Determinants of follow-up failure in patients undergoing vascular surgery procedures.

Document Type


Publication Date




Journal Title

Annals of vascular surgery

MeSH Headings

Aged, Aged, 80 and over, Aortic Aneurysm, Abdominal, Chi-Square Distribution, Endarterectomy, Carotid, Endovascular Procedures, Female, Humans, Logistic Models, Lost to Follow-Up, Male, Middle Aged, Multivariate Analysis, New England, Odds Ratio, Postoperative Care, Process Assessment (Health Care), Registries, Risk Factors, Time Factors, Treatment Outcome, Vascular Grafting, Vascular Surgical Procedures


BACKGROUND: The Vascular Study Group of New England (VSGNE) requires documentation of follow-up for >80% of patients at least 9 months postprocedure. However, many participating groups fall short of this goal. We sought to identify factors independently associated with loss to long-term follow-up (LTF).

METHODS: The VSGNE was queried from 2008 to 2012, for all carotid endarterectomy (CEA), endovascular aneurysm repair (EVAR), open abdominal aortic aneurysm repair (OPEN), infrainguinal bypass (INFRA), and suprainguinal bypass (SUPRA) procedures in patients who survived greater than 9 months postprocedure. Our primary endpoint was loss to LTF, with LTF defined as documentation of a phone call or office visit ≥9 months postprocedure. Multivariable logistic regression was used to identify independent predictors of loss to LTF. Covariates included patient and procedural characteristics, and treatment center/physician. Relative contributions of covariates to the model were assessed by evaluation of the relative model Wald chi-squared values.

RESULTS: We identified 14,452 procedures (6567 CEA, 2391 EVAR, 3356 INFRA, 979 OPEN, and 1159 SUPRA). Of those, 4669 (32%) were lost to LTF. Rates of loss to LTF varied by center, and ranged from 9.8% to 100%. Independent predictors of loss to LTF were history of coronary artery disease or percutaneous coronary artery intervention (odds ratio [OR] 1.4, 95% confidence interval [CI] 1.1-1.7), procedure type (OPEN, OR 1.4, 95% CI: 1.2-1.7; CEA, OR 1.2, 95% CI: 1.1-1.4; referent, EVAR), and discharge to rehab (OR 1.2, 95% CI: 1.1-1.4; referent, home). Center variation was the strongest determinant of loss to LTF with a model χ2 over 40 times as large as the second strongest determinant.

CONCLUSIONS: LTF is central to outcome reporting and is vital to the success of any registry effort. In the VSGNE experience, center variation is the strongest predictor of loss to LTF, outweighing patient and procedural factors. Other predictors of loss to LTF included history of coronary revascularization, procedure type, no prior history of congestive heart failure, and discharge location. High performing centers likely have specific process measures that decrease loss to LTF. As the Society for Vascular Surgery Vascular Quality Initiative continues to roll out nationally, high performing centers in VSGNE should be studied to document and propagate best practices for minimizing loss to LTF.




Division of Vascular Surgery, Maine Medical Center, Portland ME.

First Page


Last Page