Methodological concerns cast doubt on link between SIRI and diabetic foot infection severity

A detailed critique of a recent study linking systemic inflammatory response index (SIRI) to severe diabetic foot infection highlights potential overlaps, statistical issues, and the need for more robust, prospective research to validate biomarker utility.

A commentary in Diabetes, Metabolic Syndrome and Obesity has raised methodological doubts over a recent study that linked the systemic inflammatory response index, or SIRI, with severe diabetic foot infection. Qingqing Shan of Chengdu Integrated TCM & Western Medicine Hospital said the paper by Wu and colleagues was an interesting attempt to use a routine blood marker for risk stratification, but argued that its design and statistical handling may have overstated the strength of the association. The original study, from The First People’s Hospital of Changzhou, reported a strong relationship between higher SIRI values and severe infection, with an area under the receiver operating characteristic curve of 0.959. (pubmed.ncbi.nlm.nih.gov)

Shan’s central concern is that SIRI and the definition of severe diabetic foot infection may partly draw on the same laboratory signals. International guidance for diabetic foot infection classifies severe cases as those with systemic inflammatory response syndrome features, including an elevated white blood cell count and immature neutrophils, while SIRI itself is calculated from neutrophils, monocytes and lymphocytes. In Shan’s view, that overlap could make SIRI look more accurate than it really is, because the biomarker and the outcome are not fully independent. (aafp.org)

The letter also argues that the study should not have used the language of prediction so freely. Because the exposure and outcome were assessed during the same admission window in a retrospective design, Shan said the analysis can support association but not genuine prognosis. The authors studied 242 patients and divided them into severe and moderate infection groups, but the commentary says a real predictive test would need follow-up to see whether baseline SIRI forecasts later outcomes such as amputation or intensive care admission. (pubmed.ncbi.nlm.nih.gov)

A further objection is statistical. Shan said the final regression model included 17 covariates for 110 severe cases, leaving an events-per-variable ratio of about 6.5. That is below the traditional rule of thumb of 10, and while newer work suggests the threshold can sometimes be lower, the letter argues that the model would have been more convincing if the authors had used penalised methods or validation procedures to test for overfitting. (pubmed.ncbi.nlm.nih.gov)

Even so, the commentary did not dismiss the study outright. Shan said it offers a useful starting point for further research into inflammatory markers in diabetic foot care, a field in which related indices such as the systemic immune-inflammation index have already been examined for links with diabetic foot disease. The call now, the letter suggests, is for prospective, multicentre studies that can separate genuine clinical usefulness from statistical overlap. (pubmed.ncbi.nlm.nih.gov)