Study population and setting
This systematic review and meta-analysis described the rate of further infections (secondary attack rate, SAR) within households where at least one person tested positive for SARS-CoV-2. It identified 485 published articles through database searches; screening procedures removed duplicates and examined relevance, whether full-text versions were available, and whether they reported the rate of secondary transmission in households. After screening and assessment, 40 published studies were included in the meta-analysis.
Summary of Main Findings
There was substantial heterogeneity across the 40 studies. Overall, there was an estimated mean secondary attack rate of 19.0% (95% CI: 14.9 – 23.1%) in a household, and 18.1% (95% CI: 12.9 – 34.8%) among family contacts. There was no statistically significant difference between household and family secondary infection rates, however having a reported close contact with someone infected did increase the SAR by 4.3% (95% CI: 2.9 – 5.6%). There was a statistically significantly (P<0.001) higher SAR from people who were symptomatic and brought the infection home to household contacts (19.9% SAR, 95% CI: 14.0 – 25.7%) compared to those asymptomatic index cases (0.7%, 95% CI: 0.0 – 3.8%). It also estimated 32.5% (95% CI: 7.4 – 57.5%) of households with an index case had a secondary case. Spouses had statistically significant (P<0.001) increased secondary transmission (43.3% of spouses were infected) compared to other relatives of the index case (18.3%), and adult contacts had significant (P<0.001) increased transmission (31.0%) compared to children (15.7%).
The study drew upon numerous others to form the basis of its meta-analysis, which means these results are more likely to be valid estimates than if they had only found a handful. Because they had 40 studies in their pool, they were able to perform a number of further analyses to look at different dynamics of transmission, such as being a spouse versus being another relative, or differences in infection based on self-reported sex. They also controlled for major sociodemographic variables in their meta-analysis, which reduces some of the confounding between study populations and leads to less biased estimates.
As the study conducted its various secondary analyses, the sample size dwindled, and in some cases this led to imprecise estimates with large confidence intervals. There were also differences between study samples and protocols that may result in further heterogeneity that is not wholly considered. For instance, there were differences in the study protocols for frequency and type of SARS-CoV-2 test, rates of community transmission, and differences in case definition between index case versus contacts. Asymptomatic infections were likely excluded from most of the studies, leading to a potential underestimate in the rate of transmission. Many studies could not rule out secondary infections that occurred outside the home, which, if widespread, may inflate the estimates.
This is a large meta-analysis estimating the rate of secondary infection due to household transmission, which is a pressing public health concern as reopening policies are debated.
This review was posted on: 12 August 2020