Aim: To investigate whether national and multi-country SSI incidence can be estimated from ECDC PPS data.
Methods: In all, 159 hospitals were included from 15 countries that participated in both ECDC surveillance modules, aligning surgical procedures in the incidence surveillance to corresponding specialties from the PPS. National daily prevalence of SSIs was simulated from the incidence surveillance data, the Rhame and Sudderth (R&S) formula was used to estimate national and multi-country SSI incidence from the PPS data, and national incidence per specialty was predicted using a linear model including data from the PPS.
Findings: The simulation of daily SSI prevalence from incidence surveillance of SSIs showed that prevalence fluctuated randomly depending on the day of measurement. The correlation between the national aggregated incidence estimated with R&S formula and observed SSI incidence was low (correlation coefficient = 0.24), but specialty-specific incidence results were more reliable, especially when the number of included patients was large (correlation coefficients ranging from 0.40 to 1.00). The linear prediction model including PPS data had low proportion of explained variance (0.40).
Conclusion: Due to a lack of accuracy, use of PPS data to estimate SSI incidence is recommended only in situations where incidence surveillance of SSIs is not performed, and where sufficiently large samples of PPS data are available.
- surgical site infection