Surveillance system assessing the need for updating systematic reviews

Article type
Authors
Ahmadzai N1, Newberry SJ2, Maglione M2, Tsertsvadze A1, Ansari MT1, Hempel S2, Motala A2, Tsouros S1, Garritty C1, Schneider Chafen J3, Shanman R2, Skidmore B1, Moher D1, Shekelle PG4
1Knowledge Synthesis Group, Clinical Epidemiology Program, Ottawa Hospital Research Institute, Centre for Practice-Changing Research, 501 Smyth Rd, Ottawa, ON, K1H 8L6, Canada
2Southern California Evidence-based Practice Center (SCEPC), The RAND Corporation, 1776 Main Street, PO Box 2138, Santa Monica, CA 90401, USA
3Stanford University, Stanford University 117 Encina Commons, Stanford, CA 94305-6019, USA
4Southern California Evidence-based Practice Center (SCEPC), The RAND Corporation, 1776 Main Street, PO Box 2138, Santa Monica, CA 90401, USA; Veterans Affairs Greater Los Angeles Healthcare System, 11301 Wilshire Boulevard, Los Angeles, CA 90073, USA
Abstract
Background: Systematic reviews (SRs) go out of date as new evidence emerges. Instead of fixed periodic updating, an efficient SR currency surveillance program could be less resource intensive.

Objectives: Based on methods we developed previously, three Evidence-based Practice Centers were tasked to establish and execute a surveillance program to identify the need for updating a cohort of Comparative Effectiveness Reviews (CERs) commissioned by the Agency for Healthcare Research and Quality.

Methods: We incorporated two existing methods of updating signal detection into a step-by-step surveillance approach and assessed the currency of 24 CERs. New evidence was identified from: (a) literature searches, (b) expert opinion, and (c) safety alerts. We determined the currency of conclusions in the reviews, and, based on their potential impact on decision-making, the updating priority (low, medium, or high) of CERs. Low or medium priority CERs were reassessed every 6 months (Fig. 1). We did not update the reviews.

Results: The CERs mainly compared effectiveness and safety of pharmaceuticals and surgical procedures for various health conditions (Table 1). The median number of studies in the original SRs was 104 (range, 8–436), new studies were 15 (range, 0–35), and the expert response rate was 35% (71% for re-assessed CERs). Of the nine identified safety alerts only one influenced the updating priority of one CER. Of the 24 CERs, 2 were classified as high (8%), 5 as medium (21%), and 17 as low (71%) priority for updating from 11 to 62 months after their last search date. Of the 10 re-assessed CERs reassessed after 6 months, updating priority changed for only 1 CER.

Conclusions: We established a surveillance program and evaluated 24 CERs. The application of the program is practical and efficient for assessing the need for updating SRs across a wide range of health interventions.