Title
A household-level score to predict the risk of tuberculosis among contacts of patients with tuberculosis: a derivation and external validation prospective cohort study
Date Issued
01 January 2020
Access level
open access
Resource Type
journal article
Author(s)
Saunders M.J.
Wingfield T.
Evans B.E.W.
Publisher(s)
Lancet Publishing Group
Abstract
Background: The epidemiological impact and cost-effectiveness of social protection and biomedical interventions for tuberculosis-affected households might be improved by risk stratification. We therefore derived and externally validated a household-level risk score to predict tuberculosis among contacts of patients with tuberculosis. Methods: In this prospective cohort study, we recruited tuberculosis-affected households from 15 desert shanty towns in Ventanilla and 17 urban communities in Callao, Lima, Peru. Tuberculosis-affected households included index patients with a new diagnosis of tuberculosis and their contacts who reported being in the same house as the index patient for more than 6 h per week in the 2 weeks preceding index patient diagnosis. Tuberculosis-affected households were not included if the index patient had no eligible contacts or lived alone. We followed contacts until 2018 and defined household tuberculosis, the primary outcome, as any contact having any form of tuberculosis within 3 years. We used logistic regression to identify characteristics of index patients, contacts, and households that were predictive of household tuberculosis, and used these to derive and externally validate a household-level score. Findings: Between Dec 12, 2007, and Dec 31, 2015, 16 505 contacts from 3 301 households in Ventanilla were included in a derivation cohort. During the 3-year follow-up, tuberculosis occurred in contacts of index patients in 430 (13%, 95% CI 12–14) households. Index patient predictors were pulmonary tuberculosis and sputum smear grade, age, and the maximum number of hours any contact had spent with the index patient while they had any cough. Household predictors were drug use, schooling of the female head of a household, and lower food spending. Contact predictors were if any of the contacts were children, number of lower-weight (body-mass index [BMI] <20·0 kg/m2) adult contacts, number of normal-weight (BMI 20·0–24·9 kg/m2) adult contacts, and number of past or present household members who previously had tuberculosis. In this derivation cohort, the score c statistic was 0·77 and the risk of household tuberculosis in the highest scoring quintile was 31% (95% CI 25–38; 65 of 211) versus 2% (95% CI 0–4; four of 231) in the lowest scoring quintile. We externally validated the risk score in a cohort of 4248 contacts from 924 households in Callao recruited between April 23, 2014, and Dec 31, 2015. During follow-up, tuberculosis occurred in contacts of index patients in 120 (13%, 95% CI 11–15) households. The score c statistic in this cohort was 0·75 and the risk of household tuberculosis in the highest scoring quintile was 28% (95% CI 21–36; 43 of 154) versus 1% (95% CI 0–5; two of 148) in the lowest scoring quintile. The highest-scoring third of households captured around 70% of all tuberculosis among contacts. A simplified risk score including only five variables performed similarly, with only a small reduction in performance. Interpretation: This externally validated score will enable comprehensive biosocial, household-level interventions to be targeted to tuberculosis-affected households that are most likely to benefit. Funding: Wellcome Trust, Medical Research Council, Department of Health and Social Care, Department for International Development, Joint Global Health Trials consortium, Bill & Melinda Gates Foundation, Innovation for Health and Development.
Start page
110
End page
122
Volume
20
Issue
1
Language
English
OCDE Knowledge area
Sistema respiratorio
Enfermedades infecciosas
EpidemiologÃa
Scopus EID
2-s2.0-85076703736
PubMed ID
Source
The Lancet Infectious Diseases
ISSN of the container
14733099
Sponsor(s)
The authors would like to thank all members of the Innovation For Health And Development research team and members of tuberculosis-affected households in the study setting, without whom this research would not have been possible. This research and members of the research team were funded by the Wellcome Trust (057434/Z/99/Z, 070005/Z/02/Z, 078340/Z/05/Z, 105788/Z/14/Z, and 201251/Z/16/Z), Department for International Development Civil Society Challenge Fund, the Joint Global Health Trials consortium (Medical Research Council, Department for International Development, Department for Health and Social Care, and Wellcome Trust; award MR/K007467/1), the Bill & Melinda Gates Foundation (OPP1118545), the Foundation for Innovative New Diagnostics, the Sir Halley Stewart Trust, WHO, the STOP TB partnership's TB REACH initiative funded by the Government of Canada and the Gates Foundation (W5_PER_CDT1_PRISMA), and the charity Innovation For Health And Development. None of these organisations had any role in or placed any restrictions on the preparation or publication of this manuscript. This manuscript represents the research and opinion of the authors and not of any of these funding organisations
In this study of tuberculosis-affected households, which included more than 20 000 contacts from two independent cohorts, we derived and externally validated a household-level risk score that stratified households with large differences in the risk of tuberculosis occurring among contacts. This score combines data from readily collectable index patient, household, and contact characteristics into a model that could be used to target comprehensive biosocial, household-level interventions to households at highest risk of tuberculosis among contacts. Although these interventions should be considered for all tuberculosis-affected households, using a risk stratification approach could considerably improve their impact and cost-effectiveness, especially in resource-constrained settings. We envisage that this score could be used at the time of index patient diagnosis to prioritise enhanced active case finding among contacts to detect tuberculosis earlier at a less infectious stage, 20 preventive treatment, and social protection interventions to maximise access to health care and address poverty-related tuberculosis risk factors. 14 Although these interventions have potential benefits for all tuberculosis-affected households, our score could be used by decision makers to prioritise interventions in several ways. For example, one approach in severely resource-constrained settings might be to focus on the highest scoring third of households, of which around 25% are likely to have household tuberculosis within 3 years. This is a very high proportion when considering that tuberculosis typically affects less than 1% of households in a community at a given time. 1 In other settings, decision makers might use a more inclusive threshold. For example, prioritising the highest scoring two thirds of households would capture more than 90% of all tuberculosis among contacts. Prioritising a higher proportion of tuberculosis-affected households is likely to increase the epidemiological impact of household-level interventions and could be balanced against the availability of resources in specific settings. Our strategy to derive a score was based on a preconceived framework of index patient, household, and contact factors that we considered to be potential predictors of household tuberculosis. We showed an approximately linear relationship between index patient type of tuberculosis and sputum smear grade and risk of household tuberculosis. In our previous work among adult contacts of patients with tuberculosis who were nearly all sputum smear positive, 11 smear grade did not predict tuberculosis among individuals. However, in this larger study, a strength of which is the inclusion of patients with extra-pulmonary tuberculosis and bacteriologically unconfirmed pulmonary tuberculosis, type of tuberculosis and smear grade strongly predicted tuberculosis assessed at a household level. This finding might partly be because another strength of this study was the inclusion of children, whose principal exposure is more likely to be the current index patient. By contrast, our previous study only included adults, who might have had multiple exposures throughout their lives, reducing the importance of the infectiousness of the currently diagnosed index patient. 21 Although an isolated sputum smear result is probably a crude measure of infectiousness, smear microscopy is still the most widely available diagnostic test globally, particularly in resource-constrained settings. 22 Optimising microscopy by use of strategies such as viability staining 23 to identify the most infectious patients could complement and further improve our score. We showed in multivariable analysis that the maximum duration of exposure any contact had to the index patient while they had cough predicted household tuberculosis in a dose-dependent relationship, independently of index patient type of tuberculosis and sputum smear grade. We also showed an association between index patient age and risk of household tuberculosis. Households in which the index patient was younger than 20 years had the highest risk of tuberculosis, possibly because of undetected tuberculosis among adults in the household. 20 Households in which the index patient was 50 years or older had the lowest risk of tuberculosis, possibly because these patients more commonly present with atypical symptoms, 24 and because older people might be more commonly isolated when unwell. Although we did not observe an increased risk of household tuberculosis among households affected by rifampicin-resistant tuberculosis, the individual, household, and public health consequences of rifampicin-resistant tuberculosis strongly support the prioritisation of these households for interventions, independent of this risk score. The other variables included in our score show that a biosocial approach is essential to ending the tuberculosis epidemic. 25 Tuberculosis inequitably affects poorer households, principally in lower-income countries. 9 Our finding that households in which the female head had less schooling, a general marker of household poverty, 26 had a higher risk of household tuberculosis further supports this association between tuberculosis and poverty. Economic prosperity, leading to improved living conditions and better nutrition, is recognised as the most important driver of the reduction in tuberculosis incidence in western Europe during the pre-antibiotic era. 9 Since then, multiple studies have shown the inverse association between tuberculosis incidence and socioeconomic development, including government spending on social protection. 27–29 In this study, we extended our previous findings showing the role of nutritional factors in determining tuberculosis risk. 11 Our results suggest that it is the nutritional status of contacts, and not the overall number of contacts, that best predicts which households will have a contact who has tuberculosis. For example, in multivariable analysis, the number of adult contacts of lower weight greatly increased risk of household tuberculosis, the number of adult contacts of normal weight somewhat increased risk, and the number of adult contacts who were overweight did not increase risk. Although we did not observe a clear linear relationship between the number of child contacts, or their weight, and household tuberculosis, households that included children were at substantially higher risk of household tuberculosis than those without children. This finding reinforces that optimal management of child contacts is needed to reduce childhood tuberculosis morbidity and mortality. 3 Relatedly, we showed that households that spent relatively less on food per person had an increased risk of household tuberculosis. This variable might be an indicator of overall monetary poverty, reflected by the fact that the score performed equally well when replaced by income, and might also reflect food security and the quality of food consumed by household members. Furthermore, households that included anyone who used drugs were at higher risk of household tuberculosis, and households that included anyone who drank alcohol to excess were at higher risk of household tuberculosis in univariable analysis. Therefore, our score might be used to prioritise holistic interventions that aim to optimise the nutritional status of members of households at highest risk and address these harmful health behaviours. As well as reducing tuberculosis risk, improved nutrition and reduced harmful substance use are likely to have far reaching health benefits. We showed a dose-dependent increased risk of household tuberculosis among households previously affected by tuberculosis. This supports our approach of deriving a household-level risk score because tuberculosis clusters in households, frequently affecting multiple household members, which might be explained by an increased number of exposures for all contacts, previous tuberculosis conferring a high risk of subsequent tuberculosis among individuals, 30 and by the fact that households previously affected by tuberculosis are likely to be poorer than households that have never been affected by tuberculosis. 31 This study had some limitations. We might have underestimated the number of households in which a contact had tuberculosis because we did not actively follow-up contacts to establish tuberculosis diagnoses outside the study setting (eg, in private health facilities). However, our previous work 20 showed that these cases account for a small proportion of the overall tuberculosis burden among contacts so are unlikely to have affected our results. We were unable to account for censoring of households that moved away or contacts who died. However, our experiences of working in this setting since 2002 suggest migration and death are rare. 20 Similarly, we did not collect data on how variables changed over time because we aimed to derive a score that could be used at the time of index patient diagnosis using baseline data. Although a substantial proportion of households had missing data on some variables in the derivation cohort, we used robust multiple imputation methods to complete these data and facilitate increased power for score derivation, evaluated the score only among households with complete data, and externally validated the score in a distinct cohort of households for which the majority had complete data. A strength of our approach is the inclusion of variables that are likely to be consistent and easily recordable by health workers across settings, which, given their role in determining tuberculosis risk, should be integrated into routinely collected data systems by national tuberculosis programmes. In settings where some of these data are not available, the simplified score including only five variables could be used with only a small reduction in performance. We did not have data on other risk factors, such as household ventilation or HIV infection among contacts. However, HIV prevalence in Peru is low (about 0·2% of women aged 15–49 years) and is unlikely to affect the interpretation of our results. 32 In our previous study, 11 we showed an increased tuberculosis risk among contacts exposed to indoor air pollution from cooking fuels. We were unable to investigate this variable for our current cohort because there has been a near universal shift to clean, gas cookers in our setting. The use of our score in other settings should, therefore, consider local epidemiology (including HIV prevalence), health behaviours, and household characteristics. Programmatic interventions targeted to tuberculosis-affected households aim to detect and prevent all cases of tuberculosis among contacts, irrespective of the source of infection. Therefore, we did not use molecular techniques to confirm transmission from index patients to contacts because it would not affect our findings. In conclusion, we derived and externally validated a simple household-level risk score that stratifies tuberculosis-affected households with different risks of tuberculosis among contacts. The score had similar predictive performance in derivation and external validation cohorts, with excellent calibration in the external validation cohort, lending promise to the use, further validation, and impact evaluation of our score in other settings. Contributors MJS and CAE designed the study. MJS, TW, SD, RM, ER, MAT, and CAE collected data. MJS and CAE analysed the data, and MJS, TW, SD, MRB, MAT, RHG, and CAE interpreted the data. All authors prepared, reviewed, and approved the manuscript as submitted. Declaration of interests We declare no competing interests. Acknowledgments The authors would like to thank all members of the Innovation For Health And Development research team and members of tuberculosis-affected households in the study setting, without whom this research would not have been possible. This research and members of the research team were funded by the Wellcome Trust (057434/Z/99/Z, 070005/Z/02/Z, 078340/Z/05/Z, 105788/Z/14/Z, and 201251/Z/16/Z), Department for International Development Civil Society Challenge Fund, the Joint Global Health Trials consortium (Medical Research Council, Department for International Development, Department for Health and Social Care, and Wellcome Trust; award MR/K007467/1), the Bill & Melinda Gates Foundation (OPP1118545), the Foundation for Innovative New Diagnostics, the Sir Halley Stewart Trust, WHO, the STOP TB partnership's TB REACH initiative funded by the Government of Canada and the Gates Foundation (W5_PER_CDT1_PRISMA), and the charity Innovation For Health And Development. None of these organisations had any role in or placed any restrictions on the preparation or publication of this manuscript. This manuscript represents the research and opinion of the authors and not of any of these funding organisations.
Sources of information:
Directorio de Producción CientÃfica
Scopus