Pénombre, la vie publique du nombre

--

Is there such a thing as immunization against a risk ?

Vaccination against hepatitis B is not compulsory in France, and it is up to parents to decide whether or not to have their 12 year-old child vaccinated... in the light of the dim clarity shed by a series of letters and figures ?

 
First there are the letters

There are many sorts of hepatitis. They are defined by letters : A, B, C, D and so on. Hepatitis B, our subject here, is a viral disease that may be transmitted sexually and through the blood and saliva. This means that health professionals and young people with “at-risk behavior” are particularly exposed.

Fortunately, many virus carriers will never experience any special problem, but in a small proportion the evolution of the disease is quite serious : some asymptomatic cases develop symptoms, which may become chronic, a number of the latter cause cirrhosis, which may produce liver cancer.

Perfectly effective vaccines are available since 1981. At first, only “at-risk” groups were vaccinated, but in 1991, WHO discovered that this strategy did not suffice to cause regression of the disease throughout the world, and therefore recommended that national authorities prescribe the systematic immunization of young people, with the long-term goal of eradicating the disease.

France adopted this strategy in 1994, and mass immunization campaigns targeted at infants and pre-adolescents were launched, as a result of which about 40 % of the population was immunized in 1999.

 
Then there are the question marks

All vaccines have “undesirable effects”, which may be more or less unpleasant. Most are minor and do not call the beneficial character of this vaccination into question. But these side effects are conscientiously monitored by the health authorities, of course, to make sure that no more serious consequences develop. Any physician who observes serious side effects following the administration of a vaccine (or of any other medication) is under the obligation to report it to the nearest drug monitoring center. These centers analyze the event and the results of their work are put together at the national level by the Drug Administration. On the basis of available findings, this agency attempts to establish a causal link between the medication and the side effects observed.

In the case of the anti-hepatitis B vaccine, some cases of severe illness, and especially of multiple sclerosis and other ailments known as demyelinizing diseases have been reported since the introduction of mass immunization. Is there a causal relationship between these ailments and vaccination ? In October 1998, the Nanterre court said there was. Without calling for any special forensic expertise, on the basis of the observation that Miss X had developed multiple sclerosis within days of that vaccination, it considered that a causal relationship was patent and sentenced the manufacturer of the vaccine to pay damages to the victim.

With headlines denouncing “the killer vaccine”, the causal relationship seemed to be equally obvious for some journalists. The victims were of the same opinion, and felt that the statisticians were being overcautious. An association, the REVAHB, was actually set up to make contact between victims, support them, defend their material and moral interests, and contribute elements of proof in the lawsuits they bring for responsibility. The REVAHB has accumulated a great many dossiers tending to demonstrate the causal relationship between vaccination and severe neurological ailments.

 
Then, numbers are brought to bear

But if a few individuals among the 25 million people who are vaccinated in France develop multiple sclerosis within weeks of vaccination, it may be a simple matter of chance. Between 1993 and 1998, the number of notified cases actually remained much lower than would be explained by simple coincidence. Does this mean that there is no causal effect ? Alas no, since there is no guarantee that the figures are exhaustive, despite the legal obligation to report cases. With concern rising in 1997-98, the Drug Administration decided to launch four different large-scale studies using four different methods, to get to the bottom of the matter. In September 1998, the results of the 4 studies were released simultaneously, and the world’s greatest specialists were locked up in a room for an entire day to analyze the findings and draw conclusions. Several days later the ministry published the outcome, with the utmost transparency.

The first and second studies were what is called case/control studies, in which a sample of multiple sclerosis victims was analyzed to determine whether the rate of immunization against hepatitis B differed from that found in the control sample.

The first study, conducted at the Pitié-Salpétrière hospital in Paris, showed an odds ratio of 1.7. In other words, all else being equal, the vaccination rate of multiple sclerosis victims was 70 % higher than for people without that ailment in France. Does this constitute proof ? No, for there is a great margin of uncertainty attached to this 1.7 ratio. The “95 % confidence interval”, that is the range within which the actual ratio has 95 % of chances of being located, is the range [0.8 ;3.7] here. In other words, if we want to have less than one chance out of 20 of being mistaken, we must confine ourselves to the assertion that the odds ratio is somewhere between 0.8 (immunization would be 20 % less frequent among multiple sclerosis victims) and 3.7 (immunization would be 4 times as frequent then). Statistically speaking, the result is said to be “non significant at the 5 % threshold”, since 1 falls within the confidence interval. There is simply a suspicion.

The second, more powerful study covered 18 neurology departments in University Hospital Centers, the largest possible sample of its kind. It collected 242 cases of demyelinizing diseases, compared with 407 paired controls. It found that :

 — if we take only patients whose immunization is attested by an up-to-date vaccination card, the odds ratio is 1.4, with a 95 % confidence interval between 0.4 and 4.5 ;

 — if all patients are considered, the ratio is 1.8, with an interval between 0.7 and 4.6.

Still not conclusive with a risk of error of less than 1/20th, but again, some suspicion. Would the third study settle the matter ?

This study covered an extremely large number of individuals (4 million people, 520 cases of multiple sclerosis, 2,505 controls), thanks to the ongoing data bank fed by English general practitioners. It found an odds ratio of 1.4 for the occurrence of multiple sclerosis within two months of immunization, with a 95 % confidence interval between 0.8 and 2.4.

Again, a suspicion, but still nothing conclusive !

The fourth study considers notifications of severe side effects. As mentioned above, notification to the disease monitoring network is not exhaustive. Now it just so happens that the REVAHB has also been collecting notifications, and has recorded a great many suspected cases over several years. Some cases are reported to the disease monitoring network, others to the REVAHB, still others to both, and some to no-one at all. In this case a method inspired by.... a pond-fishing technique (!), the capture-recapture method, is used to assess the percentages of notifications of each sort.

If we postulate that the fact of being reported to one of the two circuits does not affect the probability of being reported to the other (the validity of which is difficult to ascertain, but which does not seem absurd, a priori), measurement of the proportion of cases notified to each agency that were also notified to the other should give some notion of how exhaustive they are, and above all, of the total number of cases. That is precisely what was done. This yielded : 236 cases notified to the disease monitoring network, 71 to the REVAHB only, and 16 cases to both. According to calculations, the number of cases was comprised between 361 and 1,834 (with a 95 % confidence interval, as usual). Comparison of these figures with the expected number of cases within the population, considering the number of vaccinated individuals in France, is again inconclusive.

It is highly improbable that any subsequent studies will be able to go much further.

 
The risk of hepatitis B

Can we at least determine the risk of hepatitis B, for the prevention of which the vaccine is administered ?

The Institute for Health Monitoring (formerly the RNSP) is in charge of this evaluation. Since there is no obligation to report hepatitis B, its incidence is not accurately known. An estimation is obtained by extrapolation of the figures given by the doctors in the “sentinel network”, who detect the symptoms of hepatitis (since they are only concerned with symptoms, they have no way of detecting symptom-free cases of hepatitis). In 1994, the annual number of new cases was estimated at 7,800, which estimation was cut down to 3,100 cases of symptoms of hepatitis B, on the basis of the 1996 data.

The RNSP had done a simulation of the benefits of vaccination for a cohort of adolescents followed up until age 30. It estimated that immunization avoids between 3 and 12 cases of fulminating hepatitis and between 12 and 58 cases of cirrhosis of the liver, some of which would evolve into lethal liver cancer. Naturally, the risk for any given person depends on a number of individual parameters including the presence or absence of at-risk behavior.

Conversely, an extrapolation, for the same cohort, of data on the side effects of the vaccine, discussed above, arrives at an estimation of between 0 and 2.2 cases of severe neurological ailments (such as multiple sclerosis) caused by vaccination.

The RNSP concluded that “the benefits to the community of vaccination against hepatitis B seem to be greater than the potential risk represented by it”.

Now that you are well informed, will you have your 12-year-old vaccinated ?

 
The solution

To avoid this terrible dilemma, and rid yourself of the need for the above statistical developments, have your child vaccinated during infancy. The specialists unanimously proclaim immunization to be effective at that time, to avoid the need for immunization at adolescence, and to be devoid of any undesirable side effect.

 
Jean-René Brunetière
March 2000

 
Penumbra, 2001 June