Recent outbreaks of diseases, including measles, have been blamed largely on changes in vaccine coverage due to the antivax movement. In many ways, this claim seems reasonable. But "seems reasonable" is not scientific. In order to assign a causative relationship, we need robust science.
The simplest way to analyze changes in vaccine coverage is to see if there's a simple change in the overall vaccination rate. As I mentioned in another question, there isn't, at least for measles vaccines in the US: https://www.researchgate.net/post/Is_the_medical_community_justified_in_assigning_blame_to_the_anti-vax_movement_wrt_measles_et_al
But this doesn't mean that coverage hasn't changed. The distribution of coverage may be different. How could we evaluate the claim that this kind of change has occurred? In terms of data, we could convert county wide data into vertices on a graph, where edges represent geographic adjacency. If there was absolutely no change, these two graphs would be homomorphic. But what would be a good method for measuring how dissimilar the graph is now to what it was in the past, in a way that would inform this question?
In a given area I would see if I can find the number of doses dispensed. This is the number purchased minus the number destroyed (if any). I could then match the yearly number of doses applied to the birth rate in the area. I can then estimate the percentage of the population at risk and the number of measles cases.
Then there are the details. Are all measles cases reported? The vaccine is not 100% effective. Will anti-vaxers be less likely to go to a doctor? In some cases I suspect the answer is yes, but is that common? Is the vaccine data accurate? Immigration, emigration issues. People in one county crossing county lines to go to a doctor of their choice, or sick people crossing county lines.
There are some fundamental problems.
1) You conclude that vaccination decreases disease incidence. The problem is that many will claim that we already knew that.
2) You conclude that vaccination was not significantly correlated with disease incidence. You will now have the pro-vaxers trying to find a way to discredit your data.
3) The anti-vaxers will not be convinced if #1 no matter what and if #2 will claim your study proves the vaccines are worthless at best.
Use very robust sample sizes. Try to make it such that some of the study results are interesting even if you conclude the expected. A map showing population, vaccine use, and disease by county might be interesting. It would help if you can identify some other (unexpected) factors that are influencing the trends. Age of parents, parent income, types of workers, and other possibly confounding factors.
Another approach would be to focus on the anti-vaxers and ask if disease incidence in this population differs significantly from the overall population based on CDC figures. Obviously the CDC figures have not (as far as I know) differentiated between pro- and anti-vaxers, so the CDC data will be slightly biased as an estimate of disease in the pro-vaxer population. As long as the pro-vaxer population is much larger than the anti-vaxer population the issue should not be a problem. The CDC value is discussed as an upper bound to the disease incidence in the pro-vaxer population.
If you have interest in publishing in a scientific journal, be aware of ethics rules regarding human subjects even if this is nothing more than asking a few people some simple questions. If this is part of a M.S. or Ph.D program you will need to comply with university ethics rules even if you have no interest in publishing in the scientific literature.
You might look at this:
https://sas.elluminate.com/p.jnlp?psid=2018-03-02.1233.M.54880B4E9513338FAC8D9708FD3107.vcr&sid=uwf0707
The link works for me as of 1-28-2019, but you will have to download Blackboard. Pasting the link into your browser will start the process. With luck the link will work for you. I say this because I was registered for this class and there may be some code (on my computer or theirs) that allows me access but will deny others.
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis