Characteristics associated with household transmission of SARS-CoV-2 in Ontario, Canada
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (ScreenIT)
- Evaluated articles (NCRC)
Abstract
BACKGROUND
Within-household transmission of SARS-CoV-2 infection has been identified as one of the main sources of spread of COVID-19 after lockdown restrictions and self-isolation guidelines are implemented. Secondary attack rates among household contacts are estimated to be five to ten times higher than among non-household contacts, but it is unclear which individuals are more prone to transmit infection within their households.
METHODS
Using address matching, a cohort was assembled of all laboratory-confirmed cases of COVID-19 residing in private households in Ontario, Canada. Descriptive analyses were performed to compare characteristics of cases in households that experienced secondary transmission versus those that did not. Logistic regression models were fit to determine index case characteristics and neighbourhood characteristics associated with transmission.
FINDINGS
Between January and July, 2020, there were 26,152 cases of COVID-19 residing in 21,226 households. Longer testing delays (≥5 days versus 0 days OR=3·02, 95% CI: 2·53 - 3·60) and male sex (OR=1·28, 95% CI: 1·18 - 1·38) were associated with greater odds of household secondary transmission, while being a healthcare worker (OR=0·56, 95% CI: 0·50 - 0·62) was associated with lower odds of transmission. Neighbourhoods with larger average economic family size and a higher proportion of households with multiple persons per room were also associated with greater odds of transmission.
INTERPRETATION
It is important for individuals to get tested for SARS-CoV-2 infection as soon as symptoms appear, and to isolate away from household contacts; this is particularly important in neighbourhoods with large family sizes and/or crowded households.
FUNDING
This study was supported by Public Health Ontario.
Research in Context
EVIDENCE BEFORE THIS STUDY
We searched PubMed and Google Scholar up to September 3, 2020 to identify individual-level cohort studies or meta-analyses on household transmission of COVID-19. We used the search terms (“COVID” OR “SARS-CoV-2”) AND (“household” [Title]), and also reviewed the reference lists of any studies found during the search to identify additional studies. We considered studies that reported secondary attack rates and/or other measures of association (i.e., relative risk, odds ratio, or hazard ratio) for household transmission. We did not consider any modelling studies, studies that focused specifically on children, or small case studies that included less than three households. The search returned 53 studies, of which 51 were included in three meta-analyses. Pooled household secondary attack rates from the three meta-analyses were 19%, 27%, and 30%; secondary attack rates in households were estimated to be five to ten times as high as in non-household settings. Most studies were conducted in Asia and identified households from contact tracing, with individual studies reporting on fewer than 6000 households. Most studies did not consider households with no secondary transmission, and focused on a limited set of secondary case characteristics.
ADDED VALUE OF THIS STUDY
We applied an address matching algorithm, which identified 21,226 private households of laboratory-confirmed cases of COVID-19 in Ontario, Canada. Ontario has the advantage of a universal healthcare system and population-wide data for the entire province. To our knowledge, this study contains the largest number of private households with at least one confirmed case of COVID-19. We compared a variety of individual- and neighbourhood-level characteristics of households with and without secondary transmission. We also applied logistic regression models to determine index case characteristics associated with transmission, which gave important insights into factors that may help reduce secondary transmission in households.
IMPLICATIONS OF ALL THE AVAILABLE EVIDENCE
Findings from this study and existing evidence suggest that testing delays and household crowding play important roles in whether household secondary transmission occurs. Odds of household transmission may be reduced by cases seeking testing as soon as symptoms appear, and self-isolating outside the home or in a room alone if possible. These strategies may be considered by public health officials to reduce household transmission and mitigate local spread of COVID-19. Future research should further investigate the role of children and youth in household transmission.
Article activity feed
-
SciScore for 10.1101/2020.10.22.20217802: (What is this?)
Please note, not all rigor criteria are appropriate for all manuscripts.
Table 1: Rigor
Institutional Review Board Statement IRB: 12 We obtained ethics approval from Public Health Ontario’s Research Ethics Board. Randomization not detected. Blinding not detected. Power Analysis not detected. Sex as a biological variable not detected. Table 2: Resources
Software and Algorithms Sentences Resources For address matching, we applied a natural language processing algorithm using Python’s sklearn library to identify unique households that contained at least one COVID-19 case. Python’ssuggested: (PyMVPA, RRID:SCR_006099)Results from OddPub: We did not detect open data. We also did not detect open code. Researchers are encouraged to share open data when possible (see Nature blog).
Results from LimitationRec…SciScore for 10.1101/2020.10.22.20217802: (What is this?)
Please note, not all rigor criteria are appropriate for all manuscripts.
Table 1: Rigor
Institutional Review Board Statement IRB: 12 We obtained ethics approval from Public Health Ontario’s Research Ethics Board. Randomization not detected. Blinding not detected. Power Analysis not detected. Sex as a biological variable not detected. Table 2: Resources
Software and Algorithms Sentences Resources For address matching, we applied a natural language processing algorithm using Python’s sklearn library to identify unique households that contained at least one COVID-19 case. Python’ssuggested: (PyMVPA, RRID:SCR_006099)Results from OddPub: We did not detect open data. We also did not detect open code. Researchers are encouraged to share open data when possible (see Nature blog).
Results from LimitationRecognizer: We detected the following sentences addressing limitations in the study:Our study has some limitations that merit discussion. First, we did not have information on the total number of individuals residing in each household or the characteristics of uninfected household members, thus we were unable to calculate the proportion of household contacts infected to generate SARs. However, we were able to control for some neighbourhood-level characteristics of household composition including economic family size and proportion of households with multiple persons per room or multi-family households. Our finding of high transmission and acquisition of SARS-CoV-2 infection between individuals in the same age group therefore likely reflects the inherent age structures of households in Ontario. Second, we may have misclassified some index cases if a previously infected individual within the household was untested (e.g., asymptomatic or symptomatic but did not seek testing), and we may have misclassified some secondary cases if their infection was acquired outside the household. We may also have missed secondary cases within a household that were untested. Third, we only considered one index case per household, and considered all subsequent cases within a 14-day period to be secondary cases (i.e., did not account for tertiary transmission). Fourth, as this study encompasses a period before schools were re-opened in the fall, there were few index cases among children (N=190) and as such, we were not able to determine the extent to which children played a role i...
Results from TrialIdentifier: No clinical trial numbers were referenced.
Results from Barzooka: We did not find any issues relating to the usage of bar graphs.
Results from JetFighter: We did not find any issues relating to colormaps.
Results from rtransparent:- Thank you for including a conflict of interest statement. Authors are encouraged to include this statement when submitting to a journal.
- Thank you for including a funding statement. Authors are encouraged to include this statement when submitting to a journal.
- No protocol registration statement was detected.
-
Our take
The objective of this study, available as a preprint and thus not yet peer reviewed, was to identify the rate of secondary transmission within households in Ontario, Canada from January to July 2020. Overall, the study identified 26,152 cases, with 7,993 (30.6%) from households with secondary transmission. Of these, 4,926 were cases of secondary transmission. Among their index cases that resulted in secondary transmission, there was a median 1 additional case. The study used machine learning techniques to address-match cases to a household, and there may be misspecification based on this algorithm. They also were limited to only laboratory-confirmed cases reported to the public health authority and thus asymptomatic cases may have been missed; nevertheless, this represents the largest study of household transmission …
Our take
The objective of this study, available as a preprint and thus not yet peer reviewed, was to identify the rate of secondary transmission within households in Ontario, Canada from January to July 2020. Overall, the study identified 26,152 cases, with 7,993 (30.6%) from households with secondary transmission. Of these, 4,926 were cases of secondary transmission. Among their index cases that resulted in secondary transmission, there was a median 1 additional case. The study used machine learning techniques to address-match cases to a household, and there may be misspecification based on this algorithm. They also were limited to only laboratory-confirmed cases reported to the public health authority and thus asymptomatic cases may have been missed; nevertheless, this represents the largest study of household transmission to date.
Study design
prospective-cohort
Study population and setting
The study objective was to describe the rates of secondary transmission within households for all laboratory-confirmed COVID-19 cases in Ontario, Canada among 21,226 private households using data from Public Health Ontario from January to July 2020. Using natural language processing, the study attempted to address-match cases living in households, apartment buildings, and multi-unit dwellings, but not those that address-matched to congregate facilities such as prisons or nursing homes. Individuals living in apartments or multi-unit dwellings without a specific apartment number were also excluded. Secondary transmission in a household was defined as address-matched cases within 1 to 14 days after the index case was identified. Households with multiple cases on the index date were also excluded, given it was not possible to determine if secondary transmission occurred within the index cluster. The study collected individual-level data such as employment status, high-risk status based on age and pre-existing comorbidities, and prior association with a known COVID-19 outbreak (such as at a workplace, or a long-term care facility), case month, case’s age, sex, and region of residence. They also assessed delay metrics: (1) delay between symptom onset and testing; (2) delay between specimen collection and receipt of test results; and (3) delay between the test report and entry into the disease reporting system. Finally, neighborhood characteristics from the 2016 Canada census were considered, including average family size, proportion of households with multiple persons per room, proportion of multi-family households, and urbanicity.
Summary of main findings
The study identified 38,984 confirmed COVID-19 cases in Ontario. After exclusion criteria based on index clustering and address-matching, there were 26,152 cases residing in private households, with 18,169 cases (69.5%) from households without secondary transmission and 7,993 (30.6%) with secondary transmission. Of these, there were 3,067 index cases, with a median of 1 case of secondary transmission per index case. The study found adults 20 to 59 years old and considered low-risk (based on comorbidities and age) were more likely to acquire and transmit infection within households. Index cases without secondary transmission were more likely to be healthcare workers (OR: 0.56, 95% CI: 0.50 – 0.62) or to have been associated with an outbreak outside the home (OR: 0.61, 95% CI: 0.55 – 0.68). Individuals without any symptoms flagged in the reportable disease system also had reduced odds of household transmission (OR: 0.48, 95% CI: 0.38 – 0.61). Living in a neighborhood with larger than average family size was also associated with increased transmission of nearly 1.88 odds per one-person increase (95% CI: 1.70 – 2.09).
Study strengths
The study used the reportable disease system to identify a large number of cases in Ontario, which likely reduced selection bias to have a more representative population. They also were able to link this data using address-matching, which made it easier to automatically identify people within households than if they had required individuals to report one another. They also used a comprehensive set of covariates that examined not only individual- or neighborhood-level risk factors, but also health system-level with their delay on testing.
Limitations
They primary limitation was that the study did not have the total number of individuals living in a given household, and therefore had to estimate using the neighborhood-level characteristics and other individual-level characteristics. They also were limited to the individuals with laboratory-confirmed testing, as opposed to COVID-19 diagnosis which would capture cases that did not get tested or universal testing strategies that would better identify asymptomatic cases in the population. Therefore, there may be some misclassification of cases. Additionally, during this time, schools were largely closed and did not reopen until the fall, after the study period had ended. Therefore, the lack of transmission among children may not be generalizable to other time periods. Finally, because the address-matching was through machine learning techniques, there may be some misspecification based on the processor and some individuals may be mis-matched.
Value added
This study is the largest of private households reported to date, as others most often used standard contact tracing methods which limited their sample.
-