groups, and the surgeons had no impression of increased bleeding tendencies in the heparin group. Unfortunately, the incidence of wound hasmatomata and subsequent sepsis was not mentioned, and ophthalmic and neurosurgical operations did not appear in the trial, no doubt excluded because of the disastrous consequences of iatrogenic bleeding into the eye or brain. Multiple injections of controls with saline would not have pleased Pappworthzbut having gone to such lengths, how interesting it would have been to compare the degree of bruising and pain at the injection site in each group. The crusade for low-dose heparin seems endless, started by Sharnoff,3 taken up by Kakkar 4and backed by advertising and television. By now the message must have been received loud and clear-yet how many surgeons in busy peripheral hospitals are offering routine heparin prophylaxis? Very fewand then only for high-risk cases. Why? Perhaps wound complications and the extra work involved are disliked by the nurses, or the discomfort from twice-daily injection deters the patients. To patients the possibility of a pulmonary embolus might seem remote, and the heparin is just an expensive nuisance. All prophylaxis must be safe, simple, and applicable to every adult patient on every operation list whatever the operation, excluding only minor surgery. Browse6 has given an admirable review of this dilemma and listed the wide choice of pharmacological and mechanical aids available. Most of these methods work, and it is high time we compared cost and convenience instead of constantly reiterating their efficacy. A technique which gives the least bother to hospital staff will win the day because it will actually be used, and used repeatedly. No surgeon writes to The Lancet without a bee in his bonnet, and I have to own up to our solution to deep-vein thrombosis and pulmonary embolus on the "costa geriatrica", developed by Doran and myself in 19737 and based on work by Doran et al,8 Clark and Cotton,9 and Browse and Negus. 10 We use electrical stimulation of the calf muscles by battery-driven portable ’Electronic Gaiters’. Easily applied by the anaesthetic room technician, these cause contractions every 3-5 s throughout the operation. Consecutive patients on every surgical list are given routine prophylaxis, with no hindrance or complica1. 2.
Kiil, J., Axelsen, F., Kiil, J., Andersen, D., Lancet, 1978, i, 1115.
Pappworth,
M. H. Human Guinea
Pigs: Experimentation
on
Man.
1967. 3. Sharnoff, J. G. De Blasio, G. Lancet, 1970, ii, 1006. 4. Kakkar, V. V. Br. J. Hosp. Med. 1977, 18, 32. 5. International Multicentre Trial Lancet, 1975, ii, 45. 6. Browse, N. L. Ann. R. Coll. Surg. 1977, 59, 138. 7. Powley, J. M., Doran, F. S. A. Lancet, 1973, i, 406. 8. Doran, F. S. A., White, M., Drury, M Br. J. Surg. 1970, 9. Clark, C., Cotton, L. T. ibid. 1968, 55, 211. 10. Browse, N. L., Negus, D. Br. med. J. 1970, iii, 615.
57, 20.
London,
262 tions. There are no trailing wires, tubes, gas cylinders, or noise distract the surgeon, and there have been no thromboembolic complications in patients so treated for the past two years. to
The
gaiters
are
available from Downs
Surgical Ltd., Mitcham,
Sur-
rey. Eastbourne District General Eastbourne, Sussex
Hospital,
J. M. POWLEY
LOW-DOSE RADIATION
all members of
a population are equally sensitive to radiation, (c) sensitivity to radiation does not vary with age, and (d) the dose-response relationship is linear. Some of these provisos, as we will now try to show, may be invalid or inapplicable to the Hanford data. Consider two equal-size groups of 18-year-olds, one receiving nothing except background radiation and the other having received a further dose of 3.6 rad. The total effective doses would thus be 1.8rad (0-1x18) for the first group and 5.4 rad (1.8+3.6) for the second. Also, by (b), (c) and (d) there would be three times as many radiogenic cancers in the second group as in the first (alternatively, any extra
in the second group would equal twice the number of radiogenic cancers in the first). Finally by our risk estimates the second group would have twice as many bone-marrow tumours as the first, so half the tumours in the first group would be caused by background radiation. By repeating this argument for older groups one would soon discover that the total number of bone-marrow tumours was smaller for men over 36 years than the estimated numbers of radiogenic cases. This reductio ad absurdum makes it appropriate to question some widely held assumptions about the mode of action of radiation and some consequences of these assumptions. In earlier papers on the Hanford data doses were measured in rems, but we gave doubling doses in rads because the more refined units (rem) seemed illusory when, for most of the study period, film badges precluded the possibility of separating the neutrons from, for example, gamma rays. Since neutrons may be 20 times as dangerous as gamma rays we decided that the first basic assumption (a) was not applicable to Hanford data. One reason why we used the doubling-dose model was because we assumed that the mode of action was the same for radiation as for other carcinogens (i.e., mutagens are potential carcinogens, and vice versa). We also knew that genetic constitution exerts a strong influence on cancer sensitivity and that cancers
SIR,-Dr Mole1 has commented
on our
analyses
of
cancer
deaths of Hanford workers in two periods (1944-722 and 1944-773). We would like to correct one error of fact and several mistaken impressions that he seems to share with others. Mole decided that, between the two analyses, the number of cancer deaths rose by only 11% from 670 to 743. However, these two figures have different bases. The smaller figure included 228 men who were either not monitored for external radiation or recorded zero doses. The larger figure was the number of monitored cases (some with zero doses) in the second analysis, by which time the total number of cancer deaths was 924. Thus the true increase was 38%. In the first analysis2 we were not in a position to distinguish between non-monitoring and zero dose. In the second, this was possible, so we excluded 181 cancer deaths of non-monitored workers. Since a third of the cancers in the first analysis were either not exposed or recorded zero dose, a distribution of our doubling-dose estimates would have been very anomalous, so we did not give any confidence limits. Once non-exposure and zero dose were distinguished quotation of confidence limits was justified although there was still a very wide gap between the upper and lower levels. Therefore the apparent anomalies noted by Mole are not at all surprising. Mole and others put great stress on the possibility that estimates of dose effects from occupational exposure might be biased because exposure period and cumulative dose are correlated. In our second analysis3 we included exposure period as one of several controlling factors in a Mantel-Haenszel analysis. If exposure period had been important the radiation effect would have been much smaller in the controlled than the crude analysis, but it was not. The difference between our results and those of Marks et al.4 (who examined Hanford data for the period 1944-74) is more a matter of interpretation and emphasis than might be gathered from Mole’s letter. We uncovered three cancers with definite radiation effects (bone-marrow, pancreas, and lung) while the approach of Marks et al., which included a strongly controlled analysis, revealed a definite association between radiation dose and multiple myeloma and pancreatic cancer. The significance levels were such that the two findings could not both be statistical flukes. Marks et al. postulate, as a cause of the association, exposure to unidentified chemicals at the same time as radiation, and we are suggesting that the association is a direct one. Can our low risk estimates, in the form of doubling doses, be reconciled with what is known about background radiation and what is widely assumed to be the role of radiation in carcinogenesis ? Our lowest estimate of doubling dose is 3.6rad for bone-marrow cancers, and the background radiation dose is approximately 0.1 rad per annum. It is generally agreed that the number of radiogenic cancers is proportional to the total radiation dose, provided (a) all ionising radiations are sufficiently similar for similar values of doubling dose to apply, (b) 1. Mole, R. Lancet, 1978, i, 1155. 2. Mancuso, T. F., Stewart, A. M., Kneale, G. Hlth Phys. 1977, 33, 369. 3. Kneale, G., Stewart, A. M., Mancuso, T. F. Paper IAEA-SM-224-510 read at I.A.E.A. symposium on the Late Biological Effects of Ionising Radiation, held in Vienna on March 13-17, 1978. 4. Marks, S., Gilbert, E. S., Brettenstein, B. D. ibid. paper IAEA-SM-224-509.
different constitutions react differently to different diseases carriers of the gene for xeroderma pigmentosa, even in a recessive form, are predisposed to skin cancer). Therefore we decided that the second basic assumption (b) was probably invalid. This conclusion prepared us for the possibility of nonlinearity of the dose-response curve (at a population level) since high doses would saturate a small proportion of highly sensitive individuals and leave the rest of the population with a different level of response. Another version of unequal sensitivity would be expected if carcinogens interact, and several surveys suggest that in relation to lung cancer the combined effects of smoking and radiation are multiplicative rather than additive. Finally the third basic assumption of equal sensitivity to the cancer induction effects of radiation at all ages was most unlikely since the immune system is clearly playing an important role in the aetiology of all diseases, including cancers. Also the first and second analysis of Hanford data contain evidence which show that assumption (c) is invalid, and a similar conclusion can be drawn from an earlier analysis of data from the Oxford Survey of Childhood Cancers which showed a 16-fold decrease in sensitivity between the first and third trimesters of fetal life.5 Comparisons between observed and expected cancer deaths will eventually show whether there is a cancer hazard for workers in the nuclear industry. However, to be certain of success, this approach requires either much larger doses than were encountered in the Hanford study (where the mean cumulative dose was only 1-38 rad for cancers and 0.99 for non-cancers)
(e.g.,
much larger data base (say, 500000 man-years of follow-up instead of 150 000 man-years3). Therefore, unlike Dr Mole and Professor Anderson (July 15, p. 161), we attach more importance to the fact that the mean cumulative dose was higher for 22 bone-marrow cancers (4-49 rad) than for 42 other reticuloendothelial neoplasms (0-99 rad) than to the fact that
or a
5.
Kneale, G., Stewart, A. M.J. nath. Cancer Inst. 1976, 57, 1009.
SIR,-The confirmation by Dr Stein and his colleagues’ of earlier observations’’ of a small weight gain in the first few days post-partum is’ relevant to discussions concerning the setiology of post-partum hypertension (P.P.H.T.) and of peri-partum cardiac failure (P.P.C.F.).3-6 There are, however, reasons why these observations are unlikely to be of great importance in this context. The observed gains are small, and are reversed by one week after delivery. They presumably relate to the sodium retention necessary to raise the relatively low plasma-sodium concentrations of pregnancy to "normal". By contrast,’8 P.P.H.T. peaks, on average, in the second week post-partum, but it may not appear until several weeks later and after considerable weight loss (see figure), and it may also persist for months. In Zaria there were no differences between the weekly weight losses after delivery of patients with P.P.H.T. and of those without. P.P.H.T. certainly occurred in some who had no fluid retention, and others who had taken no kanwa, although both almost certainly further raised blood-pressure levels. P.P.C.F., likewise, most commonly dates from the second week post-partum, but usually presents a month or more after delivery. It, too, may
Ante-partum (sitting) and post-partum (lying) blood pressures and weights of a 15-year-old girl during her first pregnancy, on discharge one day after delivery, 5 days after delivery, and weekly thereafter. Kanwa was taken for the first 11 days of the puerperium only. without salt loading, and may present as acute left-ventricular failure in the absence of hypertension or of oedema. P.P.H.T. and P.P.C.F. appear to be entities in their own right, but there is little doubt, albeit on circumstantial grounds, that both are worsened in Zaria by taking kanwa after childbirth, and it is highly likely that P.P.H.T. is itself partly responsible for the very high incidence of P.P.C.F. among Hausa women in Zaria. Whether excessive sodium retention by otherwise normal women in the puerperium is fundamental to the genesis of
occur
P.P.C.F.,b
or
whether it is due
to an
increased
liability
to myo-
cardial slippage in response to modest hypertension,5 is open to question. Despite Sanderson’s argument6 and his unpublished data suggesting that cardiac output is normal in these patients, they usually have massive oedema, a markedly raised
1. 2.
Stein, G. S., Morton, J., Keating, P. Lancet, 1978, i, 1214. Dennis, K. J., Bytheway, W. R.J. Obstet. Gynæc. Br. Commonw. 1965, 72, 94.
Davidson, N. McD., Trevitt, L., Parry, E. H. O. Bull. Wld Hlth Org. 1974, 51, 203. 4. Sanderson, J. E. Lancet, 1977, ii, 1159. 5. Davidson, N. McD. ibid. 1978, i, 145. 6. Sanderson, J. ibid 7. Davidson, N. McD, Parry, E H. O. Q. Jl Med. (in the press). 8. Davidson, N. McD. D.M. thesis, University of Oxford, 1978. 3.
venous pressure, cardiomegaly which is often gross, and severe oliguria in the absence of renal damage and which is readily reversed by digoxin or diuretics. Most observers would, I suggest, agree to call such a syndrome cardiac failure (as in our descriptive term P.P.c.F.), although some pedants (like myself) would quarrel with calling it a cardiomyopathy until the pathophysiology has been more precisely defined.
jugular
PERI-PARTUM CARDIAC FAILURE
Department of Medicine,
University of Edinburgh, Western General Hospital, Edinburgh EH4 2XU
N. MCD. DAVIDSON
PROPHYLAXIS AGAINST POSTOPERATIVE PULMONARY EMBOLISM
SiR,—Dr Kiil and colleagues’ have confirmed that postoperative pulmonary emboli are reduced by low-dose heparin. Their
patients were randomly allocated to receive twice-daily injections of sodium heparin or saline for a week. Four died from pulmonary embolism in the placebo group and one in the heparin group. The preoperative and postoperative transfusion requirements were identical in the two groups, and the surgeons had no impression of increased bleeding tendencies in the heparin group. Unfortunately, the incidence of wound hasmatomata and subsequent sepsis was not mentioned, and ophthalmic and neurosurgical operations did not appear in the trial, no doubt excluded because of the disastrous consequences of iatrogenic bleeding into the eye or brain. Multiple injections of controls with saline would not have pleased Pappworthzbut having gone to such lengths, how interesting it would have been to compare the degree of bruising and pain at the injection site in each group. The crusade for low-dose heparin seems endless, started by Sharnoff,3 taken up by Kakkar 4and backed by advertising and television. By now the message must have been received loud and clear-yet how many surgeons in busy peripheral hospitals are offering routine heparin prophylaxis? Very fewand then only for high-risk cases. Why? Perhaps wound complications and the extra work involved are disliked by the nurses, or the discomfort from twice-daily injection deters the patients. To patients the possibility of a pulmonary embolus might seem remote, and the heparin is just an expensive nuisance. All prophylaxis must be safe, simple, and applicable to every adult patient on every operation list whatever the operation, excluding only minor surgery. Browse6 has given an admirable review of this dilemma and listed the wide choice of pharmacological and mechanical aids available. Most of these methods work, and it is high time we compared cost and convenience instead of constantly reiterating their efficacy. A technique which gives the least bother to hospital staff will win the day because it will actually be used, and used repeatedly. No surgeon writes to The Lancet without a bee in his bonnet, and I have to own up to our solution to deep-vein thrombosis and pulmonary embolus on the "costa geriatrica", developed by Doran and myself in 19737 and based on work by Doran et al,8 Clark and Cotton,9 and Browse and Negus. 10 We use electrical stimulation of the calf muscles by battery-driven portable ’Electronic Gaiters’. Easily applied by the anaesthetic room technician, these cause contractions every 3-5 s throughout the operation. Consecutive patients on every surgical list are given routine prophylaxis, with no hindrance or complica1. 2.
Kiil, J., Axelsen, F., Kiil, J., Andersen, D., Lancet, 1978, i, 1115.
Pappworth,
M. H. Human Guinea
Pigs: Experimentation
on
Man.
1967. 3. Sharnoff, J. G. De Blasio, G. Lancet, 1970, ii, 1006. 4. Kakkar, V. V. Br. J. Hosp. Med. 1977, 18, 32. 5. International Multicentre Trial Lancet, 1975, ii, 45. 6. Browse, N. L. Ann. R. Coll. Surg. 1977, 59, 138. 7. Powley, J. M., Doran, F. S. A. Lancet, 1973, i, 406. 8. Doran, F. S. A., White, M., Drury, M Br. J. Surg. 1970, 9. Clark, C., Cotton, L. T. ibid. 1968, 55, 211. 10. Browse, N. L., Negus, D. Br. med. J. 1970, iii, 615.
57, 20.
London,
262 tions. There are no trailing wires, tubes, gas cylinders, or noise distract the surgeon, and there have been no thromboembolic complications in patients so treated for the past two years. to
The
gaiters
are
available from Downs
Surgical Ltd., Mitcham,
Sur-
rey. Eastbourne District General Eastbourne, Sussex
Hospital,
J. M. POWLEY
LOW-DOSE RADIATION
all members of
a population are equally sensitive to radiation, (c) sensitivity to radiation does not vary with age, and (d) the dose-response relationship is linear. Some of these provisos, as we will now try to show, may be invalid or inapplicable to the Hanford data. Consider two equal-size groups of 18-year-olds, one receiving nothing except background radiation and the other having received a further dose of 3.6 rad. The total effective doses would thus be 1.8rad (0-1x18) for the first group and 5.4 rad (1.8+3.6) for the second. Also, by (b), (c) and (d) there would be three times as many radiogenic cancers in the second group as in the first (alternatively, any extra
in the second group would equal twice the number of radiogenic cancers in the first). Finally by our risk estimates the second group would have twice as many bone-marrow tumours as the first, so half the tumours in the first group would be caused by background radiation. By repeating this argument for older groups one would soon discover that the total number of bone-marrow tumours was smaller for men over 36 years than the estimated numbers of radiogenic cases. This reductio ad absurdum makes it appropriate to question some widely held assumptions about the mode of action of radiation and some consequences of these assumptions. In earlier papers on the Hanford data doses were measured in rems, but we gave doubling doses in rads because the more refined units (rem) seemed illusory when, for most of the study period, film badges precluded the possibility of separating the neutrons from, for example, gamma rays. Since neutrons may be 20 times as dangerous as gamma rays we decided that the first basic assumption (a) was not applicable to Hanford data. One reason why we used the doubling-dose model was because we assumed that the mode of action was the same for radiation as for other carcinogens (i.e., mutagens are potential carcinogens, and vice versa). We also knew that genetic constitution exerts a strong influence on cancer sensitivity and that cancers
SIR,-Dr Mole1 has commented
on our
analyses
of
cancer
deaths of Hanford workers in two periods (1944-722 and 1944-773). We would like to correct one error of fact and several mistaken impressions that he seems to share with others. Mole decided that, between the two analyses, the number of cancer deaths rose by only 11% from 670 to 743. However, these two figures have different bases. The smaller figure included 228 men who were either not monitored for external radiation or recorded zero doses. The larger figure was the number of monitored cases (some with zero doses) in the second analysis, by which time the total number of cancer deaths was 924. Thus the true increase was 38%. In the first analysis2 we were not in a position to distinguish between non-monitoring and zero dose. In the second, this was possible, so we excluded 181 cancer deaths of non-monitored workers. Since a third of the cancers in the first analysis were either not exposed or recorded zero dose, a distribution of our doubling-dose estimates would have been very anomalous, so we did not give any confidence limits. Once non-exposure and zero dose were distinguished quotation of confidence limits was justified although there was still a very wide gap between the upper and lower levels. Therefore the apparent anomalies noted by Mole are not at all surprising. Mole and others put great stress on the possibility that estimates of dose effects from occupational exposure might be biased because exposure period and cumulative dose are correlated. In our second analysis3 we included exposure period as one of several controlling factors in a Mantel-Haenszel analysis. If exposure period had been important the radiation effect would have been much smaller in the controlled than the crude analysis, but it was not. The difference between our results and those of Marks et al.4 (who examined Hanford data for the period 1944-74) is more a matter of interpretation and emphasis than might be gathered from Mole’s letter. We uncovered three cancers with definite radiation effects (bone-marrow, pancreas, and lung) while the approach of Marks et al., which included a strongly controlled analysis, revealed a definite association between radiation dose and multiple myeloma and pancreatic cancer. The significance levels were such that the two findings could not both be statistical flukes. Marks et al. postulate, as a cause of the association, exposure to unidentified chemicals at the same time as radiation, and we are suggesting that the association is a direct one. Can our low risk estimates, in the form of doubling doses, be reconciled with what is known about background radiation and what is widely assumed to be the role of radiation in carcinogenesis ? Our lowest estimate of doubling dose is 3.6rad for bone-marrow cancers, and the background radiation dose is approximately 0.1 rad per annum. It is generally agreed that the number of radiogenic cancers is proportional to the total radiation dose, provided (a) all ionising radiations are sufficiently similar for similar values of doubling dose to apply, (b) 1. Mole, R. Lancet, 1978, i, 1155. 2. Mancuso, T. F., Stewart, A. M., Kneale, G. Hlth Phys. 1977, 33, 369. 3. Kneale, G., Stewart, A. M., Mancuso, T. F. Paper IAEA-SM-224-510 read at I.A.E.A. symposium on the Late Biological Effects of Ionising Radiation, held in Vienna on March 13-17, 1978. 4. Marks, S., Gilbert, E. S., Brettenstein, B. D. ibid. paper IAEA-SM-224-509.
different constitutions react differently to different diseases carriers of the gene for xeroderma pigmentosa, even in a recessive form, are predisposed to skin cancer). Therefore we decided that the second basic assumption (b) was probably invalid. This conclusion prepared us for the possibility of nonlinearity of the dose-response curve (at a population level) since high doses would saturate a small proportion of highly sensitive individuals and leave the rest of the population with a different level of response. Another version of unequal sensitivity would be expected if carcinogens interact, and several surveys suggest that in relation to lung cancer the combined effects of smoking and radiation are multiplicative rather than additive. Finally the third basic assumption of equal sensitivity to the cancer induction effects of radiation at all ages was most unlikely since the immune system is clearly playing an important role in the aetiology of all diseases, including cancers. Also the first and second analysis of Hanford data contain evidence which show that assumption (c) is invalid, and a similar conclusion can be drawn from an earlier analysis of data from the Oxford Survey of Childhood Cancers which showed a 16-fold decrease in sensitivity between the first and third trimesters of fetal life.5 Comparisons between observed and expected cancer deaths will eventually show whether there is a cancer hazard for workers in the nuclear industry. However, to be certain of success, this approach requires either much larger doses than were encountered in the Hanford study (where the mean cumulative dose was only 1-38 rad for cancers and 0.99 for non-cancers)
(e.g.,
much larger data base (say, 500000 man-years of follow-up instead of 150 000 man-years3). Therefore, unlike Dr Mole and Professor Anderson (July 15, p. 161), we attach more importance to the fact that the mean cumulative dose was higher for 22 bone-marrow cancers (4-49 rad) than for 42 other reticuloendothelial neoplasms (0-99 rad) than to the fact that
or a
5.
Kneale, G., Stewart, A. M.J. nath. Cancer Inst. 1976, 57, 1009.
