Columns for checking duplicated entries about hospitalnetwork

[tjibbed]

I hadn't given the auxiliary data much thought here... I guess that for the moment, we could use unique(base, by = key(base)), with the prerequisite that all four primary variables are set as key (right?).

If I understand it correctly, this would select the unique records irrespective of the auxiliary data, selecting de facto at random one of the records to be kept. We might have to think about selecting based on aux data quality (eg. the least NA values in aux data), but for now unique(base, by = key(base)) will probably suffice.

[tjibbed]

Since the rest of the adjust_overlapping_stays function does not take the aux data into account, what admission is used as first and second if only aux data differs is completely arbitrary at the moment. It therefore doesn't matter which of the records we keep.

Until we think of a way to chose which aux data to keep, the faster methods is preferable. I changed it accordingly.

To put it all in perspective:

In one of my "real" databases, I had 602 duplicates out of 9 million records. These duplicates did all differ in their aux data. However, I have no idea what the "correct" records would have to be, I'd have to go through them by hand. Still, these are records of the same patient being admitted to the same hospital for the same time period. For the reconstruction of the network it doesn't matter. Only if we need or want to use some aux variable, there would be a tiny difference (e.g. admissions for certain diagnoses in certain hospitals)