A Pilot Study of the Accuracy of Diagnostic Coding for Venous Thromboembolism within an Administrative Dataset of Emergency Department Visits.

Abstract Introduction: With the advent of low molecular weight heparins, venous thromboembolism (VTE) management has largely shifted from the inpatient to the outpatient setting. Yet there is a paucity of data addressing the incidence of outpatient VTE and the process and cost of care related to its management. Large administrative databases are often a good source of data for population-based research. However, concerns about their accuracy necessitate that they be validated first. The National Ambulatory Care Reporting System (NACRS) is an administrative database launched by the Canadian Institutes of Health Information (CIHI) in 2001. Hospitals submit abstracted information on all emergency department (ED) visits to CIHI. If the accuracy of VTE diagnosis codes within NACRS is known, this database could be used to address the above issues on a population level. Methods: Prior to a large-scale validation of VTE codes within NACRS, we conducted a pilot study at our large, tertiary care hospital - Sunnybrook Health Sciences Centre in Toronto, Canada. Our goal was to determine the accuracy of NACRS coding for VTE. To define a cohort of patients with suspected VTE who presented to our ED between January 1 and December 31, 2005 we generated a list of all visits with a procedure code of either duplex ultrasound, contrast enhanced chest CT or VQ scan. We also generated a list of all visits with a NACRS diagnosis code of VTE (ICD-10-CA codes for phlebitis and thrombophlebitis - I80.1, I80.2, I80.3, 180.8, 180.9 and pulmonary embolism - I26.0 and I26.9). The latter list captured patients who had imaging done at an outside facility. The lists were merged and duplicate entries removed. Electronic and/or paper chart review was carried out to confirm the diagnosis of VTE for all visits, based on positive diagnostic imaging results. Discrepancies were resolved by consensus between two physicians. Results: During the study period, there were over 40,000 visits to our ED. Using the above algorithm, 1149 patient visits were generated with either a procedure code for the above radiological tests or a diagnosis code for VTE. 348 visits had imaging done for reasons other than VTE (ie. trauma or malignancy), and 17 visits had no recorded diagnostic imaging. These visits were excluded. Of the remaining 784 visits, 121 had a diagnosis code of VTE and a confirmed diagnosis of VTE on chart review (true positives). 10 visits were coded as VTE but a diagnosis of VTE could not be confirmed on chart review (false positives). 30 visits were not coded as VTE but had a confirmed diagnosis of VTE (false negatives). 623 visits were neither coded as VTE nor had a confirmed diagnosis of VTE (true negatives). The prevalence of VTE in our sample was therefore 19.3%. The sensitivity of NACRS coding was 80.1% (95% CI 72.7% to 86.0%), while the specificity was 98.4% (95% CI 97.0% to 99.2%). Conclusion: NACRS coding for VTE is highly specific, but less sensitive. This suggests NACRS may be useful for studying outpatient VTE at a population level, though a multi-site validation is required. Using a search algorithm to identify patients with suspected VTE based on procedure and diagnosis codes is feasible, given the non-specific nature of the presenting symptoms of VTE. This algorithm can be used for similar validation studies.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.