1221 TABLE II-UNPAIRED ANALYSIS OF ARTIFICIAL-SWEETENER USE

Letters

to

IN MALES

the Editor

ARTIFICIAL SWEETENERS AND BLADDER CANCER

SIR,-Your editorial on our findings of a positive association between the use of artificial sweeteners and male bladder cancer (Sept. 17, p. 578) concluded that "most readers will find the case against saccharin unimpressive". You referred to our analyses as "superficial" and indicated that insufficient data were "presented on potentially important confoundfactors to allow adequate analysis". ing The premature publicity caused by the unauthorised public release in the United States of a preliminary draft of our paper interfered with the normal process of consideration and response by authors to reviewer’s comments, and we should now like to present the data which we hope will convince you and your readers that our study was designed, conducted, and analysed at least as carefully as those studies you cited in contradiction to our findings. Selection of controls is critical in case/control studies.’ Our neighbourhood controls were selected by carefully designed, executed, and supervised procedures. In British Columbia and Nova Scotia, individual control selection began from the fourth house to the right of that of the case at the same time of day as the case had been interviewed and proceeded until an eligible control was found and agreed to participate. The estimated response-rate was 80% in British Columbia and 96% in Nova Scotia. In Newfoundland an electoral list was used to identify all potential controls in the census district corresponding to that of the case. From these lists a random sample of potential controls were drawn. There were no refusals from eligible controls. Any case/control study is subject to differences in recall by cases and controls. However, our controls and cases were inter...

TABLE I-ARTIFICIAL-SWEETENER USAGE BY PROVINCE AND SEX: I

RISK RATIOS AND PERCENTAGE USERS AMONG CONTROLS

viewed in their own homes in as similar a situation as possible, and we have no evidence to suggest differential recall nor are we aware of any means whereby artificial-sweetener users should be under-represented by those controls who were at home when the interviewer called compared with those who were out. Indeed, if being at home when the interviewer called were more likely for those members of the general population who are less active or ill, it might be expected that artificialsweetener users would be over-represented in the controls. This situation would have introduced a bias against demonstrating an association with artificial-sweetener use. This was possibly more likely for females than males, since if a male control was being sought in British Columbia or Nova Scotia and a female was found at home who indicated that an eligible male resided in the household, control seeking ceased and the interviewer returned at an appointed time to conduct the interview. 1. Miller, A.

B. J. natn. Cancer

Inst.

1975, 54, 299.

i

i

i

i

i

*p=0.006.

tp-0.01. Table i shows risk-ratio estimates for artificial sweetener use and percentage of artificial-sweetener users among controls by individual province. Although the estimates for Newfoundland and Nova Scotia are based on much smaller numbers than those for British Columbia, the risk ratios are all greater than 1 for males and in two of the three provinces are less than 1 for females. The percentage of users among controls is quite consistent between the three provinces for males. This internal consistency supports the validity of control selection. Cases and controls were similar with respect to the following unmatched variables (similar enough not to need consideration as potential confounding variables): marital status; religion; years of residence in province; consumption of milk, tea, fiddlehead greens, and nitrite and nitrate containing meats; history of asthma, tuberculosis, and operations other than on the bladder; illnesses other than urinary infections; use of analgesics ; and cooperation in the interview and interest in the interview, as assessed by the interviewer, and duration of the interview. Thus there were a number of respects in which the cases and controls were similar. Differences were, however, observed with respect to schooling, occupation, history of bladder and kidney infections, water supply, instant coffee, and cigarette smoking, besides artificial-sweetener use. However, the apparent association of education and risk of bladder cancer was due to confounding by exposure to a high-risk industry (to be reported in detail later). The difference in histories of bladder infections vanished when infections reported in the five years before diag-. nosis of bladder tumour were excluded and this almost certainly represents early symptoms of bladder cancer. The difference in histories of kidney infections was much smaller for males and is not statistically significant. The association of bladder cancer with cigarette smoking and certain industrial exposures such as rubber, chemicals, and petroleum has been reported by other investigators, and we have good qualitative and quantitative agreement with their results. Our observed association of bladder cancer and the use of a non-public water supply has not been reported before, but neither has it been investigated. The association with instant coffee is also not unexpected ; however, we did not find a corresponding difference for coffee other than instant. Table in in our paper

was

intended

to

be illustrative rather

1222 than

and it represents one of many analyses that done. Another approach to controlling for cigarette smoking and instant coffee consumption is shown in table ii in this letter. The matching has been broken since any large-sample bias thus introduced seems to be fairly negligible and in2 any case would direct the estimate of risk ratio towards unity.2 These data show that instant-coffee consumption is not confounding-neither is cigarette smoking, although it does modify the association in that lifetime non-smokers and current smokers of less than 15 cigarettes a day are not, apparently, at increased risk from artificial-sweetener use, although the estimates are based on small numbers. Your editorial stressed the difference for males and females. We recognise the dangers of arbitrarily subdividing data into categories and analysing data within such arbitrarily defined strata. However, for bladder cancer the distinction between males and females seems to us to be fundamental and not arbitrary and we decided to do this analysis by sex before we knew the results. Possible reasons for the sex discrepancy are chance, a difference between the sexes with respect to carcinogenicity (which, despite the similar finding in animals, you feel to be unlikely), or a bias which particularly affected the female controls. With respect to the latter the percentage users of artificial sweeteners among the female controls in the three areas was far less consistent than for the males (table t). Male cases used more artificial sweeteners per se but did not use more dietetic foods and beverages than male controls, whereas female cases used less artificial sweeteners, less dietetic foods, and less dietetic beverages than did female controls. Dietetic foods and beverages are comparatively new; only 1 of our subjects reported using them more than ten years before interview, whereas among those 13 subjects who indicated the time period when they used artificial sweeteners in tablet or drop form, 10 used them during or before the 1939-45 war. Thus, if artificial sweeteners are carcinogenic (and have the usual long latent period for bladder carcinogenesis) one would anticipate an increased risk associated with the use of tablets or drops but not with the use of dietetic foods or beverages. The reverse effect observed in females is much weaker, because of the smaller numbers. The pattern is, however, consistent with a group of female controls who were overweight and thus used artificial sweeteners, dietetic foods, and dietetic beverages. If the observed association between artificial sweeteners and male bladder cancer is not due to chance, then it must be: (1) real; (2) due to confounding; or (3) due to some systematic bias or error in study design and execution. We feel that table u and other analyses where we have controlled for confounding in other ways adequately cover this issue. None of these analyses indicates that the observed association between artificialsweetener use and bladder cancer in males could be due to confounding by some variable for which we have quantitative information. Neither are we aware of any possible confounding factor which could explain such an association but for which we have no data. The possibility of a systematic bias or error is harder to exclude. Nevertheless, the careful design and execution of our study, the internal consistency of the results, and the agreement between our results and those of previous investigators for variables other than artificial-sweetener use point to the lack of any such bias. Our results helps to clarify those studies which failed to show an effect on bladder-cancer mortality among diabetics. 3Further, they do not contradict the lack of association between the secular trends in bladder-cancer mortality and national patterns of artificial-sweetener consumption6,7 in view of the low attributable risk observed in our study in the general population. Our results differ primarily from those of investi-

definitive,

were

2. 3. 4.

Seigel, D. G., Greenhouse, S. W. J. chron. Dis. 1973, 26, 219. Armstrong, B., Doll, R. Br. J. prev. soc. Med. 1975, 29, 73. Armstong, B., Lee, A. J., Adelstein, A. M., Donovan, J. W., White, G. C., Rittle, S. ibid. 1976, 30, 151. 5. Kessler, I. I. J. natn. Cancer Inst. 1970, 44, 673. 6. Berbank, F., Fraumeni, Jr. J. F. Nature, 1970, 227, 296. 7. Armstrong, B., Doll, R. Br. J. prev. soc Med. 1974, 18, 233.

gators who have carried out previous case/control studies. Essentially, the major and possibly critical difference in design lies in the use of "neighbourhood" as opposed to "hospital" controls. One source of bias in hospital controls is any association between being overweight and the risk of hospital admission. In two of the four reported case/control studies’-" using hospital controls, the frequency of use of artificial sweeteners among male controls

was much higher than in our study (30%8 and 18%9). We agree that the role of artificial sweeteners in the xtiology of bladder cancer requires further study in man, but we feel that our study should not be dismissed and should be given at least equal weighting with the negative findings.

Epidemiology Unit, National Cancer Institute of Canada,

University of Toronto, Toronto, Ontario, Canada

A. B. MILLER G. R. HOWE

APARTHEID AND MENTAL HEALTH CARE

SIR,-Iam surprised that The Lancet should comment (Sept. 3, p. 491) on a World Health Organisation report which "is not based on first hand experience." My Department has carefully examined the report and found it to be a mixture of untruths, half-truths, and slanted information. The Minister of Health invited both the Red Cross and W.H.O. to send a team to investigate the entire mental-health set-up in South Africa during 1976. There was no reaction from W.H.O. The International Committee of the Red Cross, however, availed itself of the opportunity to "carry out an exploratory and preliminary visit to a representative sample of South African mental institutions in order to familiarize itself with the specific problems at first hand, before deciding whether or not to carry out a full-scale inspection". In an aidememoire dated May 26, 1977, addressed to the Minister of Health, the president of the International Committee of the Red Cross said: "the I.C.R.C. delegates did not find, in any of the psychiatric institutions which they visited, any patients hospitalised for other than medical reasons." You refer to a lack of published figures; these figures have been incorporated into this Department’s annual report since 1970. Your remark concerning "the absence of a single practising Black psychiatrist in South Africa" suggests that this is due to Government policy. Nothing can be further from the truth. Due to lack of interest among Black medical practitioners they cannot be coaxed into psychiatry as a specialty, despite attractive and generous scholarships and other inducements such as study leave on full pay. The statement that the great majority of African patients are treated on an involuntary basis is not entirely correct; the following statistics indicate that there has been a steady increase in the numbers treated on a voluntary basis since 1932.

Due to overcrowding in State hospitals it became essential find alternative accommodation for long-term Black patients. The solution was to use to best advantage buildings vacated by mining companies which had ceased to function. By means of tendering between private enterprise and the State these buildings were converted into hospitals. My Department has announced a multimillion-rand mental-health to

8. 9. 10. 11.

Morgan, R. W., Jain, M. Can. med. Ass. J. 1974, 111, 1067. Kessler, I. I. J. Urol. 1976, 115, 143. Simon, D., Yen, S., Cole, P. J. natn. Cancer Inst. 1975, 54, 587. Wynder, E. L., Goldsmith, R. Cancer, 1977, 40, 1246.

Artificial sweeteners and bladder cancer.

1221 TABLE II-UNPAIRED ANALYSIS OF ARTIFICIAL-SWEETENER USE Letters to IN MALES the Editor ARTIFICIAL SWEETENERS AND BLADDER CANCER SIR,-Your ed...
310KB Sizes 0 Downloads 0 Views