Transmission characteristics of MERS and SARS in the healthcare setting: a comparative study

Abstract

Background: The Middle East respiratory syndrome (MERS) coronavirus has caused recurrent outbreaks in the Arabian Peninsula since 2012. Although MERS has low overall human-to-human transmission potential, there is occasional amplification in the healthcare setting, a pattern reminiscent of the dynamics of the severe acute respiratory syndrome (SARS) outbreaks in 2003. Here we provide a head-to-head comparison of exposure patterns and transmission dynamics of large hospital clusters of MERS and SARS, including the most recent South Korean outbreak of MERS in 2015.

Methods: To assess the unexpected nature of the recent South Korean nosocomial outbreak of MERS and estimate the probability of future large hospital clusters, we compared exposure and transmission patterns for previously reported hospital clusters of MERS and SARS, based on individual-level data and transmission tree information. We carried out simulations of nosocomial outbreaks of MERS and SARS using branching process models rooted in transmission tree data, and inferred the probability and characteristics of large outbreaks.

Results: A significant fraction of MERS cases were linked to the healthcare setting, ranging from 43.5 % for the nosocomial outbreak in Jeddah, Saudi Arabia, in 2014 to 100 % for both the outbreak in Al-Hasa, Saudi Arabia, in 2013 and the outbreak in South Korea in 2015. Both MERS and SARS nosocomial outbreaks are characterized by early nosocomial super-spreading events, with the reproduction number dropping below 1 within three to five disease generations. There was a systematic difference in the exposure patterns of MERS and SARS: a majority of MERS cases occurred among patients who sought care in the same facilities as the index case, whereas there was a greater concentration of SARS cases among healthcare workers throughout the outbreak. Exposure patterns differed slightly by disease generation, however, especially for SARS. Moreover, the distributions of secondary cases per single primary case varied highly across individual hospital outbreaks (Kruskal-Wallis test; P < 0.0001), with significantly higher transmission heterogeneity in the distribution of secondary cases for MERS than SARS. Simulations indicate a 2-fold higher probability of occurrence of large outbreaks (>100 cases) for SARS than MERS (2 % versus 1 %); however, owing to higher transmission heterogeneity, the largest outbreaks of MERS are characterized by sharper incidence peaks. The probability of occurrence of MERS outbreaks larger than the South Korean cluster (n = 186) is of the order of 1 %.

Conclusions: Our study suggests that the South Korean outbreak followed a similar progression to previously described hospital clusters involving coronaviruses, with early super-spreading events generating a disproportionately large number of secondary infections, and the transmission potential diminishing greatly in subsequent generations. Differences in relative exposure patterns and transmission heterogeneity of MERS and SARS could point to changes in hospital practices since 2003 or differences in transmission mechanisms of these coronaviruses.

Figures

Fig. 1

Transmission trees of Middle East respiratory syndrome (MERS) and severe acute respiratory syndrome (SARS) outbreaks linked to health-care settings. a MERS outbreak in Al-Hasa, Saudi Arabia, from 1 April to 23 May 2013 [11]. b MERS outbreak in South Korea from 20 May to 5 July 2015 [13, 33]. c Nosocomial SARS outbreak in Singapore from 25 February to 11 May 2003 [16]. d Nosocomial SARS outbreak in Toronto from 23 February to 5 April 2003 [17]. Numbers inside the nodes of the tree are used to indicate a group of cases rather than a single case. Colors are used to distinguish the index case from secondary cases and highlight different exposure categories among secondary cases, including patient, visitor or family member, healthcare worker, and non-clinical staff working in the hospital

Fig. 2

Total number of cases per disease generation and categorized according to exposure category (health-care worker, patient, visitor or family member, and non-clinical healthcare staff) for large nosocomial outbreaks of Middle East respiratory syndrome (MERS) and severe acute respiratory syndrome (SARS). Incidence starts at generation 0, which refers to the index case

Fig. 3

Exposure patterns among secondary cases (health-care worker, patients, visitor or family member, and non-clinical staff) throughout the entire outbreak (top panel) and for the first disease generation (bottom panel) for each of the Middle East respiratory syndrome (MERS) and severe acute respiratory syndrome (SARS) outbreaks linked to health-care settings. Bottom panel shows the total index reproduction number (R0) and a breakdown by exposure category of secondary case. The horizontal dashed line in the bottom panel at R = 1.0 is shown for reference

Fig. 4

The reproduction number according to disease generations, Rg , starting with the index reproduction number (R0) at generation 0 and according to exposure category (healthcare worker, patients, visitors or family member, and non-clinical staff) for Middle East respiratory syndrome (MERS) and severe acute respiratory syndrome (SARS) outbreaks linked to healthcare settings. Rg denotes the mean number of secondary cases stemming from cases in generation g. The horizontal dashed line at R = 1.0 is shown for reference

Fig. 5

Outbreak simulations for (top) Middle Eastern respiratory syndrome (MERS) and (bottom) severe acute respiratory syndrome (SARS) through branching process models using parameters shown in Table 3. Curves are derived from 5,000 outbreak realizations. The distributions of outbreak size, duration, and peak size are well characterized by power-law-like distributions. The cumulative and generation-based incidence curves for the five largest outbreaks are highlighted in red; while smaller outbreaks are displayed in cyan. Right panels illustrate the expected frequency of occurrence of an outbreak of a given size

Fig. 6

Scatter plot of expected outbreak size and duration for 5,000 simulated outbreaks derived via branching process models of Middle Eastern respiratory syndrome (MERS) and severe acute respiratory distress (SARS) using parameters shown in Table 3. Data points for observed MERS and SARS outbreaks are shown in color for reference