Pre-emptive correction for haemodialysis arteriovenous access stenosis

BackgroundGuidelines recommend routine arteriovenous (AV) graft and fistula surveillance (technology-based screening) in addition to clinical monitoring (physical examination) for early identification and pre-emptive correction of a stenosis before the access becomes dysfunctional. However, consequences on patient-relevant outcomes of pre-emptive correction of a stenosis in a functioning access as opposed to deferred correction, i.e. correction postponed to when the access becomes dysfunctional, are uncertain.ObjectivesWe aimed to evaluate 1) whether pre-emptive correction of an AV access stenosis improves clinically relevant outcomes; 2) whether the effects of pre-emptive correction of an AV access stenosis differ by access type (fistula versus graft), aim (primary and secondary prophylaxis), and surveillance method for primary prophylaxis (Doppler ultrasound for the screening of functional and anatomical changes versus measurement of the flow in the access); and 3) whether other factors (dialysis duration, access location, configuration or materials, algorithm for referral for intervention, intervention strategies (surgical versus radiological or other), or study design) explain the heterogeneity that might exist in the effect estimates.Search methodsWe searched the Cochrane Kidney and Transplant Specialised Register to 30 November 2015 using search terms relevant to this review.Selection criteriaWe included all studies of any access surveillance method for early identification and pre-emptive treatment of an AV access stenosis.Data collection and analysisWe extracted data on potentially remediable and irremediable failure of the access (i.e. thrombosis and access loss respectively); infection and mortality; and resource use (hospitalisation, diagnostic and intervention procedures). Analysis was by a random effects model and results expressed as risk ratio (RR), hazard ratio (HR) or incidence rate ratio (IRR) with 95% confidence intervals (CI).Main resultsWe identified 14 studies (1390 participants), nine enrolled adults without a known access stenosis (primary prophylaxis; three studies including people using fistulas) and five enrolled adults with a documented stenosis in a non-dysfunctional access (secondary prophylaxis; three studies in people using fistulas). Study follow-up ranged from 6 to 38 months, and study size ranged from 58 to 189 participants. In low-to moderate-quality evidence (based on GRADE criteria) in adults treated with haemodialysis, relative to no surveillance and deferred correction, surveillance with pre-emptive correction of an AV stenosis reduced the risk of thrombosis (RR 0.79, 95% CI 0.65 to 0.97; I-2 = 30%; 18 study comparisons, 1212 participants), but had imprecise effect on the risk of access loss (RR 0.81, 95% CI 0.65 to 1.02; I-2 = 0%; 11 study comparisons, 972 participants). In analyses subgrouped by access type, pre-emptive stenosis correction did not reduce the risk of thrombosis (RR 0.95, 95% CI 0.8 to 1.12; I-2 = 0%; 11 study comparisons, 697 participants) or access loss in grafts (RR 0.9, 95% CI 0.71 to 1.15; I-2 = 0%; 7 study comparisons; 662 participants), but did reduce the risk of thrombosis (RR 0.5, 95% CI 0.35 to 0.71; I2 = 0%; 7 study comparisons, 515 participants) and the risk of access loss in fistulas (RR 0.5, 95% CI 0.29 to 0.86; I-2 = 0%; 4 studies; 310 participants). Three of the four studies reporting access loss data in fistulas (199 participants) were conducted in the same centre. Insufficient data were available to assess whether benefits vary by prophylaxis aim in fistulas (i.e. primary and secondary prophylaxis). Although the magnitude of the effects of pre-emptive stenosis correction was considerable for patient-centred outcomes, results were either heterogeneous or imprecise. While pre-emptive stenosis correction may reduce the rates of hospitalisation (IRR 0.54, 95% CI 0.31 to 0.93; I-2 = 67%; 4 study comparisons, 219 participants) and use of catheters (IRR 0.58, 95% CI 0.35 to 0.98; I-2 = 53%; 6 study comparisons, 394 participants), it may also increase the rates of diagnostic procedures (IRR 1.78, 95% CI 1.18 to 2.67; I-2 = 62%; 7 study comparisons, 539 participants), infection (IRR 1.74, 95% CI 0.78 to 3.91; I-2 = 0%; 3 studies, 248 participants) and mortality (RR 1.38, 95% CI 0.91 to 2.11; I-2 = 0%; 5 studies, 386 participants).In general, risk of bias was high or unclear in most studies for many domains we assessed. Four studies were published after 2005 and only one had evidence of registration within a trial registry. No study reported information on authorship and/or involvement of the study sponsor in data collection, analysis, and interpretation