Tag: immortal-time bias


A reanalysis of an Italian study on the effectiveness of COVID-19 vaccination suggests that it might have unintended effects on total mortality

Publication date: 11/04/2024 – E&P Code: repo.epiprev.it/2862
Authors: Alessandria M.1, Malatesta G.2, Donzelli A.3, Berrino F.4.Sottomesso alla peer review sulla rivista Epidemiologia&Prevenzione

Abstract: Immortal-time bias (ITB) is known to be common in cohort studies and distorts the association estimates between treated and untreated groups. We used data from the last of two large studies in an Italian province on COVID-19 vaccines safety and effectiveness incurred this bias, and aligned the entire population on a single index date, to correct the ITB. We considered the “all-cause deaths” outcome to compare the survival curves between the unvaccinated group and the various vaccination statuses. The all-cause deaths Hazard Ratios in univariate analysis for unvaccinated (reference) versus vaccinated with 1, 2, 3/4 doses were 0.88 (CI95: 0.78 –1.00; p-value 0.044), 1.23 (1.16–1.32; p-value ≤0.001) and 1.21 (1.14–1.29; p-value ≤0.001), respectively. The multivariate values were 2.40 (2.00–2.88; p-value <0.0001), 1.98 (1.75–2.24; p-value <0.0001), 0.99 (0.90–1.09; ns). The possible explanations of the trend of the Hazard Ratios as vaccinations increase could be a harvesting effect; a calendar-time bias, accounting for seasonality and pandemic waves; a case-counting windows bias; a healthy-vaccinee bias; or some their combination. With two and even with 3/4 doses the calculated Restricted Mean Survival Time and Restricted Mean Time Lost have shown a small but significant downside for the vaccinated populations.

More >

COVID-19 vaccination, all-cause and non-COVID-19 mortality in an Italian Province. Data updated, re-presented biases

Publication date: 02/10/2023 – E&P Code: repo.epiprev.it/2824
Authors: Giovanni Malatesta1, Alberto Donzelli2, Marco Alessandria3, Franco Berrino4

Abstract: The SARS-CoV-2 pandemic has led to an unprecedented effort to generate real evidence on the safety and effectiveness of various treatments, mRNA vaccines included. We can now welcome with great interest the publication of a new article on mortality in the general population of the Italian province of Pescara, divided by vaccination status, with a follow-up of two years (1).
Already in articles published during the pandemic (2, 3) it has been argued that, in observational cohort studies, an incorrect management of follow-up times may introduce the so-called Immortal-Time Bias (ITB) in favour of the exposed group. Nevertheless, ITB still appears to be present in several cohort studies. As highlighted in an article (Intervention) recently published in Epidemiologia&Prevenzione (4), a study on the safety of COVID-19 vaccines in the population of an Italian province is no exception (5). A possible explanation that the ITB is still largely prevalent in such cohort studies may be that the structure of the ITB is still poorly understood (3). A new study on the effectiveness of COVID-19 vaccines (1) in the same Italian province as that examined in the article (5) we have previously considered (4) repeats the aforementioned bias, along with others, similarly common, that we have discussed before.
Errata corrige: nel grafico a pagina 3 la cifra 1,896,676 è stata coretta in 186,676 (11 ottobre 2023)

More >