Epidemiological basis of tuberculosis eradication

Bull. Org. mond. Sante 11971, 45, 667-687Bull. Wid Hlth Org.J Epidemiological Basis of Tuberculosis Eradication 11. Mortality among Tuberculosis Cases and the General Population of Greenland * ERIK IVERSEN Greenland experienced, during the 1950s, a decline in mortality such as is on recordfor hardly any other place in the world: from 24 per 1 000 in 1951 to 8 per 1 000 in 1960, a decline of more than 10 % per year. Deaths from tuberculosis were especially reduced. Whereas more than one-third ofall deaths in 1951 were considered to be due to this disease, practically no deaths are ascribed to it today. This rapid improvement in the health situation in Greenland, which coincides with a large-scale development programme, is documented in detail in the present paper. The study is based partly on official mortality statistics and partly on a 9-year follow-up study of mortality and of morbidity from tuberculosis in the total population of West Greenland registered in 1955. The existence of such data for a developing area is probably unique. BACKGROUND In 1948, a commission of Greenlanders and Danish parliament and government officials was appointed with the object of planning the transformation of Greenland into a modern community. Accordingly, in 1950, a large-scale project was initiated for the general development oftrade and industry, expansion of educational institutions, improvement of housing conditions, and establishment of adequate health services. Much effort was devoted to loosening the grip of tuberculosis on the population. Favoured by narrow dwellings; poor standards of hygiene; long, dark winters; and low resistance against disease, caused by the irregularity of food supplies, the tu- bercle bacillus flourished and spread in Greenland as probably in few other parts of the world. The direct measures taken against the tuberculosis problem have been described in detail by Stein et al. (1968). They included BCG vaccination, case- finding, and effective treatment of tuberculosis pa- tients. A modern sanatorium with some 200 beds was established in Greenland in 1954 but, during the * From the Danish Tuberculosis Index, 25 Svanem0llevej, 2100 Copenhagen, Denmark. Previous articles in this series have appeared in Bull. Wid Hlth Org., 1959, 21, 5-49; 1964, 30, 609-621; 1966, 35, 483-508; 1966, 35, 509-526; 1967, 36, 703-718; 1967, 36, 719-731; 1967, 36, 733-746; 1967, 37, 893-906; 1967, 37, 907-925; 1969, 41, 95-113. period 1951-58, as many as 1 500 patients had to be sent to Denmark 1 for treatment. A chemoprophy- lactic trial was conducted in Greenland in 1956 and 1957 (Groth-Petersen etal., 1960; Horwitz etal., 1966). These undertakings were not easy, hampered as they were by an arctic climate and by the fact that the Greenland population lived in small communities, scattered along a coastline more than 2 000 km long. More specifically: systematic case-finding covering the whole population, and thereby early diagnosis of tuberculosis, was simply not possible at first. The situation changed in 1955, when a fully equipped sailing chest clinic was launched. Since then, annual X-ray examinations and, where indicated, bacterio- logical examinations, have been made of the whole population of West Greenland-i.e., about 90% of all Greenlanders. This paper gives a detailed account of the drama- tic decline in mortality that accompanied the develop- ment programme. Quite reliable official population and mortality statistics were already being compiled before the beginning of the campaign, and the popu- lation registered in connexion with the first X-ray round has been followed-up for 9 years with regard to mortality and to morbidity from tuberculosis. 1 Greenland has been an integral part of Denmark since 1953. For brevity, however, the European part of Denmark is referred to as " Denmark " in the present paper. 2759 -667 9 E. IVERSEN STUDY MATERIAL Official statistics At the end of each calendar year the clerical authorities in Greenland draw up a census list for each community, giving the name, sex, and date of birth of each inhabitant. At the same time, the births and deaths that have occurred during the year, which are recorded in the church registers, are noted. These data have formed the basis for crude annual population and vital statistics in Greenland since 1907. The age distribution of the population is ascer- tained every fifth year through a detailed census, whereas a tabulation of deaths, according to age, is made annually from copies of the church registers. Age-specific mortality rates have been worked out since 1948. Statistics on causes of death, according to sex and broad age groups, have been compiled since 1951 on the basis of notifications from district medical officers. If the physician is not present when a person dies, the death is communicated to him by a nurse, midwife, or other authoritative person, together with a provisional cause of death. When the physician next visits the place, he decides upon the definitive cause of death, taking all available infor- mation into account. Extracts from these statistics are given in Appendix Tables 1 and 2 (Denmark, Ministeriet for Gr0nland, 1960a,b; Denmark, Lands- lkgen i Gr0nland, 1951-1965). The follow-up study Data for this study have been collected from the following sources. (1) The basic material was a file that had been established during the first round of the X-ray boat from August 1955 to June 1956. Before the boat was launched, individual record cards with identification data (name, address, sex, date of birth) had been prepared from the official census lists of persons living in West Greenland at the end of 1954. At the visit of the boat to each community, the file was brought up to date: cards for persons who had died or moved away were taken out, and cards for the new- born and for newcomers were filled in. A total of 23 179 persons were registered in this way, consti- tuting the study population for follow-up. All persons attending the examinations-93 % of those registered- were subjected to a Mantoux 5-TU tuberculin test and a 70-mm photofluorogram of the chest. Persons who did not react to the tuberculin test (induration: 0-5 mm) were vaccinated with BCG. The X-ray picture was read by two readers and, in any case where one or both of them saw a suspicious lesion, a full-size roentgenogram was made and a sputum specimen was collected. The sputa were examined by direct microscopy and culture at the Statens Seruminstitut, Copenhagen. The results of all these examinations were recorded on the individual record cards, as were theBCG vaccinations and any information on earlier vaccinations. (2) Another source of material for the study was a tuberculosis register, kept jointly by the tuberculosis sanatorium in Greenland and by the Danish Tuber- culosis Index in Copenhagen. The register was established in 1954, when information was collected from the district medical officers on all persons who were, or had recently been, under treatment for respiratory tuberculosis; since 1955, all new cases of the disease have been recorded in the register. Information on deaths among registered persons, ob- tained through the cause-of-death reporting system, is also recorded. (3) Finally, photocopies were obtained of all pages of the church registers in which deaths had been recorded since the first voyage of the X-ray boat. A single file of punched cards, with one card for each person in the study population, was created at the Danish Tuberculosis Index, using the data from all three sources. Each card contained information on sex, year of birth, and place of residence, as well as the results of examinations made on the X-ray boat. Furthermore, if the person had a record in the tuberculosis register, the date of diagnosis and the results of the relevant bacteriological and X-ray examinations were entered on the punched card. Finally, the date of death and cause of death (if known) were recorded for all deceased persons. Among the study population of 23 179, registered in the period 1955-56, a diagnosis of respiratory tuberculosis had already been made in 1 699 persons before the mass examinations were begun. Another 373 cases were found during the first voyage of the X-ray boat, so that 2 072 persons-9% of the study population-were known to be suffering, or to have suffered previously, from respiratory tuberculosis. This was the initial size of what will be referred to throughout this paper as " the tuberculous popu- lation ", persons without a history of tuberculosis- 21 107 in the first instance-making up " the non- tuberculous population ". Table 1 summarizes what befell these two popu- lations during the following 9 years, starting from the first visit of the X-ray boat to each community; 668 EPIDEMIOLOGICAL BASIS OF TUBERCULOSIS ERADICATION: 1 669 Table 1. Basic components of follow-up study Group Tuberculous Non-tuberculous Totalpopulation population Initial population 2072 21 107 23 179 New tuberculosis cases + 1 711 -1 711 Deaths -355 -1 046 -1 401 Population by end of follow-up 3 428 18 350 21 778 Person-years 26 712 175 923 202 635 roughly speaking, this period corresponds to the calendar years 1956-64. A number of deaths occur- red in each of the two groups, and 1 711 new tuber- culosis cases moved from the non-tuberculous popu- lation into the tuberculous population. This means that the mortality among persons with a history of tuberculosis could be determined from the fate of a total of 3 783 persons, observed over a period ranging from a few days to 9 years. Similarly, the mortality among persons without tuberculosis could be found by following 21 107 persons until they died or con- tracted the disease (within the 9-year period). The person-years corresponding to the observed number of deaths for each population category are shown in the bottom line of the table; they have been cal- culated under the assumption that the new cases diagnosed each year were spread evenly throughout the year. RESULTS Trend of mortality, 1907-65 The general trend of mortality in Greenland during the last half-century is shown in Fig. 1, which shows the crude death rates 1 for each of the years 1907-65 (Denmark, Ministeriet for Gr0nland, 1957; 1960 a, b; Denmark, Landslxgen i Gr0nland, 1951-1965). The picture looks completely different before and after 1950. Up to that year, the curve is dominated by a number of impressive peak rates of more than 35 per 1 000, contrasting to inter-peak rates of 20-25 per 1 000. The peak rates were caused by violent epidemics of diseases such as whooping-cough and influenza. That these visitations were not catas- 1 The term " crude " means that age has not been taken into account in the calculation of these rates as it is for age-specific and age-standardized rates. trophic to the Greenland population may be ex- plained only by its high fertility. The crude birth rates in these years varied between 37 and 47 per 1 000, thus enabling the population to increase from 13 000 in 1910 to 22 000 in 1950. After 1950, there is an al- most complete absence of mortality peaks. This does not mean that epidemic diseases had disappeared: influenza and whooping-cough still occurred now and then, and the different communities had their first outbreaks of measles between 1951 and 1962 (Christensen et al., 1952, 1953; Bech, 1965). How- ever, prophylactic and curative treatment now made it possible to limit the effects of epidemics. Furthermore, mortality declined dramatically dur- ing the 1950s: from 24 per 1 000 in 1950 and 1951, the death rate dropped continuously to a level of 8 per 1 000 in 1960. This corresponds to an average decline of more than 10% per year. Since then, mortality has remained fairly constant at about 9 per 1 000. The birth rate rose somewhat after 1950, and in recent years has remained at 45-50 per 1 000. The population therefore grew more rapidly, and by 1965 numbered 35 000. Mortality according to cause, 1951-65 In Fig. 2 the crude mortality rates for the 15-year period 1951-65 are broken down according to three groups of causes of death: tuberculosis, other dis- eases, and accidents I (Appendix Table 1). The role played by tuberculosis is seen to have changed drasti- cally during the 1950s. Whereas more than one- third of all deaths in 1951 were attributed to this disease, only a few deaths from tuberculosis have been reported in recent years. The cause of death 1 In the present paper this term covers accidents, poison- ings, and violence (Nos 800-999 in the International Classi- fication of Diseases, Injuries, and Causes of Death, WHO). 30 20 02 1910 '20 '30 '40 '50 '60 CALENDAR YEAR Fig. 1. Crude annual death rates for Greenland, 1907-65. 20 Lu 10 Lu4 CALENDAR YEAR Fig. 2. Crude annual death rates for Greenland, 1951-65, according to cause of death. EPIDEMIOLOGICAL BASIS OF TUBERCULOSIS ERADICATION: 11 P_tZA.0 tu'---'''' 'm to DENMAR K i M ALE S _E X XFE MALE S I1I3 m4 2 0 '-'21 8 12i Fig.~~~~~~3 Ditibto of pouato inGenadadDnak_ 90a codn tose!ndae was probably not ascertained with the same degree of accuracy in the early 1950s as later, so part of the decline in the number of deaths from tuberculosis may be due to an improvement in the determination of the cause of death. There is no doubt, however, that a considerable part of the drastic fall in the general mortality rate between 1951 and 1960 can be ascribed to the retreat of tuberculosis as a cause of death. Mortality from diseases other than tuberculosis also decreased during the 15 years covered by Fig. 2; indeed the rate was about halved. Accidental deaths, on the other hand, were on the increase, and today account for about 30% of all deaths in Greenland. Age-specific mortality, 1950-65 The level of 9 per 1 000, to which the crude mor- tality rate had fallen by 1960, might at first sight seem to be quite satisfactory, since it compares well with the present level in Denmark and most other European countries (United Nations, 1969). It is strongly influenced, however, by the fact that the population in Greenland is very young. This is seen clearly from Fig. 3, which shows the age/sex pyramids for the populations of Greenland and Den- mark in 1960 (Denmark, Statistiske Departement, 1963, 1965). In Greenland, 78% of the population was below 35 years of age; in Denmark, only 52%. A proper evaluation of the mortality in Greenland must therefore be made on the basis of age-specific rates (Appendix Table 2). The change in the age-specific mortality during the last 15 years is illustrated in Fig. 4 by means of three curves. Two of them, those for 1950-51 and 1959-60, refer to the beginning and end, respectively, of the period during which the crude death rates dropped so rapidly; the curve for 1964-65 shows the latest available rates. Each curve begins with the death rate for children between 1 and 2 years and ends with that for the age group 55-64 years; the very high 671 E. IVERSEN 1959-60, AND 1964-65 lz. lz.lz. QCkwQ. ti ti . IR czi _j IR 1)"I Zt ;Z! 11.1 AGE IN YEARS Fig. 4. Age-specific mortality in Greenland, 1950-51, 1959-60, and 1964-65. rates in infancy and old age have been omitted for the sake of clarity. The drastic reduction in mortali- ty during the 1950s affected all age groups, but the relative decline was greatest at the lowest ages: mor- tality in children between 1 and 10 years was re- duced to one-fifth; that in the oldest age groups, to one-third. However, the absolute decrease for the oldest ages was quite substantial. Between the periods 1959-60 and 1964-65, mortality remained practically unchanged. The significance of these mortality rates becomes clearer when they are compared with the correspond- ing rates for Denmark. The 1950-51 and 1964-65 curves for Greenland have therefore been repro- duced in Fig. 5 against a background of four Danish curves showing the changes in mortality during the period of about 120 years (1840-49 to 1961-65) for which such data are available (Den- mark, Statistiske Departement, 1905, 1966; Denmark, Danmarks Statistik, 1967). The mortality curves for Greenland and Denmark differ both in level and in shape. This is seen most clearly when the 1950-51 curve for Greenland is compared with the 1880-89 curve for Denmark. The two curves practically coin- cide where they represent the rates during childhood, whereas the rates for adults are, roughly speaking, twice as high for Greenland as for Denmark. It would seem that the adult Greenlanders were able to protect their children better than themselves against the rigours of arctic life; on the other hand, one might say that the process of aging was more rapid in Greenland than in Denmark. Because of this feature, neither of the two curves for Greenland can be characterized as corresponding fully to a particular curve for Denmark; and rates such as those for adult Greenlanders in 1950-51 have simply never been recorded in Danish mortality statistics. In 1964-65, the mortality among children and adults in Greenland corresponded to the Danish rates for the periods 1921-25 and 1911-15, respectively. The in- fant mortality in Greenland, for children below 1 year of age, was 129 per 1 000 in 1950-51 and 74 per 1 000 in 1964-65. These two rates reflect the situation in Denmark in 1895-99 and the early 1 930s. 672 EPIDEMIOLOGICAL BASIS OF TUBERCULOSIS ERADICATION: 11 lz ti.: kKtL:NLAN U I/ oz~ I OD.:DENOARK F40 I ~~~~~~~~~~~~~D.1840 49 /D. 1880-89 20 G.~~~~~~~'.1964-65 IC 0 10 20 30 40 50 60 AGE IN YEARS Fig. 5. Age-specific mortality in Greenland and Denmark during selected periods of time. Mor-tality according to sex, age, and cause Age-specific death rates for the 6-year period 1958-63 are presented in Fig. 6, separately for males and females. The upper section shows the total mor- tality from all causes of death, whereas the other two sections give the mortality from accidents and dis- eases. The rates of mortality from all causes were generally higher for males than for females; but the age trend was somewhat different for the two sexes, and at the age of 38-47 years females had a higher rate than males. This feature is explained when the accident and disease categories are considered separately. Fig. 6B shows that the risk of death from accidents was much higher for males than for fe- males. This was true even among children, where the mortality of boys was 2-3 times that of girls; and the mortality in adult males was about 4 times that in females. The curves for diseases in Fig. 6C pre- sent a more complex pattern in that the relations between the sex-specific rates were completely differ- ent during three periods of life. The rates for males and females were nearly the same during childhood and adolescence; between the ages of 25 and 50 years, the mortality of females was up to 4 times that of males, and among people above 50 years of age the rates were highest in males. The four curves in Fig. 6B and 6C have been reproduced in Fig. 7 together with corresponding curves for Denmark (Denmark, Sundhedsstyrelsen, 1961). It is seen here that the excess mortality in Greenland is due largely to deaths from accidents, which were about 5 times as frequent as in Denmark. However, mortality from diseases also was higher in Greenland, the difference being particularly pro- nounced for children of both sexes below 5 years of age and for females aged between 40 and 50 years, where the rates for Greenland were about 3 times those for Denmark. Mortality according to tuberculosis status Even if, since 1960, tuberculosis has practically disappeared as a direct cause of death in Greenland (see Fig. 2), this does not necessarily mean that the disease now has no influence upon the mortality. It could well have an indirect effect, by lowering its victims' resistance against attacks from other diseases. This question may be elucidated by data from the 9-year follow-up of the population 673 E. IVERSEN O~~~~~~~~~~~ - B. ACCIDENTS $O 10MALES a °-- FEMALES MALES 30 -C. DISEASES FEMALES 20 -/ 10_o L -- S ------ .. 0 10 20 30 40 50 60 70 AGE IN YEARS Fig. 6. Mortality in Greenland, 1958-63, according to sex, age, and cause of death. registered during the first round of the X-ray boat in 1955-56. As explained under " Study material ", this popu- lation may be represented as currently consisting of two segments: the tuberculous population-i.e., per- sons suffering, or known to have suffered, from respi- ratory tuberculosis; and the non-tuberculous popu- lation-i.e., persons without a diagnosis or history of tuberculosis. The year-by-year size of the two groups, the annual flow of new tuberculosis cases from the non-tuberculous population into the tuberculous population, and the annual number of deaths oc- curring in each of the groups are given in Appendix Tables 3 and 4, according to age. The Annex ex- plains how these data have been used for the calcu- lation of age-specific annual death rates. Fig. 8 shows such rates, representing the average mortality during the 9-year period. Rates are given only for the five 10-year age groups between 15 and 64 years, since the rates for the relatively small tuberculous population below 15 years were rather unreliable, and it was impracticable to include in the graph the high rates for persons above 65. The three shadings indicate the approximate contributions that tuberculosis, other diseases, and accidents, made to the mortality. The mortality was consistently higher, within each age group, for the tuberculous than for the non- tuberculous population. This, of course, was partly owing to the deaths caused directly by tuberculosis, but also to an excess of deaths from other diseases. Below 45 years of age, the mortality from diseases 674 EPIDEMIOLOGICAL BASIS OF TUBERCULOSIS ERADICATION: II 30 -.3C, 4 ;= R 20| C. DISEASES MALES 20 GREENLAND DENMARK i10 E - -D. DISEASES FEMALES/ 10[' 10 20 30 40 50 60 70 AGE IN YEARS Fig. 7. Mortality in Greenland, 1958-63, and in Denmark, 1960, according to sex, age, and cause of death. other than tuberculosis was 2-3 times as high among the tuberculous as among the non-tuberculous popu- lation; in the 45-54-year age group, it was 1.8 times as high, and for those above 55 years, 1.1 times. The mortality from accidents, on the other hand, tended to be lower in the tuberculous population. For the further analysis of the follow-up data, it has been judged convenient to combine the age- specific death rates for each population category into a single rate. The crude mortality rates-I 3.3 per 1 000 person-years for the tuberculous population and 5.9 per 1 000 for the non-tuberculous population -are not suitable: they give an exaggerated im- pression of the excess mortality, part of the differ- ence being due to the relatively high age of the tuberculous population. Instead, age-standardized rates, representing the mortality among persons above 15 years of age, have been calculated as described in the Annex, and the result is shown in Fig. 9. These rates are seen to reflect quite well the general pattern of the age-specific rates in Fig. 8. When evaluated according to the standardized rates, the mortality from diseases other than tuberculosis in the tuber- culous group is seen to be 1.5 times as high as among the non-tuberculous population. Age-standardized rates have been used in Fig. 10 to show the time trend in the mortality among the tuberculous and non-tuberculous populations; average annual rates are given for each of the three 3-year follow-up periods, roughly coinciding with the calendar periods 1956-58, 1959-61, and 1962-64. The most conspicuous feature is the drastic decline 675 -et a 15-24 25-34 35-44 45-54 INITIAL AGE IN YEARS 55-64 Fig. 8. Age-specific mortality among the tuberculous and non-tuberculous populations in Greenland, 1956-64, according to cause of death. TUBERCULOSIS OTHER DISEASES ACCIDENTS IZ 14 4. Fig. 9. Age-standardized death rates for the tuberculous and non-tuberculous populations in Greenland, 1956-64. 60 50 T : TU ERCULOUS POPULATION 44.9 NT:NO -TUBERCULDUS POPULATION 40- UBERCULOS S 20 EPIDEMIOLOGICAL BASIS OF TUBERCULOSIS ERADICATION: I 6 216 9.4 14 k- 12 ' IIJa 6 90 4 2 3 0- T NT 1956-58 Fig. 10. Mortality among the tuberculous 1956-58, 1959-61, and 1962-64. in the tuberculosis mortality rate for the tubercu- lous population: from 9.4 per 1 000 in the first period to 1.9 and 0.9 per 1 000, respectively, in the last two periods. Also the mortality from diseases other than tuberculosis decreased considerably between the first and second periods, in both the tuberculous and the non-tuberculous populations. In the third period, however, the rate for the tuberculous population rose again. On closer inspection, this increase is found to be due to widespread epidemics of measles and influenza in 1962. In that year, the tuberculous popu- lation experienced a mortality from other diseases that was more than twice the level in 1961 and 1963. In the non-tuberculous population, on the other hand, the corresponding increase was only about 20 %. Subgroups of the tuberculous population In Fig. 10, the tuberculous population has been treated as a whole comprising new cases as well as and non-tuberculous populations in Greenland, persons who contracted the disease several years previously, and the variation in the mortality in this group, and in the non-tuberculous population, has been demonstrated according to the follow-up period. Another approach has been attempted in Fig. 11, with the purpose of clarifying how the mor- tality among the tuberculous depended upon the year in which the disease had been diagnosed in them and upon the time that had elapsed since the diagnosis. The tuberculous population has been divided into three groups, according to the year of diagnosis, and a number of age-standardized annual death rates have been calculated for each group, covering different periods of time after the diagnosis. The pattern of the rates is best illustrated by those in Fig. llB, which relate to persons in whom tuber- culosis was diagnosed during the first mass campaign (1955-56) and during the following 3 years. In this 77 E. IVERSEN B. 1235 PATIENTS DIAGNOSED 1955-58 10°0 C. 849 PATIENTS DIAGNOSED 1959-64 TUBERCULOSIS OTHER DISEASES oL _~~~~~~~ACCIDENTSX IST 2ND_ 3RD 4TH-6TH 7TH _9TH 0TH YEAR AFTER DIAGNOSIS Fig. 11. Mortality among the tuberculous population in Greenland, 1956-64, according to the time of diagnosis and length of time that has elapsed since diagnosis. group, the mortality in the first year after diagnosis is particularly high-22 per 1 000. More than one- half of the rate-13 per 1 000-is attributable to deaths caused directly by tuberculosis; the mortality from other diseases-8 per 1 000-was slightly higher than in the non-tuberculous population during the same period, 1956-58 (see Fig. 10); on the other hand, the number of death from accidents was in- significant. During the subsequent years, different trends are seen for the three categories of deaths. The rate for deaths from tuberculosis decreased con- tinuously: the mortality from other diseases re- mained fairly constant, in contrast to the decreasing trend seen in Fig. 10 for the non-tuberculous popu- lation. Deaths from accidents increased sharply about 4 years after diagnosis, when the patients had apparently returned to normal life. Fig. 1IA shows the mortality among patients in whom the disease had been diagnosed before systematic case- finding was established. The tuberculosis mor- tality rates in this group were several times the corresponding rates in Fig. 11B, not only at the out- set but also as long as 10 years after the diagnosis. Fig. 1 IC, on the other hand, shows that, even during the first year after diagnosis, few deaths from tuber- culosis occurred among persons in whom the disease was diagnosed during the last 6 follow-up years. The result of the X-ray examination made at the time of diagnosis was known for 2 023 (97 %) of the 2 084 persons in whom tuberculosis was diagnosed during or after the first mass campaign. The extent of the lesions was indicated by a distinction between unilateral and bilateral lesions, and their severity was characterized by the absence or presence of a cavity. Fig. 12 gives age-standardized annual death rates for each of the four possible combinations of X-ray classifications. The mortality was lowest among patients with unilateral lesions without a cavity (cate- 678 EPIDEMIOLOGICAL BASIS OF TUBERCULOSIS ERADICATION: 11 1J 5 6. k G C. BILATERAL LESIWITHOUT CAVIT) %J (616 PATIENTS) m 20 k ': 15 _ 2.310 Fig. 12. Mortality among tuberculosis to the initial X-ray findings. gory A) and highest among those with bilateral lesions and a cavity (category D), whereas interme- diate rates were found for the other two categories. This applies to the mortality caused directly by tuber- culosis as well as to that from other diseases. The data thus indicate that both factors, extent and se- verity of lesions, were of relevance for the mortality level. The results of the initial bacteriological examina- tions were available for 1 912 (92%) of the same 2 084 patients. Of these 1 912 patients, 651 (about one-third) were bacteriologically negative. Quite sur- prisingly, it was not possible to demonstrate any difference in mortality between these and the bac- teriologically positive patients. Subgr-oiups of the nion-tuberculous populationi Among the non-tuberculous population of 21 107 persons, 19 895 (94%) attended the mass examina- tions in 1955-56. The X-ray examinations showed patients in Greenland, 1956-64, according that 8 791 of them (44%) had normal lungs, 6 879 (35 %) had calcified lesions, 1 409 (7%) had healed lesions, and 2 780 (140%) had suspicious lesions. Whenever the first X-ray picture was unsatisfactory, a retake was attempted. Thus, finally, the X-ray was technically inadequate in only 36 cases (0.2%). The classification of the findings is based on double read- ings of the 70-mm photofluorograms or full size roentgenograms (see under " Study material "); where the two readers disagreed about the classifi- cation, the more severe reading was accepted. Age-standardized mortality rates for the persons who, at this initial examination, had normal lungs, calcified or healed lesions, or suspicious lesions are shown in Fig. 13 together with the rates for the 1 212 non-attenders. Four rates are given for each group, representing the first follow-up year, the following 2 years, and the last two 3-year periods. As did the rates for the total non-tuberculous population, these rates represent the mortality only up to the moment 679 E. IVERSEN I03 C3 (278C 10L 1956 9g Fig. 13. Mortality among the to the initial X-ray findings. when tuberculosis was contracted. The division of the rates according to the cause of death is based on the assumption that the age-specific mortality from accidents was the same for all four categories of the non-tuberculous population. The results for persons with calcified lesions and for those with healed lesions have been combined in the presentation, because the mortality was practi- cally the same in the two groups. Fig. 13 shows, furthermore, that it was hardly different in persons with normal lungs. The situation was different, however, for persons with suspicious le- sions. During the first follow-up year-i.e., the first year after the lesions had been seen-this group had a rate of mortality from disease (13 per 1 000) that was 2-3 times the corresponding rate for persons with normal lungs. However, it was well below the rate of 23 per 1 000 observed at the same time in per- sons in whom tuberculosis had been diagnosed under the mass campaign. The highest mortality was found in the non-attenders, who during the first year had the highest rate (26 per 1 000) encountered in the whole follow-up study; it must have been severe illness that prevented many of these persons from coming for the examinations. The excess mortality non-tuberculous population in Greenland, 1956-64, according was not limited to the first year, as regards either the persons with suspicious lesions or the non- attenders: for both groups it was still discernible by the end of the follow-up study. Of the 19 895 attenders from among the non- tuberculous population, 2 240 (92% of the non- reactors to the tuberculin test) were vaccinated with BCG, whereas 6 793 had been vaccinated earlier and 8 714 were natural reactors. Only 48 persons were tuberculin-negative and not vaccinated. The status of 2 100 others was unknown. If would have been interesting to compare the mortality in the vaccinated persons with that in the natural reactors to the tuberculin test. However, most of the vaccinated were aged less than 25 years, whereas practically all the reactors were over 5 years of age. In the age span that included persons of both categories-5 to 24 years-the mortality from disease was too low for reliable rates to be established. DISCUSSION A significant expression of a population's mor- tality level at a certain time is the expectation of life- i.e., the average lifetime of a hypothetical group of 680 EPIDEMIOLOGICAL BASIS OF TUBERCULOSIS ERADICATION: 11 TUBERCULOSIS PATIENTS NON-TUBERCULOUS IST YEAR 2ND _3RD YEAR POPULATION 3 AFTER DIAGNOSIS AFTER DIAGNOSIS I ~ ~ ~ IEASES ~ACCIDENTS I20 G:GREENLAND 1959-64 256 ~~~~D:DENMARK 1961-64104 12.767 2.46 0~~~~~~~~~~~~~~~~~~. G D G D G D Fig. 14. Mortality among tuberculosis patients and the non-tuberculous population in Greenland and Denmark. newborn infants subjected throughout life to the age-specific mortality rates for the population. As recently as 1950-51, the expectation of life in Greenland was as low as 39 years, and Fig. 1 indi- cates that this must have been more or less the nor- mal level-apart from epidemics-during the pre- ceding 35 years. Nine years later, in 1959-60, the expectation of life had gone up to 62 years. Deaths from tuber- culosis, which formerly constituted about one-third of all deaths, had been almost eliminated, and the mortality from other diseases had been halved. The improvement was due partly to direct antitubercu- losis measures and partly to a rise in the general standard of living. However interesting it would have been, it is of course not possible to say how much each of the different undertakings in the development programme, started about 1950, contributed to this impressive result. Since 1960, no further mortality reduction has been observed, and the expectation of life for 1964-65 -62 years-was still well below the corresponding 72 years for Denmark. The difference was due main- ly to a high infant mortality and to a high mortality from accidents among males. More than one-third of all deaths in males today are caused by accidents, of which drowning accounts for about half. Perhaps the most significant finding in the follow- up study is that tuberculosis patients were not only exposed to the direct mortal effects of tuberculosis: their mortality from other diseases also was higher for many years after the attack of tuberculosis. Dr Krebs Lange has demonstrated similarly that per- sons affected by the measles epidemic of 1962 had a morbidity from tuberculosis, in 1962 and 1963, that was more than twice the rate for the rest of the West Greenland population (Bech, 1965). The explanation of these two findings may be that a certain segment of the population is especially prone to contract tuberculosis, measles, and other diseases because of inherited and/or environmental conditions. Another and, it would seem, more likely interpretation is that tuberculosis as well as measles lowered the patients' resistance against later attacks by other diseases. The mortality rates observed, in the first 3 years after diagnosis, in inhabitants of Greenland in whom tuberculosis was diagnosed during the period 1959-64 (Fig. 1IC), and the rate for the non-tuberculous population during the same period (Fig. 10), are compared in Fig. 14 with similar data for Denmark for the period 1961-64 (Iversen, 1967). The Danish rates were obtained as weighted averages of age- specific rates, using the same weights as in the cal- culation of the Greenland rates. Quite unexpectedly, the mortality among tuberculosis patients was con- siderably higher in Denmark than in Greenland, al- though it was the reverse for the non-tuberculous population. An explanation could be that the few tuberculosis patients in Denmark stem from certain population segments with a high mortality-apart from that caused directly or indirectly by tuberculosis -whereas in Greenland such patients differ less from the general population. Other comparisons with the mortality among Danish tuberculosis patients also are of interest. In a study of patients from a large provincial town, cover- ing the period 1925-64 (Buhl & Nyboe, 1967), the mortality was found to depend, as in Greenland, upon both the extent of the lung lesions (unilateral 681 E. IVERSEN or bilateral) and their severity (absence or presence of cavity). However, in the study of all Danish patients in 1961-64 (Iversen, 1967), cavitation seemed to have no influence upon the mortality. The study showed, on the other hand, that patients in whom the initial bacteriological examination gave a positive result had a higher mortality than those who were bacteriologically negative; strangely enough, this could not bedemonstrated for the Greenland patients. Among the non-tuberculous population of Green- land, those with suspicious lesions and, even more, the non-attenders at the 1955-56 mass examinations had particularly high mortality rates. The same was found in a Danish follow-up study based on mass examinations made in 1950-52 (Horwitz & Knudsen, 1960; Horwitz & Wilbek, 1971). ACKNOWLEDGEMENTS The author is greatly indebted to the Northern Europe University Computing Centre, Lyngby, Denmark, for the opportunity to use their IBM 7094 computer to do the calculations required for this study. R1tSUM1 FONDEMENT tPID1MIOLOGIQUE DE L'ERADICATION DE LA TUBERCULOSE: 11. MORTALITt PARMI LES MALADES TUBERCULEUX ET DANS L'ENSEMBLE DE LA POPULATION AU GROENLAND Un vaste programme destine 'a transformer le Groen- land en une collectivite moderne en assurant son develop- pement dans tous les domaines a e mis a execution en 1950. Parmi ses principaux objectifs figurait la lutte contre la tuberculose qui exerqait alors ses ravages. Un des moyens mis en aeuvre a e le lancement, en 1955, d'un bateau equipe en centre itinerant de depistage des affections respiratoires grace auquel on a pu procder chaque annee a l'examen radiologique et bacteriologique d'une grande partie des habitants. La presente etude est basee en partie sur les statistiques officielles de demogra- phie et de mortalite et en partie sur les donnees recueillies par l'observation pendant 9 ans de la population (23 179 personnes) recensee lors de la premiere tournee du centre itinerant. Jusqu'en 1950, au Groenland, la mortalite a e pro- fondement influence par les graves epidemies de grippe et de coqueluche qui frappaient les collectivites a inter- valles variables. En 1949, par exemple, le taux de la mor- talite brute a atteint 41 pour 1000. Il s'est abaisse ensuite en 1950 et 1951 a un niveau (24 pour 1000) considere comme un taux interepidemique normal. A partir de ce moment, la mortalite a decline chaque annee de faeon frappante pour n'tre plus en 1960 que de 8 pour 1000, taux qui ne variera plus par la suite que dans une faible mesure. La baisse de la mortalite durant les annees 50 a eu comme corollaire une augmentation de l'esperance de vie qui est passee de 39 a 62 ans. La mortalite par tuberculose, qui representait a l'origine 1/3 des decs, a e pratique- ment reduite a zero et la mortalite par autres maladies a diminue de 50 %. On a note par ailleurs une augmentation du nombre des decs par accidents qui contribuent actuellement pour 30% a la mortalite globale. La population etudiee a et repartie en deux groupes: 2072 personnes atteintes ou ayant souffert de tuberculose pulmonaire et 21 107 personnes indemnes de la maladie et sans antecdents tuberculeux. Pour l'ensemble de la periode 1956/64, les taux de mortalite parmi les sujets ages de plus de 15 ans ont atteint 13,4 pour 1000 dans le Ier groupe et 8,5 pour 1000 dans le second. Cette difference est due en partie aux deees par tuberculose et en partie a une surmortalite de 50% due a d'autres causes. Il semble que l'infection tuberculeuse diminue la resis- tance des sujets atteints a 1'egard d'autres maladies. Dans le groupe des malades tuberculeux, la surmorta- lite a ete particulierement elevee chez les sujets qui, a l'epoque du diagnostic, etaient porteurs de lesions bila- terales et de cavites, et relativement faible chez ceux qui ne presentaient que des lesions unilaterales sans cavite. Chez les personnes indemnes de tuberculose, mais mon- trant des images suspectes lors de l'examen radiologique initial, la mortalite a e de 100 % superieure a la normale. 682 EPIDEMIOLOGICAL BASIS OF TUBERCULOSIS ERADICATION: 11 683 REFERENCES Bech, V. (1965) Arch. ges. Virusforsch., 16, 53-56 Buhl, K. & Nyboe, J. (1967) Bull. Wld Hlth Org., 37, 907-925 Christensen, P. E., Schmidt, H., Jensen, O., Bang, H. O., Andersen, V., & Jordal, B. (1952) Acta med. scand., 144, 313-322 Christensen, P. E., Schmidt, H., Bang, H. O., Andersen, V., Jordal, B., & Jensen, 0. (1953) Acta med. scand., 144, 430-454 Denmark, Danmarks Statistik (1967) Befolkningens be- vaegelser 1965, Copenhagen Denmark, Landslkgen i Gronland (1951-1965) Sund- hedstilstanden i Gronland (annual reports), Copenhagen Denmark, Ministeriet for Gr0nland (1957) Diagram over fodselshyppighed og dodelighed 1907-1955, Beretninger vedr0rende Gr0nland Nr. 1, Copenhagen, p. 82 Denmark, Ministeriet for Gr0nland (1960a) Statistiske oplysninger on udviklingen i Gronland 1948-1958, Beret- ninger vedr0rende Gr0nland Nr. 6, Copenhagen Denmark, Ministeriet for Gr0nland (1960b) Den gron- landske befolknings dodelighed 1948-58, mimeographed report, Copenhagen Denmark, Statistiske Departement (1905) Befolknings- forholdene i Danmark i det 19. drhundrede, Copenhagen Denmark, Statistiske Departement (1963) Folke- og boligtallingen 1960, B (1), Copenhagen Denmark, Statistiske Departement (1965) Folketwllingen i Gronland 1960, Copenhagen Denmark, Statistiske Departement (1966) Befolknings- udvikling og sundhedsforhold 1901-60, Copenhagen Denmark, Sundhedsstyrelsen (1961) Dodsdrsagerne i Kongeriget Danmark 1960, Copenhagen Groth-Petersen, E., Gad, U., & 0stergaard, F. (1960) Am. Rev. resp. Dis., 81, 643-652 Horwitz, 0. & Knudsen, J. (1960) Bull. Wld Hlth Org., 23, 669-681 Horwitz, O., Payne, P. G. & Wilbek, E. (1966) Bull. Wld Hlth Org., 35, 509-526 Horwitz, 0. & Wilbek, E. (1971) Amer. Rev. resp. Dis., 103, in press Iversen, E. (1967) Bull. Wld Hlth Org., 37, 893-906 Stein, K. P. S., Krebs Lange, P., Gad, U., & Wilbek, E. (1968) Arch. environm. HIth, 17, 501-506 United Nations (1969) Statistical Yearbook 1968, New York Annex CALCULATION OF DEATH RATES IN THE FOLLOW-UP STUDY Appendix Tables 3 and 4 show the year-by-year size of the tuberculous and non-tuberculous popu- lations, the annual flow of new tuberculosis cases into or out of these populations, and the annual num- ber of deaths. The data are given separately for different cohorts-i.e., persons belonging to speci- fied age groups at the beginning of the follow-up. Let the initial size of the youngest cohort in the tuberculous population be LOT (0,4), the inflow to the cohort of new cases during the first year of ob- servation Cl (0,4), and the outflow of deaths D1T (0,4). By the end of the year, the size and age of the cohort is L1T (1,5) = LOT (0,4)+ C1 (0,4)-D1T (0,4). For the non-tuberculous population we have, simi- larly, LlNT (1,5) = LONT (0,4)-Cl (0,4)-D1NT (0,4). Corresponding age-specific annual death rates could now be calculated as T(0,4) = T (O4)+2DT(0,4)q1T(0 LOT (0,4)+½11C,(0,4) and NT (0,4 D1NT (0,4) q1 (0,4) LONT (0,4)-'/2 C, (0,4) The formulae for other cohorts and observation years are analogous. Age-standardized rates, expressing the general mortality among the tuberculous and non-tubercu- lous populations during each observation year, have been constructed as weighted averages of the age- specific rates. For both populations, the initial age composition of the total study population was used as a weight. As the material gives no information on the mor- tality in the youngest ages during the later obser- vation years-the study population aged during the follow-up, and no new birth cohorts were included in the study-the standardized rates could not cover the whole span of life. It was decided, therefore, to calculate rates representing the mortality for persons above 15 years of age, the formulae for the tuberculous population being 10 E. IVERSEN qll = [L (15,24).qI(I5,24)+ ...] /L (l5+) q,T = [L (1(). q2T (15)+L (16,25) q12T(16,25)+...]/L (15+) qgT = [L (15,22) . qgT (15,22)+L (23,32). q,T(23,32)+...] / L (15+) where L is the size of an age group in the initial total study population. For convenience, the death rates in the first column to the right of the equation mark, not readily available from Appendix Table 3, were taken to be equal to the rates in the second column. Appendix table 1. Population size, total number of deaths, and crude death rates in Greenland during the period 1948-65, according to the cause of death Year Mid-year Number population of deaths 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 21 987 22 113 22 318 22712 23 111 23 607 24 179 24 866 25 706 26712 27 709 28 690 29 791 30 841 31 796 32 847 34 058 34 922 504 910 537 548 473 392 384 371 347 329 284 278 247 284 344 269 321 327 Deaths per 1 000 population all causes 22.9 41.2 24.1 24.1 20.5 16.6 15.9 14.9 13.5 12.3 10.2 9.7 8.3 9.2 10.8 8.2 9.4 9.4 tuberculosis 8.7 8.0 5.0 4A 2.8 1.0 0.6 0.9 0.3 0.4 0.2 0.2 0.1 0.1 other diseases 13.5 10.9 9.9 9.8 10.0 9.6 8.7 7.4 6.2 6.1 6.5 8.1 5.7 6.2 6.7 accidents 1.9 1.6 1.7 2.0 2.1 2.9 3.0 1.9 3.2 1.8 2.5 2.5 2.5 3.1 2.6 _~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 684 EPIDEMIOLOGICAL BASIS OF TUBERCULOSIS ERADICATION: 1 1 Appendix table 2. Age-specific death rates, in Greenland, 1 948-65 Year 0 0 1948 135.7 1949 373.0 1950 137.1 1951 121.0 1952 113.5 1953 98.3 1954 115.1 1955 87.4 1956 86.6 1957 98.2 1958 78.7 1959 76.5 1960 71.5 1961 65.9 1962 87.3 1963 70.5 1964 69.9 1965 77.1 41.6 130.5 40.4 33.0 24.4 27.4 19.2 17.3 13.5 11.6 7.6 3.3 5.4 8.7 6.7 5.6 5.5 4.4 2-4 16.6 45.1 13.5 7.7 7.9 7.6 6.9 9.2 4.1 4.2 2.0 1 .6 2.7 2.8 2.6 2.0 2.9 4.5 5 .9 13.7 16.8 8.5 5.5 4.6 4.3 1.9 2.7 1.7 1.1 1.3 1.2 1.4 1.9 1.8 0.6 1.3 1.4 Age (years) 10-14 15-19 20-34 35-44 45-54 5.2 5.6 5.6 4.0 3.5 5.5 2.0 2.3 1.9 0.3 1.9 0.9 1.5 1 .1 1.1 1.2 1.2 11.4 13.2 9.2 8.3 5.4 6.6 4.8 3.6 5.8 4.4 4.2 3.8 3.3 1.0 1.0 3.6 5.3 3.3 13.1 18.1 18.8 18.4 12.1 9.0 7.8 5.2 5.7 6.9 3.6 4.9 2.0 5.4 5.2 3.3 5.1 3.5 20.9 18.0 21.2 26.2 23.6 17.4 14.8 14.0 12.7 8.8 10.7 7.1 6.6 6.8 5.8 6.1 6.1 4.2 10.5 6.6 9.0 10.2 12.6 9.7 9.7 a Infant mortality: deaths under 1 year of age per 1 000 live births. Other age groups: number of deaths per 1 000 population. 685 55-64 65- 40.2 45.9 47.6 60.6 44.0 35.5 34.3 47.1 32.4 20.0 25.1 22.5 13.9 18.7 23.6 16.2 19.3 18.6 107.4 122.7 94.1 151.8 148.5 95.4 118.7 127.5 121.8 79.0 70.6 59.7 60.2 63.8 96.6 53.6 64.1 79.5 E. IVERSEN Appendix table 3. The tuberculous population, according to follow-up year and age 5-14 D L C 2 160 25 1 183 69 250 39 287 51 - 338 44 - 382 28 410 29 - 436 17 1 450 14 Total 991 55 4 2 896 316 Initial age (years) 15 24 D L C 2 512 65 2 574 78 2 643 58 695 58 750 49 797 28 3 824 38 3 857 24 1 879 13 1 3 6 531 411 25-34 D L C D L: population atbeginning of follow-up year 3 555 54 7 C: inflow of new cases 9 602 75 5 of respiratory tuberculosis 6 672 38 8 D: outflow of deaths 3 702 36 5 2 733 32 5 1 760 33 4 5 789 29 6 2 812 20 5 2 827 23 7 33 6 452 340 52 Follow- up year 35-44 L C 1 359 41 2 388 54 3 431 38 4 464 35 5 491 19 6 506 25 7 525 25 8 543 15 9 550 12 45-54 CD 12 224 30 1 1 251 46 5 293 17 8 306 1 7 4 317 14 6 325 1 4 7 332 18 8 335 8 4 338 7 55-64 D L C 3 1 32 24 4 147 34 4 175 16 6 184 14 6 187 8 7 190 5 15 189 10 5 185 1 8 180 2 65- D L C 9 52 9 6 52 9 7 50 10 11 51 4 5 54 4 6 53 - 14 48 2 6 41 1 9 39 1 Total 4257 264 65 2721 171 58 1 569 114 73 440 40 All ages CD D 9 2072 265 47 11 2 290 375 49 9 2 616 222 41 1 2 797 223 34 5 2 986 175 27 5 3 134 133 29 9 3 238 1 55 59 3 13334 88 32 5 3 390 75 37 57 25857 1 711 355 686 Follow- up year 0-4 C 2 3 4 5 6 7 8 9 78 93 102 108 116 121 121 125 127 17 10 6 8 5 4 2 3 EPIDEMIOLOGICAL BASIS OF TUBERCULOSIS ERADICATION: II Appendix table 4. The non-tuberculous population, according to follow-up year and age Follow- up 0-4 year L C 1 3991 17 2 3 945 10 3 3 928 6 4 3916 8 5 3 902 5 6 3 892 - 7 3 882 4 8 3871 2 9 3 866 3 D 29 7 6 6 5 10 7 3 2 Initial age (years) 5-1 4 15-24 C D L C 5 848 25 5 809 69 5 731 39 5 684 51 5 624 44 5 568 28 5 534 29 5497 17 5470 14 Total 35 193 55 75 50 765 316 14 9 8 9 12 6 8 10 24 3 959 65 3 872 78 3 775 58 3 703 58 3 632 49 3 572 28 3 530 38 3 477 24 3441 13 100 32 961 411 25-34 D C D L: population at _ __ - beginning of - --- - . __- follow-up year 19 14 13 11 14 15 12 13 2 703 54 2 633 75 2 545 38 2 497 36 2 451 32 2411 33 2 364 29 2 327 20 2 299 23 133 22 230 340 16 13 10 10 8 14 8 8 9 96 C: outflow of new cases of respiratory tuberculosis D: outflow of deaths 45-54 D L C 10 14 13 9 11 6 12 14 8 1 413 30 1 362 46 1 307 1 7 1 272 1 7 1 244 1 4 1 218 14 1 192 18 1 163 8 1 141 7 55-64 D L c 21 9 18 11 12 12 11 14 13 915 24 858 34 807 1 6 776 1 4 739 8 716 5 693 1 0 657 1 637 2 65- D L c 33 17 15 23 15 18 26 19 25 416 9 364 9 333 1 0 293 4 265 4 241 - 220 2 190 1 165 1 D 43 22 30 24 20 21 28 24 21 All ages L C D 21 107 265 188 20 654 375 110 20 169 222 114 19833 223 105 19 505 175 94 19 236 133 101 19002 155 115 18 732 88 104 18540 75 115 Total 15032 264 97 11 312 171 121 6798 114 191 2487 40 687 Follow- up year 2 3 4 5 6 7 8 9 35-44 L C 1 862 41 1 811 54 1743 38 1692 35 1648 19 1618 25 1587 25 1550 15 1521 12 233 176778 1 711 1 046