By Helen Petousis Harris 04/09/2019

Here is a very quick and over simplified explanation about why some of the cases of measles are fully vaccinated (epidemiologists look away!).

The vaccine is not 100% protective. One dose is around 92% and two doses around 98%. Generally, things look like this pie chart, assuming the vaccine is potent and the patients relatively healthy. Most cases occur in the unvaccinated with a few in the vaccinated.


A common pitfall is to look at the number of cases by vaccine status, for example when I looked today at the recent measles report it shows that among 10-19-year-olds 17 cases are fully vaccinated and 96 are not vaccinated. A natural inclination is to go OMG! the vaccine is only 85% protective. The problem with doing this is that most people are vaccinated, you need the denominator.

The correct way to do it is to consider the attack rate among the vaccinated and the attack rate in the unvaccinated. The attack rate is the proportion of cases in the population at risk divided by the number of persons in that population. Of course, I have totally over-simplified things here, if I was going to do this in a robust manner it would be through a cohort study using data-linkage and would adjust for confounders, a bit like we did for our gonorrhea cohort study, which has quite cool methodology. Also, the formula for effectiveness uses odds ratios, but I only wish to illustrate a point here so please bear with my over simplistic explanation.

The formula for vaccine efficacy is (ARU-ARV)/ARU x 100%. ARU means attack rate in unvaccinated and ARV means attack rate in vaccinated.

We need some additional information for our calculation other than the vaccine status of the cases. We need to know how many of our population are vaccinated and how many are unvaccinated. For this 10-19-year age group, based of the number of individuals born in the years 2000-2009 I get around 480,000 vaccinated and around 108,000 unvaccinated (remember this is crude!).

Plugging the information into the basic formula I get around 96%. There are many more unknowns for this age group, for example for 75 cases the vaccine status was unknown. However, if I perform the same exercise for the 5-9-year old’s (where vaccine status is known for all but 4 cases), I get 97%.

Another exercise is to imagine that 98 in every 100 people at the raging party in a downstairs nightclub are vaccinated and measles is invited. The two unvaccinated party goers get measles. Also, two or three of the fully vaccinated people get measles. OMG! half the cases were vaccinated.

Hopefully you can see that if one wants to avoid measles it is much better to be in the vaccinated group! Also, remember that most people are in this vaccinated group. I have likely underestimated this in my assumptions.

0 Responses to “Why are vaccinated people getting measles?”

  • Thanks for this clear summary, Helen, much appreciated.
    Could I ask: what is the incidence rate of people getting measles now, if they actually suffered measles when they were children? (That is the situation for me, in that I had measles when I was about six, many years ago and well before the vaccine (first measles only, and later MMR) became available (in the UK, it happens). That is, I asking what level of immunity is rendered by previous suffering of the disease, given that some people do apparently get recurrences of diseases they have previously had.

    • Hi Duncan
      Generally, measles is one of those diseases that leaves people with life-long immunity. I am not aware of any literature describing people getting the disease more than once. In contrast, i do know that mumps has been known to strike more than once. Some diseases leave you with virtually no immunity against later infections (gonorrhoea!) while others provide a lifetime of protection, measles is one of them.

  • Hello Helen

    Your pie chart excludes those classified as unknown. these could be people who have been vaccinated pre 2004 (NIR) but can’t prove it. They get classified as uknown and make up a large proportion of cases.

    • Hi Peter
      Yes, the unknown’s could all be vaccinated, or all unvaccinated, or anywhere in between. The pie chart is based on what we know about the vaccines effectiveness, and also why I mentioned the younger age group, in whom we pretty much know the status (apart from 4) where my back of envelope was a reasonable high estimate. As mentioned in the blog, this was horribly crude simply to illustrate the importance of the denominator. The reason we know the vaccine effectiveness is via methods more akin to the example I gave and how we know a single dose is around 92% and two doses takes that up higher.

  • Thank you Helen. I read somewhere that the second dose wasn’t a booster but it is just to try and get some of those who didn’t get vaccinated when younger. Can you explain this for us? Is the second dose a booster?

    • Hi Peter
      Yes, this is correct, in that dose two is not considered a booster. Unlike non-live vaccines live vaccines work more like a digital on/off. After one dose there are a small proportion of individuals do not respond with protective immunity (about 8%). The second dose mops up most of these. This dose can be given any time at least 4-weeks after the first dose. In this sense it is all very flexible.

  • Thank you Helen. Do you know how many children have their first dose of MMR at 4 years old?

    • Hi Peter,
      No I don’t. It is reported by milestone on the Ministry NIR website but this will not provide this information. The data is there but requires a specific query that is not part of routine reporting.

  • Hi Helen. That surprises me. Wouldn’t that be important information when dealing with a measles outbreak? Do Milsestone imms rates relate to individual diseases. How would we get the MMR information from? Does IMAC have the immunisation details on individual imms?

    • Hi Peter
      The vaccine status of individuals can be checked (by a registered user of the NIR) and we know how many/the proportion of people under the age of about 13 are vaccinated and with how many doses. We know the proportion of people who have been fully immunised at 6 months, 8 months, 12 months, and two years, and five years. The ministry should also know the coverage by antigen but this is not one of the milestones on the website. Your question re the promotion of people receiving a first dose at four years would need a specific check.

  • Thanks again Helen

    Am I reading you correctly? If, say, 80% of 6 month old have been vaccinated by their milestone date and then 10% get vaccinated at 7 months, then the public figure won’t change to 90% so it will always look as if that group is under vaccinated? Secondly, I have seen no figures in publications showing the percent of children vaccinated with MMR vaccine only the milestone rates which include all vaccines including rotavirus and chickenpox. Have you just said that experts like yourself don’t know the actual vaccination rates for MMR?

    • Ashton,
      Yeah, I agree, about three comments ago! 🙂

  • Ashton & Helen, (not sure I can reply as a nested comment, hence names!)

    “I can smell the salmon and anchovies on his breath…”

    Yes, such “innocent” questions. (Reminds me of someone too!)

  • Regardless of the aroma of someone’s breath, in the interest of transparency from some of our country’s experts, shouldn’t the valid questions be answered to the best of your ability? Otherwise, where do we find ourselves – In a place where facts are kept under wraps, only visible to a select few? Surely everyone should have free access to the stats, or at the very least the data be communicated clearly when requested?

  • Hi Bindy. The issue is not the questions per se. Its the manner and intent of the asking. “Sea-lioning” is a technique used by less scrupulous actors to appear innocent while having a hidden intent.

    Some are good at it, others a bit blatant. Either way, its not usually effective to engage the sea-lion.

    The facts are not under wraps – if you get a relevant qualfication you can problably get access to them. If its published reseach, you can get it regardless of your qualification. As a layperson, I have never understood the push for the democratisation of science, especially medicine. Most people (including me) don’t have the capability or compentence to make informed comment on originating data. Lacking a relevant qualification, I am happy to allow people who trained for 20 or more years to do the heavy lifting on the data while I assess the outputs.

  • Just to add a few points to this:-

    Once it’s clear someone is trolling the wise thing to do is to not engage with them further.

    The point in Helen’s piece is simply that when people talk about how effective a vaccine is, you have to be note the ‘out of’ part: out of those who got measles is not the same as out of those who vaccinated or not.

    There are a number of people online pointing at how out of the people who got measles some weren’t vaccinated, saying that this means the vaccine isn’t very effective.

    Actually it shows that the vaccine is very effective.

    If roughly 9 times more people vaccinated than those that didn’t (and all other things are equal), measles will be 9 times more often trying to infect a vaccinated person than a not vaccinated person, yet still we see (a lot) fewer vaccinated people get the measles than people who aren’t vaccinated. The vaccine is pretty effective, and you’re much better off with it.

    “Surely everyone should have free access to the stats,”

    Hopefully you mean well in this, but, in a nutshell, no, that’s a bad idea!

    This question might help: would everyone like anyone to be able to look up their . I’d guess most people wouldn’t.

    Access to medical data is almost invariably restricted for ethical reasons – it’s private data.

    Medical data released publicly is invariably summarised for this reason (and also pragmatic reasons). When summarising it you have to choose what subsets of the data you summarise for public release, you’re can’t please everyone with that either! This doesn’t affect registered researchers’ work, as Helen pointed out.

  • Opps: “anyone to be able to look up their” had a ‘medical condition or treatment’ in angle brackets next – I forgot that text in angle brackets would get dropped. Sorry.