Is there a positive volume-outcome relationship in peripheral vascular surgery? Results of a systematic review

Objectives: to examine the evidence for the existence, otherwise, of a posi tive volume-outcome relationship in the area of peripheral vascular surgery . Design: systematic overview of prospective or retrospective volume-outcome studies. Data sources: seven bibliographic databases were searched for English-langu age articles published between 1986 and 1998. Study selection: thirty-six articles published in peer-reviewed journals; e xcluding editorials, letters or abstracts; and addressing volume and outcom e in peripheral vascular surgery. Criteria were applied and agreed by conse nsus between two of the authors. Data extraction: the articles identified were independently assessed by two of the authors. Studies were categorised into three distinct areas - "caro tid endarterectomy" (17 studies), "abdominal aortic aneurysm repair" (16 st udies) and "other vascular interventions" (four studies). Within each categ ory studies were further classified according to full adjustment, partial a djustment or no adjustment for case mix. Where discrepancies arose, decisio ns were referred to a third author for arbitration. Data synthesis: findings for carotid endarterectomy identified It positive volume-outcome relationship for both mortality and stroke at the physician level. There was less support for a positive relationship for mortality at the hospital level, and no evidence of benefits for stroke in higher volume hospitals. If only studies making a full adjustment for case mix are Inclu ded, there is no clear support from statistically significant evidence for or against a positive volume-outcome relationship. For repair of unruptured abdominal aortic aneurysms there is evidence of a positive volume-outcome relationship at both the physician and hospital level, with evidence being particularly strong at the level of the hospital. For ruptured aneurysms th e evidence is suggestive of there not being a positive volume-outcome relat ionship at the hospital level, while for physicians the evidence is move ba lanced with no clear support either way. For "other vascular interventions" there were insufficient studies (n=4) from which to draw, meaningful concl usions. Conclusions: our results shop that evidence of a relationship between volum e and outcome in peripheral vascular surgery may be attributable to factors such as lack of adjustment for case-mix, different definitions of volume a nd poor quality of studies, especially those of retrospective design. Futur e studies should address these deficiencies by making full adjustment for c ase mix and by being prospective in design.