Does C-reactive protein day 1 post-surgery have a predictive role for post-operative complications: a single-centre perspective following published meta-analysis

Abstract

Bona et al. (2019) have published a review and Bayesian meta-analysis to investigate the predictive utility of Creactive protein (CRP) for for dectecting complications post operative for bariatric surgey. post-bariatric surgical complications [1]. Identification of patients who are at risk of developing complications is of great importance to improve clinical outcomes, improve patient experience, guide resource allocation, and provide safe enhanced recovery protocols. There are no national guidelines or protocols to use CRP for this purpose but it is an attractive possibility as it is cheap, easily collected and some authors have found predictive utility. Our group in York Teaching Hospital published work used in this metaanalysis where we concluded that post-operative day (POD) 1 CRP had no diagnostic utility; however, POD 2 did [2]. Our work had a relatively small sample size and was retrospective in nature. Therefore, we commend Bona et al. on their work to combine the results of many smaller studies using metaanalysis. Some authors suggest the use of a negative result along with normal physiology to expedite discharge [3]. Bona et al. have suggested that a result of less than 61 mg/l, with a normal examination, seems associated with minimal risk of anastomotic leak or intra-abdominal infectious complications. They quote a figure of less than 2% risk [1]. We have re-analysed our data set at the York Teaching Hospital with this cut off (61 mg/l) to give a single-centre perspective on this result. One hundred patients had a POD 1, POD 1 CRP. Of these 67 had a positive result and 33 had a result below the cut off. Of the 67 positives, 57 were false positives giving a disappointing positive predictive value (PPV) of 0.14. Of the 33 negative results, 29 were true negatives giving a negative predictive value (NPV) of 0.87. In our cohort of patients, CRP on POD 1 does not appear clinically useful. If used as a screening tool to predict for complications, it was positive in 67% of the patients and was a false positive in 82.5% of these cases. If the use was to improve enhanced recovery and early discharge, it does have a good NPV; however, with only a small number of patients had a negative result. Furthermore, 14% of the negative results (n = 4) had serious complications. One of which had bowel obstruction and required ITU admission and one died of complications of bowel ischemia. We suggested that POD 1 was too early to distinguish between the CRP rise seen with surgery and the early development of complications. Given the findings that POD 2 CRP was predictive in our study population, we would be very interested to see meta-analysis of the predictive value of CRP on POD 2 and 3.