Identification of volume overload hospitalizations among hemodialysis patients using administrative claims: a validation study

Public Deposited
Analytics

Downloadable Content

Download PDF
Request Version for Screen Reader
Creator
  • Assimon, Magdalene M
    • Affiliation: School of Medicine, Department of Medicine, Division of Nephrology and Hypertension
  • Nguyen, Thuy
    • Affiliation: School of Medicine, Department of Medicine
  • Katsanos, Suzanne L
    • Affiliation: School of Medicine, Department of Medicine, Division of Nephrology and Hypertension
  • Brunelli, Steven M
    • Other Affiliation: DaVita Clinical Research, Minneapolis, USA
  • Flythe, Jennifer
    • Affiliation: Cecil G. Sheps Center for Health Services Research, School of Medicine, Department of Medicine, Division of Nephrology and Hypertension
Abstract
  • Abstract Background High rates of volume overload hospitalizations may indicate inadequate dialysis facility fluid management. Administrative claims databases are often used to study such outcomes, but these data are generated for billing purposes and may not capture clinical nuance. It is unknown if volume overload admissions can be correctly identified in administrative data and if a single claims-based definition for volume overload can be used across epidemiologic surveillance studies, observational studies of exposure-outcome associations and quality assessments. We conducted a validation study to assess the accuracy of claims-based definitions for volume overload hospitalizations among hemodialysis patients. Methods Data were taken from a random sample of 315 adult hemodialysis patients admitted to University of North Carolina Hospitals from January 2010 through June 2013. Standardized chart reviews were conducted to clinically adjudicate the presence or absence of volume overload at hospital admission. Claims-based definitions were constructed from varying combinations of fluid-related ICD-9 discharge diagnosis codes including fluid overload, pulmonary edema, pleural effusion, and heart failure. Using clinically adjudicated volume overload hospitalizations as the reference standard, validity metrics and their 95 % confidence intervals (CIs) were estimated for each definition. Results Of the 315 hospital admissions, 77 (24.4 %) were clinically adjudicated as volume overload hospitalizations. The prevalence of claims-identified volume overload admissions varied across definitions, ranging from 1.6 to 37.1 %. When definitions were constructed with discharge diagnosis codes present in any billing position, volume overload hospitalizations defined by fluid overload, pleural effusion or heart failure diagnosis codes had the highest sensitivity, 81.8 % (95 % CI: 71.4 %, 89.7 %). Volume overload hospitalizations defined by pulmonary edema diagnosis codes had the highest specificity, 98.3 % (95 % CI: 95.8 %, 99.5 %). Definitions constructed with discharge diagnosis codes present in any billing position (versus the primary position) captured more false positive events. Conclusions Prevalence and validity estimates of volume overload hospitalizations vary across claims-based definitions. A universal claims-based definition for volume overload hospitalizations may not apply to all clinical and research scenarios. Investigators and regulators need to consider the implications of misclassifying events when evaluating and monitoring hemodialysis patient volume overload admissions with administrative data. Claims-based definitions should be selected accordingly.
Date of publication
DOI
Identifier
Resource type
  • Article
Rights statement
  • In Copyright
Rights holder
  • The Author(s).
Language
  • English
Bibliographic citation
  • BMC Nephrology. 2016 Nov 11;17(1):173
Publisher
  • BioMed Central

Relations

Parents:

This work has no parents.

Items

Thumbnail Title Date Uploaded Visibility Actions
file details 12882_2016_article_384.pdf 2019-05-07 Public Download
file details 12882_2016_384_moesm1_esm.pdf 2019-05-07 Public Download
file details 12882_2016_384_moesm2_esm.pdf 2019-05-07 Public Download