|
|
Back to 2015 Annual Meeting Program
Validation of Small Bowel Obstruction and Abdominal Surgery Diagnostic Codes in an Electronic Medical Records Database
Frank I. Scott*1, 2, Ronac Mamtani2, 3, David S. Goldberg1, 2, Najjia N. Mahmoud4, James Lewis1, 2
1Gastroenterology, University of Pennsylvania, Philadelphia, PA; 2Center for Clinical Epidemiology and Biostatistics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA; 3Department of Medicine, Division of Hematology/Oncology, University of Pennsylvania, Philadelphia, PA; 4Department of Surgery, University of Pennsylvania, Philadelphia, PA
BACKGROUND: Post-operative small bowel obstructions (SBO) affect ~5% of individuals after an abdominal surgery. Post-operative adhesion formation is estimated to be responsible for up to 75% of SBOs, and the incidence has not changed over time. There are limited clinical data on clinical risk factors for adhesion-related SBO, in part due to the lack of a large population-representative database with validated surgical events and complications such as SBO.
AIMS: To determine the positive predictive value (PPV) of diagnostic codes for intra-abdominal surgery and SBO in a large computerized database.
METHODS: The study was conducted using The Health Improvement Network (THIN), a large database within the United Kingdom (UK) derived directly from general practitioner (GP) electronic medical records. THIN contains >9.5 million patients, distributed among over 480 practices. Medical diagnoses and surgical procedures are recorded by the GP within the scope of care using the Read coding system. THIN data has been validated for multiple acute and chronic medical conditions and drug exposures. Data on the reliability of surgical codes are limited. Individuals >18 years old who were registered within a THIN practice for >1 year without a surgical event who then had an incident intra-abdominal or pelvic surgery were identified using a list of 1040 Read codes for surgical procedures. We then identified patients with a potential subsequent SBO using a list of 18 Read codes. We randomly sampled 205 individuals from this cohort to confirm the index surgical procedure and subsequent SBO via completion of a detailed questionnaire to the patients' GPs. We also requested confirmatory documents such as operative notes and discharge summaries. The PPV and 95% CI of surgical SBO codes were calculated from the returned questionnaires using the additional documents as the gold standard when available. Additional analyses assessed the PPV of only the records with supplemental documents and the number of surgeries and SBOs occurring within 30 days of the coded date within THIN.
RESULTS: 178 of 205 (86.8%) of questionnaires were returned. 130(73%) responses included confirmatory documents. The median age at initial surgery was 61. 6 (IQR 47.1-70.4). The median age at SBO was 63.2 (IQR 49.6-74.2). The median time to SBO after surgery was 475 days (IQR 65-1496 days). Of the 178 returned questionnaires, 166 confirmed the surgery (PPV 93.3%, 95%CI: 0.89- 0.96), and 153 confirmed the SBO (PPV 86.0%, 95%CI: 0.80 - 0.91). The PPV was similar when looking at individuals with confirmatory documents and when restricting the returned code to within 30 days (Table 1).
CONCLUSIONS: The recording of surgery and SBO was highly reliable in the Health Improvement Network, a novel primary care database. THIN can be used to explore the epidemiology of intra-abdominal surgery and SBO
Table 1: Positive Predictive Value (PPV) for surgery and SBO
| Variable | # confirmed events/total codes | PPV (95%CI) | | Surgical events | 166/178 | 93.3% (89-96%) | | ---with confirmatory documents | 126/130 | 96.9% (94-99%) | | ---within 30d of date | 156/178 | 87.6% (82-92%) | | | | | SBO events | 153/178 | 86.0% (80-91%) | | ---with confirmatory documents | 116/130 | 89.2% (84-95%) | | ---within 30d of date | 147/178 | 82.6% (76-88%) |
Back to 2015 Annual Meeting Program
|
