Household Transmission of SARS-CoV-2

Key Points Question What is the household secondary attack rate for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)? Findings In this meta-analysis of 54 studies with 77 758 participants, the estimated overall household secondary attack rate was 16.6%, higher than observed secondary attack rates for SARS-CoV and Middle East respiratory syndrome coronavirus. Controlling for differences across studies, secondary attack rates were higher in households from symptomatic index cases than asymptomatic index cases, to adult contacts than to child contacts, to spouses than to other family contacts, and in households with 1 contact than households with 3 or more contacts. Meaning These findings suggest that households are and will continue to be important venues for transmission, even in areas where community transmission is reduced.

This supplemental material has been provided by the authors to give readers additional information about their work.

#1 AND #2
We searched PubMed using terms described in this table for studies of SARS-CoV-2, SARS-CoV, Middle East Respiratory Syndrome coronavirus (MERS-CoV), and human coronaviruses NL63 (HCoV-NL63), OC43 (HCoV-OC43), 229E (HCoV-229E), and HKU1 (HCoV-HKU1) with no restrictions on language, study design, time, or place of publication. Pre-prints were included. We also manually searched the reference lists of eligible studies from PubMed to identify further eligible studies. For influenza, we used the secondary attack rates reported in one comprehensive review. 1  We included articles with original data for estimating household secondary attack rate of SARS-CoV-2. The publication must report at least two of numerator, denominator, and secondary attack rate among household contacts. Where numerators (numbers of infected/sick contacts) or denominators (numbers of contacts) were not reported but the number of index cases and secondary attack rate were available, the denominator was calculated acknowledging limits of significant digits. We did not contact authors for additional data.
Household transmission of SARS-CoV-2 is described in case reports in narrative form of individual families or households with large numbers of cases, and contact tracing investigations whereby investigators identify the number of infected household or family members from index cases, usually over a period of 14 or 21 days. We excluded studies that 1) were case reports of individual families or households as these can bias results towards high attack rates, 2) reported infection prevalence in the household without describing transmission, 3) tested household contacts using antibody tests as antibodies may be detected many weeks after infection, 75 4) were of close contacts that did not report secondary attack rates for household or family members, 5) were preprints of published articles already included in the review, and 6) had overlapping populations with another study already included in the review. We used the same eligibility criteria for studies of other coronaviruses. One reviewer first screened studies by titles and abstracts to identify potential studies for inclusion. One reviewer subsequently evaluated full-text articles and selected those that met the inclusion criteria.

eAppendix 2. Data Extraction
One reviewer extracted the following information for studies of SARS-CoV-2: first author, location, index case identification period, index case identification method, number of index cases, index case symptom status, transmission mitigation strategies, household/family/close contact type, test used to diagnose contacts, universal/symptomatic testing, number of tests per contact, follow-up duration, number of infected household contacts, total number of household contacts, household secondary attack rate, and transmission risk factors. For studies that reported secondary attack rates for both household and family contacts, only the household contact secondary attack rate was extracted. For studies that included secondary attack rates for close contacts, we also extracted the number of infected close contacts, total number of close contacts, and close contact secondary attack rate.
For studies of other coronaviruses, one reviewer extracted: first author, location, number of index cases, number of infected contacts, total number of contacts, secondary attack rate, contact type, and test used to diagnose contacts. For influenza, we extracted secondary attack rate ranges reported in the review.
The following methods describe how secondary attack rates were calculated and how specific covariates were categorized if there was ambiguity during the process of data extraction:

eAppendix 3. Additional Description of Studies
Household index cases were identified by passive surveillance, active surveillance of key populations (e.g., travelers from areas with active SARS-CoV-2 transmission; individuals detected by neighborhood fever screenings), and both active and passive surveillance (see eTable 3). Some studies included index cases with SARS-CoV-2 infections (both symptomatic and asymptomatic), whereas others included symptomatic COVID-19 index cases only. Another targeted asymptomatic SARS-CoV-2 infected index cases, 28 some of whom developed symptoms during a follow-up period. Several studies stated they assumed all secondary cases were infected by the index case to whom they were traced, 27,36,50 others excluded secondary cases if they developed symptoms before exposure to the primary case, 13,32 another excluded household contacts assessed to have the same exposure as the COVID-19 index cases, 25 and another randomly selected one index case as the infector. 7 Ignoring tertiary transmission inflates the secondary attack rate.
Many studies only included household contacts, but others included family members, or other close contacts, including individuals outside the household. We assumed that studies of household contacts included anyone living in the same household as the index case unless stated otherwise. For example, several studies reported household contacts as family members in households. 6,20,47,48,55 Several studies further restricted household contacts to those who spent at least one night or 24 hours in the house after symptom onset of the index case. 6,25,30,48,50 Most studies involved tracing contacts and monitoring them for 14 or 21 days. Monitoring methods included phone calls, 6,10,18,22,26,32,35,41 text messages, 10,18 or direct observation by healthcare workers. 33,36 Some studies tested all contacts immediately after the index case was diagnosed at the onset of the observation period and monitored them for symptoms. 7,11,15,20,25,28,54 Others tested all contacts during or at the end of the observation period regardless of symptoms, 17,27,34,37,48,51 whereas others only tested symptomatic contacts (see eTable 4). One study only tested asymptomatic contacts. 17 Several studies tested contacts multiple times throughout the observation period irrespective of symptoms. Of those studies, several also reported extra testing of individuals who developed symptoms during quarantine. 11,15,29,30,47,55 Other studies tested all contacts, 15,21,39,40,43,56 or interviewed index cases about symptoms of household members, 5,38 immediately without additional monitoring. Many studies, particularly those in China, reported in-home quarantine of contacts during the observation period after index cases were confirmed.
Case ascertainment was primarily done via RT-PCR on nasopharyngeal or oropharyngeal samples. Several studies also reported whole-genome sequencing, 8,[47][48][49] and nucleic acid tests. 12,49,55 Three studies identified index cases via RT-PCR, but only collected symptom information about household contacts from telephone interviews with index cases. 5,38,44

Factors for infectiousness
Regarding the timing of transmission, several studies reported SARS-CoV-2 transmission prior to index case symptom onset. 13,24,31,55 One study reported higher transmission risk after symptom onset relative to the incubation period. 31 Others found no significant difference in secondary transmission between contacts exposed to index cases before or after illness onset, 13,30,50 although close contacts were quarantined for two studies, 13,30 and index cases were isolated after symptom onset in the third. 50 Some studies reported infection risk peaked during exposure to the index case 2-4 days before or within 5 days of symptom onset. 13,21,31 Another reported a higher secondary attack rate among close contacts of pre-symptomatic index cases than asymptomatic carriers. 13 Diarrhea, 48 pneumonia, 5,13 acute respiratory distress syndrome, 13 myalgia, 31 chills, 31 dizziness, 31 lymphocyte count, 52 neutrophil percentage, 52 and expectoration, 33 were associated with secondary transmission in some studies. Symptoms not shown to be associated with infectivity were fever, 30,31,33,48,50,52 fatigue, 31,33,52 dyspnea, 31 headache, 31 nasal congestion, 31 pharyngalgia, 31 arthralgia, 31 rhinorrhea, 31 nausea, 31 vomiting, 31 chest tightness, 31 palpitation, 31 poor appetite, 31 abdominal pain, 31 and white blood cell count. 52

Awareness and behavioral factors
Several studies explored whether prevention measures were associated with reduced transmission. Contacts who wore face masks and index cases who wore masks all the time after illness onset had lower odds of infection and transmission, respectively. 17,48,51 Conversely, contacts who did not apply protective measures (e.g., face mask, avoiding contact with index case) had higher odds of infection. 50 One study reported that greater frequency of chlorine/ethanol based disinfectant use for house cleaning and ventilation hours per day were associated with reduced risk, 48 whereas another did not. 50 Frequency of room cleaning (wet type) and hand hygiene were not significant. 17,48 Index case isolation after illness onset was associated with reduced secondary transmission. 30 Other studies did not find the time interval from illness onset to medical isolation, 48 hospital admission, 48,50,52 or laboratory confirmation, 48,52 to be associated with transmission. Self-awareness of being infected with SARS-CoV-2 and knowledge of COVID-19 were not significant, 48 but lack of knowledge of index case's own infectiousness was associated with transmission. 48 Health profession of the index case was a protective factor in one study. 5 Physical contact, 17,50 and sharing a vehicle, 7,31,33,50,51 living room, 50 cigarette, 17 or meal, 48,50 were associated with infection, but eating with separate tableware was not. 17,48 Smoking behavior in index cases or contacts was not associated with transmission. 29,50