675 LARYNGEAL-CANCER MORTALITY-RATES PER YEAR PER 100 000 POPULATION, 1950-69, AMONG WHITE MALES ACCORDING TO REGION AND POPULATION SIZE
county survey of
laryngeal cancer in white males revealed mortality (1950-69) in the north-east, particularly in the Philadelphia-New York-Boston corridor and along the Hudson River.3 Rates were also high in urban areas along the
throughout the U.S. Rates among non-white males this period nearly doubled in the South, in both urban and rural areas, while increasing more slowly elsewhere. There is evidence of a recent decrease in the urban-rural alcohol consumption gradientbut trends on alcohol consumption by race are unknown. Although interences based on correlation analyses such as these are limited, the data do suggest that not only alcohol consumption but also tobacco smoking and occupational exposures affect the patterns of laryngeal cancer in the United States. WILLIAM J. BLOT Environmental Epidemiology Branch, National Cancer Institute, JOSEPH F. FRAUMENI, JR Bethesda, Maryland 20014, U.S.A. LINDA E. MORRIS rural
areas
during
excess
South-eastern Atlantic Coast and Gulf Coast, in a pattern resembling that for lung cancer. We examined the association of the age-adjusted laryngeal cancer rates with demographic and industrial variables for the 3056 counties of the contiguous United States. Weighted multiple linear regression techniques were used. Rates throughout the country increased sharply with urbanisation, being nearly twice as high in highly urban compared with rural areas (see table). The urban gradient seems due, at least in part, to hea--6 vier alcohol-and tobacco consumption among city dwellers. 1 The rates were inversely proportional to socioeconomic status (e.g., county educational indices). Ethnic associations were seen, with rates tending to be higher in counties with higher percentages of residents of Irish (in particular), Polish, southern European, or British descent, in general agreement with a 1950 death-certificate survey among foreign-born populations in the U.S.’ Industrial correlations were also evident. Mortality tended to be higher, by 5-10%, in U.S. counties with heavy concentrations of chemical and printing industries. Larger increases were uncovered in a separate survey of areas with large shipyards during the 1939-45 war.8 Wartime exposures to asbestos may account in part for the higher rates of lung9 and laryngeal cancers in coastal areas of the country. ’We calculated the correlation between per-caput tax revenue data on alcohol beverage sales, which were available at the county level for six States with 217 counties, and the rates of mortality from laryngeal cancer among white male residents of these counties. The Spearman correlation coefficients were positive for five of the six States: +0.50 (P=0.006), +0.35 (p=0.06), +0.27 (P=0.05), +0.39 (P=0.001), -0-16 (P=0-30), and +0.35 (P=0.004). The correlations were reduced, but the positive association still remained, when we adjusted for demographic variables and for lung-cancer mortality (as a substitute for tax data on tobacco sales which are not available). Nationally the distribution of laryngeal cancer was correlated most strongly with lung cancer, closely followed by oesophageal and oral cancers. The association of laryngeal cancer with any one of these three tumours remained after adjusting for the others, consistent with a causal role for both tobacco and alcohol. 10 The U.S. death-rate for laryngeal cancer in white males has been more or less constant over the period 1950-69, but analysis of the county data showed a steady increase of 20-30% in 3. Mason, T. J., McKay, F. W., Hoover, R., Blot, W. J., Fraumeni, J. F., Jr. Atlas of Cancer Mortality for U.S. Counties: 1950-69. U.S. Government Printing Office, Washington, D. C., 1975. 4. Blot, W. J, Fraumeni, J. F., Jr. J. chron. Dis. 1977, 30, 745. 5. Keller, M., Alcohol and Health. U.S. Government Printing Office, Washing-
ton, D.C , 1974. Haenszel, W., Shimkin, M. B., Miller, H. P. Tobacco Smoking Patterns in the U.S. U.S. Government Printing Office, Washington, D.C., 1956. 7. Haenszel, W. J. natn. Cancer Inst. 1961, 26, 37. 8. Blot, W. J., Stone, B. J., Fraumeni, J. F., Jr. Environ. Res. (in the press). 9. Blot, W. J., Harrington, J. M., Toledo, A., Hoover, R., Heath, C. W., Jr. Fraumeni, J. F., Jr. New Engl. J. Med. (in the press). 10. Rothman, K. J. in Persons at High Risk of Cancer (edited by J. F. Fraumeni, Jr.); p. 139. New York, 1975. 6
MANAGEMENT OF THE IRRITABLE BOWEL
SIR,-Your editorial (Sept. 9,
p. 557) was marred by your of barium enema in the diagnosis of this complaint ("... much time and money is wasted on bariumenema examinations ..."). Such an attitude towards this important investigation is misplaced. The diagnosis of irritablebowel syndrome is a diagnosis based on negative findings; presumably the fewer the investigations carried out on a patient the greater the number of malicious diseases that will be included under this heading. An integral part of the management of the syndrome is, as you say, reassurance and psychological support. I find it hard unconditionally to reassure my patients -especially this particular bowel oriented and, generally speaking, medically aware group of people-in the absence of a barium-enema report. There is a place for barium enema even if only to act as a baseline so that any disease process can be seen to evolve in future investigations; I have seen little to refute the findings of Havia and Manner’ and their conclusion that the irritable-bowel syndrome may be a precursor to diverticular disease. Barium enema should keep its place in the diagnosis of this syndrome, both to reassure the patient (and his physician) that he truly deserves to be classified thus and so that a truer picture of what the disease process involves might be obtained
attitude
to
the
place
long term. Department of Surgery, Charing Cross Hospital, London W6 8RF
TREVOR
JOHN CROFTS
SCHOOL MILK AND GROWTH IN SCHOOLCHILDREN
SiR,—The preliminary information presented by Dr Baker and his colleagues (Sept. 9, p. 575) does not justify their conclusion that school milk has no effect on the physical development of primary-school children. The data seem to refer to a group of Mid-Glamorgan children in the lower half of the 7-11-year age range, and yet conclusions are drawn for the whole of this age range, which includes the beginning of the adolescent growth spurt. Furthermore, height and weight are poor indicators of bone status.2 A more reliable index can be derived from metacarpal measurements of children’s hand X-rays, if it were possible to 3 get agreement for this Ten years ago Newton-John and Morgan4 suggested that if the amount of bone in the skeleton could be increased, then a reduction in the frequency of fractures could be achieved. Provision of calcium supplements can at best delay bone loss in 1. Havia, T., Manner, R. Acta chir. scand 1971, 137, 569. 2. Garn, S. M., Wagner, B. in Adolescent Nutrition and Growth (edited F. P. Heald); p. 149. London, 1969. 3. Horsman, A. in Calcium, Phosphate and Magnesium Metabolism (edited B. E. C. Nordin). Edinburgh, 1976. 4. Newton-John, H. F., Morgan, D. B. Lancet, 1968, i, 232.
by
by
676
osteoporosis. It thus of appears logical dietary calcium at a encourage provision stage when it can be used to increase bone growth. Surely, therefore, the responsible attitude to the provision of school milk is that until it can be shown to have no effect on bone status determined radiographically, it is premature to conclude, with Baker et al., that "free school milk for 7-11-yearold children is unlikely to have any appreciable effect on the physical development of the children." postmenopausal
women, at risk from
to
previously undiagnosed and for which treatment was mandatory are noted. Perhaps, once a patient is referred to the social services department for long-stay accommodation, the medical diagnostic services become less interested, whereas patients sent to a were
day unit will, for the most part, be supervision and will be referred for
under intense medical rather than
treatment
diagnosis. Health Centre,
Stony Stratford,
Department of Biochemistry, Royal Dental School, Jenner Wing, St. George’s Hospital Medical School,
Milton
London SW17
Keynes,
J. S. COBB
Bucks
F. B. REED
H.D.L.-CHOLESTEROL AND DIABETES MEDICAL SCREENING OF OLD PEOPLE
323) takes issue with Professor Brocklehurst and his colleagues (July 15, p. 141) on the assumption that they had decreed that the screening of old people must be done in outpatients, and their reply (Sept. 2, p. 532) seems to encourage this belief. However, although in their survey the patients were seen in outpatients there is nothing in the article to prove that this is necessarily the only method of screening. Indeed in their introduction Brocklehurst et al. imply that periodic examination by the patient’s own general practitioner would be the ideal. Brocklehurst et al. make an important point-namely, that elderly people should be screened by a doctor interested in their problem at the time of their referral for residential care-and it would be a pity if this were to be obscured by argument over the precise method of screening adopted. In practice the best approach is likely to SIR,-Dr Wright (Aug. 5,
vary from area
p.
to area.
Devon and Exeter Exeter EX2 5DW
Royal
Hospital (Wonford), D. B. SHAW
SIR,-Professor Brocklehurst and his colleagues have drawn attention to the need for medical screening of elderly patients before admission to part III accommodation. In April a new social services home was opened in Milton Keynes, a year after the opening of a day hospital a mile away. MAJOR CONDITIONS
SIR,-In the light of conflicting reports’-4 of the association of
high-density-lipoprotein (H.D.L.)-cholesterol with oral hypoglycaemic agents, in particular the sulphonylureas, we thought it would be of interest to report the evidence from the Contraceptive Drug Study in Walnut Creek, California. Full data on H.D.L.-cholesterol have been published elsewhere,s but in our data tiles for 4978 women whose serum-H.D.L.-cholesterol levels were measured, we have found 6 who were taking a sulphonylurea and had a mean H.D.L.-cholesterol of 45.5 mg/dl and 11 who were taking insulin and who had a mean H.D.L. of 68.5 5 mgi dl. This difference is partly due to differences in weight, however. The women in the sulphonylurea group were all overweight, and averaged 30% above normal weight for height. After adjustment for weight, as well as for age, smoking and drinking habits, and oestrogen or progestin use (the adjustments being estimated by multiple regression from the entire cohort), the means were 51.5 mg/dl and 68.7 mg/dl for the sulphonylurea and insulin groups, respectively. The adjusted grand mean for the entire cohort was 62.3 mg/dl (S.D. 15-0). Thus, the mean for the insulin takers is not significantly different from this grand mean, while that for the sulphonylurea takers is low (t=l-8, p<0-05, one-sided test). Contraceptive Drug Study, Kaiser-Permanente Medical Center, Walnut Creek, California 94596, U.S.A.
JOHN
WINGERD
Epidemic Intelligence Service, Center for Disease Control, Atlanta, Georgia
DIANA B. PETITTI
DEMANDING TREATMENT AND DETECTED ON
SCREENING OF ELDERLY
p. 66) of a negative correlation glycosylated hxmoglobin (HbA1) concentration and, plasma-H.D.L.-cholesterol in diabetics was not confirmed by Dr Boucher and Dr Yudkin (July 29, p. 269), who asked whether, by chance, better control in female diabetics, whose H.D.L.-cholesterol concentrations would be expected to be higher than
SIR,-Our finding (July 8,
between
those in men, could account for the correlation we found. We have now analysed our results for men and for women separately and have found no significant difference between the sexes for HbA, or H.D.L.-cholesterol concentration. Mean HbAI concentration in women was 12-2±0-3% (±S.E.M.) of total haemoglobin, and in men 11-8±0-3%. Mean plasmaH.D.L.-cholesterol in diabetic women was 1-03 ±0-04 mmol/1, and in men 0-97 ±0-3mmol/1. There was no significant correlation between HbA1 and plasma-H.D.L.-cholesterol for women or men considered separately nor for any single-sex treatment group. However, we believe that this failure to reach statistical significance is a function of group size, trends towards a negative correlation being observed in all except the sulphonylureatreated groups. We do not think that the argument for the Patients admitted to either went through the same screening procedure by myself. Naturally, many abnormalities were recorded, but more major undiagnosed illness was found among those admitted to the residential home. The table compares the first block of 18 patients admitted to the home with the 18 consecutive patients seen immediately before at the day hospital. In this table only major medical conditions which
Bar-On, H., Landau, D., Berry, E. Lancet, 1977, i, 761. Stalenhoef, A. F. H., Demacker, P. N. M., Lutterman, J. A., van’t Laar, A. ibid. 1978, i, 325. 3. Calvert, G. D., Graham, J. J., Mannik, T., Wise, P. H., Yeates, R. A. ibid. 1978, ii, 66. 4. Durrington, P. ibid. 1978, ii, 206. 5. Bradley, D. D., Wingerd, J., Petitti, D. B., Krauss, R. M., Ramcharan, S. New Engl. L. Med. 1978, 299, 17. 1. 2.