Is it possible to predict the severity of acute appendicitis? Reliability of predictive models based on easily available blood variables

Abstract

Abstract Introduction Recent evidence confirms that the treatment of acute appendicitis is not necessarily surgical, and selected patients with uncomplicated appendicitis can benefit from a non-operative management. Unfortunately, no cost-effective test has been proven to be able to effectively predict the degree of appendicular inflammation as yet, therefore, patient selection is too often left to the personal choice of the emergency surgeon. Our paper aims to clarify if basic and readily available blood tests can give reliable prognostic information to build up predictive models to help the decision-making process. Methods Clinical notes of 2275 patients who underwent an appendicectomy with a presumptive diagnosis of acute appendicitis were reviewed, taking into consideration basic preoperative blood tests and histology reports on the surgical specimens. Variables were compared with univariate and multivariate analysis, and predictive models were created. Results 18.2% of patients had a negative appendicectomy, 9.6% had mucosal only inflammation, 53% had transmural inflammation and 19.2% had gangrenous appendicitis. A strong correlation was found between degree of inflammation and lymphocytes count and CRP/Albumin ratio, both at univariate and multivariate analysis. A predictive model to identify cases of gangrenous appendicitis was developed. Conclusion Low lymphocyte count and high CRP/Albumin ratio combined into a predictive model may have a role in the selection of patients who deserve appendicectomy instead of non-operative management of acute appendicitis.

Keywords

• Acute appendicitis
• Appendicectomy
• Predictive models
• Lymphocyte count
• c-reactive protein
• Albumin