Article Text

Download PDFPDF

Diabetes mellitus and latent tuberculosis infection: baseline analysis of a large UK cohort
Free
  1. Charlotte Jackson1,
  2. Jo Southern2,
  3. Ajit Lalvani3,
  4. Francis Drobniewski3,4,
  5. Chris J Griffiths5,
  6. Marc Lipman6,7,
  7. Graham H Bothamley8,
  8. Jonathan J Deeks9,10,
  9. Ambreen Imran2,
  10. Onn Min Kon11,12,
  11. Sithembinkosi Mpofu2,
  12. Vladyslav Nikolayevskyy3,4,13,
  13. Melanie Rees-Roberts3,
  14. Alice Sitch9,
  15. Saranya Sridhar3,
  16. Chuen-Yan Tsou4,
  17. Hilary Whitworth3,14,
  18. Ibrahim Abubakar1
  1. 1 Institute for Global Health, University College London, London, UK
  2. 2 TB Section, Public Health England, London, UK
  3. 3 Department of Medicine, Imperial College London, London, UK
  4. 4 National Mycobacterium Reference Laboratory, Public Health England, London, UK
  5. 5 Centre for Primary Care and Public Health, Queen Mary University of London, London, UK
  6. 6 Respiratory & HIV Medicine, Royal Free London NHS Foundation Trust, London, UK
  7. 7 UCL Respiratory, Division of Medicine, University College London, London, UK
  8. 8 Respiratory Medicine, Homerton University Hospital NHS Foundation Trust, London, UK
  9. 9 Test Evaluation Group, Institute for Applied Health Research, University of Birmingham, Birmingham, UK
  10. 10 National Institute for Health Research (NIHR) Birmingham Biomedical Research Centre, Birmingham, UK
  11. 11 Respiratory Medicine, Imperial College Healthcare NHS Trust, London, UK
  12. 12 National Heart and Lung Institute, Faculty of Medicine, Imperial College London, London, UK
  13. 13 Blizard Institute, Queen Mary University of London, London, UK
  14. 14 Department of Clinical Research, London School of Hygiene and Tropical Medicine, London, UK
  1. Correspondence to Dr Charlotte Jackson, Institute for Infection and Immunity, St George’s, University of London, London SW17 0RE, UK; cjackson{at}sgul.ac.uk

Abstract

We conducted a cross-sectional analysis of baseline data from a UK cohort study which enrolled participants at risk of latent tuberculosis infection (LTBI, defined as a positive result for either of the two interferon gamma release assays). Binomial regression with a log link was used to estimate crude and adjusted prevalence ratios (PRs) and 95% CIs for the relationship between diabetes mellitus (DM) and LTBI. Adjusted for age, sex, ethnicity, body mass index and the presence of other immunocompromising conditions, DM was associated with a 15% higher prevalence of LTBI (adjusted PR=1.15, 95% CI 1.02 to 1.30, p=0.025).

Trial registration number PREDICT is registered on clinicaltrials.gov (NCT01162265)

  • tuberculosis
  • clinical epidemiology
View Full Text

Statistics from Altmetric.com

Diabetes mellitus (DM) and tuberculosis (TB) are major global public health priorities.1 2 Many studies have assessed the relationship between DM and active TB disease.3 Data on the effect of DM on the risk of latent TB infection (LTBI) are more limited. A recent systematic review identified one cohort study, with an adjusted risk ratio of 4.40 (95% CI 0.50 to 38.55) and 12 cross-sectional studies which generated a pooled adjusted OR of 1.18 (95% CI 1.06 to 1.30).4

The Prognostic Evaluation of Diagnostic IGRAs Consortium (PREDICT) study was a prospective, multisite UK cohort study aiming to evaluate the predictive values of interferon gamma release assays (IGRAs) for the development of active TB among recent entrants to the UK from high-burden countries and contacts of active TB cases (‘contacts’). PREDICT is registered on clinicaltrials.gov (NCT01162265). In this study, we use baseline data from PREDICT to investigate the association between DM and LTBI.

Recruitment took place between January 2011 and July 2015. After giving informed consent, participants completed a questionnaire and provided blood samples for IGRAs. Participants with evidence of active TB were excluded. The main exposure of interest in this secondary analysis was a self-reported history of DM. Data were also collected on the method of DM control used. The outcome of interest was LTBI, defined as a positive result for either or both of the two commercially available IGRAs, Quantiferon-TB Gold In-Tube (QFT-GIT, Qiagen, Hilden, Germany) and TSpot.TB (Oxford Immunotec, Abingdon, UK). Participants with no valid IGRA results were excluded from this analysis. Other covariates on which data were collected are described in the online Supplementary data.

Supplemental material

Binomial regression with a log link was used to estimate crude and adjusted prevalence ratios (PRs) and 95% CIs for the relationship between DM and LTBI.5 Age and sex were treated as a priori confounders. A causal diagram of the relationships between potential confounders and outcomes using directed acyclic graphs, interpreted using dagitty.net6 (online supplementary figure S1), was used to identify the minimum set of other covariates required for adjustment. P values were derived from likelihood ratio tests. We assessed potential interactions between DM and age3 and DM and ethnicity,7 as observed for active TB.3 7 All analyses used a complete-case approach. We conducted sensitivity analyses: (1) adjusting for age as a continuous variable using fractional polynomials8; (2) using Poisson regression with robust standard errors9; (3) restricting analysis to contacts; (4) including only participants who had concordant results for the two IGRAs; (5) repeating the primary analysis additionally adjusting for country of birth. Further methodological details and the questionnaire are provided in the Supplementary data.

Nine thousand one hundred and fifty seven participants were included in the analysis (table 1, online Supplementary table S1, figure S2). Seven hundred and fifty six participants (8.3%) reported having diabetes, of whom 535 provided information about how they controlled the condition: 409 taking medication, 55 on insulin, 20 using both insulin and other medication(s) and 51 through monitoring and/or diet only.

Table 1

Characteristics of participants with and without LTBI and unadjusted prevalence ratios for the association with LTBI

Prevalence of a positive IGRA was 31.5% and 27.3% among those with and without DM, respectively (table 1: unadjusted PR=1.15, 95% CI 1.03 to 1.29, p=0.012). Characteristics associated with a positive IGRA on univariate analysis included increasing age, male sex, being born outside the UK, being a contact, having had a previous TB diagnosis or previous contact with a patient with TB and immunosuppression (table 1). IGRA positivity varied by ethnicity, being highest in the Black African ethnic group and lowest among Black Caribbean participants. There was no evidence that having a positive IGRA was associated with previous BCG vaccination, HIV status, body mass index (BMI), smoking or social risk factors (table 1).

Complete covariate data were available for 8336 participants (91.0% of the included participants; online Supplementary table S2 compares these 8336 participants with the 821 with incomplete data). Adjusting for sex, age group, ethnicity, immunosuppression and BMI, the adjusted PR for the association between DM and LTBI was 1.15 (95% CI 1.02 to 1.30, p=0.025, online Supplementary table S3). There was no evidence of interaction between DM and age group (p=0.22) and weak evidence of interaction between DM and ethnicity (p=0.055, online Supplementary table S4). Sensitivity analyses produced similar results, although the adjusted PR increased to 1.29 (95% CI 1.09 to 1.52, p=0.002) when analysis was restricted to contacts (online Supplementary table S5).

Our results are likely to be generalisable to migrants and contacts in the UK (although there were some differences between participants included and excluded from the analysis (online Supplementary tables S1 and S2)), but perhaps not to other settings with different distributions of risk factors including country of birth and ethnicity. We used both of the commercially available IGRAs and conducted a sensitivity analysis restricted to participants with concordant results, providing additional certainty regarding the diagnosis of LTBI. Limitations of the study include the self-reported nature of DM status, although this was frequently supported by reported use of insulin or oral hypoglycaemic agents. Any participants with undiagnosed DM would be misclassified; this would be non-differential with respect to IGRA status and could bias our estimates towards the null. It is also possible that DM (and other forms of immunosuppression) influences the response to IGRA.10

This is a cross-sectional analysis, so we cannot be certain whether DM onset preceded LTBI. However, the association persisted when analysis was restricted to contacts, who were considered likely to have acquired infection recently. Residual confounding (eg, by socioeconomic status) could inflate our estimated adjusted PRs. Reported HIV prevalence was low and may be an underestimate as it was based on self-report.

Consistent with a previous systematic review and meta-analysis,4 this study suggests that after adjustment for age, sex, BMI, ethnicity and immunosuppression, DM is associated with a small increase in the prevalence of positive IGRA results, among individuals at high risk of LTBI. Prospective studies are needed to further investigate the temporal relationship between DM and both infection and disease onset.

Acknowledgments

We are grateful to all members of the PREDICT Study Group: IA, David Adeboyeku, Nabeela Bari, Jack Barker, Helen Booth, Graham Bothamley, Felix Chua, Dean Creer, Mathina Darmalingam, Robert N Davidson, Martin Dedicoat, JJD, FD, Anne Dunleavy, Jose Figueroa, CJG, Pranab Haldar, Mimi Haseldean, Andrew Hayward, Norman Johnson, OMK, Heinke Kunst, AL, ML, Stefan Losewicz, Joanne Lord, William Lynn, Bobby Mann, Heather Milburn, John Moore-Gillon, Geoff Packe, Anton Pozniak, Frances Sanderson and JS. We also thank Beverley Marks (the study administrator), laboratory staff who undertook tests, clinical and nursing colleagues who contributed to participant recruitment and our study steering committee (Professor Bertie Squire—chair, Dr Sani Aliyu, Dr Stuart Baugh, Aurora Dawson) and data monitoring committee (Dr Chapman—Chair, Dr Adrian Smith and Dr Jon Innes, Dr Jonathan Deeks). We are also grateful to Sue Dart as well as all the temples, mosques, offices and other congregate settings for their assistance.

References

View Abstract

Footnotes

  • Contributors IA conceived and oversaw the design and conduct of this analysis. JS coordinated the study and contributed to study design. CJG, ML, GHB and OMK contributed to study design and recruitment. AL, FD and JJD contributed to study design. CJ conducted statistical analysis with support from AS and JJD and drafted the manuscript. AI and SM recruited participants. VN, MR-R, SS, C-YT and HW performed laboratory work supervised by AL and FD. All authors critically reviewed and contributed to the manuscript. IA is the chief investigator of the PREDICT study; FD and AL are co-PIs.

  • Funding This work was supported by the National Institute for Health Research [grant numbers NIHR HTA 08/68/01, NIHR SRF-2011-04-001, NIHR NF-SI-0616-10037 to IA]. FD was supported by the Imperial Biomedical Research Centre.

  • Competing interests CJ has undertaken paid consultancy work for Otsuka Pharmaceutical unrelated to the content of this paper. AL has several issued patents underpinning immunodiagnostics for tuberculosis. The ESAT-6/CFP-10 interferon-gamma ELISpot was commercialised by an Oxford University spin-out company (Oxford Immunotec, Abingdon, UK) from which Oxford University and AL have royalty entitlements.

  • Patient consent Not required.

  • Ethics approval PREDICT study was approved by the Brent Research Ethics Committee (reference 10/H0717/14).

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Presented at Interim results of this study were presented at the British Thoracic Society Winter Meeting, London, December 2013 (abstract number S57).

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.