An open-access database of infectious disease transmission trees to explore superspreader epidemiology

Abstract

Historically, emerging and reemerging infectious diseases have caused large, deadly, and expensive multinational outbreaks. Often outbreak investigations aim to identify who infected whom by reconstructing the outbreak transmission tree, which visualizes transmission between individuals as a network with nodes representing individuals and branches representing transmission from person to person. We compiled a database, called OutbreakTrees, of 382 published, standardized transmission trees consisting of 16 directly transmitted diseases ranging in size from 2 to 286 cases. For each tree and disease, we calculated several key statistics, such as tree size, average number of secondary infections, the dispersion parameter, and the proportion of cases considered superspreaders, and examined how these statistics varied over the course of each outbreak and under different assumptions about the completeness of outbreak investigations. We demonstrated the potential utility of the database through 2 short analyses addressing questions about superspreader epidemiology for a variety of diseases, including Coronavirus Disease 2019 (COVID-19). First, we found that our transmission trees were consistent with theory predicting that intermediate dispersion parameters give rise to the highest proportion of cases causing superspreading events. Additionally, we investigated patterns in how superspreaders are infected. Across trees with more than 1 superspreader, we found preliminary support for the theory that superspreaders generate other superspreaders. In sum, our findings put the role of superspreading in COVID-19 transmission in perspective with that of other diseases and suggest an approach to further research regarding the generation of superspreaders. These data have been made openly available to encourage reuse and further scientific inquiry.

Fig 1. We compiled infectious disease transmission trees from the literature along with reported attribute information.

Shown here are example trees in the database. (A) Ebola spread in different contexts [8]. (B) Measles spread in different locations [9]. (C) COVID-19 spread among age classes [10]. Primary sources for transmission trees are available in OutbreakTrees and listed in the Supporting information. OutbreakTrees may be accessed online at http://outbreaktrees.ecology.uga.edu. COVID-19, Coronavirus Disease 2019.

Fig 2. Characteristics of transmission trees in OutbreakTrees.

(A) Tree size varies from 2 to 286 with a median of 3 and most trees represent outbreaks taking place in the past 20 years (only trees with 10 or more cases shown in date plot due to large number of small COVID-19 trees from 2020). (B) The largest trees are from H1N1 and SARS outbreaks, while the highest proportion of trees in the database are from outbreaks of COVID-19, followed by adenovirus and Ebola. Tree size axes in both plots are shown on a log₁₀ scale to better illustrate variation in medium-sized trees. All trees are used in this analysis. The data to reproduce this figure can be found at https://doi.org/10.5061/dryad.nk98sf7w7. COVID-19, Coronavirus Disease 2019; MERS, Middle East Respiratory Syndrome; SARS, Severe Acute Respiratory Syndrome.

Fig 3. The time dependence of R, k, and the proportion of cases causing superspreading events.

(A) R decreased significantly between the first and second halves of transmission trees. (B) k did not differ significantly between the first and second halves of transmission trees. Y-axis is on a log₁₀ scale for visual aid. (C) The proportion of cases causing superspreading events decreased significantly between the first and second halves of transmission trees. (D) Decrease in R shown for each tree by disease. R was below 1 in the second half of all trees; red line denotes R = 1. The Wilcoxon rank test was used for all significance tests (*: p≤0.05, **: p≤0.01, ***: p≤0.001, ****: p≤0.0001), and results are shown in red stars. Trees were assumed to be complete and only trees with 20 or more cases and at least 2 generations of spread were used in these analyses. Results assuming tree incompleteness are shown in S3 Fig. The data to reproduce this figure can be found at https://doi.org/10.5061/dryad.nk98sf7w7. COVID-19, Coronavirus Disease 2019; MERS, Middle East Respiratory Syndrome; SARS, Severe Acute Respiratory Syndrome.

Fig 4. The importance and expected frequency of superspreading across diseases.

(A) The highest proportion of cases causing superspreading events is observed at intermediate dispersion parameters, as predicted by theory [3]. (B) Dispersion parameter (k) of a negative binomial distribution fit to the offspring distribution of trees by disease (for diseases with at least 3 trees). Lower dispersion parameters are indicative of greater variation in number of secondary infections. Vertical line and value printed in each facet shows the median k and standard error for each disease. X-axes are on a log₁₀ scale in both plots for visual aid. Trees were assumed to be complete and only trees with 20 or more cases and at least 2 generations of spread were used in these analyses. Other size cutoffs are shown in S4 and S5 Figs and results assuming tree incompleteness are shown in S6 Fig. The data to reproduce this figure can be found at https://doi.org/10.5061/dryad.nk98sf7w7. COVID-19, Coronavirus Disease 2019; MERS, Middle East Respiratory Syndrome; SARS, Severe Acute Respiratory Syndrome.

Fig 5. In two-thirds of transmission trees, superspreaders infect superspreaders more often than would be expected by chance.

The expected number of superspreader-superspreader dyads was calculated by $\frac{s(s-1)}{S-t}$ for each tree, where s is the number of superspreaders in the tree, t is the number of terminal nodes (nodes that do not cause onward transmission), and S is tree size. Ratios larger than 1 indicate more superspreader-superspreader dyads were observed than would be expected by chance. This analysis was limited to trees with more than 1 superspreader, 20 or more cases, and 2 or more generations of spread. We assumed tree completeness here, but results assuming incompleteness are shown in S7 Fig. The data to reproduce this figure can be found at https://doi.org/10.5061/dryad.nk98sf7w7. COVID-19, Coronavirus Disease 2019; MERS, Middle East Respiratory Syndrome; SARS, Severe Acute Respiratory Syndrome.