Skip to main content

Association between living with children and outcomes from COVID-19: an OpenSAFELY cohort study of 12 million adults in England

Our take —

This study of 9,157,814 people in England from February 1 to August 3, 2020, available as a preprint and thus not yet peer reviewed, sought to identify the risk of COVID-19 among adults due to living with children of different age groups. They found adults <65 years old had a reduced hazard of death (26% reduction) if they lived with children 0 to 11 years old than if they did not. They also found that living with children 12 to 18 years old increased the risk of infection by 8%. There was no significant effect of living with children for adults >65 years old on any outcomes. This was a national study from England, which has important policy implications for lockdown restrictions, including potential school closures. Caution in the interpretation is warranted as unaccounted for behavioral differences within families with small children could have affected their risk of infection, and authors did not account for timing of school closures.

Updated Review Available

This expert summary is for the non peer-reviewed preprint. We also summarized this paper after it underwent peer-review and was published in BMJ on March 18, 2021. You can find our updated review of the published article here.

Study design

Prospective Cohort

Study population and setting

The study objective was to examine the risk of infection among adults associated with living with children of different age groups, both during and following school lockdown orders in England. The study included 9,157,814 adults between >=18 and <=65 years of age, and an additional cohort of adults >65 years old (N=2,567,671), and assessed the impact on SARS-CoV-2 infection from primary care records in The Phoenix Partnership, COVID-19 hospital admission using data from the Secondary Uses Service, COVID-19 ICU admission from the Intensive Care National Audit & Research Center, and death due to COVID-19 noted in the Office for National Statistics mortality records. This was done on the OpenSAFELY data analytics platform created for the National Health Service (NHS) of England. The study population required individuals to have >=3 months of active follow-up via general practices using the Phoenix Partnerhsip software from February 1, 2020 onwards. Hospital admission data were available until May 1, 2020, while infection outcomes were available through August 3. 2020. Children were linked to households via a household identifier and enumerated based on age, and exposure categories for adults in the study were grouped as: (1) no children under 18 in the house; (2) any child 0 to 11 years of age; (3) no children 0 to 11 years of age but one or more children 12 to 18 years old.

Summary of Main Findings

The study found that living with children 0 to 11 years of age was not associated with an increase in SARS-CoV-2 infection, COVID-19 hospital admission, or ICU admission among adults <65 years old. It did significantly reduce the hazard of death from COVID-19 by 26% (95% CI: 0.60 – 0.90). Living with children 12 to 18 years of age increased the risk of infection by 8% (95% CI: 1.03 – 1.13) but was not associated with any other outcomes. For adults <65 years of age, living with children age 0 to 11 years reduced the risk of death from non-COVID-19 causes by 32% (95% CI: 0.62 – 0.74), and by 27% (95% CI: 0.66 – 0.81) for those living with children 12 to 18 years. For adults >65 years of age, there were no associations with any of the outcomes, including infection, hospital admission, ICU admission, COVID-19-specific death, or non-COVID-19-death.

Study Strengths

The study was able to use a large sample size based on attendance of a TPP general practice during the study period, which means they likely had power to identify infections and outcomes when they occurred. By linking to a number of registries, it had follow-up for the majority of this large population. They also controlled for many potential confounders, including age, sex, body mass index, smoking status, deprivation index, ethnicity, geographic area, and the total number of individuals in a household. They also controlled for chronic comorbidities associated with severe COVID-19 outcomes, further reducing the potential for confounding in these estimates.


The largest limitation was data availability—for instance, for hospital admission they only had from February 1 to May 1, 2020, while for the other outcomes, they had longer follow-up until August. Given hospital admission data is focused in the beginning of the pandemic during the first wave, it may not be reflective of more recent trends. Occupation was also unmeasured, which could reflect whether individuals stayed in the house or were essential workers that had to continue contact. There also may be differences between risk behaviors among parents of children compared to those without that may impact their risk of exposure, thereby impacting the risk ratios with unmeasured confounding. Similarly, the study could not adjust for temporality differences such as school closures or other lockdown restrictions imposed, likely due to variability across schools. Finally, their measure of children was based off simply the number of children linked in the record, and may not actually reflect contact due to children residing with a different parent or with other family members.

Value added

This is the largest and most comprehensive study from a nation-wide cohort that examines the change in risk of COVID-19 in adults due to contact with children.

This review was posted on: 7 December 2020