Vous êtes sur la page 1sur 63

Practical Strategies for Surveillance

Dr. Sanjeev Singh MBBS, DCH, MD, Mphil PGDMLS, PhD

SHEA-CDC TRAINING COURSE IN HEALTHCARE EPIDEMIOLOGY

David K. Warren, MD, MPH


Assistant Professor of Medicine Washington Univ. Sch. of Medicine Hospital Epidemiologist, Barnes-Jewish Hospital

SHEA-CDC TRAINING COURSE IN HEALTHCARE EPIDEMIOLOGY

Learning Objectives
Identify data sources for surveillance Identify important components of a surveillance system List questions infections control programs should answer when evaluating their surveillance system

SHEA-CDC TRAINING COURSE IN HEALTHCARE EPIDEMIOLOGY

He uses statistics as a drunken man uses lamp-posts - for support rather than illumination.
Andrew Lang
(18441912), Scottish author

Overview
Historical perspective Designing surveillance systems
Choosing targets Definitions & Denominators Data sources Case-finding methods

Recording & analyzing data Presenting surveillance data Evaluating surveillance systems

Core Functions of Infection Control Programs


Conduct surveillance Investigate outbreaks/clusters Report reliable data Provide information and expert guidance Train/motivate direct patient caregivers Develop policies and procedures Develop interventions to reduce infections

Historical Perspective
1958: AHA recommends routine hospitalacquired infection surveillance
(AHA Advisory Committee, Conference on Staphylococcal Disease)

1969: CDC initiates CHIP


(Comprehensive Hospital Infections Project)

1970: CDC recommends ICPs and hospital epidemiologists


National Nosocomial Infection Surveillance System (NNIS)
Garner et al, Proceedings from International Conference on Nosocomial Infections 1970

Historical Perspective (contd)


1974-83: Study on the Efficacy of Nosocomial Infection Control (SENIC)
32% of nosocomial infections (NI) preventable Different combinations of IC practices reduced infections at each site Surveillance was the component necessary to reduce infections at each site

2002: Illinois mandated reporting of healthcare-associated infections to state 2005: NNIS becomes NHSN
Haley et al, Am J of Epidemiol, 1985;121

Increasing Public Awareness of Healthcare-associated Infections


36 states have either enacted or have pending legislation mandating public reporting of HAI (Including Missouri) Deadly Hospital Infections: Which Hospitals Are Best, Worst And How You Can Tell KSDK Ch. 5

Feedback of surgical site infection rates can reduce infection rates: KISS data

Adjusted Odds Ratio

* Significant after adjusted for procedure type, age, duration of surgery, wound class, ASA score, & hospital size
Brandt C et al. Infect Control Hosp Epidemiol 2006;27:1347-51

Essential Elements of Surveillance


Assess population Select event or process to survey Choose surveillance methods, including risk adjustment Monitor for event or process Apply surveillance definitions Calculate rates & analyze surveillance data Report & use surveillance findings
Adapted from Lee TL et al , AJIC 1998;26:277-88

Before beginning surveillance, ask...


Why do we need surveillance data? What are we going to do with it? Are the data available? Who will collect the data? How? Are the data likely to influence change?

Designing A Surveillance System


Focus on target events that:
Occur in a high volume population Can be prevented Occur frequently Cause serious morbidity Increase mortality Increase length of stay Are costly to treat Are amenable to change

Extent of Surveillance
Hospital-wide, traditional Periodic surveillance Prevalence surveys Targeted surveillance Outbreak thresholds
Broad St. pump cholera outbreak, 1854

Example - Large Adult University Hospital


Entire Hospital
C. difficile VRE MRSA* TB*

Surgical services
Spinal surgery CABG* Knee & hip* implants Abd hysterctomy* C-section Outpt ortho procedures

Oncology/BMT
BSI Aspergillus

ICUs
VAP BSI*
* State-mandated reporting

OH indicators
TB exposures Flu vaccination rate Body sub exposures

Example--Small Private Hospital


Entire hospital
Primary BSI HAP MRSA* C. difficile

Surgical services
Spinal surgery Knee implants Arthroscopy Lap gastric bypass

ICU
Primary BSI* VAP

OH indicators
Flu vaccination rate PPD testing rate

* State-mandated reporting

Example of choosing targets:


BJC Healthcare Infection Control 1997 Review of system-level Infection Control activities conducted - findings:
64 indicators tracked across system Collecting too much data No consistent criteria used to select IC surveillance indicators

Example of choosing targets:


BJC Healthcare Infection Control Developed criteria-based indicator selection process & tools Decrease in indicators tracked across system to 16 Decreased number of variables collected for routine surveillance Increased time for IC to focus on education & interventions

ICP Activities - Time Allocation


e c n a l l i e v r u S
10% 10%

n o i t a c u d E . n i m d A m a r g o r P

40%

n o i t ka ai g et r s be t v un OI

25% 15%
Source: BJC HealthCare 1997

r e h t O

ICP Activities - Time Allocation


e c n a l l i e v r u S
10% 25%

n o i t a c u d E . n i m d A m a r g o r P

20%

n o i t ka ai g et r s be t v un OI s n o i t n e v r e t n I

10% 15%

20%

r e h t O

Source: BJC HealthCare 2000

Components of Surveillance
Definitions Case finding and data collection Data entry or tabulation Data analysis Data interpretation Data reporting

Components of Surveillance
Case Definitions
Written Reviewed Applied consistently Imprecise definitions can lead to incorrect conclusions

Components of Surveillance
Case Definitions
Review CDC/NHSN definitions
(http://www.cdc.gov/nhsn/PDFs/pscManual/17pscNosInfDef_current.pdf)

Use exactly as written Use some of the definitions Adapt or modify the definitions Use definitions used by comparison hospitals or system

Components of Surveillance
Denominator Data
Measure either the population at risk and/or the amount of risk faced by the population Examples Count of surgical procedures in a month is the population at risk for developing a SSI associated with surgical procedures that month The sum of all central line days in a month for all patients with central lines in the ICU is the amount of risk (device days) faced by ICU patients during that month

Determining Appropriate Denominators


If infection risk is related to exposure to a single event, use the count of those events (e.g., # admissions, # procedures) If risk of infection relates to the duration of exposure over time, use the sum of the exposure time (e.g., central line days, ventilator days, patient or ICU days)

Collection of Surveillance Data


Collect only data you will use!! Collect data needed to document the infection (based on case definitions) Collect data important to identify risk factors (ex., NHNS risk index) Collect data important to healthcare providers

Collection of Surveillance Data


Patient demographics
Name, medical record number, age, sex, ward, admission, transfer and discharge dates, attending physician

Infection
Signs, symptoms, onset date

Relevant laboratory or diagnostic data


Culture & sensitivity reports, CBC, UA, x-ray reports

Exposure to risk factors


Invasive devices surgical procedures, implants, etc.

Collection of Surveillance Data


Who should collect case (numerator) data? Passive surveillance - unit MDs or RNs report infection to epidemiology team
Generally results in under-reporting of infections May result in bias

Collection of Surveillance Data


Who should collect case (numerator) data? Active surveillance - trained epidemiology staff identify infections using medical records, unit rounds & discussions with patient care staff
Generally results in more accurate identification of infections Better surveillance consistency over time

Collection of Surveillance Data


Who should collect denominator data? Use existing records/logs if possible
Unit admission logs OR procedure logs Respiratory care ventilator use logs

If logs unavailable, partner with unit staff to develop methods


Develop log books for daily count of ICU patients w/ CVCs, have unit staff record information

Collection of Surveillance Data


When should infection data be collected? Retrospective - staff trained in applying surveillance guidelines review medical records after patient discharge, complete case forms if criteria are met
Depends heavily on completeness of the record Little opportunity to supplement data May be excessive gap between occurrence of infections & when they are discovered Commonly used in outbreak investigation

Collection of Surveillance Data


When should infection data be collected? Prospective - IC staff make ward rounds, review medical records and talk to caregivers while the patient is still in hospital
Medical record supplemented by other data Increases awareness of unit staff Allows earlier identification of clusters Give opportunity to observe unit practices More resources needed Prospective studies - tracking outcomes over time after common exposure

Case Finding Methodology


Method Sensitivity (%)

Physician self-report forms Fever Antibiotic use Fever plus antibiotic use Microbiology reports Selected chart reviews Total chart reviews Standard

14-68 47 48 59 33-65 85 90 100

Sources: S. Streed 2002; Heipel D et al. Am J Infect Control 2007

Sources of Surveillance Data


Patient medical record Activity logs (O.R, labor & delivery, ICUs) Diagnostic and operative reports Microbiology, laboratory & autopsy reports Radiology reports Pharmacy records Admission Department (admit diagnosis) Billing Department (ICD-9 codes) Risk Management Unit rounds by epidemiology staff
Discussions with unit staff Direct observation

Surveillance Case-finding Methods


Total chart review ?? gold standard Total chart review 74-94% sensitive
Records did not document all necessary data, e.g., laboratory reports missing Records were not available Reviewer could not examine patient

Total chart review


Time consuming Review many records for pts w/o infection

Case-finding Methods
Laboratory Records
Clinical laboratory reports are a primary source for identifying infections Results prompt chart review Good source for identifying BSIs, resistant organisms; not good for SSIs, pneumonias, UTIs Sensitivity depends on the number of infections from which cultures are obtained & the culture methods

Case-finding Methods
Selected Chart Review
Multicenter study compared sensitivity of:
Routine IC surveillance vs. Post-op antibiotic exposure within 60 days of surgery, + ICD-9 discharge dx consistent w/ infection

Evaluated CABG (9 abx days), C-sections (2 abx days), & Breast surgery (6 abx days) SSIs
Yokoe D et al. Emerging Infect Dis 2004

Sensitivity* of IC routine surveillance vs. screening by abx exposure + ICD-9 codes for identifying SSI
Routine surveillance CABG 59% Abx exposure 91% ICD-9 code Abx + ICD-9

54%

93%

C-section

38%

84%

78%

97%

Breast surgery

33%

94%

70%

96%

* Compared to a gold standard comprised of all infections identified either during prospective surveillance or medical record review

Yokoe D et al. Emerging Infect Dis 2004

Case-finding Methods
Computerized Screening
Advantages
Automatic Use data in available databases Provides data with little effort after programming is completed

Disadvantages
Accuracy of the data in other data bases cannot be assumed Necessary data may not be available in computer databases

Example of computerized screening: GermWatcher


A culture-based surveillance system Uses micro data downloaded from the BJH mainframe Applies coded rules, based on NNIS criteria to distinguish between contaminants vs. likely true nosocomial infections. In validation studies, GermWatcher coded cultures the same as an ID specialist 96.4% of the time.
Kahn et al. J Am Med Inform Assoc 1996

Using computer algorithms to detect catheter-associated BSIs

Trick W et al. Emerging Infect Dis 2004

Case-finding Methods
Post-Discharge Screening
Supplement for inpatient case-finding Useful for post-op, postpartum, & neonates Challenging - decreased LOS, fragmented healthcare delivery Surgical implant SSI surveillance (1 yr.)

Case-finding Methods
Post-Discharge Screening (contd)
Various methods
Direct patient contact (phone, postcard, etc) Physician contact ED & hospital admission records Computerized screening methods (HMOs)

No agreed upon standard exists patient & MD surveys have sensitivity of 15-64%
Sands et al. J Infect Dis 1996; Heipel D et al. Am J Infect Control 2007;35:2002.

Recording Surveillance Data

Recording Surveillance Data


Develop a standardized form for collecting data Include all desired data elements Types of forms:
Line listing - allows recording of many patients on one page Case form - allows recording of many data elements about a single patient

Type of data collection form depends on # of data elements needed Computerized databases/spreadsheets should be utilized (ease of manipulation and analysis)

Segment of a Case Form:


Host Factors Cancer Diabetes Drug Abuse ETOH Abuse Obesity Treatments Central Line Respirator Surgery Infection Type N C Infection Site Bloodstream Lower Resp SSI UTI Name Adm Dr Unit __________ __________ __________ __________

Infection Signs Organism Drainage E. coli Erythema P. aeruginosa Fever >38.5 S. aureus

Data Analysis
Consider Stratification (Grouping) of Data Population under study often not homogenous
Necessary to control for intrinsic and some extrinsic risk factors when comparing rates Analysis within a risk strata helps avoid problem of confounding

Stratification of SSI Risk Using the NHSN Surgical Site Infection Risk Index
Wound Class If clean or clean-contaminated If contaminated or dirty infected American Society of Anesthesiologists physical status (ASA) score If 1 or 2 If 3, 4, or 5 Duration of Surgery If < Time T* If >= Time T* Score 0 1

0 1 0 1

Range of scores is from 0 to 3, and risk of infection increases as the score increases *T = the 75th %tile for the duration of the procedure in NNIS hospitals, rounded to the nearest hour

Presentation of Data: Epicurve


P. aeruginosa Nosocomial Bacteremia January 1995- January 2004
5

No. cases

0
Y E Y E Y E Y E Y E Y E 0 Y E 5 9 4 7 Y E 8 2 Y E 3 6 1 -9 M S n-9 M S n-9 M S n-9 M S n-9 M S n-0 M S n-0 M S n-0 M S n-0 M S n-0 n Ja Ja Ja Ja Ja Ja Ja Ja Ja Ja

Month

Presentation of Data: Incidence density


P. aeruginosa Nosocomial Bacteremia January 1999- January 2004

6.0 Rate Mean

No. bacteremia/1,000 pt. days


Mean=0.4

5.0

4.0

3.0

2.0

1.0
Mean 0.4

0.0

JY

JY

JY

O Ja C n02

JY

O Ja C n01

O Ja C n03

JY

O Ja C n00

Ja n99

Month

O Ja C n04

A P

A P

A P

A P

A P

Surgical ICU Primary Bloodstream Infection Rates & Femoral Line Utilization Percentage 2002-2004
40 35 BSI Rate (per 1000 line days) 30 25 20 20 15 15 10 5 0 Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan 10 5 0 35 30 25

2002 BSI Rate BSI Mean

2003 Fem Line %

2004 Fem Line % Mean

Femoral Line Utilization Percent

Points to remember with presenting data


Present data!!
Medical & nursing staff IP committee Hospital administration

Have a consistent data presentation format

Comparing Data
Comparisons are valid only if all parties:
Used the same surveillance intensity Used the same data collection methods Used the same definitions Risk-adjusted for differences in population

Pitfalls of Comparing Data


Trauma service closes

Ongoing Evaluation of a Surveillance System


Did system detect clusters/outbreaks? Did you change patient care practices? Were data used to:
Decrease endemic rate? Assess interventions? Ensure that rates did not increase when P/P changed?

Do administrative & clinical staff use data?

Ongoing Evaluation of a Surveillance System


Can you reduce or stop some aspect? Can you automate screening for high risk patients? Can you automate data collection?

Take Home Messages


Select surveillance projects to meet needs of your institution Design a flexible system by combining methods of surveillance & case finding Use computerized databases

Take Home Messages


Take advantage of information technologies to automate aspects of surveillance Analyze & report your data Use your data to improve patient care

Early Infection Control

Thank you

I dont think its sanitary to drink from the same footprint

Vous aimerez peut-être aussi