Administrative data have high variation in validity for recording heart failure.

Area of Science:

• Health Informatics
• Clinical Epidemiology
• Public Health Data

Background:

• Administrative data is frequently utilized for identifying patients with heart failure (HF).
• The accuracy of administrative data for HF case ascertainment requires systematic evaluation.

Purpose of the Study:

• To conduct a systematic review of studies assessing the validity of administrative data for recording heart failure (HF).
• To evaluate the performance of International Classification of Diseases (ICD) codes in administrative datasets for HF identification.

Main Methods:

• Systematic review of English peer-reviewed articles published between 1990 and 2008.
• Inclusion of studies validating ICD-8, -9, and -10 codes from administrative data.
• Analysis of HF code frequencies in Canadian hospital discharge data (2000-2006).

Main Results:

• Over 70 different ICD codes were used to define HF across 25 studies.
• Sensitivity ranged from 29% to 89%, while specificity and negative predictive values generally exceeded 70%.
• Positive predictive values varied widely (12% to 100%), with kappa values ranging from 0.39 to 0.84.

Conclusions:

• International Classification of Diseases (ICD)-9 and -10 codes 428 and I50 are the most prevalent for defining HF in hospital discharge data.
• The validity of administrative data for recording HF is inconsistent and depends on the data source and specific codes used.