INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In industrialized countries, incidence of tuberculosis among the foreign born may be 10-20 times higher than among native born and accounts for a growing proportion of all cases (1). The majority of the native-born population do not have tuberculosis infection (3, 4) and so could be susceptible if exposed. There is limited evidence that transmission of tuberculosis occurs from foreign born adults to native-born children within households (5, 6), where risk of transmission is known to be highest (4, 7). Although risk is lower, transmission in community settings could result in more new tuberculosis infections because of the much larger number of persons potentially exposed. There are case reports of transmission in schools, at work, or other community settings from highly contagious individuals (8). Studies using DNA fingerprinting techniques have demonstrated significant, unrecognized, transmission from casual, even chance encounters (9), and one Dutch study suggested that a significant proportion of cases in the native born resulted from transmission from foreign-born (12). Studies of disease incidence may underestimate the extent of transmission because of the low proportion of those infected who will develop disease and the long latency between transmission and disease. We hypothesized that in a young Canadian-born population, positive tuberculin reactions would be associated with indices of potential exposure to active tuberculosis among the foreign-born.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Overall Study Design

A cross-sectional study was conducted, which utilized individual as well as group or community level (ecologic) information, to test the hypothesis that, among Canadian-born children and young adults, prevalence of tuberculin reactions was associated with indicators of exposure to active tuberculosis among foreign-born persons. The health outcome of interesttuberculin reactionswas measured directly among participants along with the two major causes of false positive reactions. Household exposure to tuberculosis was estimated from questionnaire items, and non-household contact was estimated indirectly from group or community level information derived primarily from census data.

Study Population

Four population groups were selected for study: (1) and (2) schoolchildren in grades 6 and 10: There are seven health districts (each with a population of roughly 150,000) on the island of Montreal. Of these, three were selected, on the basis of previously published information (13) so as to include populations with a wide range of socioeconomic status and demographic characteristics. All public schools located within these three health districts were selected for the study, and within these schools all children in grades 6 and 10 were considered eligible; (3) Students in their pre-clinical years in health professional programs in community colleges and universities on the island of Montreal; and (4) Workers aged up to 25 yr in three work forces, selected to represent a spectrum of socio-economic background and proportion of foreign-born: (1) office workers (expected to have higher socioeconomic background than workers in manufacturing industries); (2) heavy manufacturing (in Montreal this work-force has a very low proportion of foreign-born); and (3) garment industry (in Montreal this workforce has a high proportion of foreign-born). Lists of all companies in these three sectors, located on the island of Montreal, were obtained from occupational health officials. Companies on this list were approached for agreement to participate, until an estimated eligible population of 1,000 workers for each of the three work-forces was attained. In each company which agreed to the study all workers up to the age of 25 yr were considered eligible.

Written informed consent was obtained from parents for the grade six children, parents as well as participants for grade ten, and from participants for workers and health professional students. This project was approved by an ethics committee of the Montreal Chest Institute Research Centre.

In each of the study groups, all were invited to participate. Persons in the workplaces and schools surveyed were eligible even if their current and/or permanent home address was off the island of Montreal. The study was conducted among the schoolchildren in 1987-1988, the workers in 1988-1989, and the health professions students in 1989- 1991. Tuberculin test results among the foreign-born (14) as well as impact of BCG vaccination on tuberculin reactions among Canadian-born (15) have been reported elsewhere. This report will be restricted to analysis of the results among the non-BCG vaccinated Canadian born participants.

Tuberculin Testing and BCG Vaccination

Tuberculin skin-testing was performed on all those who gave written consent and did not have a history of prior treatment or a documented prior positive tuberculin test. Over the course of the study, tuberculin testing was performed by two of the authors (B.V. and D.M.), as well as one of six nurses. After 48-72 h the transverse diameter of the induration was demarcated with the ballpoint technique (16), and measured with machinists calipers. Readings were performed by the authors (D.M. and B.V.) or one of three other nurses (from among the six testers). All nurses had many years of experience in tuberculin testing, and to ensure standardized techniques, all testers and readers were re-trained by the principal investigator (D.M.), who also verified initial readings by other readers.

Two types of testing materials were usedfive tuberculin units (5TU) of PPD-T, bio-equivalent to PPD-S (Connaught Laboratories, Toronto, Canada), was given on the right forearm and five tuberculin units of PPD-B (PPD-Battery from Mycobacterium intracellulare, supplied by Connaught Laboratories, Toronto, Canada) was given at the same time on the left forearm. (Health professional students in 1989 and 1990 were not given simultaneous PPD-B.) PPD-B was selected because of the availability of extensive epidemiological information regarding interpretation of these simultaneous reactions (17, 18) and because in previous Canadian surveys reactions to PPD-B were among the most frequent of reactions to nontuberculous mycobacterial antigens (19).

In the provinces of Quebec and Newfoundland, between 1948 and the mid-1970s, BCG vaccination was included in the routine vaccinations given to much of the general population but in the other Canadian provinces has been given only to aboriginal Canadians. Records of participants born in Quebec or Newfoundland were forwarded to the Institut Armand Frappier which has maintained a registry of all those vaccinatedthe accuracy of which has previously been verified (15). For those born in Canada outside of Quebec and Newfoundland, (who represented 8.5% of the total Canadian born study population) it was assumed that no BCG had been received.

Self-Administered Questionnaires

A self-administered questionnaire was completed by all participants regarding prior skin-tests, known household contact with TB, possible travel exposure as well as demographic information regarding their parents. This questionnaire was translated by two persons independently into French and back translated to verify accuracy of translation. It was also translated into six other languagesItalian, Spanish, Portuguese, Greek, Vietnamese, and Chinese. Each of these translations was verified by a second person whose mother tongue was that language.

Census Information

Within all major urban centers in Canada, Statistics Canada has defined census tracts to have average population of 4,000 persons (range 1,500-8,000) and be as homogeneous as possible with respect to social and economic living conditions. In urban areas the Canadian postal service has assigned postal codes that are unique for each side of the street on each city block, or for large buildings, such as apartments with 20 or more family units. Using the postal code conversion file of Statistics Canada, the postal code of the current home address for grade six and ten students, and workers, or permanent address for health professional students was used to assign individuals to the census tract of their residence. Socioeconomic and demographic information compiled by Statistics Canada for each census tract in the greater Montreal census metropolitan area was taken from the 1986 census.

Incidence of Tuberculosis Disease

Information on all reported incident cases of tuberculosis disease on the islands of Montreal and Laval were abstracted in a non-nominal fashion from two sources: the provincial data base for reportable diseases for the years 1986 to 1993, and direct review of public health records from 1982 to 1992 at each of the departments of community health on the islands of Montreal and Laval. Information abstracted included: full postal code of their home address (to be able to assign each case to the census tract of residence), date and country of birth, date of diagnosis, as well as site and bacteriologic status of tuberculous disease.

Data Analysis

Outcome variables. Tuberculous and nontuberculous mycobacterial infections are all or nothing events. Therefore, after taking into account the dual testing results, the reactions for the two tuberculin antigens were dichotomized as negative or positive as follows:

PPD-T considered positive if PPD-T  10 mm AND PPD-T  PPD-B. (17, 20)

PPD-T considered positive if PPD-T  12 mm IRREGARDLESS of PPD-B. (18)

PPD-B considered positive if PPD-B  5 mm AND PPD-B  PPD-T. (17)

Annual risk of infection (ARI) was calculated based on the prevalence of infection in each 3-year interval using the formula suggested by Styblo (21):

Independent variables. Socioeconomic factors were considered covariates that could potentially confound the relationship between tuberculin status and indices of contact with foreign born. Measures of socioeconomic status used in the analysis were the average income or value of housing units in the neighborhood of residents taken from Statistics Canada census data. Household exposure to tuberculosis was ascertained from questionnaire responses regarding known household contact, and estimated from incidence of tuberculosis in the country of birth of the parents. Nonhousehold exposure to tuberculosis was estimated from: (1) group informationthe proportion of foreign born from tuberculosis endemic areas in the same school or work place; and (2) community level informationin the neighborhood of home residence the proportion of persons born in tuberculous endemic areas, as well as actual incidence of active tuberculosis among foreign-born. The countries considered endemic for tuberculosis included: Haiti, all countries in Central America and Africa, all countries in Asia except Japan, and certain countries in South America with reported incidence of all forms of active tuberculosis greater than 100 per 100,000 (22).

Associations with tuberculin skin test reactions were tested for significance using chi-squared tests for categorical and t-tests for continuous independent variables (23). Multivariate logistic regression was performed to estimate the effect of different indices of contact with foreign born and tuberculin reactions after adjusting for age, gender, and other potential confounders. Parameter estimates obtained from logistic regression were used to calculate odds ratios and 95% conference intervals for comparison of (mean value) versus (mean + 1 standard deviation) for continuous variables (24).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

As shown in Table 1, of 10,234 initially eligible participants 7490 (73%) participated. Participation was higher in the grade 6 and health professional students and lowest in grade 10 (adolescent) students. In total, there were 3,710 Canadian-born participants who had never been BCG vaccinated. Census data was available for 3,615 of these who had current and/or permanent residence in the greater Montreal census area. TB incidence data was available for 3,097 with residence on the islands of Montreal and Laval.

As of the 1986 census, there were 716 census tracts in the greater Montreal area. Study participants lived in 626 of these, for an average of 5.7 participants per census area. Based on this census information in the neighborhood where participants lived: average income was $19,009 (interquartile range: 15,754 to 20,778), housing value was $91,065 (66,617 to 100,754), percent of population from TB endemic areas was 6.8% (1.8% to 9.2%), and incidence of tuberculosis was 4.1 per 100,000 among Canadian-born (0 to 5.6), and 24.8 among foreign-born (0 to 34). The study was conducted in 176 schools or workplaces, for an average of 21 participants in each site. On average 12.1% (interquartile range: 0 to 18.7%) of the population at these sites had immigrated from TB endemic countries.

All personal characteristics shown in Table 2 were significantly different between the 4 major population groups and were significantly different between the three subgroups of workers. All indices of socioeconomic status were significantly higher among health professional students. Among the workers, the proportion of foreign-born from tuberculosis endemic areas in the same workforce was significantly higher in the garment industry, as anticipated, compared to the other manufacturing and office workforces.

As shown in Table 3, reactions to PPD-B were relatively uncommon. They had a modest effect on PPD-T reactions through cross-reactivity. Larger simultaneous reactions to PPD-B were seen for 15 (19%) of the 80 reactions to PPD-T in the 5-9 mm range, only two of the 90 (2.2%) reactions to PPD-T of 10+ mm, and none of the reactions to PPD-T of 15 mm or greater. Reactions to PPD-B were not associated with older age, gender, indices of socioeconomic status, nor with visits to the southern United States (a common winter destination of Canadians) although reported travel to this region was usually brief. Positive PPD-B reactions were seen in 6% of the 647 whose parents were born in countries with high or intermediate rates, compared with 3% of the 2,313 whose parents were from Canada, or other low incidence countries (p < 0.001). Information regarding travel to TB endemic areas was not sought in the questionnaires.

Positive tuberculin reactions (PPD-T) were significantly associated with older age, population group and household contact (Table 4), but not with any indices of potential contact with foreign born. As shown in Figure 1, there was a general downward trend in the annual risk of infection for each population group. The annual risk of infection estimated separately for health professional students was significantly lower than the other groups. Over the interval of birth years of participants from 1960 to 1976 the estimated annual risk of infection decreased by 3% per year. As shown in Table 5, after adjustment for age, group, and household contact, tuberculin reactions were not associated with any indices of potential contact, including incidence of reported cases of tuberculosis among foreign-born or Canadian-born on the islands of Montreal and Laval between 1982 to 1993. 

View larger version (K): [in this window] [in a new window]   Figure 1.   Prevalence and annual risk of TB Infection. The annual risk was calculated from prevalence and mean age of groups. Solid line (small squares) = prevalence of tuberculin reactions in Grades 6 and 10 students and workers. Dashed line (crosses) = calculated annual risk of infection in grade 6 and 10 students and workers. Dotted line (stars) = calculated annual risk of infection in health professions students.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Four findings deserve commentthe low prevalence of sensitivity to PPD-B, decreasing prevalence of tuberculin reactions, and the failure to detect associations between tuberculin reactions in the study population, and indices of contact with foreign-born active TB in Canadian-born.

The low prevalence of sensitivity to PPD-B in this young Canadian-born population is consistent with previous surveys in Canada (19) as well as in the northern United States (25). Prevalence of reactions to PPD-B is much higher among residents of the southern United States (25), or immigrants from tropical or subtropical countries (14), which has been attributed to better survival in these warm environments of the responsible mycobacteria (26). Interestingly, participants whose parents had immigrated from TB endemic areas had higher prevalence of sensitivity to PPD-B. Since person to person transmission has never been documented, we speculate that sensitization may have occurred through travel to the parents' country of origin, although no data was gathered regarding such travel in this study. Sensitivity to PPD-B had an important effect on reactions to PPD-T in the range of 5-9 mm, through cross- reactivity, but negligible effect on reactions of 10+ mm, reinforcing the utility of the 10 mm cut-point (20) in regions with similar climate to Montreal.

In this nonvaccinated Canadian-born population, the prevalence of tuberculin reactions was low and the estimated annual risk of infection appeared to be declining slowly. Tuberculin reactions were significantly associated with older age and household contact, although the population attributable risk for such contact was only 5% because this was reported by very few participants. After adjustment for age, health professional students had significantly lower prevalence, which may be explained by their higher socioeconomic status. By contrast, tuberculin reactions were only weakly associated with socioeconomic indicators from census information. The apparent discrepancy may reflect the relative accuracy of measurement, i.e., membership in the health professional student group was a better overall indicator of socioeconomic status than aggregate information about income and housing in the small census area where they lived.

In this study nonhousehold exposure to tuberculosis was estimated from group or community information, which may not accurately reflect individual exposure (27), although such data has been used successfully to examine mortality or cancer incidence of groups (28, 29). The main justification for use of this data is that the primary study question was (as it is commonly expressed): "What is the risk of tuberculosis for me, or my children, from working or attending school, where there are many foreign-born?" The indicators of exposure, based on school and neighborhood information should have provided the best estimates to address this question. Questioning regarding exposure to tuberculosis in the community is unlikely to be accurate since the vast majority of those exposed will not be aware of this; and most transmission in the community goes unnoticed (9). Direct questioning regarding contact with foreign-born was considered too intrusive.

Although these group or community indices of exposure to tuberculosis in the foreign-born seemed the most appropriate, such ecologic variables may result in misclassification of exposure. This would reduce the likelihood of finding a relationship with tuberculin reactions. For example, some participants may have recently changed their home address, school, or work. In addition, colleagues at school or work may not have posed any substantial risk of exposure, particularly true for grade 6 schoolchildren in whom active tuberculosis is seldom contagious (30). However, even in these schoolchildren, exposure could result from contact with older siblings or parents.

Evidence regarding the extent of tuberculosis transmission from foreign to native born is limited as this question has seldom been addressed. Among children born in Britain, those whose parents had immigrated from Asia had 20 times higher incidence of tuberculosis disease than children whose parents were born in Britain (6). However, there may have been significant overdiagnosis among children of Asian ethnic origin because such cases are infrequently microbiologically confirmed (31) and clinical diagnosis of disease in children is notoriously unreliable (30). A recent study in New York found that among caretakers of U.S.-born children aged 0-5 with active tuberculosis 39% were foreign-born (5), although in the same study among 10 adults with documented transmission to children, only one was foreign-born. In a study utilizing small areas analysis, the incidence of tuberculosis disease was higher in the British-born population who lived in neighborhoods with higher proportion of foriegn-born (32), but the authors attributed this to confounding with lower socioeconomic status. A recent study in the Netherlands utilizing RFLP "DNA finger printing" estimated that 16-25% of tuberculosis strains isolated from Dutch-born adults aged 18-50 were strains usually isolated from foreign-born (12). This was interpreted as evidence of transmission from foreign-born to Dutch-born, although the investigators had no idea where or how this transmission was occurring. The majority of the Dutch-born young adults with evidence of recently transmitted tuberculosis (i.e., in clusters of identical strains) were intravenous drug users, HIV-infected, homeless or other urban marginalized populations (12)very different from the population of the present study.

Tuberculin reactions in study participants were not associated with reported cases of tuberculosis disease among Canadian-born living in the same neighborhood. This appears to contradict evidence from studies utilizing DNA finger printing which detected significant transmission occurring in neighborhoods (9). However, these studies may have overestimated the extent of transmission because it was detected predominantly in HIV infected individuals in whom infection may lead to disease rapidly in a high proportion. It is also possible that transmission occurred only within members of the same high risk populations. Within large urban centres in industrialized countries, tuberculosis in the native-born is increasingly concentrated in two groupsthe elderly and the urban poor, the latter meaning unemployed, homeless, ex-prisoners, intravenous drug users and HIV-infected persons (2, 33). Transmission has been documented within these populations (34), but there may be very limited contact between these subgroups and the general population. We failed to detect any association between tuberculin reactions among the young Canadian-born population attending school or working, and incidence of TB disease in Canadian-born. We speculate this may reflect the simple fact that these populations are very different and rarely in contact, therefore minimizing opportunities for transmission. Tuberculin screening of populations in Canada similar to those of the present study would not be recommended since prevalence of positive reactions was very low (20, 35). Such efforts should be targeted to the high risk subgroups (20, 35).

Tuberculosis infection in the young Canadian-born population is now uncommon and the risk appears to be declining slowly. Our results suggest that tuberculosis in high risk groups foreign-born, and Canadian-born elderly or urban poordoes not result in detectable transmission in community settings to young Canadian-born persons who are attending school or employed.