ATLAS Tutorial: Data Sources - Condition Occurrence

Summary

This video script introduces the Conditioned Reporting feature in Atlas, a tool for analyzing data sources standardized to the OMOP Common Data Model. It showcases a tree map and tabular view to visualize the prevalence and intensity of health conditions, such as type 2 diabetes and essential hypertension, across a database. Detailed graphs provide insights into concept prevalence trends, seasonal variations, and age distribution, highlighting Atlas's ability to drill down into specific health concepts for a comprehensive analysis.

Takeaways

• 🗺️ The video introduces the 'conditioned reporting' feature within the data sources capability in Atlas, which helps in analyzing and visualizing data.
• 📊 The 'condition occurrence report' is selected to demonstrate the tree map representation, where the size of boxes indicates the prevalence of a condition within the database.
• 🌡 The color in the tree map represents the intensity of the condition, as measured by the number of records per person with that concept.
• 🔍 Selecting a box, such as 'Type 2 diabetes mellitus', reveals detailed information about the prevalence and average records per person affected by the condition.
• 📝 There is also a tabular view available that lists concept ID, name, person counts, prevalence, and records per person in a more traditional format.
• 📊 The report includes graphs that show the prevalence of a condition stratified by year, age, and gender, indicating trends over time.
• 📈 The prevalence graph for 'essential hypertension' shows an increase over time and a higher prevalence in older adults, with relative similarity between genders.
• 📅 Another graph highlights the prevalence by month, showing a spike in October 2015, which may reflect a change in data source vocabularies from ICD-9 to ICD-10.
• 📋 The report also details the type of records for a concept, such as outpatient claims, and the distribution of age at the first occurrence of the concept, stratified by gender.
• 🔎 The condition occurrence report allows for detailed exploration of any concept, with the ability to sort and search the table for specific conditions.
• 🔑 The tool is described as powerful for understanding the occurrence of concepts within a condition domain and for comparing different data sources.

Q & A

What is the purpose of the 'Condition Occurrence' report in Atlas?

The 'Condition Occurrence' report in Atlas is used to analyze and visualize the prevalence and intensity of various health conditions within a database, standardized to the OMOP Common Data Model.

How is the tree map representation in the 'Condition Occurrence' report helpful?

The tree map representation provides a graphical display of the prevalence and intensity of conditions, where the size of the boxes indicates the prevalence and the color represents the intensity, measured by the number of records per person.

What does the prevalence percentage in the tree map signify?

The prevalence percentage signifies the proportion of patients within the database that have a particular condition, such as 7.28% for type 2 diabetes mellitus affecting over 6.2 million patients.

How can the 'records per person' metric be interpreted in the context of the report?

The 'records per person' metric indicates the average number of records associated with a specific condition for each patient who has that condition, reflecting the intensity of the condition's documentation.

What additional information is available in the tabular view of the report?

The tabular view, or 'table' tab, offers a line listing of concepts with their IDs, names, person counts, prevalence, and records per person, allowing for a detailed examination of condition occurrence data.

How does one explore a specific condition in more detail within the report?

By clicking on a row in the tabular view or selecting a box in the tree map, users can drill down into a detailed report for that specific condition, revealing additional graphs and data.

What does the first graph in the detailed report represent?

The first graph in the detailed report represents the concept prevalence of a selected condition, stratified by year, age, and gender, providing insights into the condition's distribution over time and among different demographics.

What does the prevalence by month graph indicate about the stability of a condition?

The prevalence by month graph shows the concept prevalence per thousand persons, indicating the stability or changes in the condition's occurrence over different months, which can reveal trends or anomalies like the spike in October 2015.

Why might there be a spike in condition prevalence in a specific month, as seen in the example?

A spike in condition prevalence in a specific month, such as October 2015, might reflect changes in the data source vocabularies, like a transition from ICD-9 to ICD-10, rather than an actual increase in the condition's occurrence.

What does the 'records for this concept stratified by their type' graph show?

This graph shows the distribution of concept records by their type, such as outpatient claims in primary or secondary diagnosis fields, and inpatient claims, indicating where the majority of the data for a condition is coming from.

How can the 'age at the first occurrence' graph help in understanding the condition's impact?

The 'age at the first occurrence' graph, stratified by gender, provides insights into the median age and distribution of the first occurrence of a condition, helping to understand its impact across different age groups and genders.

What functionalities does the 'Condition Occurrence' report offer for data analysis?

The 'Condition Occurrence' report allows for sorting based on prevalence or records per person and searching for specific concepts, enabling users to compare and contrast data across different sources.

Outlines

00:00

📊 Condition Occurrence Reporting in Atlas

The video script introduces the conditioned reporting feature within the data sources capability of Atlas. It begins by selecting the data sources report from the left-hand side menu, where users can choose from various standardized sources following the O mob common data model. The script focuses on the 'condition occurrence' report, which presents data in a tree map format. The tree map's size of boxes indicates the prevalence of a condition within the database, while the color intensity represents the number of records per person for that condition. For instance, type 2 diabetes mellitus is highlighted with its prevalence and average records per person. The script also mentions a tabular view for a more detailed listing of concepts, including concept ID, name, person counts, prevalence, and records per person. Further exploration is possible by clicking on rows to drill down into more detailed reports, including graphs that show concept prevalence stratified by year, age, gender, and other factors. The script notes an interesting spike in prevalence for essential hypertension in October 2015, likely due to a transition in data source vocabularies from ICD-9 to ICD-10.

05:00

🔍 In-depth Analysis with Condition Occurrence Report

The second paragraph delves deeper into the condition occurrence report's capabilities, allowing for a detailed analysis of specific concepts within the condition domain. It describes the ability to sort and search the condition occurrence table based on prevalence or records per person. The script provides an example of searching for 'diabetes mellitus', which leads to a drill-down report for type 2 diabetes mellitus. This report includes various graphs that offer insights into the concept's prevalence, such as age at first occurrence stratified by gender, and the distribution of age within that gender. The report is praised as a powerful tool for understanding the occurrence of concepts across different data sources and for making comparisons. The script concludes by directing viewers to Odyssey.org for more information about Atlas, its data sources reporting, and other features.

Keywords

💡Conditioned Reporting

Conditioned Reporting refers to the process of generating reports based on specific conditions or criteria. In the context of the video, it is a feature within Atlas that allows users to analyze data based on certain health conditions. The video demonstrates how to use this feature to understand the prevalence and intensity of conditions within a database.

💡Data Sources

Data Sources are the origins of the data that are being analyzed. In the video, the speaker selects data sources that have been standardized to a common data model and configured within Atlas. These sources are crucial for generating accurate and meaningful reports.

💡Omop Common Data Model

The Omop Common Data Model is a standardized framework for representing healthcare data. It is used to ensure consistency across different datasets. In the video, the data sources are mentioned to be standardized to this model, which facilitates the analysis and reporting of health conditions.

💡Tree Map

A Tree Map is a graphical representation where rectangles are nested and sized to represent different categories of data. In the video, the tree map is used to visualize the prevalence and intensity of health conditions, with the size of the box indicating prevalence and color intensity reflecting the number of records per person.

💡Prevalence

Prevalence in the context of the video refers to the proportion of a particular population found to be affected by a medical condition at a specific time. It is a key metric used in the condition occurrence report to understand how common a condition is within the database, such as the 7.28% prevalence of type 2 diabetes mellitus mentioned.

💡Concept

In the video, a Concept represents a specific health condition or characteristic within the data model. Concepts are used to categorize and analyze data, such as 'type 2 diabetes mellitus' or 'essential hypertension', and their prevalence and records are examined in the condition occurrence report.

💡Tabular View

A Tabular View is a way of presenting data in rows and columns, which is an alternative to the graphical tree map representation. The video describes a 'table' tab that lists concepts with their IDs, names, person counts, prevalence, and records per person, providing a detailed breakdown of condition data.

💡Drill Down

Drill Down is the process of selecting a specific piece of data and retrieving more detailed information related to it. In the video, the speaker drills down on concepts like 'essential hypertension' to explore additional graphs and statistics, providing a deeper understanding of the condition within the data source.

💡Trellis Plots

Trellis Plots are a type of graph that displays data in a matrix of small multiples or small charts. The video describes how these plots are used to stratify the prevalence of a condition by year, age, and gender, offering insights into how the prevalence varies across different demographics.

💡ICD-9 and ICD-10

ICD-9 and ICD-10 refer to the International Classification of Diseases, Ninth and Tenth Revisions, respectively. These are standard medical classification systems used worldwide for recording and analyzing health conditions. The video mentions a spike in prevalence in October 2015, which may reflect a transition from ICD-9 to ICD-10 in the data source.

💡Concept Records

Concept Records in the video refer to the individual data entries associated with a specific health condition. The speaker discusses how these records are stratified by their type and source, such as outpatient claims, providing insights into where and how conditions are being recorded within the database.

Highlights

Introduction to conditioned reporting in data sources capability within Atlas.

Selection of data sources standardized to the O mob common data model.

Condition occurrence report selection leads to a tree map representation.

Tree map size indicates the prevalence of a condition concept within the database.

Color in the tree map represents the intensity of the condition concept.

Example of type 2 diabetes mellitus concept prevalence and records per person.

Tabular view provides a line listing of concepts with ID, name, person counts, prevalence, and records per person.

Drilling down on rows provides additional information on the concept.

Graphs represent concept prevalence stratified by year, age, and gender.

Trellis plots show the prevalence of essential hypertension over time and across different demographics.

Concept prevalence can reflect changes in data source vocabularies, such as the transition from ICD-9 to ICD-10.

Prevalence by month graph shows stability with a notable spike in October 2015.

Records for the concept are stratified by type and concept ID.

Outpatient claims are a significant source of concept records.

Age at first occurrence graph provides insights into the distribution by gender and age.

Median age and interquartile range for essential hypertension concept by gender.

Condition occurrence report allows for sorting and searching within the table for specific concepts.

Drill-down report for type 2 diabetes mellitus provides detailed graphs and information.

The condition occurrence report is a powerful tool for understanding and comparing concepts across different data sources.

For more information, visit Odyssey.org.

Transcripts

[Music]

today we're going to introduce the

conditioned reporting inside of the data

sources capability within Atlas on the

left-hand side I will select the data

sources report and here we can select

any of our sources that have been

standardized to the O mob common data

model and configured inside of Atlas and

we're going to select any of a series of

data sources reports today we've

selected condition occurrence when you

select the condition occurrence report

you will be brought to this tree map

representation where you see a large

series of boxes hovering over the boxes

will reveal the information about what

they share here the size of the box

represents the prevalence of a condition

concept within the database and the

color represents his intensity as

represented by the number of records per

person who have that concept so for

example here I've selected a box which

is the type 2 diabetes mellitus concept

the prevalence of that concept in this

database is 7.2 8% which is over 6.2

million patients in this data source and

for people who have a dykon cept of type

2 diabetes mellitus we can see that

those people on average have 15.5

records per person this tree map

representation provides you a graphical

display to see which concepts occur more

or less frequently as well as a general

sense of the intensity there's also a

tabular view represented in this tab

called table which provides you the same

information as a tabular line listing so

here each row provides you the concept

ID the concept name and then the person

counts prevalence and records per person

here we can see this first row shows the

concept of essential hypertension in

this data source occurs in 15.6 million

patients which is a prevalence cot of

the concept of 18 percent and on average

people who have this essential

hypertension concept have it eleven

point three times

each of these rows can be explored

further by clicking on the row and

drilling down to the report here I've

selected the essential hypertension

concept and so I can scroll down to see

the additional information that is made

available the first graph represents the

concept prevalence of the concept of

essential hypertension the graph is

stratifying that prevalence by year age

and gender in each of these trellis

plots we can see the x-axis represents

the year of observation here this data

sources showing me information from 2000

to 2017 each trellis is representing an

age decile and the y axis of each of

these plots represents the concept

prevalence per thousand persons so on

this graph we can see for essential

hypertension that the prevalent concept

prevalence of that is higher in older

adults it seems to be growing over time

and it seems to be relatively similar

between men and women represented by the

colored line series if I scroll down

further we see a graph here representing

prevalence by month the x-axis is

showing us the calendar month the y-axis

is showing us the concept prevalence per

thousand persons this graph is showing

us a relative stability of this

particular concept in the data source

here we can see it is growing at a

relatively constant clip over time until

there seems to be an interesting spike

at October 2015 here we can see in

October 2015 the prevalence is 60

persons per per thousand persons it's

important to reinforce that the

information provided here is a concept

prevalence not a disease or phenotype

prevalence in this particular case the

illustration of the change in October of

2015 may be a reflection of the source

data source vocabularies and in this

particular case transition from icd-9 to

icd-10

the next graph down below represents the

records for this concept stratified by

their type concept ID here we can see in

this data source that 68% of the concept

records come from an outpatient claim in

a secondary diagnosis field while 27% of

the data is coming from an outpatient

claim in a primary diagnosis field we

can additionally see information coming

from inpatient claims as well

finally the graph to the right shows us

the age at the first occurrence of this

concept here we can see stratified by

gender on the x-axis and age represented

on the y axis and if hovering over any

of these box plots provides me

information about the distribution of

age within that particular gender so

here we can see that the median age of

people who are females who receive a

essential hypertension concept is 60 and

the distribution represented by the 10th

and 90th percentile in the interquartile

range the 25th and 75th percentile the

condition occurrence report here allows

drill down on any particular concept of

interest in the condition occurrence

table this table allows for both sorting

based on the prevalence or records per

person as well as searching for any

particular concept if I search for

diabetes mellitus we can see that the

concept of type 2 diabetes mellitus

appears in the table and selecting that

row will again bring up a drill down

report once the report is loaded for

that particular concept again we can

scroll down to see each of those same

graphs represented consistently in this

tool the condition occurs report is a

powerful tool to allow you to understand

what concepts are occurring in the

condition domain within a particular

source and to be able to compare and

contrast across different sources

you may have available in your

environment for more information about

Atlas in the data sources reporting and

the rest of Odyssey check us out at

Odyssey org