We used a workgroup consensus process to define NFCOI, suggest ways to elicit information on NFCOI, and propose methods to mitigate the bias from NFCOI. The workgroup for this chapter included 16 individuals from 10 EPCs and AHRQ. All members of the workgroup had specifically expressed interest in working on the guidance, and many had prior experience assessing and handling COI of staff, consultants, and expert panelists. The workgroup lead (MV) set the scope and timeline, scheduled and led conference calls, developed and distributed meeting summaries, assigned and coordinated tasks of group members, drafted sections of the guidance chapter, and edited materials. All members participated in monthly conference calls, made suggestions regarding the scope of the chapter, and submitted written contributions to the chapter.

The workgroup identified several examples of NFCOI, and in discussions, expanded or collapsed categories as relevant. For instance, an early draft separately discussed NFCOI in cases of intense advocacy and policy debate, but feedback from reviewers and workgroup members suggested that these two sources of NFCOI should be combined because the central issues were similar. We developed questions to elicit responses for each source of bias separately, and then combined and revised questions to flow in a logical sequence. This document, including specific examples and NFCOI questions, was revised in response to peer review. We also posted the document for public comment but received no feedback. We reviewed published COI policies and recommendations from several prominent organizations that sponsor, conduct, publish, or are otherwise involved in the production of SRs. These organizations include the IOM,7,9 the Cochrane Collaboration,10 the U.S. Department of Health and Human Services,6 the International Committee of Medical Journal Editors,16 the National Institute for Health and Clinical Excellence,12 and PCORI.11 From these policies, we developed our working definition for NFCOI specifically for SR teams. We also searched for empirical evidence and well-established theoretical frameworks on NFCOI but found none. Therefore, we cannot unequivocally suggest guidance about managing NFCOI across all topics and contexts.

Our approach is to highlight circumstances in which NFCOI may be present; use real-world examples whenever possible, based primarily on our experience with some details altered for reasons of confidentiality. Our purpose was to demonstrate how they may be considered conflicts; and provide guidance on assessing whether the conflicts pose a risk of bias.

Definition of Conflict of Interest

In crafting a definition of COI, we considered and built on definitions other organizations use, particularly the IOM.612,16 We considered whether or not to incorporate three dichotomies: (1) financial versus nonfinancial COI, (2) institutional versus individual COI, and (3) perceived versus real COI. In rejecting these sharp distinctions, we offer a general definition of NFCOI that can be applied across clinical areas and topics to address the wide spectrum of EPC work and EPC contributors, while emphasizing that the primary interest is always the quality and integrity of the SR.

Financial Versus Nonfinancial COI

Financial and nonfinancial COI are not necessarily mutually exclusive. Financial concerns may indirectly drive NFCOI. For example, an individual’s concerns about his or her professional reputation may be nonfinancial, but may be driven in part by concerns about his or her ability to compete for future funding or receive increased compensation. In contrast to some other literature in this area, rather than drawing sharp distinctions between financial and nonfinancial COI, we recognize that these interests may be interrelated.

Individual Versus Institutional COI

In considering whether or not to distinguish between individual and institutional COI, we noted that institutional COI may lead to NFCOI for individuals, although the literature generally defines institutional conflicts in financial terms. For instance, when faculty members review the evidence for a medication in which their university holds the patent, institutional COI may lead to NFCOI for individual investigators with no involvement in the patent because of the risk that the employer might have indirect bearing on the faculty member’s judgment and action in the SR. In addition to these conceptual overlaps, practical considerations also support the disclosure of these interests under NFCOI. Existing FCOI policies fails to elicit these conflicts because they frame disclosure in relation to the individual rather than the institution. Such conflicts are more likely to emerge in evaluating NFCOI. In considering financial versus nonfinancial COIs and individual versus institutional COIs, we do not distinguish among financial, institutional, and individual interests. Instead, we support the use of definitions that collectively call out all such interests as secondary or competing interests that are not mainly financial.

Real Versus Perceived COI

We also considered the issue of real versus perceived COI, particularly in light of recent recommendations in the IOM’s report, Standards for Systematic Review (2011).7,9 Standard 2.2 lays out the IOM’s expectations for SR teams: each team member should disclose potential COI and professional or intellectual bias, and the project lead should exclude individuals with a clear financial conflict and individuals whose professional and intellectual bias diminish the SR’s credibility in the eyes of the intended user. This guidance frames intellectual and professional bias as perceived conflict—in the eyes of the intended user—and thereby dramatically expands the potential range of intellectual perspectives and professional affiliations that might be considered grounds for exclusion. This guidance requires the SR team or the sponsor to speculate about the possible perceptions of the many intended users regarding NFCOI. If the standard is interpreted strictly, SR teams could result that include no participants with expertise in the field and, thus, limited ability to interpret the evidence appropriately. In contrast, another IOM report, “Conflict of Interest in Medical Research, Education, and Practice,” rejects the distinction between actual and perceived COI on two grounds.7 First, such a distinction suggests that perceived COIs are not actual conflicts until the decisionmaker “favors secondary interests over primary interests.” Second, the distinction leads to “overly broad and excessively subjective rules.”7 In other words, the distinction between real and perceived conflicts should be irrelevant to the management of conflicts. According to this view, the goal of the research team should then be to identify and manage the risk of undue influence, not necessarily to eliminate all perceived conflicts, since the research team cannot control the perceptions of all possible readers of the review.

Based on these considerations, the EPC workgroup has chosen to define NFCOI broadly as follows:

A set of circumstances that creates a risk that the primary interest—the quality and integrity of the systematic review—will be unduly influenced by a secondary or competing interest that is not mainly financial.

This definition is largely based on the definition of COI in the IOM report, “Conflict of Interest in Medical Research, Education, and Practice.”7 It departs from definitions in the IOM report, “Standards for Systematic Review,” and the Cochrane Collaboration manual in that it does not maintain a distinction between real and perceived conflicts.7,10