Optimal statin therapy reduces cardiovascular events. We have shown that optimal statin therapy is underused in Veterans with CVD. This can be due to clinical inertia or statin associated side effects which providers usually document in text notes and may not be well captured in structured datasets.
Our objectives are to (Aim 1) identify reasons for suboptimal statin use using structured data and provider text notes using NLP; (Aim 2) understand provider and patient perspectives on statin intolerance and refine a communication aid targeting providers to improve guideline-concordant statin use; and (Aim 3) pilot test an intervention to improve optimal statin use in Veterans with CVD using the communication aid refined in Aim 2.
For Aim 1, we will randomly identify VA patients with CVD on optimal, suboptimal and no statins, and partition them into training and test sets. We will train our NLP system to achieve sensitivity and specificity of >90%, compared with manual chart review to identify reason for suboptimal stain use. In Aim 2, we will conduct interviews with providers and patients to elicit their perspectives on clinical inertia and statin intolerance. These interviews will help refine the communication aid for Aim 3. In Aim 3, we will conduct a pilot trial with Houston and Nashville VAMC PACTs serving as intervention sites. PACTs at intervention sites will receive the communication aid to assist them with statin initiation and/or titration in CVD patients on suboptimal statins. At usual care sites, PACT providers will only receive a quarterly report of the proportion of their CVD patients on suboptimal statins. Our primary outcome is change in the proportion of CVD patients receiving optimal statin therapy.
We developed an initial corpus of 465 patients for training/testing the NLP system. We selected notes with standard titles that were most likely to contain documentation of adverse events related to statin usage. We used an iterative process to develop an annotation schema consisting of concepts of interest within the documents. Two nurse annotators identified an initial group of concepts within a small batch of 10-20 documents. Upon completion, annotation was discussed among the research team for further refinement and finalization.
To create document sets for annotation, we used one document from each patient. Documents were stratified so that each set of documents contained approximately 25% African Americans and 75% Non-African Americans, 25% females and 75% males, and 50% were on no statins, 25% on low statin, and 25% on moderate statin. Currently, a total of 171 documents (3 groups of 57) from 171 patients were selected for annotation. Within the first 57 documents, we identified 8 patients with statin-related adverse events, which corresponds to a prevalence rate of 14%.
Our results identify the vast majority of high-risk Veterans not on optimal statin therapy due to intolerance versus clinical inertia. Our communication aid will identify strategies to initiate or titrate statins in high-risk Veterans. These results will be important for the VA Health Care System to identify patients with "true statin intolerance" who will be future candidates for expensive new drugs recently approved by the FDA.
- Okunrintemi V, Khera R, Spatz ES, Salami JA, Valero-Elizondo J, Warraich HJ, Virani SS, Blankstein R, Blaha MJ, Pawlik TM, Dharmarajan K, Krumholz HM, Nasir K. Association of Income Disparities with Patient-Reported Healthcare Experience. Journal of general internal medicine. 2019 Feb 19.
- Mishra SR, Ghimire S, Shrestha N, Shrestha A, Virani SS. Socio-economic inequalities in hypertension burden and cascade of services: nationwide cross-sectional study in Nepal. Journal of Human Hypertension. 2019 Jan 18.
- Virani SS, Akeroyd JM, Nambi V, Michos ED, Morris PB, Nasir K, Smith SC, Stone NJ, Petersen LA, Ballantyne CM. Applicability and Cost Implications for Proprotein Convertase Subtilisin/Kexin Type 9 Inhibitors Based on the ODYSSEY Outcomes Trial. Circulation. 2019 Jan 15; 139(3):410-412.
- Hira RS, Kataruka A, Akeroyd JM, Ramsey DJ, Pokharel Y, Gurm HS, Nasir K, Deswal A, Jneid H, Alam M, Ballantyne CM, Petersen LA, Virani SS. Association of Body Mass Index With Risk Factor Optimization and Guideline-Directed Medical Therapy in US Veterans With Cardiovascular Disease. Circulation. Cardiovascular quality and outcomes. 2019 Jan 1; 12(1):e004817.
- Virani SS, Ballantyne CM. Low-Density Lipoprotein Cholesterol: Is 160 the New 190? Circulation. 2018 Nov 20; 138(21):2326-2329.
Prevention, Prognosis, Treatment - Implementation
Best Practices, Care Management Tools, Implementation