65
faecal extract from volunteers infected with MS1. The technique is informative and precise, but can hardly be described as quantitative-or even semiquantitative-so that quantitative assessment of rises in antibody titre is not really practicable. The findings in the 22 patients studied here are entirely negative for this hepatitis-A antigen-i.e., there were no seroconversions. The apparent discrepancy between the relative
A, and the material used
was a
proportions of non-hepatitis-B post-transfusion hepatitis and the distribution of viruses in this group, in different countries, is at present unexplained. Of course, geographical differences may apply-i.e., there may be an agent or agents at large in the U.S.A. but not in the U.K. However, before a hypothetical hepatitis-C virus is produced, a touch of Occam’s razor is not inappropriate. These patients might have a hepatitis A differing antigenically from the MSI of FEINSTONE et al. Morphologically, the supposed MS1 particle is like an enterovirus,l1 and the major groups of enteroviruses are notoriously divided into distinct serological subtypes, particularly the members of the ECHO and Coxsackie virus groups. On more certain ground in terms of hepatitis-A antigen, and contributing a real advance in technique, is the work of DEINHARDT, HILLEMAN, and others. DEINHARDT 13 and HILLEMAN 14 and co-workers have transferred strains of human hepatitis A to marmosets, and complement-fixation 15 and immuneadherence tests 16 have been developed, using extracts of infected marmoset liver as antigen. These tests KRUGMAN are sensitive, quantitative, and specific. and others 17 have applied both these techniques to patients who have had MS1 hepatitis A and they have shown seroconversion to this antigen. It will be of considerable interest to apply these two tests to patients with post-transfusion hepatitis. Meanwhile, it would be a great pity to disregard the contribution made by the screening of blood for HBAg and the exclusion of HBAg-positive donations. This has made big inroads into post-transfusion hepatitis 4,5 and has helped in the almost complete clearing of hepatitis from haemodialysis units in Great Britain.18,19 Nevertheless, there is still, at least in the U.S.A., a sizable and somewhat puzzling hepatitis gap to be plugged. The implication or exoneration of hepatitis A cannot, at the present rate of progress, be far off. 13.
Deinhardt, F., Holmes, A. W., Capps, R. B., Popper, H. J. J. exp. Med. 1967, 125, 673. 14. Provost, P. J., Ittensohn, O. L., Villarejos, V. M., Arguedas, G. J. A., Hilleman, M. R. Proc. Soc. exp. Biol. Med. 1973, 142, 1257. 15. Provost, P. J., Ittensohn, O. L., Villarejos, V. M., Hilleman, R. M. ibid. 1975, 148, 962. 16. Provost, P. J., Wolanski, B. S., Miller, W. J., Ittensohn, O. L., McAleer, W. J., Hilleman, M. R. ibid. p. 532. 17. Krugman, S., Friedman, H., Lattimer, C. New Engl. J. Med. 1975, 292, 1141. 18. Polakoff, S. (Public Health Laboratory Service Survey) Br. med. J. 1974, iv, 751. 19. Marmion, B. P., Tonkin, R. W. Br. med. Bull. 1972, 28, 169.
PREVENTION
OF
HÆMOPHILUS
INFLUENZÆ
MENINGITIS INFECTION with a pathogenic microbe causes illness in some persons whereas others remain well. Their immunity may be due to their good fortune in having at some time carried in their microbial flora avirulent bacteria with antigens similar to those of the pathogen. Such natural immunity may be induced by poorly virulent strains of the pathogenic species (as may happen with immunity to meningococcal disease 1,2) but also by completely unrelated organisms such as the coliform bacilli of the normal intestinal flora. 3,Thus the surface antigens of certain serotypes of Haemophilus, Meningococcus, and Pneumococcus may cross-react immunologically with antigens of other bacteria, including Escherichia coli.55 Attenuated strains of virulent microorganisms have been used for vaccination of man and animals since the days of Pasteur, and the possibility is now suggested that natural immunisation by the commensal body flora could also be exploited-by feeding suitable strains of E. coli to immunise against H. infiuenze meningitis.4,6-8
Schneerson and Robbins8 have reported that the feeding to adult volunteers of E. coli with the antigenic structure 075 :K100 :H5 caused an increase in the amount of serum-antibody against the capsular antigen of H. influenza type b. Earlier, suitable strains of E. coli had been found by culturing rectal swabs on an agar medium containing antiserum against the H. infiuenzae type b capsular antigen; a visible halo of antigen-antibody precipitate formed around colonies of bacteria with a similar antigen.4 The E. coli capsular antigen, designated K100, has since been shown to be immunochemically similar to the polysaccharide capsule of H. influenzce type b.9 The E. coli were fed to 16 healthy adult volunteers before breakfast, after taking 2 g. of sodium bicarbonate, in a dose of about 107-108 bacteria. The strain could be recovered from their faeces for three to eight weeks afterwards, and the volunteers responded, somewhat variably, with the development of antibodies bactericidal for H. infiuenzae type b. The feeding of young babies with the E. coli might well immunise them against Hcemophilus meningitis, since almost all strains of H. influenza from cases of meningitis have a type b capsule, 10 and there is evidence that the corresponding antibody is 1.
protective. 11, 12 Meningitis is
most
liable
to
develop
Goldschneider, I., Gotschlich, E. C., Artenstein, M. S. J. exp. Med.
1969, 129, 1327. Reller, L. B., MacGregor, R. R., Beaty, H. N. J. infect. Dis. 1973, 127, 56. 3. Wilson, G. S., Miles, A. A. Topley and Wilson’s Principles of Bacteriology, Virology and Immunology; chap. 41. London, 1975. 4. Bradshaw, M. W., Schneerson, R., Parke, J. C., Jr., Robbins, J. B. Lancet, 1971, i, 1095. 5. Robbins, J. B., Myerowitz, R. L., Whisnant, J. K., Argaman, M., Schneerson, R., Handzel, Z. T., Gotschlich, E. C. Infect. Immun. 1972, 6, 651. 6. Myerowitz, R. L., Handzel, Z. T., Schneerson, R., Robbins, J. B. ibid. 1973, 7, 137. 7. Br. med. J. 1974, ii, 462. 8. Schneerson, R., Robbins, J. B. New Engl. J. Med. 1975, 292, 1093. 9. Schneerson, R., Bradshaw, M., Whisnant, J. K., Myerowitz, R. L., Parke, J. C., Jr., Robbins, J. B. J. Immun. 1972, 108, 1551. 10. Turk, D. C., May, J. R. Hæmophilus Influenza: Its Clinical Importance. London, 1967. 11. Alexander, H. E., Heidelberger, M., Leidy, G. Yale J. Biol. Med. 1944, 16, 425. 12. Schneerson, R., Rodrigues, L. P., Parke, J. C., Jr., Robbins, J. B. J. Immun. 1971, 107, 1081. 2.
66
antibody and before they develop their own natural antibodies against the capsular antigen that is, between the ages of about 3 months and 5 years-and oral vaccination at the age of 3-6 months might overcome this gap in immunity. The strains of E. coli used have shown no evidence of enteropathogenicity, nor do they appear to be invasive. They are not entirely avirulent-they may be able to cause urinary-tract infection, for examplebut they do not appear to be any different in this respect from many other strains of E. coli that are to be found in large numbers in the normal human intestine.9 The investigators have not yet tried to immunise babies in this way; but the procedure has been shown to work in baby rabbits,and E. coli strains have been safely fed to newborn babies by workers in Czechoslovakia, who also showed that an antibody response developed when the intestine was colonised.13 Since babies are bound to be colonised by similar organisms quite naturally, the process should be safe provided the strains do not cause diarrhoea or other illness. The incidence of Haemophilus meningitis can be estimated only roughly. About 1500 cases of acute meningitis are notified in England and Wales to the Registrar General each year, and returns to the Public Health Laboratory Service suggest that about 20-25% of laboratory-diagnosed cases are due to H. influenzæ.14,15 If it is accepted that meningitis is undernotified by about 50%, there may be 800 cases each year in England and Wales. Since the disease mostly affects children under 5 years of age, perhaps 1 child in 1000 is liable to contract the disease before the 5th birthday. The mortality-rate may be as high as 10%, and the incidence of sequelse, which can include impaired intelligence, behaviour disorders, or hydrocephalus, has been reported to be 30% in the U.S.A.16 and 22% in Sweden.l’ Thus Haemophilus meningitis, though not common, is a serious illness, and the use of a vaccine given simply by mouth might be worth while provided it could be shown to be both safe and protective. Proof of safety may be obtainable by careful clinical trials, first in small groups and later in larger; but efficacy may be hard to establish since the disease has such a low incidence. Babies may not respond to the vaccination so well as immunologically mature adults,18 most of whom already possess some antibody against the shared antigen, and susceptibility to Haemophilus infection may be influenced by factors other than circulating antibody. 19. 20But the exploitation of what is probably the natural process by which infants become immune to this serious infection might prove a valuable and economical approach to infectious-disease control, with possibilities for other pathogens-meningococci and pneumococci, for example-that share antigens with bacteria of the commensal flora. 13. Lodinova, R., Jouja, V., Lanc, A. J. Bact. 1967, 93, 797. 14. P.H.L.S. Br. med. J. 1974, i, 453. 15. P.H.L.S. ibid. 1973, i, 623. 16. Sell, S. H. W., Merrill, R. E., Doyne, E. O., Zimsky, E. P., Jr. Pediatrics, Springfield, 1972, 49, 206. 17. Jonsson, M., Alvin, A. Scand. J. infect. Dis. 1971, 3, 141. 18. Smith, D. H., Peter, G., Ingram, D. L., Harding, A. L., Anderson, P. Pediatrics, Springfield, 1973, 52, 637. 19. Feigin, R. D., Richmond, D., Hosler, M. W., Shackelford, P. G. Am. J. med. Sci. 1971, 262, 338. 20. Lancet, 1971, ii, 914.
INTRAVENOUS UROGRAPHY IN HYPERTENSION THE morbidity and mortality associated with a high blood-pressure are a considerable incentive to clinicians to try to cure the hypertension by finding a surgically correctable abnormality of the renal, vascular, or endocrine system. Today’s drugs, however, can now control the blood-pressure of a large proportion of hypertensive patients and so the more elaborate investigations, such as renal arteriography or divided renal-function tests, are reserved for the occasional patient. Intravenous urography, however, is normally recommended as a routine investigation for those who 1 may need treatment. A greater awareness of hypertension and organised screening programmes 2-4are leading to the detection of an increasing number of patients with mild hypertension. Can we justify the cost and the inconvenience and risk to the patient of intravenous urography in all these cases ? Atkinson and Kellett5 made a retrospective analysis of the value of intravenous urography in 952 patients attending the Glasgow blood-pressure clinic. Although 120 (17-6%) renal abnormalities were demonstrated, these led to a change in patient management in only 9 cases (0-9%). A few of their patients, however, had been followed up for as little as four months, and so it is reassuring to see their results confirmed by Dr Bailey and others (on p. 57 this week). They followed up a group of patients for at least three years. In 80 patients, urography revealed 15 renal abnormalities (19%), but in only 1 was patient management changed by unexpected urographic findings. This patient underwent nephrectomy for unilateral pyelonephritis, but this did not lead to a permanent cure of her hypertension. Can 80 urograms be justified to pick up 1 patient ? An attractive proposal is to limit urography to selected patients in whom the yield is likely to be higher. Kreel et al.’studying a large series of mostly normotensive patients investigated by urography, proposed that this investigation should not be ordered until the results of simple investigations, such as urine examination and culture, have been studied; and Bailey et al. suggest that the indications in hypertension should be the same as in the normotensive population. The Glasgow group,5 however, found that this policy would miss a small number of significant lesions in younger patients, and, while agreeing with these indications in the older patients, suggest that urography remain routine in patients up to the age of 40. Can these results be extrapolated to other investigations for hypertension ? Is there now ever a place for admitting a patient with other than the most severe hypertension to the ward for investigation ? Answers to these questions are urgently needed, especially since here may lie an opportunity for large financial
savings. 1. Pickering, G. W. High Blood Pressure. London, 1968. 2. Tudor Hart, J. Lancet, 1970, ii, 223. 3. Lovell, R. R., Prineas, R. J. Int. J. Epidemiol. 1974, 3, 25. 4. Hawthorne, V. M., Greaves, D. A., Beevers, D. G. Br. med. J. 1974, iii, 600. 5. Atkinson, A. B., Kellett, R. J.Jl R. Coll. Phycns Lond. 1974, 8, 175. 6. Kreel, L., Elton, A., Habershon, R., Mason, A. M. S., Meade, T. W. Br. med. J. 1974, iv, 31.
‘
faecal extract from volunteers infected with MS1. The technique is informative and precise, but can hardly be described as quantitative-or even semiquantitative-so that quantitative assessment of rises in antibody titre is not really practicable. The findings in the 22 patients studied here are entirely negative for this hepatitis-A antigen-i.e., there were no seroconversions. The apparent discrepancy between the relative
A, and the material used
was a
proportions of non-hepatitis-B post-transfusion hepatitis and the distribution of viruses in this group, in different countries, is at present unexplained. Of course, geographical differences may apply-i.e., there may be an agent or agents at large in the U.S.A. but not in the U.K. However, before a hypothetical hepatitis-C virus is produced, a touch of Occam’s razor is not inappropriate. These patients might have a hepatitis A differing antigenically from the MSI of FEINSTONE et al. Morphologically, the supposed MS1 particle is like an enterovirus,l1 and the major groups of enteroviruses are notoriously divided into distinct serological subtypes, particularly the members of the ECHO and Coxsackie virus groups. On more certain ground in terms of hepatitis-A antigen, and contributing a real advance in technique, is the work of DEINHARDT, HILLEMAN, and others. DEINHARDT 13 and HILLEMAN 14 and co-workers have transferred strains of human hepatitis A to marmosets, and complement-fixation 15 and immuneadherence tests 16 have been developed, using extracts of infected marmoset liver as antigen. These tests KRUGMAN are sensitive, quantitative, and specific. and others 17 have applied both these techniques to patients who have had MS1 hepatitis A and they have shown seroconversion to this antigen. It will be of considerable interest to apply these two tests to patients with post-transfusion hepatitis. Meanwhile, it would be a great pity to disregard the contribution made by the screening of blood for HBAg and the exclusion of HBAg-positive donations. This has made big inroads into post-transfusion hepatitis 4,5 and has helped in the almost complete clearing of hepatitis from haemodialysis units in Great Britain.18,19 Nevertheless, there is still, at least in the U.S.A., a sizable and somewhat puzzling hepatitis gap to be plugged. The implication or exoneration of hepatitis A cannot, at the present rate of progress, be far off. 13.
Deinhardt, F., Holmes, A. W., Capps, R. B., Popper, H. J. J. exp. Med. 1967, 125, 673. 14. Provost, P. J., Ittensohn, O. L., Villarejos, V. M., Arguedas, G. J. A., Hilleman, M. R. Proc. Soc. exp. Biol. Med. 1973, 142, 1257. 15. Provost, P. J., Ittensohn, O. L., Villarejos, V. M., Hilleman, R. M. ibid. 1975, 148, 962. 16. Provost, P. J., Wolanski, B. S., Miller, W. J., Ittensohn, O. L., McAleer, W. J., Hilleman, M. R. ibid. p. 532. 17. Krugman, S., Friedman, H., Lattimer, C. New Engl. J. Med. 1975, 292, 1141. 18. Polakoff, S. (Public Health Laboratory Service Survey) Br. med. J. 1974, iv, 751. 19. Marmion, B. P., Tonkin, R. W. Br. med. Bull. 1972, 28, 169.
PREVENTION
OF
HÆMOPHILUS
INFLUENZÆ
MENINGITIS INFECTION with a pathogenic microbe causes illness in some persons whereas others remain well. Their immunity may be due to their good fortune in having at some time carried in their microbial flora avirulent bacteria with antigens similar to those of the pathogen. Such natural immunity may be induced by poorly virulent strains of the pathogenic species (as may happen with immunity to meningococcal disease 1,2) but also by completely unrelated organisms such as the coliform bacilli of the normal intestinal flora. 3,Thus the surface antigens of certain serotypes of Haemophilus, Meningococcus, and Pneumococcus may cross-react immunologically with antigens of other bacteria, including Escherichia coli.55 Attenuated strains of virulent microorganisms have been used for vaccination of man and animals since the days of Pasteur, and the possibility is now suggested that natural immunisation by the commensal body flora could also be exploited-by feeding suitable strains of E. coli to immunise against H. infiuenze meningitis.4,6-8
Schneerson and Robbins8 have reported that the feeding to adult volunteers of E. coli with the antigenic structure 075 :K100 :H5 caused an increase in the amount of serum-antibody against the capsular antigen of H. influenza type b. Earlier, suitable strains of E. coli had been found by culturing rectal swabs on an agar medium containing antiserum against the H. infiuenzae type b capsular antigen; a visible halo of antigen-antibody precipitate formed around colonies of bacteria with a similar antigen.4 The E. coli capsular antigen, designated K100, has since been shown to be immunochemically similar to the polysaccharide capsule of H. influenzce type b.9 The E. coli were fed to 16 healthy adult volunteers before breakfast, after taking 2 g. of sodium bicarbonate, in a dose of about 107-108 bacteria. The strain could be recovered from their faeces for three to eight weeks afterwards, and the volunteers responded, somewhat variably, with the development of antibodies bactericidal for H. infiuenzae type b. The feeding of young babies with the E. coli might well immunise them against Hcemophilus meningitis, since almost all strains of H. influenza from cases of meningitis have a type b capsule, 10 and there is evidence that the corresponding antibody is 1.
protective. 11, 12 Meningitis is
most
liable
to
develop
Goldschneider, I., Gotschlich, E. C., Artenstein, M. S. J. exp. Med.
1969, 129, 1327. Reller, L. B., MacGregor, R. R., Beaty, H. N. J. infect. Dis. 1973, 127, 56. 3. Wilson, G. S., Miles, A. A. Topley and Wilson’s Principles of Bacteriology, Virology and Immunology; chap. 41. London, 1975. 4. Bradshaw, M. W., Schneerson, R., Parke, J. C., Jr., Robbins, J. B. Lancet, 1971, i, 1095. 5. Robbins, J. B., Myerowitz, R. L., Whisnant, J. K., Argaman, M., Schneerson, R., Handzel, Z. T., Gotschlich, E. C. Infect. Immun. 1972, 6, 651. 6. Myerowitz, R. L., Handzel, Z. T., Schneerson, R., Robbins, J. B. ibid. 1973, 7, 137. 7. Br. med. J. 1974, ii, 462. 8. Schneerson, R., Robbins, J. B. New Engl. J. Med. 1975, 292, 1093. 9. Schneerson, R., Bradshaw, M., Whisnant, J. K., Myerowitz, R. L., Parke, J. C., Jr., Robbins, J. B. J. Immun. 1972, 108, 1551. 10. Turk, D. C., May, J. R. Hæmophilus Influenza: Its Clinical Importance. London, 1967. 11. Alexander, H. E., Heidelberger, M., Leidy, G. Yale J. Biol. Med. 1944, 16, 425. 12. Schneerson, R., Rodrigues, L. P., Parke, J. C., Jr., Robbins, J. B. J. Immun. 1971, 107, 1081. 2.
66
antibody and before they develop their own natural antibodies against the capsular antigen that is, between the ages of about 3 months and 5 years-and oral vaccination at the age of 3-6 months might overcome this gap in immunity. The strains of E. coli used have shown no evidence of enteropathogenicity, nor do they appear to be invasive. They are not entirely avirulent-they may be able to cause urinary-tract infection, for examplebut they do not appear to be any different in this respect from many other strains of E. coli that are to be found in large numbers in the normal human intestine.9 The investigators have not yet tried to immunise babies in this way; but the procedure has been shown to work in baby rabbits,and E. coli strains have been safely fed to newborn babies by workers in Czechoslovakia, who also showed that an antibody response developed when the intestine was colonised.13 Since babies are bound to be colonised by similar organisms quite naturally, the process should be safe provided the strains do not cause diarrhoea or other illness. The incidence of Haemophilus meningitis can be estimated only roughly. About 1500 cases of acute meningitis are notified in England and Wales to the Registrar General each year, and returns to the Public Health Laboratory Service suggest that about 20-25% of laboratory-diagnosed cases are due to H. influenzæ.14,15 If it is accepted that meningitis is undernotified by about 50%, there may be 800 cases each year in England and Wales. Since the disease mostly affects children under 5 years of age, perhaps 1 child in 1000 is liable to contract the disease before the 5th birthday. The mortality-rate may be as high as 10%, and the incidence of sequelse, which can include impaired intelligence, behaviour disorders, or hydrocephalus, has been reported to be 30% in the U.S.A.16 and 22% in Sweden.l’ Thus Haemophilus meningitis, though not common, is a serious illness, and the use of a vaccine given simply by mouth might be worth while provided it could be shown to be both safe and protective. Proof of safety may be obtainable by careful clinical trials, first in small groups and later in larger; but efficacy may be hard to establish since the disease has such a low incidence. Babies may not respond to the vaccination so well as immunologically mature adults,18 most of whom already possess some antibody against the shared antigen, and susceptibility to Haemophilus infection may be influenced by factors other than circulating antibody. 19. 20But the exploitation of what is probably the natural process by which infants become immune to this serious infection might prove a valuable and economical approach to infectious-disease control, with possibilities for other pathogens-meningococci and pneumococci, for example-that share antigens with bacteria of the commensal flora. 13. Lodinova, R., Jouja, V., Lanc, A. J. Bact. 1967, 93, 797. 14. P.H.L.S. Br. med. J. 1974, i, 453. 15. P.H.L.S. ibid. 1973, i, 623. 16. Sell, S. H. W., Merrill, R. E., Doyne, E. O., Zimsky, E. P., Jr. Pediatrics, Springfield, 1972, 49, 206. 17. Jonsson, M., Alvin, A. Scand. J. infect. Dis. 1971, 3, 141. 18. Smith, D. H., Peter, G., Ingram, D. L., Harding, A. L., Anderson, P. Pediatrics, Springfield, 1973, 52, 637. 19. Feigin, R. D., Richmond, D., Hosler, M. W., Shackelford, P. G. Am. J. med. Sci. 1971, 262, 338. 20. Lancet, 1971, ii, 914.
INTRAVENOUS UROGRAPHY IN HYPERTENSION THE morbidity and mortality associated with a high blood-pressure are a considerable incentive to clinicians to try to cure the hypertension by finding a surgically correctable abnormality of the renal, vascular, or endocrine system. Today’s drugs, however, can now control the blood-pressure of a large proportion of hypertensive patients and so the more elaborate investigations, such as renal arteriography or divided renal-function tests, are reserved for the occasional patient. Intravenous urography, however, is normally recommended as a routine investigation for those who 1 may need treatment. A greater awareness of hypertension and organised screening programmes 2-4are leading to the detection of an increasing number of patients with mild hypertension. Can we justify the cost and the inconvenience and risk to the patient of intravenous urography in all these cases ? Atkinson and Kellett5 made a retrospective analysis of the value of intravenous urography in 952 patients attending the Glasgow blood-pressure clinic. Although 120 (17-6%) renal abnormalities were demonstrated, these led to a change in patient management in only 9 cases (0-9%). A few of their patients, however, had been followed up for as little as four months, and so it is reassuring to see their results confirmed by Dr Bailey and others (on p. 57 this week). They followed up a group of patients for at least three years. In 80 patients, urography revealed 15 renal abnormalities (19%), but in only 1 was patient management changed by unexpected urographic findings. This patient underwent nephrectomy for unilateral pyelonephritis, but this did not lead to a permanent cure of her hypertension. Can 80 urograms be justified to pick up 1 patient ? An attractive proposal is to limit urography to selected patients in whom the yield is likely to be higher. Kreel et al.’studying a large series of mostly normotensive patients investigated by urography, proposed that this investigation should not be ordered until the results of simple investigations, such as urine examination and culture, have been studied; and Bailey et al. suggest that the indications in hypertension should be the same as in the normotensive population. The Glasgow group,5 however, found that this policy would miss a small number of significant lesions in younger patients, and, while agreeing with these indications in the older patients, suggest that urography remain routine in patients up to the age of 40. Can these results be extrapolated to other investigations for hypertension ? Is there now ever a place for admitting a patient with other than the most severe hypertension to the ward for investigation ? Answers to these questions are urgently needed, especially since here may lie an opportunity for large financial
savings. 1. Pickering, G. W. High Blood Pressure. London, 1968. 2. Tudor Hart, J. Lancet, 1970, ii, 223. 3. Lovell, R. R., Prineas, R. J. Int. J. Epidemiol. 1974, 3, 25. 4. Hawthorne, V. M., Greaves, D. A., Beevers, D. G. Br. med. J. 1974, iii, 600. 5. Atkinson, A. B., Kellett, R. J.Jl R. Coll. Phycns Lond. 1974, 8, 175. 6. Kreel, L., Elton, A., Habershon, R., Mason, A. M. S., Meade, T. W. Br. med. J. 1974, iv, 31.
‘
