The article you link to is all over the place and largely references itself rather than linking to primary sources. It's hard to mount an argument against such a profligate barrage of claims. The best you can do is patiently take claims one at a time and search the scientific literature for studies that refute the various hypotheses presented. The worst thing you can do in an argument (in which you wish to convince the adversary) is dismiss their concerns without evidence.
There are simple things to point out about the article. For instance,
Health professionals and doctors with government ties were also blamed
in Finland and Sweden after a H1N1 vaccination program was halted
following a 300 per cent increase in cases of the neurological
disorder narcolepsy amongst children and young people who had received
the shot over the last six months.
A 300% INCREASE! Responsible epidemiologists, I believe, would report something like "4 times greater risk of developing narcolepsy". If you look at the data, you see that this "300% increase" amounts to 20 patients, it appears. 1 So, instead of 20 in 100,000, you might expect 80 in 100,000. It's still worth a look and worth striving to improve; but, this fact alone might not justify violent revolution.
Gardasil appears to be safe, according to my brief look at the literature. Although, I'm not a pediatric neurologist and, at least one thinks that Gardasil was brought to the market too soon and that it is causing serious long-term side-effects.
I have a humble degree in mathematics with a concentration in epidemiology. I'm very interested in vaccine safety and cost/benefit. I would like to be strongly pro-vaccine. However, dismissive statements implying that there is no debate do not push me in the pro-vaccine direction. Flatly saying "there's a significantly greater risk in not getting vaccinated", "it's been refuted countless times", "the claim ... is based solely on .. a complete scam" without any reference is unbecoming of a true skeptic. Use references to peer-reviewed papers.
The case of universal Hepatitis B vaccination for newborn Americans may put me slightly at odds with those who would suggest that following the AAP schedule to the letter is the absolute best approach to vaccination.
The incidence of HepB before vaccination in the United States for those under 15 was on the order of 1 in 100,000. At $100 a series, we are spending $10,000,000 to prevent 15 cases of HepB when we vaccinate 100,000 low-risk Americans? (please correct me if this is not the right ballpark).
Edit: the low incidence reported in the above paper might be due to the fact that those under 5 are far less likely to suffer from an acute HepB infection. People who are infected as children, on the other hand, are more likely to develop chronic life-long infections resulting in severe outcomes such as liver cancer and death. The CDC reports that 45,000 children < 10 were infected with HepB every year before universal vaccination. Meanwhile, the CHOP reports "18,000 children were infected with hepatitis B virus by the time they were 10 years old". These numbers suggest an incidence rate more on the order of 70/100,000 and 30/100,000, respectively. According to the CHOP report, 1/2 of these infections were from infected mothers, while the CDC's numbers have about 1/4 coming from infected mothers. Using the CDC's high number of vaccine-preventable incidence, it looks like vaccinating 100,000 children could prevent as many as 1,000 cases of HepB.
Using the incidence estimates from the CDC, universal HBV looks like a slam dunk in cost/benefit. BUT.
There is a lot of scary stuff out there, at first glance.
If any of those peer-reviewed papers have one grain of reality in them, then universal HepB vaccination may not be best thing to do in terms of risk and benefit. A high-risk child going immediately into daycare in NYC who has an infected father should almost certainly still bear the risk of receiving HBV in the first month. On the other hand, a child with a stay-at-home mom, living in a rural area in the US, with almost no chance of contracting HepB, may be better off not getting HBV within 12 hours of birth. If HBV within the first day of birth does indeed put the child at greater risk of autoimmune disorders and special education needs, the extremely low-risk child perhaps should not receive HBV at such a young age.
(*) UPDATE: I wanted to check on the "9 times higher risk of needing special education given HBV" numbers. The NHANES database is available as csv here . Doing a simple model in R to look at the relationship between HBV and special education, adjusting for age, health and neonatal ICU (TL; DR: Any claim of 9X from this data is highly misleading.):
> summary(hbv)
2or3 1 NA's
228 678 65
> summary(spec_ed)
Not SpecEd NA's
912 58 1
> cs(spec_ed, hbv)
Exposure
Outcome Non-exposed Exposed Total
Negative 221 632 853
Positive 7 46 53
Total 228 678 906
Rne Re Rt
Risk 0.03 0.07 0.06
Estimate Lower95ci Upper95ci
Risk difference (attributable risk) 0.04 0.01 0.08
Risk ratio 2.21 0.98 5
Attr. frac. exp. -- (Re-Rne)/Re 0.55
Attr. frac. pop. -- (Rt-Rne)/Rt*100 % 47.52
Number needed to harm (NNH) 26.92 12.69 149.36
or 1/(risk difference)
> cc(spec_ed, hbv)
hbv
spec_ed 2or3 1 Total
Not 221 632 853
SpecEd 7 46 53
Total 228 678 906
OR = 2.3
Exact 95% CI = 1.01, 6.12
Chi-squared = 4.27, 1 d.f., P value = 0.039
Fisher's exact test (2-sided) P value = 0.049
I don't understand how these numbers are stretched for an 8.6 OR with p=0.0003!
When I make a model with their covariates, I can come up with a 2.77 OR:
> model = glm(spec_ed ~ hbv + ecq + age + health, family=binomial)
> print(summary(model))
Call:
glm(formula = spec_ed ~ hbv + ecq + age + health, family = binomial)
Deviance Residuals:
Min 1Q Median 3Q Max
-1.0094 -0.3549 -0.2697 -0.1989 2.9430
Coefficients:
Estimate Std. Error z value Pr(>|z|)
(Intercept) -4.08455 0.68606 -5.954 2.62e-09 ***
hbv1 1.02191 0.45226 2.260 0.02385 *
ecqNot -1.51885 0.33112 -4.587 4.50e-06 ***
age 0.16108 0.05827 2.764 0.00570 **
health 0.40134 0.14618 2.746 0.00604 **
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
(Dispersion parameter for binomial family taken to be 1)
Null deviance: 391.85 on 899 degrees of freedom
Residual deviance: 351.47 on 895 degrees of freedom
(71 observations deleted due to missingness)
AIC: 361.47
Number of Fisher Scoring iterations: 6
> exp(1.02)
[1] 2.773195
But, it's worth noting that using only those who got the triple series ("1") and those who got no HBV ("3") (instead of putting "2" -those who got some- in with those who got none, "3"), a statistically significant correlation is not achieved. Also, learning disabilities and attention deficit disorder are uncorrelated with HBV in this NHANES data.
I made a little tool to test hypothesis of any variable in NHANES with a simple contingency table. The protective effect of HBV on females is actually more pronounced than the supposed negative effect on the boys.
The girls were clearly protected by HBV:
In [12]: ct, p, nhanes = test_kwargs(NHANES, only_test={"riagendr": ["2"]}, float_limits={"ridageyr": [0,9]}, imq020=(["1"], ["3"]), pfq040=(["1"], ["2"]))
pfq040 in ['1'] pfq040 in ['2']
imq020 in ['1'] 12 559
imq020 in ['3'] 10 140
p = 0.00381724204472
Whereas, the boys were only borderline statistically significant.
In [13]: ct, p, nhanes = test_kwargs(NHANES, only_test={"riagendr": ["1"]}, float_limits={"ridageyr": [0,9]}, imq020=(["1"], ["3"]), pfq040=(["1"], ["2"]))
pfq040 in ['1'] pfq040 in ['2']
imq020 in ['1'] 46 632
imq020 in ['3'] 4 147
p = 0.0535442211328
I believe you should isolate each claim and judge it on the evidence. Even peer-reviewed articles can be misleading, sadly.
