Related Experiment Video
Updated: Nov 13, 2025
02:28
Author Spotlight: Self-Assessment Protocol for Predicting Psoriatic Arthritis in Psoriasis Patients
Published on: March 1, 2024
603
Validation of claims-based algorithms to identify patients with psoriasis
Hemin Lee1, Mengdong He1, Soo-Kyung Cho1,2
1Division of Pharmacoepidemiology and Pharmacoeconomics, Brigham and Women's Hospital and Harvard Medical School, Boston, Massachusetts, USA.
Pharmacoepidemiology and Drug Safety
|March 14, 2021
Summary
Accurate identification of psoriasis (PsO) patients is vital for real-world evidence. Claims-based algorithms combining diagnosis codes and treatments show moderate-to-high positive predictive value for PsO identification.
Area of Science:
- Real-world evidence generation
- Pharmacoepidemiology
- Health outcomes research
Background:
- Accurate identification of patients with psoriasis (PsO) is crucial for generating reliable real-world evidence on disease course and treatment patterns.
- Existing methods for identifying PsO patients in large datasets may have limitations in accuracy and efficiency.
Purpose of the Study:
- To develop and validate claims-based algorithms for identifying patients with psoriasis (PsO).
- To assess the positive predictive value (PPV) of these algorithms using electronic health records (EHR) data.
- To evaluate the disease activity of identified PsO patients.
Main Methods:
- Nine claims-based algorithms were developed using International Classification of Diseases (ICD)-9 codes, specialist visits, and medication dispensing data.
- Medicare data linked to EHRs from two healthcare networks (2013-2014) were utilized.
- The gold standard for validation was the treating physician's PsO diagnosis confirmed by chart review; PPV and 95% confidence intervals (CI) were calculated.
Main Results:
- The nine algorithms identified 990 unique patient records, with 918 (92.7%) reviewed.
- Algorithm positive predictive values (PPV) ranged from 65.1% to 82.9%.
- The highest PPV (82.9%) was achieved by an algorithm requiring at least one ICD-9 diagnosis code for PsO and one prescription claim for topical vitamin D agents. Algorithms combining diagnosis codes and specific treatments demonstrated moderate-to-high PPV.
Conclusions:
- Claims-based algorithms incorporating PsO diagnosis codes and specific treatment dispensing show moderate-to-high positive predictive value.
- These validated algorithms can be effectively utilized for identifying PsO patients in future real-world data pharmacoepidemiologic studies.
- The developed algorithms provide a valuable tool for enhancing the accuracy of PsO patient identification in large-scale research.
