In ohdsi-studies/MentalHealthEquity: Assessing Health Equity in Mental Healthcare Delivery Using a Federated Network Research Model

Initial Set-Up

Activating Project Environment

knitr::opts_chunk$set(echo = TRUE, message = F, results = 'hold', cache = T)
options(scipen=999) 
devtools::load_all()
library(MentalHealthEquity)

Packages Used in Analysis

library(DatabaseConnector)

DatabaseConnector

Package Description:

This package will be used to create connections to the database and execute queries to obtain characterizations about the cohorts in the database.

Introduction

Background

The intent of this report is to investigate the characterization statement:

Characterization Statement 1: Characterize the individuals being seen for mental health care services (related to depression, bipolar disorder, and suicidality) at least one time – including hospitalization events.

This characterization statement is founded on the central research topic for this study:

Research Topic: Based on [CLAIMS], we see X% of all persons with at least one claim indicating [DEPRESSION/BIPOLAR DISORDER/SUICIDALITY] are not seen again.

By which the phrases "…all persons…" refer to those seen by patience care provider, etc. and "…are not seen again." implies lack of adherence to care.

Data Analysis Preparation

Creating Initial Connection

Defining connection details for connecting to a given database:

connectionDetails <- createConnectionDetails(
  dbms = "postgresql",
  server = "data.hdap.gatech.edu/synpuf_v5",
  user = keyring::key_list("synPUF")[1,2],
  password = keyring::key_get("synPUF", keyring::key_list("synPUF")[1,2]),
  port = 5434,
  pathToDriver = "../utils"
)

Creating Initial OMOP Tables

In this case, the schema follows the OMOP CDM v5 schema:

For this analysis, we will work with the following tables from the schema:

Per the [Book of OHDSI], here are the break downs for these tables:

• PERSON -
• LOCATION -
• OBSERVATION_PERIOD -
• CONDITION_OCCURRENCE -

General Characterization of Data

# The next SQL query takes 3 parameters: 
# the table/domain, 
# concept name (demographic is default concept name, but location is diff)
# a bool var for specific demographic or entire table

sql <- "
  SELECT {@x} ? {@table.@concept_name, count(DISTINCT PERSON_ID)} : {*}
  FROM person 
      LEFT JOIN (SELECT LOCATION_ID, STATE FROM location) AS location ON 
        person.location_id = location.location_id
      LEFT JOIN (SELECT CONCEPT_ID, 
                    CONCEPT_NAME as RACE_CONCEPT_NAME  
                  FROM concept WHERE DOMAIN_ID = 'Race' 
                  AND STANDARD_CONCEPT = 'S') AS race ON
        person.race_concept_id = race.concept_id
      LEFT JOIN ( SELECT CONCEPT_ID, 
                    CONCEPT_NAME as ETHNICITY_CONCEPT_NAME 
                  FROM concept WHERE DOMAIN_ID = 'Ethnicity' 
                    AND STANDARD_CONCEPT = 'S') AS ethnicity ON
        person.ethnicity_concept_id = ethnicity.concept_id
      LEFT JOIN (SELECT CONCEPT_ID, 
                    CONCEPT_NAME as GENDER_CONCEPT_NAME 
                  FROM concept WHERE DOMAIN_ID = 'Gender' 
                  AND STANDARD_CONCEPT = 'S') AS gender ON
        person.gender_concept_id = gender.concept_id
  {@x}? {GROUP BY @table.@concept_name
        ORDER BY COUNT DESC} :
        {LIMIT 10}
"

Cohort Populations

General Cohort

Overall Trends

conn <- connect(connectionDetails)
date_sql <- "
  SELECT  MIN(visit.visit_end_date) as first_visit,
          MAX(visit.visit_end_date) as last_visit
  FROM visit_occurrence visit
"
dates <- renderTranslateQuerySql(conn, date_sql)
disconnect(conn)

The dates in the dataset range from r dates$FIRST_VISIT and r dates$LAST_VISIT

Patient Spread

This calculates the number of patients in the dataset:

conn <- connect(connectionDetails)
count <- querySql(conn, sql = "SELECT COUNT(DISTINCT PERSON_ID) FROM PERSON")
disconnect(conn)

In this dataset, there are r count patients. To further examine this data, we can break them down across the following axes:

• State
• Age
• Race
• Gender
• Care setting

State Breakdown

conn <- connect(connectionDetails)
state <- renderTranslateQuerySql(conn, sql, table = "location", 
                                 concept_name = "state", x = T)
disconnect(conn)

The top states are: r head(state, n = 10)$STATE

r knitr::kable(head(state, n = 10))

Age Breakdown

conn <- connect(connectionDetails)
age_sql <- "
  SELECT  MIN(YEAR(visit.visit_end_date) - person.year_of_birth) as min_age, 
          MAX(YEAR(visit.visit_end_date) - person.year_of_birth) as max_age
  FROM person
    JOIN visit_occurrence visit ON
      person.person_id = visit.person_id
"
age <- renderTranslateQuerySql(conn, age_sql)
disconnect(conn)

The age of the patient at the time of visit ranged from r age$MIN_AGE years of age to r age$MAX_AGE years of age.

Racial Breakdown

conn <- connect(connectionDetails)
race <- renderTranslateQuerySql(conn, sql, table = "race", 
                                concept_name= "RACE_CONCEPT_NAME", x = T)
disconnect(conn)

In the dataset, there are r mapply(paste, race$COUNT, race$RACE_CONCEPT_NAME, "persons") r knitr::kable(race)

Gender Breakdown

conn <- connect(connectionDetails)
gender <- renderTranslateQuerySql(conn, sql, table = "gender", 
                                  concept_name = "GENDER_CONCEPT_NAME", x = T)
disconnect(conn)

In the dataset, there are r mapply(paste, gender$COUNT, gender$GENDER_CONCEPT_NAME, "persons") r knitr::kable(gender)

Care Site Breakdown

conn <- connect(connectionDetails)
caresite_sql <- "
  SELECT concept.concept_name, count(*) VISIT_COUNT, 
    count(DISTINCT visit.person_id) PERSON_COUNT
  FROM visit_occurrence visit 
    JOIN (SELECT CONCEPT_NAME , CONCEPT_ID
          FROM concept 
          WHERE DOMAIN_ID = 'Visit' 
              AND STANDARD_CONCEPT = 'S') concept ON
      visit.visit_concept_id = concept.concept_id
  GROUP BY concept.concept_name
  ORDER BY count(*) DESC
"
care_site <- renderTranslateQuerySql(conn, caresite_sql)
disconnect(conn)

In the dataset, there are r care_site$PERSON_COUNT unique patients with an indicated care site. The care site setting r mapply(paste, care_site$CONCEPT_NAME, "had", care_site$VISIT_COUNT, "visits") r knitr::kable(care_site)

\newpage

Inpatient Cohort

Overall Trends

Patient Spread

This calculates the number of patients in the dataset:

conn <- connect(connectionDetails)
inpatient_sql <- "
  SELECT @query
  FROM (SELECT *
        FROM visit_occurrence visit
        JOIN
          ( SELECT CONCEPT_NAME , CONCEPT_ID
            FROM concept
            WHERE DOMAIN_ID = 'Visit'
            AND STANDARD_CONCEPT = 'S') concept
        ON visit.VISIT_CONCEPT_ID = concept.CONCEPT_ID
        WHERE LOWER(concept.CONCEPT_NAME) LIKE '%@a%'
        ) VISIT_INFO
  JOIN
    ( SELECT *
      FROM person
          LEFT JOIN (SELECT LOCATION_ID, STATE FROM location) AS location ON
            person.location_id = location.location_id
          LEFT JOIN (SELECT CONCEPT_ID,
                        CONCEPT_NAME as RACE_CONCEPT_NAME
                      FROM concept WHERE DOMAIN_ID = 'Race'
                      AND STANDARD_CONCEPT = 'S') AS race ON
            person.race_concept_id = race.concept_id
          LEFT JOIN ( SELECT CONCEPT_ID,
                        CONCEPT_NAME as ETHNICITY_CONCEPT_NAME
                      FROM concept WHERE DOMAIN_ID = 'Ethnicity'
                        AND STANDARD_CONCEPT = 'S') AS ethnicity ON
            person.ethnicity_concept_id = ethnicity.concept_id
          LEFT JOIN (SELECT CONCEPT_ID,
                        CONCEPT_NAME as GENDER_CONCEPT_NAME
                      FROM concept WHERE DOMAIN_ID = 'Gender'
                      AND STANDARD_CONCEPT = 'S') AS gender ON
            person.gender_concept_id = gender.concept_id) DEMO_INFO
  USING(PERSON_ID)
  {@x} ? {GROUP BY DEMO_INFO.@concept_name
          ORDER BY COUNT DESC}
"
inpatient <- renderTranslateQuerySql(conn, 
                                     inpatient_sql, 
                                     x = F, a = "inpatient", 
                                     query = "COUNT(*) VISIT_COUNT, 
                                     COUNT(DISTINCT PERSON_ID) PERSON_COUNT
                                     ")
disconnect(conn)

In the inpatient cohort, there are a total of r inpatient$PERSON_COUNT distinct patients, and there are a total of r inpatient$VISIT_COUNT entries. Compared to the original unique person count in the overall dataset, which is r count, r count - inpatient$PERSON_COUNT are not accounted for in the inpatient count.

To further examine this data, we can break them down across the following axes:

• State
• Age
• Race
• Gender
• Care setting

State Breakdown

conn <- connect(connectionDetails)
inpatient_state <- renderTranslateQuerySql(conn, inpatient_sql, x = T, a = "inpatient",
                        query = "DEMO_INFO.state, COUNT(DISTINCT PERSON_ID)",
                        concept_name = "state")
disconnect(conn)

In the inpatient cohort, the top 5 states by person count are r knitr::kable(head(inpatient_state, n = 10))

Age Breakdown

conn <- connect(connectionDetails)
inpatient_age <- renderTranslateQuerySql(conn, inpatient_sql,
                                   x = F, a = "inpatient",
                                   query = "
                MIN(YEAR(VISIT_INFO.visit_end_date) -                                   
                  DEMO_INFO.year_of_birth) as min_age, 
                MAX(YEAR(VISIT_INFO.visit_end_date) - DEMO_INFO.year_of_birth) 
                  as max_age, 
                SUM(CASE WHEN YEAR(VISIT_INFO.visit_end_date) - 
                  DEMO_INFO.year_of_birth < 18 THEN 1 END) AS less_than_18,
                SUM(CASE WHEN YEAR(VISIT_INFO.visit_end_date) -                              
                  DEMO_INFO.year_of_birth BETWEEN 18 AND 24 THEN 1 END) AS age_18_to_24,
                SUM(CASE WHEN YEAR(VISIT_INFO.visit_end_date) -                              
                  DEMO_INFO.year_of_birth BETWEEN 25 AND 34 THEN 1 END) AS age_25_to_35,
                SUM(CASE WHEN YEAR(VISIT_INFO.visit_end_date) -                              
                  DEMO_INFO.year_of_birth BETWEEN 35 AND 44 THEN 1 ELSE 0 END) AS age_35_44,
                SUM(CASE WHEN YEAR(VISIT_INFO.visit_end_date) -                              
                  DEMO_INFO.year_of_birth BETWEEN 45 AND 54 THEN 1 ELSE 0 END) AS age_45_54,
                SUM(CASE WHEN YEAR(VISIT_INFO.visit_end_date) -                              
                  DEMO_INFO.year_of_birth BETWEEN 55 AND 64 THEN 1 ELSE 0 END) AS age_55_64,
                SUM(CASE WHEN YEAR(VISIT_INFO.visit_end_date) -                              
                  DEMO_INFO.year_of_birth >= 65 THEN 1 ELSE 0 END) AS age_65_and_more")
disconnect(conn)

In the inpatient dataset, it takes in the visit encounters, and the ages for the individuals were found according to the time of the inpatient visit. The youngest individual was of age r inpatient_age$MIN_AGE, and the oldest individual was of age r inpatient_age$MAX_AGE

Racial Breakdown

conn <- connect(connectionDetails)
inpatient_race <- renderTranslateQuerySql(conn, inpatient_sql, 
                                          x = T, 
                                          a = "inpatient", 
                                          concept_name = "RACE_CONCEPT_NAME", 
                                          query = "
                                          DEMO_INFO.RACE_CONCEPT_NAME, 
                                          COUNT(DISTINCT PERSON_ID)")
disconnect(conn)

In the inpatient cohort, there are r mapply(paste, inpatient_race$COUNT, inpatient_race$RACE_CONCEPT_NAME, "persons") r knitr::kable(inpatient_race)

Gender Breakdown

conn <- connect(connectionDetails)
inpatient_gender <- renderTranslateQuerySql(conn, inpatient_sql, 
                        x = T, a = "inpatient", 
                        concept_name = "GENDER_CONCEPT_NAME", 
                        query = "
                        DEMO_INFO.GENDER_CONCEPT_NAME, 
                        COUNT(DISTINCT PERSON_ID)
                        ")
disconnect(conn)

In the inpatient cohort, there are r mapply(paste, inpatient_gender$COUNT, inpatient_gender$GENDER_CONCEPT_NAME, "persons") r knitr::kable(inpatient_gender)

ohdsi-studies/MentalHealthEquity documentation built on March 9, 2023, 12:39 p.m.