Presence of 2024 cases in the study

[lilliniroberto]

Dear colleagues,

looking at the results of incidence/prevalence and DUS analyses for my centre (INT - Italian National Cancer Institute in Milan), I wonder why I have found that also cases from 2024 were considered, while the research protocol limited the study to 2023 cases.

Examining the R script, I have found no limitations for incident cases.

Let me know if it's only an issue of mine or if we need some new controls in the R script or anything else.

Thank you and best wishes Roberto

[tiozab]

@lilliniroberto do not worry about this, while the protocol states 2023, we consider everybody until the last data capture . @martapineda maybe the protocol needs an update on the end date of the study (last data capture) ?

[lilliniroberto]

Thank you. Still two more questions: 1) Looking at the results, in incidence our data show a steep increasing for some medicines. Do you think this could be acceptable or due to a bias in considering data fron an uncomplete year? 2) The denominator for incidence/prevalence is defined as any individual in the database who is present for ≥1 day in the time period (i.e., in the month, quarter or year of the measurement). Is it correct that this means every patients in the db regardless of having assumed or not any of the considered medicines once at least? If yes, in your opinion is this a potential bias in comparing the results from different db?

[tiozab]

@lilliniroberto thanks for your questions: @martapineda also to consider for the manuscript (although we do not put much emphasis on incomplete years, like 2024, or we should consider removing the 2024 altogether)

1. If the data is from an incomplete year, this can definitely introduce bias. For example, if the data only includes the first few months of a year and those months typically have higher sales for antibiotics.
2. exactly, that is the definition of the denominator. you are right, that caution is warranted when comparing results from different databases: However, therefore we stratify by health care setting knowing that secondary care will have a different patient population than primary care e.g. Yet, taking a step back, we are looking at medicine use and in particular are interested in time trends, thus, our focus is not on the comparison between databases, but before and after shortage announcement. Yet, it is interesting to see that some countries had more suggested shortages defined by a drop in 33% in use after the shortage announcement (results are also stratified by health care setting). I hope you can join the DUS webinar on the 5th of september where we showcase the results (soon to be in a preprint) from the Incidence Prevalence (results will be updated of course after the final deadline August 26)