Minor Changes to Previously Approved Research: A Study of IRB Policies

The U.S. federal human research regulations allow institutional review boards (IRBs) to use an expedited review procedure to consider an investigator’s request to make minor changes (or amendments) to a protocol the IRB previously approved.¹ Under an expedited review procedure, the chair of the IRB or one or more experienced members of the IRB designated by the chair can approve minor changes (such as adding new investigators or sites, correcting grammatical or typographical errors in the protocol or consent form, or slightly increasing the amount of blood collected) without involving the full IRB at a convened meeting. The rationale for allowing minor changes to be approved by expedited review is to help reduce administrative burdens on the full IRB without undermining the protection of research participants. The expedited review procedure may also lead to quicker turnaround times for full IRB review of protocols.

Since federal research regulations do not currently define the term “minor change,”² the onus falls on research institutions to develop policies that balance the potentially conflicting goals of protecting research participants and minimizing burdens on the full IRB. Policies that classify too many types of changes as minor may fail to provide adequate protection for research participants because full IRB review of protocols often can provide additional safeguards beyond those afforded by review solely by the IRB chair. For example, the full IRB might not approve an amendment that the expedited reviewer would approve, such as a change that significantly increases the risks of a minimal-risk pediatric study in which participants receive no direct benefits. The full IRB might also require protections not mandated by the expedited reviewer, such as additional clinical monitoring or additions to the consent form. Policies that classify too few types of changes as minor may waste the IRB’s time by requiring the full IRB to review changes that can be handled by the chair. For example, there would be no need for the full IRB to review an amendment seeking to add urine collection to an approved study, since collecting urine poses no known risks to participants.

Consistency—both internal and external—is another important consideration for policy development. IRB policies should be internally consistent so that investigators at the institution will view the IRB’s decisions as fair. Policies should also be applied consistently. An investigator may view an IRB as arbitrary and unfair if she discovers that her protocol amendment was sent to the full IRB for review while a similar amendment was approved by expedited review. Evidence suggests that when scientists perceive that they are being treated unfairly by an organization, they are more likely to violate ethical norms or legal rules.³

External consistency is important so that research participants receive equal protection at different research institutions. If research institutions have different understandings of what constitutes a minor change, then amendments that receive full IRB review at some institutions may not receive the same review at others. As noted above, full IRB review can often provide additional protections beyond those afforded by review by the chair or other qualified IRB members. It follows, therefore, that interinstitutional differences in the level of IRB review could result in unequal protections for research participants.⁴

Studies of inconsistent interpretations of “minimal risk” in pediatric research provide further support for the idea that inconsistent interpretations of “minor changes” could lead to unequal protections research for participants. For example, Shah and colleagues found that 23% of IRB chairs classified allergy skin testing as minimal risk, while 43% percent regarded it as a minor increase over minimal risk, and 27% viewed it as more than a minor increase over minimal risk.⁵ If an investigator plans to add allergy skin testing to a pediatric protocol, this change could be approved as a minor amendment by the expedited reviewer, even though it might have been considered a major amendment (or even disapproved) by the full IRB, depending on how the terms “minor change” and “minimal risk” are interpreted. Thus, there are compelling reasons for encouraging IRBs to develop clear and consistent policies concerning minor changes to previously approved research. Federal agencies have taken some steps to encourage policy development in this area.

The Office for Human Research Protections (OHRP), which oversees IRBs that review research funded by the Department of Health and Human Services (DHHS), states that institutions should have written policies concerning the approval of minor changes to research.⁶ OHRP also advises that any change to the research that materially affects the balance of benefits and risks should not be considered minor.⁷ The Food and Drug Administration (FDA), which oversees research submitted in support of marketing approval for new medical products, advises that institutions should follow written procedures for expedited review and states that any change that increases the risk or discomfort of the study should not be considered minor.⁸

The Association for the Accreditation of Human Research Protection Programs (AAHRPP), the organization that accredits organizations with human research protection programs, has also taken steps to encourage institutions to adopt policies related to minor changes in approved research. Although AAHRPP standards do not define minor changes, they do require that organizations follow written policies and procedures for expedited review of changes to previously approved studies.⁹ Over 200 organizations, including universities, commercial IRBs, hospitals, and health care systems, have received full AAHRPP accreditation, and many more are currently seeking it.¹⁰

Do IRBs have policies concerning minor changes to approved protocols? What do those policies say? And are the policies consistent across IRBs? Our study addressed these questions. Our specific aims were to: 1) determine the proportion of IRBs at top U.S. research institutions with policies that define or provide examples of minor changes to previously approved research; 2) describe differences among definitions of minor changes; and 3) ascertain whether funding level, accreditation, public versus private status, and geographic region are associated with the presence of policies that include definitions or examples of minor changes. To our knowledge, there has been no previously published study of U.S. IRB policies concerning minor changes to approved research.

Materials and Methods

We attempted to access IRB policies from the 200 most highly funded research institutions in theUnited States and Puerto Rico (a U.S. territory). The rankings are based on funding information from the Center for Measuring University Performance (CMUP). The CMUP is a program associated with Arizona State University that collects and analyzes data on American research institutions.¹¹ We used the most recent CMUP data (from 2007) on research funding levels. We excluded four institutions (Desert Research Institute, University of California System Central Office, University of Maryland Biotechnology Institute, and Woods Hole Oceanographic Institution) that did not have their own IRB. We searched the public Web sites of the remaining 196 institutions for IRB policies. If an institution did not post IRB policies on its public Web site, we asked it for a copy of its policies (if any).

We examined IRB policies to determine whether they define a minor change or provide examples of a minor change. We considered a definition to be: 1) an explicit attempt to say what is meant by a minor change, e.g., “a minor change is”; 2) a paraphrasing of minor, e.g., “a minor change, a change that does not increase risk”; or 3) a contrasting of minor with its opposite, e.g., “a major change is.” We included definitions that were confusing or poorly worded, as long as it was clear that the institution was attempting to provide a definition. In examining the definitions, we considered whether any of the following features or elements were mentioned: 1) risks or discomforts; 2) research procedures, methods, aims, or design; 3) balance or assessment of benefits and risks; 4) research team; 5) facilities or collaborating institutes; 6) study population or expected enrollment; 7) informed consent, recruitment, or other factors that affect an individual’s decision to participate; and 8) other. This list of elements was based on preliminary data collection and an authoritative textbook on IRBs.¹² To ensure consistency in recording and categorizing the data, two of us independently coded the policies and resolved disagreements. We also gathered data on: 1) total amount of 2007 research funding; 2) AAHRPP accreditation; 3) public versus private status; and 4) geographic region as defined by the U.S. census bureau (Northeast, South, Midwest, or West).

We used statistical procedures (FREQ and LOGISTIC) from the SAS software package (version 9.2, SAS Institute Inc., Cary, NC, USA) to investigate how the proportion of IRBs that defined minor changes to previously approved research was affected by funding, accreditation (yes or no), status (public or private), and location (Northeast, South, Midwest, or West). We treated funding as a continuous variable, either in terms of dollars or national rankings. In separate (unadjusted) analyses of accreditation, status, and location, we calculated an exact 95% confidence interval for the response proportion at each level of these categorical predictors. We also used logistic regression to assess the effects of the predictors (categorical and continuous), both individually and simultaneously, on the response proportion. In addition to formulating the response variable as the proportion of IRBs that defined minor changes, we also examined the proportion that gave examples of minor changes and the proportion that gave either definitions or examples. All procedures were performed at the nominal 0.05 significance level.

Study Results

We obtained policies from 184 of the 196 institutions with an IRB (93.9% response rate); the listof those responding is available upon request. Of the 184 respondents, 96 defined minor changes (52.2%; confidence interval: 44.7%, 59.6%), 80 gave examples of minor changes (43.5%; confidence interval: 36.2%, 51.0%), and 125 provided definitions or examples (67.9%; confidence interval: 60.7%, 74.6%). The provided definitions focused on a lack of significant change in the following elements: risks or discomforts (82.3%); research methods, aims, or procedures (50.0%); risk/benefit ratio (34.4%); research team (18.8%); research facilities (18.8%); study population size or composition (18.8%); consent process or willingness to participate (10.4%); and other (20.8%). Nearly a third (33.2%) of the institutions were accredited by the AAHRPP, 74.5% were public, 25.5% were private, 37.0% were from the South, 22.8% were from the Northeast, 20.7% were from the West, and 19.6% were from the Midwest. The average level of research funding was $234 million (range: $36 million to $1.5 billion).

As seen in Table 1, the fact that an IRB defined or illustrated minor changes to previously approved research was positively associated with higher levels of research funding. The same qualitative results were obtained when examining funding ranks rather than absolute dollar amounts (Table 1). As indicated in Table 2, accredited institutional programs were more likely to include a definition or example of minor changes than were nonaccredited programs. However, the slightly higher percentage of private institutions that provided definitions or examples compared to public institutions was not statistically significant, nor were the minor differences between geographic regions (Table 2).

Table 1.

Summary of Total Research Funding According to Whether an IRB Provides Definitions or Examples of Minor Changes to Previously Approved Research

Response1	Level	N	Range	Median	Mean	95% C.I. for the Mean2
			Total Research Funding in Millions of Dollars
Definitions	Yes	96	36.4 – 842.8	267.4	314.1	(268.4, 359.8)
	No	88	36.4 – 1,554.1	94.0	147.0	(108.6, 185.5)
Examples	Yes	80	36.4 – 1,554.1	211.1	290.3	(232.5, 348.2)
	No	104	36.5 – 823.1	117.6	191.0	(156.7, 225.2)
Definitions or examples	Yes	125	36.4 – 1,554.1	216.0	287.9	(245.2, 330.7)
	No	59	36.5 – 531.8	83.4	120.3	(93.2, 147.4)
			National Rank with Respect to Total Research Funding
Definitions	Yes	96	2 – 200	63.5	76.0	(64.8, 87.2)
	No	88	1 – 199	136.0	128.3	(117.9, 138.6)
Examples	Yes	80	1 – 200	79.5	85.7	(73.0, 98.4)
	No	104	4 – 198	120.5	112.7	(101.7, 123.7)
Definitions or examples	Yes	125	1 – 200	78.0	84.2	(74.4, 94.1)
	No	59	22 – 198	144.0	136.5	(124.3, 148.7)

¹Three responses were analyzed with respect to minor changes to previously approved research according to whether an institution provided definitions, examples, or at least one of the two.

²The entries in parentheses are the lower and upper limits of a 95% confidence interval.

Table 2.

Percent of IRBs Providing Definitions or Examples of Minor Changes to Previously Approved Research by AAHRPP Accreditation Status and Selected Institutional Characteristics

			Percent (95% C.I.)1 of IRBs Providing:
Predictor	Level	N	Definitions	Examples	Definitions or examples
AAHRPP accreditation	Yes	61	80.3 (68.2, 89.4)	65.6 (52.3, 77.3)	98.4 (91.2, 100.0)
	No	123	38.2 (29.6, 47.4)	32.5 (24.4, 41.6)	52.9 (43.6, 61.9)
Institutional control	Public	137	49.6 (41.0, 58.3)	40.9 (32.6, 49.6)	64.2 (55.6, 72.2)
	Private	47	59.6 (44.3, 73.6)	51.1 (36.1, 65.9)	78.7 (64.3, 89.3)
Geographic location	Northeast	42	50.0 (34.2, 65.8)	47.6 (32.0, 63.6)	66.7 (50.5, 80.4)
	South	68	50.0 (37.6, 62.4)	36.8 (25.4, 49.3)	69.1 (56.7, 79.8)
	Midwest	36	52.8 (35.5, 69.6)	55.6 (38.1, 72.1)	66.7 (49.0, 81.4)
	West	38	57.9 (40.8, 73.7)	39.5 (24.0, 56.6)	68.4 (51.4, 82.5)

¹The entries in parentheses are the lower and upper limits of an exact 95% confidence interval.

In an alternative approach, we used logistic regression to model the response proportion as a function of each (continuous or categorical) predictor individually. We also used logistic regression to model response as a function of multiple predictors simultaneously, which allowed us to investigate which ones remained influential when considered together. When the predictors were examined individually, the results were similar to those summarized in Tables 1 and 2. Higher funding—whether represented in dollar amounts or as national rankings—and accreditation were both associated with a higher proportion of IRBs providing a definition, an example, or either a definition or an example of minor changes. As before, there was no evidence that any of the response proportions varied significantly with either public versus private status or geographic location. The same qualitative results were obtained when multiple predictor variables were considered simultaneously: funding and accreditation appeared important, but public versus private status and geographic location did not.

Discussion

While our study indicates that most IRBs in our sample have a policy that defines or provides examples of minor changes to previously approved research, there is considerable variation in the definitions of minor change that have been adopted by different institutions. Although most of the polices we examined mentioned no significant change in risk as a defining feature of a minor change, a high percentage also mentioned other elements, including no significant changes in the research, the risk/benefit ratio, the research team, and the study population. This observed variation in the definition of minor changes may be a cause for concern. For instance, an amendment that is approved by expedited review at one IRB may be sent to the full board for review at another institution, which could raise issues of unequal protection for research participants and inconsistent handling of research proposals across institutions.

Federal agencies could promote consistent interpretations of the research regulations by providing additional guidance for the review of minor changes to previously approved research. Policies related to minor changes should include clear definitions or detailed examples that provide consistent direction for IRB chairs, IRB staff, and investigators. Such policies should balance the potentially conflicting goals of protecting research participants and minimizing burden on the full IRB. Policies should also address six key areas identified in our study: risk/benefit, research methods, research staff, research facilities, study population, and consent/willingness to participate.

Our study also indicates that AAHRPP accreditation and level of research funding have a significant impact on the development of definitions and examples of minor changes. Public versus private status and geographic location had no significant effect. This result is expected, since AAHRPP requires that institutions have written policies dealing with expedited review in order to receive accreditation. It is likely that the proportion of IRBs with definitions or examples of minor changes will increase as more institutions receive AAHRPP accreditation, which suggests that accreditation has a positive impact on IRB policy development. One would expect that institutions with higher levels of research funding would be more motivated to develop IRB policies than institutions with lower levels of funding, since they face greater pressure from regulatory agencies and sponsors to ensure compliance with regulations and ethical guidelines.

It also appears that FDA guidance regarding minor changes has had an important influence on definitions adopted by university IRBs, since 82.3% of those with a definition included no significant increase in risk as a defining feature of a minor change, and the FDA has stated that any change that increases risks or discomforts should not be considered minor. OHRP also may have influenced how minor change is defined, since that agency advises that any change that materially affects the risk/benefit ratio should not be considered minor, and 34.4% of the IRB policies in our sample included “no significant change in risks/benefits” as a defining feature of a minor change.

One limitation of our study is that most of the policies we examined were obtained from publicly available Web sites rather than from requests for them from the IRB. It is possible that some IRBs had policies that defined or provided examples of minor changes in research that they did not put on their Web sites. While we recognize that this is a potential limitation of our study, we do not think that it had a significant impact on our findings. The institutions that placed their IRB policies on Web sites listed many different policies, ranging from the criteria for IRB approval to adverse event reporting, and it makes little sense to put all of these policies on a Web site and not include policies dealing with minor changes, especially since one of the main purposes of publishing a policy on a Web site is to provide guidance for investigators, IRB members, and IRB staff.

Another limitation of our study is that we focused on the top U.S. institutions in terms of research funding and did not collect data on other types of institutions with IRBs, such as universities or colleges with less funding or hospitals. It is possible that trends we observed in our sample would be found in other samples—i.e., that the development of policies or definitions of minor changes is positively associated with AAHRPP accreditation and higher funding—but this is only speculation on our part. Another study of IRB policies with a broader sample could determine if this is the case.