O. P. Owodunni1, B. D. Lau2,6,10, K. L. Florecki1, K. L. Webster1,9, D. L. Shaffer3,7, D. B. Hobson3,7, P. S. Kraus8, C. G. Holzmueller1,9, J. K. Canner4, M. B. Streiff5,9, E. R. Haut1,9,10,11 1The Johns Hopkins University School Of Medicine,Acute Care/Surgery,Baltimore, MD, USA 2Johns Hopkins University School Of Medicine,Radiology,Baltimore, MD, USA 3Johns Hopkins University School Of Medicine,Surgery,Baltimore, MD, USA 4Johns Hopkins University School Of Medicine,Surgery Center For Outcomes Research,Baltimore, MD, USA 5Johns Hopkins University School Of Medicine,Hematology/Medicine,Baltimore, MD, USA 6Johns Hopkins University School Of Medicine,Health Sciences Informatics,Baltimore, MD, USA 7Johns Hopkins University School Of Medicine,Nursing,Baltimore, MD, USA 8Johns Hopkins University School Of Medicine,Pharmacy,Baltimore, MD, USA 9The Armstrong Institute for Patient Safety and Quality,Baltimore, MD, USA 10The Johns Hopkins Bloomberg School of Public Health,Health Policy And Management,Baltimore, MD, USA 11Johns Hopkins University School Of Medicine,Anesthesiology And Critical Care Medicine,Baltimore, MD, USA
Introduction: Acute myocardial infarction (AMI) and venous thromboembolism (VTE) are common cardiovascular disorders and a significant cause of morbidity and mortality in surgical patients. These disorders are commonly monitored and publicly reported quality-of-care measures. Many health outcomes research rely on readily available, inexpensive data, originally collected for administrative and billing purposes. However, the inherent limitations of administrative databases are often overlooked. We hypothesized that the National Inpatient Sample (NIS) would have a high prevalence of underreporting of diagnostic procedures for AMI and VTE.
Methods: We retrospectively analyzed inpatient encounters with a primary diagnosis of AMI or VTE (comprising deep vein thrombosis [DVT], or pulmonary embolism [PE]) in the 2012-2014 Healthcare Cost & Utilization Project NIS database. We identified standard diagnostic procedures for each diagnosis: venous duplex ultrasound (US) for DVT, electrocardiogram (EKG) for AMI, and chest computer tomography (CT) scan, pulmonary angiography, echocardiography, and nuclear medicine ventilation/perfusion (V/Q) scan for PE. We appropriated survey weighting statistical approaches and calculated the proportions of patients with a documented corresponding diagnostic test for their diagnosis.
Results: All three cardiovascular outcomes of interest had a similarly low proportion of patients with documentation of a corresponding diagnostic procedure. Only 0.26% (n = 4,800) of patients with reported AMI had an EKG. Just 2.20% (n = 9,655) of patients with reported DVT events had a peripheral vascular ultrasound. For patients with a PE diagnosis, 4.10% (n= 20,825) had a CT scan, 1.60% (n= 7,965) had pulmonary angiography, 5.20% (n= 26,390) had echocardiography, and 0.48% (n= 2,460) had a V/Q scan (see table 1).
Conclusion: An extremely small proportion of diagnostic procedures are documented for inpatient encounters with a documented diagnosis of AMI, DVT, and PE. This finding calls into question the validity of using NIS and other administrative databases to examine health care use and outcomes for at least some categories of diagnostic tests. Unfortunately, NIS does not provide enough granular data to control for differences in diagnostic procedure use which may lead to surveillance bias in outcomes reporting. Clinical researchers and policy makers must understand and acknowledge the limitations inherent in these data when using them for pay-for-performance initiatives and hospital benchmarking.
