How the media can lie about neuroscience statistics in the news

"H1N1 Vaccine Associated With Small but Significant Risk of Guillain-Barre Syndrome" screams the headline of a neuroscience news site.

Well, okay at first glance.  After all, it is a fact that one can get GBS after immunization. But what are the ACTUAL ODDS? What are the actual numbers, based on a recently published (in JAMA) study done in Quebec?

The population of Quebec: 7825803 (about 8 million), in 2009 (see here). Those vaccinated for the H1N1 strain of inflenza in Quebec in 2009, during or the study time interval: about 4400000 (4.4 million).

There were 83 cases confirmed of Guillain-Barre syndrome around then, of which 25 were from individuals dosed with the H1N1 vacccine within 8 weeks of disease onset. So, we have 25 cases out of 4.4 million and (83 - 25), or 58 cases out of ( 7825803 - 4400000 ) = 3825803.

Relative risk calculation: ( 25 / 4400000 ) / ( 58 / 3825803 ) = 0.375.

So, the statistics indicate that the vaccine was associated with a DECREASED risk of Guillain-Barre compared to not getting the vaccine. 

In fact, those who got H1N1 got Guillain-Barre during the period just after immunization had 2.6 times LESSER risk of getting Guillain-Barre versus someone who did not get the vaccine in that same period.

This is not first class evidence, of course: the population was not randomized.  Furthermore, if we include the 17 cases of GBS in those who were immunized prior to the 8 week interval, who were therefore counted in the 58 cases felt to be unrelated to immunization, we get a more even distribution, befitting the altered time intervals: 42 immunized cases and 41 non-immunized cases. The resulting trend is still biased by 11% toward immunization protecting against GBS even then: relative risk ( 42 / 4400000 ) / (41 / 3825803 ) = 0.89.

Thus, looking at the numbers, we can see how the headline given in the news can show how badly the media can be biased against belief in the safety of immunizations today. It is concerns about safety, promoted by media scare stories, that may drive resistance to immunization in some Americans.

-----------------------------

ABSTRACT

AUTHORS: Philippe De Wals, PhD; Geneviève Deceuninck, MD; Eveline Toth, MSc; Nicole Boulianne, MSc; Denis Brunet, MD; Renée-Myriam Boucher, MD; Monique Landry, MD; Gaston De Serres, PhD

TITLE: Risk of Guillain-Barré Syndrome Following H1N1 Influenza Vaccination in Quebec.

Context  In fall 2009 in Quebec, Canada, an immunization campaign was launched against the 2009 influenza A(H1N1) pandemic strain, mostly using an AS03 adjuvant vaccine. By the end of the year, 57% of the 7.8 million residents had been vaccinated.

Objective  To assess the risk of Guillain-Barré syndrome (GBS) following pandemic influenza vaccine administration.

Design  Population-based cohort study with follow-up over the 6-month period October 2009 through March 2010. The investigation was ordered by the chief medical officer of health in accordance with the Quebec Public Health Act.

Setting  All acute care hospitals and neurology clinics in Quebec.

Population  Suspected and confirmed GBS cases reported by physicians, mostly neurologists, during active surveillance or identified in the provincial hospital summary discharge database. Medical records were reviewed and cases classified according to Brighton Collaboration definitions (categorized as level 1, 2, or 3, corresponding to criteria of decreasing certainty in diagnosis). Immunization status was verified and denominators were estimated from the provincial immunization registry (4.4 million vaccinated) and census data (total target population aged ≥6 months, 7.8 million), with a total of 3 623 046 person-years of observation.

Main Outcome Measures  Relative and attributable risks were calculated using a Poisson model and the self-controlled case-series method.

Results  Over a 6-month period, 83 confirmed GBS cases were identified, including 71 Brighton level 1 through 3 cases. Twenty-five confirmed cases had been vaccinated against 2009 influenza A(H1N1) 8 or fewer weeks before disease onset, with most (19/25) vaccinated 4 or fewer weeks before onset. In the Poisson model, the age- and sex-adjusted relative risk was 1.80 (95% CI, 1.12-2.87) for all confirmed cases during the 8-week postvaccination period and was 2.75 (95% CI, 1.63-4.62) during the 4-week postvaccination period. Using the self-controlled case-series method, relative risk estimates during the 4-week postvaccination period were 3.02 (95% CI, 1.64-5.56) for all confirmed cases (n = 42) and 2.33 (95% CI, 1.19-4.57) for Brighton level 1 through 3 cases (n = 36). The number of GBS cases attributable to vaccination was approximately 2 per 1 million doses. There was no indication of an excess risk in persons younger than 50 years.

Conclusions  In Quebec, the 2009 influenza A(H1N1) vaccine was associated with a small but significant risk of GBS. It is likely that the benefits of immunization outweigh the risks.

JAMA. 2012;308(2):175-181. doi:10.1001/jama.2012.7342.