Administrative or clinical data?

Hospitals within the Dallas-Fort Worth region have reduced Catheter Associated Urinary Tract Infection (CAUTI) rates based upon the Center for Medicare and Medicaid Services (CMS) definition significantly since 2010, the baseline year for the CMS Hospital Engagement Network (HEN) contract.



Utilizing CMS definition for Hospital Acquired Condition (HAC) 7, the decrease is greater than 70 percent. National Healthcare Safety Network (NHSN) CAUTI Standard Infection Ratio (SIR) started being tracked for ICU in 2012. NHSN SIR shows an increase in CAUTI SIR both nationally and regionally.

In an attempt to explain differences in CAUTI Rates vs NHSN SIR, a comparison was done utilizing administrative data. Within the data set, the rate was based upon CMS administrative collection for CAUTI with the denominator being acute patient discharges rather than device days. NHSN SIR is a ratio of the count of actual incidences/projected incidences. Within large databases, the trend line is generated from multiple like facilities within a region. The projected trend line for this comparison was based upon the incidence within the region.

By utilizing our data resource, our graph for 2012 approximates the NHSN SIR graph, but by extending the comparison back to 2011, the regional SIR showed a decrease of 20 percent. As improvements are implemented and facilities approach zero, comparing actual rates to predicted rates can lead to higher ratios until the trend line adjusts to the slope.

Within healthcare there is an ongoing debate related to data and data sets. One camp insists the most accurate and therefore best data comes from clinical sources. Clinical sources can be extractions from electronic health/medical records or clinical personnel performing manual reviews regularly depending on the data collected. One such database is the National Healthcare Safety Network (NHSN). Staff members collect data meeting definitions and perform manual data entry into the repository.

Administrative data is based upon claims data which includes diagnosis codes for the hospital or physician visit. Currently, the International Classification of Diseases book 9 (ICD-9) is used in the U.S. while ICD-10 was endorsed by the World Health Organization (WHO) and by 1994 was used by many European states. In a search for comparison studies, most articles originating in the U.S. identify clinical data as better at identifying higher risk patients and specific complications while British and European journals find differences to be insignificant. In 1993, CMS turned to clinical data for improved identification of risk adjusted mortality. ICD-9 was adopted in 1975 and contains 17,000 codes. ICD-10 contains 155,000 codes. ICD-11 is currently in development, with 2015 the anticipated year for endorsement.

With implementation of ICD-10, which allows greater specificity with significantly more coded events available, the perceived difference between administrative and clinical data may become moot.