«Neonatal BCG protection against leprosy: a study in Manaus, Brazilian Amazon ´ SE RGIO S. CUNHA*, **, LAURA C. RODRIGUES*, VALDERIZA PEDROSA þ, ...»
Lepr Rev (2004) 75, 357– 366
Neonatal BCG protection against leprosy: a study
in Manaus, Brazilian Amazon
SE RGIO S. CUNHA*, **, LAURA C. RODRIGUES*,
VALDERIZA PEDROSA þ, INEZ M. DOURADO**,
MAURICIO L. BARRETO** &
SUSAN MARTINS PEREIRA**
*London School of Hygiene and Tropical Medicine, Keppel Street,
**Instituto de Saude Coletiva, Universidade Federal da Bahia, Brazil
þFundacao Alfredo da Matta, Manaus, Brazil ¸˜ Accepted for publication 21 September 2004 Summary There is clear evidence that BCG protects against leprosy, but cross- immunity with environmental mycobacteria can interfere with vaccination protection. Some have cast doubts whether BCG vaccination can offer a signiﬁcant impact against leprosy in the Brazilian Amazon, which is an endemic area for leprosy and with a high prevalence of environmental mycobacteria. This study was designed to estimate the vaccine effectiveness of neonatal BCG against leprosy in Amazon region, in Brazil. This is a cohort study nested in a randomized community trial. The study had two main results. First, neonatal BCG vaccination in Brazilian Amazon elicited protection of 74% (95% CI 57 – 86) against all forms of leprosy cases.
Second, the highest protection was observed for multibacillary cases, 93% (95% CI 71 – 98). It is concluded that the study provides evidence that neonatal BCG may have an important and overlooked impact on the occurrence and transmission of leprosy, maybe even more in the future when the cohort which received a high coverage of BCG reaches the age of high incidence of leprosy.
Introduction There is clear evidence that BCG protects against leprosy, but the level of protection has ranged from 20% to 90% in different studies.1,2 One hypothesis to explain this variation is cross-immunity with environmental mycobacteria (EM).3,4 All authors fulﬁlled the criteria for authorship proposed by the Vancouver Conference: concept of the study, drafting the article or revising it critically for important intellectual content and all had the ﬁnal approval for its content.
´ ´ Correspondence to: S. S. da Cunha, Instituto de Saude Coletiva, Rua Padre Feijo, no. 29, Andar 4,Canela, Salvador, Bahia, Brazil, CEP 40110–170
Brazil is a large country and has the second greatest number of cases of leprosy in the world detected each year.5 The Brazilian leprosy control programme advocates the identiﬁcation and treatment of all cases to control the disease, and as a complementary measure the BCG vaccination among household contacts of leprosy cases.6 The disease has an uneven geographical distribution in the country, and the Brazilian Amazon region is an important endemic area where a high prevalence of EM is assumed.7 Because of the hypothesis that infection with EM can interfere in the protection elicited by BCG 4, some have cast doubts on whether BCG vaccination can offer a signiﬁcant impact against leprosy8 or against tuberculosis9 in the Brazilian Amazon. Indeed, in Manaus, the largest urban centre in Brazilian Amazon, the vaccination coverage among household contacts of leprosy cases is low.
The present study took opportunity of an ongoing community trial of BCG vaccination of school children in a population with high coverage of neonatal BCG (REVAC-BCG trial) to estimate the vaccine effectiveness of neonatal BCG against leprosy in Amazon region in Brazil.
Materials and methodsSTUDY DESIGN
This is a cohort study nested in one of the study sites of the REVAC-BCG community trial.
This trial aims to evaluate the vaccine effectiveness of BCG given to schoolchildren against tuberculosis and leprosy and has two study sites, the cities of Salvador and Manaus. This cohort study was conducted in Manaus, where the trial was implemented in 1998. Details of the trial have been described elsewhere.10,11 Brieﬂy, it was a cluster randomized without placebo community trial. The study population of this cohort study was followed up in two periods and had thus two components (see Figure 1). First, from 1989 to 1997 (before implementation of the trial) corresponded to a historical cohort study involving the trial population from both control and intervention trial arms. Second, from 1999 to August 2002, during the trial follow-up, corresponded to a prospective cohort study involving only the trial population in the control arm, as this study is concerned with the neonatal dose and most individuals of the intervention arm received two doses.
STUDY SITE This cohort study was conducted in the city of Manaus, with about 1,500,000 inhabitants (1999 census) on the banks of the Negro River, Brazilian Amazon region, with a tropical
STUDY POPULATIONThe cohort participants consisted of the schoolchildren recruited into the trial, who were born between 1983 and 1991 (aged 7– 14 years at recruitment into the trial), and residing in Manaus. The trial, and this study, was restricted to children attending state schools. Only children with no BCG scar or one scar entered in this cohort study (n ¼ 112,744 in total study population; n ¼ 60,458 in control group). Identiﬁcation and vaccination data recorded in the REVAC-BCG trial database were used in this cohort study.
The presence of a scar compatible with BCG vaccination was used as a surrogate of BCG neonatal vaccination in the REVAC-BCG trial and in this cohort study. BCG scar reading was performed during the recruitment phase of the trial in 1998 (Manaus) by trained health workers who visited the schools, and data entered in the trial database. Validation of the scar as a marker of BCG vaccination in the trial population in Manaus is published elsewhere.12 In summary, BCG scar reading had a sensitivity between 94% and 98%, depending on age group, when the gold standard was the agreement between parental information about past vaccination and vaccination card.
CASE ASCERTAINMENT AND CLASSIFICATION
Data on all cases regarding identiﬁcation data, clinical form, and date of diagnosis of all leprosy cases in the city of Manaus are routinely entered in a computerized database of the local control programme. In this cohort study, leprosy cases diagnosed between 1989 and 1997 (historical cohort), and between 1999 and August 2002 (prospective cohort), in children eligible to the trial population, were ascertained from the local leprosy control programme. All cases reported as indeterminate, tuberculoid and BT (bordeline tuberculoid) were classiﬁed as paucibacillary (PB) cases; and LL (lepromatous lepromatous), BB (borderline borderline), BL (borderline lepromatous) as multibacillary (MB). Cases reported as dimorphous were classiﬁed as multibacillary, but those with negative baciloscopy and who received PB multi-drug therapy (MDT) regime were classiﬁed as paucibacillary.
The leprosy cases recorded in the database of the local leprosy control programme were linked to the records in the REVAC-BCG trial database. The matching between cases and trial study population was done based on variables present in both databases: subject’s name, date of birth, sex, and mother’s name.
360 S. S. Cunha et al.
INCIDENCE OF LEPROSY AND TUBERCULOSIS FOR GEOGRAPHICAL AREAThe city of Manaus is divided in 56 administrative areas (districts), and two categorical variables were created indicating if the schools were the students attended in 1998 were located in districts with incidence of leprosy (NCDR) and tuberculosis below or above to the incidence of the city in 1996 as a whole. These variables were taken as a proxy of the baseline risk and socio-economic status (as leprosy and tuberculosis are diseases related to poverty) of the trial study population and in the prospective cohort. However, chosen was made to not use these variables in the analysis of the historical cohort because they may not express the baseline risk during the longer period before the trial in the historical cohort, which included period until 6 –7 years before 1996.
Incidence of leprosy per 10,000 (NCDR) was estimated in children separately by BCG scar, current age, sex, and leprosy classiﬁcation. Conﬁdence intervals (95% CI) of crude rates and rate ratios were estimated according to an approximation to the Poisson log-likelihood for the log rate parameter.13 If the number of cases was smaller than 30, the conﬁdence intervals of rates were estimated according to exact conﬁdence limits for a Poisson-distributed variable.14 Interaction was done using the log likelihood ratio test.13 Age was categorized in groups and was treated as time-varying variable, expressed as current age (age during the cohort followup) through expansion of the data-set in time scales.13 BCG vaccine effectiveness was estimated as (1 –rate ratio) £ 100. Adjusted rate ratios for current age, sex, year of birth and incidence of leprosy and tuberculosis for geographical areas were obtained by using standard Poisson regression. All the statistical analysis was done using STATA version 7.015.
In the historical cohort study, 128 leprosy cases diagnosed between 1989 and 1997 in Manaus were linked to the trial study population. One hundred and seven (107) cases were classiﬁed as paucibacillary cases (21 indeterminate, 78 tuberculoid, three dimorphous, ﬁve BT) and 21 as multibacillary cases (three LL, one BL, 17 dimorphous). The three reported dimorphous cases classiﬁed as paucibacillary had negative baciloscopy and received the paucibacillary MDT regime.
In the prospective cohort study, 53 leprosy cases diagnosed between 1999 and August 2002 were linked to the trial study population. Forty-three (43) were classiﬁed as paucibacillary cases (23 tuberculoid, nine indeterminate, 10 BT, one just reported as paucibacillary case) and 10 cases as multibacillary (one BB, four dimorphous, three BL, two LL). All these four dimorphous cases had positive baciloscopy and were treated with multibacillary MDT regime.
Among cases in the historical cohort, 35 (27·3%) had no BCG scar and 93 (72·7%) had one scar; 75 (58·6%) were females and 53 (41·4%) males. In the prospective cohort, 22 (41·5%) had no BCG scar and 31 (58·5%) had one scar; 24 (45·3%) were females and 29 (54·7%) males. These ﬁgures compare with 84·0% with BCG scar and 51·3% of females in the total trial population (84·8% and 51·4% in control group, respectively).
For the whole trial population, the prevalence of BCG scar was higher for those born in recent years: 79·8%, 85·6% and 88·9%, for those born in 1983– 1986, 1987 –1988, and Neonatal BCG protection against leprosy 361 1989 –1991, respectively. For the control group these ﬁgures were 80·9%, 86·3%, 89·5%, respectively.
Table 1 shows the crude rates according to sex, current age and incidence of leprosy and tuberculosis in geographical areas in 1998. In historical cohort, females and older individuals had higher rates than males (rate of 1·51 versus 1.13) and younger people (rate of 3·22 versus 0.28). In prospective cohort, males presented higher rates than females (rate of 2·93 versus 2.30) and older individuals yet had higher rates than younger (rate of 2·93 versus 1.91). In the prospective cohort, rates of leprosy were higher among those attending schools in 1998 located in areas with higher rates of leprosy and tuberculosis (but without statistical signiﬁcance).
In the historical cohort, the rate per 10,000 for multibacillary cases were 0·22 (21/968,369 person years; 95% CI: 0·17 –0·37) and for paucibacillary cases was 1·04 (107/968,369; 95% CI: 0·86 –1·27). In the prospective cohort, the rates were 0·49 (10 cases/203,507 person years;
95% CI: 0·26 –0·91) and 2·11 (43 cases/203,507; 95% CI: 1·57– 2·84), respectively (data not shown).
Table 2 shows the crude and adjusted (current age, sex and calendar year of birth) rate ratios for leprosy in historical cohort, according to presence of BCG scar and case
classiﬁcation. The adjusted estimates of vaccine effectiveness were: 41% (95% CI:
12—60%) for all leprosy cases, 77% (95% CI: 45 –90%) for multibacillary cases, 26% (95% CI: –16 – 53%) for all paucibacillary cases, and no statistically signiﬁcant protection for paucibacillary cases excluding intermediate and for indeterminate cases. Except for multibacillary cases, the difference between crude and adjusted rate ratios varied between Table 1. Rates of leprosy per 10,000 separately by sex, current age and incidence of leprosy and tuberculosis in geographical areas (prospective cohort)
a Excluding 1074 individuals and one leprosy case with missing data about incidence of tuberculosis.
b Excluding 623 individuals (without leprosy case) with missing data about incidence of leprosy.
362 S. S. Cunha et al.
Table 2. Rate ratios for leprosy according to presence of BCG scar, separately by clinical forms, in historical cohort
a Total person year of 811,554 and 156,815, for vaccinated and unvaccinated, respectively.
b Estimate controlled for current age, sex and year of birth (categorical variables).
13% and 18%, indicating confounding effect (taking the threshold of 10% to deﬁne confounding), attributed to the association of year of birth with presence of BCG scar and leprosy rates. There was no statistically signiﬁcant interaction between presence of BCG scar and current age, year of birth and sex (data not shown).
Table 3 shows the crude and adjusted rate ratios for leprosy in prospective cohort, according to presence of BCG scar, and case classiﬁcation. The adjusted estimates of vaccine Table 3. Rate ratios for leprosy according to presence of BCG scar, by clinical forms, in prospective cohort
a Total person year of 172,505 and 31,003, for vaccinated and unvaccinated, respectively.
b Estimate controlled for current age, sex, year of birth and incidence of leprosy and tuberculosis in geographical areas before the trial follow-up (categorical variables).