1. Epidemiology and Global Health

Effect of an enhanced public health contact tracing intervention on the secondary transmission of SARS-CoV-2 in educational settings: The four-way decomposition analysis

  1. Olivera Djuric  Is a corresponding author
  2. Elisabetta Larosa
  3. Mariateresa Cassinadri
  4. Silvia Cilloni
  5. Eufemia Bisaccia
  6. Davide Pepe
  7. Laura Bonvicini
  8. Massimo Vicentini
  9. Francesco Venturelli
  10. Paolo Giorgi Rossi
  11. Patrizio Pezzotti
  12. Alberto Mateo Urdiales
  13. Emanuela Bedeschi
  14. The Reggio Emilia Covid-19 Working Group
  1. Epidemiology Unit, Azienda Unità Sanitaria Locale – IRCCS di Reggio Emilia, Italy
  2. Centre for Environmental, Nutritional and Genetic Epidemiology (CREAGEN), University of Modena and Reggio Emilia, Italy
  3. Public Health Unit, Azienda Unità Sanitaria Locale-IRCCS di Reggio Emilia, Italy
  4. Department of Infectious Diseases, Istituto Superiore di Sanità, Italy
https://doi.org/10.7554/eLife.85802
Share this article
Cite this article
  1. Olivera Djuric
  2. Elisabetta Larosa
  3. Mariateresa Cassinadri
  4. Silvia Cilloni
  5. Eufemia Bisaccia
  6. Davide Pepe
  7. Laura Bonvicini
  8. Massimo Vicentini
  9. Francesco Venturelli
  10. Paolo Giorgi Rossi
  11. Patrizio Pezzotti
  12. Alberto Mateo Urdiales
  13. Emanuela Bedeschi
  14. The Reggio Emilia Covid-19 Working Group
(2024)
Effect of an enhanced public health contact tracing intervention on the secondary transmission of SARS-CoV-2 in educational settings: The four-way decomposition analysis
eLife 13:e85802.
https://doi.org/10.7554/eLife.85802
  2. Download BibTeX
  3. Download .RIS
5 figures, 5 tables and 5 additional files

Figures

Figure 1
Download asset Open asset
Simplified timeline of transmission in household and among school contacts in the presence of standard contact tracing and backward contact tracing.

(A) In standard contact tracing, all close contacts were quarantined after identifying a case in the community. Contacts were only tested at the end of the quarantine or if symptomatic. Only for school contacts, immediate testing of all classmates was performed; if one or more classmates resulted positive, the whole class was quarantined. (B) In backward contact tracing, close contacts were also immediately tested, independently from the presence of symptoms. The tracing and quarantine policy in schools was similar. In the proposed example, after the diagnosis of a symptomatic household member, backward tracing would identify an asymptomatic child, thus allowing the extension of investigation to his school contacts and eventually stopping secondary transmission in the class.

Figure 2
Download asset Open asset
Simplified illustration of the pre and post intervention scenarios.

In panel A we report the scenario without prompt contact testing in community and its effect on the SARS-CoV-2 transmission in educational setting. Day 0: One of the children in a household became infected (primary case) but asymptomatic (gray). Day 5: One parent and one classmate became infected, also asymptomatic (gray). Day 10: The infected parent became symptomatic (orange), tested positive (red circle), and considered an index case of the household. Entire family is quarantined (bold line) but not tested immediately. Meanwhile, the primary case transmits infection further to two other classmates. Classmates of the primary case are not tested because they are not identified as school contacts due to late testing of the household contacts. Day 20: Family members of the index case are tested at the end of the quarantine. One positive classmate of the primary case became symptomatic, tested positive, and considered an index case in the school cluster given that the classmates were not considered contacts of the primary case since he was already isolated. Other classmates are tested only when an index case occurs. Panel B illustrates the scenario with prompt contact testing in community. Day 10: The infected parent became symptomatic, tested positive, and entire family was quarantined and tested at the beginning of quarantine. Primary case is identified promptly, his classmates are identified as contacts, tested, and isolated preventing further transmission of the virus.

Figure 3
Download asset Open asset
Weekly notification rates of new COVID-19 cases per 100,000 inhabitants, ages 0–19, by age class, Reggio Emilia Province, 1 September 2020 – 4 April 2021.

The graph also reports the main changes in school opening and school closures and the proportion of Alpha variants (green area) among sequenced cases reported by the Italian National Institute of Health.

Figure 4
Download asset Open asset
Directed acyclic graph for the association of intervention and number of secondary cases, mediated by contact tracing delay and known contact of the index case, adjusted for the type of school, type of index case, and class size.
Figure 5
Download asset Open asset
Upper graph: Weekly notification rates of COVID-19 per 100.000 inhabitants of Reggio Emilia Province, 1 September 2020 – 4 April 2021 and number of classes investigated.

Lower graph: Weekly average contact tracing delay and percentage of index cases with a known source of infection.

Tables

Table 1
Characteristics of 1884 classes and 1604 index cases for which a school contact with COVID-19 cases was suspected, before, and after the intervention.
n (%)Before interventionn=490After interventionn=1394
Classes (n=1884)
Type of school
 Infant-toddler centre350 (18.5)107 (21.8)243 (17.4)
 Primary school540 (28.7)125 (25.5)415 (29.8)
 Middle school496 (26.3)128 (26.1)368 (26.4)
 High school478 (25.4)129 (26.3)349 (25.0)
 Other educational services20 (1.1)1 (0.2)19 (1.4)
Calendar period
 September/October248 (13.1)
 November263 (13.9)
 December316 (16.8)
 January265 (14.1)
 February523 (27.8)
 March/April269 (14.3)
Class size
 <21862 (45.7)191 (39.0)671 (48.1)
 ≥211011 (53.7)293 (59.8)718 (51.5)
 Missing11 (0.6)6 (1.2)5 (0.4)
Secondary transmission
 No1157 (61.4)307 (62.6)850 (61.0)
 Yes727 (38.6)183 (37.4)544 (39.0)
 Number of secondary cases*1 (1-3)2 (1-3)1 (1-3)
 Mean attack rate0.1 (0.04–0.12)0.1 (0.04–0.12)0.1 (0.04–0.12)
Part of a school cluster
 No1 367 (72.6)368 (75.1)999 (71.7)
 Yes517 (27.4)122 (24.9)395 (28.3)
 Tracing delay*3 (2-5)7 (5-10)3 (2-4)
 Testing delay*4 (2-8)5 (3-8)4 (2-7)
Index cases (n=1604)n=429n=1,175
Type of index case
 Student1213 (75.6)321 (74.8)892 (75·9)
 Teacher391 (24.4)108 (25.2)283 (24.1)
Index case symptomatic
 No298 (18.6)63 (14.7)235 (20)
 Yes1306 (81.4)366 (85.3)940 (80)
Potential source of infection
 Unknown814 (50.7)283 (66·0)531 (45.2)
 Known790 (49.3)146 (34·0)644 (54.8)
Type of source
 Household outbreak614 (77.7)97 (66.4)517 (80.3)
 Social contact26 (3.3)7 (4.8)19 (2.9)
 Sport contact18 (2.3)7 (4.8)11 (1.7)
 Unidentifiable contact132 (16.7)35 (24.0)97 (15.1)
  1. *

    Median (IQR), calculated only in classes with secondary transmission.

Table 2
Association between class or index case characteristics (potential confounders) and number of secondary cases.
Number of classes with secondary transmissionNumber of secondary casesIRR* (95% CI)
Classes (n=1884)n=727n=1706
Type of school
 Infant-toddler centre131 (18.2)349ref
 Primary school217 (29.8)5531.03 (0.80–1.31)
 Middle school172 (23.7)3860.78 (0.60–1.01)
 High school202 (27.8)4090.86 (0.66–1.11)
 Other educational services5 (0.7)90.45 (0.17–1.18)
Class size
 <21316 (43.5)751ref
 ≥21411 (56.5)9551.08 (0.91–1.29)
Index cases (n=1604)n=640
Type of index case
 Student477 (74.5)1047ref
 Teacher163 (25.5)6580.75 (0.61–0.92)
 Screening01na
  1. *

    Relative risks are computed with negative binomial models with the count of secondary cases as a dependent variable.

Table 3
Association between class or index case characteristics and the process indicators (potential mediators).
TotalnKnown source of infection of the index casen (%)*Index case symptomaticn (%)*Tracing delayMedian (IQR)
Classes (n=1884)
Type of school
 Infant-toddler centre350157 (44.9)281 (80.3)3 (2-5)
 Primary school540255 (47.2)355 (65.7)3 (2-5)
 Middle school496206 (41.5)293 (59.1)3 (2-6)
 High school478216 (45.2)360 (75.3)3 (2-6)
 Other educational services207 (35.0)10 (50.0)3 (1.5–5)
 P value†0.3780.0010.147
Class size
 <21862375 (41.4)587 (81.6)3 (2-5)
 ≥211011461 (45.6)707 (69.9)3 (2-6)
 Missing115 (45.5)5 (45.5)6 (3-7)
 p value†0.6610.7820.367
Index cases (n=1604)
Type of index case
 Student1213683 (56.3)957 (78.9)3 (2-5)
 Teacher391104 (26.7)342 (87.5)3 (2-6)
 p value†<0.001<0.0010.486
  1. *

    Values are numbers with row percentages.

  2. †

    Kruskal-Wallis test.

Table 4
Negative binomial regression of the association between the number of secondary cases (outcome) and intervention promptness indicators (mediators).
Entire period(n=1884)Before Alpha variant (n=827)
IRR*95% CIIRR*95% CI
Tracing delay1.010.99–1.041.030.99–1.07
Known source of infection of the index case0.750.63–0.910.730.55–0.96
Index case symptomatic1.210.96–1.531.300.93–1.82
  1. *

    Adjusted for the type of school, type of index case, and class size.

Table 5
Four-way decomposition mediation analysis of the association between intervention and the number of secondary cases.
ERR*95% CI
Total effect–0.1–0.35–0.15
Controlled direct effect
 Known contact (M=1)–0.29–0.61–0.03
 Unknown contact (M=0)0.31–0.49 to –0.02
Pure indirect effect–0.02–0.10–0.07
Mediated interaction–0.14–0.28 to –0.01
Reference interaction
 Known contact (M=1)0.350.03–0.68
 Unknown contact (M=0)–0.25–0.49 to –0.02
  1. *

    Adjusted for the type of school, type of index case, and class size.

  2. ERR, excess relative risk; M, mediator (known source of infection of the index case).

Additional files

Supplementary file 1

Plain language definitions of the mediation analysis.

https://cdn.elifesciences.org/articles/85802/elife-85802-supp1-v1.docx
Supplementary file 2

STATA code.

https://cdn.elifesciences.org/articles/85802/elife-85802-supp2-v1.docx
Supplementary file 3

Weekly notification rates of new COVID-19 cases in Reggio Emilia.

https://cdn.elifesciences.org/articles/85802/elife-85802-supp3-v1.xlsx
Supplementary file 4

Weekly average contact tracing delay and percentage of index cases with a known source of infection.

https://cdn.elifesciences.org/articles/85802/elife-85802-supp4-v1.xlsx
MDAR checklist
https://cdn.elifesciences.org/articles/85802/elife-85802-mdarchecklist1-v1.docx

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Olivera Djuric
  2. Elisabetta Larosa
  3. Mariateresa Cassinadri
  4. Silvia Cilloni
  5. Eufemia Bisaccia
  6. Davide Pepe
  7. Laura Bonvicini
  8. Massimo Vicentini
  9. Francesco Venturelli
  10. Paolo Giorgi Rossi
  11. Patrizio Pezzotti
  12. Alberto Mateo Urdiales
  13. Emanuela Bedeschi
  14. The Reggio Emilia Covid-19 Working Group
(2024)
Effect of an enhanced public health contact tracing intervention on the secondary transmission of SARS-CoV-2 in educational settings: The four-way decomposition analysis
eLife 13:e85802.
https://doi.org/10.7554/eLife.85802