SEARCH

SEARCH BY CITATION

Keywords:

  • annual risk of infection;
  • tuberculosis;
  • tuberculin;
  • treatment outcome;
  • notification rates;
  • South Africa
  • risque annuel d’infection;
  • tuberculose;
  • tuberculine;
  • résultat du traitement;
  • taux de notification;
  • Afrique du Sud
  • riesgo anual de infección;
  • tuberculosis;
  • tuberculina;
  • resultado del tratamiento;
  • tasas de notificación;
  • Sudáfrica

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. References

Objective  To estimate the change in annual risk of tuberculosis infection (ARTI) in two neighbouring urban communities of Cape Town, South Africa with an HIV prevalence of approximately 2%, and to compare ARTI with notification rates and treatment outcomes in the tuberculosis (TB) programme.

Methods  In 1998–1999 and 2005, tuberculin skin test surveys were conducted to measure the prevalence of Mycobacterium tuberculosis infection and to calculate the ARTI. All 6 to 9-year-old children from all primary schools were included in the survey. Notification rates and treatment outcomes were obtained from the TB register.

Results  A total of 2067 children participated in the survey from 1998 to 1999 and a total of 1954 in 2005. Based on a tuberculin skin test cut-off point of 10 mm, the ARTI was 3.7% (3.4–4.0%) in the 1998–1999 survey and 4.1% (3.8–4.5%) in 2005. The notification rate for pulmonary TB increased significantly from 646 per 100 000 in 1998 to 784 per 100 000 in 2002. In Ravensmead, there was no significant change in ARTI [first survey: 3.5% (3.1–3.9%), second survey: 3.2% (2.9–3.6%)], but in Uitsig the ARTI increased significantly from 4.1% (3.6–4.6%) to 5.8% (5.2–6.5%). The difference in ARTI between the two areas was associated with differences in reported case rates and the proportion of previously treated cases.

Conclusion  Tuberculosis transmission remains very high in these two communities and control measures to date have failed. Additional measures to control TB are needed.

Objectif:  Estimer les changements dans le risque annuel d’infection de la tuberculose (RAIT) dans deux communautés urbaines voisines de Cape Town en Afrique du Sud avec un taux de prévalence du VIH d’environ 2% et comparer le RAIT aux taux de notifications et aux résultats du traitement dans le programme de lutte contre la tuberculose (TB).

Methodes:  En 1998-1999 et 2005, des études sur le test cutanéà la tuberculine ont été menées afin de mesurer la prévalence de l’infection par Mycobacterium tuberculosis et pour le calcul du RAIT. Tous les enfants de 6 à 9 ans dans toutes les écoles primaires ont été inclus dans l’enquête. Les taux de notifications et les résultats du traitement ont été obtenus à partir du registre de la TB.

Resultats:  2067 enfants ont participéà l’enquête de 1998-1999 et 1954 en 2005. Sur la base d’une valeur seuil du test cutanéà la tuberculine de 10 mm, le RAIT a été estiméà 3,7% (3,4% - 4,0%) dans l’étude menée en 1998-1999 et à 4,1% (3,8% - 4,5%) en 2005. Le taux de notification de la TB pulmonaire a augmenté de 646 pour 100000 en 1998 à 784 pour 100000 en 2002. A Ravensmead il n’y avait pas de changement significatif dans le RAIT (première enquête: 3,5% (3,1% - 3,9%), seconde enquête: 3,2% (2,9% - 3,6%)), mais à Uitsig le RAIT a augmenté de 4,1% (3,6% - 4,6%) à 5,8% (5,2% - 6,5%). La différence de RAIT entre les deux zones était associée à des différences dans le taux de cas rapportés et la proportion de cas déjà traités.

Conclusion:  La transmission de la TB demeure très élevée dans ces deux communautés et les mesures de contrôle à ce jour ont échoué. Des mesures supplémentaires pour le contrôle de la TB sont nécessaires.

Objetivo:  Estimar el cambio en el riesgo anual de infección por tuberculosis (RAIT) en dos comunidades urbanas vecinas de Ciudad del Cabo, Sudáfrica, con una prevalencia de VIH de aproximadamente 2%, y comparar el RAIT con las tasas de notificación y los resultados del tratamiento en el programa de tuberculosis (TB).

Métodos:  En 1998-1999 y 2005, se realizaron pruebas dérmicas de tuberculina con el fin de medir la prevalencia de infección por Mycobacterium tuberculosis y calcular el RAIT. Todos los niños de 6 a 9 años de escuelas primarias fueron incluidos en el estudio. Las tasas de notificación y los resultados del tratamiento fueron obtenidos del registro de TB.

Resultados:  2,067 niños participaron en el estudio entre 1998 y 1999 y 1,954 en el 2005. Basándose en un punto de corte para la prueba de tuberculina de 10 mm, el RAIT fue 3.7% (3.4% - 4.0%) en el estudio de 1998-1999 y del 4.1% (3.8% - 4.5%) en el 2005. La tasa de notificación para TB pulmonar aumentó significativamente de 646 por100,000 en 1998 a 784 por 100,000 en el 2002. En Ravensmead no había un cambio significativo para el RAIT (primer estudio: 3.5% (3.1% - 3.9%), segundo estudio: 3.2% (2.9% - 3.6%)), pero en Uitsig el RAIT aumentó significativamente de 4.1% (3.6% - 4.6%) a 5.8% (5.2% - 6.5%). La diferencia en RAIT entre las dos áreas estaba asociada a diferencias en las tasas de casos reportados y en la proporción de casos previamente tratados.

Conclusión:  La transmisión de TB continua siendo muy alta en estas dos comunidades, y las medidas de control a día de hoy han fallado. Se requieren medidas adicionales para el control de la TB.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. References

Tuberculosis (TB) remains an important public health problem (Corbett et al. 2003). WHO targets for the Stop-TB strategy are to detect at least 70% of incident cases of smear-positive TB and to cure at least 85% of these patients (Stop TB Partnership & WHO 2006). Mathematical models predict that reaching these targets will lower the transmission of TB, the incidence of TB, and the annual risk of TB infection (ARTI) (Dye et al. 1998; Borgdorff et al. 2002).

The prevalence of TB infection and the ARTI, which are estimated from repeated tuberculin surveys of children, are considered the best epidemiological indicator to monitor ongoing TB transmission in a community (Arnadottir et al. 1996). Changes in the ARTI over time reflect the impact of TB control measures. In Africa, a number of tuberculin surveys have been published in the last 10 years, and ARTIs between 0.3% and 3.1% have been reported (Bosman et al. 1998; El Ibiary et al. 1999; Odhiambo et al. 1999; Salaniponi et al. 2004; Trébucq et al. 2005; Egwaga et al. 2006). For South Africa, ranking 7th among the 22 highest TB burden countries in the world (WHO 2007), no recent national ARTI estimates are available, although a recent TST survey in a small community in Cape Town showed an ARTI of 4.1% (Middelkoop et al. 2008). The aim of this study was to assess the trend in ARTI in Ravensmead and Uitsig, two high incidence urban communities in Cape Town with a prevalence of HIV of approximately 2%. We performed two tuberculin skin test (TST) surveys among school children. The first survey was done in 1998–1999 and the second in 2005. Then we compared the ARTI data with the trends in pulmonary TB (PTB) notifications and treatment outcomes over the period 1993–2002.

Methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. References

Study population

Ravensmead and Uitsig have a total surface area of 3.4 km2 and had a census population of 34 294 in 1991 and 36 343 in 2001 (Statistical Support and Informatics 2001). The average monthly income per household was R6,750 in Ravensmead and R3,290 in Uitsig in 2001 (1 USD ≈ 7.8 Rand). Employment rates were approximately 51% in Ravensmead and 40% in Uitsig between 1996 and 2001. The infant mortality rate in these communities is 19 per 1000 live births and two primary health care clinics and an adjacent tertiary hospital serve the area. A TB survey in 2002 showed a prevalence of 10 per 1000 for bacteriologically confirmed TB (Den Boon et al. 2007). BCG vaccination in the neonatal period has been compulsory since 1972 and the vaccination coverage is estimated at 80–99% (Fourie 1987; Corrigal et al. 2008).

Study design and sampling

The two TST surveys were conducted according to the WHO/IUATLD guidelines (Arnadottir et al. 1996). The first TST survey was conducted in 1998–1999 and it was repeated using the same method in 2005. Using the exact average calendar time of the survey, and the exact average age of the children, we determined that the 1998–1999 survey gives the average ARTI in 1994 and the 2005 survey represents the average ARTI in 2000.

All primary schools, nine in 1998–1999 and eight in 2005 (one school in Ravensmead closed), were included in the survey and all children (grade 1–7) were asked to have a skin test. The Committee for Human Research, Faculty of Health Sciences, Stellenbosch University, the Department of Health, City of Cape Town and the Department of Education of the Western Cape Province gave permission for the study. Written informed consent was obtained from parents/legal guardians and assent from participants was sought.

Data collection

Sex and the presence of a BCG scar were recorded for each child. In the 1998–1999 survey, the age of the child was recorded while in the 2005 survey we calculated age using the date of birth and the date of the survey. A trained and experienced field worker performed the TST. A quantity of 0.1 ml from a single batch of PPD RT 23 (Statens Serum Institut, Copenhagen, Denmark), at a dose of 2TU was injected intradermally on the ventral aspect of the left forearm. The maximum transverse diameter of the induration was read with a calibrated calliper after 72 h. Parents received a letter that informed them of their child’s result and symptomatic children were asked to attend the local health facility where they were screened for active TB and managed accordingly.

Statistical analyses

All 6 to 9-year old children whose BCG scar status and TST induration were recorded were included in the data analysis. Children older than 9 were excluded from the analysis to ensure that no child included in the first survey could be included in the second survey. The frequency distributions of the TST indurations of children with and without BCG-scar were compared using the Mann–Whitney test. Analyses were done on all children irrespective of BCG scar status, but we also did sensitivity analysis excluding children with a BCG scar. Reactions greater than or equal to 10 mm were considered to be attributable to infection with Mycobacterium tuberculosis. Besides the cut-off point of 10 mm we also explored the cut-offs of 7, 8, and 9 mm. Prevalence of TB infection and its 95% confidence intervals (95% CI) were adjusted for non-response in each school. The formula 1−(1−prevalence)1/mean age was used to calculate the ARTI (Rieder 1995, 2005; Arnadottir et al. 1996). In the 1998–1999 survey we obtained age and not birth date, therefore mean age was calculated after adding 0.5 years to the reported age.

All data were dually entered in a Microsoft Access database and the two entries were compared with Epidata. The dataset with the least errors was corrected after verification with the original data. Statistical calculations were performed with STATA version 8.0 (StataCorp 2003).

Notification rates and treatment outcomes

Notification rates were calculated from the number of recorded TB cases at a clinic and the population size of the community served by the TB clinic. The number of PTB cases and the treatment outcomes of smear-positive PTB cases were obtained from the provincial TB database for Ravensmead and Uitsig TB clinics. Unfortunately the recorded PTB cases and the treatment outcome of these cases were only available for 1998–2002. We therefore used additional data from our epidemiological research database to study the trend in smear-positive TB notification rates for the period 1993–1998. This epidemiological database was based on the notification data 1993–1996 and on the data from the TB treatment registers form the clinics in Ravensmead and Uitsig. For the calculation of rates we obtained the population size from the 1991, 1996 and 2001 census data for Ravensmead and Uitsig. The population size for the interim years was calculated with the inter-census growth rate.

Cure and completion of treatment were considered as successful treatment outcomes. Unfavourable treatment outcomes included failure, default and transfer. Death was considered separately.

We compared the trend in notification rates and treatment outcomes for the two areas using logistic regression analysis with the population as those exposed, and the TB cases as those with the disease outcome. Area, year and the interaction term area*year were included as independent variables in the model.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. References

TST survey 1998–1999

In the first TST survey 5779 children were registered as primary school attendees; 2566 being between 6 and 9 years old. Children without parental consent, not present on both days of the TST survey and those with missing data were excluded. Of the 6 to 9-year old children 2067 (81%) were included in the analyses. Seven to nine year old children were more often included (82%) than six year old children (75%) (OR: 1.51, 95% CI: 1.21–1.89). Boys (80%) and girls (81%) had an equal participation rate (OR: 1.06, 95% CI: 0.87–1.29). The participation rate in Uitsig (89%) was higher than in Ravensmead (76%) (OR = 2.63, 95% CI: 2.08–3.32). 65% of children had a BCG scar. The distributions of TST indurations of children with and without BCG scars were not significantly different (Figure 1a) (Z = 0.501, P = 0.616). In this survey, 958 (72%) of children with a BCG scar and 491 (68%) of children without a BCG scar had no measurable skin test reaction (OR = 1.21, 95% CI: 0.99–1.47).

image

Figure 1.  (a) Frequency distribution of tuberculin skin test indurations in the 1998–1999 survey. (b) Frequency distribution of tuberculin skin test indurations in 2005.

Download figure to PowerPoint

TST survey 2005

In the second TST survey, 4759 children were registered as primary school attendees of whom 2460 were between 6 and 9 years old. Children without parental consent, not PTB resent on both days of the TST survey and those with missing data were excluded. Of those 1954 (79%) were included in the survey. Seven to nine year old children were more often included (81%) than 6-year-old children (72%) (OR: 1.68, 95% CI: 1.34–2.10). Boys (81%) had a higher participation rate than girls (78%) (OR: 1.22, 95% CI: 1.01–1.49). The participation rate was 82% in Uitsig and 78% in Ravensmead (OR = 1.34, 95% CI: 1.10–1.65). The proportion of children with a BCG scar was 68% and there were no significant differences between the distribution of TST indurations of children with and without a BCG scar (Figure 1b) (Z = 1.439, P = 0.150). In this survey, 944 (71%) of children with a BCG scar and 409 (66%) of children without a BGC scar had no measurable skin test reaction (OR = 1.29, 95% CI: 1.05–1.58).

Prevalence of infection and ARTI

The prevalence of M. tuberculosis infection was 26.2% (95% CI: 24.3–28.1%) in 1998–1999, corresponding to an ARTI of 3.7 (95% CI: 3.4–4.0%). In 2005, the prevalence of infection was 28.9% (95% CI: 27.0–30.9%), corresponding to an ARTI of 4.1% (95% CI: 3.8–4.5%) (Table 1). For Ravensmead the ARTI was calculated to be 3.5% (95% CI: 3.1–3.9%) in 1998–1999 and 3.2% (95% CI: 2.9–3.6%) in 2005; a difference that was not statistically significant. The ARTI in Uitsig increased significantly from 4.1% (95% CI: 3.6–4.6%) to 5.8% (95% CI: 5.2–6.5%). In 2005, the ARTI in Uitsig was significantly higher than in Ravensmead. Using cut-off points of 7, 8 and 9 mm increased the estimates for the prevalence of infection and the ARTI when compared with 10 mm but it did not change the trend over time or the differences between the two areas (data not shown).

Table 1.   Prevalence of Mycobacterium tuberculosis infection (TST ≥10 mm) and annual risk of tuberculosis infection (ARTI) in Ravensmead and Uitsig
 1998–19992005
Analyzed nAverage age yearsPrevalence of infection % (95% CI)ARTI % (95% CI)Analyzed nAverage age yearsPrevalence of infection % (95% CI)ARTI % (95% CI)
Total area20678.0826.2 (24.3–28.1)3.7 (3.4–4.0)19548.0728.9 (27.0–30.9)4.1 (3.8–4.5)
Ravensmead12378.0524.8 (22.4–27.2)3.5 (3.1–3.9)11978.0623.3 (20.9–25.7)3.2 (2.9–3.6)
Uitsig8308.1328.6 (25.6–31.7)4.1 (3.6–4.6)7578.0838.4 (35.0–41.9)5.8 (5.2–6.5)

We repeated our analysis including only children without a BCG scar. ARTI estimates obtained in these analyses were slightly higher than the ARTI estimates reported for all children irrespective of BCG scar status but the trend over time or the difference between the two areas did not change (data not shown).

Notification rates and treatment outcomes

Both Ravensmead and Uitsig had high and increasing PTB notification rates (Figure 2). In Ravensmead the notification rate for total TB increased from 596 to 686 per 100 000 between 1998 and 2002. In Uitsig it increased from 750 to 987 per 100 000 over the same period. The PTB notification rate in Uitsig was higher than in Ravensmead (P < 0.001). Over the period 1998–2002 Ravensmead had a significantly higher proportion of new PTB cases, and thus a lower proportion of previously treated PTB cases than Uitsig (P = 0.008) (Figure 3).

image

Figure 2.  Pulmonary tuberculosis notification rates for Ravensmead and Uitsig.

Download figure to PowerPoint

image

Figure 3.  Proportion new pulmonary tuberculosis cases among total PTB cases.

Download figure to PowerPoint

The notification rates of smear-positive TB were available from 1993 to 2002. The notification rates for the first period (1993–1998) show a similar trend as for the second period (1998–2002) (Figure 4). The smear positive notification rates increased significantly between 1993 and 2002 in both Ravensmead and Uitsig (P < 0.001). The notification rates were higher in Uitsig compared to Ravensmead both for smear-positive TB (P < 0.001) as for new smear-positive TB (P < 0.001). There was no significant difference in the proportion of new cases among smear positive cases between the two areas (P = 0.627) or over time (P = 0.083).

image

Figure 4.  Notification rates for new and total smear positive pulmonary tuberculosis 1993–2002.

Download figure to PowerPoint

Treatment success rates were similar in Ravensmead and Uitsig (P = 0.788) and increased during the period 1998–2003 (P = 0.055) (Figure 5). This was the case for new as well as previously treated smear positive cases (data not shown). Unfavourable outcomes decreased in both areas, but only became below 15% in 2003. Unfavourable treatment outcomes were more often seen among previously treated cases. The death rate was approximately 3%, and remained the same (data not shown).

image

Figure 5.  Percentage successful treatment among all smear positive pulmonary tuberculosis cases.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. References

In this study, we show that the TB transmission rate remains high in Ravensmead and Uitsig. An ARTI of more than 3.5% is one of the highest reported in the world in the last 10 years (Tupasi et al. 2000; Dubuis et al. 2004; Trébucq et al. 2005; Egwaga et al. 2006; Gopi et al. 2006). Obviously the TB epidemic is not under control. This is also reflected in the high and rising notification rates. Uitsig had the highest notification rates, which corresponded with a higher ARTI in this area. The increasing notification rates could be partly or in total due to increased case detection. Although information about this is lacking, we think that case detection might have improved between 1993 and 2002, especially after the Stop TB Strategy (then known as the DOTS Strategy) was introduced in 1996–1997. However, we do not think that better case detection can explain the total increase in notification rates.

Approximately 60% of all PTB cases were new TB cases. This means that as much as 40% of PTB cases had been treated for TB before. In Uitsig the proportion of previously treated PTB patients was higher than in Ravensmead over all 5 years studied. These previously treated cases might be at least partially responsible for the ongoing and increasing transmission in Uitsig. A possible explanation for this high proportion of previously treated patients is low cure rates in the past. If patients are not cured they have a higher risk to relapse or to develop drug resistance. Unfavourable treatment outcomes were falling in Ravensmead and Uitsig, but were below 15% only in the last year that we studied. A continuation of this trend is necessary to have a significant effect on transmission (Maher et al. 2005). Alternatively, the high proportion of previously treated patients could be due to the high transmission and re-infection rate in the area (Verver et al. 2004).

Rising HIV prevalence could partly explain the increase in the ARTI (Odhiambo et al. 1999). In some settings however, although TB notification rates increase due to HIV, the prevalence of TB and the ARTI are less affected by HIV (Corbett et al. 2004; Egwaga et al. 2006). A possible explanation for this is that HIV-infected persons with TB may be infectious for a shorter period because fast progression to disease leads to death or early case detection and treatment. The shorter duration of infectiousness reduces the prevalence of undetected cases as well as the rate of transmission. Alternatively, the effect of an increasing HIV prevalence is not yet reflected in the ARTI estimate because the ARTI is a measure in the past – an average estimate centred on a point at half the age of the children tested.

Because an accurate measure of the HIV prevalence in the study area is lacking, we can only estimate this using other data available: In the district where Ravensmead and Uitsig are located the prevalence of HIV infection among mothers attending antenatal clinics has been reported to increase from 7.9% in 2001 to 15.1% in 2004 (Shaikh et al. 2006). However, we think the prevalence of HIV is lower in the communities we studied: Voluntary counselling and testing (VCT) of registered adult (≥15 years) TB patients during 2002 and 2003 showed a HIV seroprevalence among TB patients of 7.9% at Ravensmead clinic and 10.9% at Uitsig clinic (Personal communication Caldwell). However, the prevalence of HIV in the general population is likely to be lower than in TB patients (Corbett et al. 2003; WHO 2007). We therefore estimate that the prevalence of HIV among adults in the study area is approximately 2%, which corresponds with the reported estimate of 1.9% for the Western Cape Province in the population older than 2 years (Shisana et al. 2005). Death rates due to TB do not suggest an increase in HIV since the death rates have not increased between 1998 and 2002. The prevalence of HIV in these communities and the Western Cape Province in general is lower than the national prevalence of 10.8% in South Africa (Shisana et al. 2005) and can therefore not be generalized to the whole country. However, our findings of high TB transmission rates in these communities raise concerns about the ARTI in areas that are more affected by HIV.

Our study has several limitations. The interpretation of tuberculin survey data has some methodological difficulties that influence sensitivity and specificity of the test. These include small TST indurations following BCG vaccination (Farhat et al. 2006), misclassification of vaccination status (not all vaccinated children develop a typical scar) (Fjallbrant et al. 2008,Fine et al. 1989), loss of TST sensitivity due to HIV infection, and the influence of environmental mycobacteria on TST induration (Rieder 1995, 2005). In our setting however, there are little environmental mycobacteria which resulted in very few children with TST indurations between 7 and 10 mm. The prevalence of TB infection varied therefore little by the chosen cut-off points. The prevalence of infection and the ARTI were also very similar among BCG-vaccinated and non-vaccinated children and therefore we think that our trend estimates are fairly robust. We expect that very few 6 to 9-year old children in this community are infected with HIV, which is therefore unlikely to have influenced our estimates of prevalence of infection and ARTI.

Another limitation is that we had to obtain the smear positive notification rates for the period 1993–2002 from two sources. However, since both sources of information are based on the data in the treatment registers from the clinics, we expect they are comparable. By doing a school survey we only included the children that attended school which could have introduced a bias. Children from less poor families may go to school outside the study area, while children from poorer families may not attend school at all. But school attendance rates are high and we think they were unlikely to differ between the two surveys.

Although there seem to be associations between the ARTI, the high notification rates, the estimated HIV prevalence, and a high proportion of previously treated PTB cases, the ecological nature of this comparison warrants careful interpretation. In addition, our ARTI data represent the epidemiological situation of on average 4–5 years before the surveys and because of that notification and outcome trends may not correspond completely with the ARTI trends.

Conclusion

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. References

This study shows that there is an ongoing high transmission rate of M. tuberculosis to children in these high burden communities. TB control measures to date have failed to control TB and an intensified effort is needed to decrease the transmission of M. tuberculosis and to lower TB incidence in Ravensmead and Uitsig. Improving health services and the diagnostic process, active or intensified case finding, and active tracing of defaulters could all contribute to control of TB. Although treatment outcomes have been improving, a continuation of this trend is necessary to reach the 85% cure rate target and stop transmission. A follow up TST survey in 2011 will enable us to measure changing trends in the ARTI and to reassess the impact of the program.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. References

We thank Dr. Bernard Fourie from the South African Medical Research Council for his assistance with the first TST survey, and Dr. Ivan Toms, Director of Health of the City of Cape Town, and the Department of Education Western Cape province for permission to work in the communities. Judy Caldwell is kindly acknowledged for providing us with the notification and treatment outcome data of the clinics in Ravensmead and Uitsig. We are grateful to Katherine Lawrence for the data management and Nico Nagelkerke for advice on the statistical analysis. We thank the field team for collecting the data and the teachers and children of the primary schools in Ravensmead and Uitsig for their participation.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. References
  • Arnadottir T, Rieder HL, Trebuc QA & Waaler HT (1996) Guidelines for conducting tuberculin skin test surveys in high prevalence countries. Tubercle and Lung Disease 77, Suppl. 120.
  • Borgdorff MW, Floyd K & Broekmans JF (2002) Interventions to reduce tuberculosis mortality and transmission in low- and middle-income countries. Bulletin of the World Health Organization 80, 217227.
  • Bosman MC, Swai OB, Kwamanga DO et al. (1998) National tuberculin survey of Kenya. International Journal of Tuberculosis and Lung Disease 2, 272280.
  • Corbett EL, Watt CJ, Walker N et al. (2003) The growing burden of tuberculosis: global trends and interactions with the HIV epidemic. Archives of Internal Medicine 163, 10091021.
  • Corbett EL, Chaqralambous SC, Moloi VM et al. (2004) Human immunodeficiency virus and the prevalence of undiagnosed tuberculosis in African gold miners. American Journal of Respiratory and Critical Care Medicine 170, 673679.
  • Corrigal J, Coetzee D & Cameron N (2008) Is the Western Cape at risk of an outbreak of preventable childhood diseases? Lessons from an evaluation of routine immunisation coverage South African Medical Journal 98, 4145.
  • Den Boon S, Van Lill SWP, Borgdorff MW et al. (2007) High prevalence of previously treated tuberculosis among undetected cases of tuberculosis in Cape Town, South Africa. Emerging Infectious Diseases 13, 11891194.
  • Dubuis M, Fiekert K, Johnston M, Neuenschwander BE & Rieder HL (2004) A tuberculin skin test survey among Afghan children in Kabul. International Journal of Tuberculosis and Lung Disease 8, 10651072.
  • Dye C, Garnett GP, Sleeman K & Williams BG (1998) Prospects for worldwide tuberculosis control under the WHO DOTS strategy. Directly observed short-course therapy. Lancet 352, 18861891.
  • Egwaga SM, Cobelens FG, Muwinge H, Verhage C, Kalisvaart N & Borgdorff MW (2006) The impact of the HIV epidemic on tuberculosis transmission in Tanzania. AIDS 20, 915921.
  • El Ibiary S, De Coster EJ, Tolba FM et al. (1999) Trend in the annual risk of tuberculous infection in Egypt, 1950–1996. International Journal of Tuberculosis and Lung Disease 3, 294299.
  • Farhat M, Greenaway C, Pai M & Menzies D (2006) False-positive tuberculin skin tests: what is the absolute effect of BCG and non-tuberculous mycobacteria? International Journal of Tuberculosis and Lung Disease 10, 11921204.
  • Fine PE, Ponnighaus JM & Maine N (1989) The distribution and implications of BCG scars in northern Malawi. Bulletin of the World Health Organization 67, 3542.
  • Fjallbrant H, Ridell M & Larsson LO (2008) BCG scar and tuberculin reactivity in children and adults. Scandinavian Journal of Infectious Diseases 40, 387392.
  • Fourie PB (1987) BCG vaccination and the EPI. South African Medical Journal 72, 323326.
  • Gopi PG, Subramani R & Narayanan PR (2006) Trend in the prevalence of TB infection and ARTI after implementation of a DOTS programme in South India. International Journal of Tuberculosis and Lung Disease 10, 346348.
  • Maher D, Borgdorff M & Boerma T (2005) HIV-related tuberculosis: how well are we doing with current control efforts? International Journal of Tuberculosis and Lung Disease 9, 1724.
  • Middelkoop K, Bekker LG, Meyer L et al. (2008) Rates of tuberculosis transmission to children and adolescents in a community with a high prevalence of HIV infection among adults. Clinical Infectious Diseases 47, 349355.
  • Odhiambo JA, Borgdorff MW, Kiambih FM et al. (1999) Tuberculosis and the HIV epidemic: increasing annual risk of tuberculous infection in Kenya, 1986-1996. American Journal of Public Health 89, 10781082.
  • Rieder H (1995) Methodological issues in the estimation of the tuberculosis problem form tuberculin surveys. Tubercle and Lung Disease 76, 114121.
  • Rieder H (2005) Annual risk of infection with Mycobacterium tuberculosis. European Respiratory Journal 25, 181185.
  • Salaniponi FM, Kwanjana J, Veen J, Misljenovic O & Borgdorff MW (2004) Risk of infection with Mycobacterium tuberculosis in Malawi: national tuberculin survey 1994. International Journal of Tuberculosis and Lung Disease 8, 718723.
  • Shaikh N, Abdullah F, Lombard CJ, Smit L, Bradshaw D & Makubala L (2006) Masking through averages – intraprovincial heterogeneity in HIV prevalence within the Western Cape. South African Medical Journal 96, 538543.
  • Shisana O, Rehle T, Simbay LC et al. (2005) South African National HIV Prevalence, HIV Incidence, Behaviour and Communication Survey, 2005. HSRC Press, Cape Town.
  • StataCorp. (2003) Stata Statistical Software: Release 8.0. Stata Corporation, College Station, TX.
  • Statistical Support and Informatics (2001) Statistics. Western Cape, South Africa. Census 2001.
  • Stop TB Partnership and WHO (2006) Global Plan to Stop TB 2006-2015 (WHO/HTM/STB/2006.35). World Health Organization, Geneva.
  • Trébucq A, Guérin N, Ali Ismael H, Bernatas SS, Sèvre JP & Rieder HL (2005) Prevalence and trends of infection with Mycobacterium tuberculosis in Djiboeti, testing an alternative method. International Journal of Tuberculosis and Lung Disease 9, 10971104.
  • Tupasi TE, Radhakrishna S, Pascual ML et al. (2000) BCG coverage and the annual risk of tuberculosis infection over a 14-year period in the Philippines assessed from the nationwide prevalence surveys. International Journal of Tuberculosis and Lung Disease 4, 216222.
  • Verver S, Warren R, Munch Z et al. (2004) Transmission of tuberculosis in a high incidence urban community in South Africa. International Journal of Epidemiology 33, 351357.
  • WHO (2007) Global tuberculosis control. Surveillance, planning, financing. WHO report 2007. World Health Organization (WHO/HTM/TB/2007.376), Geneva.