A Study of Reliance Agreement Templates Used by U.S. Research Institutions

Abstract

We performed a content analysis of reliance agreement templates from 73 of the top U.S. research institutions ranked by research funding. 67.1% of institutions in our sample use the Office of Human Research Protections (OHRP) template and 8.2% use the SMART IRB template. Although a significant percentage of institutions (45.2%) use their own custom template, many of these also use the OHRP template or the SMART IRB template. 21.9% of institutional templates have only the OHRP minimum 10 contractual provisions (or elements), 27.4% have 11 elements, 26% have 12-19 elements, and 24.7% have 20 or more elements. 5.5% of institutional templates require the relied-upon institution to indemnify the relying institution, 6.8% require the relying institution to indemnify the relied-upon institution, 8.2% require the relied-upon institution to have liability insurance, 15.1% require the relying institution to have liability insurance, 2.7% require the relied-upon institution to have accreditation, and none require the relying institution to have accreditation. The number of elements in the template was positively associated with total research funding and private institutional control. Our data indicate that institutions which are seeking reliance agreements for cooperative research involving human subjects are not likely to encounter many difficulties because most are using commonly employed templates and are not imposing potentially disputable requirements on parties to the agreement.

Introduction

On January 19, 2017, the Obama Administration published amendments to federal regulations known as the Common Rule which require collaborating institutions to designate a single institutional review board (IRB) to review and oversee research involving human subjects. Barring any changes by the Trump Administration, this section of the Common Rule becomes effective on January 20, 2020.¹ On June 21, 2016, the National Institutes of Health (NIH) announced a single-IRB policy for multisite research funded by the agency which is virtually identical to the Common Rule’s requirements. The NIH policy, which was developed in anticipation of the amendments to the Common Rule, becomes effective on January 25, 2018.² The Common Rule and the NIH policy both include some exceptions to the single-IRB requirement. The single-IRB mandate does not apply to international research collaborations or where single-IRB review conflicts with tribal, state, or local law. Agencies following the Common Rule can also make exceptions for compelling justifications.³

Federal agencies have promulgated the single-IRB mandate in response to complaints from investigators and institutions that multiple-IRB review is redundant, wasteful, expensive, and time-consuming.⁴ For example, a study of health services at Veterans Affairs (VA) medical centers required 115 submissions to 17 IRBs at a total cost of $170,000 in staff time. It took 17 months for all the IRBs at participating medical centers to approve the study.⁵ Another study of health services involving IRBs at 43 VA primary care clinics consumed 4,680 hours of staff time over the 19-month period needed to obtain final approval from all IRBs. Approval times ranged from 52 to 798 days.⁶ A multisite clinical trial submitted to 45 institutions incurred $107,544 in research review costs, including $82,610 in IRB fees and $24,934 in staff time.⁷

Although the single-IRB mandate has widespread support among investigators, funding agency leaders, and IRB members, some have raised concerns about the policy and its implementation.⁷ One of these is that understanding the local context, such as cultural traditions, community values, economic conditions, or language barriers, often plays an important role in the review of research.⁸ In a multisite study, an IRB unfamiliar with the local context may make decisions that fail to adequately protect human subjects. In response to this concern, some have proposed that IRBs take steps to ensure that they receive adequate information concerning local context, such as soliciting advice from investigators associated with local collaborating institutions, local community members or local advisory boards.⁹

Our paper will focus on a different concern with the single-IRB mandate, namely, the difficulties that collaborating institutions may face in negotiating reliance agreements.¹⁰ A reliance agreement (or authorization) is a legally binding contract between collaborating institutions, in which one party (the relying institution) relies on the other (the relied-upon institution) for review and oversight of research with human subjects. Reliance agreements may cover a single study, a group of studies, or an indefinite number of studies and may include multiple relying institutions.¹¹ The federal regulations allow institutions engaged in cooperative research to form reliance agreements.¹² Difficulties with negotiating reliance agreements may occur when institutions disagree about the terms of the agreement or its interpretation.¹³ Disagreements can also arise when institutions seek to minimize their legal or financial liability by including indemnification, insurance, or accreditation clauses in the agreement.¹⁴

The Office of Human Research Protections (OHRP), which oversees research funded by the National Institutes of Health (NIH), has taken steps to facilitate the negotiation of reliance agreements by developing a template that institutions can use. Under the terms described in the template, the relied-upon institution must abide by the human research regulations that apply to relying institutions and follow written procedures for reporting its actions and findings concerning the research to the relying institution. The relied-upon institution must also make IRB minutes available to the relying institution. Relying institutions forego their right to review the research and are responsible for ensuring that their research complies with federal regulation.¹⁵

Since 2016, over 150 institutions have agreed to participate in a collaborative research arrangement known as SMART IRB (Streamlined, Multisite, Accelerated Resources for Trials IRB). SMART IRB, which is funded by the NIH’s Clinical and Translational Science Awards Program, includes a reliance agreement template that members of the group can use for multisite research involving human subjects.¹⁶ The SMART IRB template is much longer than the OHRP template and addresses numerous issues not explicitly covered by the OHRP template, such as monitoring of research, quality assurance and quality improvement, insurance, indemnification, reporting of non-compliance and unanticipated problems, education of IRB members and research staff, recordkeeping, and audits.¹⁷

The objectives of our study were to: 1) analyze the content of reliance agreement templates used by institutions involved in cooperative research with human subjects; 2) determine whether institutions are seeking to minimize their liability by including indemnification, insurance, or accreditation clauses in the agreements; and 3) determine whether the number of elements covered by these templates is associated with total research funding or public vs. private status. Hopefully, the results of our study will provide researchers, institutions, funding agencies, and IRBs with information useful for anticipating the difficulties that may arise in negotiating reliance agreements. If there is evidence of significant variation in IRB templates used by institutions, then one would expect that difficulties in negotiating reliance agreements may arise. If many institutions are using similar templates, then one would expect that difficulties are likely to be minimal.

Methods

We attempted to contact human research protection officials (e.g., research compliance officers, IRB chairs or IRB managers) from the top 200 U.S. institutions ranked by total research funding for the most recent available year, using data provided by the Center for Measuring University Performance.¹⁸ We eliminated 9 institutions from the list because they do not oversee research with human subjects (i.e., College of Nanoscale Science and Engineering at University of Albany, State University of New York; Woods Hole Oceanographic Institution; Cold Spring Harbor Laboratory; and University of California System Admin Central Office) or because their human subjects oversight is covered by a larger (parent) institution (i.e., University of Medicine & Dentistry of New Jersey; Indiana University-Purdue University; Pennsylvania State University, Hershey Medical Center; University of Tennessee, Institute of Agriculture; and University of Maryland Center for Environmental Science). That left us with 191 institutions to contact. We emailed the human research protections officials during February and March 2017 and sent two reminder emails if we did not receive a response. In the email, we asked the officials to provide us a copy of their reliance agreement template. We also collected data on total research funding and public vs. private institutional control, which was provided by the Center for Measuring University Performance.

We developed a coding system for describing the content of the reliance agreements based on the OHRP template, the SMART IRB template, and an initial review of the documents we obtained. Sometimes the officials sent more than one template, in which case we combined coding for the multiple templates, eliminated redundancies, and noted the number of templates used by the institution. Two members of the research team independently coded the documents and resolved disagreements after inter-rater agreement was assessed statistically. The coding framework was a set of 38 yes or no questions related to contractual provisions (i.e. elements) or the type of reliance agreement (see Appendix). We also measured word counts for the templates. If an institution used more than one template, we took the average word count for the institution.

We used Kappa statistics to assess inter-rater agreement for the coding of the reliance agreement data. We used linear regression to assess the relationship between the total number of elements and total research funding. We used Cook’s distance to evaluate outliers. Due to the presence of outliers, we also used Spearman’s correlation to evaluate the relationship between the total number of elements and total funding. We the used Mann–Whitney test to compare the total number of elements in public vs. private institutions.

The National Institutes of Health’s Office for Human Subjects Research Protections determined that our study does not qualify as human subjects research because it does not involve the collection of private information about individuals.

Results

Of the 191 institutions contacted, 77 institutions (40.3%) responded to our request. 3 institutions (1.6%) declined to participate and 111 (58.1%) never answered our email. We did not code data from four institutions because they sent a document that was not a reliance agreement template (such as an executed reliance agreement or memorandum of understanding) or they do not enter into reliance agreements. Thus, we coded data for 73 institutions. The average annual research funding for coded institutions was $278.4 million (range $49.5 million to $1.1 billion). 54 institutions (74%) were public and 19 (26%) were private. See Figure 1 and Table 1.

Figure 1:

Number of Elements in Reliance Agreement Templates, Total Research Funding Dollars (x 1000), and Institutional Control for 73 Academic Institutions

Table 1:

Characteristics of 73 Coded Institutions

Average total research funding	$278.4 million (range = $49.5 to $1,112.5 million)
Public institutions	54 (74%)
Private institutions	19 (26%)

Inter-rater agreement for the coding of the reliance agreement data was high. Average agreement for coded items was 86.7% (range: 49-100%). Agreement was 49% for one item (scope of the agreement), 70-79% for 8 items, 80-89% for 11 items, and 90% or higher for 18 items. Complete inter-rater agreement data and statistics are available on request.

49 institutions (67.1%) use the OHRP template; 33 (45.2%) use their own (i.e., custom) template; and 6 (8.2%) use the SMART IRB template. 10 (13.7%) use the OHRP template and a custom template; 3 (4.1%) use the OHRP template and the SMART IRB template; 3 (4.1%) use a custom template and the SMART IRB template; and 1 (1.4%) uses templates from OHRP and SMART and a custom template. The average number of elements included in the templates was 15 (range: 10-28). 16 institutions (21.9%) use templates that have 10 elements (the OHRP minimum). Among these, 14 include all the OHRP elements and 2 include 9 of these elements. 20 institutions (27.4%) have 11 elements. Among these, all 20 include the OHRP minimum elements, and all 20 include scope of the agreement as the additional element. 19 institutions (26.0%) have 12-19 elements and 18 (24.7%) have 20 or more elements. Many institutions use more than one template and one uses 7 templates to cover different situations (such as agreements for when the institution is the relied-upon institution, for when it is relying upon another institution, for a single investigator, etc.). See Table 2. The average word count was 1027 (range: 219-6,584).

Table 2:

Characteristics of Reliance Agreement Templates (n =73)

Use the OHRP template	49 (67.1%)
Use a custom template	33 (45.2%)
Use the SMART IRB template	6 (8.2%)
Use the OHRP template and a custom template	10 (13.7%)
Use the OHRP template and the SMART IRB template	6 (8.2%)
Use the SMART IRB template and a custom template	3 (4.1%)
Use OHRP, SMART, and custom templates	1 (1.4%)
Average number of elements discussed in templates	15
10 elements discussed	16 (21.9%)
11 elements discussed	20 (27.4%)
12-19 elements discussed	19 (26.0%)
20 or more elements discussed	18 (24.7%)
Average word count in templates	1027 (range = 219 to 6584)

4 institutions (5.5%) require the relied-upon institution to indemnify the relying institution and 5 (6.8%) require the relying institution to indemnify the relied-upon institution. 6 institutions (8.2%) require the relied-upon institution to have liability insurance and 11 (15.1%) require the relying institution to have liability insurance. 2 institutions (2.7%) require the relied-upon institution to have accreditation, and no institutions require the relying institution to have accreditation. See Table 3.

Table 3:

Requirements to Reduce Legal Liability in Templates (n =73)

Relied-upon institution must indemnify relying institution	4 (5.5%)
Relying institution must indemnify relied-upon institution	5 (6.8%)
Relied-upon institution must carry liability insurance	6 (8.2%)
Relying institution must carry liability insurance	11 (15.1%)
Relied-upon institution must have accreditation	2 (2.7%)
Relying institution must have accreditation	0 (0.0%)

In evaluating the relationship between the number of elements in the reliance agreement templates and total research funding, two institutions were overly influential in the regression analysis (names withheld to protect confidentiality, Cook’s distance > 0.09). We therefore removed the two institutions before performing linear regression and observed a positive association (p-value: 0.039). The Spearman’s correlation between the total number of element and total funding using all 73 institutions was 0.267 (p-value: 0.023), agreeing with the result from linear regression.

Discussion

Our data indicate that institutions which are seeking reliance agreements for cooperative research involving human subjects are not likely to encounter many difficulties because 67.1% of institutions in our sample use the OHRP template and 8.2% use the SMART IRB template. The use of common templates, such as these, facilitates the negotiation of reliance agreements by minimizing potentially contentious issues. Although a substantial percentage of institutions (45.2%) use their own custom template, many of these also use the OHRP template or the SMART IRB template. Moreover, most institutions are not imposing potentially disputable requirements on parties to the reliance agreement. 5.5% require the relied-upon institution to indemnify the relying institution, 6.8% require the relying institution to indemnify the relied-upon institution, 8.2% require the relied-upon institution to have liability insurance, 15.1% require the relying institution to have liability insurance, 2.7% require the relied-upon institution to have accreditation, and none require the relying institution to have accreditation.

The number of elements in the reliance agreement template was positively associated with total research funding and private institutional control. It may be the case that institutions with higher levels of research funding are more concerned about minimizing liability than those with less funding, because they are more involved in risky human research collaborations, such as clinical trials, which can increase liability by including numerous research sites. Private institutions may be more concerned about liability than public ones, because they do not have government immunities from certain types of lawsuits.¹⁹ However, these explanations of the associations were observed are speculative and more research is needed to better understand how institutions are addressing liability concerns.

The response rate (40.3%) is a potential limitation of our study. However, we do not believe that it invalidates our results. First, some of the emails we sent may have been rejected as spam or junk mail, so our real response rate may be higher than 40.3%. Second, we compared respondents and non-respondents in terms of total research funding, funding rank, and public vs. private status and found no response bias for these variables. Statistical analysis for response bias is available on request.

Another potential limitation is that we sampled the top 200 academic institutions ranked by total research funding and did not seek data from institutions with less research funding. However, we do not think this is a significant limitation of our study because the top 200 institutions are more likely to be involved in cooperative research with human subjects than lower-ranked institutions. Thus, our sample probably represents most institutions which are dealing with issues related to negotiating reliance agreements.

We encourage others to conduct additional research on ethical, legal, and practical issues related to negotiating reliance agreements in cooperative research involving human subjects. Valuable data could be obtained by conducting interviews with officials who are involved in or impacted by these negotiations, such as institutional attorneys, IRB chairs, IRB management staff, or investigators, to learn more about the difficulties they encounter and how they navigate around them.

Appendix: Coding Questions for Reliance Agreement Templates

Type of Template

1. Does the institution use the OHRP template or a derivative with only minor changes (such as adding the institution’s name)?

2. Does the institution use SMART IRB template?

3. Does the institution use its own custom template?

Template Elements

OHRP Minimum Elements

• 4.Does the template include names of the institutions?
• 5.Does the template include names of the investigators?
• 6.Does the template include names of sponsors (if any)?
• 7.Does the template include names for projects or studies?
• 8.Does the template include listings for Federalwide Assurance (FWA) numbers?
• 9.Does the template include listing for IRB numbers?
• 10.Does the template include signature lines?
• 11.Does the template require that relied-upon institution will comply with requirements of the FWA (i.e., compliance with ethical principles and federal regulations)?
• 12.Does the template require that the relied-upon institution will follow written procedures for reporting its findings and actions to the relying institution?
• 13.Does the template require that the relying institution will comply with its FWA?
• 14.Does the template require the relied-upon institution to have accreditation for its human research protection program (or equivalent)?

Additional Elements

• 15.Does the template address standard operating procedures for review and oversight of research with human subjects?
• 16.Does the template address local context issues, such as local laws or other considerations?
• 17.Does the template address reporting non-compliance, unanticipated problems, adverse events, or injuries?
• 18.Does the template address communications with regulatory agencies?
• 19.Does the template address audits or investigations?
• 20.Does the template address monitoring, quality assurance, or quality improvement?
• 21.Does the template address conflicts of interest?
• 22.Does the template address confidentiality?
• 23.Does the template address compliance with the Privacy Rule of the Health Insurance Portability and Accountability Act (HIPAA)?
• 24.Does the template address education in human research protections?
• 25.Does the template address dispute resolution or mediation?
• 26.Does the template address governing law for the agreement?
• 27.Does the template address breach of the agreement?
• 28.Does the template address termination or length of the agreement?
• 29.Does the template address scope of the agreement?
• 30.Does the template address severability of parts of the agreement?
• 31.Does the template address indemnification?
• 32.Does the template address insurance for human research liability?

Special Requirements to Reduce Legal Liability

• 33.Does the template require the relied-upon institution to have accreditation?
• 34.Does the template require the relying institution to have accreditation?
• 35.Does the template require the relied-upon institution to indemnify the relying institution?
• 36.Does the template require the relying institution to indemnify the relied-upon institution?
• 37.Does the template require the relied-upon institution to carry liability insurance?
• 38.Does the template require the relying institution to carry liability insurance?