Materials and Methods

We reviewed the characteristics and outcomes of the review of 5 common protocols submitted to individual IRBs at NPRC centers over a 5 year period from 2007 to 2012. The 5 NPRC centers included Columbia University, University of Alabama at Birmingham, University of North Carolina at Chapel Hill (UNC), University of Texas Medical Branch, Galveston, and University of Utah/Intermountain Healthcare (Utah).

The 5 studies reviewed were projects to be implemented as part of a larger National Institutes of Health–funded perinatal multicenter network in which all NPRC centers (and affiliated sites) participate. Each study protocol was developed by a subcommittee comprised of clinician investigators, research coordinators, and biostatisticians. The study protocol, including a sample consent form, was approved by a steering committee prior to submission to the IRBs at each center.

This review was exempt from IRB review at all participating centers. For each selected study, complete IRB logs and communications at each site (including those for the separate IRBs at subsites) were reviewed. Data abstraction was performed by trained research team members at each of the sites using a pretested data collection form. The data form covered 13 numbered items with several items having multiple subitems. The items covered study design, IRB submission format and review type, submission to final decision time and its subcomponent (submission-to-review, review-to-initial IRB letter and IRB letter-to-final approval) times, and IRB review results including stipulations, risk classification, and age requirements to give consent.

Descriptive measures including proportions and medians (range) were used to summarize the data. The Kruskal Wallis test was used to compare numerical data across site IRBs and across protocols. We used generalized linear regression to explore the relationship between IRB processing times and variables including the number of stipulations and number of IRB meetings per month. SAS software (SAS Institute, Cary, NC) was used to compute descriptive statistics and both SAS and STATA (StataCorp, College Station, TX) were used for linear regression.

Results

The 5 study protocols we examined (designated A-E) and selected characteristics of each study including the goal, study design, and sample size are summarized in Table 1 . There was 1 observational study (A) that involved a request for a waiver of consent from the IRB to conduct chart abstraction after delivery (without any patient interaction). Three studies (B-D) were primary randomized clinical trials (RCTs) (the cytomegalovirus trial, D, involved a preliminary screening phase to identify eligible women with primary maternal cytomegalovirus), and 1 study (E) involved secondary follow-up of women and their offspring (5-10 years old) who had previously participated in a mild gestational diabetes mellitus (GDM) treatment trial.

Data for 4 protocols (A, B, D, and E) were derived from 7 IRBs: the 5 main center IRBs and 2 subsite IRBs (Intermountain Healthcare System affiliated with Utah and Wake Forest University, affiliated with UNC). Data concerning study C came from 8 IRBs (including St Mark’s Hospital, another Utah affiliate). Thus, cumulatively there were 36 individual IRB study reviews.

The frequency of IRB review meetings for the 8 individual IRBs was a median of 2 per month (range, 1–6 per month). The 3 primary RCT protocols all required a full board review at each site, but only 50% of IRBs required a full board review for the other 2 (1 observational and 1 secondary RCT follow-up) protocols. During the 5 year span, IRB submission changed from a predominantly paper format (at 4 of 7 IRBs) for the earliest study (A) to predominantly electronic submission (6 of 7 IRBs) for the most recent protocols (D and E).

All study protocols were ultimately approved by each site IRB without any protocol changes. The key IRB processing times and number of stipulations per IRB for each protocol are presented in Table 2 . Processing times and number of stipulations from the IRB reviews vary both across studies and across IRBs. Specifically, the total duration from submission to approval times at individual IRBs varied from less than a week to 2 months or more for individual studies; median IRB approval times differed significantly across studies ( P = .02). The subcomponent times also varied across studies and IRBs, but the differences did not attain statistical significance.

Finally, the number of stipulations differed significantly across individual studies ( P < .01). Close examination of the ranges of processing times and number of stipulations observed for the IRBs also suggest considerable variation across IRBs (see the following text).

The processing times and number of stipulations for each IRB site, regardless of study design, are summarized in Table 3 . Median times from protocol submission to final IRB approval (ranging from 8 to 24 days) did not differ across individual IRB sites ( P = .63). However, the subcomponent times from submission-to-IRB review ( P = .01) and IRB review-to-notification letter ( P < .01) but not review to approval ( P = .19) differed across IRBs.

The median number of stipulations ranged widely from 1 to 23 across site IRBs, although this did not attain statistical significance ( P = .07). The site with the highest median overall time to approval also had the highest median number of review stipulations (n = 23). The subsite with the lowest median time to approval (8 days) reviewed only one protocol and also matched the lowest median number of stipulations (ie, 1). On linear regression both overall submission-to-approval ( P < .001) and initial review-to-approval ( P < .001) times increased with increasing number of IRB stipulations on initial review. On the contrary, submission-to-approval ( P = .216) and submission-to-review ( P = .189) times were not associated with the number of monthly IRB meetings.

Of the 36 individual IRB study reviews, sites reported at least 1 stipulation, request, or question from the IRB for 24 of those reviews (66.7%). Among these 24 instances, 15 (62.5%) were requests to change format or wording on the consent form (most instances) or to include additional information (a full list of exclusion criteria and, frequently, to further address risks of study interventions and to provide more details about standard clinical practice vs research procedures). The remainder were administrative requests including obtaining Health Insurance Portability and Accountability Act privacy board waiver, to provide copy of data coordinating center IRB approval, describe logistics for counseling patients who test positive for CMV, and to show evidence of study-related training.

There was no direct request for a protocol change. Examples of format or wording changes to the consent form included to add spaces, delete multiple signature lines, and word changes from “you and the child” to “you and your child,” “research assent form” to “child assent form,” and “compensate” to “reimburse.”

There were 5 instances of significant delays in the IRB review process (defined as submission-to-approval time of ≥60 days) occurring at 5 different individual sites. Three study protocols (A, D, and E) were affected, and 3 of the instances involved the same protocol (D), but the reasons for the delays were all different. Delays were attributed to the following: (1) Food and Drug Administration investigational new drug requirements (specific to protocol D), (2) delays in submitting phone follow-up scripts for reviews (protocol E), (3) IRB concerns about the handling of screening results (protocol D), and (4) a site-specific request to the IRB for a more streamlined consent process (protocol D).

The fifth instance of delay (protocol A) was not directly related to the IRB: the institutional sign-off was delayed by concerns relating to the handling of medical records and Health Insurance Portability and Accountability Act. These delays were not consistent across individual study protocols or specific sites. All protocols required revisions except the observational protocol (A), which received full approval at all sites with the initial IRB review.

Most IRB stipulations or questions involved only minor revisions or clarifications regarding consent documents or patient confidentiality. The minimum enrollment age (none to 18 years) and the age requirements for parental consent (none up to 18 years) varied by IRB site. Overall, the study protocols, regardless of study design, warranted only 1 full IRB review session.