Abstract
Background We aimed to evaluate the effectiveness of contact investigation in comparison with passive case detection alone, and estimate the yield of co-prevalent and incident tuberculosis (TB) and latent TB infection (LTBI) among contacts of patients with TB.
Methods A systematic search was undertaken of studies published between 1 January 2011 and 1 October 2019 in the English language. The proportion of contacts diagnosed with co-prevalent TB, incident TB and/or LTBI was estimated. Evaluation of the effectiveness of contact investigation included randomised trials, while the yield of contact investigation (co-prevalent/incident TB and LTBI) was assessed in nonrandomised studies.
Results Data were extracted from 244 studies, of which 187 studies measured the proportion of contacts diagnosed with TB disease and 135 studies measured LTBI prevalence. Individual randomised trials demonstrated that contact investigation increased TB case notification (relative risk 2.5, 95% CI 2.0–3.2) and TB case detection (OR 1.34, 95% CI 0.43–4.24) and decreased mortality (relative risk 0.6, 95% CI 0.4–0.8) and population TB prevalence (risk ratio 0.82, 95% CI 0.64–1.04). The overall pooled prevalence of TB was 3.6% (95% CI 3.3–4.0%; I2=98.9%, 181 studies). The pooled prevalence of microbiologically confirmed TB was 3.2% (95% CI 2.6–3.7%; I2=99.5%, 106 studies). The pooled incidence of TB was highest in the first year after exposure to index patients (2.0%, 95% CI 1.1–3.3%; I2=96.2%, 14 studies) and substantially lower 5 years after exposure to index patients (0.5%, 95% CI 0.3–0.9%; one study). The pooled prevalence of LTBI among contacts was 42.4% (95% CI 38.5–46.4%; I2=99.8%, 135 studies).
Conclusions This systematic review and meta-analysis found that contact investigation was effective in high-burden settings. The higher pooled prevalence estimates of microbiologically confirmed TB compared with previous reviews suggests newer rapid molecular diagnostics contribute to increased case detection.
Abstract
Contact investigation is associated with increased case detection, reduced mortality and decreased community prevalence of TB, and resulted in a high yield of co-prevalent, incident and latent TB infection among contacts https://bit.ly/3h9jW2H
Introduction
Tuberculosis (TB) remains a major global public health challenge owing to its high morbidity and mortality rates. In 2018, 10 million people fell ill with the airborne respiratory infection caused by Mycobacterium tuberculosis; however, only 7 million of these cases were diagnosed [1]. This substantial case detection gap leads to substantial morbidity and mortality [1]. Consequently, the scale-up of strategies to identify people with TB disease and increase access to TB preventive therapy (TPT) is urgently required [2]. Targeted active case finding is a top priority for healthcare providers and policy makers [3].
Contacts of patients with TB have a substantially increased risk of developing TB compared with the general population [4]. This is reflected by the high prevalence of latent TB infection (LTBI) among contacts. Contact investigation entails screening for TB, with or without screening for LTBI. This typically includes a combination of symptom screening, chest radiography and/or the tuberculin skin test (TST). Microbiological tests used to confirm the diagnosis of TB include direct sputum smear, culture or the nucleic acid amplification test (NAAT). Contact investigation is also a priority for drug-resistant TB patients and scale-up of NAATs enables the rapid detection of drug-resistant TB. Contact investigation has been shown to increase case detection [5], facilitate earlier treatment initiation, and enable the diagnosis and treatment of LTBI in low- and high-incidence settings [6, 7]. Contact investigation may also reduce M. tuberculosis transmission in the community [8]. Recent randomised trials have provided new insights into the effectiveness of contact investigation for individuals and communities [5, 9]. Furthermore, the scale-up of rapid molecular diagnostic tests, such as GeneXpert MTB/RIF and Truenat, has been increasingly used to increase the sensitivity of screening algorithms to detect TB [10].
This systematic review and meta-analysis was undertaken to inform updated World Health Organization (WHO) guidelines regarding TB screening among contacts of TB patients. This review aimed to 1) evaluate evidence for the effectiveness of contact investigation, 2) estimate co-prevalent and incident TB and the prevalence of LTBI among contacts, and 3) evaluate the sensitivity of alternative algorithms for identifying co-prevalent TB among contacts of patients with TB.
Methods
Search strategy and study eligibility
We conducted a systematic review and meta-analysis of study-level observational data and reported these according to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) and MOOSE (Meta-analysis Of Observational Studies in Epidemiology) guidelines [11, 12]. A review protocol was developed and approved by the WHO secretariat. We searched four databases: MEDLINE (via PubMed), Embase, LILACS and Web of Science. A database-specific search strategy was used that included MeSH (Medical Subject Headings) terms or free-text words such as “Mycobacterium tuberculosis”, “tuberculosis”, “contact tracing”, “contact screening”, “disease outbreaks” and “case finding”. The complete search strategy is listed in the supplementary material. We searched the reference lists of included studies to identify additional papers for consideration.
Included studies were published between 1 January 2011 and 1 October 2019 in the English language, and provided a quantitative measure of the proportion of contacts diagnosed with co-prevalent TB, incident TB or LTBI. The evaluation of the effectiveness of contact investigation included randomised trials in which contact investigation was compared with standard “passive” approaches to case detection. The evaluation of the yield of contact investigation (co-prevalent and incident TB and LTBI) included nonrandomised studies. Studies were excluded if <10 index cases were evaluated, if index patients had not been identified as the starting point for contact investigation, the number of contacts screened was not reported (i.e. no denominator data was available), or TB among contacts was not diagnosed on microbiological or standard clinical grounds (e.g. suitable grounds include screening algorithms such as chest radiography plus symptoms and microbiology plus chest radiography). Conference abstracts were excluded.
Screening and data extraction
Each title and abstract identified in the literature search was independently screened by two reviewers (R.V.S. and M.V.). When the two reviewers’ assessments of eligibility differed, a third reviewer (K.V.) made a final decision. Manuscripts considered for full-text review were uploaded onto Endnote X8 (https://endnote.com) and independently screened for inclusion by two reviewers (R.V.S. and K.V.). Cohen's κ was used to measure inter-coder agreement at each screening phase; a poor correlation was defined as a κ<0.5.
A standardised data extraction form in REDCap [13] captured data elements from each included manuscript. Data extraction was completed by three reviewers (R.V.S., J.H. and K.V.). Extracted data included: 1) assessment of eligibility criteria and study design; 2) methods used for screening (timing of TB screening in relation to the diagnosis of the index case, TB screening and diagnostic testing modalities, and location of exposure); and 3) screening outcomes (characteristics of the index case, number of index cases in the study, number of contacts evaluated for TB disease and LTBI, number of contacts tested for TB or LTBI and number of contacts diagnosed with TB or LTBI, and HIV prevalence).
Definitions
An index case was defined as the initially identified case of new or recurrent TB, in a person of any age, in a specific household or other comparable setting in which others may have been exposed [14].
A contact was defined as any person who had been exposed to an index case [14]. A household contact was defined as any person who shared the same enclosed living space for 7 nights, or for frequent or extended periods during the day, with the index case during the 3 months before commencement of the current treatment episode [14]. A “close” contact was defined as any person who was not in the household but shared an enclosed space with the index case (such as a place of social gathering, workplace or facility) for extended time periods during the 3 months prior to the index commencing the current treatment episode [14].
The income classification of the country in which studies were conducted was classified by World Bank criteria for Gross National Income per capita: high, upper-middle, lower-middle and low income [15]. Countries were classified as low (<20 cases per 100 000 population), medium (20–99 cases per 100 000 population) or high (≥100 cases per 100 000 population) incidence for TB [1].
“Co-prevalent TB” was defined as TB that was diagnosed clinically and/or microbiologically (using any combination of smear microscopy, mycobacterial culture, NAAT, chest radiography and clinical grounds) among contacts during the baseline contact investigation, usually within 3 months of index patient diagnosis. “Incident TB” was defined as TB diagnosed clinically and/or microbiologically (using any combination of smear microscopy, mycobacterial culture, NAAT, chest radiography and clinical grounds) among contacts at any time, post the initial baseline contact investigation. Where co-prevalent and incident TB could not be separated based on available data, these were regarded as co-prevalent TB. Where a study separated the contacts into co-prevalent and incident TB, we allocated TB as either co-prevalent or incident.
LTBI was defined as a positive TST or interferon-γ release assay (IGRA) result (where TST was not available); if the study reported both results, TST was used.
Assessment of study quality
A risk-of-bias assessment was performed for all included studies using the Newcastle–Ottawa Scale (NOS), as described in the supplementary material [16].
Data synthesis and analysis
Characteristics of included studies were analysed using descriptive statistics. A meta-analysis was performed where two or more studies were included that evaluated similar outcomes. Effectiveness was measured at the level of the general population and of contacts, and outcome measures included TB case notification, TB case detection, mortality and TB prevalence. Outcomes of individual randomised studies are presented when meta-analyses were not possible.
Pooled estimates of co-prevalent TB, incident TB and LTBI among contacts were calculated for nonrandomised studies using a random effects model. Pooled estimates were weighted by the number of included participants and 95% confidence intervals calculated using Wilson's method. Within-study standard errors were calculated as the square root of the sum of the inverse of the number of contacts with each outcome of interest and the inverse of the number of contacts without that outcome. Pooled estimates were also calculated for specific subpopulations described in supplementary table E1. TB incidence estimates among contacts were stratified by the number of years between the diagnosis of the index case and the onset of TB diagnosed in contacts, up to 5 years.
Publication bias was investigated using funnel plots and the Egger test [17, 18]. Between-study heterogeneity was measured using the I2 statistic. A statistical significance threshold of p<0.05 was used [19].
Statistical analyses were performed using Stata version 16 (StataCorp, College Station, TX, USA).
Results
Study characteristics
Between January 2011 and October 2019, 244 eligible papers were identified (figure 1). Of these, 187 papers measured the proportion of contacts diagnosed with TB disease, comprising 1 404 453 individuals from 61 countries (table 1). The prevalence of LTBI was measured in 135 studies, comprising 473 075 individuals. Among studies evaluating TB disease (n=187), the median (interquartile range (IQR)) number of included contacts per study was 646 (255–2666). The majority of studies (62.6%) were conducted in countries with a high incidence of TB. Six randomised studies (3.2%) and 181 nonrandomised studies (96.8%) were identified. A summary of included studies, grouped by the country incidence of TB, is shown in table 1.
Evidence for the effectiveness of contact investigation
Three randomised control trials met the eligibility criteria for the assessment of effectiveness of contact investigation [5, 9, 20]. A meta-analysis was not performed for these studies owing to heterogeneity in their methods and outcome measures (supplementary table E2). Three other randomised controlled trials were excluded for reasons outlined in supplementary table E3 [21–23].
The effect of household contact investigation upon TB case detection was reported in one randomised trial, performed in Uganda [20]. This study found that seven out of 471 individuals were diagnosed with TB in the intervention group versus five out of 448 individuals in the control group (OR 1.34, 95% CI 0.43–4.24).
A second randomised trial evaluated the effect of household contact investigation, integrated with HIV care, upon TB prevalence in the general community [9]. The study found that 443 out of 43 941 individuals were diagnosed with TB with household screening compared with 451 out of 45 763 individuals in the control group (risk ratio 0.82, 95% CI 0.64–1.04).
TB case notification was reported in a third randomised trial performed in Vietnam. In this study, 180 out of 10 069 screened household contacts developed TB versus 110 out of 15 638 contacts in the control group in which passive case finding alone was performed. In the screening group, the relative risk of notified TB was 2.5 (95% CI 2.0–3.2) compared with the control group [5]. Additionally, 160 out of 10 069 individuals in the intervention group developed microbiologically confirmed TB versus 39 out of 15 638 individuals in the control group (relative risk 6.4, 95% CI 4.5–9.0). The screening intervention was associated with 40% reduction in mortality compared with the control group (relative risk 0.6, 95% CI 0.4–0.8) in a post hoc analysis.
The included randomised trials were assessed as having “good” quality (supplementary table E4).
Co-prevalent and incident TB in nonrandomised studies
We included 181 nonrandomised studies that evaluated the proportion of patients with co-prevalent or incident TB. Table 2 shows the pooled estimates of co-prevalent TB detected among contacts, stratified by subgroups of index cases and contacts. Figure 2 shows the pooled estimates for incident TB detected among contacts up to 5 years after exposure to the index patient. Among the 181 nonrandomised studies, the median (IQR) NOS score was 4 (3–5), out of a maximum of 9 for cohort studies and 7 for cross-sectional studies. Using the Agency for Healthcare Research and Quality scale, only 37 studies (19%) were considered “good” quality (supplementary table E5).
Co-prevalent TB detected among contacts
The overall pooled TB prevalence (clinically and/or microbiologically diagnosed) was 3.6% (95% CI 3.3–4.0%; I2=98.9%, 181 studies). The pooled TB prevalence of microbiologically confirmed TB was 3.2% (95% CI 2.6–3.7%; I2=99.5%, 106 studies). The pooled TB prevalence was 3.8% (95% CI 3.3–4.3%; I2=98.4%, 100 studies) among contacts of microbiologically confirmed index patients and 4.1% (95% CI 2.8–5.6%; I2=98.1%, 20 studies) among contacts of multidrug-resistant (MDR)/extensively drug-resistant (XDR)-TB index patients (table 2).
Pooled prevalence estimates, stratified by income classification and TB incidence of the country in which studies were performed, are shown in table 2. Studies conducted in low-income countries had a pooled TB prevalence of 5.0% (95% CI 4.0–6.1%; I2=98.7%, 35 studies), while high-income countries had a pooled TB prevalence of 1.8% (95% CI 1.3–2.3%; I2=98.8%, 42 studies). Studies conducted in low-incidence countries had a pooled TB prevalence of 1.9% (95% CI 1.5–2.4%; I2=97.7%, 37 studies), while high-incidence countries had a pooled TB prevalence of 5.0% (95% CI 4.5–5.5%; I2=99.0%, 111 studies).
Pooled TB prevalence was highest among contacts infected with HIV (11.6%, 95% CI 8.2–15.4%; I2=81.7%, 17 studies), followed by contacts ≥15 years old (5.2%, 95% CI 3.7–6.8%; I2=99.0%, 24 studies) (table 2).
Symptom screening as part of an initial algorithm was included in 113 studies (60.4%), while smear microscopy was the most reported microbiological test used (n=91 (48.6%)). The pooled TB prevalence among contacts varied according to the initial diagnostic test performed and the confirmatory microbiological test used to detect TB (table 3 and supplementary table E8).
Incident TB detected among contacts
The pooled annual TB incidence was highest in the first year after exposure to index patients (2.0%, 95% CI 1.1–3.3%; I2=96.2%, 14 studies) and significantly reduced 5 years after exposure to index patients (0.5%, 95% CI 0.3–0.9%, one study) (figure 2).
LTBI prevalence among contacts
We identified 135 studies which reported on LTBI testing and prevalence, comprising 473 075 individuals (table 4). The overall pooled prevalence of LTBI among contacts was 42.4% (95% CI 38.5–46.4%; I2=99.8%, 135 studies). When stratified by TB incidence of study setting, pooled LTBI prevalence was 28.6% (95% CI 22.7–34.8%; I2=99.8%, 37 studies) in low-incidence settings, 44.7% (95% CI 36.5–53.1%; I2=99.4%, 39 studies) in medium-incidence settings and 50.5% (95% CI 46.7–54.3%; I2=99.2%, 59 studies) in high-incidence settings.
Evaluation of algorithms for identifying co-prevalent TB among contacts
The pooled TB prevalence by initial screening and microbiological test(s) is shown in table 3. Studies which reported the use of symptom screening and chest radiography had the highest pooled prevalence at 4.4% (95% CI 3.8–5.2%; I2=97.9%, 56 studies). Variations in pooled TB prevalence were observed by microbiological test(s) used to diagnose TB, with the combination of NAAT and culture reporting the highest prevalence at 11.6% (95% CI 9.0–14.5%; two studies).
Assessment of publication bias
A visual inspection of funnel plots for nonrandomised studies reporting on TB disease and LTBI illustrated significant asymmetry indicating an under-reporting of studies with low case detection (supplementary figures E1 and E2). Evidence of publication bias was noted for studies reporting TB disease (p<0.001 using the Egger test) but not LTBI prevalence (p<0.2 using the Egger test).
Discussion
This systematic review and meta-analysis found evidence that contact investigation was effective and associated with increased case detection [5], reduced mortality and decreased community prevalence of TB [9]. Contact investigation also resulted in a high yield of cases of co-prevalent and incident TB, as well as a high prevalence of LTBI. The prevalence of TB among contacts was found to be high, particularly in low-income and high TB incidence settings, and among contacts of MDR/XDR-TB index patients, contacts ≥15 years old and contacts infected with HIV.
An accurate estimate of the effectiveness of contact investigation requires both an intervention group and also a comparator group where systematic screening was not performed. Randomised trials provide the most robust evidence of effectiveness, accounting for both measured and unmeasured confounding. Despite identifying a large number of studies in this review, most lacked a valid comparison group where screening was not performed, i.e. nonrandomised studies, which likely overestimate the benefits of contact investigation. Only three randomised controlled studies included a valid comparison group [5, 9, 20]. Differences in the outcome measures precluded the ability to conduct a meta-analysis of the outcomes of these three studies, of which two demonstrated effectiveness. In the first, a study from sub-Saharan Africa, a high-prevalence setting for TB and HIV, the community prevalence of TB was reduced by 18% owing to the intervention [9]. The second study from Vietnam, a high-incidence setting with a low prevalence of HIV [5], found that notification of TB and microbiologically confirmed TB increased, while all-cause mortality decreased. This study provides high certainty that contact investigation increases case detection. Together, these randomised trials provide strong evidence for the effectiveness of contact investigation in high-prevalence settings.
The prevalence of TB among contacts was shown to be high. The overall pooled prevalence of any TB in contacts included in this review was 3.6%, slightly higher than the 3.1% estimate in a previous meta-analysis of studies published between 1946 and 2011 [24]. Our findings may reflect scaling up of more sensitive diagnostic tests, particularly GeneXpert MTB/RIF, over the past decade [25]. Stratified estimates demonstrated a high pooled prevalence among certain groups of contacts, including contacts of MDR/XDR-TB index patients, in keeping with the prolonged duration of infectiousness of these patients [26]. Although this finding suggests that contacts of MDR/XDR-TB index patients may be at higher risk for developing TB, few studies have compared transmission of drug-susceptible versus drug-resistant TB [27]. While drug-resistant TB accounts for only 5% of the total TB burden globally, its high morbidity and mortality coupled with high costs associated with treatment make it necessary to investigate contacts exposed to drug-resistant index patients. The results from ongoing TPT trials [28, 29] will also provide important insights into selection of regimens and other implementation issues for this group of contacts.
The yield of contact investigation varied considerably between settings. We found the prevalence of TB detected among contacts was highest in high-incidence and low-income settings, consistent with high levels of community transmission [30]. Studies conducted in high-incidence settings had the highest pooled TB prevalence compared with low-incidence settings (5.0% and 1.9%, respectively). There are several possible explanations for this finding, including earlier TB screening and treatment of LTBI among contacts to prevent disease in low-incidence settings [31] and the lower risk of transmission to contacts from sources outside of the household [31]. This highlights the value of implementing strategies such as contact investigation in high-incidence settings. The higher TB prevalence among contacts in high-incidence settings suggests that transmission outside of the household plays an important role in household infection [30]. This indicates the need to expand case finding beyond known household contacts in order to reduce transmission in these settings. We also note a trend in increased case detection from 2011 to 2016 reported in studies; however, a reduction in prevalence was found for the period 2017–2019. These estimates may represent the scale-up of NAATs and the subsequent global decline in TB incidence, respectively.
Individuals who are HIV-infected are also known to be at high risk for developing TB [32]. We found the pooled prevalence of TB in this population was 11.6%, the highest among any subgroup. This finding confirms and supports the current recommendations for early TB screening and preventive therapy in this population [33]. Regardless of exposure to a known TB patient, HIV-infected individuals should always be targeted for ongoing screening, particularly in high TB incidence settings where the likelihood of infection remains high.
The use of highly sensitive screening and diagnostic algorithms provides important predictors of increased yield of contact investigation. In our review, studies that combined symptom screening and chest radiography for identifying presumptive TB patients had a higher pooled prevalence of TB detected than those which used symptoms only (4.4% versus 1.7%). Smear microscopy was used to confirm a diagnosis of TB in most studies. Rapid NAATs such as GeneXpert MTB/RIF were used in just 16% of studies, reflecting limited dissemination of this more sensitive, but also more costly, technology [34]. Importantly, the pooled prevalence of microbiologically confirmed TB among contacts of 3.2% in our study was substantially higher than reviews of studies before 2011 [24, 35], suggesting a potential benefit from using molecular rapid diagnostics for contact screening. We found that pooled prevalence of TB increased considerably when GeneXpert MTB/RIF was combined with mycobacterial culture. This suggests that if a more sensitive confirmatory test(s) is used, the yield and impact of contact investigation might be further improved. New-generation NAATs with increased sensitivity [36] may further strengthen active case finding. Finding contacts with paucibacillary TB is an expected outcome from contact investigation. Therefore, the selection of sensitive test algorithms must ensure that results are available rapidly while maintaining detection capabilities.
Our finding that TB incidence remained high throughout the first 5 years following exposure indicates that TPT and/or serial screening are both important for the management of contacts. Repeat screening has the advantage of detecting missed prevalent cases [5, 37]; however, it may not be necessary if effective TPT is given. We also found a prevalence of LTBI among contacts of 42.4%, slightly lower than previous reviews [24, 35]. It is possible that these differences reflect temporal changes in TB incidence over the past 20 years or the increased use of the IGRA, which is less likely to cause false positives compared with the TST. The expansion of screening to include contacts at a lower risk of infection, such as contacts of patients with culture-positive and smear-negative TB, may also explain the observed fall in prevalence of LTBI in low- and middle-income settings since 2011. In contrast, we found LTBI prevalence among contacts remained similar in high-income settings. Nevertheless, these findings highlight the high risk of LTBI among contacts, and the importance of TPT for contacts, to prevent incident TB. This supports recommendations to combine scale-up of contact investigation combined with TPT in high-incidence settings [38]. The included studies also demonstrate the feasibility of the TST and IGRA in high- and low-resource settings [39]. Efforts to increase access to tuberculin, coupled with health systems strengthening and staff training in the management of LTBI, can help to support the widespread adoption of TPT in resource-limited settings [40].
This systematic review and meta-analysis has several limitations. First, high levels of heterogeneity were observed among nonrandomised studies, as reflected by the I2 statistic. This is consistent with previous meta-analyses of contact investigation [24, 35]. Our study does characterise several factors which may explain inter-study variability, including variations in contact and index patient characteristics, screening algorithms employed, and differences in the epidemiology of TB between settings. However, the substantial heterogeneity precluded further subgroup analyses. Second, there is known attrition prior to completing screening of contacts during contact investigation, likely resulting in an underestimation of TB prevalence. In addition, contacts with symptoms are usually the only contacts evaluated, thus true prevalence is underestimated in most studies where this occurs. Third, the classification of TB among contacts as co-prevalent or incident depends upon whether TB is detected at the time of initial contact screening. A higher proportion of cases will be co-prevalent if delays in contact investigation occur, since contacts with LTBI may develop active disease prior to screening. Hence, contacts with co-prevalent or incident TB may be misclassified as a result of local differences in the performance of screening algorithms.
Our study has important policy implications. First, we have identified evidence for the effectiveness of contact investigation. Randomised trials published in the last decade provide greater confidence in the benefit of contact investigation for contacts. Second, despite limitations in using nonrandomised studies, we identified specific settings and specific subgroups at an increased risk of TB. Third, our comprehensive search strategy identified a large number of recent studies to evaluate LTBI among contacts. This provides a lower, and likely more robust, estimate compared with a recent review including 27 studies [41]. Our study also provides further support for why contacts of active TB patients should be prioritised for TPT, as part of wider TB control efforts.
In light of the existing data from randomised trials showing a benefit for contact investigation, it is unlikely to be ethical to conduct future randomised trials of contact investigation that include no screening as a comparator. Further operational research is now required to optimise the delivery of contact investigation in different settings (e.g. ensuring maximal enrolment and follow-up of contacts). High-quality studies comparing different screening and confirmatory testing modalities on the detection of TB among contacts are lacking. Priorities for future research include the evaluation of interventions to improve the accuracy of chest radiography interpretation [42], use of molecular diagnostics (including whole-genome sequencing) for the timely detection of drug-resistant TB among high-risk contacts, and evaluation of interventions that expand access to testing and treatment for LTBI.
Conclusions
Recent randomised controlled trials have shown that contact investigation is an effective intervention for increasing TB case detection among contacts, reducing mortality and decreasing community prevalence of TB. The pooled prevalence estimate of microbiologically confirmed TB from our study is higher than previous reviews [24, 35], suggesting that newer rapid molecular diagnostics have improved case detection. The scale-up of contact investigation promises to increase early diagnosis and prevention, contributing to global TB elimination efforts.
Supplementary material
Supplementary Material
Please note: supplementary material is not edited by the Editorial Office, and is uploaded as it has been supplied by the author.
Supplementary material ERJ-00266-2021.Supplement
Shareable PDF
Supplementary Material
This one-page PDF can be shared freely online.
Shareable PDF ERJ-00266-2021.Shareable
Acknowledgements
We would like to acknowledge Mariana Velleca (University of California San Francisco, San Francisco, CA, USA) for assistance on aspects of the initial abstract and title screening, and Cecily Miller (Global TB Programme, World Health Organization, Geneva, Switzerland) for contributions to the development of the research questions and the protocol.
Footnotes
This article has supplementary material available from erj.ersjournals.com
Author contributions: K. Velen had full access to all the data collected in this study and takes full responsibility for the integrity of the data and accuracy of the analysis. R.V. Shingde, J. Ho and G.J. Fox are all equal contributors to the manuscript. Concept and design: K. Velen and G.J. Fox. Acquisition, analysis or interpretation of data: all authors. Drafting of the manuscript: K. Velen and G.J. Fox. Critical revision of the manuscript: all authors. Statistical analysis: K. Velen. Obtained funding: G.J. Fox. Study supervision: K. Velen and G.J. Fox.
Conflict of interest: K. Velen has nothing to disclose.
Conflict of interest: R.V. Shingde has nothing to disclose.
Conflict of interest: J. Ho has nothing to disclose.
Conflict of interest: G.J. Fox has nothing to disclose.
Support statement: The study was supported by funding provided by the World Health Organization (WHO). G.J. Fox was supported by a National Health and Medical Research Council (NHMRC) Career Development Fellowship (APP 1148372). The WHO secretariat developed the research question and provided input into the protocol. The funder had no role in data collection, data analysis, data interpretation or writing of the manuscript. Funding information for this article has been deposited with the Crossref Funder Registry.
- Received January 28, 2021.
- Accepted April 29, 2021.
- Copyright ©The authors 2021. For reproduction rights and permissions contact permissions{at}ersnet.org