Single center assessment of the role of Oakland score among patients admitted for acute lower gastrointestinal bleeding – BMC Gastroenterology

The Oakland score has been validated in the UK and US to predict safe discharge in patients who present with LGIB. However, this study demonstrated findings that question the generalizability of the Oakland score for certain patient populations. Oakland score should not be used as a sole criterion to determine discharge for patients with LGIB. In this retrospective study, at the time of discharge, almost all (96.2%) patients had a higher risk Oakland score of greater than 10, despite an overall low rate of readmission at 1-month of only 3.7% (significantly lower than national average rate of 15%). This is in large part due to baseline anemia that has the highest contribution for the Oakland score. In fact, the median Oakland score before hospitalization due to GI bleeding was at least 13 given the median hemoglobin of 10.8% at baseline. Since only 2/108 (1.8%) of patients demonstrated improved Oakland score (≤ 10) at discharge, the use of this score does not represent a helpful tool in our patient population.

The Oakland score is a clinical predictor model which assess the safe outpatient management for patients with LGIB, it was developed and validated in United Kingdom in 2017 and externally validated in the US in 2020. At the time of validation, this score was compared with other risk scores (Blatchford, AIMS65, BLEED, Strate, NOBLADS, and Rockall) and showed a better prediction in safe discharge [2, 5]. This score model contained 7 (age, sex, previous hospitalization with LGIB, digital rectal exam, SBP, pulse and Hgb) and from those, Hgb and SBP contribute with the highest point representation, 0–22 points and 0–5 points, respectively [2, 5].

In this study of 108 patients admitted for LGIB, where the majority were African Americans, we retrospectively evaluated their admission and discharge criteria and analyzed the hypothetic use of the Oakland score to predict safe discharge. We considered the threshold point for safe discharge of less than or equal to 10, as patients with a threshold of more than 10 are believed to carry an increased adverse events risk and should be hospitalized [5]. Lower risk patients, whose bleeding had stopped can have a close follow up with a gastroenterologist in an outpatient setting [9, 10]. Spontaneous resolution of the majority of LGIB has been well documented, supporting non-hospital management of these low-risk patients [11, 12]. Assessing the main two variables of this score, the mean SBP at admission and discharge did not variate significantly, 129.0 and 130.7 respectively but represented 1-point drop in the score assessment. While the mean Hgb was 8.8 g/dL on arrival and 9.4 g/dL at discharge, also a small improvement but represented 4 points drop in the score assessment. During the original validation of Oakland score, a mean of 9.7 g/dL was found in the not safely discharge group compared with 12.9 g/dL in the safely discharged group, this represented a moderate improvement in Hgb, reflecting 5 points drop in the score assessment [2]. The external validation of this score in the US, a mean of 8.5 g/dL was found in the not safely discharge group compared with 12.2 g/dL in the safely discharged group, this represented a considerable improvement in Hgb, reflecting 9 points drop in the score assessment [5]. With this important impact of Hgb in the Oakland score, we evaluated baseline Hgb in our population and nearly 60% of the patients had history of anemia which we believe might have contribute to the differences seen in our study.

Considering our findings, the use of the Oakland score was not a helpful tool for safe discharge in African American patients who present with LGIB. Comparing our discharged group with Oakland score greater than 10 with the same group in the original development cohort [2], similar mortality and readmission rate was found, 1.9% and 3.7% vs. 2% and 5%, respectively. Nevertheless, if we take in consideration patient baseline anemia, a criteria not contemplated in the original study, more than 30% of the patients that were admitted with Oakland score > 10 and did not require blood transfusion could have been followed in the outpatient setting, avoiding an unnecessary hospital admission. Assessing readmission, we found that history of admission for previous LGIB was associated with readmission (p = 0.046). Similar findings have been found in 30-day readmission rate assessment in a prospective observational study of a single center and retrospective evaluation of a national sample, showing a high readmission rate due to recurrent LGIB [13, 14]. Another study, using the United States National Readmission Database, showed that LOS, associated comorbidities, and GI diseases were the major predictors of readmission in 30 days, with higher rate (14.6%) of another episode of GI hemorrhage among all GI diseases [15]. In the secondary analysis of our study, important findings as PPI use in LGIB was assessed. The inappropriate use of PPI in hospitalized patients has been of great research interest due increase risk of infections and costs [16, 17]. Data from PPI use in the setting of LGIB is scarce, a retrospective study showed that more than 30% of patients admitted for LGIB were on PPI, around of 46% of those patients received this medication without indication [18]. In our study, 74% of patients received PPI during hospital stay. An important finding to increase awareness among clinicians of appropriate use of PPI as a systematic review and meta-analysis showed an increased risk of small bowel damage using this medication in patients taking nonsteroidal anti-inflammatory drug (NSAID) [19].

There are several limitations to our study. First, this was a single center retrospective study, the sample size was small, and the retrospective description of this data can only determine association but not causality. Second, patient inclusion criteria were based on ICD-10 codes and some patients can potentially be included as LGIB, but may have had small-bowel bleeding. Third, nearly 30% of patients did not undergo colonoscopy during hospitalization, making some final diagnoses not completely reliable. Fourth, around 14% of patient’s baseline Hgb data was not available. This missing data could increase the percentage of patients with previously diagnosed anemia. Lastly, patients can have readmission to an outside hospital which would not have been captured in our readmission rates.