The Australian Government Analyzed Covid Vaccines: Discard the Results

A critical review of an observational study should begin by looking for incoherent results. If found, the study should be ignored. It proved to be untrustworthy.

The logic behind this principle is simple. We found evidence that something is inherently wrong with the data. Why should we trust any inference from “misbehaving” data?

Not long ago, I demonstrated this rule on a study of Covid vaccines and mortality in Norway. We have another opportunity. This time it is an official analysis by the Australian government that includes a comparison of Covid-related hospitalization and death in vaccinated and unvaccinated people during the Omicron wave. Of course, like all official reports, the authors show data that is compatible with remarkable effectiveness of Covid vaccines.

For example, they write (in bold print):

“Among people with COVID-19, hospitalisations related to COVID-19 and COVID-19 case fatality rates were lower for those who were vaccinated.”

How good is their data?

There is one undisputed observation about Covid vaccines. Their effectiveness, whatever it might have been, has declined over several months. That’s why the first booster and subsequent boosters were promoted. Do we have consistent evidence of waning effectiveness in the official report from Australia?

We do not.

The report shows the case fatality rate and the hospitalization rate in relation to the interval between the last injection and a Covid diagnosis. Waning effectiveness implies that both rates should increase as the interval increases. For example, the mortality rate of people who contracted Covid six months after their last dose should be higher than the mortality rate of people who contracted Covid two months after their last dose. The former should have lost some protection against death when infected. Moreover, we expect to observe monotonic (i.e., gradual) decline of effectiveness over time.

My tables below were extracted from Table S25 (Covid death) and Table S22 (Covid hospitalization) in the Excel file that is linked to the official report.

Neither the case-fatality rate (top table) nor the hospitalization rate (bottom table) has increased as the interval between the last dose and infection increased. Notice that all the rates are based on large numbers. They are “statistically stable.”

The case fatality rate was 0.12% in those who were diagnosed within one month of the last dose and 0.10% in subsequent intervals of increasing length since the last dose. The hospitalization rate has decreased — not increased — with increasing intervals since the last dose.

As far as the shortest interval is concerned, it may be argued that immunity has not been built up by the time of infection. Well, for two doses and more, the official claim is that full immunity is reached by the 7th day post-injection. What percentage of the infections within one month of the last dose happened within the first week? And what about peak protection in the following three weeks as compared with longer intervals? At any rate, we still don’t have evidence of waning effectiveness in the subsequent three intervals (2–3 months, 4–6 months, >6 months).

These are not the expected results from vaccines whose effectiveness is reduced over time. If the vaccines were effective, the data failed a basic test of credibility, and we can’t learn about their effectiveness from this data. If they weren’t effective, then data showing effectiveness is false. In either case, we can officially discard the official report of the Australian government on Covid vaccines.

I could have ended the post here, but age-specific graphs show more peculiar results.

Look first at this graph from the official report. I added an arrow that shows a view we will examine shortly.

One-dose recipients were more likely to be hospitalized than the unvaccinated, if they were 20–79 years old. But at both sides of this age range (<20, >79), the trend was reversed. Although there might be a signal here, diverging results indicate that the data cannot be trusted.

The results for two-dose recipients (purple) and booster recipients (green) are consistent in all age groups. The official report appears to show effectiveness against severe Covid due to Omicron. Should we trust this graph?

I clicked the button “Time since last dose prior to diagnosis,” and that’s what we get. (No, it was not displayed in the PDF file.)

Up to age 69, effectiveness generally increases as the interval from the last dose increases. Then, the direction abruptly changes as we cross the 70-year-old mark and again at the last two intervals for 90+ year olds.

My graph below is a summary of the official graph, combining all ages below 70 and adding question marks to indicate peculiar trends.

How should we summarize the pattern above? Intriguing? Inconsistent? Erratic? Senseless? Why do the bars line up so well only in one age group?

That’s not a biological pattern. The vaccination data “misbehaves” — overall and by age group.

If you want an example of picking a narrative-serving result, read this quote from the official report (my italics). That’s what you find on hospitalization and the time since the last dose.

“Among people aged 70–89 years who had received at least one dose of the vaccine, COVID-19-related hospitalisations were higher if their last dose of a COVID-19 vaccine were 6 or more months prior, compared to those whose last vaccine dose was up to 3 months prior (Figure 7).”

One selected age range. No reference to many other possible contrasts nor to the general inconsistent pattern that I showed above. (And they refer you to Figure 7 which does not show rates by the time since the last dose in the PDF file.)

Turning last to the case fatality rate. Overall, it was 0.12% in those who were diagnosed within one month of the last dose and 0.10% in subsequent intervals of increasing length since the last dose. No evidence of waning effectiveness. The official report does not show a breakdown below age 70 (small numbers), and the scale obscures the result in that age range.

Here are my graphs. (Note the different scales of the Y-axis.)

These are not biology-driven results either. Interestingly, in the oldest age group, the processes behind the data operated in opposite directions for the two outcomes. Compare the trend of hospitalization with the trend of death of 90+ year olds. This alone is sufficient to discredit the data, and we have ample evidence.

Throughout the report, we can tell that vaccine effectiveness is implied, but the authors make an official disclaimer:

· “data in this report cannot be used to infer vaccine effectiveness.”

Indeed. The data is useless. For other reasons, too.