Literature Update - April 2008
Compiled by Dr Leon Worth, ID Physician
Postoperative Hyperglycaemia and Risk for Surgical Site Infections Following Surgery for Hepato-biliary-pancreatic Cancer
A number of studies have identified an association between hyperglycaemia and sternal wound infection following cardiac surgery, prompting guidelines
and recommendations for maintenance of normoglycaemia as an intervention for SSI risk reduction. Ambiru S et al (J Hosp Infect 2008; 68:230-233)
prospectively evaluated patients undergoing biliary surgery at a single Japanese centre, to determine risk factors for the development of surgical
site infection (SSI), including postoperative glycaemic control.
Over a 2 year period, 265 consecutive patients undergoing elective surgery for hepato-biliary-pancreatic cancer were studied. A single surgeon
performed all procedures. Patients were observed preoperatively and followed for 30 days postoperatively. Blood glucose was monitored every 6
hours until postoperative day 2, regardless of whether a patient was diabetic. Short-acting insulin was administered as a subcutaneous dose for
blood glucose levels >200mg/dL, or as an infusion in cases of diabetes and liver cirrhosis.
Rates of SSI were 38/147 (26%), 32/53 (60%) and 18/65 (28%) in cases of liver, biliary and pancreatic cancer, respectively. 38% of all infections were
identified during post-discharge surveillance. Univariate analysis identified the following factors to be significantly associated with development
of SSI: obstructive jaundice, risk index score, number of enteric anastomoses during surgery, postoperative blood glucose control, type of cancer,
and requirement for blood transfusion. Multivariate analysis identified number of enteric anastomoses, postoperative blood glucose control and type of
cancer to be independently associated with SSI occurrence. When blood glucose levels were stratified (<200, 200-249, 250-299, ≥300 mg/dL), SSI
rates were directly correlated with degree of postoperative hyperglycaemia.
These findings suggest that maintenance of postoperative blood glucose at <200 mg/dL (11 mmol/l) is independently associated with reduced SSI rates
in this population. Remaining questions include the impact on SSI risk of improved postoperative blood glucose control beyond day 2 and the
optimisation of diabetic control in the pre-operative period.
ICD-9 Administrative Data for Surveillance of Healthcare-associated Infections
As a readily available means of data collection with less intensive resource requirements than traditional surveillance methods,
administrative datasets have been proposed for surveillance of healthcare-associated infections (HAIs). A number of previous studies, however,
have questioned the accuracy of determining infection rates from administrative databases containing International Classification of Diseases,
9th Revision, Clinical Modification (ICD-9-CM) codes. At a single centre, Stevenson KB et al (Am J Infect Control 2008; 36:155-64) retrospectively
compared the accuracy of HAI case detection by ICD-9-CM with traditional methods outlined by the CDC/HNSN.
During the study period (2005), 5 infection control practitioners (ICPs) conducted infection surveillance at the study centre.
At-risk patient populations were defined according to ICD-9-CM procedure codes for the following surgical procedures:
CABG, peripheral vascular, colorectal, head and neck, hysterectomy, spinal surgeries, craniotomy, ventricular shunt, knee and hip surgery.
Patients with central venous catheter placement and mechanical ventilation were also identified as at risk. ICD-9-CM discharge codes
for surgical site infections, bloodstream infections and pneumonia were used to screen for the presence of infection. Patients identified
by traditional surveillance methodology were used as a reference, to determine positive predictive vale (PPV) and negative predictive value
(NPV) of coding data. Discordant cases were reviewed independently by ICPs, and cases re-classified according to CDC/NHSN definitions and methodology.
For surgical procedures (n = 3882), 457 (12%) were identified as having SSI by coding data, and 144 (4%) by traditional methods. For patients
with centrally-placed central venous catheters and positive blood culture (n = 1599), 569 (36%) were identified as having a catheter-associated
BSI by coding data, and 150 (9%) by traditional methods. For patients with mechanical ventilation and intubation (n = 193), 41 (21%) were
identified as having ventilator-associated pneumonia by coding data, and 24 (12%) by traditional methods. Of discordant cases, 15% of patients
undergoing a surgical procedure were found to meet CDC/NHSN criteria. Among a random sample of discordant potential catheter-associated BSI,
none were re-classified as meeting CDC/NHSN criteria, while 31 discordant cases of ventilator-associated pneumonia, with 4 (13%) re-classified
as having ventilator-associated pneumonia according to CDC/NHSN criteria. Overall, the PPV of coding data was low (0.14 to 0.51), with an
aggregate of 0.23, and no significant difference between surgical sub-groups. The NPV of coding data was high (0.91 to 1.00), with an aggregate of 0.96.
Findings are consistent with a number of other comparative studies which have used CDC methods as a gold standard. Although ICD-10 has not
been extensively evaluated, it is likely that similar findings would be observed. The low PPV for all studied infections (SSI, catheter-associated
BSI, ventilator-associated pneumonia) would indicate that administrative data alone is a poor tool for widespread use as a surveillance tool.