Household COVID-19 secondary attack rates in India are high
In the wake of the second-wave of COVID-19 in India, there is a need to re-examine several public health strategies, from vaccination to education on reducing the spread of COVID-19.
The public service awareness campaigns in India on social distancing, the use of masks and hand sanitization are designed to be highly conspicuous, an example being the 30-second educational “caller tune” that is played when any one places a phone call to anyone else. These awareness campaigns do not sufficiently reinforce the education for the appropriate behavior in one of the most important settings, home.
One reason may be to avoid alarmist messaging, but a more probable reason is that risk of COVID-19 transmission at home in the Indian context was not quantified. How having one or more earning members or educated members benefits the rest of the family is usually discussed in a progressive sense. Without being an alarmist, an equally important discussion on how infections can in-principle spread or be arrested when one person at home is infected, is needed.
Seconday attack. An infected person is likely to spread the infection, in this case COVID-19, during the course of interactions with others in different settings including hospitals, work place, classroom or even at home. The probability that a person interacting with the index patient in a specific setting is likely to develop the infection is quantified using the Secondary attack rate (SAR).
SAR at home = (number of infected people at home - 1 person who first had it)/(total number of people at home - 1 person who first had it)
Globally average of this number is about 0.166 or 16.6%, i.e., if there is one COVID-19 infected person at home, 16% of the rest of the family members are likely to develop the COVID-19 infection. While we were discussing this with Ms. Sandhya Ramesh, she pointed out that many families she is aware of had most members infected. A sentiment which was echoed by many others later. This feedback prompted us to run a survey on this very important epidemiological parameter.
Our survey and its results. We conducted an online survey to collect the data required to estimate the household SAR in India. Many including Santosh Ansumali and Rahul Raj, were very helpful in spreading out the message. The participants were allowed to report either about their family or about a family for which they know the detailed COVID-19 history. The data on the size of the city and state to which the affected family belongs, the number of bedrooms in the house, the size of the household, whether at least one member of the family has had a Bachelor’s level education, ages of the all the members of the household, age of the index patient and of those who were infected within 2–3 weeks of the primary infection, the possibility of home isolation and number of hospitalizations or deaths were sought.
Responses were received from 128 households comprising of 610 individuals, from across India. Most of them live in a house with at least 2 bed rooms, and have at least one person with a Bachelor’s level education or more at home, suggesting the possibility that the survey responses are mostly from the middle class demographic, which mostly had an option to self-isolate. The total number of COVID-19 infections among the participants were 443, 87 from the first wave (March 2020-Jan 2021) and 356 from the second wave (Feb 2021-Jun 2021). Brijesh Saraswat performed the statistical analyses of this data.
Exceedingly High SAR. The average SAR at home (SARH) was 0.653, and the SARH for the different subgroups was also estimated (Table above). What this suggests is that on average 65% of the remaining members of the family are likely to be infected. The SARH was slightly lower if the index patient could immediately self-isolate with the suspicion of COVID-19 symptoms 0.593 compared to those who could not (0.744). There was no significant difference in the between the SARH from the first (0.681) and the second (0.647) waves, suggesting that the high SARH may not be due to the virus variants. Families living in 2 bed rooms had a slightly higher SARH (0.724) compared to those in 4 bed room houses (0.622).
The average SARH from different studies across the world varied from 0 to 0.45, with overall estimate of 0.166 (Ref. Madewell et al. 2020) with the exception of 0.898 from Egypt (Ref. Gomaa et al. 2021). The average household size in India is larger than the typical average of around 2.5 in USA and 2.3 in Europe. The SARH and the number of household contacts NH are both high relative to those in the western countries, making the contribution SARH x NH (Ref. Liu et al. 2020) of the household transmissions to the basic reproduction rate (R0) also high.
Is it true, are there avoidable factors, how to spread message about them. Considering that the COVID-19 susceptible population in India may still be high, the present estimation of high SARH suggests three important actionable steps to reduce the risks at home —
Firstly, the survey needs to be expanded. We are deeply thankful to the participants of our humble survey, which is still available for those who wish to contribute to the learnings about the SAR-at home.
https://docs.google.com/forms/d/1m3jKlFwzlvheokVes6huNlcVmfm9P8pDZZu-GOM-ioo/
Looking beyond our survey, which to our knowledge is the first one in India, the data from across the different social strata need to be collected through conventional and online surveys, or made public if it has already available. Only a larger, and expanded data set may advance the analysis beyond the equally valid argument-counter argument, that there may be an over-representation in the current data from the worst affected families, and that a survey of this nature may be too sensitive for such families to participate.
Secondly, the possible sociological underpinnings of the unavoidable and avoidable factors behind the high household transmissions need to be understood. The clearly unavoidable aspects are the high density of occupancy at home, the practical difficulty of isolation with a separate room + bathroom and/or a lack of education which prevents early identification of symptoms.
In some of our interviews the following curious, perhaps avoidable, interactions with the member to be in isolation, with some symptoms but mostly fit for minor physical activities, also surfaced.
Case study of a young teenager who is infected. The teenager who never was used to cleaning dirty dishes leaves them outside the room, and one of the other members of the family cleans them for this person. Or another member of the family enters the room to clean the room for the person in isolation. What could easily be done is that the teen in isolation could clean the plates and leave them out to dry.
Case study of a senior member of the family infected. The family does not feel it is appropriate to subject the elderly person to isolation, and they wait at least until the test results come out. Even when the person is placed in isolation, the person was allowed to dine with the rest of the family “carefully”. What could be done is to begin early isolation and not lose a few days between the suspicion of early symptoms to obtaining the test results. A clear non-negotiable protocol on not eating together until the infected or suspected recover fully is needed.
Case study of a cooking member of the family infected. The mother/wife who usually cooks the meal wears a mask and prepares the meal while the rest of the family waits elsewhere “carefully”. After preparing the meal, the infected person retires to her isolation room and the meal is eaten by the rest of the family members. While some of these are sadly hailed this as testimonies of love (Ref. https://www.dnaindia.com/viral/report-viral-virender-sehwag-offers-help-to-woman-seen-cooking-while-being-on-oxygen-support-2891878), the tough love of asking the family members to be content with alternative arrangements for a few days is absolutely required.
These case studies seem to echo with what happened at many homes. The interesting thing is that all these people think they do it “carefully”. For those who can not afford a space for isolation the reality is very different. But the educated who can afford to dedicate a space for those with symptoms should do so in a non-negotiable way.
And finally, the right public awareness campaigns need to be created not so much to alarm the public but to educate them about minor changes which can have a significant impact.
Scientific experts should use their scholarship in english (“.. keep the doors ajar ..”) or in technical jargon (“ .. the role of air, aerosols in COVID-19 spread ..”) with moderation when the messages are intended for public communication. The air, aerosol debate in public forums led many lay people confused about whether they should even open their windows as COVID-19 spreads through air!
The “caller tune” message system is remarkable. However it is a fairly static message of 30 seconds, mainly in english or hindi, with secondary references to seek information elsewhere “please call this number or visit this site for more details”.
In stead, many short caller tune messages should be prepared and played randomly so there is a novelty, and learning each time we hear the message. Some of the messages could be -
- Do not remove your mask, especially to speak (4 to 5 seconds)
- If someone is sick, please do not ask them to cook (4 to 5 seconds)
- If you take a vaccine it protects you directly, and other members at home indirectly (4 to 5 seconds)
- If a person is in isolation, do not clean their used dishes (4 to 5 seconds)
- One infected person at home can transmit to others. Try your best not to bring infections home. Not everyone has a good immune system (10 seconds)
These are just suggestions to go beyond a static public health messaging. A fruitful collaboration between public health and communications experts can lead to messaging which is correct, clear and impactful.
Acknowledgements. We would like to thank Sandhya Ramesh for the asking engaging questions on the high secondary attack rates in India, which prompted the present study. We would also like to thank Santosh Ansumali for helpful discussions and Rahul Raj for help with surveys. And of course, special thanks to Mr. Brijesh Saraswat for performing calculations and analyses.
