INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The cost-effectiveness of the BCG program is inversely proportional to the annual risk of tuberculous infection (1), which in most technically advanced countries has already reached a low level (2). Consequently, critical evaluation of the BCG vaccination programs has therefore been started (3).

General BCG vaccination program of newborns was discontinued in Sweden (7) and West Germany (8) about 20 yr ago, and later in Austria, Israel, Switzerland, and in selected areas of the Czech Republic (9). Discontinuation has usually resulted in an increase of tuberculosis among newborns, but this has been temporary and predictable. Consequently, there have been studies of the effects of discontinuing BCG programs among newborns but the scientific evidence on the efficacy of BCG revaccination is deficient (10). However, BCG revaccination, usually of tuberculin-negatives, has been, and still is, an integral part of many tuberculosis programs, some of which even use repeated revaccinations. In Finland BCG vaccination at birth and revaccination of tuberculin-negative schoolchildren has been the policy since the 1950s, and the Finnish people belong to the most comprehensively BCG-vaccinated populations, the coverage at birth still being close to 100%. Because of the steadily improving tuberculosis situation, especially among the age groups below 20 yr, the BCG revaccination program was evaluated and discontinued from 1990 onwards, seeing that the risk of increase was estimated to be very low (11). The follow-up data at 6 yr are reported here.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Population

In 1995, the population of Finland was 5.116 million. The incidence of tuberculosis has been decreasing; it was 46.9/100,000 in 1980 (all cases, all forms), 15.4/100,000 in 1990 at the time of discontinuing the BCG revaccination, and is now 12.9 /100,000. The National Tuberculosis Register covers the whole country and notification of cases is compulsory by law. If tuberculosis is confirmed bacteriologically or histologically or if the clinical evaluation of symptoms and signs indicate a full course of tuberculosis chemotherapy, the case is notified.

BCG revaccination was given to schoolchildren at the age of 11-13 yr if they were tuberculin-negative (induration under 5 mm) after two-step Mantoux testing, first with 2 TU followed by 10 TU PPD RT 23. BCG vaccination has always been given intradermally according to the instructions of the World Health Organization (12). Until 1978 the Danish (Gothenburg strain) vaccine was used with the dose of 0.1 ml. Because of high BCG osteitis rate the vaccine was changed to the British (Glaxo 1077) and the dose was reduced to half 0.05 ml for newborns, but 0.1 ml was used for revaccination.

The BCG revaccination program was officially discontinued on October 4, 1989. Our follow-up in the study group was started from the beginning of 1990, when the children born in 1979, who were the first cohort to remain without revaccination, reached 11 yr. The study group consisted of six birth cohorts, born 1979-1984, vaccinated at birth with the British BCG vaccine, altogether 388,089 children. They have been followed-up separately for tuberculosis cases in 1990-1995.

Because no parallel control group was possible, it was made of six age cohorts born in 1969-1974 (10 yr earlier than children in the study group), vaccinated at birth with the Danish vaccine, and revaccinated, if tuberculin-negative, at 11-13 yr with the British vaccine, altogether 371,199 children (Table 1). Tuberculosis cases among them in 1980- 1985 were collected for comparison.

If BCG revaccination was effective, the benefit would affect children after the revaccination age (11-13 yr); however, any increase occurring after discontinuation would affect those who were left without revaccination (i.e., today 11-16 yr). Therefore, all cases of tuberculosis (excluding the foreign-born) in the age groups of 10-14 and 15-19 yr have been checked from the National Tuberculosis Register to demonstrate the overall trend.

Statistical analysis: The annual incidence rates per person years and 95% confidence intervals (95% CI) were calculated for incidence rates and the relative risk using Taylor series confidence intervals (13).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Almost a fifth of the children born in the early 1970s were subjected to BCG revaccination at the age of 11-13 yr (Table 1), and this proportion remained constant until 1975. Thereafter revaccinations clearly started to decrease, although the official policy was unchanged, and by the time the program was discontinued, only 2-3% children were being revaccinated.

In the study group, three cases of tuberculosis were found during 6 yr after stopping BCG revaccination (Table 2): an 11-yr-old with spinal tuberculosis in 1990; a 14-yr-old with pulmonary tuberculosis in 1994; and a 16-yr-old with pulmonary tuberculosis in 1995. The study group included 388,089 children who generated 1,347,266 follow-up person years. Thus, the estimated annual incidence rate per million person years is 2.23 cases. The approximated 95% Taylor series confidence interval for the true incidence rate is from 0.72 to 6.90. 

In the control group, five cases of tuberculosis were registered during 6 yr: a 12-yr-old and a 13-yr-old both with lymph node tuberculosis in 1982, a 13-yr-old with abdominal tuberculosis in 1983, a 13-yr-old with bronchial tuberculosis in 1984, and a 16-yr-old with pulmonary tuberculosis in 1985. The control group included 371,199 children who generated 1,324,401 follow-up person years. Accordingly, the estimated annual incidence rate per million person years is 3.78 cases. The approximated 95% Taylor series confidence interval for the true incidence rate is from 1.57 to 9.07.

The relative risk for tuberculosis among non-BCG-revaccinated children is 0.59 of that for the historical controls. The approximated 95% Taylor series confidence interval for the true relative risk is from 0.14 to 2.47, showing that we can be 95% confident that the risk of tuberculosis among the non-vaccinated is at least one seventh and at most two and a half of that in the historical controls.

In 1995, the whole age group of 10-14 yr (Table 3) was non-BCG-revaccinated, the age group of 15-19 yr consisted of non-BCG-revaccinated age cohorts born 1979 and 1980 and of age cohorts born 1976-1978 with low rate of revaccination, 2- 5% (Table 1). The reduction in the tuberculosis rate from 1980 to 1995 has been noticeable (Table 3). No change in the trend can be seen in the age groups of 10-14 and 15-19 yr since 1990, after stopping the revaccination program.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Before revaccination can be of benefit, it is necessary both that the immunity given by the first vaccination has waned with time, and that it is boosted again by the second. Tuberculin sensitivity is not a marker for protection and so waning tuberculin sensitivity with time does not necessarily signify loss of immunity (14). Revaccinating tuberculin-negative children, therefore, has no scientific basis. It has even been suspected that children with weak tuberculin sensitivity may be at risk of reactivation from focal reactions after BCG vaccination (15). All this indicates the difficulties in interpretation of tuberculin tests and selection for BCG (re)vaccination.

The predicted yearly increase of tuberculosis after discontinuation of the Finnish revaccination program was estimated to be so low that by running the program for 2 yr one case can be prevented (11). The follow-up data indicate that in the present conditions in Finland, the efficacy of the BCG revaccination program may be even lower. Although no immediate effects of discontinuing BCG revaccination were detected, it may increase protection in later life, and the follow-up must, therefore, be continued to definitively answer this question.

The benefit of BCG revaccination, if any, seems to be considerably smaller than vaccination at birth. This is supported by a rise in tuberculosis among children when BCG vaccination programs at birth were discontinued. This increase could be demonstrated in Sweden within 4 yr (3, 7). The same phenomenon was seen in Czechoslovakia (9). In Finland, the BCG revaccination rate had started to decrease from 1975, already before the program was officially discontinued, and if the BCG boosting were effective, this should have resulted in a rise already before the final discontinuation, or at least after it; but no increase has so far been recorded. A retardation of the downward trend cannot be excluded, but it seems unlikely since the cases per million person years are now clearly lower than 10 yr ago. However, the historical control series is not fully adequate, although it is from the period when revaccination rate was at its highest. The incidence of tuberculosis among adolescents seems to be continuing the decrease which began during the revaccination period. The fewer tuberculosis cases in study group cannot be explained by change of BCG vaccine in 1978 because children had received BCG at birth alone with lower dose of the British vaccine, which is classified as weak, whereas the Danish vaccine is classified as strong (12).

Discontinuation of BCG revaccination may influence the tuberculosis rate among adolescents in one of three ways. First, if the policy is effective, stopping will be followed by an increase; second, the downward trend will be retarded; or third, discontinuation may have no influence at all. The last finding would not necessarily mean that the policy is useless in all conditions, because the epidemiological situation may already be so favorable in Finland that the effect of BCG revaccination might be absent.

Observations on the efficacy of BCG revaccination have been controversial. In Hungary in the 1960s the decrease of tuberculosis in BCG-revaccinated children was more rapid than in non-vaccinated adults, and when these children reached adolescence, the incidence fell further (16). However, no parallel control group was included and other factors may have biased the results; besides, epidemiological data usually display a more pronounced decrease of tuberculosis among children than in adults when the tuberculosis situation is improving. Retrospective analysis in Poland (1965-1977) demonstrated higher incidence of tuberculosis among non-revaccinated children, but the groups were not randomized and the cases were few (17). In Chile no difference was seen in tuberculosis rates in children having one or more BCG scars, proving no efficacy of repeated BCG (18). Revaccination increased tuberculin responsiveness and recognition of environmental mycobacteria in Kuwait and might indirectly have increased immunity (19).

The latest results of the first controlled trial in Malawi (20) showed that repeated BCG gave no protection against tuberculosis. It seemed to be even harmful in HIV-infected persons, by increasing the risk of pulmonary tuberculosis. In Malawi, a single BCG vaccination gave no protection either, and, therefore, BCG seemed to be without efficacy. This is in contrast to the Scandinavian experience where BCG for newborns has been effective; for example, in Sweden BCG at birth demonstrated about 80% efficacy in studies made at discontinuation of the program (7, 12, 21), and, therefore it is possible that repeat BCG vaccination would have been indicated. However, we think in the light of our results that efficacy of BCG revaccination programs is doubtful, and they should be phased out. The World Health Organization has also taken a negative attitude to repeated BCGs in its new recommendations (22).