|
|
|
# Create a New Cohort definition
|
|
|
|
Follow these steps to build a cohort in the user interface of [ATLAS](https://atlas-demo.ohdsi.org/#/home).
|
|
|
|
|
|
|
|
|
|
|
|
1. On the left hand side, click on *Cohort Definitions* module. \
|
|
|
|
When the module loads, we will see a list of existing cohort definitions.
|
|
|
|
|
|
|
|
|
|
|
|
2. On the top right, click on *New Cohort* button. \
|
|
|
|
This will open up a new cohort definition as you can see below.
|
|
|
|
.
|
|
|
|
|
|
|
|
3. Give a name to the cohort definition and Save it. \
|
|
|
|
`
|
|
|
|
New users of ACE inhibitors who have a prior diagnosis of hypertension.
|
|
|
|
`
|
|
|
|
|
|
|
|
> **Warning**
|
|
|
|
> ATLAS will not allow two cohorts to have the same exact names. ATLAS will give you a pop-up error message if you choose a name already used by another ATLAS cohort.
|
|
|
|
|
|
|
|
### Cohort definition components
|
|
|
|
We need to define the components of a cohort definition. There are three main sets of criteria we need to establish:
|
|
|
|
|
|
|
|
#### 1. Cohort entry events
|
|
|
|
"What do you need to observe so that someone enters the cohort "
|
|
|
|
|
|
|
|
1. Click at the *+ Add Initial Event ...* dropdown to see all the domains available of the old OMOP Common Data Model.
|
|
|
|
|
|
|
|
|
|
|
|
In this case, we are interested in `ACE inhibitors drug`. So, we will add a `drug exposure event` by clicking at "*Add Drug Exposure*" button.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2. Once selected, it creates a new record that says that: *We are looking for cohort entry events which are events having any of the following criteria: a drug exposure of "**Any Drug**"*
|
|
|
|
|
|
|
|
ATLAS does not know what drug we are looking for, we need to specify which concept set (`drug of interest`) is associated.
|
|
|
|
To do so, click "*Any Drug*" dropdown button and then click "*Import Concept set*", in order to see a list of all the concept sets available in the ATLAS platform.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
As we can search in the filter box, we are going to search for `ACE inhibitors` and select a concept set from the list.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
*`Now, we can see that our cohort entry events are going to be records of the drug exposure ACE inhibitors.`*
|
|
|
|
|
|
|
|
If we click *`"ACE Inhibitors"`* dropdown and hover over the concept set, we can see what concept set expression is and a list of all of the active ingredients within the `ACE inhibitor` class.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Now, this gets us all the records of `ACE inhibitors`. However, we may want to apply additional domain-specific attributes.
|
|
|
|
|
|
|
|
3. We can click "*+ Add attribute...*" button to decide what characteristics we want to add.
|
|
|
|
|
|
|
|
|
|
|
|
4. `To ensure that we are identifying new users`, we may be interested in finding `the first exposure` which is "*Add First Exposure Criteria*" option. If we click that option, the new criteria will be added to the cohort.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
*`Now, we are saying that we are looking for entry events that are a drug exposure record of ACE inhibitors requiring that the record is the first time in a person's history (i.e. the first time we see an ACE inhibitor record for any given individual).`*
|
|
|
|
|
|
|
|
Additional attributes that we could consider would be: age criteria, gender, start date, end date and all of the domain specific attributes that are available in `the drug exposure domain`.
|
|
|
|
All these attributes become criteria that will then be imposed on this `drug exposure` rule.
|
|
|
|
|
|
|
|
5. We may want to limit our population to `people who are aged greater than 18` by adding the attribute "*Add Age at Occurrence Criteria*" and specifying *`"Greater Than: 18`"*.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
6. We may want to limit the cohort to people `who are in the database sometime after some calendar period of time`, by adding *Add Start Date Criteria* attribute.
|
|
|
|
|
|
|
|
|
|
|
|
7. Finally, we need to establish whether or not we are going to pose any "*continous observation*" requirement.
|
|
|
|
We may want some period of time for which `we do not observe the drug`, in order to be confidence that `this is truly new use` and to distinguish it from `prevalent use`.
|
|
|
|
We are going to specify `at least one year (or 365 days) of prior observation time, before the first ACE inhibitor` is actually observed.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
8. Also, we can "*Limit initial events to*". \
|
|
|
|
In our case, we are looking for the `first time of a new use of an ACE inhibitor` by selecting the "*earliest event*" option per person.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
> NOTE: A cohort entry event could in fact have multiple events for a particular individual, it depends on the phenotype you are trying to specify.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
#### 2. Inclusion criteria
|
|
|
|
Once the cohort entry events are specified, we can define any inclusion criteria that we would like to apply to this cohort.
|
|
|
|
It will be applied to those cohorts entry events to identify what subpopulation is going to be used.
|
|
|
|
|
|
|
|
1. Click *New inclusion criteria* button to define the criteria we are interested in.
|
|
|
|
|
|
|
|
|
|
|
|
2. We would like for these people to `have a prior diagnosis of hypertension`. So, we should name the criteria and, if desired, put a little description of what we are looking for.
|
|
|
|
|
|
|
|
|
|
|
|
3. Now, it is time to define a criteria that represents that clinical idea by clicking "*+ Add criteria to group...*" dropdown.
|
|
|
|
|
|
|
|
> NOTE: We can see all the domain of the OMOP CDM available.
|
|
|
|
|
|
|
|
In this case, `a diagnosis of hypertension` would be represented in the `condition` domain. Then, we need to select "*Add Condition Occurrence*" criteria.
|
|
|
|
|
|
|
|
|
|
|
|
Above, we can see that the criteria says: *`We are going to find all of those events that satisfy the criteria of having at leat one occurrence of a condition occurrence record of "Any Condition" where the event starts between all time before and all time after`*.
|
|
|
|
|
|
|
|
4. Effectively, we will be looking for any particular condition record just to exist for a person. Then, we are going to click "*Any Condition*" dropdown, which provides the ability to either select concept sets that are part of this cohort definition or to import new concepts sets.
|
|
|
|
|
|
|
|
5. Click on "*Import Concept Set*" option and search for the concept set definition `hypertensive disorder`.
|
|
|
|
|
|
|
|
|
|
|
|
6. Now, if we want this to be `prior to the exposure`, we need to specify that the event of `hypertensive disorder` must occure `all` time `before` and `zero` times `before` the `index event`, which is the `first exposure to the ACE inhibitor`.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
In this regard we have now found: *`New users of ACE inhibitors and we have restricted to the subset of those new users who have at least one diagnosis of hypertension`*.
|
|
|
|
|
|
|
|
#### 3. Cohort exit
|
|
|
|
The Cohort exit defines how does a person leave the cohort of interest.
|
|
|
|
|
|
|
|
1. Define the "*Event Persistance*" where multiple options are available.
|
|
|
|
|
|
|
|
|
|
|
|
In our case, we should select the option *`end of a continous drug exposure`*.
|
|
|
|
|
|
|
|
Once selected, we need to specify:
|
|
|
|
- "*Concept set containing the drug(s) of interest+". Here, it may be `ACE inhibitors`.
|
|
|
|
- "*Persistence window*", representing the maximum allowable gap between successive records that we would consider to be a continuous period of exposure. So, if we tolerate `30-day gap` between prescriptions, we can specify that.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Now, we created a cohort definition of: *`Patients who enter at the time of their first ever ACE inhibitor exposure, who are over 18 and the first exposure happened after 01/01/2020. They have to have at least one year of prior observation. There is an additional inclusion criteria that they have to have a diagnosis of hypertension prior to their exposure. They will leave the cohort when they stop being continuosly exposed to ACE inhibitors`.
|
|
|
|
|
|
|
|
Then, we need to save this cohort.
|
|
|
|
|
|
|
|
|
|
|
|
## Other components
|
|
|
|
Up top, next to the definition, there are other components that can be evaluated:
|
|
|
|
- Concept Sets: can examine the concept sets that are contained whitin each definition.
|
|
|
|
|
|
|
|
|
|
|
|
- Generation: we can generate this cohort on any of the sources configured in the ATLAS platform by clicking "*Generate*" button. We would be able to run this analysis on a particular data source.
|
|
|
|
|
|
|
|
|
|
|
|
- Export.
|
|
|
|
- Text View: textual description that fully scpecifies what this cohort definition represents.
|
|
|
|
- Graphical View: representation of what it means to be in this cohort.
|
|
|
|
|
|
|
|
- JSON : this json expression will then be able to be renderred into an SQL statement that can be executed across one of many different database platforms.
|
|
|
|
|
|
|
|
|
|
|
|
- Messages: provides you warnings as you have defined your cohort definition in case you have applied any logic that might not make sense.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
> Descargar json y ponerlo en el repo (para poder importarlo a otro ATLAS) |
|
|
\ No newline at end of file |
