Estimating Infection Rates

advertisement
Developing a Framework for Estimation of
Healthcare-Associated Infection Burden at the
National and State Level
Matthew Wise, MPH, PhD
Epidemiologist, Office of Prevention Research and Evaluation
CSTE Annual Conference
June 4, 2012
National Center for Emerging and Zoonotic Infectious Diseases
Division of Healthcare Quality Promotion
Importance of HAI Burden Estimates

Defining the public health impact of HAIs
 Morbidity, mortality, and cost
 Where is burden greatest?
 How should public health resources be allocated?

How has burden changed with implementation of
prevention programs or policies?

Useful communications tool
 Policymakers may relate better to numbers than rates
 Can aid in advocating for resources
Major Healthcare-Associated
Infection (HAI) Types

Device-associated infections:
 Bloodstream infections in patients with central lines (CLABSI)
 Urinary tract infections in patients with catheters (CAUTI)
 Pneumonias in ventilated patients (VAP)

Surgical site infections (SSI):
 Superficial and complex infections following surgical procedures

Multidrug-resistant and other important pathogens:
 Methicillin-resistant Staphylococcus aureus (MRSA)
 Clostridium difficle infections (CDI)
Previous HAI Burden Estimates
Expanding on Previous Burden Estimates

Ability to project to the state level

Focus on HAI types that are targets of prevention
efforts

Take advantage of more robust HAI data
What’s Changed?

HAI surveillance is much more comprehensive
 From hundreds of facilities in the 1990s to thousands of facilities
currently
 Data collected on a larger number of infection types

Greater access to National Healthcare Safety Network
data at the state level
 State reporting requirements
 Group user function
CMS Reporting Incentive Timeline
HAI type
Setting/description
Date implemented
CLABSI
Acute care hospital critical care units
January 2011
CAUTI
Acute care hospital critical care units
January 2012
SSI
Acute care hospitals: COLO and HYST
January 2012
Dialysis
Events
Outpatient dialysis centers: IV
antimicrobial starts, BSI, access infection
January 2012
CLABSI
Long-term acute care hospitals
October 2012
CAUTI
Long-term acute care hospitals
October 2012
CAUTI
Inpatient rehabilitation facilities
October 2012
MRSA BSI
Acute care hospitals: LabID event
January 2013
CDI
Acute care hospitals: LabID event
January 2013
Median State-Specific Percent of Acute Care
Facilities Participating in HAI Surveillance
2009
2011
CLABSI
9%
61%
CAUTI
6%
27%
SSI
4%
18%
VAP
7%
15%
CDI
0%
3%
A Common Approach

CDC and some states already producing HAI burden
estimates or exploring burden estimation

Benefits of a common (or at least coordinated)
approach:
 (Relatively) comparable estimates across states
 Internally consistent estimates (e.g., sum to the ~national total)
 Greater efficiency by developing standard methods and data
sources
What is needed to produce HAI burden
estimates?

Is there a source of data on the frequency of infections
that is generalizable to the population I want to
calculate burden for?
 Example: “Do I have information on the rate of CLABSIs in
hospitalized critical care patients in the United States?”

Do data exist to define the entire population at risk for
the outcome of interest?
 Example: “Do I know the total number of central line-days in
hospitalized critical care patients in the United States?”
Simple Approach to HAI Burden Estimation
Define the
denominator:
Estimate
infection rates:
Patient-days
Device-days
Procedures
CDI
CLABSI/CAUTI/VAP
SSI
Multiply
Number of
infections
Simple Approach to HAI Burden Estimation
Define the
denominator:
Estimate
infection rates:
Patient-days
Device-days
Procedures
CDI
CLABSI/CAUTI/VAP
SSI
Multiply
Number of
infections
Defining the Denominator:
Data Sources

AHRQ Healthcare Cost and Utilization Project

State hospital discharge data

CMS Healthcare Cost Reports
Defining the Denominator:
AHRQ Healthcare Cost and Utilization Project

Source of national data on non-Federal short-stay
community hospital discharges
 Also state-specific data available for 35 states


Information can be used to estimate patient-days and
surgical procedure denominators
HCUPnet web query system
http://hcupnet.ahrq.gov/
Defining the Denominator:
State Hospital Discharge Data




“Raw” state-specific discharge data files that HCUP
uses to create its databases
Data on patient-days and surgical procedures
Ability to design more complex queries
Can be difficult/cumbersome to access in some states
Defining the Denominator:
CMS Healthcare Cost Reports



Filed by all Medicare-eligible hospitals, nursing homes,
dialysis facilities, hospice, and home health agencies
Publicly available, but files difficult to work with
Patient-day data stratified by hospital type and critical
care status
http://www.cms.gov/Research-Statistics-Data-and-Systems/Filesfor-Order/CostReports/Cost-Reports-by-Fiscal-Year.html
Defining the Denominator:
Complications

General issues
 Most data sources exclude Federal facilities
 Administrative data can lag by 1-3 years

Device-associated infections
 Need to stratify patient-day denominators by critical care status
 Must take device utilization into account

Surgical site infections
 NHSN procedures may not map directly to ICD-9-CM procedure
codes used in hospital discharge data
An Example of Estimating Burden:
CLABSIs in Critical Care Patients, US, 2010
Estimate critical care
patient-days from CMS
Hospital Cost Reports and
inflate by 4% to account for
Federal hospitals
20.8 million*1.04=21.7 million total patient-days
An Example of Estimating Burden:
CLABSIs in Critical Care Patients, US, 2010
Obtain device utilization
ratio from NHSN and
convert patient-days to
central line-days
21.7 million*0.50 =
10.8 million central line-days
21.7 million US critical care patient-days
Simple Approach to HAI Burden Estimation
Define the
denominator:
Estimate
infection rates:
Patient-days
Device-days
Procedures
CDI
CLABSI/CAUTI/VAP
SSI
Multiply
Number of
infections
Estimating Infection Rates:
Data Sources

Hospital discharge data

Emerging Infections Program

National Healthcare Safety Network (NSHN)
Estimating Infection Rates:
Discharge Data

Few HAIs can be accurately identified using
administrative data sources

CDI
 ICD-9-CM code 008.45 does a reasonable (but not perfect) job of
identifying CDI
 Primary diagnosis correlated with community-onset infection
 Secondary diagnosis correlated with hospital-onset infection

Some surgical site infections
 Example: Some success in identifying post-CABG mediastinitis
using a combination of ICD-9-CM diagnosis and procedure codes
Estimating Infection Rates:
Emerging Infections Program

Captures infections occurring in community and
healthcare settings
 Rates generally calculated per 100,000 population

Active Bacterial Core Surveillance (ABCs)
 Invasive MRSA surveillance

Healthcare-Associated Infections-Community Interface
 CDI surveillance
 HAI and antimicrobial use prevalence survey of hospitalized
patients
Estimating Infection Rates:
National Healthcare Safety Network

Voluntary, incentivized, and mandatory reporting of
HAIs to CDC by healthcare facilities and organizations

Outcomes under surveillance (selected):
 Hospital-onset CLABSI, CAUTI, and VAP rates per 1,000 device-days
 Surgical site infections per 1,000 procedures (40 different
procedure types)
 Dialysis events (IV antimicrobials, BSI, access infection) per 100
patient-months by vascular access type
 Multidrug-resistant organism and CDI rates based on patient-days
or admissions
Estimating Infection Rates:
Complications

Discharge data is useful in only specific circumstances

EIP data only collected from (at most) ten geographic
areas and may not represent the locality for which
estimates are being generated

NHSN
 The units/facilities participating in surveillance may be
systematically different than non-participants
 Reported data from participants may not represent “ground truth”
 Primarily captures infections with onset in hospitals and other
inpatient healthcare facilities (some exceptions)
Simple Approach to HAI Burden Estimation
Define the
denominator:
Estimate
infection rates:
Patient-days
Device-days
Procedures
CDI
CLABSI/CAUTI/VAP
SSI
Multiply
Number of
infections
An Example of Estimating Burden:
CLABSIs in Critical Care Patients, US, 2010
~16,000
critical care
CLABSIs in 2010
Multiply critical care
CLABSI rate by central linedays to estimate infections:
*10.8 million*(1.46/1000)
10.8 million US central line-days
21.7 million US critical care patient-days
Additional Considerations

For point estimates
 Is infection data representative of in my entire jurisdiction
 Are there reasons the data might not represent “ground truth”?

When examining trends
 Definition and surveillance system changes
 Changes in the types of units/facilities participating in surveillance
 Growing “at risk” population  may need a counterfactual
comparison

Uncertainty
 Sensitivity analyses
 Monte Carlo simulation
Summary

HAI infection rate data is increasingly robust enough to
produce estimates at the state level
 More infection types
 Greater number of settings

Numerous supplemental (and often publicly available)
data sources exist to facilitate extrapolation of
infection rates to estimate burden at the state level
Future Burden Estimation Efforts

When can we just start counting infections reported to
NHSN?

How can reliable estimates be produced for less
populous areas?

Can we produce more comprehensive HAI burden
estimates (i.e., less piecemeal)?

Could state-specific HAI denominators (e.g., patientdays, device-days, procedures) be made publicly
available?
Contact Information:
Matthew Wise, MPH, PhD
Prevention and Response Branch
Division of Healthcare Quality Promotion, CDC
cxx4@cdc.gov
For more information please contact Centers for Disease Control and Prevention
1600 Clifton Road NE, Atlanta, GA 30333
Telephone: 1-800-CDC-INFO (232-4636)/TTY: 1-888-232-6348
E-mail: cdcinfo@cdc.gov Web: http://www.cdc.gov
The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the
Centers for Disease Control and Prevention.
National Center for Emerging and Zoonotic Infectious Diseases
Division of Healthcare Quality Promotion
Download