Statistical Method to Detect Tuberculosis Outbreaks among Endemic Clusters in a Low-Incidence Setting

Identification of growing tuberculosis incidence clusters in a region with a decrease in tuberculosis prevalence in Moscow (2000-2019)

Background

The control of tuberculosis (TB) may benefit from a prospective identification of areas where the incidence may increase in addition to the traditionally identified foci of high incidence. We aimed to identify residential areas with growing tuberculosis incidence rates and assess their significance and stability.

Methods

We analysed the changes in TB incidence rates using case data georeferenced with spatial granularity to apartment buildings in the territory of Moscow from 2000 to 2019. We identified sparsely distributed areas with significant increases in the incidence rate inside residential areas. We tested the stability of found growth areas to case underreporting via stochastic modelling.

Results

For 21 350 cases with smear- or culture-positive pulmonary TB among residents from 2000 to 2019, we identified 52 small-scale clusters of growing incidence rate responsible for 1% of all registered cases. We tested clusters of disease growth for underreporting and found them to be relatively unstable to resampling with case drop-out, but their spatial displacement was small. Territories with a stable increase in TB incidence rate were identified and compared to the rest of the city, which is characterised by a significant decrease in incidence.

Conclusions

Identified areas with a tendency for an increase in the TB incidence rate may be important targets for disease control services.

Epidemiological cluster identification using multiple data sources: an approach using logistic regression

In the management of infectious disease outbreaks, grouping cases into clusters and understanding their underlying epidemiology are fundamental tasks. In genomic epidemiology, clusters are typically identified either using pathogen sequences alone or with sequences in combination with epidemiological data such as location and time of collection. However, it may not be feasible to culture and sequence all pathogen isolates, so sequence data may not be available for all cases. This presents challenges for identifying clusters and understanding epidemiology, because these cases may be important for transmission. Demographic, clinical and location data are likely to be available for unsequenced cases, and comprise partial information about their clustering. Here, we use statistical modelling to assign unsequenced cases to clusters already identified by genomic methods, assuming that a more direct method of linking individuals, such as contact tracing, is not available. We build our model on pairwise similarity between cases to predict whether cases cluster together, in contrast to using individual case data to predict the cases' clusters. We then develop methods that allow us to determine whether a pair of unsequenced cases are likely to cluster together, to group them into their most probable clusters, to identify which are most likely to be members of a specific (known) cluster, and to estimate the true size of a known cluster given a set of unsequenced cases. We apply our method to tuberculosis data from Valencia, Spain. Among other applications, we find that clustering can be predicted successfully using spatial distance between cases and whether nationality is the same. We can identify the correct cluster for an unsequenced case, among 38 possible clusters, with an accuracy of approximately 35 %, higher than both direct multinomial regression (17 %) and random selection (< 5 %).

Quantifying Mycobacterium tuberculosis Transmission Dynamics Across Global Settings: A Systematic Analysis

The degree to which individual heterogeneity in the production of secondary cases ("superspreading") affects tuberculosis (TB) transmission has not been systematically studied. We searched for population-based or surveillance studies in which whole genome sequencing was used to estimate TB transmission and in which the size distributions of putative TB transmission clusters were enumerated. We fitted cluster-size-distribution data to a negative binomial branching process model to jointly infer the transmission parameters $R$ (the reproduction number) and the dispersion parameter, $k$, which quantifies the propensity of superspreading in a population (generally, lower values of $k$ ($<1.0$) suggest increased heterogeneity). Of 4,796 citations identified in our initial search, 9 studies from 8 global settings met the inclusion criteria (n = 5 studies of all TB; n = 4 studies of drug-resistant TB). Estimated $R$ values (range, 0.10-0.73) were below 1.0, consistent with declining epidemics in the included settings; estimated $k$ values were well below 1.0 (range, 0.02-0.48), indicating the presence of substantial individual-level heterogeneity in transmission across all settings. We estimated that a minority of cases (range, 2%-31%) drive the majority (80%) of ongoing TB transmission at the population level. Identifying sources of heterogeneity and accounting for them in TB control may have a considerable impact on mitigating TB transmission.

Using Machine Learning Techniques and National Tuberculosis Surveillance Data to Predict Excess Growth in Genotyped Tuberculosis Clusters

The early identification of clusters of persons with tuberculosis (TB) that will grow to become outbreaks creates an opportunity for intervention in preventing future TB cases. We used surveillance data (2009-2018) from the United States, statistically derived definitions of unexpected growth, and machine-learning techniques to predict which clusters of genotype-matched TB cases are most likely to continue accumulating cases above expected growth within a 1-year follow-up period. We developed a model to predict which clusters are likely to grow on a training and testing data set that was generalizable to a validation data set. Our model showed that characteristics of clusters were more important than the social, demographic, and clinical characteristics of the patients in those clusters. For instance, the time between cases before unexpected growth was identified as the most important of our predictors. A faster accumulation of cases increased the probability of excess growth being predicted during the follow-up period. We have demonstrated that combining the characteristics of clusters and cases with machine learning can add to existing tools to help prioritize which clusters may benefit most from public health interventions. For example, consideration of an entire cluster, not only an individual patient, may assist in interrupting ongoing transmission.

Molecular surveillance for large outbreaks of tuberculosis in the United States, 2014-2018

OBJECTIVE

This study describes characteristics of large tuberculosis (TB) outbreaks in the United States detected using novel molecular surveillance methods during 2014-2016 and followed for 2 years through 2018.

METHODS

We developed 4 genotype-based detection algorithms to identify large TB outbreaks of ≥10 cases related by recent transmission during a 3-year period. We used whole-genome sequencing and epidemiologic data to assess evidence of recent transmission among cases.

RESULTS

There were 24 large outbreaks involving 518 cases; patients were primarily U.S.-born (85.1%) racial/ethnic minorities (84.1%). Compared with all other TB patients, patients associated with large outbreaks were more likely to report substance use, homelessness, and having been diagnosed while incarcerated. Most large outbreaks primarily occurred within residences among families and nonfamilial social contacts. A source case with a prolonged infectious period and difficulties in eliciting contacts were commonly reported contributors to transmission.

CONCLUSION

Large outbreak surveillance can inform targeted interventions to decrease outbreak-associated TB morbidity.

Logically Inferred Tuberculosis Transmission (LITT): A Data Integration Algorithm to Rank Potential Source Cases

Understanding tuberculosis (TB) transmission chains can help public health staff target their resources to prevent further transmission, but currently there are few tools to automate this process. We have developed the Logically Inferred Tuberculosis Transmission (LITT) algorithm to systematize the integration and analysis of whole-genome sequencing, clinical, and epidemiological data. Based on the work typically performed by hand during a cluster investigation, LITT identifies and ranks potential source cases for each case in a TB cluster. We evaluated LITT using a diverse dataset of 534 cases in 56 clusters (size range: 2–69 cases), which were investigated locally in three different U.S. jurisdictions. Investigators and LITT agreed on the most likely source case for 145 (80%) of 181 cases. By reviewing discrepancies, we found that many of the remaining differences resulted from errors in the dataset used for the LITT algorithm. In addition, we developed a graphical user interface, user's manual, and training resources to improve LITT accessibility for frontline staff. While LITT cannot replace thorough field investigation, the algorithm can help investigators systematically analyze and interpret complex data over the course of a TB cluster investigation.

Code available at:https://github.com/CDCgov/TB_molecular_epidemiology/tree/1.0; https://zenodo.org/badge/latestdoi/166261171.

Developing National Genotype-Independent Indicators for Recent Mycobacterium Tuberculosis Transmission Using Pediatric Cases-United States, 2011-2017

INTRODUCTION

Pediatric tuberculosis (TB) cases are sentinel events for Mycobacterium tuberculosis transmission in communities because children, by definition, must have been infected relatively recently. However, these events are not consistently identified by genotype-dependent surveillance alerting methods because many pediatric TB cases are not culture-positive, a prerequisite for genotyping.

METHODS

We developed 3 potential indicators of ongoing TB transmission based on identifying counties in the United States with relatively high pediatric (aged <15 years) TB incidence: (1) a case proportion indicator: an above-average proportion of pediatric TB cases among all TB cases; (2) a case rate indicator: an above-average pediatric TB case rate; and (3) a statistical model indicator: a statistical model based on a significant increase in pediatric TB cases from the previous 8-quarter moving average.

RESULTS

Of the 249 US counties reporting ≥2 pediatric TB cases during 2009-2017, 240 and 249 counties were identified by the case proportion and case rate indicators, respectively. The statistical model indicator identified 40 counties with a significant increase in the number of pediatric TB cases. We compared results from the 3 indicators with an independently generated list of 91 likely transmission events involving ≥2 pediatric cases (ie, known TB outbreaks or case clusters with reported epidemiologic links). All counties with likely transmission events involving multiple pediatric cases were identified by ≥1 indicator; 23 were identified by all 3 indicators.

PRACTICE IMPLICATIONS

This retrospective analysis demonstrates the feasibility of using routine TB surveillance data to identify counties where ongoing TB transmission might be occurring, even in the absence of available genotyping data.

Pathogen Genomics in Public Health

Rapid advances in DNA sequencing technology ("next-generation sequencing") have inspired optimism about the potential of human genomics for "precision medicine." Meanwhile, pathogen genomics is already delivering "precision public health" through more effective investigations of outbreaks of foodborne illnesses, better-targeted tuberculosis control, and more timely and granular influenza surveillance to inform the selection of vaccine strains. In this article, we describe how public health agencies have been adopting pathogen genomics to improve their effectiveness in almost all domains of infectious disease. This momentum is likely to continue, given the ongoing development in sequencing and sequencing-related technologies.

Understanding Tuberculosis Transmission in the United Kingdom: Findings From 6 Years of Mycobacterial Interspersed Repetitive Unit-Variable Number Tandem Repeats Strain Typing, 2010-2015

Genotyping provides the opportunity to better understand tuberculosis (TB) transmission. We utilized strain typing data to assess trends in the proportion of clustering and identify the characteristics of individuals and clusters associated with recent United Kingdom (UK) transmission. In this retrospective cohort analysis, we included all culture-confirmed strain-typed TB notifications from the UK between 2010 and 2015 to estimate the proportion of patients that clustered over time. We explored the characteristics of patients in a cluster using multivariable logistic regression. Overall, 58.5% of TB patients were concentrated in 2,701 clusters. The proportion of patients in a cluster decreased between 2010 (58.7%) and 2015 (55.3%) (P = 0.001). Being a clustered patient was associated with being male and UK-born, having pulmonary disease, having a previous TB diagnosis, and having a history of drug misuse or imprisonment. Our results suggest that TB transmission in the UK decreased between 2010 and 2015, during which time TB incidence also decreased. Targeted cluster investigation and extended contact tracing should be aimed at persons at risk of being in a transmission chain, including UK-born individuals with social risk factors in clusters with a high proportion of patients having pulmonary disease.