Get documents about "How to calculate a rate
Document Sample
How to interpret nosocomial
infection rates ?
Joseph HAJJAR (Valence)
&
Anne SAVEY (Lyon)
How to calculate a rate ?
Numerator
Rate =
Denominator
• Difficulties to obtain this indicator
– Organization of data collection
– Case-finding / numerator
– Agreement on the denominator
1
Data collection
• Multiplicity of “time and place”
– Surgical unit (hospitalization)
– Operating theater (procedure)
– Consultation (follow-up)
– Laboratory (diagnosis)
• Multiplicity of actors
– Health care workers
– IC committee / IC Team (CLIN / EOH)
– Surveillance supervisor
– Network coordinator
• Multiple sources of information
Numerator identification
Case-finding
• Definitions of infections
– Lack of gold standard for SSI
– Subjectivity of criteria
• Diagnosis validated by the surgeon only
• Duration of the follow-up
– From 30 days to 1 year
• Post-discharged surveillance
– From 40 to 50 % of SSI occur after discharge
– Shorter length of stay, ambulatory
RPC. Infections nosocomiales : comment interpréter les taux ? L’exemple des ISO.
ANAES. Mars 2003
2
Agreement on the denominator
• Definition of the population under surveillance
– All surgical specialties
• Even those at very low risk ?
– All procedures from a given specialty
• Or focusing on a representative single procedure ?
– All wound class contaminations
• Including "contaminated" (3) and "dirty/infected" (4) ?
A rate, what for ?
• Having an indicator included
– In an internal evaluation process
– For a continuous quality improvement of quality
and security of care
Final objective = To decrease infection rates
Risk
identification
Evaluation Information
of their impact & training
Implementation of of HCW
corrective measures
3
A rate, what for ?
• Reduction of NI (%) according to the type of
control program (Senic Project – 1985)*
surveillance with feed-back to surgeons 20
SSI & implementation of control measures
+ a trained infection control team 35
UTI hospital-wide surveillance since one year 38
and 1 IC Nurse / 250 beds
PNE … …
*The efficacy of infection surveillance and control programs in preventing nosocomial infections in
US hospitals. Haley et al. Am J Epidemiol 1985;121:182-205
A rate, how to interpret it ?
• First, do not forget what it represents and
what can generate variations
– Factors influencing exhaustiveness and data
quality
– Factors associated to infectious risk (case-mix)
– Factors associated to quality of care
4
Exhaustiveness & quality of data
SSI + SSI -
• Numerator: infections Surv+ a b
(true+) (false+)
– Were all SSI declared ? Surv - c d
• Sensitivity = true + / SSI+ (false-) (true-)
• = a / a+c
– Did I declare wrong SSI ?
• Specificity = true - / SSI-
• = d / b+d
• Denominator: population
• Exhaustiveness = % of patients included
• Completeness = % of complete files or % missing data
• Correctness = % of correct data
What is behind a rate ?
• Numerous risk factors
– Endogenous:associated to patient
– Exogenous: associated to the procedure
• Risk adjustment
– NNIS Index
• Take into account of 3 independent risk factors
– WCC, ASA score, duration of procedure
– Homogenous types of patients (case-mix)
5
Host Risk Factors*
Relationship between host risk factors and SSI
(1) definite (2) likely (3) possible
low albumin (2)
immunosuppression
malnutrition (2) therapy (3)
cancer (3) disease severity morbid
index (1) obesity (1)
diabetes mellitus (3) old age (1)
ASA score (1)
prolonged
infections preoperative stay (1)
at other sites (1)
SSI
* Consensus paper on the surveillance of surgical wound infections. Sheretz et al. AJIC 1992;20: 263-270
Operation-related Risk Factors*
glove
inexperience (3) low procedure drains (3) punctures (3)
volume (1)
unskilled
surgeon (3) low abdominal
prolonged site (1)
failure to obliterate foreign
hosp. admission (2)
dead space (3) material (3)
no prophylactic
tissue razor no preoperative antibiotics (1)
trauma (2) shaves (1) shower or scrub (3)
poor
hemostasis (3)
specific type
number of of surgical procedure (1)
people in OR (3)
multiple emergency (3)
procedures (2)
prolonged duration intraoperative surgical wound
of surgery (1) contamination (1) class (1)
SSI
* Consensus paper on the surveillance of surgical wound infections. Sheretz et al. AJIC 1992;20: 263-270
6
Quality of care
• Variability
– Resources
• Equipment / human resources
– Organization of practices
• Protocols
– Staff qualification and performance
• Training
• Experience / skill
Factors of variability of infection rates
Variations Variations Quality Case-mix
due to hazard due to seasons of data collection and type of care
Staff Ratio
Variations of indicators Performance Staff
« nosocomial infections » Quality of care qualification
System
for production & use Quality of
of indicators organization
Modalities of the feed-back,
others evaluative processes
Efficiency of
the information system
risk reduction
7
A ratio, how to use it ?
• To compare itself to its own previous results:
– Increase ?
– Decrease ?
– From where to where ?
• To look at:
– Potential problems with data collection and methodology
– Specificity of the case-mix
– Change in practices
Methodology and data collection
• Numerator
– Misinterpretation of definitions
– Under-declaration
– Lack in diagnosis methods, case validation
• Denominator
– Small population
– Lack of exhaustiveness
– Errors in inclusion criteria
8
Patient case-mix
• Patients with a specific severity or risk factor
– Age, impaired immunity, underlying diseases
– Surgery for cancer
• New type of cares
– Transfer of high risk patients (transplants …)
Practices and performance
• Changes in organization
– Units, operating theater, consultations
• Modification of procedures associated with
SSI prevention
– Preoperative cares (shower, hair removal, antisepsis …)
– Antibioprophylaxis
– Environment
9
A ratio, how to use it ?
• Compare its results with others
– Literature
– Other units in the network
• Be very cautious !
– Difficulties in statistical comparisons
– Methodological differences
Statistical difficulties
• SSI = statistically rare event within
certain specialties (< 1 %)
• "Time and place"-dependant variations
• Comparison ►confidence interval
T1 T2
20 / 200 = 10 % 30 / 200 = 15 %
CI 95 % = [5,8 % - 14,2 %] CI 95 % = [10,1 % - 19,9 %]
10
Differences in methodology
• Follow-up of 4 773 patients (5 804 procedures)*
• Use of 4 different definitions for SSI
• Important variation of the SSI rates due to
– Minor changes in definitions or their interpretation
– 19,2% / 14,6% / 12,3% / 6,8%
• Decrease of the cases after feed-back to the
surgeons
• Absence of concordance discourage
comparisons
* SWI as a performance indicator: agreement of common definitions of wound infection in 4773
patients. Wilson et al. B M J, doi:10.1136/bmj.38232.646227.DE (14 septembre 2004)
Crude rate > 7% OR
"NNIS 0" rate > 4%*
Part. Type Type chir All NNIS 0 Previous Comments
period n n results
Tx Tx NNIS 0
2e Priv. DIG 107 71 50 Small pop.
1 7.5 % * 8.5 % 4% Great variability
1 Priv. UROL 266 177 - Difficulty SSI
2 7.1 % * 6.8 % or UTI ?
2e Hosp. ORTH 151 101 86 Problem / surv.
3 4.6 % 4.9 % 1.2 % Exhaustiveness = 50 %
1 Priv. ORL 88 62 - Small population
4 4.5 % 4.8 % cancer surgery ++
1 Hosp. POLYV 301 182 - Half SSI < UROL
5 4.6 % 4.4 % (UTI ?)
* ISO SUD-EST / CCLIN Sud-Est
11
Conclusion (temporary)
• Importance of implementation of corrective measures
• Rates = Real interest for a given unit
• Premature utilization of SSI ratio as a direct
indicator of performance or quality of care
• Necessity of complementary studies
– Strict respect of the methodology
– Agreement on solid definitions
– Internal and external quality control of data
– Improvement of the risk adjustment
12
