Understanding of intervention description

[ReneGeci]

Hello,

I am a bit confused by the way interventions are defined. Especially "constant infusions" and "multiple doses".

If there is an IV infusion that was given constantly over some duration of X min, then how can I read out this duration? I see the entries "time" and "time_end" and would have expected this to give the start and endpoint of the administration but it seems "time_end" is actually empty for basically all entries. Instead, in some cases "time" seems to give the correct duration (for example 15 min), but then in other cases this is a negative value (like "-1 h"). this is confusing to me...

Similarly, i have problems understanding "multiple doses". How do I correctly read out how often and for how long a dose was given? For example, I have no idea what "1 days" would mean...

[janekg89]

Hi Rene, yes, for a constant infusion we use the key 'time' for the starting time and 'time_end' for ending time. So the infusion intervall is described by the difference. For oral administration, the 'time_end' value is not necessary. Typically, we recommended to use 'time'=0 for single interventions. For multiple subsequent interventions, the time will obviously differ from 0 at least for some interventions. The times chosen always should be chosen constistently across interventions and outputs. For example, let's say caffeine was administrated 10hr before midazolam and concentration of caffeine and midazolam were measured 1hr after the administration of midazolam. This can be encoded as caffeine intervention at time=-12hr, midazolam administration at time=0hr, and concentration of caffeine and midazolam at 1hr. For multiple applications you also might come across 'time' values like this: S-48T48R2. This is an abbreviation for multiple interventions with S start, T period/time, R number of repeats. I hope I could help. Greetings Jan

[ReneGeci]

@janekg89, thanks for the explanation! however, there seem to be some inconsistencies because not all the entries follow the system you described.

for example if u look at

PKDB00195, Clements1984, intervention_PK: 234

this is a "constant-infusion" "intravenous (iv)" intervention. and if u look at the paper, it states that the IV was given over 2h. the entry has time = 2h. and no time_end entry. so the duration is not given by the difference of "time" and "time_end". and this not just an exception, but many entrie are like this.

on the other hand:

PKDB00441 Calatayud1995 intervention_pk: 1292 is also a constant infusion IV with time = 0 and time_end = 20 min. So, this then follows the system you described. And the paper confirms that indeed they gave a 20 min long infusion.

further, this gets complicated when there are negative entries in "time". so now for example if t = -1h (e.g. PKDB00263 Shapiro1987 2674) then I am unsure how to interpret this correctly... obviously the IV started at t = -1h but is it also 1h long? or is the duration unknown?

further, i am confused by what the difference is between "constant infusion" and "single-dose". at first, i assumed "single-dose" is always a bolus IV, then at least it would be clear that time always stands for the time of administration. but then there are also entries with time and time_end (like Duedahl2005, intenvetion 2436). so then i am unsure if i can just trust "time" for "single dose" to only be the start time or if this is also sometimes actually the duration of the infusion?

i am sorry, if this sounds like criticism, actually I really like the database but uncertainities like this make it unfortunately difficult for me to use it. which is why i would really like to clarify/improve this.

Best René

[matthiaskoenig]

Hi Rene,

sorry for the slow replies. We are currently organizing a large conference for next week. Afterwards we can address your issues. I suggest we have a short online meeting to discuss. Then it is more clear for us what you need and what your use case is. Please contact me via mail konigmatt@googlemail.com.

Best Matthias

[ReneGeci]

no worries, i fully understand! :) will contact you via email.

[janekg89]

PKDB00195, Clements1984, intervention_PK: 234 this is a "constant-infusion" "intravenous (iv)" intervention. and if you look at the paper, it states that the IV was given over 2h. the entry has time = 2h. and no time_end entry. so the duration is not given by the difference of "time" and "time_end" and this is not just an exception, but many entries are like this.

Indeed this is not curated correctly, will be updated. Probably time_end encoding style is new then this study. Possible solution: add validation rule: required value for time_end if constant infusion.

PKDB00441 Calatayud1995 intervention_pk: 1292 is also a constant infusion IV with time = 0 and time_end = 20 min. So, this then follows the system you described. And the paper confirms that indeed they gave a 20 min long infusion.

In later studies (see PKDB identifier) our curators should know the encoding pattern. Will be updated.

further, this gets complicated when there are negative entries in "time". so now for example if t = -1h (e.g. PKDB00263 Shapiro1987 2674) then I am unsure how to interpret this correctly... obviously the IV started at t = -1h but is it also 1h long? or is the duration unknown?

In the case of multiple interventions typically the most relevant intervention for the curator is encoded as t=0. In this case, it is the glucose application. In the publication, they write that peptides and insulin are administrated 60 min earlier. Same as Clements1984, probably time_end encoding is missing.

Further, i am confused by what the difference is between "constant infusion" and "single-dose". at first, i assumed "single-dose" is always a bolus IV, then at least it would be clear that time always stands for the time of administration. but then there are also entries with time and time_end (like Duedahl2005, intenvetion 2436). so then i am unsure if I can just trust "time" for "single dose" to only be the start time or if this is also sometimes actually the duration of the infusion?

Descriptions for most of the used types can be found at https://pk-db.com/curation. The description for “single-dose” is “Single dose of substance. In case of iv route this corresponds to a bolus injection. In case of oral route this is a single dose taken orally (often as tablet or solution).” Further, "time_end" is still allowed also in the case of bolus injection. Sometimes in the text some info is given on infusion duration.