INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Published recommendations for the administration of preventive therapy to individuals with a positive tuberculin skin test (TST) and a clear chest X-ray, who have not had recent exposure to a known source case, are based on assessment of the size of the TST reaction, the prior probability of infection, the likelihood of disease progression or reactivation, and the risks of adverse effects arising from the administration of preventive therapy (1). However, there is evidence to suggest that the practice of administration of preventive therapy in this setting varies (4, 5).

The variation in practice may stem, at least in part, from uncertainty about the contemporary relevance of the data on which these recommendations are based. A number of decision analyses have been undertaken to integrate information on risk of tuberculosis, case fatality rate for tuberculosis, effectiveness of isoniazid chemoprophylaxis, risk of isoniazid-induced hepatitis, and case fatality rate for isoniazid-induced hepatitis (6). Although the conclusions of these analyses differ in important respects because of differences in some of these assumptions, the tuberculosis incidence rates used for most of these analyses are those estimated from discrete populations in Georgia and Alabama by Comstock and Edwards (10). However, there are substantial differences among populations in the reported incidence of tuberculosis in TST-positive persons. Comstock and Edward's population subgroup-specific estimates ranged from 35 to 104 per 100,000 per year (10). Other estimates vary even more widely (8, 12).

BCG vaccination further complicates the interpretation of tuberculin testing for decisions on preventive therapy. Both the relation between TST reaction size and the likelihood of tuberculosis infection and the relation between tuberculosis infection and the risk of subsequent active disease may be influenced by the presence of a BCG vaccination scar.

Australia has one of the lowest rates of tuberculosis in the world: 5.4 per 100,000 persons per year (15). Migrants to Australia probably have a low risk of acquiring tuberculous infection after arrival in Australia (16) and cases arising after migration are likely to represent the reactivation of remotely acquired, latent disease. Several large groups of migrants to Australia have originated from countries with a high incidence of tuberculosis and have been subject to surveillance for tuberculosis after arrival in Australia. It has been common practice not to give preventive therapy to adults and older children with a positive TST and a clear chest X-ray in the absence of a history of recent exposure to a known infectious case. This has given us the opportunity to reexamine existing assumptions about the rates of late reactivation of tuberculosis in people with positive TSTs.

The aim of this study was to estimate the risk or incidence rate for tuberculosis in a general population sample of predominantly Southeast Asian, TST-positive, refugee migrants to Australia. We sought to assess the influence of TST reaction size, presence of a BCG scar, age at migration, and time since migration on the incidence rate for tuberculosis. Assessment of the risk of reactivation of tuberculosis will inform the debate on policy and practice for the administration of preventive therapy in TST-positive migrants from countries with a high incidence of tuberculosis.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The Cohort

A historical cohort study was conducted among 24,653 refugees who were screened for tuberculosis after arrival in Sydney, Australia during the period 1984 to 1994.

All refugees had a chest X-ray performed before departure for Australia. Those in whom currently or recently active tuberculosis was suspected were commenced on antituberculous drug therapy at that time. Soon after their arrival in New South Wales (NSW) all refugees had further screening for tuberculosis at the Refugee Screening Unit or at the Liverpool Chest Clinic. The results of this initial postmigration screening, which comprised a TST, examination for a BCG scar, and a chest X-ray, were recorded in a log book. TST reaction size was recorded for all reactions  10 mm. TST reaction sizes in the range 0 to 9 mm were recorded as "negative." Chest X-rays were recorded as "satisfactory," that is, demonstrating no pleural or parenchymal abnormality, or "not satisfactory."

To exclude those refugees who may have had chemotherapy or preventive therapy for tuberculosis, the cohort for this study was limited to those refugees who were not receiving treatment for tuberculosis and were aged 12 yr and over at the time of initial screening in Australia. The cohort was further limited to those whose chest X-rays at the time of postarrival screening were classified as satisfactory. Refugees meeting these criteria would not have been prescribed preventive therapy by the physician responsible for the refugee screening program during the study period (C. M. Mukerjee, personal communication). However, as there are no comprehensive data available on the administration of prior antituberculous drug therapy, this possibility cannot be excluded for all subjects.

Tuberculosis Notification Data

Cases of tuberculosis arising in the cohort were identified from among those who were notified to the NSW Department of Health Infectious Diseases Surveillance System and its predecessor, the Tuberculosis Register, during the period 1984 to 1998.

Matching Screening and Notification Databases

Members of the cohort who were notified as cases of tuberculosis during the study period were identified by a record linkage analysis using a database of screened refugees and a database of notified cases of tuberculosis (17). A computer-generated search was used to identify records with an identical date of birth in the two databases. These pairs of records were then manually checked for similar names, sex, country of origin, and date of arrival. A further search, based on the surnames and given names, identified a small number of additional matching records with differing dates of birth. These differences appeared to be due to recording errors or incomplete dates. A further computer-generated search based on date of arrival did not identify any additional cases.

Specificity of the Diagnosis of Tuberculosis in Notified Cases

We reviewed notification data, case notes, and X-ray films of the notified cases of tuberculosis identified from the refugee cohort in order to confirm or refute the diagnosis of active tuberculosis. The diagnosis of active tuberculosis was accepted if any of the following were recorded (17):

1. Positive culture for Mycobacterium tuberculosis

2. Positive direct smear for acid-fast bacilli (AFB), and the patient was not notified as "atypical"

3. Histopathological report of caseating granulomas or findings "consistent with TB"

4. Radiological pulmonary infiltrate that regressed after antituberculosis drug treatment

5. A positive TST, clinical features of extrapulmonary tuberculosis, and response to treatment

Integrity of the Cohort in 1998

To be certain that cases of tuberculosis arising in this cohort would have been reported to the NSW Department of Health, we sought to confirm that the members of the cohort were predominantly alive and living in NSW throughout the study period. We attempted to trace the current location of 200 cohort members selected at random from the study population.

Data Analysis

Incidence rates were calculated per person-years of follow-up. Ninety-five percent confidence intervals around the estimated rates were calculated according to the asymptotic method of Fleiss (18).

Subject age at arrival was classified into quartiles. TST reaction size was classified as negative (< 10 mm), 10-14, 15-19, 20-29, or 30 mm and over. The effects of age, TST reaction size, the presence of a BCG scar, and the interactions among these variables were tested by logistic regression.

The study protocol was approved by the Research Ethics Committees of the South Western Sydney Area Health Service and the NSW Department of Health.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Description of the Cohort

The study population comprised 15,489 subjects, which represented 84.3% of all refugees aged more than 12 yr at initial screening. Eighty-six percent of this population originated from Southeast Asia (Vietnam, 67.5%; Cambodia, 10.5%; and Laos, 7.8%) and 12.9% came from Central and South America (El Salvador, 6.3%; and Chile, 3.9%). The remaining 1.3% came from other countries in the Middle East, Europe, and Africa.

The members of this study population were young: the median age was 27 yr, with an interquartile range of 20 to 33 yr. Fifty-eight percent of the cohort were male. A BCG scar was identified in 53.3% of the cohort and the TST was read as positive in 57.7%, including 51% of those with no visible BCG scar and 63% of those with a BCG scar. No TST result was recorded for 505 (3.3%) members of the study population.

Integrity of the Cohort

Of 200 randomly selected members of the original cohort, we were able to establish the whereabouts of 181. Of these, 171 (85.5% of 200) were confirmed as living in New South Wales in 1998.

Incidence of Tuberculosis

The mean duration of follow-up to June 1998 or to the time of diagnosis of tuberculosis was 10.3 yr.

There were 151 notified cases of tuberculosis among members of the study cohort. Of these, 122 (81%) were confirmed as cases of tuberculosis. Eighty-six cases were confirmed on the basis of a positive culture for M. tuberculosis. Another nine cases had positive direct smears for AFBs and were not notified as atypical. A further 10 cases had histopathological findings consistent with tuberculosis. In 12 cases the diagnosis was based on clearing of a radiological infiltrate during treatment for tuberculosis. Finally, five extrapulmonary cases were diagnosed on the basis of typical clinical features, a positive TST, and a positive response to treatment. In total, 41 (34%) of the confirmed cases were extrapulmonary including 32 cases of tuberculous lymphadenitis.

The crude annual incidence of tuberculosis in this cohort was 76.2 (95% confidence interval [95% CI], 63.5 to 91.3) per 100,000 person-years. Figure 1 shows that the incidence of tuberculosis was greatest in the first 3 yr after migration, during which time the average annual incidence was 112 (95% CI, 85 to 148) per 100,000 person-years. The rate subsequently decreased and over the next 10 yr averaged 66 (95% CI, 52 to 84) per 100,000 person-years. Among those whose initial TST reaction was 15 mm or greater, the annual incidence rate in the first 3 yr was 213 (95% CI, 150 to 300) per 100,000 person-years and in the subsequent 10 yr the rate averaged 122 (95% CI, 90 to 165) per 100,000 person-years.

View larger version (16K): [in this window] [in a new window]   Figure 1.   Annual incidence of tuberculosis (95% confidence interval) by years since arrival in Australia.

The annual incidence rates for tuberculosis ranged from fewer than 40 per 100,000 among those whose initial TST reaction was read as negative, to more than 200 per 100,000 among those with strongly positive TST reactions (Table 1). The effect of TST reaction size on risk of tuberculosis was not affected by the presence of a BCG scar (Table 2). There was a significant linear trend for increased risk of tuberculosis with increasing TST reaction size category (Table 2).

There was no significant linear trend in the risk of tuberculosis across the age quartiles (Table 2).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In this cohort of predominantly Southeast Asian refugees who arrived in Sydney, Australia during the period 1984 to 1994, the annual incidence of tuberculosis, over an average period of 10 yr of follow-up, was 76 per 100,000. The risk of subsequent tuberculosis was related to size of the TST reaction on arrival but was not influenced by the presence of a BCG scar.

This cohort represents a uniquely valuable population in which to study reactivation of latent tuberculosis. Over a period of 4 to 14 yr after the identification of infection and exclusion of existing active pulmonary disease by chest X-ray, the members of this cohort have been living in a country with one of the lowest tuberculosis incidence rates in the world. Although it is possible that transmission and hence reinfection has occurred in Australia, or during visits back to their country of origin, the former has probably been even less common in the Sydney environment than it was in San Francisco (19). There is anecdotal evidence that Vietnamese refugees in Australia were reluctant to travel to back to Vietnam before 1993, although many have visited since that time.

The other strength of this study is the sensitivity and specificity of case ascertainment. There is evidence that the great majority of patients treated for tuberculosis in NSW have been notified to the Department of Health and will have been identified as cases in our study (20, 21). This is, in part, attributable to the fact that tuberculosis treatment (for all ages and for all sites) is given free of charge through chest clinics in NSW. The validity of this database as a complete, and hence sensitive, record of cases of tuberculosis occurring in NSW is supported by the observation that recent intensified surveillance, including tracing through records of sale of antituberculous drugs, did not result in any detectable increase in the notification rate. The specificity of the diagnosis of tuberculosis cases in this analysis was enhanced by review of case records of notified cases and reassessment of the diagnosis based on criteria. Finally, the finding that the cohort has remained intact, within New South Wales, during the period of follow-up supports the likelihood of complete ascertainment of incident cases.

Our finding that there was a significant linear trend in incidence of tuberculosis with increasing TST reaction size was expected. The analysis of this relation is limited by the method of recording TST reaction size in these subjects. All reactions < 10 mm were simply classed as "negative." It is likely that some of the cases arising in TST-"negative" subjects were, in fact, attributable to prior infection. These cases would include those whose TST reaction size was between 5 and 9 mm, at least some of whom would be infected, and others who were anergic at the time of initial screening due to deprivation in the refugee camps or during their travel to Australia.

We found that the presence or absence of a BCG scar did not alter the relation between TST reaction size and risk of subsequent tuberculosis. Although this study was not designed to assess the effectiveness of BCG vaccination in preventing tuberculosis, our finding is consistent with the conclusions of two meta-analyses: that BCG administered in childhood is not effective in preventing tuberculosis in adults (22, 23). It also supports the view that, after a prolonged interval, BCG does not influence TST reaction size (24, 25).

The disease incidence rates for subjects with 10- to 20-mm TST reactions in this cohort were similar to those reported in the Comstock and Edwards analysis (10), although lower than reported in other settings (8, 12, 26).

It is possible, with some assumptions, to estimate the lifetime risk of tuberculosis for a member of this refugee cohort. The observed annual incidence rate for a member of the cohort with a TST reaction  15 mm was 213 per 100,000 for the first 3 yr and 122 per 100,000 for the following 10 yr. Hence, for a 25 yr old, the expected cumulative incidence to age 38 yr is 1,859 per 100,000 (1.86%). If we assume, based approximately on the data from Comstock and Edwards (10), that the latter incidence rate applies to age 50 yr and that the rate increases by 50% (i.e., to 183 per 100,000) for the period from age 50 to 75 yr, then the cumulative incidence to age 75 yr would be 7.8%. Applying similar assumptions to a 35 yr old with a > 15-mm TST reaction, the cumulative risk to age 75 yr is 6.7%. These estimates of lifetime cumulative risk of tuberculosis are slightly higher than the lifetime risk estimates calculated by Comstock and Edwards (10).

The application of these risk estimates to the task of bridging the gap between practice and policy on the administration of preventive therapy in this context requires reanalysis of the decision trees referred to above. While this study has shown that previous estimates of lifetime risk for tuberculosis are applicable in this population, the other inputs into this decision require re-estimation in the Southeast Asian migrants. In particular, the influence of the relatively high prevalence of isoniazid resistance observed in some Southeast Asian countries on the effectiveness of isoniazid chemoprophylaxis is not known. At least some failures of prevention could be expected due to isoniazid resistance (27). Further, the risk of isoniazid-induced hepatitis and other adverse effects, the case fatality rates for tuberculosis and hepatitis, and risk of transmission of infection before diagnosis and treatment have not been established in this population. It is likely that, in a community such as ours, which is well serviced with primary health care and in which treatment for tuberculosis is directly observed, delays in diagnosis and fatalities will be less than previously estimated.

This is the first study to have examined the incidence of reactivation of latent tuberculosis in a population of predominantly Southeast Asian refugees living in an environment where reinfection rates are almost certainly very low. The observed rates are similar to those estimated in the general population of the United States in the 1950s and 1960s. We have confirmed that the risk increases with TST reaction size and that, in this cohort of adults and older children, it is independent of BCG vaccination status. Further data on the prognosis of tuberculosis and the effects of isoniazid preventive therapy in Southeast Asian migrants to Australia and similar countries are required to inform policy and practice for the prevention of tuberculosis in this population.