Rapid Responses to:
|
|
Rapid Responses: Rapid Responses published:
-
![[Read Rapid Response]](/icons/misc/sectionDOWN.gif)
Borderline significance of relative risk for association between type 1 diabetes and Hib vaccine
- Hans Thulesius (22 October 1999)
-
Parents'' Response
- Brenda Brandt (1 November 1999)
-
Not coincidence
- Sandy Stanford (2 November 1999)
-
Keep Researching this please.
- Stephanie Haven (6 November 1999)
-
Association between type 1 diabetes and Hib vaccine
- J Claire Bramley (12 November 1999)
-
Diabetes and haemophilus influenzae vaccination
- A Juche, W Becker-Bruser (22 November 1999)
-
Re: Association between type 1 diabetes and Hib vaccine
- John P Heptonstall (23 November 1999)
-
Response
- Jinny Willis (6 December 1999)
-
Re: Response
- Bart Classen (28 January 2000)
|
|
|||
|
Hans Thulesius, General practitioner Växjö, Sweden
Send response to journal:
|
EDITOR Classen and Classen urges readers of the article of Karvonen et al (1) to check their calculations of relative risks. I agree that it would be more appropiate to compare the two vaccinated cohorts with the unvaccinated cohort. Doing this shows the following: The relative risk for diabetes type 1 diagnosed between 0 and 10 years of age in the the two cohorts of Hib-vaccinated children compared with the un-vaccinated cohort using EpiInfo Statcalc (WHO freeware) is 1.14 (1.00 <RR< 1.30; uncorrrected Chi-Square = 3.75, p=0.0528). This result is obtained by using the numbers presented in table 1 of Karvonens article (1). The same calculations in the subgroups of children diagnosed between 0 and 4.9 years gives a relative risk of 1.11 (0.91 <RR< 1.37), and for the children diagnosed between 5 and 10 years the relative risk is 1.16 (0.97 <RR< 1.37). My reflection is that Karvonen et al have underestimated a possible connection between diabetes type 1 and and Hib vaccine, and that Classen and Classen have overestimated it. More research is thus still needed! 1. Karvonen M, Cepaitis Z, Tuomilehto J. Association between type 1 diabetes and Haemophilus influenzae type b vaccination: birth cohort study. BMJ 1999; 318: 1169-1172[Abstract/Ful |
|||
|
|
|||
|
Brenda Brandt, Full-time mom
Send response to journal:
|
Thanks for publishing this study. Our 10 year old daughter was diagnosed with diabetes a few days after her third birthday (February 1992). She received her HIB vaccine on 9/20/91. Consequently, we are very interested in any further research that is conducted. Please keep us informed if possible. We have previously read about possible dangers of various vaccines but have never heard about this link. Sincerely, |
|||
|
|
|||
|
Sandy Stanford Homemaker
Send response to journal:
|
This study is very interesting to me. I have a son who was diagnosed with diabetes 6 months after his first hib shot. two of his friends were also diagnosed 6 months after their first hib shot. There is no family history of diabetes in any of these families,this is too coincidental. Please, continue to study this. Thank you, Sandy stanford |
|||
|
|
|||
|
Stephanie Haven Full Time Stay At Home Mom
Send response to journal:
|
I found this article very interesting as my daughter had a vaccine a few months before she was diagnosed with type 1. Please keep working on this. |
|||
|
|
|||
|
J Claire Bramley, Senior Scientist, Immunisation Scottish Centre for Infection and Environmental Health
Send response to journal:
|
Using the same raw data, one group of researchers support an association between type 1 diabetes and Hib vaccine1, while colleagues refute it2. Judicious selection and presentation of data was used to influence perceived study outcome. The tables below show the relative and attributable risks for diagnosis of type 1 diabetes in each of the two cohorts scheduled to received Hib in comparison to a cohort born prior to commencement of the vaccine study, with 95% confidence intervals. Table 1. Relative risk of diagnosis of type 1 diabetes [95% confidence interval] _________________________________________________________ Age at diagnosis (years Cohort 1* Cohort 2 † Cohort 3 ‡ _________________________________________________________ 0 - 1.9 - 1.39 [0.8, 2.39] 1.44 [0.83, 2.48] 0 - 4.9 - 1.19 [0.93, 1.52] 1.04 [0.80, 1.35] 0 - 7 - 1.26 [1.05, 1.51] 1.14 [0.95, 1.38] 0 - 10 - 1.17 [1.00, 1.37] 1.11 [0.94, 1.30] _________________________________________________________(Raw data from (1) used, except for 0 - 7 years, where no raw data available and rates per 100,000 used as given in (2)). Table 2. Attributable risk of diagnosis of type 1 diabetes per 100,000 children [95% confidence interval] _________________________________________________________ Age at diagnosis (years) Cohort 1 * Cohort 2 † Cohort 3 ‡ _________________________________________________________ 0 - 1.9 - 10 [-8, 3] 11 [-7, 29] 0 - 4.9 - 26 [-13, 64] 6 [-32, 43] 0 - 7 - 54 [12, 96] 30 [-11, 71] 0 - 10 - 58 [-2, 118] 36 [-24, 95] __________________________________________________________ * Born 1 from October 1983 to 1 September 1985; ie before Hib study vaccination period (n = 128 936). † Born on odd numbered days from 1 October 1985 to 37 August 1987; scheduled to receive 4 vaccinations at 3, 4, 6 and 14-18 months (n = 59 238). ‡ Born on even numbered days from 1 October 1985 to 37 August 1987; scheduled to receive 1 vaccination at 24 months (n = 57 114). Using published data, it is only for cumulative diagnosis of type 1 diabetes in cohort 2 to age seven that the 95% confidence intervals do not span the null hypothesis. This was a cut-off employed by Classen and Classen, raising the question as to whether this was the only point at which statistical significance was attained. Simon was right to insist on the value of confidence intervals3. These were omitted by both Classen and Classen and White when speaking of the attributable risk of Hib vaccination in highly emotive terms 1 4. In both examples, the 95% confidence intervals include zero. No published analysis of the Finnish study to date has taken account of actual vaccine status of children in relation to the diagnosis of type 1 diabetes, making it essentially ecological. Vaccine uptake in cohorts 2 and 3 was not 100%. Moreover, cohort 1 is historical to cohorts 2 and 3. This is problematic as there is an established temporal trend toward increased incidence of type 1 diabetes. Furthermore, the non-independence of consecutive year data in analysis of infectious disease dynamics is well-known. The re-examination of statistics promotes scientific rigour, but this is no academic exercise, as recent letters from concerned parents demonstrate 5. The power of statistics, no matter how spurious, to undermine vaccination programmes of proven worth has a long historical precedent. References 1. Classen JB, Classen DC. Association between type 1 diabetes and Hib vaccine. Causal relation is likely. BMJ 1999;319(23 October):1133. 2. Karvonen M, Cepaitis Z, Tuomilehto J. Association between type 1 diabetes and Haemophilus influenzae type b vaccination: birth cohort study. BMJ 1999;318(1 May):1169-1172. 3. Electronic responses. Association between type 1 diabetes and Hib vaccine: birth cohort study. eBMJ 1999;318 www.bmj.com/cgi/eletters/318/7192/1169. 4. White H. Association between type 1 diabetes and Hib vaccine. More research is still needed. BMJ 1999;319(23 October):1133. 5. Electronic responses. Association between type 1 diabetes and Hib vaccine. Causal relation is likely. eBMJ 1999;319 www.bmj.com/cgi/eletters/319/7217/1133#EL2 |
|||
|
|
|||
|
A Juche, Doctor for general medicine; physician and pharmacist arznei-telegramm, W Becker-Bruser
Send response to journal:
|
EDITOR - reading the correspondence and electronic responses to the article by KARVONEN et al (1) concerning type I diabetes and haemophilus influenzae vaccination, we find it disturbing that despite wide public interest (e.g. electronic responsec s to the letters of CLASSEN, CLASSEN and WHITE), the fundamental questions and criticism regarding the Finnish study are left unanswered both by the authors and by ELLIMAN, D., writer of the accompanying editorial (2). As CLASSEN, CLASSEN (3) and WHITE (4) pointed out in their letters we urge the authors to answer the following questions. What is the relative risk (including confidence interval) between children receiving no vaccine and children receiving vaccine in any dose (one or four doses)? As one can calculate the "Number Needed to Treat" from the original Finnish study, what is the "Number Needed to Harm" concerning this particular adverse reaction? Why are no statistical comparisons made between children with no vaccination and with four vaccinations? What are the reasons for not showing the data of cohort 1 in figure 1 in the article by KARVONEN et al? Medical doctors need clarification and answers to these questions to be confident in recommending the vaccination to their patients. Without any answer by the authors or the editorialist to this contradictory conclusions and worrying questions, one cannot help but suspect risks are swept under the carpet. Dr. med. Aaron JUCHE, Wolfgang BECKER-BRÜSER A.T.I. Arzneimittelinformation Berlin GmbH & Co. KG
Bismarckstr. 63, D-12169 Berlin
1 KARVONEN M, CEPAITIS Z, TUOMILEHTO J. Association between type 1 diabetes and Haemophilus influenzae type b vaccination: birth cohort study. BMJ 1999; 318: 1169 - 1172 2 ELLIMAN D. Vaccination and type 1 diabetes mellitus. BMJ 1999; 318: 1159 - 1160 3 CLASSEN JB, CLASSEN DC. Association between type 1 diabetes and HIB- vaccine. Causal relation is likely. BMJ 1999; 319: 1133 4 WHITE H. Association between type 1 diabetes and HIB-vaccine. More research is still needed. BMJ 1999; 319: 1133 A.T.I. Arzneimittelinformation Berlin GmbH & Co. KG Bismarckstr. 63, D-12169 Berlin, Fax: +30-79 49 02-20 http://www.arznei-telegramm.de, E-Mail: ati@berlin.snafu.de |
|||
|
|
|||
|
John P Heptonstall, Director of The Morley Acupuncture Clinic and Complementary Therapy Centre West Yorkshire
Send response to journal:
|
This topic interests me both as a parent whose children are 'expected' to be vaccinated and a practitioner whose patients often ask for advise on vaccination matters. Your response to Classen and Classen (C & C) mentions 'the power of statistics ...to undermine vaccination programmes of proven worth'. I suggest that 'the power of statistics' works both ways and (thanks to Mark Twaine) people have begun to acknowledge that statistics can be made to 'prove' virtually any hypothesis given the ideal 'n' and sufficient variables in the data to protect the researcher's conclusions. Is that what C & C have done in this case as you suggest? Despite their data providing for their conclusion, your implication that the cut-off point ensured the result 'they desired' is technically non-existent does not preclude the probability that their figures may still identify a link between Hib vaccine and Diabetes; your argument merely shows that the (lack of) confidence supplied by their statistical analysis induces doubt. Parents' comments should be heeded, after all who 'knows' the children but the parents - statistics rarely take into account the strength of parental experience of their own children, and the subjective observations of a causal relationship between post-vaccination changes in a child and the pre-vaccine state of that child. Modern medical analysis disregards this yet I belive will be increasingly proved wrong once the parents' input is considered - eg. autism, M.E. for which thousands of parents have gone to/ are going to law. Diabetes will probably be the next major issue, along with asthma, SIDS and the increasingly prevalent 'neurological impairments' in children I hear physicians and speech therapists proclaiming have increased exponentially this past decade or more. I would ask you a few questions:- 1. What was/is the incidence of diabetes in Finland, in 0 to 10 year olds for the nine years a. pre-Hib vaccination b. post-Hib vaccination 2. What are the figures for the UK for the nine years pre. and post Hib vaccination introduction. I note you say 'there is an established temporal trend towards increased incidence of type 1 diabetes' as if this somehow adds to the negation of C & Cs findings. 3. The parents say their children (and childrens' friends) developed diabetes about 6 months after a Hib shot. What is the pre. and post Hib introduction rate of incidence of Type 1 diabetes in children aged 0 to 2.5 years of age in the nine years before and after introduction? a. For Finland b. For the UK 4. If the Hib. shot is causing diabetes, what possible causal mechanism might be involved in your experience? 5. I note your expertise in the vaccination field, your comment re. proven worth of vaccines, the fact that the 'new meningitis C vaccine' campaign is imminent and the only study data I have been able to find on the new type vaccine describes over 200 'toddlers' being split into 'vaccinated' (V) and 'control' (C) groups who were given:- V group - 2 injections 2 months apart of either Chiron's MenC vaccine or a meningococcal polysaccharide vaccine. C group - Hepatitis B vaccine. Then ALL toddlers received a 12-month follow-up dose of a licensed meningococcal polysaccharide vaccine. Antibodies were evaluated before the first vaccination and at 2, 3, 14 and 15 months after the initial inoculation. In view of comments of Doctor for general medicine, physician and pharmacist A. Juche: W. Becker-Bruser, above "why are no statistical comparisons made between children with no vaccination and with the vaccinations".....why would Chiron use a trial method that clearly bombards ALL trial toddlers quite heavily with meningococcal antibodies when tesing the efficacy/effectiveness of the 'new vaccine'? As our government spokespersons have been quick to inform us that the new Men. C vaccines are produced and tested similar to the Hib vaccines, was the Hib vaccine tested in simple terms of - those injected with Hib versus those not injected with Hib - or in a format similar to Chiron's 'strange' trial protocol? Kind regards John H. |
|||
|
|
|||
|
Jinny Willis
Send response to journal:
|
EDITOR- Classen and Classen's comments (BMJ 23 October 1999) reflect a similar and ongoing debate concerning Hepatitis B vaccination and type 1 diabetes. Classen claimed (NZMJ, 24 May 1996) that "asked to evaluate the effect of a recent hepatitis B immunisation programme in New Zealand ... found a large epidemic of diabetes, 60% increase, occurred in New Zealand following this immunisation programme and believe that the most likely explanation is that the immunisation programme caused the diabetes epidemic". The incidence rate of type 1 diabetes in Canterbury (NZ) has shown a significant increase with time since 1970, equivalent to 0.4 cases per year, or one extra new case every two years (p<0.001). An incidence peak of 21.2 cases/100 000 person years was recorded in 1990. Hepatitis B immunisation for newborn babies was introduced in New Zealand on 29 Feb 1988 and from 1 Dec 1989 the plasma derived vaccine was replaced by a recombinant vaccine administered at 6 weeks. We examined the relationship between introduction of the Hepatitis B vaccine and the increased incidence of type 1 diabetes in a geographically defined cohort of children and adolescents aged < 20 years. Of the 154 cases identified since 29 February 1988, 126 cases were born prior to the introduction of the vaccine, 11 cases were born between 29 February 1988 and 1 December 1989 (1.75 birth cohort years) and 17 were born after 1 December 1989 (7 birth cohort years). The cumulative incidence rates expressed as cases/birth cohort year were 6.8 (95% CI, 4.6-7.9) in cases never immunised, 6.3 (95% CI, 4.5-6.5) in cases immunised at birth and 2.4 (95% CI, 0-4.9) in cases immunised after 6 weeks of age. It is acknowledged that the immunised group has a much shorter follow-up period than the group that was never immunised. This analysis based on high acceptance rates for this vaccine does not provide evidence that the increase in the incidence of type 1 diabetes in this cohort is attributable to immunisation for Hepatitis B, or that immunisation at birth has a protective effect. The incidence of type 1 diabetes is increasing in almost all populations studied. In Finland during 1965-1992, that is during the period of the Haemophilus influenzae type b vaccination trial, the increase in incidence was almost linear (1). The use of historical controls therefore creates an artificially low background incidence rate. In addition, incidence patterns are characterised by episodic peaks and troughs in presentation. Predictably, some of the peaks in incidence correspond to the introduction of various vaccination regimes, but this coincidence should not be confused with causality. Jinny Willis Russell Scott Brian Darlow 1 Tuomilehto J, Virtala E, Karvonen M, Lounamaa R, Pitkaniemi J, Reunanen A, Tuomilehto-Wolf E, Toivanen L. Increase in incidence of insulin-dependent diabetes mellitus among children in Finland. Int J Epidemiology 1995; 24:984-92. |
|||
|
|
|||
|
Bart Classen, Classen Immunotherapies Classen Immunotherapies
Send response to journal:
|
Dear Sir, Jinny Willis questions the association we discovered (New Zealand Medical Journal, 109: 195, 1996) between the hepatitis b vaccine and the development of IDDM in New Zealand. Her analysis is completely flawed because it assumes that there was no catch up program in New Zealand . In fact the vaccine was offered to all children 18 and under shortly after hepatitis B vaccination was started in 1988. Willis analyzes the incidence of IDDM in children born before February 1988 to children born after this time, however all children whether born before February 1988 or after were offered the hepatitis b vaccine. Only children older than 18 in 1988, ie those born before around 1970, were not offered the vaccine. Willis' analysis only proves that the incidence of IDDM is higher in older children (those born before 1988) than the very young children (those born after December 1989). It is well documented that the incidence of IDDM is generally higher in children aged 5-10 then children age 0-4. Bart Classen |
|||