0
Obstructive Lung Diseases |

A Validated Claims-Based Prediction Model for COPD Severity

Dendy Macaulay*, PhD; Shawn Sun, PhD; Rachael Sorg, MPH; Sherry Yan, PhD; Gourab De, PhD; Eric Wu, PhD; Paul Simonelli, MD
Author and Funding Information

Analysis Group, Inc., New York, NY


Chest. 2012;142(4_MeetingAbstracts):675A. doi:10.1378/chest.1387734
Text Size: A A A
Published online

Abstract

SESSION TYPE: COPD: Severity and Risk Predictors

PRESENTED ON: Monday, October 22, 2012 at 11:15 AM - 12:30 PM

PURPOSE: Develop and validate a claims-based regression model to predict chronic obstructive pulmonary disease (COPD) severity using health insurance claims linked to spirometry results from electronic health records (EHR).

METHODS: Patients with ≥1 diagnosis of COPD (ICD-9: 491, 492, 496) and spirometry results from January 2004 to May 2011were identified from EHR data linked to healthcare claims data provided by Medmining. Claims data from 3 months prior to and following (observation period) the most recent spirometry test date (index date) were used to construct the model. Spirometry tests were excluded if the patient had an asthma diagnosis during the observation period or if the test was within 1 week of a COPD exacerbation. Patients were classified into one of three severity levels based on EHR spirometry results (FEV1 %predicted, FEV1/FVC) and GOLD guidelines: Unclassified, Mild/Moderate, and Severe/Very severe. Using a random selection of 90% of patients in each severity level (training sample), a multinomial logistic regression model was developed using a combination of clinical guidance and stepwise selection (thresholds of p-value<0.5 for variable selection/elimination) that evaluated patient characteristics recorded in claims as potential predictors of severity. Model predictions were evaluated for sensitivity/specificity/accuracy as well as concordance (using Cohen’s Kappa) between predicted and observed severity. The remaining 10% of patients (test sample) were used for validation.

RESULTS: 2,028 COPD patients met sample selection criteria: 886 Unclassified, 683 Mild/Moderate, 459 Severe/Very severe. The final model included demographics, comorbidities, COPD-related resource utilization, and all-cause healthcare visits. Within the training sample, the model correctly predicted COPD severity for 59.6% of all patients (accuracy for predicting Unclassified: 70.2%; Mild/Moderate: 66.6%; Severe/Very severe: 82.3%) with kappa=0.36. In the test sample, the model correctly predicted COPD severity for 60.5% of all patients (accuracy for predicting Unclassified: 71.2%; Mild/Moderate: 68.3%; Severe/Very severe: 81.5%) with kappa=0.38.

CONCLUSIONS: The prediction model provides a simple, validated method to classify patient COPD severity based on claims data.

CLINICAL IMPLICATIONS: Developing a validated claims-based COPD severity algorithm provides a useful tool for assessing COPD severity in the absence of clinical measures and identifying patients who require additional intervention.

DISCLOSURE: Dendy Macaulay: Consultant fee, speaker bureau, advisory committee, etc.: Consultant

Shawn Sun: Employee: Employee of Forest Research Institute

Rachael Sorg: Consultant fee, speaker bureau, advisory committee, etc.: Consultant

Sherry Yan: Consultant fee, speaker bureau, advisory committee, etc.: Consultant

Gourab De: Consultant fee, speaker bureau, advisory committee, etc.: Consultant

Eric Wu: Consultant fee, speaker bureau, advisory committee, etc.: Consultant

Paul Simonelli: Consultant fee, speaker bureau, advisory committee, etc.: Consultant

No Product/Research Disclosure Information

Analysis Group, Inc., New York, NY

Sign In to Access Full Content

MEMBER & INDIVIDUAL SUBSCRIBER

Want Access?

NEW TO CHEST?

Become a CHEST member and receive a FREE subscription as a benefit of membership.

Individuals can purchase this article on ScienceDirect.

Individuals can purchase a subscription to the journal.

Individuals can purchase a subscription to the journal or buy individual articles.

Learn more about membership or Purchase a Full Subscription.

INSTITUTIONAL ACCESS

Institutional access is now available through ScienceDirect and can be purchased at myelsevier.com.

Sign In to Access Full Content

MEMBER & INDIVIDUAL SUBSCRIBER

Want Access?

NEW TO CHEST?

Become a CHEST member and receive a FREE subscription as a benefit of membership.

Individuals can purchase this article on ScienceDirect.

Individuals can purchase a subscription to the journal.

Individuals can purchase a subscription to the journal or buy individual articles.

Learn more about membership or Purchase a Full Subscription.

INSTITUTIONAL ACCESS

Institutional access is now available through ScienceDirect and can be purchased at myelsevier.com.

Figures

Tables

References

NOTE:
Citing articles are presented as examples only. In non-demo SCM6 implementation, integration with CrossRef’s "Cited By" API will populate this tab (http://www.crossref.org/citedby.html).

Some tools below are only available to our subscribers or users with an online account.

Sign In to Access Full Content

MEMBER & INDIVIDUAL SUBSCRIBER

Want Access?

NEW TO CHEST?

Become a CHEST member and receive a FREE subscription as a benefit of membership.

Individuals can purchase this article on ScienceDirect.

Individuals can purchase a subscription to the journal.

Individuals can purchase a subscription to the journal or buy individual articles.

Learn more about membership or Purchase a Full Subscription.

INSTITUTIONAL ACCESS

Institutional access is now available through ScienceDirect and can be purchased at myelsevier.com.

Related Content

Customize your page view by dragging & repositioning the boxes below.

Find Similar Articles
CHEST Journal Articles
PubMed Articles
  • CHEST Journal
    Print ISSN: 0012-3692
    Online ISSN: 1931-3543