Event inclusion in Cohort Summary Statistics in Atlas
Last updated
Last updated
You can analyze events in Characterizations before or after people enter the cohort. This may help you to make sure that your cohorts have expected characteristics while not having any unexpected or unwanted features.
The time when a person enters the cohort is defined using Cohort Entry Events.
There are four options to define a time window before (prior) a person enters the cohort:
Any Time Prior
Long Term (365 days prior window)
Medium Term (180 days prior window)
Short Term (30 days prior window)
It is also possible to extend the end time for events to be included in the summary statistics.
To analyze drug purchases before a person enters the cohort, you can import the following Feature analyses in Characterizations:
Drug Exposure Any Time Prior
Drug Exposure Long Term (365 days prior window)
Drug Exposure Medium Term (180 days prior window)
Drug Exposure Short Term (30 days prior window)
Save your Characterization
Open the Executions tab
Click the Generate button on the right side of a FinnGen data release
When the generation is complete, click the View latest result button
Below, we see the results for drugs purchased any time before the people were diagnosed with asthma and included in an asthma demo cohort (Drug Exposure Any Time Prior). Budesonide and Mometasone are typical asthma medications. We see many antibiotics, and ibuprofen as the fourth most frequently purchased medication.
When the prior time window is shortened, ibuprofen is no longer found among the top 10 most frequently purchased medicines and budesonide moves upwards on the top 10 list. For the shortest time window of 30 days (Drug Exposure Short Term), the top 10 most frequently purchased medicines in the asthma demo cohort are:
To analyze conditions before a person enters the cohort, we import the following Feature analyses in Characterizations:
Condition Occurrence Any Time Prior
Condition Occurrence Long Term (365 days prior window)
Condition Occurrence Medium Term (180 days prior window)
Condition Occurrence Short Term (30 days prior window)
The results below show the top ten most frequent conditions within the asthma cohort when the time window is set to any time prior to the asthma diagnosis. Asthma is found in all cases because the cohort was defined for this condition:
When the time window is set to 30 days prior to the asthma diagnosis the top ten most frequent conditions show conditions typically coexisting with asthma:
Viewing medications, visits, conditions, and operations without any time window restrictions can be useful, for example, to find out if there are medications contradicting the diagnosis. In this case, additional criteria (e.g. drug usage) may be defined for the cohort.
In this example, we explore conditions and drugs for asthma cases and controls. The controls cohort contains all people not included in the cases cohort.
First, we select all drugs and conditions prior to cohort start, we import Drug Exposure Any Time Prior and Condition Occurrence Any Time Prior to Feature analyses.
Second, we extend the time window by using the endDays parameter in the Feature analysis parameters.
Below, we extended the time window, onwards from the day when people enter the cohort, by 100 years (36500 days). We now include all events for all people regardless of when the people enter the cohort.
The results below show that Allergic asthma and Intrinsic asthma are present in 19261 cases and 4070 controls. This indicates that at least the control cohort is not well defined. We should revise the cohorts before using them in analyses.
