Acute severe colitis (ASC) from Inflammatory Bowel Disease (IBD) is a medical emergency that is difficult to predict outcome. Hospitalisation rates for IBD are high in OECD countries and in New Zealand, this contributed to an estimated cost of $17 million NZD from 2001 to 2013.[[1]] Intravenous corticosteroid therapy remains the first-line treatment for ASC. However, 26% of cases fail steroid treatment and require rescue therapy according to a recent Auckland study.[[2]] The Day 3 Oxford Criteria is an older composite prediction tool that defines steroid failure to direct rescue therapy but its predictive value is controversial in the biologics era.[[3,4]] Therefore, recent studies have been focused on biomarkers such as C-reactive protein/albumin ratio (CAR) or faecal calprotectin (fCal) to provide earlier rescue therapy prediction.[[5–7]]

The multi-centre cohort study on ASC from 2016 to 2019 across all three district health boards (DHBs) in Auckland showed significantly less need for rescue therapy in first-line intravenous hydrocortisone treatment compared to intravenous methylprednisolone.[[2]] We present a subsequent retrospective study based on the data from one of the DHBs to examine the predictive value of CAR and fCal for rescue therapy in ASC.

Methods

Cases of ASC admissions to North Shore hospital as defined by Truelove and Witts criteria in the Waitematā DHB protocol were identified in the study by Schauer et al. between June 2017 and June 2018.[[2,8]] This cohort was retrospectively examined for their age, gender, ethnicity, smoking status, type of IBD, admission C-reactive protein (CRP), admission albumin, admission fCal or recent fCal within 4 weeks prior to admission. CAR was calculated from admission CRP and albumin. The primary outcome was the need for rescue therapy defined as cases that required rescue biologic agents, cyclosporin or colectomy. The Day 3 Oxford Criteria is part of the protocol to guide rescue therapy, but recordings of its utility were not consistently found in the retrospective data.[[3]] The length of stay (LOS) was recorded as a secondary outcome.

Continuous data are presented as means with standard deviations if variables have normal distribution, otherwise presented as medians and interquartile ranges. Bivariate analyses were carried out for all variables in relation to the need for rescue therapy. This is done individually using the appropriate statistical tests (Table 1) via the Statistical Package for the Social Sciences. If a biomarker was found to be significantly different between rescue and non-rescue groups, Area under the Receiver operating characteristic analysis was used to evaluate the optimal cut-off and multivariate analysis through multiple logistic regression was performed to verify statistical significance when potential confounding variables are included. Linear regression was performed for each biomarker in relation to LOS.

Results

There were 108 cases (63% men; mean age 39.9 years) included in the analysis (Table 1). All cases had admission CRP and albumin for CAR calculation but only 35 cases (32.4%) had an admission or recent fCal. Demographic variables are comparable between the fCal group and the overall cohort. There are 22 cases (20.4%) that needed rescue therapy and out of these, five cases (4.6%) had colectomy. The demographic variables were similar between the rescue and non-rescue groups while the median LOS is understandably longer in the rescue group (p<0.01).

View Table 1 and Figure 1.

The median CAR is 1.12 (IQR 0.50–3.92) in the rescue group, which is significantly higher than the non-rescue group median CAR of 0.51 (IQR 0.16–1.87), p=0.03 (Figure 1). This finding is confirmed in the multivariate analysis involving age, gender, ethnicity, smoking status and type of IBD, p=0.03. CAR has a significant positive linear relationship with LOS (r=0.45, p<0.01). The optimal CAR cut-off for our cohort is 0.62 with a positive likelihood ratio of 1.56, p<0.05.

The median fCal between the rescue group (1,180mcg/g, IQR 234–3,860) and the non-rescue group (1,090mcg/g, IQR 244–3658) were not significantly different in our cohort, p=0.96. This was the case even if admission fCal (p=0.88) and recent fCal (p=0.68) were analysed separately. The fCal also did not have any significant correlation with LOS.

Discussion

In this study, we confirm that admission CAR is a useful early prediction tool for rescue therapy in ASC. This is supported by previous studies and the optimal cut-off for CAR is likely between 0.5 to 1.5.[[5,6]] CAR is also validated as a marker of disease severity in chronic IBD as well as a prognostic marker in a wide range of critical illnesses such as malignancy and sepsis.[[9,10]] Therefore, its use needs to be interpreted appropriately in distinct clinical contexts.

In comparison, fCal in this study did not demonstrate significant correlation with the need for rescue therapy. This is despite its established role as a disease activity marker in IBD and other studies demonstrating its predictive value in ASC.[[5,7,11]] The lack of significant correlation may be attributable to our relatively small group of cases that had admission or recent fCal. The collection of fCal on admission for ASC is part of the protocol and the paucity of data within this cohort highlights the need for clinicians to emphasise its importance with patients and nursing staff.

Limitations of the study include the small group of cases that had fCal and its retrospective design. Strengths include a pragmatic study design with well-defined variables as well as the consistency of a protocol on ASC in this single-centre study. The cohort is reflective of the real-world IBD population and gives the study generalisability. Further studies with larger prospective fCal data or longitudinal follow-up data after admission for ASC will be valuable in this area.