Background: Although the majority of national quality initiatives utilize electronic health record (EHR) or administrative data, their ability to adequately discriminate performance has been brought into question and it is unclear certain outcomes, such as postoperative complications, are accurately ascertained. By comparison, clinical registry data, like the VA Surgical Quality Improvement Program (VASQIP), are widely considered robust for performance evaluation and quality improvement (QI). But, VASQIP data collection is resource intensive?data are manually abstracted by trained local Surgical Quality Nurses (SQNs) for a systematic sample of surgical cases performed at all VA hospitals. VASQIP then uses the data to characterize the quality and safety of surgical care at each hospital based on risk-adjusted 30-day morbidity and mortality rates. Significance: VASQIP data collection practices present two important limitations. First, perioperative outcome rates have significantly decreased the past two decades making it unclear whether systematic case sampling is adequately powered to identify underperforming hospitals. Second, the time required for VASQIP data collection detracts from SQNs? ability to engage in other important job functions, like local QI activities. Because SQNs spend substantial time working with VASQIP data, this represents an important missed opportunity to identify a quality problem when it is evolving rather than when it has already occurred. As such, alternative approaches that can provide reliable data and decrease the burden of data collection would have tangible benefits for other national surgical and non-surgical QI initiatives within VA and the private sector. Innovation: This project is novel because it can change the paradigm regarding the collection of QI data from purely EHR or clinical registry to a more efficient hybrid model that could address reliability concerns associated with the use of EHR (or administrative) data alone. It will also provide real-world, generalizable data that can only be obtained within VA's data platform and can inform VA and the private sector national surgical and non-surgical QI initiatives. We have two national operational partners: 1.) VA National Surgery Office (NSO); 2.) Office of Reporting, Analytics, Performance, Improvement, and Deployment (RAPID). Specific Aims: The overall goal is to address two important questions. First, given low perioperative outcome rates across VA, is systematic sampling robust enough to inform surgical QI? Second, are hybrid data (i.e.: EHR combined with clinical registry variables) a potentially reliable alternative for measuring VA hospital surgical performance? These questions will be explored through the following specific aims: (1) Evaluate whether analyzing all VASQIP-eligible surgical cases, relative to current systematic case sampling, improves negative predictive value (i.e.: decreases false negative rates) for identifying VA hospitals with outlier performance; (2) Compare the use of hybrid EHR and clinical registry data, relative to clinical registry alone, for evaluating risk-adjusted surgical performance at VA hospitals; (3) Explore how more efficient VASQIP data collection could enhance local QI efforts through in-depth, key informant interviews with SQNs. Methodology: This mixed-methods proposal will involve hospital-level, observational studies using VASQIP and Corporate Data Warehouse (CDW) data from patients who underwent non-cardiac surgery (2016-2019) as well as qualitative interviews with SQNs. With comparative effectiveness in mind, these data will be used to explore what would be observed if data from all surgical cases were included in VASQIP and to understand whether other existing VA data sources might improve VASQIP data collection efficiency and enhance local QI. Next Steps: With the NSO, we will prospectively compare the fidelity of hand-abstracted variables to automatable variables from CDW. The implementation plan (supported by the VA National Director of Surgery) will utilize VASQIP?s existing infrastructure by partnering with VINCI to provide the NSO with centralized CDW access (using RAPID?s data access model as a template) allowing automated data collection.

I01HX003127

2021-05-01

2025-04-30