OHDSI Home | Forums | Wiki | Github

HowOften: Community contributions wanted

Comparative Burden of Illness: An Analysis of Symptom Incidence and Health Service Utilization by Major Disease

Questions:
To assess and compare the burden of selected major diseases (e.g. right heart failure, inflammatory bowel disease, treatment-resistant depression) based on:

  1. The incidence of specific symptoms (e.g. dyspnea, chest pain, and cough) post diagnosis.
  1. Service utilization rates, specifically emergency room visits and inpatient services, after the index date.

Sub-questions:

  1. What is the differences in incidence of the specified symptoms and service utilization compared to others?
  2. Are there discernible trends in symptom incidence and service utilization across calendar years for each disease?
  3. How does the burden of disease differ by gender and across age groups?
  4. Are these rates in line with any expected published rates.

Justification: Understanding the variability in symptom incidence and service utilization by disease can provide insights into the differential burden these diseases impose on patients and the healthcare system. This knowledge can help policymakers, allowing for more informed decisions about resource allocation, tailored interventions, and research prioritization to mitigate the impact of these diseases.

FYI if anyone wants to read the protocol it is here: https://github.com/ohdsi-studies/HowOften/blob/master/Documents/HowOften%20protocol%20v1.0.pdf

It looks like there are two sources for target (a.k.a exposure) and outcome cohorts.

Source 1: phenotype library.
@Gowtham_Rao Should I assume that any cohort in the phenotype library with the value eventCohort = 1 is considered an outcome cohort and any cohort with eventCohort = 0 will be considered a target cohort? Iā€™m wondering how to know which cohorts in the library will be used for target and outcome cohorts. Also will every combination of target cohort with outcome cohort be used in the study or just a subset of possible target-outcome pairs?

Source 2:
Create a standard target cohort for every drug ingredient.
Create a standard outcome cohort for every snomed disorder (condition?) concept. Would it be possible to include a list of these concept IDs in the protocol? Iā€™m not 100% sure what is meant by disorder concepts.
Measure incidence of every combination of target and outcome pairs for all stratifications.

Am I understanding the study design correctly?

See here

1 Like
t