key: cord-325546-bgdr25z1
authors: Pham, T. M.; Mo, Y.; Cooper, B.
title: The Potential Impact of Intensified Community Hand Hygiene Interventions on Respiratory tract Infections: A Modelling Study
date: 2020-05-27
journal: nan
DOI: 10.1101/2020.05.26.20113464
sha: 
doc_id: 325546
cord_uid: bgdr25z1

Increased hand hygiene amongst the general public has been widely promoted as one of the most important non-pharmaceutical interventions for reducing transmission during the ongoing COVID-19 pandemic and is likely to continue to play a key role in long-term efforts to suppress transmission before a vaccine can be deployed. For other respiratory tract infections community hand hygiene interventions are supported by evidence from randomised trials, but information on how effectiveness in reducing transmission scales with achieved changes in hand hygiene behaviour is lacking. This information is of critical importance when considering the potential value of substantially enhancing community hand hygiene frequency to help suppress COVID-19. Here, we developed a simple model-based framework for understanding the key determinants of the effectiveness of changes in hand hygiene behaviour in reducing transmission and use it to explore the potential impact of interventions aimed at achieving large-scale population-wide changes in hand hygiene behaviour. Our analyses show that the effect of hand hygiene is highly dependent on the duration of viral persistence on hands and that hand washing needs to be performed very frequently or immediately after hand contamination events in order to substantially reduce the probability of infection. Hand washing at a lower frequency, such as every 30 minutes or with a delay of 15 minutes after contamination events, may be adequate to reduce the probability of infection when viral survival on hands is longer, such as when hands are contaminated with mucus. Immediate hand washing after contamination is more effective than hand washing at fixed-time intervals even when the total number of hand washing events is similar. This event-prompted hand washing strategy is consistently more effective than fixed-time strategy regardless of hand contamination rates and should be highlighted in hand hygiene campaigns.

Family members: Wash hands with soap and water regularly, especially after coughing or sneezing, before, during and after food preparation, before eating, after toilet use, before and after caring for ill persons and when hands are visibly dirty.

Patients: Wash hands immediately and thoroughly after coughing, sneezing, removing face mask. Stay in a separate room from other family members.

[11]

ECDC Rigorous hand-washing with soap and water >20 seconds, or alcohol-based solutions, gels or tissues is recommended in all community settings in all possible scenarios, especially after coughing or sneezing, disposal of used tissues.

Family members: Wash hands frequently, especially after contact with the patient or with any surface frequently touched by the patient, e.g., before and after preparing food, before eating, after using the toilet, removing face mask/ gloves, handling waste.

Patients: Wash hands immediately and thoroughly after coughing, sneezing, removing face mask.

[12, 13] PHE Washing hands more often, especially after arriving at work or home, after blowing nose, coughing or sneezing, before eating or handling food.

Wash hands frequently with soap and water for 20 seconds or using hand sanitiser, especially after coughing/sneezing and disposal of used tissue.

[14, 15] CDC Wash hands often with soap and water for >20 seconds especially after being in a public place, or after blowing nose, coughing, or sneezing.

Clean frequently touched surfaces and objects daily (e.g., tables, countertops, light switches, doorknobs, and cabinet handles) using a regular household detergent and water. Wash hands often with soap and water for >20 seconds or hand sanitizer, especially after going to the bathroom, before eating, and after blowing nose, coughing, or sneezing. Always wash your hands with soap and water if your hands are visibly dirty.

[16, 17] An immediate consequence of this conceptualisation is that the time interval between the hands becoming 78 contaminated and making infectious contact with the host's mucosa can have a critical impact on how 79 effective a given frequency of hand washing will be at interrupting transmission ( Figure 2 ). If this time 80 interval is relatively long in the absence of hand hygiene, regular effective hand hygiene will have a 81 high chance of blocking potential transmission events (red diamonds in Figure 2 panel A). In contrast, 82 if this time interval is short much more frequent hand hygiene will be needed to block an appreciable 4 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 27, 2020. . https://doi.org/10.1101/2020.05.26.20113464 doi: medRxiv preprint 1. uniformly at fixed time intervals (fixed-time hand washing), or 89 2. with a delay after hand contamination events (event-prompted hand washing). 

Hands of susceptible individuals are assumed to become contaminated at random. These contamination events are assumed to occur independently of each other, and to follow a Poisson distribution with a mean of λ c events per hour. The probability of the virus persisting on hands at time t after contamination,

, is assumed to decay exponentially with a half-life of T 1/2 . This is consistent with experimental data for influenza A (see [19] ). Individuals touch their face at random leading to potential infection events that are assumed to occur independently of each other, and follow a Poisson distribution with a mean of λ f events per hour. The probability that a single face-touching contact with contaminated hands actually leads to transmission is .

Assume the face-touching events occur at times t 1 , . . . , t F during the given time period T . Then the cumulative probability of infection over the time period T is given by:

5 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted May 27, 2020. When available, parameter estimates were obtained from the literature. Otherwise, we performed sen-98 sitivity analyses where parameters were varied within plausible ranges (see Table 2 ). In the fixed-time 99 hand washing scheme, we varied time intervals between hand washing to be 5 min to 6 hours. For event-100 prompted hand washing, the delay of hand washing after hand contamination events was varied from 1 101 min to 6 hours.

There is little published data on the rate of hand contamination events susceptible individuals are exposed 104 to when in contact with infected individuals who are shedding respiratory viruses, and none specific to 105 SARS-CoV-2. In a direct observation study conducted by Zhang et al [20] , surface touching behaviour 106 in a graduate student office was recorded. Approximately 112 surface touches per hour were registered.

Another study by Boone et al [21] found that the influenza virus was detected on 53% of commonly 108 touched surfaces in homes with infected children. Informed by these values, we took 60 events per hour 109 as the upper bound for the rate of hand contamination events λ c . We chose 1 event per hour as the lower 110 bound. In our main analyses, we used a rate of 4 hand contamination events per hour.

To date, it is not known how long SARS-CoV-2 can persist on human fingers. In [19] , the survival 113 of influenza A on human fingers was experimentally investigated. We fitted exponential decay curves 114 to these results in order to determine the half-life of probability of persistence of H3N2 for two viral 115 volumes of 2 µL and 30 µL (see Table 2 and supplementary material). In addition, we vary the half-life 116 of probability of persistence from 1 to 60 min in our analysis.

Model outcomes 118 The model output is the cumulative probability of a susceptible person becoming infected in twelve hours 119 and we will refer to it subsequently as simply the probability of infection. We investigated the impact 120 of hand washing on the probability of infection for different hand contamination rates. In addition, 121 we compared the two hand washing schemes (fixed-time vs. event-prompted) to find the optimal hand 122 washing strategy that will lead to the greatest reduction of the probability of infection. The model CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted May 27, 2020. . respectively), the same hand washing frequency decreases the probability of infection to 6.9 % and to 132 4.6 %, respectively. Consequently, fewer hand washes are necessary to reduce the probability of infection 133 by 50 % for long compared to short durations of viral persistence (see Figure S2 ). This observation can be 134 explained by the fact that the shorter the virus persists on hands, the shorter the intervals between hand 135 contamination and transmission events tend to be (with a higher transmission probability per contact, 136 see Figure S4 ) and therefore, the less likely hand washing is able to interrupt infection events. Figure S3 The second notable finding from the model is that event-prompted hand washing is more effective than 8 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 27, 2020. . https://doi.org/10.1101/2020.05.26.20113464 doi: medRxiv preprint and 2% when hand washing is performed every 15 min and one minute after hand contamination events, 151 respectively. The differences between the two hand washing schemes are less pronounced if hand washing 152 is performed less frequently or with a longer delay after hand contamination events since the two hand 153 washing schemes become more similar. It follows that delays between hand contamination and hand 154 washing decrease the effect of hand washing on reducing the probability of infection. Another important parameter that affects the effect of hand hygiene is the hand contamination rate.

157 Figure 5 shows the increase in hand hygiene frequency required to half the probability of infection from 158 10% (no hand wash) to 5%. When hand contamination rate is relatively rare, at less than 10 times per . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 

Our study provides new insights into factors that affect the effectiveness of hand hygiene behaviour in 174 reducing the probability of infection. Firstly, we found that the shorter the virus survives on hands, the 175 less effective increasing hand hygiene frequency is in reducing infection. The logic behind this is that 176 when the virus dies off quickly before hand washing is performed, the time intervals between hand con-177 tamination and transmission tend to be shorter and the respective transmission probability per contact 178 tends to be higher for the same cumulative probability of infection. Secondly, contaminated surfaces are 179 crucial for the effect of hand hygiene. The more often hands become contaminated, the more frequently 180 hands need to be washed to reduce infection risk. Lastly, when hands are not constantly contaminated, 181 event-prompted hand washing is more efficient than fixed-time hand washing given the same hand washing 182 frequency. This is because delays in hand washing after contamination of hands in fixed-time compared 183 to event-prompted hand washing tend to be longer, and, during this delay, susceptible hosts may become 184 infected through face-touching. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 27, 2020. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 27, 2020. . https://doi.org/10.1101/2020.05.26.20113464 doi: medRxiv preprint

Hands of susceptible individuals are assumed to get contaminated at random. These contamination events are assumed to occur independently of each other, and follow a Poisson distribution with a mean of λ c events per hour. The probability of the virus to persist on hands at time t after contamination, P (t), is assumed to decay exponentially with a half-life of T 1/2 . This is consistent with experimental data for influenza A (see [19] ). Individuals touch their face at random leading to potential infection events that are assumed to occur independently of each other, and follow a Poisson distribution with a mean of λ f events per hour. The probability that a single face-touching contact with contaminated hands actually leads to transmission is .

Assume the face-touching events occur at times t 1 , . . . , t F during the given time period T . Then the cumulative probability of infection over the time period T is given by:

We assume that when hand washing is performed after the last hand contamination event and before a 312 face-touching event at time t i , the respective probability of virus persistence P (t i ) is reduced to zero.

Probability of viral persistence on contaminated hands 314 The decay of the probability of viral persistence on contaminated hands is modeled as an exponential decay with probability distribution:

where λ d is the decay constant. The probability that virus will die off within time t is given by the integral of the decay distribution function from 0 to t:

The probability that the virus will persist at time t is one minus the probability that it will die off within the same period:

The average survival time (or mean lifetime) is given by:

15 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 27, 2020. 

where λ d is the decay rate. The decaying quantity, n(t), represents the number of fingers with recoverable 316 infectious viral particles and is assumed to have an initial value of n 0 at time zero.

In the experiment with 2 µL inoculum, 18 contaminated fingers from six individuals were tested for the 319 presence of infectious virus at 1, 3, 5, 15 and 30 min after initial contamination. Figure S1 depicts the 320 data and the fitted curve for the 2 µL inoculum. The decay rate was estimated to be λ 

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 27, 2020. . https://doi.org/10.1101/2020.05.26.20113464 doi: medRxiv preprint

The shorter the half-life of virus persistence, the higher the frequency of hand washing necessary in order 328 to prevent 50% of infections (see Figure S2 ). In addition, the time intervals between hand contamination 329 and hand washes have to be shorter in order to prevent 50% of the infections (see Figure S3 . 18 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 27, 2020. . https://doi.org/10.1101/2020.05.26.20113464 doi: medRxiv preprint Figure S4 shows that the shorter the virus persists on hands, the higher the probability of transmission 332 per face-touching contact has to be if the cumulative probability of infection is assumed to be fixed. 

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 27, 2020. . https://doi.org/10.1101/2020.05.26.20113464 doi: medRxiv preprint minutes or the time delay of hand washing after hand contamination events is decreased to one or five 351 minutes. However, due to the the high rate of hand contamination events of every 4 minutes or every 352 minute, respectively, such an uptake seems infeasible. Hence, when susceptible individuals are exposed 353 to continuous contamination, the best strategy would be to wash their hands as frequently as possible, 354 especially after touching potentially contaminated surfaces, and to reduce the rate of contamination by, 355 e.g., cleaning surfaces in their environment or isolating the infectious person. Comparison of number of hand washes 358 Figure S11 shows the average number of hand washes per hour for the two hand washing schemes in 359 the scenario used in the main analysis, i.e. for a hand contamination rate λ c = 4 hour -1 . For a fair 360 comparison between the two hand washing schemes, fixed-time hand washing should be compared to 361 event-prompted hand washing using approximately the same average number of hand washer per hour.

For example, hand washing every fifteen minutes may be compared to event-prompted hand washing one 363 minute after hand contamination.

23 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted May 27, 2020. Hand contamination events are assumed to occur on average 4 times per hour.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted May 27, 2020. . https://doi.org/10.1101/2020.05.26.20113464 doi: medRxiv preprint

Ignaz Semmelweis and the birth of infection control

Hand hygiene and patient care: pursuing the Semmelweis legacy