The potential impact of intensified community hand hygiene interventions on respiratory tract infections: a modelling study

Hand hygiene is among the most fundamental and widely used behavioural measures to reduce the person-to-person spread of human pathogens and its effectiveness as a community intervention is supported by evidence from randomized trials. However, a theoretical understanding of the relationship between hand hygiene frequency and change in risk of infection is lacking. Using a simple model-based framework for understanding the determinants of hand hygiene effectiveness in preventing viral respiratory tract infections, we show that a crucial, but overlooked, determinant of the relationship between hand hygiene frequency and risk of infection via indirect transmission is persistence of viable virus on hands. If persistence is short, as has been reported for influenza, hand-washing needs to be performed very frequently or immediately after hand contamination to substantially reduce the probability of infection. When viable virus survival is longer (e.g. in the presence of mucus or for some enveloped viruses) less frequent hand washing can substantially reduce the infection probability. Immediate hand washing after contamination is consistently more effective than at fixed-time intervals. Our study highlights that recommendations on hand hygiene should be tailored to persistence of viable virus on hands and that more detailed empirical investigations are needed to help optimize this key intervention.


3
Another issue is that if viral survival on hands and surfaces is longer then this may lead to higher secondary attack rates but this does not seem to be factored into the model. By callibrating long and short survival viruses to the same baseline secondary attack rate the model is effectively presenting findings in terms of relative risk reductions rather than absolute risk reductions -if the absolute baseline risk is higher in longer survival viruses then the same relative risk reduction represents a bigger reduction in absolute risk for long-surviving viruses. This issue is not really factored into the results or policy implications. It might be usefully addressed by calibrating longer surviving viruses to a higher baseline attack rate than shorter surviving viruses.
The other main observation is that event prompted handwashing i.e. after the contamination event is more effective than regular handwashing at the same frequency.
This makes sense, however the authors have set the event prompted washing rate at 4 times per hour whereas the observational study they previously cite for touching of surfaces within an office is 112 times per hour. This really raises the question of whether event prompted hand hygiene is feasible. Perhaps this could be discussed.
The models focus on acquisition rather than spread of infection via the hand hygiene route. When the event in question is, for example blowing or wiping one's nose (catch it, kill it, bin it) then there is a clear event of interest. It would be interesting to model this or at least highlight in the discussion that that the strategies discussed do not take account of onward transmission from an infected subject.
The other important point that is perhaps underemphaised is that reducing the hand contamination rate is also effective in reducing risk. The authors mention the public health strategy of cleaning surfaces but do not mention reduction in surface touching as a strategy to reduce transmission.
Overall, I see this a very useful paper and my comments are aimed at allowing the authors to either refine their analysis or nuance their discussion, rather than seeing them as undermining their main conclusions.

09-Dec-2021 Dear Mrs Pham
The Editor of Proceedings A has now received comments from referees on the above paper and would like you to revise it in accordance with their suggestions which can be found below (not including confidential reports to the Editor).
Please submit a copy of your revised paper within four weeks -if we do not hear from you within this time then it will be assumed that the paper has been withdrawn. In exceptional circumstances, extensions may be possible if agreed with the Editorial Office in advance.
Please note that it is the editorial policy of Proceedings A to offer authors one round of revision in which to address changes requested by referees. If the revisions are not considered satisfactory by the Editor, then the paper will be rejected, and not considered further for publication by the journal. In the event that the author chooses not to address a referee's comments, and no scientific 5 Referee: 2 Comments to the Author(s) The paper is a useful addition to the handwashing literature in that it models the potential effect of hand hygiene interventions according to timing and frequency of handwashing and survival of pathogen on hands. The conclusions have useful implications for the design of handwashing interventions in different settings to maximise impact.
As with all models a key challenge is the development of appropriate parameters to include in the model, The kay insight is that if pathogens survive on the hand for only short periods then more frequent handwashing is needed to reduce risk compared to pathogens with longer survival on hands. This finding is highly robust to different values of parameters chosen.
The authors calibrate the model to a 10% secondary attack rate over 12 hours as the baseline attack rate -they say this is based on household secondary attack rates -however household secondary attack rates are driven by multiple routes of transmission not just surface to hand to mucous membrane -the direct droplet and aerosol part of the transmission are not amenable to interruption through hand hygiene -in this case, surely the model should be calibrated to the proportion of this secondary attack rate the authors think is not due due to direct droplet or aerosol spread -could the authors clarify this point?
Another issue is that if viral survival on hands and surfaces is longer then this may lead to higher secondary attack rates but this does not seem to be factored into the model. By callibrating long and short survival viruses to the same baseline secondary attack rate the model is effectively presenting findings in terms of relative risk reductions rather than absolute risk reductions -if the absolute baseline risk is higher in longer survival viruses then the same relative risk reduction represents a bigger reduction in absolute risk for long-surviving viruses. This issue is not really factored into the results or policy implications. It might be usefully addressed by calibrating longer surviving viruses to a higher baseline attack rate than shorter surviving viruses.
The other main observation is that event prompted handwashing i.e. after the contamination event is more effective than regular handwashing at the same frequency. This makes sense, however the authors have set the event prompted washing rate at 4 times per hour whereas the observational study they previously cite for touching of surfaces within an office is 112 times per hour. This really raises the question of whether event prompted hand hygiene is feasible. Perhaps this could be discussed.
The models focus on acquisition rather than spread of infection via the hand hygiene route. When the event in question is, for example blowing or wiping one's nose (catch it, kill it, bin it) then there is a clear event of interest. It would be interesting to model this or at least highlight in the discussion that that the strategies discussed do not take account of onward transmission from an infected subject.
The other important point that is perhaps underemphaised is that reducing the hand contamination rate is also effective in reducing risk. The authors mention the public health strategy of cleaning surfaces but do not mention reduction in surface touching as a strategy to reduce transmission.
Overall, I see this a very useful paper and my comments are aimed at allowing the authors to either refine their analysis or nuance their discussion, rather than seeing them as undermining their main conclusions.

Is the paper of sufficient general interest? Excellent
Is the overall quality of the paper suitable? Good Can the paper be shortened without overall detriment to the main message? Yes Do you think some of the material would be more appropriate as an electronic appendix? No

Recommendation? Accept as is
Comments to the Author(s) I believe that the authors have successfully addressed all my comments, and I am happy to recommend publication as it is.

07-Apr-2022
Dear Mrs Pham I am pleased to inform you that your manuscript entitled "The potential impact of intensified community hand hygiene interventions on respiratory tract infections: a modelling study" has been accepted in its final form for publication in Proceedings A.
Our Production Office will be in contact with you in due course. You can expect to receive a proof of your article soon. Please contact the office to let us know if you are likely to be away from e-mail in the near future. If you do not notify us and comments are not received within 5 days of sending the proof, we may publish the paper as it stands.
As a reminder, you have provided the following 'Data accessibility statement' (if applicable). Please remember to make any data sets live prior to publication, and update any links as needed when you receive a proof to check. It is good practice to also add data sets to your reference list. &lt;em&gt;Statement (if applicable): &lt;/em&gt;The code reproducing the results of this study is available at https://github.com/tm-pham/ handhygiene_modelling. Note that if you have opted for open access then payment will be required before the article is published -payment instructions will follow shortly.

Open access
If you wish to opt for open access then please inform the editorial office (proceedingsa@royalsociety.org) as soon as possible.
Your article has been estimated as being 14 pages long. Our Production Office will inform you of the exact length at the proof stage.
Proceedings A levies charges for articles which exceed 20 printed pages. (based upon approximately 540 words or 2 figures per page). Articles exceeding this limit will incur page charges of £150 per page or part page, plus VAT (where applicable).
Under the terms of our licence to publish you may post the author generated postprint (ie. your accepted version not the final typeset version) of your manuscript at any time and this can be made freely available. Postprints can be deposited on a personal or institutional website, or a recognised server/repository. Please note however, that the reporting of postprints is subject to a media embargo, and that the status the manuscript should be made clear. Upon publication of the definitive version on the publisher's site, full details and a link should be added.
You can cite the article in advance of publication using its DOI. The DOI will take the form: For tips on promoting your accepted paper see our blog post: https://royalsociety.org/blog/2020/07/promoting-your-latest-paper-and-tracking-yourresults/

Reviewer recommendation
Manuscript Number: RSPA-2021-0746 Title: The potential impact of intensified community hand hygiene interventions on respiratory tract infections: a modelling study.
In this work, authors aim to quantify the impact of interventions related to hand hygiene on exposure to respiratory tract infections. To do this, they consider a relatively simple mathematical model which considers exposure for a single individual through the fomite route. In particular, the model represents: (i) contamination of hand(s) via contact with contaminated surfaces, but where the contamination of these surfaces is not explicitly modelled; (ii) two different hand-hygiene regimes based on regular hand washing vs hand washing prompted after surface contact with some delay; (iii) persistence of virus on hands in terms of an exponential decay; and (iv) exposure via contacts of the hand with membranes.
I think this is a nice contribution to the field, and I think it should be published after authors address the comments below. The role played by the fomite transmission route has not been very extensively analysed yet, and this type of approach strikes a nice balance between model complexity, feasibility of analysis and interpretability of results. Still, I have a number of comments/questions/concerns that I believe the authors should address.
Major comments, mainly related to model assumptions: It is not clear (not to me), from the description in Figures 1 and 2 and within the text, what happens when a sequence of hand contamination events occur. Would the "Probability of infection" curve in Figure 1 go back to its baseline peak original value if another hand contamination event occurs before hand washing? λ c is the rate of the Poisson process for the number of hand contamination events. This parameter will effectively depend on: (a) the rate at which the individual touches surfaces; and (b) the percentage of surfaces that are contaminated in that given environment at any given time. One could just vary this parameter to account for heterogeneity in (a) (across individuals and situations in real life). However, more thought needs to be made for (b). If the authors are considering as a major factor in the study the "persistence of the pathogen on hands", it is to be expected that the persistence (eg half-life) on hands and on surfaces would be highly correlated for many pathogens. This means that pathogens that survive for longer on hands would likely survive also for longer on surfaces, and then a fair comparison would imply considering larger values for λ c for these pathogens (i.e. when T 1/2 is large)? I think some numerical results should focus on this, and some analysis would be needed for the conclusions in the paper to be robust.
It is a standard approach in Quantitative Microbial Risk Assessment (QMRA) to explicitly model the level of contamination on hands and surfaces (e.g. P F U/cm 2 ). The authors have instead followed a simpler approach, which has limitations but leads to a simple model which provides useful insights. Still, the fact that the level of contamination (at least) on hands is not explicitly modelled is a very important limitation of the study, which is currently not mentioned. In general, and related to this, I got the feeling that some relevant literature is missing in the manuscript: The fact that the level of contamination on hands is not explicitly modelled leads to a number of relatively strong assumptions in the model, which relate to my following points.
is the probability of infection per hand-mucosa contact. In a real scenario, this probability would depend on: (a) the level of contamination on hands (P F U/cm 2 ); (b) the hand area involved on the hand-mucosa contact; (c) the number of contacts in a relatively short period of time, leading to an accumulation of dose; and (d) the infectivity of the pathogen and susceptibility of the host (eg in terms of a dose-response curve). Merging these factors into a single fixed parameter is a limitation of the study which should be highlighted. In particular, is unlikely to be independent across sequential contacts. If is large due to high hand contamination levels in a given contact, it will likely be large in the next contact for the same reason. Some of this is partially accounted for through the exponential decay of virus on hands, but not fully (eg very large initial contamination levels on hands would lead to a very large probability of infection for particular dose-reponse curves, regardless of the decay happening).
-Doses can accumulate on membranes across sequential contacts in the short-term, leading to an accumulation of risk (i.e. risk for the cumulative dose).
-The approach followed by the authors to set (e.g. Figure S4) is not fully clear to me. Why would depend on the half-life? I understand that they are trying to make things comparable by fixing the cumulative probability of transmission, but is this a fair comparison? A particular pathogen would have a given half-life for persistence on hands, and a given based on (a)-(d) above. Does it make sense to force the two pathogens with different half-lifes to be "equally infectious" by accordingly modifying ? This might be affecting some of the conclusions. But I might have missunderstood something here.
Authors should clarify when they talk about infectious virus vs viral RNA. In Figure S1, they mention infectious virus, but in Section 2(d) they mention a study [15] which detected virus (I think here it is just RNA) on 53% of surfaces. If it is RNA, I imagine this might lead to an overestimate of contaminated surfaces for the purposes of infection transmission modelling, since I believe the ratio between viral RNA and infectious virus can be as small as 1:1000, and not all those 53% of surfaces would be "infectious" (or infectious enough to contaminate the hand enough to cause infection via later on).
Authors choose a cumulative probability of infection of 10% over a period of 12h based on attack rates for Influenza from [10][11][12][13]. Are these quantified rates for fomite transmission? or in general regardless of the route? If these are general, then 10% in 12h might be a significant over estimate of infection risk via the fomite route. Some sensitivity analysis for λ f is needed, I believe, unless I have missed it. The face-touching rate can vary across individuals, ages (e.g. kids vs adults), but also depending on environmental conditions (public vs private spaces; facemasks etc). This parameter might have a significant impact on some of the exposure predictions, where hand hygiene might not be effective when λ f is large, regardless of the hand washing frequency strategy.

Minor comments:
For consistency, authors might want to use the same symbol (X) for transmission blocked in Figures 1 and 2.
Authors state "We assumed that when hand washing is performed... P (t i ) is reduced to zero". It is probably a sensible assumption given high efficacy of hand washing, but authors might want to support this statement with some references/arguments. Is this implicitly assuming not only that the sanitizer has large efficacy, but also that the hand-washing is performed correctly?
Computed instead of cmputed in Figure S4 caption.
The Github link in the Data Accessibility statement is broken, probably because of the index h. In this repository, the authors might want to add some extra description of the codes in the README file.
Please explain how the cumulative probability of infection expression in line 55, Page 12, is obtained.

Dear Raminder Shergill and Professor Matjaz Perc,
Herewith we would like to resubmit the revised manuscript entitled "The potential impact of intensified community hand hygiene interventions on respiratory tract infections: a modelling study" (RSPA-2021-0746) for your consideration for publication in Proceedings of the Royal Society A.
We would like to thank the editors and the reviewers for the valuable comments and suggestions, and the opportunity to revise our manuscript. We have carefully revised the manuscript with changes marked in blue and a point-by-point reply to the reviewers. Sincerely,

Thi Mui Pham on behalf of all authors
Appendix B

Reviewer 1
In this work, authors aim to quantify the impact of interventions related to hand hygiene on exposure to respiratory tract infections. To do this, they consider a relatively simple mathematical model which considers exposure for a single individual through the fomite route. In particular, the model represents: (i) contamination of hand(s) via contact with contaminated surfaces, but where the contamination of these surfaces is not explicitly modelled; (ii) two different hand-hygiene regimes based on regular hand washing vs hand washing prompted after surface contact with some delay; (iii) persistence of virus on hands in terms of an exponential decay; and (iv) exposure via contacts of the hand with membranes.
I think this is a nice contribution to the field, and I think it should be published after authors address the comments below. The role played by the fomite transmission route has not been very extensively analysed yet, and this type of approach strikes a nice balance between model complexity, feasibility of analysis and interpretability of results. Still, I have a number of comments/questions/concerns that I believe the authors should address.

Major comments, mainly related to model assumptions:
1. It is not clear (not to me), from the description in Figures 1 and 2  Response: Thank you for pointing out that our manuscript did not address the consequences of a sequence of hand contamination events. We indeed assume in our model that, in case of a sequence of hand contamination events, the probability of viral persistence is reset to one after each hand contamination event while the probability of transmission per face-touching contact is kept the same. The probability infection is, thus, reset to its ϵ baseline value. We, therefore, assume no accumulation of infection risk. In addition, we also assume that the functional shape of this probability is always the same. Our model, therefore, might underestimate the effect of hand washing in reducing the cumulative probability of infection. We have added the following sentence to the Methods section ("Mathematical model", page 4-5): "In case of a sequence of hand contamination events, we assumed that the probability P(t) is reset to its baseline value of one and that the transmission probability is kept the same. As ϵ such, the probability P(t) is assumed to always have the same functional shape even after sequential hand contamination events. We, therefore, do not account for accumulation of infection risks neither for heterogeneity of hand contamination events." We also added the following paragraph to the Discussion section (page 10): "We assumed no accumulation of the probability of infection for a sequence of hand contamination events and did not account for heterogeneity of hand contamination events.
Instead, the probability of viral persistence is reset to its baseline value of one and the transmission probability is kept equal for consecutive hand contamination events. Our ϵ model, therefore, might underestimate the reduction in the probability of infection induced by hand washing. However, since we do not expect this to affect the hand washing schemes differently, we do not expect this to change our qualitative conclusions."

2.
is the rate of the Poisson process for the number of hand contamination events. This λ parameter will effectively depend on: (a) the rate at which the individual touches surfaces; and (b) the percentage of surfaces that are contaminated in that given environment at any given time. One could just vary this parameter to account for heterogeneity in (a) (across individuals and situations in real life). However, more thought needs to be made for (b). If the authors are considering as a major factor in the study the "persistence of the pathogen on hands", it is to be expected that the persistence (eg half-life) on hands and on surfaces would be highly correlated for many pathogens. This means that pathogens that survive for longer on hands would likely survive also for longer on surfaces, and then a fair comparison would imply considering larger values for for these pathogens (i.e. when is large)? I λ 1/2 think some numerical results should focus on this, and some analysis would be needed for the conclusions in the paper to be robust.
Response: We agree that depends both on (a) and (b) and we agree that there might be a λ correlation between the persistence on hands and the hand contamination rate . However, λ it's not clear what exactly this relationship would look like as pathogens may behave very differently on human tissue than on artificial surfaces. In addition, this can vary across pathogens and depend on the surface type and environmental conditions. As we will discuss in our response to 4. iii), our study highlights that for a new pathogen with observed secondary attack rates (e.g., through household studies) we would need to find out about the half-life on contaminated hands to make statements about how transmission would be expected to change with increasing hand hygiene. For the analyses in the main text, we fixed the baseline probability of infection and the hand contamination rate, assuming that these parameters could be determined through other (household) studies. If indeed, there is a correlation between the half-life and the hand contamination rate, our analyses need to be extended and we have explored this in the supplementary material (page 34ff. "Correlation between half-life of viral persistence on hands and hand contamination rate"). The results in Figure S20 and Figure  (for the Diamond princess cruise) or Jones (2020) for a hospital setting.
Response: Thank you for pointing out the missing literature in the manuscript. We have incorporated many of the suggested articles along with additional literature on that topic in the introduction (page 2) and the discussion (page 10). In particular, we have also added literature on the relative contribution of fomite-mediated transmission routes for influenza in the Methods section (see "Parameters", page 5). We agree that our decision to neglect a dose-response relationship and the level of contamination of hands is a limitation of our study. However, we think that our simpler model provides a sufficient framework to assess the effect of the two hand washing schemes that we considered.
4. The fact that the level of contamination on hands is not explicitly modelled leads to a number of relatively strong assumptions in the model, which relate to my following points. it is unlikely to be independent across sequential contacts. If is large due to ϵ high hand contamination levels in a given contact, it will likely be large in the next contact for the same reason. Some of this is partially accounted for through the exponential decay of virus on hands, but not fully (eg very large initial contamination levels on hands would lead to a very large probability of infection for particular dose-response curves, regardless of the decay happening).

Response: We assumed that is fixed and constant for all infection events. ϵ
As such, each self-infection event is maximally correlated as it is assumed to be the same. However, as we did not account for different levels of hand iii. The approach followed by the authors to set (e.g. Figure S4) is not fully ϵ clear to me. Why would depend on the half-life? I understand that they are trying to make things comparable by fixing the cumulative probability of transmission, but is this a fair comparison? A particular pathogen would have a given half-life for persistence on hands, and a given based on (a)-(d) above. Does it make sense to force the two pathogens with different half-lifes to be equally infectious" by accordingly modifying ? This might be ϵ affecting some of the conclusions. But I might have misunderstood something here.
Response: Indeed the reviewer is correct: in Figure 3 we are interested in comparing the impact of hand washing frequency on pathogens of equal transmissibility but with different half-lives, and this motivated our decision to vary to achieve the same overall transmission probability for the ϵ different pathogens. The purpose of this was to highlight the impact of the half-life (which is often poorly quantified) on the effectiveness of hand hygiene, thus highlighting that for a new pathogen of known transmissibility we would need to find out about the half-life on contaminated hands to make statements about how transmission would be expected to change with increasing hand hygiene. We certainly didn't mean to imply that increased persistence of a pathogen on hands would not increase transmissibility. To address this potential confusion we have now added the following text to the Methods section ("Parameter", page 5): "The probability of transmission per face-touching event was constrained to meet a fixed probability of infection to reflect the fact that we are interested in how our beliefs about how the potential impact of enhanced hand hygiene for a pathogen of known transmissibility will vary according to what we know or believe about its survival on hands." Figure S1, they mention infectious virus, but in Section 2(d) they mention a study [15] which detected virus (I think here it is just RNA) on 53% of surfaces. If it is RNA, I imagine this might lead to an overestimate of contaminated surfaces for the purposes of infection transmission modelling, since I believe the ratio between viral RNA and infectious virus can be as small as 1:1000, and not all those 53% of surfaces would be "infectious" (or infectious enough to contaminate the hand enough to cause infection via later on).

Authors should clarify when they talk about infectious virus vs viral RNA. In
Response: Indeed, in the paper by Boone et al (2004) that we cite viral viability was not assessed. Assuming 100% of detected RNA is from viable virus, we used this paper to inform the upper bound for the hand contamination event rate. However, we acknowledge that a direct translation from detected RNA to viable virus is likely unrealistic and should be seen as a theoretical upper bound. We have added the following sentence to the Methods section (see "Parameters", page 5) to highlight this assumption: "Note that hour -1 is based on a RNA to viable virus ratio of 1:1 and should be seen as λ = 60 a theoretical upper bound as in practice, this ratio is likely much smaller." In our main analysis, we used a much lower hand contamination rate of 4 events per hour.
We performed sensitivity analyses for hand contamination rates of 1, 20 and 60 (page 28ff. in supplementary material). In fact, for very high hand contamination rates the effect of hand washing on the probability of infection is small and fixed-time hand washing may become more efficient (reduction in probability of infection when compared to similar average number of hand washing events per hour) and event-prompted hand washing with short delays may become infeasible. For low hand contamination rates, the conclusions from our main analysis remain unchanged.
6. Authors choose a cumulative probability of infection of 10% over a period of 12h based on attack rates for Influenza from [10][11][12][13]. Are these quantified rates for fomite transmission? or in general regardless of the route? If these are general, then 10% in 12h might be a significant overestimate of infection risk via the fomite route.
Response: The secondary attack rates from the cited studies are in fact not adjusted for the fomite transmission route. Hence, the actual cumulative probability of infection might be lower than what has been reported. Unfortunately, little is known about the relative contribution of transmission routes for respiratory viruses and may vary for different pathogens (as discussed in the Discussion section, page 10ff.). As secondary attack rates also varied across studies and across age groups (values range from 8% to 50%), we performed additional sensitivity analyses with respect to this parameter. We have now added sensitivity analyses for the cumulative probability of infection of 0.01%, 1%, and 5% to the supplementary material (additional to the ones for 30% and 50%). Together with the main analyses, we now cover a broad spectrum of values and show in the supplementary material (page 18ff.) that our conclusions remain unchanged when this parameter is varied (and all others are fixed). 7. Some sensitivity analysis for is needed, I believe, unless I have missed it. The face-touching λ rate can vary across individuals, ages (e.g. kids vs adults), but also depending on environmental conditions (public vs private spaces; facemasks etc). This parameter might have a significant impact on some of the exposure predictions, where hand hygiene might not be effective when is large, regardless of the handwashing frequency strategy. showed that this rate depends also on the activity with generally higher contact rates for eating vs non-eating activities as well as the area of the face (lower contact frequencies for mouth, eyes, and nose vs all other face areas). Informed by these additional studies, we added sensitivity analyses for of 5, 20 and 50 times per hour to the supplementary λ material (page 21ff.). Note that in our main analysis we used =10 based on the study of λ Kwok et al and adjusted for the face touches that included mucosal areas as face touching events might not necessarily directly translate to infection events. Our sensitivity analyses showed that the main conclusions did not change if was varied but all other parameters λ remained fixed.

Minor comments:
8. For consistency, authors might want to use the same symbol (X) for transmission blocked in Response: Thank you for this suggestion. We have changed the square to the symbol X in 9. Authors state "We assumed that when hand washing is performed... P(ti) is reduced to zero".
It is probably a sensible assumption given high efficacy of hand washing, but authors might want to support this statement with some references/arguments. Is this implicitly assuming not only that the sanitizer has large efficacy, but also that the hand-washing is performed correctly?
Response: We have added a reference by Grayson et al (2009) andLarson et al (2012) to support this assumption. This indeed implicitly assumes a high efficacy of hand washing itself but also that it is performed correctly. We agree that this assumption might be optimistic and our results, therefore, show the maximum effect of hand washing on the probability of infection. We have added this limitation to the Discussion section (page 10): "Fourthly, we assumed that hand washing reduced the probability of persistence on hands to zero and hence that hand washing is maximally efficacious in removing the virus from hands.

Although a high efficacy of hand hygiene with soap and water and alcohol-based hand rubs
has been demonstrated [28,29], its real world effectiveness depends on how well it is practised by the individual. Our model, therefore, demonstrates the maximal effect of the two different hand washing schemes in this regard, but since we made this assumption for both, we do not expect it to impact our overall qualitative conclusions." 10. Computed instead of cmputed in Figure S4 caption.
Response: We have corrected this spelling mistake.
11. The Github link in the Data Accessibility statement is broken, probably because of the index h. In this repository, the authors might want to add some extra descriptions of the codes in the README file.
Response: We had changed the name of the Github repository to https://github.com/tm-pham/handhygiene_modelling but the link in the manuscript was not updated. We have corrected this mistake and added some descriptions of the code in the README file.
12. Please explain how the cumulative probability of infection expression in line 55, Page 12, is obtained.
Response: We added the following text to further explain this expression on page 14: "The cumulative probability of infection over the time period T is given by As with all models a key challenge is the development of appropriate parameters to include in the model, The key insight is that if pathogens survive on the hand for only short periods then more frequent handwashing is needed to reduce risk compared to pathogens with longer survival on hands. This finding is highly robust to different values of parameters chosen. 4. The models focus on acquisition rather than spread of infection via the hand hygiene route.
When the event in question is, for example blowing or wiping one's nose (catch it, kill it, bin it) then there is a clear event of interest. It would be interesting to model this or at least highlight in the discussion that the strategies discussed do not take account of onward transmission from an infected subject.
Response: We agree with the reviewer that it would be interesting to explore onward transmission from an infected individual and its interdependence with hand washing strategies. However, this would require a full transmission model and is out of the scope of this study. Here, we focus on the effect of hand washing on the individual's risk of infection rather than its effect on a population level. We have added this limitation to the Discussion section of our manuscript (page 10): "We have evaluated the effectiveness of hand washing on an individual's risk of infection mediated by the fomite-hand contamination route and did not take onward transmission into account." 5. The other important point that is perhaps underemphasised is that reducing the hand contamination rate is also effective in reducing risk. The authors mention the public health strategy of cleaning surfaces but do not mention reduction in surface touching as a strategy to reduce transmission.
Response: We agree with the reviewer that the reduction of the hand contamination rate represents an alternative strategy to reduce fomite-mediated transmission. We have added this point in the respective paragraph in our discussion: "Furthermore, in the second case, where hands become contaminated very frequently, a substantial reduction in the probability of infection is unlikely to be attained unless hand washing frequency is increased drastically, i.e., every one to five minutes. Because hand washing at such a high rate is not practical (neither for fixed-time nor event-prompted hand washing), the recommendation in this scenario is to regularly clean the environment (such as surfaces), to reduce the rate of surface touching (if possible), and/or isolate infected individuals to reduce hand contamination events." In addition, we have explored the impact of hand washing on the probability of infection when both the half-life of viral persistence as well as the hand contamination rate is varied in our supplementary material (page 32ff.). There we show that for long time intervals between hand washing events or long delays between hand washing and hand contamination events, hand washing is only effective with relatively low hand contamination events. This emphasises that reducing the hand contamination rate may be an effective alternative intervention (see page 34 in supplementary material).
Overall, I see this a very useful paper and my comments are aimed at allowing the authors to either refine their analysis or nuance their discussion, rather than seeing them as undermining their main conclusions.