Infection Fatality Rate – A Critical Missing Piece for Managing Covid-19

coronavirusby Rich Condit

Rich Condit is a virologist and emeritus Professor, University of Florida, Gainesville and a host on This Week in Virology.

Modeling done by the Institute for Health Metrics and Evaluation (IHME) at the University of Washington suggests that if stringent social distancing measures are kept in place, the “first wave” of covid-19 disease in the US may subside by mid-June, with a total accumulation of ~93,000 deaths.  The IHME states that: “By end the of the first wave of the epidemic, an estimated 97% of the population of the United States will still be susceptible to the disease, so avoiding reintroduction of COVID-19 through mass screening, contact tracing, and quarantine will be essential to avoid a second wave.

I find it more than a little unnerving that after a grueling “first wave” of disease, a mere 3% of the population may be infected and thus be (at least relatively) immune to re-infection.  I seriously doubt if after the first wave the virus will be “gone”, rather just beaten back but nevertheless broadly if thinly distributed throughout the population and poised for a comeback when social distancing restrictions are lifted.  If at that time, 97% of the population remains uninfected, then we are destined for a very significant second wave of infection.

But is that 3% infection ratio accurate, and if not, what is the real number?  To understand this, we need to understand the difference between case fatality rate (CFR) and infection fatality rate (IFR).  CFR is the ratio of the number of deaths divided by the number of confirmed (preferably by nucleic acid testing) cases of disease.  IFR is the ratio of deaths divided by the number of actual infections with SARS-CoV-2.  Because nucleic acid  testing is limited and currently available primarily to people with significant indications of and risk factors for covid-19 disease, and because a large number of infections with SARS-CoV-2 result in mild or even asymptomatic disease, the IFR is likely to be significantly lower than the CFR.  The Centre for Evidence-Based Medicine (CEBM) at the University of Oxford currently estimates the CFR globally at 0.51%, with all the caveats pertaining thereto.  CEBM estimates the IFR at 0.1% to 0.26%, with even more caveats pertaining thereto.

Crunching some numbers, if IHME estimates 3% of the US population (~330M) resistant to the disease after the first wave, that implies 9.9M infections.  The IHME prediction of ~93,000 deaths implies an IFR of 0.9%, close to the commonly estimated 1% CFR globally.  Note that IHME is really only projecting deaths based on real data, and therefore should not be faulted for choosing a conservative estimate of infection rate (~1%), especially given that the IFR is, at this point in time, a particularly slippery number.  However, if, just for argument’s sake, the IFR is really as low as 0.1%, then it follows that after the first wave of disease, as much as 30% of the US population will have been infected and thus resistant to reinfection.  It seems to me that this would give the virus considerably less fertile ground to grow on and significantly dampen the impact of a second wave.

This is not just an intellectual exercise.  Understanding the true infection fatality rate has major consequences for planning control measures following our first encounter with SARS-CoV-2.

We can and should obtain a direct measure of the SARS-CoV-2 IFR to help plan for the future of the pandemic.  Simple and relatively inexpensive kits have become available which test for antibodies to SARS-CoV-2 in a drop of blood in a few minutes.  These kits adapt existing technology developed for other pathogens to SARS-CoV-2.  The kits allow for “point-of-care” testing, that is, they can be used anywhere and results are obtained in real time, and they require only minimal instruction for use. (An example of how such a kit works can be found here.  The FDA has given Emergency Use Authorization to Cellex, Inc. for use of their kit.  Dozens of other companies and laboratories are developing similar tests.)  The kits test for both IgM and IgG antibodies specific for SARS-CoV-2 in the blood.  IgM is the first antibody to appear after an infection, usually within about one to two weeks.  Within about three weeks IgM disappears and is gradually replaced with IgG.  Therefore, the kit can determine whether a test subject has ever been infected with SARS-CoV-2, and if so, whether that infection was recent.

If I were king, I would distribute SARS-CoV-2 test kits free of charge to pharmacies nation-wide, and consider setting up testing stations at commonly frequented locations such as schools, sporting events, polling stations, and yes, even bars.  I would also establish a centralized, web based database for collecting and collating the test results. The data collected could be used to determine exactly the IFR in the US, along with a geographic distribution and possibly other demographic information.

We were asleep at the switch for the first wave.  Let’s wake up and catch the second wave.

http://covid19.healthdata.org/

https://www.cebm.net/covid-19/global-covid-19-case-fatality-rates/

https://www.newscientist.com/article/2239497-why-we-still-dont-know-what-the-death-rate-is-for-covid-19/

Comments on this entry are closed.

  • Virus Watcher 6 April 2020, 7:32 am

    “CEBM estimates the IFR at 0.1% to 0.26%”

    This is inconsistent with cruise ship studies that estimated IFR in 0.5-1.5% CI

    Its also inconsistent with experiences in small Italian towns that showed herd immunity (70% of blood dnors have antibodies) only after 1.5% of the population died (many still in ICU)

    It is also inconsistent with the WHO assessment that there is *not* a very large proportion of very mild infections.

    “Herd Immunity by deliberate infection” would be worth a thought if IFR below 0.1% (Think Swine Flu pandemic) but it is out of the question for a IFR>0.5% pathogen unless you are comfortable with millions of deaths.
    It is also inhumane and irresponsible to deliberately expose humans to a virus that we know little about its potential for lasting lung damage even in younger age groups.

    The virus must *first* be eardicated locally and temporally in as much communities as possible. Then can be controlled much better in a localized way.

    Thats the way China or South Korea managed the crisis *sucessfully* so far.

  • Jane crawford 7 April 2020, 3:41 pm

    Being a person with a chronic viral desease whom takes anti viral drugs everyday I would recommend irradiation of any new virus which has multiple infection points that can hook on to dogs ears cars noses live in feaces infect monkeys and rats

  • Virus Watcher 2 8 April 2020, 9:00 am

    A few rebuttals to “Virus Watcher”

    If you actually read Dr. Condit’s links, you’ll quickly find your arguments addressed.

    – Cruise ships A) are a very small sample size and B) on average have more elderly than the general population. When we’ve done more random population sampling and adjusted for age & gender, it suggests lower IFR in the general population.

    – You don’t cite evidence for “small Italian towns with high IFR” and needing 70% infection to achieve heard immunity. Same considerations: Sample Size & Demographic adjustments. Everything I’ve found about Italian towns (ex. the town of Vo) suggests IFR < 1% for the general population.

    Regarding the WHO, they ignored this whole thing and trusted China for a month and they still don't acknowledge aerosol transmission. I'm not exactly looking to them for guidance.

    Finally, Dr. Condit is not suggesting "inhumane and irresponsible" exposition of the population to a virus with a high IFR.

    He IS suggesting that if the IFR is low enough (for the sake of argument, let's use your arbitrary number of 0.1%), he is suggesting that relaxing mitigation and containment measures might be reasonable.

  • Martin Lindh 8 April 2020, 4:51 pm

    Germany is currently doing antibody testing. It would be great if you could talk about the details of their testing in the next TWIV. They are aiming to do 100 000 antibody tests in april.

    https://www.spiegel.de/wissenschaft/medizin/coronavirus-muenchen-testet-3000-menschen-auf-antikoerper-a-ca87a994-237a-4f7b-8050-9162515ddfef

  • David Vanstone 9 April 2020, 11:46 am

    Dear Doctor,

    Concerning testing with the kits: Why not have teams go out to randomly selected points, starting immediately, and collect samples. 2,000 samples could be collected within a few days. Then we’d have on-going, up-to-date and reasonably certain stats. That’s what I would do if king!

    Dave

  • David Vanstone 9 April 2020, 12:44 pm

    Also, if the first wave infects 30%, then a second or maybe a third wave would have to occur before we reach herd immunity. Is there any way of predicting a herd immunity. If it’s 70%, wouldn’t take 3 “good” waves to reach herd immunity?

  • Steve 9 April 2020, 11:42 pm

    Unfortunately I think that your testing strategy is what will indeed have to occur to ascertain a reasonably accurate IFR reasonably quickly as the first wave subsides. I say unfortunately because I don’t think it’s actually going to happen. That said, with the heavy emphasis in many areas, both in the United States and globally as well, on diagnosing, treating and isolating (to various degrees) the clearly symptomatic I think mild cases are being missed in high volume. These individuals are not being sought out and the last thing many of them, who have been ordered to shelter in place anyway, want to do is go out and risk exposure to something that they may not already have to get tested in response to symptoms they self evaluate as mild.

    The above speculation not withstanding, based on the data at the link below, not accounting for duplicate and unusable tests, which I have no way to account for, about 19.9% of tests run are positive. If one were to do some rough and simplistic calculations and were to extrapolate that to a population of 331,000,000 that would be as many as 65,869,000 people infected or soon to be infected. Using the 93000 fatalities that would be an IFR of 0.14% which puts it within the range estimated by CEBM, albeit at the low end.

    https://www.worldometers.info/coronavirus/country/us/

  • Steve 9 April 2020, 11:50 pm

    Virus Watcher wrote: “Herd Immunity by deliberate infection” would be worth a thought if IFR below 0.1% (Think Swine Flu pandemic) but it is out of the question for a IFR>0.5% pathogen unless you are comfortable with millions of deaths.
    It is also inhumane and irresponsible to deliberately expose humans to a virus that we know little about its potential for lasting lung damage even in younger age groups.”

    At some point it’s no longer going to be a choice. An economic tipping point is going to be reached where the current strategy, at least to the degree currently implemented, is simply no longer tenable. Exposure, or at least risk of exposure on the part of significantly greater than current numbers is going to be unavoidable unless the cure is going to on net be worse than the disease. That said, to the degree we have a choice, we do have Sweden already taking that approach, so, since we have few other choices anyway, let’s watch them and see what happens, and if no better data is available factor their results into any chosen strategy.

  • Steve 10 April 2020, 12:00 am

    “Concerning testing with the kits: Why not have teams go out to randomly selected points, starting immediately, and collect samples. 2,000 samples could be collected within a few days. Then we’d have on-going, up-to-date and reasonably certain stats. That’s what I would do if king!”

    We need to do way more testing than that and what would the criteria be? Completely random? Completely random in known high infection rate areas? At any rate according to the data linked below we’ve already run over 2.3 million tests and even at that everybody, including me above, is still really just making quasi educated guessing. Next time we are going to have to err on the side of caution and close the borders to all non citizen entrants and citizens are going to need to quarantine. The economic damage from that will exist but it will be far far less than it is now.

    https://www.worldometers.info/coronavirus/country/us/

  • Benjamin Blumberg, PhD 10 April 2020, 5:25 am

    The probable success of the hydroxychloroquine+azithromycin combo, plus a novel “Patch Vaccine” being readied at U/Pittsburgh consisting of 400 microneedles made with synthetic antigenic SARS-CoV-2 peptides embedded on a band-aid, means that we may have at hand both a cure for COVID-19 and a vaccine for the susceptible herd within 6 months of the start of the epidemic. If everything works, and if the CDC and the FDA approve, this will have made flattening the curve with the risk of extending the epidemic and the economic dislocation worthwhile. IF.

  • Hojo 10 April 2020, 1:31 pm

    Would you plesae comment on the study from Shanghai where they collected multiple plasma samples from Covid-19 patients being discharged from hospital and found that 30% generated very low antibodies and 6% generated none? Thank you!

    Here is the study
    https://www.medrxiv.org/content/10.1101/2020.03.30.20047365v1.full.pdf

  • Martin Lindh 10 April 2020, 5:49 pm

    First preliminary results from the german serology test. CFR = 0.37.

    https://www.land.nrw/sites/default/files/asset/document/zwischenergebnis_covid19_case_study_gangelt_0.pdf

    Since there are still people who are sick in the town I would guess IFR will be higher.

  • Jim Welch 10 April 2020, 6:58 pm

    Though the CFR is commonly estimated to ~1%, 1% seems to me too low for the US data. CEBM’s latest projection is 61,545 deaths. The current number of US cases is around 500,000. If the CFR is 1% we should see 61,545/0.01 ~ 6 million cases in the next month or so. That would seem to be unlikely given the current trend in cases. Maybe 3-5% is more realistic?

  • Jim Welch 10 April 2020, 7:01 pm

    CEBM in my previous comment should have been IHME …

  • Ken 10 April 2020, 11:05 pm

    The best way to be done with this is to protect the old, the frail, and otherwise compromised people while the virus is allowed to run its course through the population. The result would be herd immunity and an end to these nightmarish, unsustainable, and draconian measures that are not only insane, but are dragging this on way longer than need be. You know, what we do with that pandemic we have every year — the seasonal flu.

  • Rob 11 April 2020, 2:51 am

    The IFR (Infection Fatality Rate) is best measured with the South Korean data. After all, South Korea tested EVERYONE who could have been in contact with any COVID-19 patient. And in doing so, they got the pandemic under control. So they likely tested EVERYONE who got infected. When tested against the number of infected people that died, they obtain 2.0 %. So that’s the best estimate of the COVID-19 IFR and a metric for the rest of the world :
    https://www.cdc.go.kr/board/board.es?mid=a30402000000&bid=0030

  • Lidya Garcia 11 April 2020, 4:04 pm

    La tasa de mortalidad por este virus se ha estado aumentado veloz en todos los países, es muy importante que todos estemos en confinamiento y tomar nuestras debidas precauciones para así no poder contraer este virus, y mas las personas con enfermedades graves o con una edad muy avanzada. Seamos héroes quedándonos en casa.

  • Wlodek Zski 11 April 2020, 10:50 pm

    The incidence of Covid-19 in S. Korea is so far around 0.0002 (10,500 per 50 M of population). The incidence of possible secondary re-infection in the population of the survivors of Covid-19 is minimum 0.01 (91 reinfected per 10,500 cases of Covid (Minus the deaths To be more exact)). Does it mean that a susceptibility to Covid-19 increases about 50-fold AFTER one had survived the first infection (0.01/0.0002). What do you think?

  • Rob 11 April 2020, 11:20 pm

    Wlodek, you are confused. The South Korea data does not say anything about people that never got infected.

    South Korea became one of the few countries that got to control the spread of this virus. They tested everybody who could have possibly be infected by each known COVID-19 patient, and isolated them if THEY tested positive. So they likely tested ALL infected people, and got the virus under control like that in 4 weeks. As a result of this, their data also shows the best estimate for the IFR (Infection Fatality Rate) of this virus, which is 2.0 %.

    That means if you catch this virus, you have a 2% chance of dying (if your health-care system is working).

  • Steve 11 April 2020, 11:26 pm

    “Though the CFR is commonly estimated to ~1%, 1% seems to me too low for the US data. CEBM’s latest projection is 61,545 deaths. The current number of US cases is around 500,000. If the CFR is 1% we should see 61,545/0.01 ~ 6 million cases in the next month or so. That would seem to be unlikely given the current trend in cases. Maybe 3-5% is more realistic?”

    That 500,000 doesn’t really mean much. It’s really just a baseline for absolute minimum infected. That is just identified cases that have actually been tested and confirmed. Assuming no duplicate tests, which there of course are, only about 0.8% of the population has actually been tested. If they are picking up a couple percent of the total infections that can mimic allergies or a common cold in mild cases they are doing good. Consider that just under 20% are testing positive if you were to apply this to the population as a whole of 330,000,000 you’d get as many as 66,000,000 positive. It may not really be that high due to the fact that testing is not random but based on symptoms or believed exposure but still that gives you some idea. It’s WAY more than 500,000.

    https://www.worldometers.info/coronavirus/country/us/

  • Steve 11 April 2020, 11:38 pm

    Rob wrote: “South Korea became one of the few countries that got to control the spread of this virus. They tested everybody who could have possibly be infected by each known COVID-19 patient, and isolated them if THEY tested positive. So they likely tested ALL infected people, and got the virus under control like that in 4 weeks. As a result of this, their data also shows the best estimate for the IFR (Infection Fatality Rate) of this virus, which is 2.0 %.

    That means if you catch this virus, you have a 2% chance of dying (if your health-care system is working).”

    Unlikely. CEBM gives a CFR of 1.96% for South Korea. For the IFR and CFR to be the same it would mean that they had either identified all the infections or missed as many deaths as they missed infections. Neither possibility is at all likely. It is simply not possible to identify and test every person who had contact with a known infected person in a country of 51.6 million people, and even then what about the unknown infected persons? Missing a large number of COVID-19 deaths is even less likely with surveillance systems on high alert for anyone who died after even showing COVID-19 symptoms.

    https://www.cebm.net/covid-19/global-covid-19-case-fatality-rates/

  • Rob 12 April 2020, 3:38 am

    Steve said : ” CEBM gives a CFR of 1.96% for South Korea. For the IFR and CFR to be the same it would mean that they had either identified all the infections or missed as many deaths as they missed infections. ”

    Well, they did not miss many. If they did miss a lot of infected people, COVID-19 would have re-emerged in South Korea. So the IFR number is unlikely to be much below 2.0 %.

  • Eric Miller 13 April 2020, 5:45 pm

    Rob,

    South Korea is still confirming more than 30 new cases per day. They obviously have not caught every infected person, and certainly did not at the beginning and middle of their local epidemic.

  • ZannBee 16 April 2020, 2:01 am

    It may be difficult to accurately estimate CFC and IFC while there are so many unresolved cases. At this point in time, the only accurate measure is the CLOSED case fatality rate.

  • TB 16 April 2020, 3:44 am

    We can get some upper bound estimates on US infection % based on the attack rate in certain areas. NYC has the highest attack rate. It has a population of 8.4M. If we say herd immunity is reached at 80% of the population, then the upper bound on infected people in NYC is around 6.7M. NYC currently makes up 40% of US covid deaths, so the upper bound of infected people in the US maps to 16.75M or about 5% of the population.

    I’d say the 3% estimate is more plausible than the 30%.

  • TB 16 April 2020, 4:06 am

    “about 19.9% of tests run are positive. If one were to do some rough and simplistic calculations and were to extrapolate that to a population of 331,000,000 that would be as many as 65,869,000 people infected or soon to be infected.”

    There’s significant sampling bias in the testing, since in most places people only people with symptoms will be eligible to get tested.

  • Dan Ward 18 April 2020, 6:31 am

    Very interesting and useful blog, and useful comment too.

    I would like to reply and add to TB’s comment from 16 April in particular, mentioning sampling bias in COVID-19 testing.

    I have recently published this statistical report analysing this very issue:

    https://www.researchgate.net/publication/340539075_Sampling_Bias_Explaining_Wide_Variations_in_COVID-19_Case_Fatality_Rates

    This report used ratios between tests and deaths to estimate the extent of sampling bias, and found that countries with less sampling bias (higher tests/deaths ratios) have significantly lower CFR estimates. This implies that the underlying IFR will be at the lower end of the current CFR range.

  • Virus Watcher 3 23 April 2020, 1:07 pm

    The critical piece here that all of you are missing is the percentage of people who are currently infected by any virus in the USA. On any given day somewhere between 2-11 percent of people are infected with a virus. We also know that about 13 percent of people who are tested for the virus are positive. We can use these pieces of data to give a decent estimate of the total number of infections in the US. Let’s assume 7 percent have a cold or flu or covid. Let’s also assume all of them go for testing. We know if they get tested then 13 percent will be positive for COVID. Multiply 330 mill x .07 = 23.1 mill with a virus. Multiply 23.1 mil x .13 =3.003 mill people with covid in the US. Divide 41,000 / 3.003 mill = .0136 or 1.36% CFR. Keep in mind that this does not take into consideration the number of asymptomatic cases. Recall about 30 percent of cases will be asymptomatic. If we divide the total number of our estimated symptomatic cases by .3 or 30 percent. 3.003 mill / .3 = 10.01 mill infected. The 10.01 mill represent both symptomatic and asymptomatic. If we then divide total dead 41,000/ 10.01 mill = .0040 or .4% CFR. This is a reasonable assumption based on the data at hand. Remember this assumes that EVERYONE with symptoms in this country went for a test. We know not everyone will test so the actual CFR must be much lower. Perhaps 10 fold lower? So assume that 1 in 10 people with any symptoms of any virus would go for a test. If we assume that then we can estimate the current number of infections in this country at 9 percent. If so that would give us a CFR of 0.04%. Should we really have a country closed for this? Going further if we assume that 70 percent of the population will become infected over the next two years then a total of 132,000 people will die. This suggest 65,000 a year. Sounds like the flu. So we are approximately 1/3 of the way thru our pandemic. Expect 2 more waves. These numbers sound reasonable and I propose that the best thing to do for the country would be to open so the millions who are suffering from depression, unemployment, and hunger don’t die as well.

  • Jospeh Steak 24 April 2020, 11:11 am

    A long follow up to Dan Ward’s insightful comment/link.

    An assessment of the most recent CFR data from different countries is useful to inform what should be the upper end of the IFR for COVID-19. The CFRs in Singapore and Qatar are 0.11% and 0.13%, respectively, as of yesterday. As we know, the numerator (COVID-19 deaths) is far more accurate than the denominator (COVID-19 infections). We also know that the denominator is far more likely to increase relative to the numerator, as more of the population is tested, and we get a better estimate of the true percentage of people infected. This indicates that the CFR will most likely keep decreasing as these countries test more of their populations. Therefore, the estimates from Singapore and Qatar represent the upper end of the IFR (0.11 – 0.13 %), based on the information we currently have. There are various other factors that could influence these estimates up or down; for example the quality of healthcare, the fitness of the population, biased reporting, etc. However, these would have a minimal effect relative to the far greater possibility that the CFR estimates will go down as these countries test more of the population; and get closer to the true IFR with time. Another important point is that these estimates come from two very different countries with very different demographics, cultures, and approaches to healthcare.

  • Jane Wilson 25 April 2020, 10:28 am

    From cohort IFRs in the Imperial College March ~26 report, one can tease out this fact. Ninety percent of deaths will be those over 60 and 70% to 80% will be over 70 years of age. Never mind the appalling ethics of destroying the economy and the lives of the young to postpone infection of us seniors, there is a statistical problem. When almost all the deaths are in a cohort with a very high natural mortality rate, there is significant double counting. Many had a high probability of dying in the same period from something else. Because of this, it never made sense to take an average IFR, apply it to a population, and then conclude that the result represented incremental deaths. The suicidal mitigation lockdowns do nothing except slightly postpone infection, worse still, possibly postpone it to the fall, to coincide with the flu season. The evidence is growing that many of us are carrying the virus. Do we push the world into complete economic collapse, into crime, institutional failure, because we are waiting for more data? The predicted great die off did not occur as expected. WHO is trying to credit mitigation measures. This is not correct. The great die off did not occur because the original projections were flawed. But they’d rather save face than admit this.

  • Kurt E. Walberg 28 April 2020, 6:50 pm

    I am hoping to understand the concepts of IFR and CFR a little more fully, relating them to the numbers of cases and fatalities that are being reported. Having spent some time looking at CEBM’s recently updated Covid site, https://www.cebm.net/covid-19/global-covid-19-case-fatality-rates/, I realized that IFR and CFR are not based on raw numbers alone, but are the result of a number of operations which effectively reduce the raw numbers by orders of magnitude.

    My layman’s translation of this is that for both IFR and CFR, CEBM is guessing (or estimating) that there are many more cases and many more infections than show up in the raw data. So for New York State, which reported on 4/27/20 (via https://www.livescience.com/coronavirus-updates-united-states.html) 298,004 cases with 22,623 deaths (yielding a raw, or crude, fatality rate (not to be confused with Case Fatality Rate) of roughly 7.5%), CEBM figures that there are actually more than 4,500,000 cases in New York State (yielding a CaseFR of 0.5% (22,623/4,500,000)), and that there are actually more than 19 million infections in New York State (22,623/19,000,000), which means almost the entire state is infected.

    That can’t be right and I would sure appreciate some clarification.

    Thanks in advance.

  • Barry Rodgers 29 April 2020, 2:20 pm

    Frustrated. Simply, the CFR is based on actual confirmed cases whereas the IFR is based on estimated cases.

  • Kurt E. Walberg 30 April 2020, 12:37 pm

    Thanks for the clarification. You’re saying CaseFR is equivalent to Crude Fatality Rate, which makes sense. Is my reverse calculating sound? If so, the CEBM’s IFR estimate of 0.1% can’t be reconciled with the data on the ground in NYS, where the per capita death rate statewide currently shows as just under 0.12%, meaning we are approaching more estimated infections in NYS than there are people. Still trying to make sense out of this as well as of the whole situation.

  • Kurt E. Walberg 30 April 2020, 4:29 pm

    My reverse calculating question was dumb, so I take it back. Its deaths per infections, and we only know the current reported deaths and reported infections, we don’t know the final numbers. It seems like IFR is what people who know guess that rate will eventually be.

  • SPK 4 May 2020, 3:36 am

    Not a virologist, but a numbers guy, and I agree with author’s take. Per COVID tracking project data, after about 2 million tests, on average about 19% test positive. Implies 60 million infected out of population of 330 million. IFR .1%. Still results in a lot of loss if second or third wave are another 19% infected.