Public Reporting of Central-line Associated Bloodstream Infections (CLABSIs) using Administrative Billing Codes (ABC) versus Standard Infection Control (IC) Surveillance: Apples and Oranges

Background:

Administrative billing codes have been proposed as a method to capture hospital associated infections (HAIs) with intention of simplifying the resource-intensive process of traditional surveillance. Under pressure of recent changes in reporting regulations, several states are considering using electronic surveillance that incorporates administrative codes. However, the validity of using billing codes to identify HAI has not been established.

Objective:

Identify the number of CLABSI identified with ABC and/or IC surveillance using electronic databases. Compare the amount of concordant cases between the two methods.

Methods:

We performed a comparative analysis of CLABSIs detected using standard IC surveillance and administrative billing codes from 10/1/2007 through 12/31/2009 at three study hospitals (2 community and 1 tertiary care). CLABSIs diagnosed using standard IC surveillance were identified using the DICON Surveillance Database (2 community hospitals) and the Duke University Hospital (DUH) Surveillance Database.  Methods for standard IC surveillance for CLABSIs were identical at all 3 hospitals and followed CDC/NHSN definitions.  The Duke University Health System Enterprise Data Repository was queried for all technical billing cases with ICD-9 code 999.31, “Infection due to central venous catheter” for each of the 3 hospitals.  CLABSIs identified by each method were compared to determine the proportion of concordance between the two methods.

Results:

A total of 1490 unique CLABSIs were identified using both methods during the study period; 873 (64%) were identified by ABC and 745 (55%) were identified using IC surveillance. Only 128 (9%) of cases were concordant between the 2 methods.  In community hospitals, 155 (66%) were identified by ABC and 93 (40%) were identified using IC surveillance; only 15 (6%) were concordant. In the tertiary care hospital, 718 (57%) were identified by ABC and 652 (52%) were identified using IC surveillance; only 113 (9%) were concordant. Using CLABSIs identified by ABC instead of IC surveillance would increase the CLABSI episodes by 67% in the community hospitals, and 10% in the tertiary hospital.

Conclusions:

There was astoundingly little overlap in CLABSIs identified by the two methods. Additionally, counts of CLABSI were higher using ICD-9 codes than with IC surveillance method. These differences were evident despite the increase in education for billing personnel that occurred after Center for Medicare and Medicaid Services (CMS) rule for no reimbursement for HAIs went into effect. If adopted, reporting using ABC would erroneously raise the rate of CLABSI for these hospitals. Our results suggest that using ABC to identify CLABSIs is an unreliable and misleading method, and even less reliable when used in the community setting.