Tube&e 65 (1984) 8>91 @ Longman Group Ltd.

CHANGES IN TUBERCULOSIS NOTlFlCATlOtU BATES IN ETHNIC GROUPS IN ENGLAND BElWEEN 1971 and 1978/79

Ian Sutherland

MRCBiostatistics Unit, Medical Research Council Centre, Hills Road, Cambridge C622QH

V. H. Springett

Birmingham

A. J. Nunn

MRC Tuberculosis and Chest Diseases Unit, Brompton Hospital, Fulham Road, London SW3 6HP

Summary Between special surveys in 1971 and 1978179 the estimated annual tuberculosis notification rate for males in England fell from 30.5 per 100 000 to 22.6 per 100 000 (a decline of 3.8 % per year). For females the rate fell from 18.4 per 100 000 to 15.8 per 100 000 (1.9 % per year).

In each survey the lowest rates were those for the white ethnic group born in the United Kingdom. The highest rates, some more than 50 times as great, occurred in immigrants from the Indian sub-continent (Indian or Pakistani/Bangladeshi). The rates for immigrants from the West Indies were 3 to 4 times as great as than those for the white group.

The most rapid reductions in rate between the surveys, of about IO % per year, occurred in West Indian immigrants of both sexes and in Pakistani/Bangladeshi male immigrants. The rate for Pakistani/Bangladeshi females fell by 6.5 % per year. For whites born in the U.K. the annual rate of decline was 5.1 % in each sex. There was very little change for Indian immigrants of either sex.

Between the surveys, continued immigration of groups from the Indian sub- continent with high notification rates considerably slowed the decline in notification rate for the whole population.

The steep downward trend in notification rate for the white ethnic group may be expected to continue, but changes in the other ethnic groups are more difficult to assess because they are influenced by so many uncertain factors. In addition, the trends in the non-white ethnic groups born in the United Kingdom cannot yet be ascertained, but will become of increasing importance.

Risum6 Entre les enquetes speciales de 1971 et 1978-79, le taux annuel estime des cas declares ches les hommes en Angleterre tomba de 30,5 pour 100 000 a 22,6 pour 100 000 (une diminution de 3,8 % par an). Pour les femmes, la taux tomba de 18,4 pour 100000 a 15,8 pour 100000 (1,9% par an).

Dans chaque enquete, les taux les plus bas dtaient ceux du groupe ethnique blanc ne dans le Royaume-Uni. Les taux les plus hauts, certains plus de 50 fois plus eleves, apparurent chez les immigrants du sous-continent indien (indiens ou

84 Sutherland and others

pakistanais/bangladeshis). Les taux pour les immigrants des Caraibes furent 3 a 4 fois plus eleves que ceux concernant le groupe des blancs.

Entre les deux enquetes, les reductions les plus rapides des taux, d’environ IO % par an, se produisirent chez les immigrants des Carai’bes des deux sexes at chez les immigrants males du Pakistan et du Bangladesh. Chez les femmes du Pakistan et du Bangladesh, le taux baissa de 6,5 % par an. Pour les blancs nes au Royaume-Uni, le taux annuel de declin fut de 5,l % dans chaque sexe. II y a eu tres peu de changement pour les immigrants in&ens de chaque sexe.

Entre les deux enquetes, il y a eu une immigration importante et permanente de groupes venant du sous-continent indien, groupes parmi lesquels les taux de cas declares sont eleves. Ce fait a considerablement ralenti la diminution du taux des cas declares pour I’ensemble de la population.

On peut esperer que la forte diminution du taux des cas declares parmi les groupes ethniques blancs se poursuivra; par contre des modifications de taux dans les autres groupes ethniques sont plus difficiles a evaluer; elles sont en effet influencees par un grand nombre de facteurs difficiles a prevoir. De plus, la tendance evolutive des taux parmi les sujets des groupes ethniques non blancs nes au Royaume-Uni ne peut pas etre encore appreciee, mais son influence deviendra de plus en plus importante.

Resumen Entre las encuestas especiales realizadas en 1971 y en 1978/79, la tasa anual de notification de la tuberculosis en lnglaterra disminuyo en 10s sujetos del sexo masculine de 30.5 por 100 000 a 22.6 por 100 000 (disminucion de 3.8 % por ario). En el sexo femenino la tasa bajb de 18.4 por 100 000 a 15.8 por 100 000 (1.9 % por aho).

En cada encuesta las tasas mas bajas fueron las del grupo etnico correspondiente a 10s blancos nacidos en el Reino Unido. Las mas altas, algunas de las cuales fueron 50 veces mas elevadas, se encontraron en 10s inmigrantes procedentes del subcontinente lndio (India o Paquistan/Bangladesh). Las tasas de 10s inmigrantes del Caribe fueron 3 6 4 veces mas elevadas que las del grupo blanco.

Las reducciones m&s rapidas de las tasas entre las dos encuestas, de alrededor de IO % por ario, se observaron en 10s inmigrantes del Caribe, de ambos sexos y en aquellos de Paquistan/Bangladesh, de sexo masculine. La tasa de las mujeres procedentes de Paquistan/Bangladesh, bajb en 6.5 % por ario. En 10s blancos, nacidos en el Reino Unido, la disminucion anual de la tasa fue de 5.1 % en ambos sexos. En 10s inmigrantes de la India hubo muy pocas modificaciones en ambos sexos.

La continua y substantial inmigracion de grupos provenients del sub-continente Indio, con tasas elevadas de cases declarados de tuberculosis, ocurrido entre las dos encuestas, ha producido una lentitud en la disminucion de la tasa de cases declarados de la poblacidn total.

La pronunciada tendencia decreciente de la tasa de cases notificados en 10s grupos blancos es muy probable que continue, pero las modificaciones en 10s otros grupos etnicos son mas dificiles de evaluar a causa de la influencia de muchos factores dificilmente previsibles. Ademas, no se puede alin determinar la tendencia en 10s grupos Btnicos no blancos nacidos en el Reino Unido pero se piensa que esta adquirira cada vez m&s importancia.

Tuberculosis in ethnic groups 85

Introduction

Between 1971 and 1979 tuberculosis notification rates for males in England and Wales fell from 30.0 to 22.3 per 100 000 per year, an annual decline of 3.6 %; for females the notification rates fell over the same period from 18.2 to 15.5 per 100 000 per year, an annual decline of 2.0 % (based on data supplied by the Office of Population Censuses and Surveys-OPCS). These rates for England and Wales are for the total resident population, combining all ethnic groups. From the routinely collected notification figures it is not possible to obtain separate rates for ethnic or immigrant groups, though it is known from special surveys made in 1971[1] and 1978/79[2] that immigrants from the Indian sub-continent made a disproportionately large contribution to the national notification rates at both times. It is also known that there was substantial immigration from the Indian sub-continent between 1971 and 1979.

A comparison of the results of the two surveys should give information on the changes in rates occurring in separate ethnic or immigrant groups. However, the two surveys used quite different criteria and methods, and it is not possible to make a valid assessment of the changes directly from the published reports. The major problem is that, whereas in the 1978/79 survey patients were classified by ethnic groups, in the 1971 survey they were classified only by place of birth.

The records of both surveys have been made available to the authors of this paper and parts of the data have been re-analysed to study the changes in notification rates between the surveys in similarly defined groups. Attention has been directed particularly to the groups making the largest contributions to the total of notifications.

Methods Four specific population groups have been studied because they can be identified with sufficient accuracy in both surveys to make a valid comparison. They are the white ethnic group born in the United Kingdom (U.K.) and the Indian, the Pakistani/Bangladeshi and the West Indian ethnic groups born outside the U.K.

Population estimates

The population estimates on which the 1978/79 survey rates are based are derived from the National Dwelling and Housing Survey131 (NDHS) carried out in 1978. Information on ethnic groups is given only for England in the NDHS, so the population estimates for both surveys in this paper are for those resident in England at the time of the surveys. The population estimates on which the 1971 survey rates are based are derived from the census held on the night of April 26726, 1971.

In the NDHS individuals were classified by a member of their household into one of 11 ethnic groups, which included white, Indian, Pakistani, Bangladeshi and West Indian as well as other categories. Place of birth was recorded as U.K. or elsewhere.

The census of 1971 gives no direct information on ethnic groups, but does record place of birth of persons enumerated and of their parents. This information on place of birth has been used as described below to estimate the sizes of the principal ethnic groups in the population of England in 1971, which are as nearly as possible comparable to the estimates for 1978779.

In deriving both sets of estimates, groups with incomplete information have been distributed proportionately among the groups for whom full information is available.

86 Sutherland and others

Ethnic groups identified among subjects resident in England

1. White ethnic group born in the U.K. For 1971, born in the U.K. with one or both parents born in the British Isles or the Old Commonwealth (Australia, Canada, New Zealand), but neither in the New Common- wealth (all other Commonwealth countries). For 1978/79, born in the U.K. and classified as of white ethnic origin in the NDHS.

2. Indian ethnic group born outside the U.K. For 1971, born in India or Kenya with one or both parents born in the New Commonwealth. For 1978/79, born outside the U.K. and classified as of Indian ethnic origin in the NDHS.

3. Pakistani/Bangladeshi ethnic group born outside the U.K. For 1971, born in Pakistan (which then included the area now known as Bangladesh) with one or both parents born in the New Commonwealth. For 1978/79, born outside the U.K. and classified as of Pakistani or Bangladeshi ethnic origin in the NDHS.

4. West Indian ethnic group born outside the U.K. For 1971, born in the West Indies with one or both parents born in the New Commonwealth. For 1978/79 born outside the U.K. and classified as of West Indian ethnic origin in the NDHS.

5. Others All those not included in the 4 groups defined above, and therefore including those of white ethnic origin born outside the U.K. and those of Indian, PakistanVBangladeshi or West Indian ethnic origin born in the U.K., as well as those belonging to other ethnic groups, or of mixed ethnic origin.

Notifications of tuberculosis

Patients notified with tuberculosis in the 1971 and 1978/79 surveys and resident in England were classified into the above 5 groups on the same basis as the populations.

Denotifications and renotifications could not be identified in the 1971 survey, so these have been included in the 1978/79 results. In 1971 there was no evidence of notifications solely because of chemoprophylaxis, but there were a number of such notifications in 1978/79 which have been excluded from the present study. Duplicate notifications have been removed from both surveys.

To calculate annual notification rates, the number of notifications reported in each survey has been scaled up by the ratio of the number of notifications reported to the OPCS in a full year, to the number recorded during the survey period. The midpoint of the year used for scaling up the 1971 survey was the end of March 1971 and for the 1978/79 survey it was the end of December 1978, these corresponding closely to the midpoints of the survey periods. The interval between the surveys was therefore 7.75 years.

These procedures have been adopted to obtain the best possible comparison between the two surveys and the rates calculated here are relevant only to this comparison. Changes in notification rates in the 7.75 years between the surveys have been expressed for comparative purposes as annual percentage changes.

Tuberculosis in ethnic groups 87

Results

Estimated populations In Table I the population estimates for the 4 defined ethnic groups, the residual group and all groups combined are shown separately for each sex in 1971 and 1978179. There was a small decrease in the total number of males, a rather larger decrease in the number of females. The change in population for each group is also shown; there were increases in the 3 main immigrant groups, which were greater among females than among males.

The largest group, more than 92 % of the total population on both occasions, is the white group born in the U.K. White males and white females both showed a small decrease in population size. The next largest group, and the only other to decrease in size, is the residual group, representing 5 to 6 % of the total population. This is an extremely heterogeneous group, incl,uded only for completeness. It is unfortunately not possible to obtain satisfactory estimates of the size of the different groups contributing to it.

The Indian, PakistanVBangladeshi and West Indian immigrant ethnic groups each represent less than 1 % of the total population. The increases in these populations, as a proportion of the population in 1978/79, give an approximate indication of the extent to which immigration was occurring between the surveys. The influx was largest for PakistanVBangladeshi females, half of whom entered in the period between the surveys, compared with about one quarter for Indian and West Indian females, and about one-fifth of males in each of these 3 groups. The proportions of the 1971 populations entering during the 7 years prior to 1971 can be estimated similarly from the census data. Again the largest influx occurred among Pakistani/Bangladeshi females, about four-fifths of whom had entered during the preceding 7 years, as had half the PakistanVBangladeshi males. This compares with half of Indian females, two-fifths of Indian males, one-third of West Indian females and one-fifth of West Indian males. All these ethnic groups were thus still undergoing substantial change by continuing immigration at the time of both surveys, to a greater extent for some groups than others. By comparison the white ethnic group was a very stable population, the total changing by less than 1 % between the surveys.

The age distributions of the white and non-white ethnic groups (not tabulated here) show substantial differences. The non-white immigrant groups include relatively few older persons, the great majority being under 45 years of age.

Table I. Estimated populations of identifiable ethnic groups in England in 1971 and 1978l79.

Ethnic group Ma/es Females

Estimated population Difference Estimated population Difference in thousands in thousands in thousands in thousands 1971 1978/79 (1978/7%19711 1971 1978/79 (1978/79-1971)

White born in U.K. 20 543 20 535 -8 21 894 21 642 -252 Indian born abroad 180 214 +34 146 192 +46 PakistaniBangladeshi

born abroad 97 120 +23 35 70 +35 West Indian

born abroad 118 152 +34 119 159 +40 Others 1417 1315 -102 1469 1381 -88

All groups 22 355 22 335 -20 23 663 23 444 -219

88 Sutherland and others

Survey notifications A direct comparison of the total number of notifications received in each survey is not helpful as the 1971 survey covered 4 calendar months whereas that in 1978/79 covered 6 calendar months. Table II therefore shows the mean number of notifications per calendar month in each of the groups studied.

The total number of notifications received each month fell from 508 to 399 for males, and from 324 to 291 for females; however the numbers increased in some ethnic groups and decreased in others.

Notification rates Table III shows the estimated annual notification rates by sex in each ethnic group in the two surveys. The rates for the white group are considerably lower than those for the immigrant groups, some of which are more than 50 times as great as the corresponding rates for the whites born in the U.K.

Table II. Mean monthly number of tuberculosis notifications for identified ethnic groups in England in the 1971 and 1978/79 surveys.

Ethnic group Males

Notifications per calendar month 1971 1978/79

White born in U.K. 325 229 Indian born abroad 56 68 PakistanilBangladeshi

born abroad 64 37 West Indian

born abroad 9 5 Others 54 60

Difference (197@79-1971)

-96 195 135 -60 +12 62 85 +23

-27 29 36 +7

-4 5 3 -2 +6 32 32 0

Females

Notifications per calendar month 1971 1978I79

Difference (1978/7%1971)

All groups 508 399 -109 324 291 -33

Table III. Estimated annual tuberculosis notification rates for identified ethnic groups in England in 1971 and 1978/79.

Ethnic group

Annual notification rate per 100000 1971 1979/79

Change in rate (percent per year)

White born in U.K. Indian born abroad PakistanilBangladeshi

born abroad West Indian

born abroad Others

21.2 14.1 -5.1 414.1 402.1 -0.4

880.0 390.7 -9.9

102.2 44.6 -10.1 51.7 57.6 +1.4

Females

Annual notification rate per 100000 1971 197w79

11.9 7.9 572.8 562.9

1109.0 657.6

56.6 25.3 -9.9 29.5 29.1 -0.2

Change in rate (percent per year)

-5.1 -0.2

-6.5

All groups 30.5 22.6 -3.8 18.4 15.8 -1.9

Tuberculosis in ethnic groups 89

The highest rates recorded are for the Pakistani/Bangladeshi groups, with annual rates over 800 per 100 000 for males and over 1100 per 100 000 for females in 1971, falling to under 400 for males and under 700 for females in 1978/79. The rates for the Indian ethnic groups are also very high, at just over 400 per 100 000 for males and over 500 per 100 000 for females in each survey.

In contrast the rates for West Indians (and for the residual group) are much lower, though above the rates for the white group.

Changes in notification rates The annual percentage change in notification rate between the 1971 and 1978/79 surveys is also shown in Table III. For all ethnic groups combined the rates declined by 3.8 % per year for males and by 1.9 % per year for females. These annual percentage decreases between the surveys are similar to the decreases shown by the national notification rates for England and Wales over a similar period, quoted at the beginning of this report (see also (41).

The most rapid decreases occurred in immigrant groups, about 10 % per year in West Indians of each sex and in Pakistani/Bangladeshi males and 6.5 % among the Pakistani/ Bangladeshi females. In the white group the annual decline was 5.1 % for each sex. The Indian immigrant group, in contrast, showed only a slow decrease of less than 0.5 % per year for each sex.

The effect of combining results for different groups

Because of the wide range of notification rates and the different sizes of the ethnic groups, combining the figures may lead to apparently anomalous results. For example the result of combining the figures for Pakistani/Bangladeshi females with those for all other females is shown in Table IV. The annual percentage decline in rate for Pakistani/Bangladeshi females is 6.5 % per year. For all other females combined the annual rate of decline is 2.4 % per year. Combining these two groups gives an annual decline of 1.9 % per year, which is less than the annual decrease for either of the groups considered separately. The explanation is that the doubling in size of the population of Pakistani/Bangladeshi females between the surveys outweighed the decrease in their high notification rate, so that there were more cases among them in 1978/79 than in 1971 (Table II). When this group is combined with all other groups, the increase in the number of cases diminishes the annual decline in the notification rate for all groups. Thus, the change in rate for the combined group is influenced not only by the changes in rate in the subgroups but also by changes in the relative size of those subgroups.

Table IV. Changes in estimated annual notification rates in Bangladeshi/Pakistani females and in other females in England between 1971 and 1978/79.

Ethnic group 1971 7979/79 Change in rate

Estimated Survey Annual Estimated Survey Annual (percent population notifications notification population notifications notification per year) in thousands rate in thousands rate

per 100 000 per 100 000

PakistaniIBangladeshi born abroad 35 117 1109.0 70 218 657.6 -6.5

All others 23 628 1177 16.7 23 374 1520 13.8 2.4

All groups 23 663 1294 18.4 23 444 1746 15.8 -1.9

90 Sutherland and others

The way in which the changes in size and the changes in rate in the subgroups affect the combined rate is illustrated most simply by the changes in the monthly numbers of notifications in Table II. The main determinant of the downward trend in the combined rate is the white ethnic group, in which there was a substantial downward trend in notification rate with virtually no change in population size. The decline in the combined rate is slowed by the experience of the Indians of both sexes and by the PakistanVBangladeshi females, in whom the increases in population had a greater effect than the decreases in their notification rates. For PakistanIBangladeshi males, however, the substantial decline in notification rate outweighed the increase in population, thus slightly accelerating the decline in the combined group of all males. The West Indian and the residual groups exerted only very small influences on the combined rates.

The magnitude of the effect of continued immigration from high incidence areas on the annual decline in notification rate for the combined population can be assessed by direct standardisation for ethnic group, i.e. by applying the notification rates for the ethnic groups in 1971 and 1978/79 to a population of unchanged ethnic constitution. Using the mean of the 1971 and 1978/79 populations as the standard, the standardised annual rate of decline for males is 4.3 %, compared with 3.8 % without standardisation. In females the standardised annual rate of decline is 3.7 %, compared with 1.9 % without standardisation.

Discussion The observations on which this paper is based are drawn from surveys of tuberculosis notifications in 1971 and 1978779, the results of which have already been published.[l,2] The main finding of each of those surveys was the very much higher notification rates in groups from the Indian subcontinent than in other immigrant or non-immigrant groups. The object of this paper is to relate the findings of the two surveys in order to study the changes in the notification rates in identified ethnic groups.

An additional finding of the 1971 survey[5] was that the notification rates for immigrant groups tended to be highest for recent arrivals, that is within about 5 years, and became progressively lower for immigrants who had been in the U.K. for longer periods. The observation in the present analysis that the highest notification rates in both surveys occurred in the group with the greatest proportion of relatively recent arrivals is consistent with this observation. Thus in 1971 the highest rate of 1109 was in the PakistanVBangladeshi females of whom four/fifths were recent immigrants, the lowest rate among the Indian sub-continent ethnic groups occurred in the Indian males (414) of whom only two-fifths were recent immigrants. The corresponding figures for 1978/79 were 658 in Pakistani/ Bangladeshi females of whom about half were recent immigrants compared with rates of 402 and 391 in Indian and PakistanVBangladeshi males of whom only one-fifth were recent immigrants.

fhe steep downward trend in notification rate for the life-time resident white ethnic group, which was established before any large-scale immigration from the New Commonwealth occurred, has continued. The present analysis gives direct evidence that notification rates for the identifiable immigrant groups have also declined. The most rapid declines in notification rate between the two surveys, of about 10 % per year, were shown by West Indians of both sexes and by PakistanVBangladeshi males. The next largest decline of 6.5 % per year was shown by Pakistani/Bangladeshi females and this is greater than the 5.1 % per year for the white group, both males and females. However the Indian ethnic group showed only a small decline of less than 0.5 % per year in each sex.

With most of the identified ethnic groups showing substantial decreases in rate, the annual decreases in the notification rate for all groups combined, 3.8 % for males and 1.9 %

Tuberculosis in ethnic groups 91

for females, are less than might be expected. This is the result of combining figures for groups with widely differing rates when the size of the groups with the highest rates is increasing. Continued immigration from high incidence areas, particularly the Pakistani/ Bangladeshi females, thus had a substantial effect on slowing the downward trend in the national rate. In the absence of further immigration between the surveys, it is estimated that the notification rates for all groups combined would have decreased by 4.3 % annually for males and 3.8 % for females.

It is difficult to assess future changes in notification rate. It is likely that the downward trend in the white ethnic group will continue, but the rates in the other ethnic groups will depend on a number of factors. Continued immigration will have effects which will depend not only upon its extent (which has decreased since 1979) but also on the original environment and the age and sex of those entering this country. The long term trends in the rates among the immigrants as they become older, and live in this country longer, cannot yet be ascertained. At the same time children born in the U.K. of immigrant parents are reaching adult ages in increasing numbers. It is known that in 1978179 their notification rates lay between those for children of the white ethnic group and children born abroad[21, but the trends in their rates cannot yet be ascertained. The influence of return visits to countries of origin where the prevalence of tuberculosis remains high is also unknown. There is clearly a need for future studies to estimate trends in ethnic groups, and their effects on the notification rate in the total population.

Acknowledgements

We wish to acknowledge the valuable assistance of Mr Marc Coghlan (MRC Biostatistics Unit) and Miss Deborah Johnson (MRC Tuberculosis and Chest Diseases Unit) who calculated the population estimates for the 1971 and 1978/79 surveys respectively.

References 1 British Thoracic and Tuberculosis Association (1973). A tuberculosis survey in England and Wales 1971; the

influence of immigration and country of birth upon notifications. Tubercle, 54, 249. 2 Medical Research Council Tuberculosis and Chest Diseases Unit (19801. National survey of tuberculosis

notifications in England and Wales 1978-79. British Medical Journal, 2, 895. 3 Department of the Environment (1979). National Dwelling and Housing Survey. London: HMSO. 4 Nunn, A. J., Springett, V. H., 81 Sutherland, I. (1984). Changes in tuberculosis notification rates in the white ethnic

group in England between 1971 and 1978/79. British Journal of Diseases of the Chest, 78, in press. 5 British Thoracic and Tuberculosis Association (1975). Tuberculosis among immigrants related to length of

residence in England and Wales. British Medical Journal, 3, 698.