Skewed Statistics

If the Vaccinated Child Gets a Disease They Class It as Unvaccinated

In a study of measles infection rates in Senegal they determined that 40% of those vaccinated still came down with measles. However, this figure was really higher because they were counting some vaccinated children with measles as unvaccinated.

Patterns of measles transmission at school and at home were studied in 1995 in a rural area of Senegal with a high level of vaccination coverage. Among 209 case children with a median age of 8 years, there were no deaths, although the case fatality ratio has previously been 6-7% in this area. Forty percent of the case children had been vaccinated against measles.

Vaccine efficacy was found to be 57% (95% Cl -23 to 85) in the schools and 74% (95% Cl 62-82) in the residential compounds.

Starting in mid-1987, vaccinations were
offered and systematically recorded by the project at vaccination sessions organized once per month at each of the area's three health centers as part of trials of measles and pertussis vaccines (12). Children who had onset of disease within 2 weeks of measles vaccination have been classified as unvaccinated.

Source: American Journal of Epidemiology, Volume 149, number 4, 15th February 1999 - http://aje.oxfordjournals.org/content/149/4/295.full.pdf

And again.....

Documentation provided during the previous campaign may have been incomplete or lost subsequently. Hence some individuals were probably immunized without this being recorded by the project.

Children who had measles within 2 weeks of
vaccination were classified as unvaccinated.

Source: Serologic status and measles attack rates
among vaccinated and unvaccinated children
in ruFa1 Senegal, Pediatr Infect Dis J, 1995;14:203-9 - http://horizon.documentation.ird.fr/exl-doc/pleins_textes/pleins_textes_7/b_fdi_53-54/010016512.pdf

And again.....

If there was no vaccination linked to a child in the demographic database in Tari, it was assumed that the child had not been vaccinated. Hence, ‘unvaccinated’ is a default status. This would not be a problem in a system with perfect information on all vaccinations or assessment of vaccination cards of all children who die. But this was not the case in Tari where there were delays in the collection of information, so that vaccinations might have been preferentially registered among children followed-up intensively (children who survived), whereas vaccinations for children who travelled or died might be missing. As a consequence, vaccinated children who died may have been wrongly classified as unvaccinated, and the analysis might suffer from survival bias.

Source: Commentary: Contrary findings from
Guinea-Bissau and Papua New Guinea, International Journal of Epidemiology 2005;34:149–151 - http://www.hawaii.edu/hivandaids/Contrary_Findings_from_Guinea-Bissau_and_Papua_New_Guinea.pdf

And again....

Patients were excluded if they did not recall influenza testing or experiencing symptoms of illness or if they were vaccinated 1-13 days before influenza illness onset. Children with no influenza vaccination since September 2003 were classified as unvaccinated. (This study was in 2004).

Source: Assessment of the Effectiveness of the 2003-04 Influenza Vaccine Among Children and Adults—Colorado, 2003, MMWR. 2004;53:707-710 - http://jama.ama-assn.org/content/292/14/1674.full

And again! We'll Just Pretend they Didn't Have Shots

We investigated an outbreak of varicella in a population of children with a high proportionof vaccinees who were attending a day-care center in a small community in New Hampshire.

Children were considered to have been vaccinated if more than 42 days had elapsed since vaccination. Two children who were vaccinated on December 26, 2000, were classified as unvaccinated.

Source: New England Journal of Medicine, Volume 347, number 24, December 12th 2002 - http://www.cpnlac.org/pdfs/brote%20de%20varicela%20en%20pac%20vacunados.pdf

Partially Vaccinated or Children Vaccinated at Earlier Times Than Usual, Classed as Unvaccinated

Children who were inappropriately vaccinated were classified as unvaccinated. Children without vaccine cards..or incomparability of exposure since both vaccinated and unvaccinated children were assumed to have been equally likely to have been exposed.

Source: International Journal of Epidemiology, volume 19, number 4, 1990 - http://cidbimena.desastres.hn/docum/crid/Noviembre2005/pdf/eng/doc7188/doc7188-contenido.pdf

And if They Kill Your Child in a Vaccine Trial They'll Just Put Him in the Unvaccinated Group!

Survival bias arises if ‘missingness’ of the vaccination records (as defined in (b) in the
previous paragraph) is associated with the outcome (namely, death). This can arise,
for example, if dead children’s cards are destroyed soon after death, as is the case in
some of the studies undertaken in West African countries: (Vaugelade et al., 2004,
Kristensen et al., 2000), and the second study of (Elguero et al., 2005).
This bias has been described in detail (Jensen et al., 2007). Briefly, in such
circumstances, treating ‘missing’ as unvaccinated will differentially place dead
children in the unvaccinated group, and hence will bias the mortality ratio towards
zero. The magnitude of the survival bias increases as the intervals between surveys
(to ascertain vaccination status) increase, as the proportion of dead children whose
vaccination cards are destroyed (and who are therefore classified as unvaccinated)
increases, and as vaccination rates increase.

In some circumstances, intervals between surveys are short (for example, in the
study described by (Moulton et al., 2005), the surveys took place every two weeks),
and hence survival bias is unlikely to play a major role. In other situations, for
example those described by (Vaugelade et al., 2004, Kristensen et al., 2000), there
were substantial intervals between surveys, and in such circumstances there may be substantial survival bias.

Note also that, whatever the frequency of surveys, survival bias will only be present
if dead children’s records are more likely to be missing than live children’s records.

In some circumstances, however, there are large numbers of ‘missing’ records due,for example, to the destruction of dead children’s records (or other informative mechanism). In such circumstances, survival bias is likely to be substantial, and sensitivity analyses are unlikely to be useful, because they yield too wide a range of values of the mortality ratio. In this situation, which has commonly arisen in studies undertaken in West Africa, a landmark analysis is recommended (Jensen et al., 2007).

Source: Epidemiological studies of the non-specific effects of vaccines: II -
methodological issues in the design and analysis of cohort studies, C. Paddy Farrington1, Martin J Firth 2, Lawrence H. Moulton3, Henrik Ravn4, Per
Kragh Andersen5 and Stephen Evans6 on behalf of the Working Group on Nonspecific Effects of Vaccines, http://stats-www.open.ac.uk/TechnicalReports/Child%20survival.pdf

PAGE UNDER CONSTRUCTION.