Innovations in clinical documentation integrity practice: Continual adaptation in a data-intensive healthcare organisation

Background Numbers of clinical documentation integrity specialists (CDIS) and CDI programs have increased rapidly. CDIS review patient records concurrently with patient admissions and visits to ensure that information is accurate, complete and non-ambiguous, and query clinicians when they see opportunities for improving data. The occupation was initially focused on improving data for reimbursement, but rapid changes to clinical coding requirements, technologies and payment systems led to a quickly evolving role for CDI programs and changes in CDIS practice. Objective This case study seeks to uncover the ongoing innovation and adaptation occurring in a CDI program by tracing the evolution of a single CDI program over time. Method We present a case study of the CDI program at the HonorHealth hospital system in Arizona. Results The HonorHealth CDI program holds a unique hybrid expertise and role within the healthcare organisation that allows it to rapidly adapt to support emergent demands both internal and external to the organisation, such as supporting accurate data collection for the COVID-19 pandemic. Conclusion CDIS are a vital component in present data-intensive resourcing efforts. The hybrid expertise of CDIS and capacity for adaption and relationship building has enabled the HonorHealth CDI program to adapt rapidly to meet a growing array of clinical documentation integrity needs, including emergent needs during the COVID-19 pandemic. Implications The HonorHealth case study can guide other CDI programs in adaptation of the CDI role and practices in response to changing organisational needs.

Identification of root causes of clinical coding problems in Iranian hospitals

Background Improving the quality of coded data requires the identification and evaluation of the root causes of clinical coding problems to inform appropriate solutions. Objective The objective of this study was to identify the root causes of clinical coding problems. Method Twenty-one clinical coders from three cities in Iran were interviewed. The five formal categories in Ishikawa's cause-and-effect diagram were applied as pre-determined themes for the data analysis. Results The study indicated 16 root causes of clinical coding problems in the five main themes: (i) policies, protocols, and processes (lack of clinical documentation guidelines; lack of audit of clinical coding and feedback to clinical coders; the long interval between documentation and clinical coding; and not using coded data for reimbursement; (ii) individual factors (shortage of clinical coders; low-skilled clinical coders; clinical coders' insufficient communication with physicians; and the lack of continuing education; (iii) equipment and materials (incomplete medical records; lack of access to electronic medical records and electronic coding support tools; (iv) working environment (lack of an appropriate, dynamic, and motivational workspace; and (v) management factors (mangers' inattention to the importance of coding and clinical documentation; and to providing the required staff support. Conclusion The study identified 16 root causes of clinical coding problems that stand in the way of clinical coding quality improvement. Implications The quality of clinical coding could be improved by hospital managers and health policymakers taking these problems into account to develop strategies and implement solutions that target the root causes of clinical coding problems.

DeepDRG: Performance of Artificial Intelligence Model for Real-Time Prediction of Diagnosis-Related Groups

Nowadays, the use of diagnosis-related groups (DRGs) has been increased to claim reimbursement for inpatient care. The overall benefits of using DRGs depend upon the accuracy of clinical coding to obtain reasonable reimbursement. However, the selection of appropriate codes is always challenging and requires professional expertise. The rate of incorrect DRGs is always high due to the heavy workload, poor quality of documentation, and lack of computer assistance. We therefore developed deep learning (DL) models to predict the primary diagnosis for appropriate reimbursement and improving hospital performance. A dataset consisting of 81,486 patients with 128,105 episodes was used for model training and testing. Patients’ age, sex, drugs, diseases, laboratory tests, procedures, and operation history were used as inputs to our multiclass prediction model. Gated recurrent unit (GRU) and artificial neural network (ANN) models were developed to predict 200 primary diagnoses. The performance of the DL models was measured by the area under the receiver operating curve, precision, recall, and F1 score. Of the two DL models, the GRU method, had the best performance in predicting the primary diagnosis (AUC: 0.99, precision: 83.2%, and recall: 66.0%). However, the performance of ANN model for DRGs prediction achieved AUC of 0.99 with a precision of 0.82 and recall of 0.57. The findings of our study show that DL algorithms, especially GRU, can be used to develop DRGs prediction models for identifying primary diagnosis accurately. DeepDRGs would help to claim appropriate financial incentives, enable proper utilization of medical resources, and improve hospital performance.

EP.FRI.132 Current practice in the investigation and management of post-operative ileus

Post-operative ileus is non-obstructive impairment of gastrointestinal motility after surgery. Treatment involves nil-by-mouth, correction of electrolytes and stopping aggravating medications. There is no national/trust guideline for investigation and management of post-operative ileus. Aim To assess current practice in investigation and management of post-operative ileus. Method All patients with a diagnosis of post-operative ileus on surgical wards at a London hospital between November - December 2020 were identified using clinical coding. Their clinical notes were reviewed and common themes explored. Results 16 patients were diagnosed with post-operative ileus. 12 (75%) patients were male, with a median age of 60 (IQR: 28). 10 (63%) were open procedures, majority being colorectal (10; 63%), cardiothoracic (3; 19%), urological (2; 13%), vascular (1; 6%) operations. Average time between operation and diagnosis was 5.2 (range: 2-14) days. Only 2 (12.5%) patients had documented auscultation of bowel sounds at initial assessment. 8 (50%) patients had imaging to confirm diagnosis (AXR; 5, CT scan; 4). Majority (15; 94%) of patients had serum magnesium and potassium checked at diagnosis. Patients with serum potassium <4.0mmol/L (5) and magnesium <0.7mmol/L (2) had intravenous supplementation. 14 (88%) were administered Hartmann’s solution. 10 (63%) patients were made nil-by-mouth and 15 (94%) had a Ryles tube inserted. 2 (12.5%) patients were prescribed a prokinetic, and only 2 had either opioids or laxatives stopped. No patients were offered gum. Conclusion There is apparent clinical variation in investigation and management of post-operative ileus. We plan to develop an evidence-based trust guideline to reduce unwarranted clinical variation.

SP4.1.12 The impact of COVID-19 on the pattern of presentations with acute lower gastrointestinal bleeding

Aims Following the initial surge of COVID-19 cases in Spring 2020, data began to emerge on the negative impact of the disease on outcomes for serious medical conditions such as heart disease and stroke. However, the impact of COVID-19 on the pattern of emergency presentations with lower gastrointestinal bleeding (LGIB) has not been published to date. Methods We designed a clinical coding search strategy to identify all adult patients with acute LGIB presenting to the Emergency Department (ED) and the Surgical Emergency Unit (SEU) at a UK university hospital from January to July 2020. For context, data on number of overall presentations to ED and SEU were also collated for the same period. Results 169 patients (median age 63 (16-94) years, 54.4% male) with acute LGIB were identified across the six months. A graphical representation of these data demonstrated notable trends. Overall weekly ED attendances dropped by 52.0% after the national lockdown (from 1500 to 720 patients), before returning to 77.7% of pre-lockdown levels by the end of the study period (1165 patients). Pre-lockdown, there was a fluctuating number of weekly attendances with acute LGIB. After lockdown, consistently fewer patients presented and there was a reduction in the variability of numbers week on week. Conclusions These novel data support recently published trends demonstrating a post-lockdown fall in emergency attendances, although the decrease in weekly attendance with acute LGIB was less marked, perhaps reflecting the concerning nature of this symptom for patients.

921 Management of Acute Cholecystitis - the MACHO Study

Background Acute cholecystitis is a common surgical condition. Gold standard treatment is index cholecystectomy, although there are reasons this might not be offered. The aim of this study was to explore treatments and outcomes in patients with acute cholecystitis. Method A multicentre retrospective study was carried out to identify a historic three-month cohort. Patients were identified through clinical coding. Demographics, clinical outcomes, comorbidities, Tokyo grade, and intervention descriptors were collected. Logistic regression was performed to identify characteristics of patients receiving a drain, and to propensity match for clinical outcomes. Results Seven centres reported on 1130 patients. Median age was 62 years, and 145 (12.8%) had grade III cholecystitis. Grade III cholecystitis was present in 19 (25.6%) of those who underwent cholecystostomy, 34 (9.3%) of those who underwent index cholecystectomy, and 92 (13.3%) of those who were conservatively managed. Overall complication rates were higher for those managed with cholecystostomy (36.5%) or conservatively (22.6%) vs index cholecystectomy (7.5%) (p < 0.001). Logistic regression found CCI and grade III cholecystitis were associated with increased rates of any complication. Increased CCI and grade II/III cholecystitis were associated with increased rates of major complications. Conclusions 'Hot' laparoscopic cholecystectomy seems to be offered to mild cases in fit patients. Patients with grade III disease and moderate comorbidities may not have cholecystectomy in a timely manner, leaving them at risk of repeated severe episodes.

1364 Quality Control of Discharge Summaries for Endocrine Surgery Patients; A Closed Loop Audit

Aim A patient's discharge summary (TTO) should be accurate. Most of them are conducted by junior doctors at the beginning of their medical training. The information mentioned in a TTO ensures patient safety, continuity of care as well as correct clinical coding for the NHS. Therefore, a re-audit was designed to check the quality of the discharge summaries of endocrine surgical patients In Nottingham City Hospital i.e., the type of operation, diagnosis, or postoperative instructions. Method The first cycle included all the TTOs for the endocrine surgical patients operated from April 2018 to November 2018. Then we re-audited those who had endocrine surgeries from April 2019 to November 2019. NOTIS e-TTO, Bluespier theatre lists and Medway were used to retrieve the data. All general surgery patients were excluded. Results 142 and 104 patients TTOs were included in each audit cycle, respectively. Type of operation was improved from 84% to 95% in the second cycle. Correct diagnosis was reported from 68% to 72% in the second cycle audit. Conclusions The introduction of electronic operation notes in our practice improved the correct clinical coding for the type of operation mentioned in the TTO. The accuracy of correct diagnosis remains suboptimal. Therefore, education of junior doctors and an idea of double-checking from a more senior colleague should be assessed.