Article Text

Download PDFPDF

P09 Estimation of the causal effect of church attendance on risk of Mycobacterium tuberculosis infection in young children in rural Malawi using targeted maximum likelihood estimation
  1. Palwasha Khan1,
  2. Kathy Baisley2,
  3. Leo Martinez3,
  4. Themba Mzembe4,
  5. Regina Chiumya4,
  6. Katharina Kranzer1,
  7. Paul Fine2,
  8. Katherine Fielding2,
  9. Amelia Crampin2,5,
  10. Judith Glynn2
  1. 1Clinical Research, London School of Hygiene and Tropical Medicine, London, UK
  2. 2Infectious Disease Epidemiology, London School of Hygiene and Tropical Medicine, London, UK
  3. 3Epidemiology, Boston University, School of Public Health, Boston, USA
  4. 4Karonga site, Malawi Epidemiology and Intervention Research Unit, Karonga, Malawi
  5. 5Lilongwe site, Malawi Epidemiology and Intervention Research Unit, Lilongwe, Malawi


Background Important gaps exist in our understanding of Mycobacterium tuberculosis transmission, especially in high HIV prevalence settings. There is significant uncertainty about where most transmission takes place in the community, impeding control efforts. M. tuberculosis infection in young children is a sensitive indicator of recent transmission and may provide a means of identifying locales of community transmission. We estimate the causal effect of church attendance on incident M. tuberculosis infection in young children.

Methods Children aged under six years residing within a demographic surveillance site in Malawi were recruited. Tuberculin skin testing (TST) was performed at baseline and repeated after 1–2 years. At the time of the 2nd TST, structured guardian interviews were undertaken to ascertain any known contact with tuberculosis within the previous 12–18 months. Detailed household socioeconomic and demographic data were available. Very few children had never attended church in the previous year (<4%) so the unexposed group was combined with the next category (1–3 attendances) to form the baseline. Incident infection was defined as an increase in TST induration of ≥13mm from 1st TST to 2nd TST among those with indurations of <10mm at baseline (based on mixture analysis). We used hierarchical targeted maximum likelihood estimation (TMLE) to estimate the causal effect of church attendance on infection incidence accounting for clustering at the community-level. Confounders included age, sex, household socioeconomic status and time between TSTs.

Results 2349 children were eligible for inclusion, of whom 2019 (86%) had a 2nd TST placed and data on church attendance. 66 (3.2%: 95% CI 2.5–4.1) children had evidence of TST-conversion; 3/1795 (1.4%: 95% CI 0.3 – 4.0) among those with <4 attendances and 633.3%: 95% CI 2.5–4.2) in those with ≥ 4 attendances. The estimated adjusted risk ratio was 2.8 (95%CI 1.2 – 6.7; p-value 0.023) for church attendance compared to baseline group.

Conclusion High levels of church attendance (≥ 4 per year) increased the risk of incident infection by 2.8-fold compared to church attendance of less than 4 attendances per year in young children in rural Malawi. Simple infection control practices, such as opening windows or even holding congregations in outdoor spaces, may be highly beneficial in mitigating the risk of community M. tuberculosis transmission and other respiratory/airborne pathogens.

  • causal effect
  • community M.tuberculosis transmission
  • church attendance

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.