Frequency of transmission, asymptomatic shedding, and airborne spread of Streptococcus pyogenes among schoolchildren exposed to scarlet fever: a longitudinal multi-cohort molecular epidemiology contact tracing study

Introduction

Since 2014, England has experienced rates of scarlet fever that are unprecedented in modern times (1, 2). Scarlet fever is a highly communicable exanthem caused by Streptococcus pyogenes that predominantly affects younger children. Over 30,000 cases were notified in 2018, the highest number since 1960, with an age-specific incidence of 523 per 100,000 among 1-4 year olds in England and Wales (3). Infections have long-been recognised to seed outbreaks in schools and nurseries, creating a significant public health burden (2). Though benign if treated promptly, there is a 20-fold increased risk of invasive streptococcal infection in households that include a recent case of scarlet fever (4). In 2016, a near 2-fold increase in invasive disease notifications was accompanied by emergence of a new sublineage of emm1 S. pyogenes (M1_UK) expressing increased amounts of scarlet fever toxin (5).

The literature highlights a role for fomite surfaces in propagation of S. pyogenes outbreaks (6,7). Guidance for management of school/nursery outbreaks of scarlet fever emphasises the importance of maintaining good hand hygiene, prompt antibiotic treatment of index cases, and exclusion from school until 24h after antibiotic treatment has started (8). Escalation of interventions including daily cleaning of toys, utensils, and frequent touch points is recommended if necessary, along with daily vacuuming of soft furnishings (8).

To understand why outbreaks might continue to occur despite following these interventions, we investigated S. pyogenes transmission during the scarlet fever seasons of 2018 and 2019, focussing on educational settings with two sequential notifications of scarlet fever. Our primary aim was to determine if existing guidance that focuses on hygiene, alongside case exclusion, prevented transmission of S. pyogenes in the classroom. Our secondary aim was to investigate modes of transmission, to inform future guidance and provide insight into S. pyogenes outbreaks more generally.

Methods

Results

During the study, 156 educational outbreaks in London were reported in year 1 (2018), and 47 in year 2 (2019). Six settings were included, and 12 cases were recruited (Figure 1, Appendix, p3). Two cases were recruited per setting in settings 1-3; setting 2 and 3 were different classes in the same school. Three cases were recruited from each of settings 4 and 6. Cases declined to participate in setting 5, although contact and environment testing was undertaken based on one confirmed and one probable case. Sampling was completed by June 7^th in year 1 and June 28^th in year 2. During the study, participating schools reported increased prevalence of pharyngitis in addition to the reported cases of scarlet fever. The median attack rate for confirmed and probable cases was 5% (IQR 4.1-12.6%) and 10.1% (IQR 7.3-24.9%) with inclusion of possible cases. (Appendix, p3)

• Download figure
• Open in new tab

Figure 1

Flowchart indicating sample types sought from participants and schools in each setting. Samples provided, timings, and any breaks in sampling schedule are listed in Appendix pages 4-6.

Of the scarlet fever cases, 6/12 had been swabbed by primary care physicians prior to antibiotic treatment. S. pyogenes genotypes associated with each outbreak were derived from the swabs obtained from these cases, or inferred from the S. pyogenes isolates cultured from household contacts of cases. Genotypes were emm6, emm1, emm4, and emm3.93 (Figure 2). Emm1 isolates all belonged to the newly described M1_UK lineage (5), while emm4 isolates all belonged to the M4-complete lineage (10).

• Download figure
• Open in new tab

Figure 2. Phylogenetic relationship between S. pyogenes isolates sequenced from cases, contacts and environment in each of six settings.

Maximum likelihood phylogenetic tree constructed from 20,229 core SNPs (without recombination regions) extracted after mapping 136 S. pyogenes isolates to the reference sequence H293 (emm89, HG316453.2). The outer rings (from outermost to innermost) represent the settings (S1-S6), and the year of collection. The tree was drawn in a circular layout with tips coloured based on emm types. The shape of the tip indicates the source (sample type) of individual isolates: Case; Classroom contact (CC); household contact (HHC); and environmental samples from settle plate (Air); or fomite (Toy). In some cases and contacts, multiple isolates were detected per participant (Appendix pages 4-6 and 9-13). Uncommon strain subtypes (emm3.143 and emm6.9) within an emm-type are indicated by *. For detailed analysis of each setting, see supplementary figures.

All bar one of the 12 cases received antibiotic treatment prescribed by the primary care physician. S. pyogenes was not detected in swab samples obtained during the first week, in all cases where antibiotic treatment had been initiated prior to the study visit (Table 1). The child who did not receive antibiotic treatment remained throat swab positive up to and including week 2 but had cleared emm4 S. pyogenes by week 3 (Table 1, appendix p4). One contact who had a positive throat swab and cough plate on day 1 of week 1 with emm3.93 S. pyogenes, developed scarlet fever the same day, so was also recruited as a case. Where the duration of antibiotic treatment was known (8/12 cases), 4 children completed a 10-day course, and 4 took an incomplete course (including 2 cases where a 7-day course was prescribed). Of those who had received antibiotic treatment, including those who completed a full 10-day course, the throat swab grew the prevailing outbreak strain of S. pyogenes again in 4/11 cases (36.3%); by week 2 (one with emm3.93), and by week 3 (two with M1_UK and one with emm3.93). One of the cases, carrying M1_UK, was also cough-plate and hand-swab positive in week 3 despite previously testing negative. (Table 1 and Appendix p4).

Seventeen household contacts were enrolled in the study, representing 9 households (Figure 1). Of these, 3/17 (17.6%) had throat swabs that yielded S. pyogenes, two of whom had symptomatic pharyngitis with M1_UK and both demonstrated carriage spanning 2 or 3 weeks. (Appendix p5)

Among all classroom contacts swabbed in schools (Figure 1), asymptomatic throat carriage of the genomically-defined outbreak S. pyogenes strains among swabbed children rose from 11/115 (9.6%) in week 1, to 34/126 (26.9%) in week 2, to 26/108 (24.1%) in week 3, then 5/35 (14.3%) in week 4. This fell to 6/53 (11.3%), then 0/18 (0%) in classes where delayed swabbing was undertaken (Table 2). In contrast to carriage of the outbreak strain, median carriage of non-outbreak S. pyogenes was 2.8% (IQR 0-6.61%) during the study (Table 2). Transmission appeared particularly intense in setting 2, where 8/18 (44.4%) of children swabbed were infected with the outbreak M1_UK strain in week 2, in addition to the two household contacts (Figure 3, appendix p6). Overall, across all six settings, more than half of eligible children participated in the study (median participation rate 52.5%, IQR 43.71-65.83%; appendix p6).

• Download figure
• Open in new tab

Figure 3 Acquisition of M1_UK and emm3.93 S. pyogenes by classroom contacts.

Classroom contacts in (A) setting 2 and (B) Setting 6. Classroom contacts had throat swabs weekly for at least three weeks after identification of case; each icon represents an individual child, grouped by week of swabbing. Colour of icons (adioma.com) indicates throat swab result: Blue, negative; dark pink, outbreak strain S. pyogenes; pale pink, non-outbreak strain or genotype not confirmed; grey, participant not swabbed in that week. Pink glow behind icon indicates cough plate and/or hand swab positive. For (A) (n=22 classroom contacts) S. pyogenes was identified in 10/18 (55.5%) and 8/16 (50.0%) in weeks 2 and 3 respectively. The outbreak strain (emm1/M1_UK) was identified in 8/18 (44.4%) and 6/16 (37.5%) in weeks 2 and 3. For (B) (n=34 classroom contacts) S. pyogenes was identified in 7/28 (25.0%), 12/34 (35.2%), and 9/30 (30.0%) in weeks 1, 2, and 3 respectively. The outbreak strain emm3.93 was identified in 6/28 (21.4%), 9/34 (26.5%), and 6/30 (20.0%) of children who had swabs taken in weeks 1, 2, and 3 respectively. Fourth week not shown.

Whole genome sequencing was undertaken on all viable S. pyogenes isolates identified. Within each setting, isolates of the same genotype were clonal, with individual strains being a median of 0 SNPs (range 0-5 SNPs) different to other strains in that setting, consistent with a common source of transmission, but a median of 12 SNPs (range 4-55 SNPs) different from the closest strain of the same genotype sequenced in 2014. Core genomes exhibited more similarity than expected. (Appendix p8, p15-18).

Environmental swabs taken in settings 1-3 in week 1 of the first year of the study yielded mixed bacterial growth (∼1 × 10² cfu – 1 × 10⁴ cfu per swab; appendix p7). S. pyogenes was identified from just one toy (train track) from setting 2: Five colonies/swab of the outbreak strain emm1 (M1_UK) were 1-3 SNPs different to isolates obtained from children in the same class (appendix p16).

To determine the source of ongoing transmission in the classrooms, the protocol was amended in year 2 to include hand-swabs and cough-plates from classroom contacts in addition to throat swabs, while settle plates were used to sample air in the classroom. In setting 4, where emm4 predominated, one of four children carrying S. pyogenes was cough-plate positive in the second week, but not in subsequent weeks (appendix p6); all isolates in this setting were 0 SNP apart (appendix p17). Cough-plates were negative in setting 5 (also emm4 where again isolates were 0 SNP apart; appendix p6, p17). Overall 1/14 (7.1%) emm4 positive throat swabs was associated with a positive cough plate. In setting 6, where emm3.93 S. pyogenes predominated, cough-plates were positive for the outbreak emm3.93 strain in one third of children with positive emm3.93 throat swabs (week 1, 2/6; week 2, 3/9, week 3, 2/6) (Figure 3 and Appendix p6). The vast majority of strains were 0 SNP different from one another, bar three strains that were 1, 2, and 5 SNP apart (Appendix p18). SNPs differentiating strains within each outbreak are listed (Appendix p8).

In setting 4, where prevalence of the outbreak emm4 S. pyogenes had been 0/18(0%), 4/20(20%), and 4/24(16.6%) in weeks 1-3 respectively, final samples from classroom contacts were taken in week 16 after an intentional break in sampling, and all (0/18) were negative. It was only possible to incorporate a 4-week gap for settings 5 and 6 for logistic reasons related to the timing of the school term. In setting 5, carriage of the outbreak emm4 strain remained steady throughout the study, with a prevalence of 0% (0/17), 13.6% (3/22), 13.6% (3/22) in weeks 1-3 respectively and 12.5% (3/23) at week 7. Notably 2/3 of emm4 carriers who were positive in week 7 had also been positive in week 3. In setting 6, where prevalence of the outbreak emm3.93 strain in weeks 1, 2, and 3 was 21.4% (6/28), 26.5% (9/34), 19.4% (6/31) (Figure 3) respectively in asymptomatic classroom contacts, carriage of emm3.93 fell to 10% (3/30) by week 8. Notably in this outbreak, throat swabs from two of the cases and a single household contact were positive in week 3, emm3.93 (Appendix p4-5). To determine the likelihood that culture-based sampling might be insensitive, DNA extracted from culture-negative swabs from setting 6 was subject to S. pyogenes ProS PCR; 50/65 (76.9%) culture negative swabs were deemed qPCR negative.

Four settle plates were used per classroom in settings 5 and 6 to sample room air, sited above child head height. In setting 5, one of four settle plates (placed above a whiteboard) was positive for emm4 S. pyogenes in each of weeks 2 and 3 of the study (Table 3). In setting 6, two of four settle plates were positive for emm3.93 S. pyogenes in weeks 1, 2, and 3 of the study, including plates on top of 2m cupboards (Table 3). In both settings, S. pyogenes strains on settle plates were identical to strains identified in children (emm4 and emm3.93, appendix p17-18).

Discussion

This prospective contact tracing study of S. pyogenes transmission was undertaken in response to an unprecedented rise in scarlet fever notifications in England. Using genome sequencing to confirm common sources of transmission, we found a high prevalence of the outbreak strain among asymptomatic classroom contacts, peaking in the second week of our investigations. Despite antibiotic treatment and isolation of index cases for 24h after initiation of antibiotic treatment, and implementation of standard hygiene measures within the classrooms, transmission within the class was observed. Enhanced sampling in year 2 revealed evidence of prominent S. pyogenes shedding by some children, and airborne dispersal of genomically identical strains in the classroom.

In the six settings investigated, emm1 (M1_UK) strains were involved in two outbreaks, emm4 in two outbreaks, and emm3.93 and emm6 strains were involved in one outbreak each. New scarlet fever-causing lineages have been associated with national upsurges in invasive S. pyogenes (5, 11), exemplified by emergence of M1_UK, a lineage that expresses increased levels of the scarlet fever toxin SpeA (5). There are however international differences in approach to treatment of streptococcal pharyngitis (12), and transmission risks are not widely considered (13). Coupled with the public health impact of outbreaks (2), high attack rates in schools (14), and increased risk of invasive infections (4), there is a rationale to limit the spread of S. pyogenes in the population, particularly where lineages such as emm1 and emm3, that are independently associated with high case fatality or severe manifestations (15), are involved.

Current public health guidance on scarlet fever management focusses on treatment, exclusion, and hygiene interventions, with escalation to include daily cleaning of surfaces (8). Although contamination of fomites including toys is no doubt important, we found that transmission was ongoing despite such cleaning and in the absence of frequent surface S. pyogenes contamination. Indeed, the main source of S. pyogenes appeared to be the children themselves. Other than foodborne scarlet fever outbreaks (16), there are remarkably few recent investigations that examine transmission routes. Four contemporary reports describe scarlet fever outbreaks with high attack rates (23-72%) in educational settings that were not controlled by standard interventions (17-20); transmission routes were investigated in just one (17). Three describe use of antibiotic treatment to end the outbreaks, treating all colonised children (17-19). Microbiological detection of S. pyogenes to enable treatment of children found to be colonised would however be challenging to undertake routinely for all scarlet fever outbreaks. Our study underlines a need for research to evaluate a potential role for molecular point of care tests for rapid detection of colonisation and outbreak management.

By week 3, prevalence of the genomically-confirmed outbreak strain was 14%-47% across all six settings, while carriage of non-outbreak strains was infrequent. Carriage rates detected are remarkably similar to data from higher attack rate outbreaks (17-20) and studies from the pre-antibiotic era, where 26-33% of streptococcal pharyngitis classroom contacts were observed to carry the outbreak serotype (21). Our study draws attention to the importance of context when measuring S. pyogenes carriage rates as, outside of the springtime outbreak season, asymptomatic carriage in healthy children in England is estimated at <6% (22) and <1% in healthy adults (23). Studies that report average annual rates of asymptomatic carriage therefore fail to recognise the very major impact of seasonal variation and outbreaks and may provide misleading contextual information when providing recommendations.

We frequently recovered the outbreak strain among classroom contacts in week 1 of sampling, indicating that transmission had already occurred in the classroom, whether from one of the index cases of scarlet fever or another unknown source. It is possible that treatment may have been delayed in some cases; a recent study identified one fifth of scarlet fever cases to be initially diagnosed as a viral infection, allowing transmission to continue if the affected child remains in school, untreated (24). Whether this is sufficient to explain scarlet fever outbreaks merits further analysis and modelling.

It is generally believed that asymptomatic carriage is unlikely to lead to transmission, although the term ‘carrier’ is more often used to refer to a case that has been treated for S. pyogenes pharyngitis but demonstrates microbiological failure (25). In our study, the systematic increase in prevalence of outbreak strains between weeks 1 and 2 in classroom contacts pointed to ongoing transmission from asymptomatic carriers, despite index cases having been excluded and treated. Indeed, after returning to school, four cases re-acquired the outbreak strain by week 3, presumably through contact with other colonised contacts or classroom air, although did not develop symptomatic illness; these cases re-acquired emm1 (M1_UK, 2 cases) and emm3.93 (2 cases). Our investigations using cough plates and hand swabs demonstrated that at least one third of asymptomatic emm3.93 carriers were shedding more S. pyogenes than other children in exhaled cough and on hands. These findings raise a question about the definition of colonisation and infection in young children, who may not be able to identify their symptoms, or may be infectious without symptoms. The study confirms the existence of so-called heavy shedders among those who are apparently asymptomatic at the time of sampling and resonates with historic reports of heavy nasal shedding by S. pyogenes nasal carriers (26).

Settle plates placed in elevated locations for just 2-3 hours, provided evidence of S. pyogenes dispersal in classroom air in settings 5 and 6; the timing of settle plate positivity coincided with more intense periods of asymptomatic shedding. Settle plates provide an easy read out in an outbreak setting and could be used to indicate a need for improved ventilation, social distancing, or surveillance for heavy shedders. Airborne spread of scarlet fever-causing streptococci was previously recognised as a threat in hospitals in the 1930’s and in military barracks in the 1940’s (28). Contemporary outbreaks of S. pyogenes that are not explained by direct contact alone suggest a need to take not only indirect contact but also airborne transmission into consideration when developing guidelines.

The scarlet fever outbreak setting has provided an unexpected model for understanding S. pyogenes transmission and immunity. Children of the same age were exposed to a presumed similar inoculum of S. pyogenes in class, yet only ∼5% developed scarlet fever, some of whom re-acquired the outbreak strain without illness three weeks later, while some anecdotally developed pharyngitis. Others were found to asymptomatically shed S. pyogenes for several weeks, while others showed transient colonisation, or no infection whatsoever. Our understanding of immunity to S. pyogenes is heavily dominated by factors that influence susceptibility to invasive disease. The findings highlight a gap in our knowledge regarding mucosal immunity to S. pyogenes, specifically whether full immunity requires antibodies to prevent streptococcal adherence, promote bacterial clearance, and inhibit streptococcal virulence factors during pharyngitis. This is potentially critical to vaccine development. Coupled with differences in genetic susceptibility and oral microbiota, differing layers of immunity could go some way to explain why children express a range of disease phenotypes.

There are limitations to our study. Sampling of fomites was limited to single timepoints in year 1, while settle plates to sample air were only used in year 2, restricting ability to determine the relative importance of each transmission route. Our study was based in London, so findings may not be relevant to rural, or low- and middle-income settings, or regions with different climate. Furthermore, transmission intensity may be seasonal; we attempted to sample outside of the main scarlet fever season, but schools were unwilling to participate. Indeed, engagement was high when anxiety about scarlet fever was greatest.

The study has demonstrated that heavy asymptomatic shedding by a proportion of children may account for persistence of S. pyogenes outbreaks in classroom settings. The phenomenon of super-shedders and super-spreaders is increasingly recognised as a source of heterogeneity in infectious disease modelling (29, 30) and is consistent with the explosive nature of some streptococcal outbreaks. Relevance to other respiratory infections is unclear. Our findings may explain the periodic failure of interventions focussed on hygiene alone to curtail outbreaks of S. pyogenes, both in the classroom and other institutional settings. The recognition of heavy asymptomatic shedding highlights a potential role for social distancing, improved respiratory hygiene, and better ventilation in reducing S. pyogenes transmission during outbreaks. As an unforeseen consequence of the COVID-19 pandemic, implementation of these measures has proven highly successful in halting England’s scarlet fever upsurge, at least in 2020 (31), highlighting a need for better understanding of transmission routes in preventing future upsurges.

Declaration of interests

The authors declare no competing interests

Funding

Action Medical Research (GN2596); UKRI, Medical Research Council (MR/P022669/1); NIHR Health Protection Research Unit in Healthcare-associated Infection and AMR; NIHR Imperial College Biomedical Research Centre (BRC).

Author contributions

Conceptualisation and funding acquisition SS, RC, TL; Project supervision and methodology SS; Recruitment supervision RC; data collection LB, AKP, MM, EM, MKS, RCYL, DR, PH; data analysis AV, EJ; data visualisation RC, SS, AV, EJ; LB, AKP, and RC verified the identifiable participant data; EM, EJ, and AV verified linked non-identifiable microbiological and genomic data.

Original draft RC, SS; Review and editing AKP, EJ, AV, PH, DR, RCYL, TL; Final manuscript all authors. All authors had full access to all non-identifiable data in the study and agreed with the decision to submit for publication.

Data sharing

All data are included in the supplementary materials or are available from European Nucleotide Archive using the reference PRJEB43915.