Reducing the incidence of surgical site infection:
A personal story in three acts
Robert G. Sawyer, M.D.Professor of Surgery
and Public Health Sciences
At the Midpoint of the Odyssey
• Defining the real rate of surgical site infection
• Getting better using current recommendations
• Can we get even better?
Defining the Real Rate of Surgical Site Infection- Act I
Hospitalized Patients
• Crabtree TD et al. Shock 2007;17:258-262
• January to December 1997
• Compared infection rates determined by infection control practitioners (ICP) and 3x weekly chart survey by RGS
• Identical CDC definitions used
• Hospitalized patients only
Who Were These People?
Crabtree et al, Shock 2002
Differences in Rates
Crabtree et al, Shock 2002
Why?
Crabtree et al, Shock 2002
We are Not Geniuses (at all)
Crabtree et al, Shock 2002
Defining the Real Rate of Surgical Site Infection- Part II
Risk of Surgical Site Infection After Colorectal Surgery
13.9
10.5
6.5
4.3
NNIS rates, %
0.04
0.88
0.77
0.19
P
8
12
14
9
No. of infections
30.8
9.8
6.9
6.7
Infection rates, %
0
3
2
1
Risk Index
Weiss CA et al. Arch Surg. 1999;134:1041–1048
Curiosity Kills the Dog(ma)?
• Chip Foley is a very honest colorectal surgeon who was convinced his SSI rate was significantly higher than the 5-8% predicted by the CDC
• He reviewed his results based on the premise that his outcomes were bad
• He consider a failure any wound that was closed primarily but did not heal without incident
Chip’s story
• Smith RL et al, Ann Surg 2004;239:599-607
• 2 years
• Retrospective experience of a single colorectal surgeon
• 176 colorectal resections
• No simple creation or reversal of stomata
Major Outcomes
• 26% SSI rate (45 SSI)
• 49% of SSI (22 SSI) diagnosed after discharge
• Median time to diagnosis 9 days, interquartile range = 5-19 days
• 5 SSI diagnosed more than 30 days post-operatively (37, 43, 48, 65, and 73 days)
Univariate Risk Factors
Smith RL et al, Ann Surg 2004
Univariate Risk Factors
Smith RL et al, Ann Surg 2004
Predictors of SSI
Smith RL et al, Ann Surg 2004
Timing of Diagnosis of SSI
0
1
2
3
4
5
6
0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 68 72Days from Operation to Infection
Smith RL et al, Ann Surg 2004
Summary 1
• Current SSI rates are probably a lot higher than those documented either from drug studies or standard surveillance
• Colorectal surgery SSI rates may be remarkably high
• The harder you look, the more you find
Getting Better Using Current Recommendations-Act II
SCIP and General Surgery
• Hedrick TL et al, Surg Infect 2007;8:425-435.
• 4 months of baseline data, four months after change in protocol
• 379 patients before, 390 after
• General surgery patients followed as part of the National Surgical Quality Improvement Project (NSQIP)
SCIP Inspired Changes
Hedrick TL et al, Surg Infect 2007
SCIP Inspired Changes
Hedrick TL et al, Surg Infect 2007
Timing of Changes
Hedrick TL et al, Surg Infect 2007
Process Changes
Hedrick TL et al, Surg Infect 2007
SSI Rates, All Comers
Hedrick TL et al, Surg Infect 2007
SSI by Type of Procedure
Hedrick TL et al, Surg Infect 2007
SSI by Type of Procedure
Hedrick TL et al, Surg Infect 2007
Predictors of SSI
Hedrick TL et al, Surg Infect 2007
SCIP and Colorectal Surgery
• Hedrick TL et al. J Am Coll Surg2007;205:432-438
• Remember Chip?
• February 2000 to January 2002 versus January 2005 to August 2005
• SCIP measures, plus penrose drain in subcutaneous tissue if BMI ≥ 25
Patient Characteristics
Hedrick TL et al, J Am Coll Surg 2007
Patient Characteristics
Hedrick TL et al, J Am Coll Surg 2007
Operative Characteristics
Hedrick TL et al, J Am Coll Surg 2007
Operative Characteristics
Hedrick TL et al, J Am Coll Surg 2007
Process Measures
Hedrick TL et al, J Am Coll Surg 2007
Surgical Site Infection
25.6%
15.9%
0.0%
5.0%
10.0%
15.0%
20.0%
25.0%
30.0%
BaselineStudy
* (p ≤ 0.05)*In
cide
nce
↓39%
Hedrick TL et al, J Am Coll Surg 2007
Predictors of SSI
Hedrick TL et al, J Am Coll Surg 2007
Summary 2
• Application of evidence-based interventions, as exemplified by the SCIP process, can lead to a significant improvement in SSI rates
• Some standards are easier to achieve than others
• It takes a lot of hard work
Can We Get Even Better?-Act III
The effects of preoperative skin preparation on postoperative
wound infection:
a prospective study of three skin preparation protocols
Published in Infection Control and Hospital Epidemiology October 2009
Brian R. Swenson, M.D., M.S., Traci Hedrick, M.D. , Hugo Bonatti, M.D. , Rosemarie Metzger, M.D. , Timothy L. Pruett, M.D. , and
Robert G. Sawyer, M.D.
Background-Joseph Lister
• Inspired by Louis Pasteur
• Proved that gangrene was caused by microorganisms
• Used a sewage deodorizer (5% carbolic acid) to sterilize instruments/clean wounds
• Lowered surgical site gangrene incidence (Lancet (90) 2299, 1867)
Joseph Lister 1827-1912
Background
• Many compounds and formulations are currently approved for preoperative surgical site skin preparation
• Most commonly used include:– Chlorhexidine (most commonly used biocide today)
• Broad spectrum, low irritation
• Thought to collapse bacterial cell membrane potential
– Iodine based compounds (iodophors)
• Broad spectrum, rapid killing
• mechanism of action unknown
– Isopropyl Alcohol
• Broad spectrum
• Limited residual activity (usually used as an adjunct)
Background
• Skin prep research over the past several decades has focused on surrogate endpoints
– In vitro studies
– Post-prep skin cultures– Post procedure skin/wound cultures
– Serial skin swab cultures after prep treatment
• Very few studies examine SSI as the endpoint
– Central venous catheter studies
Study Aim
• As an institution, can we affect the SSI rates in general surgery patients by changing the standard skin preparation modality?
– Given reports of lower catheter-associated bloodstream infections with ChloraPrep® we hypothesized that we would see similar results in general surgery
Methods
• ‘Official’ preferred skin prep in general surgery patients at UVA was standardized for six month blocks
– 6 months UVA traditional prep
• betadine soap/EtOH/Betadine paint
– 6 months ChloraPrep®
– 6 months DuraPrep™
• Local ACS-NSQIP database used to track comorbidies, perioperative data, and 30 day outcomes, specifically SSI as defined by the CDC
• Analysis by intent-to-treat
Results
• January ‘06 – June ‘07, 3,209 patients were followed• 182 SSI (5.7%) identified
• The three periods were well matched in terms of age, race, ASA class, weight classification, wound classification, and most major comorbidities
• Minor variations were seen in gender, smoking status, preoperative sepsis, and OR time
Results-outcomes
>2.5%p<0.05
>1.8%p<0.05
Results-Univariate analysis
Variables associated with higher SSI rate
• Female gender
• Diabetes
• Cancer
• Preoperative sepsis
• Recent weight loss
• Wound classification
• OR time
• Prep solution actually used
Results-outcomes by prep received
3.4%p<0.05
~2.2%p<0.05
Results-subgroup analysis
• Where are we seeing a difference?
• Analysis of SSI outcomes comparing iodophor based preps to ChloraPrep®stratified by wound classification
– Reduction in superficial SSI in clean cases
– Reduction in all SSI in dirty cases
Logistic regressionVariable OR P Value
Female gender 1.54 0.0057
Medical history
Diabetes 1.46 0.051
Cancer 1.39 0.32
Sepsis 1.35 0.28
Weight loss 1.28 0.44
OR time (per min) 1.003 <0.001
Variable OR P Value
Wound class
Clean 1.00 -
Clean-contam. 5.35 <0.001
Contaminated 6.84 <0.001
Dirty 6.59 <0.001
Prep used
Iodophor 1.00 -
ChloraPrep® 1.35 0.073
Conclusions
• We report a large, single center, prospective, unblinded, phase IV comparison of three skin preparation modalities in an attempt to lower SSI rates
• Primary objective was successful– Significantly lower SSI rates (≥1.8%) seen in period 3 when
DuraPrep™ was utilized as the official prep solution
– Contrary to our hypothesis!
Conclusions
• In subgroup analysis no difference was identified between the traditional betadine soap/EtOH/Betadine paint and DuraPrep®
• Significant differences, however, were seen between iodophor based preps and ChloraPrep®
– 2-3% lower rates with iodophor based preps
• Differences were seen in superficial SSI in clean cases
Conjecture
• Why are our results different from what has been published about central venous catheter infection rates?
– Fundamentally different environment
• Why better results with Iodophor preps?– Does betadine kill more (no), better (not really), or longer?
– Mechanical component to the prep?– Study not designed to answer this question
Moving forward
• DuraPrep™ has been adopted at our institution as the preferred skin preparation modality for general surgery cases
• Additional multicenter studies are needed to confirm theses results
Summary 3
• Perhaps iodine-based skin preparations perform better than other skin preparations
• Confirmatory data are required
• Other interventions may be available to further reduce the risk of SSI
Overall Conclusions
• SSI rates are still too high
• 0% is probably impossible, but let’s pretend it isn’t
• There are dozens of areas for intervention, many of which have not been tested
• Improvement will depend on research AND changes in practice
Thank you
Questions?