- Email: [email protected]
CANCER NEAR NUCLEAR INSTALLATIONS
855
Letters
to
included in the inner zone. Figures are also given for the sum of the inner and outer zones, giving the broad cumulative zone to which most consideration was given in the discussion chapter of the OPCS report. Probability levels are marked to indicate the statistical significance of deviations from the standard level and of differences between the SRRs/SMRs for the installation areas and the control
the Editor
CANCER NEAR NUCLEAR INSTALLATIONS
SIR,-Dr Beral (March 7,
p 556) has presented calculations, data from an Office of Population Censuses and Surveys (OPCS) report,l which highlight differences between incidence and mortality rates in the vicinity of nuclear installations other than Sellafield in England and Wales. Both for leukaemia and for all cancers at ages 0-24 standardised registration ratios (SRR) were above 100 for the installation areas but not for the control areas-namely, 111 (p. < 005) compared with 97 for leukaemia and 108 (p<0’01) compared with 100) for all cancers. Standardised mortality ratios (SMR) showed no such difference (102 vs 106 and 99 vs 98, respectively). ’ The discussion chapter in the OPCS report draws attention to the raised registrations:deaths ratios that had been noted during preliminary analyses for districts in the vicinity of nuclear installations set up before 1955 (Sellafield, Springfields, Capenhurst, Amersham, Harwell, and Aldermaston). It was noted that such an effect could be due to local variation in the efficiency of cancer registration. Since 81 % of the population defmed as living close to nuclear installations were resident in the vicinity of those pre-1955 installations, any such bias will affect the whole data set. Beral suggests that there could be other reasons for the difference between incidence and the mortality ratios. Further detail is, therefore, given here of the results that suggested to us1 that a registration effect is the most likely explanation. The accompanying table gives SRRs and SMRs (based on expected values derived from regional age/sex specific rates) for certain categories of malignancy. Figures are given for installation areas and for their control areas for the pre-1955 installations (excluding Sellafield) and for Central Electricity Generating Board (CEGB) installations as a group. The data cover the total period during which cancer statistics have been considered subsequent to the start-up of each installation, and two discrete distance zones. These two zones are an inner one, including all districts with at least two-thirds of their population living within 8 miles of an installation, and an outer zone including all districts with at least one-third of their population resident within 10 miles except those
based
on
STANDARDISED REGISTRATION RATIOS ’
areas.
For the inner distance zone in the vicinity of the pre-1955 installations SRRs are above 100 and above the control SRRs at all ages and for all categories of malignancy considered, apart from cancer of the lung at age 25-74. No such consistent effect is apparent in the mortality figures for the inner distance zone nor in any of the figures for the outer distance zone. When figures for the two distance zones are added together the pattern for the inner zone tends to be reflected in the SRRs for the combined zones. In particular, the SRR for leukaemia and the SRR for "other malignancy" at all ages are above both regional level and control levels. The lung cancer SRR at ages 25-74 for the installation areas in the zones combined is below that for the control areas. For the CEGB installations there is no consistent pattern of increase for either SRRs or SMRs in the installation areas for either distance zone. Thus for the inner distance zone around the pre-1955 installations there is a 10-20 % increase in SRR for almost all the types of cancer considered across all age groups that is not apparent in the mortality figures. This suggests local variation in standards of cancer
registration.
Difficulties were encountered in matching installation and control areas from within the same cancer registration region (a point discussed in chapter 4 of the OPCS report). However, increased ratios were found for installations where matching was entirely from within the same registration region, and in these instances it would seem that there must be local variation in registration practice. A reason for the apparently discrepant registration patterns could be reflected in the results for cancer of the lung at ages 25-74. For the inner distance zone around the pre-1955 installations the mortality ratio is significantly below both the regional level and the control level. National mortality figures for cancer of the lung show an inverse association of rate with social classy and the results for lung cancer could suggest that districts in the vicinity of the grouped pre-1955 installations are of higher socioeconomic status than either
registration:deaths
(SRR) AND STANDARDISED MORTALITY RATIOS (SMR) IN VICINITY OF NUCLEAR INSTALLATIONS (I) (C) FOR INNER, OUTER, AND COMBINED ZONES (SEE TEXT)
AND IN CONTROL AREAS
Local areas higher lower than region used as standard: ’0-01 > P >005, Installation area higher than control area or vice versa.’001 >p>005,
’0
05
>p>0-01,
’0 01
> P > 0’001,
°005>p>001,’001>p>0001,p<0001-
"p < 0.001
I
856 their control districts or the districts which make up the regions used as standards. Such differences are marginally apparent in information on social class structure set out in appendix 4 of the OPCS report. Perhaps cancer registration is more efficient in more prosperous localities. Beral suggests that the excess incidence rates at young ages in the vicinity of nuclear installations could also indicate better survival rates. It is difficult to envisage, however, that better survival rates could be equally responsible for the 10-20 % increase of incidence that is so generally apparent at different ages and for different types of cancer for the inner distance zones around the pre-1955
installations. A further suggestion from Beral is that patients may have moved away from nuclear installations once their cancer had been diagnosed, and we are now examining this possibility. Her suggestion that an increase in incidence might be too recent to be reflected in the mortality figures seems not to be supported by Beral’s own figures: raised SRRs in installation areas are evident for all the time periods studied. P. C.-M. is
a
£ member of the Medical Research Council’s external staff.
ICRF Cancer Epidemiology Unit, Gibson Laboratories, Radcliffe Infirmary, Oxford OX2 6HE
impossible. The NRPB claims that the countermeasures adopted in West Germany will have reduced the thyroid dose by 36%, but this should not dissuade epidemiologists from looking.
it is
not
In East Europe and Soviet Russia, individual doses will have been much higher than those in West Germany, though reliable figures have not yet been forthcoming. At the IAEA meeting in Vienna the Russians calculated a collective dose of 2million man-sieverts to the population of Soviet Europe from the releases of caesium-137 only.’ On this basis, even the ICRP risk estimates predict an extra 25 000 fatal cancers, though the total cancer detriment is closer to 200 000, for the reasons outlined already. The ICRP has been reluctant to accept the Russian estimate of dose, and tried to reduce it by a factor of 10 at the Vienna meeting. The scientific basis for this attempt has yet to be produced. What is clear is that the ICRP estimate for both radiation dose and cancer risk are at the bottom end of a range of values which, multiplied together, produce a figure two orders of magnitude greater than that allowed by ICRP. This is a strange situation for an International Commission which claims always to "err on the side of caution". The Old Cottage, Wexham Street, Stoke Poges, Bucks SL3 6NB
PAULA COOK-MOZAFFARI
ROBIN RUSSELL JONES, Chairman, Fnends of the Earth Pollution Advisory Committee
Morrey M, Brown J, Williams J, et al. A preliminary assessment of the radiological impact of the Chernobyl reactor accident on the population of the European Community. Chilcot, Oxon: NRPB, 1987. 2. Committee on Biological Effects of Ionising Radiation. Beir III Committee. The 1.
PJ, Ashwood FL, Vincent T, Forman D, Alderson M. Cancer incidence and mortality in the vicinity of nuclear installations. England and Wales, 1959-80 (Stud Med Popul Subj no 51). London: HM Stationery Office, 1987. 2. Office of Population, Censuses and Surveys. Occupational mortality: the Registrar General’s decennial supplement for England and Wales, 1970-72 (series DS no 1). London HM Stationery Office, 1978. 1. Cook-Mozaffari
CHERNOBYL AND CANCER EPIDEMIOLOGY
SiR,—Your note (March 28, p 758) cites an estimate by the National Radiological Protection Board (NRPB) of 1000 extra cancer deaths within the EEC population as a result of the Chernobyl accident.’ Your account echoes the view of the secretary of NRPB, Roger Clarke, who said at the International Conference on the Biological Effects of Ionising Radiation, held at Hammersmith Hospital, London, on Nov 25, 1986: "The conclusion I want to draw here is that we shall never see any effects on Chernobyl from any epidemiological studies that might be established in Western Europe." Whilst the idea that Chernobyl cancers are undetectable will be welcomed by the nuclear industry, the NRPB position is more difficult to justify on scientific grounds. The cancer risk estimates used by NRPB are based mainly on the old atomic bomb data, even though the dosimetry and health effects are now undergoing extensive revision. The NRPB used the figure of 125 cancer deaths per 10 000 man-sievert because this is based on a worldwide scientific consensus". This is untrue. In 1980, the BEIR III Committee produced an upper estimate of 501 cancer deaths per 10 000 man-sievertand if UNSCEAR data are analysed without the atomic bomb data an upper figure of 440 cancer deaths is produced.3 Last year the NRPB’s epidemiological consultant, Sir Richard Doll, stated that the ICRP risk estimate for fatal cancer is too low by a factor of 2, and since then the most recent data from Japan give a cancer doubling dose of 133 rem at doses below 20 rem.’ This translates into a risk estimate for all cancers (excluding skin) of at least 1000 per man-sievert, eight times the ICRP estimate for fatal cancer. These calculations are consistent with the data from studies of occupationally exposed workers.5,6** In the EEC population 8000 extra cancers will be no easier to detect than 1000, but the situation may be different in individual EEC countries such as West Germany, and it will be very different in Eastern European countries and Soviet Russia. In West Germany the NRPB calculate a collective population dose of 30 000 man-sieverts (whole body), which is rather more than one-third of the total EEC dose of 78 000 man-sieverts. Without countermeasures, the thyroid dose in West Germany would have been 82 000 man-sieverts. Although the NRPB use a thyroid cancer risk estimate of 100 per 10 000 Sv, 300 seems more appropriate. This would produce almost 2500 extra thyroid cancers against an expected number of 30 000 to 40 000, an increase of 5-10%. Whilst this might be difficult to detect epidemiologically,
effects of populations of exposures to low levels of ionising radiation. Washington, DC: National Academy of Sciences, 1980. 3. Charles MW, Lindop PJ. Risk assessment without the bombs.J Soc Radiol Prot 1981; 1: 15-19. 4. Radford EP. Recent evidence of radiation induced cancer in the Japanese atomic bomb survivors. In: Russel Jones R, Southwood R, eds. Radiation and health the biological effects of ionising radiation at low levels of exposure. Chichester: John Wiley and Sons, 1987. 5. Beral V, Inskip H, Fraser P, Booth M, Coleman O, Rose G. Mortality of employees of the United Kingdom Atomic Energy Authority 1946-1979. Br Med J 1985; 291: 440-47. 6. Mancuso T, Stewart A, Kneale G. Radiation of Hanford workers dying of cancer and other causes. Health Phys 1977; 33: 369-85. 7. Anno. Tracking radiation releases. Nature 1986; 323: 29.
EPIDURAL BLOOD PATCH DOES NOT PREVENT HEADACHE AFTER LUMBAR PUNCTURE
SIR,-Headache after lumbar puncture probably results from leakage of cerebrospinal fluid (CSF) from the puncture hole in the dura.1,2 This explanation accounts for many of the clinical features of this headache, including the ability of epidural blood patching (EBP) to relieve the headache.1,2 The EBP technique, first described by Gormley in 1960involves insertion of a spinal needle at the site of the lumbar puncture and injection of a small volume of the patient’s blood. EBP usually relieves headache after lumbar puncture, but does it prevent such headaches too?2,1While studying CSF neuropeptides in manic-depressive patients and healthy volunteers5 we investigated the prophylactic value of EBP. Lumbar puncture was done in 41 healthy volunteers and in 32 euthymic bipolar patients as an outpatient procedure.5 20 G needles were used. 30 ml CSF was withdrawn with the subject in the lateral decubitus position after 2 h of bedrest. Bedrest continued for about 6 h after the procedure, when the subject went home. For the first hour after the lumbar puncture, subjects were kept in reverse Trendelenberg position, lying prone. After that food and water were offered. Subjects were advised of the risk of headache and were asked to remain in bed and to contact us if a headache developed. The characteristics of post lumbar puncture headache were not described. A prophylactic EBP was done on a consecutive series of 16 subjects (9 healthy volunteers, 7 bipolar patients). After CSF had been collected the needle was withdrawn slowly until CSF flow ceased. The needle was withdrawn an additional 1 mm to ensure that the needle bevel was completely outside the subarachnoid space. About 3 ml blood (drawn from an arm vein into a polypropylene syringe at the end of the CSF collection) was injected slowly through the spinal needle. Then the needle was removed. All other procedures were identical for the two groups.
Letters
to
included in the inner zone. Figures are also given for the sum of the inner and outer zones, giving the broad cumulative zone to which most consideration was given in the discussion chapter of the OPCS report. Probability levels are marked to indicate the statistical significance of deviations from the standard level and of differences between the SRRs/SMRs for the installation areas and the control
the Editor
CANCER NEAR NUCLEAR INSTALLATIONS
SIR,-Dr Beral (March 7,
p 556) has presented calculations, data from an Office of Population Censuses and Surveys (OPCS) report,l which highlight differences between incidence and mortality rates in the vicinity of nuclear installations other than Sellafield in England and Wales. Both for leukaemia and for all cancers at ages 0-24 standardised registration ratios (SRR) were above 100 for the installation areas but not for the control areas-namely, 111 (p. < 005) compared with 97 for leukaemia and 108 (p<0’01) compared with 100) for all cancers. Standardised mortality ratios (SMR) showed no such difference (102 vs 106 and 99 vs 98, respectively). ’ The discussion chapter in the OPCS report draws attention to the raised registrations:deaths ratios that had been noted during preliminary analyses for districts in the vicinity of nuclear installations set up before 1955 (Sellafield, Springfields, Capenhurst, Amersham, Harwell, and Aldermaston). It was noted that such an effect could be due to local variation in the efficiency of cancer registration. Since 81 % of the population defmed as living close to nuclear installations were resident in the vicinity of those pre-1955 installations, any such bias will affect the whole data set. Beral suggests that there could be other reasons for the difference between incidence and the mortality ratios. Further detail is, therefore, given here of the results that suggested to us1 that a registration effect is the most likely explanation. The accompanying table gives SRRs and SMRs (based on expected values derived from regional age/sex specific rates) for certain categories of malignancy. Figures are given for installation areas and for their control areas for the pre-1955 installations (excluding Sellafield) and for Central Electricity Generating Board (CEGB) installations as a group. The data cover the total period during which cancer statistics have been considered subsequent to the start-up of each installation, and two discrete distance zones. These two zones are an inner one, including all districts with at least two-thirds of their population living within 8 miles of an installation, and an outer zone including all districts with at least one-third of their population resident within 10 miles except those
based
on
STANDARDISED REGISTRATION RATIOS ’
areas.
For the inner distance zone in the vicinity of the pre-1955 installations SRRs are above 100 and above the control SRRs at all ages and for all categories of malignancy considered, apart from cancer of the lung at age 25-74. No such consistent effect is apparent in the mortality figures for the inner distance zone nor in any of the figures for the outer distance zone. When figures for the two distance zones are added together the pattern for the inner zone tends to be reflected in the SRRs for the combined zones. In particular, the SRR for leukaemia and the SRR for "other malignancy" at all ages are above both regional level and control levels. The lung cancer SRR at ages 25-74 for the installation areas in the zones combined is below that for the control areas. For the CEGB installations there is no consistent pattern of increase for either SRRs or SMRs in the installation areas for either distance zone. Thus for the inner distance zone around the pre-1955 installations there is a 10-20 % increase in SRR for almost all the types of cancer considered across all age groups that is not apparent in the mortality figures. This suggests local variation in standards of cancer
registration.
Difficulties were encountered in matching installation and control areas from within the same cancer registration region (a point discussed in chapter 4 of the OPCS report). However, increased ratios were found for installations where matching was entirely from within the same registration region, and in these instances it would seem that there must be local variation in registration practice. A reason for the apparently discrepant registration patterns could be reflected in the results for cancer of the lung at ages 25-74. For the inner distance zone around the pre-1955 installations the mortality ratio is significantly below both the regional level and the control level. National mortality figures for cancer of the lung show an inverse association of rate with social classy and the results for lung cancer could suggest that districts in the vicinity of the grouped pre-1955 installations are of higher socioeconomic status than either
registration:deaths
(SRR) AND STANDARDISED MORTALITY RATIOS (SMR) IN VICINITY OF NUCLEAR INSTALLATIONS (I) (C) FOR INNER, OUTER, AND COMBINED ZONES (SEE TEXT)
AND IN CONTROL AREAS
Local areas higher lower than region used as standard: ’0-01 > P >005, Installation area higher than control area or vice versa.’001 >p>005,
’0
05
>p>0-01,
’0 01
> P > 0’001,
°005>p>001,’001>p>0001,p<0001-
"p < 0.001
I
856 their control districts or the districts which make up the regions used as standards. Such differences are marginally apparent in information on social class structure set out in appendix 4 of the OPCS report. Perhaps cancer registration is more efficient in more prosperous localities. Beral suggests that the excess incidence rates at young ages in the vicinity of nuclear installations could also indicate better survival rates. It is difficult to envisage, however, that better survival rates could be equally responsible for the 10-20 % increase of incidence that is so generally apparent at different ages and for different types of cancer for the inner distance zones around the pre-1955
installations. A further suggestion from Beral is that patients may have moved away from nuclear installations once their cancer had been diagnosed, and we are now examining this possibility. Her suggestion that an increase in incidence might be too recent to be reflected in the mortality figures seems not to be supported by Beral’s own figures: raised SRRs in installation areas are evident for all the time periods studied. P. C.-M. is
a
£ member of the Medical Research Council’s external staff.
ICRF Cancer Epidemiology Unit, Gibson Laboratories, Radcliffe Infirmary, Oxford OX2 6HE
impossible. The NRPB claims that the countermeasures adopted in West Germany will have reduced the thyroid dose by 36%, but this should not dissuade epidemiologists from looking.
it is
not
In East Europe and Soviet Russia, individual doses will have been much higher than those in West Germany, though reliable figures have not yet been forthcoming. At the IAEA meeting in Vienna the Russians calculated a collective dose of 2million man-sieverts to the population of Soviet Europe from the releases of caesium-137 only.’ On this basis, even the ICRP risk estimates predict an extra 25 000 fatal cancers, though the total cancer detriment is closer to 200 000, for the reasons outlined already. The ICRP has been reluctant to accept the Russian estimate of dose, and tried to reduce it by a factor of 10 at the Vienna meeting. The scientific basis for this attempt has yet to be produced. What is clear is that the ICRP estimate for both radiation dose and cancer risk are at the bottom end of a range of values which, multiplied together, produce a figure two orders of magnitude greater than that allowed by ICRP. This is a strange situation for an International Commission which claims always to "err on the side of caution". The Old Cottage, Wexham Street, Stoke Poges, Bucks SL3 6NB
PAULA COOK-MOZAFFARI
ROBIN RUSSELL JONES, Chairman, Fnends of the Earth Pollution Advisory Committee
Morrey M, Brown J, Williams J, et al. A preliminary assessment of the radiological impact of the Chernobyl reactor accident on the population of the European Community. Chilcot, Oxon: NRPB, 1987. 2. Committee on Biological Effects of Ionising Radiation. Beir III Committee. The 1.
PJ, Ashwood FL, Vincent T, Forman D, Alderson M. Cancer incidence and mortality in the vicinity of nuclear installations. England and Wales, 1959-80 (Stud Med Popul Subj no 51). London: HM Stationery Office, 1987. 2. Office of Population, Censuses and Surveys. Occupational mortality: the Registrar General’s decennial supplement for England and Wales, 1970-72 (series DS no 1). London HM Stationery Office, 1978. 1. Cook-Mozaffari
CHERNOBYL AND CANCER EPIDEMIOLOGY
SiR,—Your note (March 28, p 758) cites an estimate by the National Radiological Protection Board (NRPB) of 1000 extra cancer deaths within the EEC population as a result of the Chernobyl accident.’ Your account echoes the view of the secretary of NRPB, Roger Clarke, who said at the International Conference on the Biological Effects of Ionising Radiation, held at Hammersmith Hospital, London, on Nov 25, 1986: "The conclusion I want to draw here is that we shall never see any effects on Chernobyl from any epidemiological studies that might be established in Western Europe." Whilst the idea that Chernobyl cancers are undetectable will be welcomed by the nuclear industry, the NRPB position is more difficult to justify on scientific grounds. The cancer risk estimates used by NRPB are based mainly on the old atomic bomb data, even though the dosimetry and health effects are now undergoing extensive revision. The NRPB used the figure of 125 cancer deaths per 10 000 man-sievert because this is based on a worldwide scientific consensus". This is untrue. In 1980, the BEIR III Committee produced an upper estimate of 501 cancer deaths per 10 000 man-sievertand if UNSCEAR data are analysed without the atomic bomb data an upper figure of 440 cancer deaths is produced.3 Last year the NRPB’s epidemiological consultant, Sir Richard Doll, stated that the ICRP risk estimate for fatal cancer is too low by a factor of 2, and since then the most recent data from Japan give a cancer doubling dose of 133 rem at doses below 20 rem.’ This translates into a risk estimate for all cancers (excluding skin) of at least 1000 per man-sievert, eight times the ICRP estimate for fatal cancer. These calculations are consistent with the data from studies of occupationally exposed workers.5,6** In the EEC population 8000 extra cancers will be no easier to detect than 1000, but the situation may be different in individual EEC countries such as West Germany, and it will be very different in Eastern European countries and Soviet Russia. In West Germany the NRPB calculate a collective population dose of 30 000 man-sieverts (whole body), which is rather more than one-third of the total EEC dose of 78 000 man-sieverts. Without countermeasures, the thyroid dose in West Germany would have been 82 000 man-sieverts. Although the NRPB use a thyroid cancer risk estimate of 100 per 10 000 Sv, 300 seems more appropriate. This would produce almost 2500 extra thyroid cancers against an expected number of 30 000 to 40 000, an increase of 5-10%. Whilst this might be difficult to detect epidemiologically,
effects of populations of exposures to low levels of ionising radiation. Washington, DC: National Academy of Sciences, 1980. 3. Charles MW, Lindop PJ. Risk assessment without the bombs.J Soc Radiol Prot 1981; 1: 15-19. 4. Radford EP. Recent evidence of radiation induced cancer in the Japanese atomic bomb survivors. In: Russel Jones R, Southwood R, eds. Radiation and health the biological effects of ionising radiation at low levels of exposure. Chichester: John Wiley and Sons, 1987. 5. Beral V, Inskip H, Fraser P, Booth M, Coleman O, Rose G. Mortality of employees of the United Kingdom Atomic Energy Authority 1946-1979. Br Med J 1985; 291: 440-47. 6. Mancuso T, Stewart A, Kneale G. Radiation of Hanford workers dying of cancer and other causes. Health Phys 1977; 33: 369-85. 7. Anno. Tracking radiation releases. Nature 1986; 323: 29.
EPIDURAL BLOOD PATCH DOES NOT PREVENT HEADACHE AFTER LUMBAR PUNCTURE
SIR,-Headache after lumbar puncture probably results from leakage of cerebrospinal fluid (CSF) from the puncture hole in the dura.1,2 This explanation accounts for many of the clinical features of this headache, including the ability of epidural blood patching (EBP) to relieve the headache.1,2 The EBP technique, first described by Gormley in 1960involves insertion of a spinal needle at the site of the lumbar puncture and injection of a small volume of the patient’s blood. EBP usually relieves headache after lumbar puncture, but does it prevent such headaches too?2,1While studying CSF neuropeptides in manic-depressive patients and healthy volunteers5 we investigated the prophylactic value of EBP. Lumbar puncture was done in 41 healthy volunteers and in 32 euthymic bipolar patients as an outpatient procedure.5 20 G needles were used. 30 ml CSF was withdrawn with the subject in the lateral decubitus position after 2 h of bedrest. Bedrest continued for about 6 h after the procedure, when the subject went home. For the first hour after the lumbar puncture, subjects were kept in reverse Trendelenberg position, lying prone. After that food and water were offered. Subjects were advised of the risk of headache and were asked to remain in bed and to contact us if a headache developed. The characteristics of post lumbar puncture headache were not described. A prophylactic EBP was done on a consecutive series of 16 subjects (9 healthy volunteers, 7 bipolar patients). After CSF had been collected the needle was withdrawn slowly until CSF flow ceased. The needle was withdrawn an additional 1 mm to ensure that the needle bevel was completely outside the subarachnoid space. About 3 ml blood (drawn from an arm vein into a polypropylene syringe at the end of the CSF collection) was injected slowly through the spinal needle. Then the needle was removed. All other procedures were identical for the two groups.