Variation in the Mortality from Cancer Amongst Persons in the Different Districts of Glasgow and Its Relationship to Social Status By Matthew Young Published in: Glasgow Medical Journal Dr. M. Young?Variation in Mortality from Cancer. 205 VARIATION IN THE MORTALITY FROM CANCER AMONGST PERSONS IN THE DIFFERENT DIS- TRICTS OF GLASGOW AND ITS RELATIONSHIP TO SOCIAL STATUS. By MATTHEW YOUNG, M.D., Statistical Department, National Institute for Medical Research. The Medical Officer of Health for the city of Glasgow, in his annual report for the year 1924 issued in August, 1925, states, in a brief reference to the varying incidence of cancer in the municipal wards of the city, that it may be observed that the death-rate from cancer during the year was 1,748 per million in Blythswood, 1,659 in Park, 1,571 in Langside, and 1,594 in Camphill, as compared with 598 per million in Maryhill, 625 in Springburn, and 743 in Govan. His comment on these figures is:?"The most obvious contrast here is that the former group of wards may be regarded as mainly residential and the other as artisan, but that without making an intensive study of classification and its age and sex incidence nothing definite can be said regarding the incidence of the disease in various o o classes of the population." The above-mentioned figures, but without the added reservation, were quoted in the London Times in October, 1925, as showing that cancer in Glasgow among the leisured classes is more than twice as prevalent as it is among the wage earners. This is certainly the impression the crude figures would convey, but it is opposed to the more generally accepted view regarding cancer and social position; as, although one or two observers, including Heron (1907),4 Davies (1924),2 and Roger Williams (1908),7 maintain that cancer is most prevalent in people of high social status; most of those who have studied the problem, including Stevenson (1923),6 Brown and Mohan Lai (1914),1 Maynard (1909),5 and Dublin (1924),3 have come to the conclusion that the mortality from malignant disease 206 Dr. M. Young?Variation in the Mortality from Cancer as a whole in the community declines with ascent in the social scale. Heron's conclusion was originally based on the crude mortality-rates from cancer in the metropolitan boroughs of London for the year 1901. In a later paper, however, he examined the statistics for the four years' period 1901-04, making a rough correction for the varying age constitution of the population in the different London Boroughs by cal- culating the cancer death-rates on the populations between the ages of 30 and 70 years, as the bulk of the cancer deaths in both sexes occur in this age interval. Although the co-efficients of correlation he obtains from these data can only be regarded as doubtfully significant, his final conclusion is that the incidence of cancer is most heavy amongst the well-to-do classes. Brown has made a more detailed examination of the cancer data for the metropolitan boroughs of London for the year 1901 and the four years' period 1906-10, using standardised death-rates, and finds that this procedure reduces the correlation obtained by Heron by the use of crude mortality-rates to such a degree that the co-efficient between the proportion of pro- fessional men in the population and the cancer death-rate becomes doubtfully significant and that between the death-rate and the proportion of domestic servants in private families quite insignificant, as it is less than its probable error. From 1911 onwards the deaths recorded as occurring in the different districts in the annual reports of the Registrar-General have been adjusted where necessary by the .transfer of deaths to the districts to which they belonged. This is of special importance in cancer, as many cases are transferred to hospital for treatment and may die away from their proper districts. For the decennium 1911-20, for which cancer statistics in districts are thus more reliable than in earlier years, correlation co-efficients have been calculated between the standardised death-rates from cancer in males, females, and persons aged 25 years and upwards in the metropolitan boroughs of London and three indices of the general social status of the residents in these boroughs, namely, the number of rooms per person, the number of domestic servants per 100 private families, and the proportion of professional men. The co-efficients are all negative, suggesting a direct association between an excess amongst Persons in the different Districts of Glasgow. 207 of cancer and lower social status, but they are all so small in comparison with their probable errors that no significance can be attached to them. They thus support the inferences drawn by Brown from the data in previous years. Brown also finds a significant negative correlation between the standardised death-rates from cancer in twenty-six divisions of the city of Hamburg for the period 1906-12 and the average income of the inhabitants therein. This result, he states, serves to strengthen the general conclusion that cancer is not more, but less fatal among the well-to-do classes. Stevenson's conclusion is that, generally speaking, there seems to be evidence of a moderate increase of cancer mortality down the social scale, an opinion which is based on the mortality data for cancer amongst males in England and Wales during the three years' period 1910-12. In view of the difference of opinion that still exists regarding the influence of social status on the incidence of cancer, it seemed desirable to ascertain if the cancer statistics for the city of Glasgow in other years supported the general inference that is likely to be drawn from the mortality-rates for the disease in the municipal wards for the year 1924. Unfortunately, the statistics relating to this disease that have been published and made available for Glasgow are not very suitable for such an investigation. No annual totals of deaths from cancer in the different wards of the city have been published for the years subsequent to 1913, and the sexes are not tabulated separately for the earlier years, where such figures are given. In an attempt to make the best use possible of the data that are available, the standardised death-rates from cancer amongst persons aged 25 years and upwards for the four years' period 1910-13, were calculated by the indirect method, i.e., by apply- ing the death-rates from cancer at the several ages for Glasgow as a whole to the populations at the corresponding ages in the respective wards and obtaining factors of correction for the crude mortality-rates. The cancer death-rates in the several age groups in the city of Glasgow for the two years 1911-12 were first applied to the populations at the corresponding ages in the municipal wards, but the standardised death-rates were re-estimated by using the deaths from cancer and the 208 Dr. M. Young?Variation in the Mortality from Cancer populations at the different ages that are tabulated by the Registrar-General for Scotland in his annual reports for the four years 1910, 1911, 1912, and 1913 for Glasgow as a whole, after making such an adjustment of the figures as appeared necessary. The standardised death-rates from cancer in the respective wards derived in this way showed little variation from those already determined. As definite fluctuations in the number of deaths from cancer in the different districts occurred from year to year, four years' statistics were taken to obtain more reliable averages. The figures for the municipal wards that were incorporated in the city at the extension of its boundaries in 1912 were only available for the year 1913; these wards are consequently not included in the analysis. The choice of appropriate measures of the general social status of the residents in the different wards presented some difficulty and was restricted by the kind of information available. There are no particulars for these districts regard- ing measures of social status that are sometimes used, such as the proportion of professional men in the community or the proportion of domestic servants in private families. Certain details exist, however, relating to the distribution of the population residing in the several wards in houses of different size, and the first two of the three measures of social status that have been selected are based on these data and are intimately related to one another. These measures of social condition in the several municipal wards are:? 1. The.average number of persons per room. 2. The percentage of the total population residing in three or more apartment houses. 3. The infantile mortality-rate. They can only be regarded as approximate criteria of social status, but, in general, it may be said that descent in social status is accompanied by increasing density of population and a smaller proportion of the population living in three or more apartment houses. It is also to some extent coincident with a general increase in the infantile mortality-rate. The standardised death-rates from cancer for the four years' period 1910-13, the values of the three selected indices of social status, and the numbers of deaths from cancer in the amongst Persons in the different Districts of Glasgow. 209 corresponding municipal wards are shown in Table I. Blyths- wood Ward, as in 1924, shows the highest death-rate but it has a relatively small population and the number of deaths from TABLE I.-tSHOWING THE STANDARDISED DEATH-RATES FROM CANCER PER 1,000 AMONGST PERSONS AGED 25 YEARS* AND UPWARDS FOR THE FOUR YEARS PERIOD 1910-13, THE AYERAGE NUMBER OF PERSONS PER ROOM, THE PER- CENTAGE OF THE POPULATION RESIDING IN THREE OR MORE APARTMENT HOUSES, AND THE AVERAGE NUMBER OF CANCER DEATHS PER ANNUM IN THE SEVERAL MUNICIPAL WARDS OF THE CITY OF GLASGOW. Municipal Wards. Standardised death-rates per 1,000 from age of 25 upwards, 1910-13. Dalmarnock, Calton, Mile End, . White vale, Dennistoun, Springburn, Cowlairs, . Townhea4j Blackfriars, Exchange, Blythswood, Broomielaw, Anderston, Sandyford, Park, Cowcaddens, Woodside, Hutchesontown Gorbals, . Kingston, . Govanhill, Langside, . Pollokshields, Kelvinside, Maryhill, . Kinning Park, Glasgow (as a whole), (1) 1*672 1-993 2 297 1*747 1-814 1 -724 1-836 1*798 1-745 1*590 2-315 2-299 1-983 1-931 1-686 2*000 1-686 2-274 1-872 1-956 1-681 1-772 1-662 1-728 2 088 2 063 1-929 Percentage Mean Average No. of population infantile of persons residing in three vjortai;tv. per room or more apart- 1 ^" (Census, 1911). ment houses (Census, 1911). (3) 11 rate, 1910-13. (2) 2-7 2-2 2-7 2-3 1-5 2-5 2-4 2 2 2-2 1-2 0-9 2 0 21 1-6 10 21 1-9 2-6 1-9 2*0 2-0 11 0-8 0-7 1-8 2-6 1-8 30 12 29 53 17 24 31 36 70 89 47 34 52 88 34 34 11 52 48 34 81 93 96 33 18 38 (4) 139 156 146 147 81 118 124 140 159 190 106 160 125 142 82 155 118 129 137 136 97 61 61 40 99 144 127 Average No. of cancer deaths per annum, 1910-13. * (5) 35 36 41 26 32 29 23 32 17 2 4 8 26 26 27 33 34 37 31 33 26 40 21 23 34 11 * The death-rates were calculated on the populations from the age of 25 upwards as only 15 per cent of the deaths from cancer in Glasgow occur below this age. cancer is only four per annum on the average; in the Exchange Ward, the number of cancer deaths averages only two per annum. As no reliance can be placed on death-rates based on such No. 3. 0 Vol, CV. 210 Dr. M. Young?Variation in the Mortality from Cancer small numbers, the death-rates in these two wards have been omitted from further consideration. From the data for the remaining twenty-four municipal wards, correlation co-efficients were calculated between the cancer death-rates and the measures of social status already named; the results are shown in Table II. The correlation co-efficient between the standardised TABLE II.?SHOWING CORRELATION OF STANDARDISED MORTALITY FROM CANCER WITH MEASURES OF SOCIAL STATUS IN MUNICIPAL WARDS. Variables. Standardised death-rates from cancer amongst persons aged 25 years and upwards in twenty-four of the muni- cipal wards of Glasgow for the four years'period 1910-13. and 1. Average No. of per- sons per room (Census 1911). 2. Percentage of popula- tion in three or more apartment houses (Census 1911). 3. Average infantile mortality-rate for four years 1910-13. Correlation Co-efficients. + '382+ 118 - *397+*116 + *516+101. death-rate from cancer and the average number of persons per room is + *382jf *118, and that between the cancer mortality-rate and the infantile mortality-rate + *516jf 101. The correlation between the standardised mortality-rate and the percentage of the residents living in three or more apartment houses is -'SQTjf'llG. These values are all statistically significant, as they exceed three times their probable errors, and indicate that in the municipal wards a decided tendency exists for cancer mortality to increase as the number of persons per room ?or the density of the population?increases, and as infantile mortality increases; and to decrease as the percentage of the population living in three or more apartment houses increases; in other words, cancer mortality tends to increase with descent in social status, in so far as this can be interpreted by the criteria chosen. Some further data regarding the incidence of cancer mortality in the two sexes in houses of different size in Glasgow are contained in the Health Report for the city for the year 1912. In this report (p. 172), the deaths from malignant disease in each sex for the triennial period 1909-11 have been divided amongst Persons in the different Districts of Glasgow. 211 into four categories according as they occurred in one, two, three, or four or more apartment houses and the death-rates in the different age periods and for all ages in each group are given in a series of tables. From these data, standardised death-rates from cancer at the age of 25 years and upwards were calculated for the sexes in the different groups. These are given in Table III in apposition with the crude mortality-rates TABLE III.?SHOWING THE CRUDE DEATH-RATES FROM CANCER PER 1,000 AT ALL AGES AND THE STANDARDISED DEATH-RATES PER 1,000 AT THE AGES 25 YEARS AND UPWARDS AMONGST THOSE RESIDING IN ONE, TWO, THREE, AND FOUR OR MORE APARTMENT HOUSES IN GLASGOW DURING THE PERIOD 1909-1911. No. of Apartments in house. 1. 2. 3. 4 or more. Crude death-rates from cancer per 1,000 at all ages. Standardised death-rates from cancer per 1,000 at the ages 25 and upwards. Males. 0*61 0*62 0-79 103 Females. 0*96 0 92 1-22 111 Males. 1-473 1*652 1*508 1-412 Females. 1-895 2174 2131 1-525 at all ages which were extracted from the tables in the report. The crude mortality-rates seem to suggest the existence of a tendency for cancer mortality in both sexes to increase as the number of apartments in the house increases, or as social status of the occupants improves, if the number of apartments in the house may be accepted as a rough measure of social status. When reference is made to the standardised death-rates, however, it is found that, though these appear to increase slightly in passing from the one-apartment to the two-apartment house groups, no appreciable increase takes place in passing from two to three-apartment houses; whereas, in passing from three to four or more apartment houses such a decline in mortality occurs that, in the latter, the death-rate in males becomes practically equivalent to that of males in one-apartment houses, and in females falls to a level at least 20 per cent below that in this group, It has been pointed out by Dr. Chalmers that the mortality-rates given in the tables for the one-apartment houses are probably understated, as the number of deaths in institutions which cannot be allocated to houses of different size, occur among persons largely recruited from the one-apartment population. If some allowance be 212 Dr. M.Young?Variation in Mortality from Cancer. made for this source of error in the mortality-rates of the lowest social class, the results in Table III would seem to lend some support to the conclusion that has been drawn by the application of the method of correlation to the available data, namely, that there is no evidence of coincidence between a relatively excessive mortality-rate from cancer in the municipal wards of Glasgow and high social status of the residents therein, such as might be inferred from their crude mortality-rates for this disease; indeed, the correlation co-efficients that have been calculated appear to suggest that a definite tendency to the converse relationship exists. REFERENCES. 1 Brown, J. W., and Mohan Lai, "An inquiry into the relation between social status and cancer morality," "Jour. Hyg.," 1914, vol. ii, p. 186. 2 Davies, S., " A note on cancer occurrence in Hampstead and Shoreditch," "Public Health," 1924, vol. xxxvii, p. 139. 3 Dublin, L., "Kober and Havhurst," "Industrial Health," 1924, p. 957. 4 Heron, D., " Note on class incidence of cancer," " Brit. Med. Jour.," 1907, vol. i, p. 621. 5 Maynard, G. D., "A statistical study of cancer death-rates," "Biometrika," 1909, vol. vii, p. 278. 6 Stevenson, T. H. C., "The social distribution of mortality from different causes in England and Wales in 1910-12," "Biometrika," 1923, vol. xv, p. 382. 7 Williams, W. R., "The Natural History of Cancer," 1908, p. 354.