In this nation-wide register study, we used SARS-CoV-2 PCR test results from all public and private Norwegian laboratories during 2020, combined with administrative population register data that identified all family members (parents and children) living at the same address.
The index cases are mainly the parents (66%), even though there are fewer parents than children in the families. The children aged 17–20 years comprise 22% of all the children, but as many as 42% of the child index cases. The younger children are not often the index case. This could be because they display less symptoms, because it is easier to test an adult than a small child if both have symptoms, or because small children are less often infected outside of the family.
Subsequent transmissions within the families follow different patterns depending on the age of the index case. SAR7 was high both when a parent was the index (24%, 95% CI 24 to 25) and when a young child aged 0–6 was index (24%, 95% CI 20 to 28). However, it was low when an older child aged 17–20 was the index (11%, 95% CI 10 to 13). In short, parents and older children contribute the most to the introduction of SARS-CoV-2 into the family, while parents and young children contribute the most to transmitting the disease within the family. It could be that older children more often have another residence than the one they are registered at, i.e. in student accommodation or due to shared custody between parents. However, older children may also behave in ways that restrict viral transfer more than young children.
Parents transmit to other parents (35%, 95% CI 33 to 36) and less to children (21%, 95% CI 20 to 22), while children transmit similarly to parents (15%, 95%CI 14 to 16) and other children (12%, 95%CI 11 to 13).
Most cases of household transmission are detected by day seven after the index tested positive (SAR7 21%), increasing only somewhat to day 14 (SAR14 24%). In this study, the share of the non-index family members who had been tested was 69% (95%CI 69 to 70) by day 7, increasing to 72% (95% CI 72 to 73) by day 14.
The percent of the non-index family members who were tested within seven days was low in March and April (20%), but increased to 50% in May and to above 80% in December (Fig. 1). While SAR7 grew in parallel with the percent tested from March to May, when testing rates grew beyond 50% from May, SAR7 remained around 20%.
Comparison with related studies
Previous literature on intra-family transmission of SARS-CoV-2 is inconclusive, as the studies are few and small, with different designs, and report widely varying SARs (1–7, 11–13).
We observed an overall SAR in line with previous studies, e.g. in two systematic reviews Madewell et al. reported household SAR of 17% (1) and Lei et al. of 27% (12), both suggesting lower SARs to children than to adults. Viner et al. also estimate lower SAR from children than adults (7). However, comparison across studies is hard due to varying follow-up times, unclear handling of co-index cases, different testing regimes and small samples, especially for young child index cases. Very few studies calculate SARs across characteristics of the index or separately to parents and children. Grijalva et al. is a notable exception, but they only have five index children below the age of 12 (5). Viner et al. find that their data were insufficient to conclude whether transmission of SARS-CoV-2 by children is lower than by adults, and they conclude their review by stating that studies “that investigate secondary infections from child or adolescent index cases compared with secondary infections from adult index cases are particularly needed» (7).
Norway has based much of its pandemic response on a demanding strategy of coordinated application of testing everyone with minor symptoms, isolation of positive cases, careful tracing of probable contacts and quarantine through the incubation period. The indications for testing, definitions of close contacts and length of quarantine have been regulated by law and adjusted over the course of the pandemic.
The observed increase in testing of family members reflects the increasing availability and reliance on testing throughout the pandemic. PCR-testing has been widely available in Norway, though in the first few months of the pandemic only symptomatic cases and healthcare personnel had wide access to the tests. From May and onwards, however, anyone wanting a test could get one by contacting their local municipal test-station, where testing was free of charge. This resulted in intensive testing as the infection rate started to grow from August, increasing the probability of persons with few or no symptoms to be included as index cases, and increasing the probability of secondary cases to be identified. It is worth noting that the frequency of testing for children, whilst generally lower than for the whole population, has been relatively high in relation to outbreaks in schools and nurseries.
Our study found that the recorded SAR7 was nearly as high as the SAR14. Almost all detected intra-family transmission occurs within the first seven days after the detection of an index case. This supports the current Norwegian strategy of testing and quarantine for 7–10 days for all family members after infection within the household.
An aggressive test/quarantine/isolate/-trace strategy can influence SAR and explain some of the variation in SAR in different studies. One would suspect that SAR7 is lower if the index case was tested in the presymptomatic period, as part of a contact tracing regimen, than if the index case was tested after symptom development, when the family would have been exposed for a longer time without prevention measures. We do not know how many of the index cases in our study who had symptoms, or whether they were tested because they were included in contact tracing around another case outside of the family.
We found substantial differences in the SAR depending on the characteristics of both the index case and the family composition. SAR was higher when young children (0–6 years) compared with older children were index, probably reflecting that the youngest need more close contact with their caregivers. Other studies have suggested that children have a lower attack rate and lower predisposition to serious disease and onward spread (1, 3, 6, 11). After the lockdown period in March and April, when the nurseries and schools reopened, strict infection control measures were applied to prevent transmission in these institutions. To which extent these measures have been successful needs to be examined further, but our results underlines that children should be kept at home when they have symptoms that could indicate infection, and, moreover, that grandparents and other caregivers in risk groups for severe COVID-19 should not provide childcare for symptomatic children.
When the index case was a parent, there was a higher SAR towards the other parent than towards the children, and the SAR from young children was higher towards their parents than towards their siblings. The lower SAR rate we observe with increasing age of the index child most likely reflects the ability to identify cases earlier due to symptomatology, and also less close contact among the older children and the adults in the house. For very young children, it will likely be difficult to reduce contact even when contagion is detected. We also found a higher SAR associated with older parents than with younger ones. This may be due to a higher level of symptoms, and perhaps a higher level of caution resulting in more testing due to higher risk for serious disease among the oldest parents. However, we do not have data to explore this question.
As could be expected, a priori, we saw an increased SAR associated with the index living with a larger family. This reflects the larger number of contacts and probably more cramped living conditions. Norwegian advice has been that when isolating at home, cases should where possible, have their own bathroom, meals brought to them and as little contact with the rest of the house as possible. This is harder in situations of large families sharing smaller living spaces. Alternative housing has only been offered and accepted to a limited extent. Measures that make alternative housing more appealing, for instance moving the whole household to a larger dwelling rather than splitting out the index, may be considered.
The current Norwegian recommendations are that all close contacts should be tested at least once, preferably twice within a ten-day (recently seven day) period after the diagnosis of the index. In our study, the proportion of the non-index family members who had been tested by seven days after index date, is very high, but not 100%. It is thus possible that asymptomatic cases are sometimes not tested, and there is reason to assume that children below six years of age are overrepresented in this group, because they are more difficult to test. This may affect who we identify as index cases, and maybe also which secondary cases are identified. For example, when a young child (0–6) was index, about 90% of the family members were tested within seven days after index date, while this number was about 70% when a parent was index, which might suggest that more secondary cases were identified when index was a child than when index was a parent.
Better knowledge of actual directions of transmission within families requires prospective studies where all family members are tested daily with the same method in the week following index identification, preferably also with reporting of symptoms and genome identification of the viruses. Genome analysis would also help to reveal exposure to multiple infection events within the same family, which could interfere with the detected SAR. However, the incidence rate of SARS-CoV-2 has been low in Norway during the pandemic, estimated to peak at 74 per 100 000 per week toward the end of 2020 (9), which makes several transmission incidents into the same family at the same time less likely. A clear advantage of our study to such prospective studies, is that we do not have attrition: We observe every family, and we can observe all family members in the follow-up period, regardless of motivation to participate in a study or not. Indeed, our data stem from a real-world situation, where detection of secondary cases relates to a combination of the actual transmission of the virus and the behavioral responses to disease and the actual testing regime. This point is illustrated by us seeing lower SAR in the two first months of the pandemic, when testing capacity was limited, than later, when testing of family members was widely available. It seems that a test capacity where about 50% of the non-index family members are being tested, results in roughly the same SAR as a test regime where 80% of the non-index family members are tested. Another way of putting this, is that testing of cases with mild symptoms captures most of the cases. It should also be noted that we intentionally excluded families where there were “co-indexes” that could result in a greater infection pressure within the household.
Another limitation to our study is that the observation period does not include the coldest winter months, where people usually spend more time indoors, and when the climate is more favorable for viral sustainability on surfaces, and maybe also for transmission. Similar factors may also explain some of the variation in SARs between different studies. We observed a small increase in SAR during December, which may be a coincidence, or perhaps due to colder weather or closer contact between family members during the celebration of Christmas. Moreover, the introduction of variants of new and more easily transferable virus mutants could have played a role.