Yesterday I reported on the vax vs. unvax study that was recently published in Germany. In short, researchers used data from a previous survey, and were able to compare 94 unvaccinated children vs. over 13,000 vaccinated children. They looked at incidence of vaccine preventable diseases, and also incidence of other infectious diseases not covered by vaccines, as well as incidence of atopic diseases (basically allergies of some sort). They found that unvaccinated children had three times as many vaccine preventable diseases compared to vaccinated children, while incidence for the other non-vaccine-preventable diseases was the same. This study suggests that vaccinations are effective in protecting from the diseases they are meant to protect and do not make a child more susceptible to other infectious disease.
People on the other side of the vaccine debate have presented a list of issues in regards to this study. Let us look and see what these concerns are.
(1) A control group many tens of times larger than the test (unvaccinated) group means that no particular trend can emerge. Doing this weakens the statistical likelihood of seeing the trend you are testing for. If getting equal size groups to test is impractical for some reason, then having twice as many controls as test kids is workable – but having many thousands more in the controls versus less than 100 kids in the unvaccinated group assures a weak finding if any.
I disagree; my understanding is that the two groups need to be well matched as far as their characteristics are concerned, so biases can be minimized. The larger control group will actually give more realistic data to compare against. To put this into context, the survey covered 17,000+ children and adolescents from 167 German localities. Of these, only 94 were identified as unvaccinated. The sample was judged to be representative of the entire population. As such the large group of vaccinated children is in fact representative of the total vaccinated population. The problem rest with the small number of unvaccinated children, that group may be too small to be representative of the entire unvaccinated population, but that can go either way.
You cannot pick and choose 100 vaccinated kids to compare with the 94 unvaccinated ones, because you then risk introducing more bias in your selection. Just how exactly do you pick your 100? How do you make sure that they provide proper control? The numbers are what they are in the study not by choice but simply because that’s the proportion of children that are unvaccinated. It is true that the 94 unvaccinated children might not be representative of the total unvaccinated population, but by reducing the vaccinated numbers you also make that group not representative of their respective population. How exactly does that make things better?
The ideal solution would be to have a group of unvaccinated children in the hundreds, or thousands, whatever size the matching control group happens to be. That might be doable, but given that it took 3 years to collect data on the current sample, that might take even longer.
2) The paper is vague on exactly what the vaccine records were. If kids got newborn hepatitis B shots, that usually resides only in the hospital (birth) record, and not in the pediatrician’s office record. Were some of the unvaccinated kids actually vaccinated? It’s not really clear – a potentially huge confounder.
The paper is not vague on the vaccine records: vaccine card information was used to determine vaccination status.
The questions about diseases were followed by data collection on the basis of medical records in the vaccination card, about data concerning the administered vaccinations and the timing of the vaccination.
However, the concern that such data may be incomplete is valid; it is impossible to be 100% sure that lack of documentation, or parental recollection, of vaccine receipt translates to an absolute certainty that the child did not receive even a single dose that went unrecorded. This shortcoming was pointed out by me on my entry, and the authors themselves:
we cannot entirely exclude the possibility that some children without a vaccination card were vaccinated without adequate documentation.
On the other hand, concerning disease incidence on vaccinated children, the survey only asked if the child was diagnosed with disease X, not concerning itself with when the disease afflicted the child. That introduces another uncertainty: the vaccinated children may have gotten sick prior to vaccination, or prior to satisfactory vaccination (example one dose in a 5 dose series) thus potentially the disease occurrence rates in the vaccinated population might be overstated in this study. This uncertainty about the records is a knife that cuts both ways.
(3) It’s a retrospective, not prospective design; considerably weaker
That is correct, but as explained in my previous entry, a prospective study will not pass ethical standards for scientific experimentation in humans. That is just not going to be done, and it is a reality we have to live with.
(4) I see errors on kids’ vaccine records in my practice regularly, which makes me further question the reliability of reviewing records and retrospectively at that
That may be true, but then maybe this person’s practice is particularly messy with records, or maybe we’re just not too good at keeping paper trails here in the U.S. The point is that how bad or good the records are is an unknown and cannot be controlled for. Furthermore, bad record keeping would affect both groups equally; there is no reason to believe that it would favor one group over the other.
(5) The German vaccine schedule is not the same as the US schedule, so we can conclude nothing about our schedule in the US from this
Not the same, but similar enough. the 5 vaccines the German study looked at are also part of the US schedule. So on top of the safety and efficacy evidence we have for each and every vaccine on the US schedule, we can now add this new evidence of efficacy and safety for 5 of the vaccines combined. Far from not being able to conclude nothing about the US schedule, we now have even more evidence for its safety and efficacy.
This all-or-nothing attitude that the antivaxer takes is intellectually dishonest: the medical sciences do not work that way, after all you’re dealing with human beings and sources of unintentional biases are practically unlimited. We look at all the evidence available and see where the majority of it points. One study is never going to prove, or disprove, anything. Certainly this one German study does not prove that the US schedule is safe, but it adds to the existing evidence of its safety.
(6) The authors appear at one point to lump “illness” all together, and count discreet illnesses like measles as they do chronic conditions like asthma, which is directly counter to the purpose of the study. Discreet infections like measles don’t equate allergic illness/inflammatory conditions that last a lifetime
I cannot comment on this; I am not sure what is meant by this paragraph.
In conclusion, the reviewer closes with this:
Long short this is a weakly designed study that looks like it intended from the outset to satisfy the bias of the authors.
I would like to hear the reasons behind such a big charge. How could the study have been designed differently to make it better? Given the existing data from the survey, what would the reviewer had done differently that would have produced better results? Going through the 5 points: reducing the control group makes things worse; the vaccine records are what they are and nothing can be done about it; the study has to be retrospective because of ethical research standards; imperfect record-keeping is what it is and cannot be changed, and the 5 vaccines used are all part of the US schedule.
It seems to me the researchers did all that could be done with the available data, and I welcome a design for a better study, within the same limitations the researchers had to work with.
Could it be that the antivaxer is biased herself against any study showing vaccines to be beneficial that she is projecting that onto the study’s authors?
--- 
RECENT COMMENTS