Claims Based Indicators of Cardiac Surgical Site Infection

Profiling hospitals' risks of surgical site infections after cardiac procedures

Susan Huang, James Livingston, Richard Platt
Eastern Massachusetts Prevention Epicenter
Department of Ambulatory Care and Prevention,
Harvard Medical School and Harvard Pilgrim Health Care,
Channing Laboratory, Department of Medicine,
Brigham and Women's Hospital,
Harvard Medical School
Boston, MA

Supported in part by grant UR8/CCU115079 from the Centers for Disease Control and Prevention

For additional information, contact:
Richard_Platt@harvard.edu
Jim.Livingston@channing.harvard.edu (technical issues)

This work is licensed under the Creative Commons Attribution License.
To view a copy of this license, visit http://creativecommons.org/licenses/by/2.0/
or send a letter to Creative Commons, 559 Nathan Abbot Way, Stanford, CA, 94305, USA.

PDF of the entire manual  |  README PDF  |  SAS Programs  |  Datasets  |  Test Runs  ]

Goal

Identify hospitals that may have unusually high or low risks of surgical site infection following invasive and non-invasive cardiac procedures.

Rationale

A typical health plan's claims data contain indicators of infections following surgery that can be used to estimate hospital-specific infection rates. These indicators include selected diagnoses, procedures, and prescriptions for antibiotics. These claims-based markers of infection can be used to guide further assessment of hospitals with potentially elevated risks of infection. Hospitals identified as having high percentages of patients with likely infection through this mechanism may not necessarily have high risks; additional evaluation would be required to determine the actual risk, and then to determine whether these risks could be reduced. This use of claims data has been described by Platt et al. (Emerg Infect Dis. 2002;8:1433-41, http://www.cdc.gov/ncidod/EID/vol8no12/02-0039.htm).

Routine assessment of hospitals' infection indicator scores may complement existing patient safety programs. Hospital-based infection control programs typically identify only a minority of infections, since most infections occur after discharge from the hospital and infected patients typically do not return to the hospital for care. In addition, it is difficult to compare hospitals' reported infection rates because hospital-based surveillance programs use different methods to identify infections.

Health plans' benefits and claims systems differ in ways that affect the accuracy (predictive value) of these indicators as predictors of infections. However, because differences between hospitals appear to be relatively consistent across health plans, it is possible to control for planto-plan differences. Information can also be combined and compared across health plans.

How these programs work

Every six or twelve months, health plans identify patients who have had a qualifying cardiac procedure since the previous time profiles were created. The initial profile should include all historical data that is readily accessible to the health plan.

Health plans create the following files for each patient who underwent cardiac procedures within the period of interest:

The included programs accomplish the following:

How these programs should be used

We recommend running these programs every six or twelve months and comparing data across half-years. As a general guideline, when comparing two half-years, at least 500 total procedures are needed for these analyses to be meaningful. In addition, at least 4 hospitals with at least 40 procedures each are most likely needed. Hospitals identified as possibly having higher than expected risks should be assessed further.

It is likely to be worthwhile for health plans operating in a single region to pool their tabulated results, in order to obtain more precise estimates of hospital-specific rates.

Interpreting and using the results

Remember that these claims based results are surrogate markers of infection. On average, approximately half of the patients with one of these indicators has the infection confirmed when the medical record is reviewed. Given this fact, these claims-based surrogate infection rates cannot be compared to rates reported by individual hospitals. Instead, the value of this analysis is in ranking hospitals according to their surrogate infection rates and in using a high threshold to guide further inquiry about the possibility that a hospital's rate is actually unusually high.

A high rate of these surrogate measures does not prove that a problem exists.

Some reasons that a high rate of these surgical site infections might be misleading include: