Literature Update - February 2008
Compiled by Dr Leon Worth, ID Physician
Differences in Implementation of Surveillance Strategies Limit International Comparison of SSI Surveillance Data
Mannien J et al (J Hosp Infect 2007; 66:224-231) compared surgical site infection (SSI) data for the Netherlands and Germany. In the Netherlands (‘PREZIES’),
data were captured from 68% of Dutch hospitals, and in Germany (‘KISS’), 14% of all hospitals participated in surveillance activities.
Surveillance definitions in both countries were based upon NNIS protocol, using definitions developed by the CDC. For both, postdischarge
surveillance was voluntary. PREZIES data for 1996-2004 and KISS data for 2004-2005 were analysed in the current study, and 9 comparable
surgical procedure categories were included from several specialties.
KISS data showed a significantly higher ASA classification than PREZIES for all studied surgical procedures. Differences for SSI data were
found for duration of surgery, wound contamination class, and postoperative duration of hospitalisation. For PREZIES, a significantly higher
superficial SSI rate was evident for 7 surgical procedures, and a higher deep SSI rate for 5 procedures. If only deep SSI during hospitalisation
were analysed, differences in SSI rates were much smaller. In KISS, 21% of SSI were detected postdischarge, whereas 34% of SSI were detected after
discharge for PREZIES, likely due to differences in intensity of postdischarge surveillance.
Caution is required when comparing SSI date between countries, even if similar infection surveillance protocols are adopted. Findings of this
study suggest that the reliability of comparisons may be influenced by a number of factors: duration of hospitalisation, intensity of
postdischarge surveillance, proportion of participating hospitals, and intensity of validation studies performed in individual countries.
Previous studies have suggested many of these factors to be relevant. If SSI surveillance data are to be reliably used for international
benchmarking, surveillance protocols in individual countries and other significant factors (e.g. external validation of data) must be comparable.
Rates of deep SSIs during hospitalisation may provide a reliable means of comparison, unaffected by intensity of postdischarge surveillance.
Use of Administrative Data for Surveillance of Invasive Aspergillosis
The clinical definition of invasive aspergillosis requires diagnostic evaluation by imaging, biopsy, or non-culture based tests
(Aspergillus galactomannan, Aspergillus PCR assays). It is not known if simplified or
less-resource intensive definitions can be reliably applied. To estimate the accuracy of ICD-9 codes for detecting invasive aspergillosis, Chang
DC et al (Infect Control Hosp Epidemiol 2008; 29:25-30) compared prospective active surveillance (TRANSNET) at a single centre with ICD-9
coded discharge codes consistent with invasive aspergillosis (117.3, 117.9, 348.8, 484.6, 484.7, 495.4) between 2001 and 2006. Medical
record review was used to classify suspected cases as proven, probable or ‘other’, using EORTC/MSG definitions.
Of 1736 patients who received haematopoietic stem cell or solid organ transplantation at the study centre, 64 were identified as having been assigned
the relevant ICD-9 codes. An additional 3 patients were identified by TRANSNET surveillance measures during the study period. Of the 67 patients with
suspected infection, medical record review classified 19 (28%) as proven or probable invasive aspergillosis, and 48 (72%) as other infections or no
infection or insufficient evidence to be classified as proven or probable invasive aspergillosis. Of the 19 with proven/probable infection, 3 were
identified by TRANSNET but not ICD-9 coding, 2 were identified by ICD-9 coding but not TRANSNET, and 14 were identified by both strategies. The
sensitivity and positive predictive value (PPV) of the ICD-9 code ‘117.3’ for detecting proven/probable infection were 63% and 71%, respectively.
Combinations of ICD-9 codes (2, 3 or 4 code combinations) resulted in improved sensitivity (84%), but reduced PPV (28-30%). Overall, ICD-9 codes
triggered review of medical records for 64 patients, only 16 (25%) of whom had proven/probable invasive aspergillosis.
The poor PPV of ICD-9 codes means that this cannot be used as the sole method for case detection. ICD-9 codes may be used for initial screening of
at-risk populations, allowing more detailed medical record review of identified cases. However, multi-centre evaluation and adoption of ICD-10 codes is
required prior to use as part of a formal surveillance strategy or inter-institutional comparison.