Online Program

Identification of Deliveries in Claims Datasets

Tuesday, November 3, 2015

Linh Nguyen, M.S, Division of Management, Policy and Community Health, University of Texas School of Public Health, Houston, TX
Suja Rajan, PhD, Division of Management, Policy and Community Health, University of Texas School of Public Health, Houston, TX
Ellerie Weber, PhD, Division of Management, Policy and Community Health, University of Texas School of Public Health, Houston, TX
Cecilia Ganduglia, MD DrPH, Management Policy and Community Health Department, University of Texas Health Sciences Center Houston, School of Public Health, Houston, TX
Maria Ukhanova, MD, MPA, School of Public Health. Management, Policy & Community Health Devision, University of Texas Health Science Center at Houston, Houston, TX
Correct identification of events in administrative claims datasets has become critical given its increased utilization in research.

As part of a study on deliveries, we used Blue Cross Blue Shield Texas (BCBSTX) and Medicaid Texas datasets to identify births occurring between 2007 and 2012. We selected validated algorithms based on clinical diagnosis (ICD-9-CM), procedures and diagnostic related groups (DRGs) codes. These were applied to all claims data (BCBSTX and Medicaid) as well as to Medicaid managed care encounter data. The analysis was limited to facility claims in women aged 12 to 55. We accounted for code modifications that occurred during the studied period. We compared the total number of deliveries identified using each search algorithm and analyzed the reasons behind the differences.

Among BCBSTX, 154,256 deliveries were identified using DRGs, 156,158 using ICD-9-CM and 140,118 using relevant ICD-9 procedures. On the Medicaid FFS, we detected 583,976 deliveries using DRGs and 27,526 more when using selected ICD-9-CM. Among encounters, however, there were considerable differences, with 285,927 deliveries identified using DRGs and 486,859 deliveries using ICD-9-CM.  The over 200,000 cases difference was mainly driven by missing DRGs among encounter observations (96% of all not identified with DRGs). Other reasons include non-delivery related DRGs and errors in data entry (partial codes and non-valid characters).

Selection of algorithms did not considerably affect delivery identification among commercially insured population but did affect that in Medicaid. It appears that best methods for identifying cases vary across types of administrative data as well as payers.

Learning Areas:

Conduct evaluation related to programs, research, and other areas of practice
Public health or related organizational policy, standards, or other guidelines

Learning Objectives:
Compare different algorithms to identify deliveries in private and public administrative claims datasets

Keyword(s): Women's Health, Medicaid

Presenting author's disclosure statement:

Qualified on the content I am responsible for because: My scientific interests include study of the prevalence of multiple chronic conditions in the adult population, and the economics of chronic care management as well as implementing algorithms to identify health conditions in administrative data. My ongoing research (leading to doctoral dissertation) focuses on studying the multiple chronic conditions in the commercially insured working–age adults with emphasis on economics of care coordination.
Any relevant financial relationships? No

I agree to comply with the American Public Health Association Conflict of Interest and Commercial Support Guidelines, and to disclose to the participants any off-label or experimental uses of a commercial product or service discussed in my presentation.