Often the most difficult task in a
clinical trial involves obtaining sufficient study participants
within a reasonable time. Time is a critical factor for both
scientific and logistical reasons. From a scientific viewpoint,
there is an optimal window of time within which a clinical trial
can and should be completed. Changes in medical practice, including
introduction of new treatment options, may make the trial outdated
before it is completed. Other investigators may answer the
questions sooner. In terms of logistics, the longer recruitment
extends beyond the initially allotted recruitment periods, the
greater the pressure becomes to meet the goal. Lagging recruitment
will also reduce the statistical power of the trial. Selective
recruitment of a lower proportion of eligible participants may
increase the non-representative nature of the sample. Costs
increase, frustration and discouragement often follow. The primary
reasons for recruitment failure include overly optimistic
expectations, failure to start on time, inadequate planning, and
insufficient effort.
Approaches to recruitment of
participants will vary depending on the type and size of the trial,
the length of time available, the setting (hospital, physician’s
office, community), whether the trial is single- or multicenter,
and many other factors. Because of the broad spectrum of
possibilities, this chapter summarizes concepts and general methods
rather than elaborating on specific techniques. Emphasis is placed
on anticipating and preventing problems. This chapter addresses
plans for the recruitment effort, common recruitment problems,
major recruitment strategies and sources, use of electronic health
records for screening, and monitoring of recruitment.
Fundamental Point
Successful recruitment depends on developing a
careful plan with multiple strategies, maintaining flexibility,
establishing interim goals, preparing to devote the necessary
effort, and obtaining the sample size in a timely
fashion.
Considerations Before Participant Enrollment
Selection of Study Sample
In Chap. 4, we define the study population as
“the subset of the population with the condition or characteristics
of interest defined by the eligibility criteria.” The group of
participants actually recruited into the trial, i.e., the study
sample, is a selection from this study population. Those enrolled
into a trial do not represent a random sample of those eligible for
enrollment. Eligible individuals who volunteer to participate in a
randomized trial may be different from eligible non-participants
(see below). The impact of this potential selection bias on the
results of a trial is not well understood. A better understanding
of the factors that influence either willingness or unwillingness
to participate in a research project can be very helpful in the
planning of recruitment efforts.
The public is generally willing to
participate in clinical trials [1].
A survey of around 1,000 Americans conducted in May 2013 showed
that 64% would “take part in a clinical trial if I was asked by
someone I trust.” Yet only about 15% of Americans report that they
have participated in a clinical trial. In this survey, lack of
trust was the major barrier to participation in trials. When asked
to rate factors involved with decision to volunteer, nearly 70%
reported reputation of the people or institution conducting the
research and whether medical bills resulting from injury from the
study would be covered as very important factors. Opportunity to
possibly improve one’s own health was noted as very important for
61%, privacy and confidentiality issues for 53%, and opportunity to
improve health of others for 50%. Complementary information was
reported in a review of 14 studies through 2001 that had addressed
the question—What reasons do people give for participating and not
participating in clinical trials [2]? The answers came from 2,189 participants and
6,498 who declined. The variability was large, but trial
participants gave as their major reason for participating potential
health benefit (45%), physician influence (27%), and potential
benefit to others (18%). Less commonly mentioned reasons given by
participants in other studies included a desire to learn more about
their condition, get free and better care, encouragement by family
members and friends, favorable impression of and trust in clinical
staff, and even to help promote the investigators’ careers
[3–6].
Several reasons for declining
participation in research projects have also been reported. In the
Emergency Care Research Institute (ECRI) survey [2], the major general reasons for not
participating were inconvenience (25%), concern over
experimentation (20%), potential lack of health benefit (19%), and
physician influence (14%). Many patients also lacked interest and
preferred to stay with their own physicians. In another survey,
fear was given as a major reason by half of those declining
participation and the use of a placebo by almost one-quarter
[6].
Logistical issues are sometimes
given—demands on time, conflicts with other commitments, and
problems with travel/transportation and parking. Barriers to
participation in cancer trials include concerns with the trial
setting, a dislike of randomization, presence of a placebo or
no-treatment group, and potential adverse effects [7].
Common Recruitment Problems
The published experience from
recruitment of participants into clinical trials through 1995 is
nicely summarized in a literature review and annotated bibliography
[8]. Over 4,000 titles were
identified and 91 articles considered useful for formulation of
recruitment strategies in clinical trials are annotated. The review
focuses on experiences recruiting diverse populations such as
ethnic minorities, women, and the elderly. Also discussed are
successful recruitment approaches, which include use of registries,
occupational sites, direct mailing, and use of media. The article
highlights the value of pilot studies, projecting and monitoring
recruitment, and the use of data tracking systems. Many of these
issues are covered in more detail later in this chapter.
A review from the United Kingdom of
114 clinical trials that recruited participants between 1994 and
2002 explored the factors related to good and poor recruitment
[9]. Approximately one-third of all
trials met their original recruitment goal within the proposed time
frame while approximately half had to be extended. Among those
failing to make the original target, one half revised the goals.
About 40% of all trials did not initiate recruitment as planned,
mostly due to staffing and logistical issues. Almost two-thirds of
the trials acknowledged early recruitment problems. More than half
of the reviewed trials, a remarkably high number, had a formal
pilot study that led to changes in the recruitment approach for the
main trial. The written trial materials were revised, the trial
design altered, the recruitment target changed, the number of sites
increased, and/or the inclusion criteria broadened.
A systematic review of recruitment
methods identified 37 trials describing four broad categories of
recruitment strategies: novel trial designs (including different
consent strategies), recruiter interventions (including training),
incentives, and provision of trial information to potential
participants [10]. Strategies that
increased recruitment rates included increasing awareness of the
health problem being studied, educational sessions, health
questionnaires, and monetary incentives. A study using
semistructured investigator interviews in the Palliative Care
Research Cooperative Group identified five effective recruitment
strategies: systematic screening of patient lists or records,
messaging to patients to make research relevant, flexible protocols
to account for patient needs, clinical champion support, and
involvement in the cooperative group [11]. There is little published information on
cost-effectiveness of interventions to improve recruitment,
although one review of ten studies found that directly contacting
potential participants seemed to be an efficient strategy
[12].
Electronic health records, combined
with programming to systematically screen for eligible
participants, provide an important tool for certain trials. A
single-center study showed that when an automated electronic health
record alert system was put into place in clinics to identify
patients for a trial of type II diabetes, referral rates increased
by tenfold and enrollment rate by twofold [13]. An electronic screening tool was effective,
and it performed particularly well in excluding ineligible patients
at Columbia University for the National Institutes of Health
(NIH)-sponsored Action to Control Cardiovascular Risk in Diabetes
(ACCORD) Trial [14]. A hybrid
approach of electronic screening coordinated with paper screening
forms was successful in the Veterans Affairs VA-STRIDE trial to
recruit patients for a physical activity intervention
[15].
A review of the challenges and
opportunities of use of electronic health records to support
clinical trial enrollment identifies regulatory issues such as use
of screening information preparatory to research and barriers
related to desire to approach a patient’s treating doctor before
approaching the patient directly [16]. An overarching goal is to use electronic
systems to better integrate clinical trials into clinical practice.
An example of success is the Thrombus Aspiration in ST-Elevation
myocardial infarction in Scandinavia (TASTE) trial, where all
interventional hospitals (and cardiologists) in Sweden agreed to
approach all eligible patients for enrollment using the national
registry platform for randomization and data collection
[17]. During the recruitment
period (see Fig. 5.1), 59.7% of all the patients
presenting with ST segment elevation myocardial infarction and
referred for percutaneous coronary intervention and 76.9% of all
the patients potentially eligible for enrollment in Sweden and
Iceland were included in the trial, at a low per-patient cost. Not
only did this remarkably high participant capture allow the trial
enrollment to be completed in 2 years and 9 months, but it enhanced
generalizability of the results.
But it is more typical that even when
carefully planned and perfectly executed, recruitment may still
proceed slowly. Investigators should always expect problems to
occur, despite their best efforts. Most of the problems are
predictable but a few may be completely unforeseen. In one
multicenter study, there were reports of murders of inpatients at
the hospital adjacent to the study clinic. It is hardly surprising
that attendance at the clinic fell sharply.
Overestimation of eligible
participants is a common reason for recruitment difficulties. A
group of Finnish investigators [18] conducted a retrospective chart review. The
typical eligibility criteria for clinical trials of patients with
gastric ulcer were applied to 400 patients hospitalized with that
diagnosis. Only 29% met the eligibility criteria but almost all
deaths and serious complications such as gastric bleeding,
perforation, and stenosis during the first 5–7 years occurred among
those who would have been ineligible. Clearly, the testing of
H2–blockers or other compounds for the prevention of
long-term complication of gastric ulcer in low-risk participants
should not be generalized to the entire ulcer population. Troubling
in this report is the evidence that the eligibility criteria can
have such a dramatic effect on the event rates in those qualifying
for participation.
Reliance on physician referrals is
common and often problematic. Usually this technique results in
very few eligible participants. In 2005, a survey of 7,000
physicians reported that only 31% of them had ever referred a
patient to a clinical trial [6]. In
one multicenter trial an investigator invited internists and
cardiologists from a large metropolitan area to a meeting. He
described the study, its importance, and his need to recruit men
who had had a myocardial infarction. Each of the physicians stood
up and promised to contribute one or more participants. One hundred
fifty participants were pledged; only five were ultimately
referred. Despite this, such pleas may be worthwhile because they
make the professional community aware of a study and its purpose.
Investigators who stay in close contact with physicians in a
community and form a referral network have more success in
obtaining cooperation and support.
When recruitment becomes difficult,
one possible outcome is that an investigator will begin to loosely
interpret entry criteria or will deliberately change data to enroll
otherwise ineligible participants or even “enroll” fictitious
subjects. Unfortunately, this issue is not merely theoretical. Such
practices have occurred, to a limited extent, in more than one
trial [19–21]. The best way to avoid the problem is to
make it clear that this type of infraction harms both the study and
the participants, and that neither science nor the investigators
are served well by such practices. An announced program of random
record audits by an independent person or group during the trial
may serve as a deterrent.
Planning
In the planning stage of a trial, an
investigator needs to evaluate the likelihood of obtaining
sufficient study participants within the allotted time. This
planning effort entails obtaining realistic estimates of the number
of available potential participants meeting the study entry
criteria. However, in the United States, access to available
patient data from paper and electronic medical records requires
compliance with the Health Insurance Portability and Accountability
Act (HIPAA) and similar regulations apply in many other countries.
Access can be granted but many community practices do not have such
a mechanism in place and tend to be reluctant to release patient
information. Even if those restrictions are overcome, census tract
data or hospital and physician records may be out of date,
incomplete, or incorrect. Information about current use of drugs or
frequency of surgical procedures may not reflect what will occur in
the future, when the trial is actually conducted. Records may not
give sufficient—or even accurate—details about potential
participants to determine the impact of all exclusion criteria.
Clearly, available data certainly do not reflect the willingness of
people to enroll in the trial or comply with the
intervention.
After initial record review, an
investigator may find it necessary to expand the population base by
increasing the geographical catchment area, canvassing additional
hospitals, relaxing one or more of the study entrance criteria,
increasing the planned recruitment time, or any combination of
these. The preliminary survey of participant sources should be as
thorough as possible, since these determinations are better made
before, rather than after, a study begins.
Investigator and study coordinator
commitment is key to success. Since lack of trust is a major
barrier to patient agreement to participate, a strategy for
effective communication from the treating physician to the patient
about the relevance and importance of the study is critical. A
concern is that investigators keep adding new trials to those they
already have committed to. Trials with higher payments seem to get
more attention. The investigator also needs strong support from his
institution and colleagues. Other investigators in the same
institution or at nearby institutions may compete for similar
participants. Since participants should generally not be in more
than one trial at a time, competing studies may decrease the
likelihood that the investigator will meet his recruitment goal.
Competition for participants may necessitate reappraising the
feasibility of conducting the study at a particular site.
Announcements of the trial should
precede initiation of recruitment. The courtesy of informing area
health professionals about the trial in advance can facilitate
cooperation, reduce opposition, and avoid local physicians’
surprise at first hearing about the study from their patients
rather than from the investigator. Talks to local professional
groups are critical, but these and any notices regarding a trial
should indicate whether the investigator is simply notifying
physicians about the study or is actively seeking their assistance
in recruiting participants.
Planning also involves setting up a
clinic structure for recruitment with interested and involved
co-investigators, an experienced and organized coordinator in
charge of recruitment, and other staff required for and dedicated
to the operations. A close working relationship between the clinic
staff and investigators, with regular clinic meetings, is crucial
from the very beginning to enrollment of the last participant.
Careful planning and clear delineation of staff responsibilities
are essential features of well-performing recruitment units.
Recruitment in most trials is
curvilinear, particularly in multicenter trials, with a gradual
acceleration of enrollment as centers start up and refine
successful strategies for enrollment. But the calculation of a
sample size estimate typically assumes a constant rate of
enrollment. A slow start can reduce the statistical power of the
trial by reducing the average participant follow-up time. Thus,
ideally, recruitment should begin no later than the first day of
the designated recruitment period. As important as the best
planning is, commitment and willingness by everyone to spend a
considerable amount of time in the recruitment effort are equally
important. Just as investigators usually overestimate the number of
participants available, they often underestimate the time and
effort needed to recruit. Investigators must accommodate themselves
to the schedules of potential participants, many of whom work.
Thus, recruitment is often done on weekends and evenings, as well
as during usual working hours.
The need for multiple recruitment
strategies has been well documented [22, 23]. The
first randomization should take place on the first day of the
identified recruitment period. Therefore, if there is a lengthy
prerandomization screening period, adjustments in the timing of the
first randomization should be made. Because it is difficult to know
which strategies will be productive, it is important to monitor
effort and yield of the various strategies. A successful strategy
in one setting does not guarantee success in another. The value of
multiple approaches is illustrated by one large study in which the
investigator identified possible participants and wrote letters to
them, inviting them to participate. He received a poor response
until his study was featured on local radio and television news.
The media coverage had apparently “legitimized” the study, as well
as primed the community for acceptance of the trial.
Contingency plans must be available in
case recruitment lags. Experience has shown that recruitment
yields, in general, are much lower than anticipated. Thus, the
identified sources needed to be much larger than the recruitment
goals. Hence, additional sources of potential study participants
should be kept in reserve. Approval from hospital staff, large
group practices, managed care organizations, corporation directors,
or others controlling large numbers of potential participants often
takes considerable time. Waiting until recruitment problems appear
before initiating such approval can lead to weeks or months of
inaction and delay. Therefore, it is advisable to make plans to use
other sources before the study gets underway. If they are not
needed, little is lost except for additional time used in planning.
Most of the time these reserves will prove useful.
If data concerning recruitment of
potential participants to a particular type of trial are scanty, a
pilot or feasibility study may be worthwhile. Pilot studies can
provide valuable information on optimal participant sources,
recruitment techniques, and estimates of yield. In a trial of
elderly people, the question arose whether those in their 70s or
80s would volunteer and actively participate in a long-term,
placebo-controlled trial. Before implementing a costly full-scale
trial, a pilot study was conducted to answer these and other
questions [24]. The study not only
showed that the elderly were willing participants, but also
provided information on recruitment techniques. The success of the
pilot led to a full-scale trial.
Recruitment Sources
The sources for recruitment depend on
the features of the study population; sick people versus well,
hospitalized versus not, or acute versus chronic illness. For
example, enrollment of acutely ill hospitalized patients can only
be conducted in an acute care setting, while enrollment of healthy
asymptomatic individuals with certain characteristics or risk
factors requires a community-based screening program. Following the
introduction of the HIPAA and other privacy regulations, readily
available sources for recruitment have changed. Identification of
potential participants through review of electronic health records
requires active involvement of those patients’ own physicians.
Thus, focus has shifted to direct participant appeal.
Direct invitation to study
participants is an attractive approach, since it avoids many
confidentiality issues. Solicitation may be done through mass
media, wide dissemination of leaflets advertising the trial, or
participation by the investigator in health fairs. None of these
methods is foolproof. The yield is often unpredictable and seems to
depend predominantly on the skill with which the approach is made
and the size and kind of audience it reaches. One success story
featured a distinguished investigator in a large city who managed
to appear on a local television station’s evening news show.
Following this single 5-min appeal, thousands of people volunteered
for the screening program. Experience, however, has shown that most
individuals who respond to a media campaign are not eligible for
the trial.
The recruitment into the Systolic
Hypertension in the Elderly Program (SHEP) was a major undertaking
[25]. A total of almost 450,000
potential participants were contacted in order to enroll 4,736
(1.1%). One of the major recruitment approaches in SHEP was mass
mailings. A total of 3.4 million letters were sent by 14 of the
SHEP clinics and the overall response rate was 4.3%. Names were
obtained from the Departments of Motor Vehicles, voter registration
lists, health maintenance organizations, health insurance
companies, AARP, and others. Endorsement was obtained from these
organizations and groups; many of them issued the invitations on
their own letterheads. Each mailing included a letter of
invitation, a standard brochure describing SHEP, and a
self-addressed stamped return postcard. Experience showed that the
response rates varied by mailing list source. It was also clear
that clinics with experienced recruitment staff did better than the
others.
A U.S. survey of 620 previous trial
participants asked where they first learned about the trials
[6]. Media the most common answer,
was given by 30%, but 26% said the internet. Web-based strategies
seem to be growing in importance, although the yield appears to
vary by type of trial. Only 14% in the survey first learned of the
trial via physician referral.
Participants may also be approached
through a third party. For instance, local chapters of patient
organizations may be willing to refer members. Another approach is
through physician referrals. To draw physicians’ attention to a
particular study, an investigator may send letters, make telephone
calls, present at professional society meetings, publish notices in
professional journals, or exhibit at scientific conferences. The
hope is that these physicians will identify potential participants
and either notify the investigator or have the potential
participant contact the investigator. As noted earlier, this
usually yields few participants. To overcome the problem with
physician referral, sponsors are offering financial incentives to
referring physicians. The value of this practice has not been
properly evaluated, but it has raised ethical issues concerning
conflict of interest, disclosure to potential participants, and
implications for the informed consent process [26].
The recruitment targets have to be
adjusted if special subgroups of the population are being
recruited. In response to a relative paucity of clinical trial data
on women and minorities, the U.S. Congress in 1995 directed the NIH
to establish guidelines for inclusion of these groups in clinical
research. The charge to the Director of the NIH to “ensure that the
trial is designed and carried out in a manner sufficient to provide
valid analysis of whether the variables being studied in the trial
affect women and members of minority groups, as the case may be,
differently than other subjects in the trial” has major
implications depending on the interpretation of the term “valid
analysis” [27].
To document a similar effect,
beneficial or harmful, separately for both men and women and
separately for various racial/ethnic groups could increase the
sample size by a factor ranging from 4 to 16. The sample size will
grow considerably more if the investigator seeks to detect
differences in response among the subgroups. We support adequate
representation of women and minorities in clinical trials, but
suggest that the primary scientific question being posed be the
main determinant of the composition of the study population and the
sample size. When the effort is made, successful enrollment of
women and minorities can be accomplished. An example is the
Selenium and Vitamin E Cancer Prevention Trial [28].
An increasingly common approach to
meeting the need for large sample sizes in multicenter trials with
mortality and major event response variables has been to establish
clinical centers internationally [29]. This experience has been positive and the
number of participants enrolled by such centers often exceeds those
in the country of the study’s origin. The success in recruitment
may, however, come at a cost. Trial findings may differ among
countries (see also Chap. 21). Possible reasons include
differences in the baseline characteristics of the study
population, in the practice of medicine as a reflection of the
quality of care, research traditions, and socioeconomic and other
factors [30, 31]. O’Shea and Califf analyzed the
international differences in cardiovascular trials and reported
important differences in participant characteristics, concurrent
therapies, coronary revascularizations, length of hospital stay,
and clinical outcomes in the U.S. and elsewhere [32]. Importantly, they pointed out that, in
general, the differing event rates would not be expected to affect
the relative effects of a treatment. But there are examples of
possible differences in treatment effects according to enrollment
within versus outside the U.S., including less benefit of beta
blockers in heart failure [33] and
of ticagrelor following acute coronary syndromes [34]. The authors of a review of 657 abstracts
from trials of acupuncture and other interventions concluded that
some countries published unusually high proportions of positive
results [35]. Possible
explanations include publication biases, level of care, and
differences in study populations.
Can findings from low and middle
income countries be extrapolated to high income countries and
regions and vice versa? It is important that the publications from
large international studies address this question by presenting
findings by geographic region.
Conduct
Successful recruitment of participants
depends not only on proper planning but also on the successful
implementation of the plan. Systems should be in place to identify
all potential participants from the identified recruitment pool and
to screen them for eligibility. For hospital-based studies, logging
all admissions to special units, wards, or clinics is invaluable.
However, keeping such logs complete can be difficult, especially
during evenings or weekends. During such hours, those most
dedicated to the study are often not available to ensure accuracy
and completeness. Vacation times and illness may also present
difficulties in keeping the log up to date. Therefore, frequent
quality checks should be made. Participant privacy is also
important and is guided by ethics committees, and in the U.S.,
HIPAA regulations. At what point do the investigators obtain
consent? For those who refuse to participate, what happens to the
data that had been collected and used to identify them? The answers
to this will vary from institution to institution and depend on who
is keeping the log and for what reason. Information recorded by
code numbers can facilitate privacy. Electronic health records can
be used. Electronic medical records permit software algorithms to
search for patient profiles that match a particular protocol and
automatically identify for the health care team those eligible for
a specific trial. A group at Cincinnati Children’s Hospital
reported a retrospective study estimating that the electronic
health records in the emergency department could be searched with
natural language processing, information extraction, and machine
learning techniques to reduce screening workload by over 90% for 13
randomly selected, disease-specific trials [36]. However, the broad experience of electronic
systems for clinical trial recruitment shows inconsistent evidence
of value. The integration of the systems with human workflow may be
more important than sophisticated algorithms [37].
For community-based studies, screening
large numbers of people is typically a major undertaking especially
if the yield is low. Prescreening potential participants by
telephone to identify those with major exclusion criteria (e.g.,
using demographics, medical history) has been employed in many
projects. In the Lung Health Study, investigators used prescreening
to reduce the number of screening visits to approximately half of
those projected [38,
39]. Investigators need to
identify the best times to reach the maximum number of potential
participants. If they intend to make home visits or hope to contact
people by telephone, they should count on working evenings or
weekends. Unless potential participants are retired, or
investigators plan on contacting people at their jobs (which,
depending on the nature of the job, may be difficult), normal
working hours may not be productive times. Vacation periods and
summers are additional slow periods for recruitment.
The logistics of recruitment may
become more difficult when follow-up of enrolled participants
occurs while investigators are still recruiting. In long-term
studies, the most difficult time is usually towards the end of the
recruitment phase when the same staff, space, and equipment may be
used simultaneously for participants seen for screening, baseline,
and follow-up examinations. Resources can be stretched to the limit
and beyond if appropriate planning has not occurred.
The actual mechanics of recruiting
participants needs to be established in advance. A smooth clinic
operation is beneficial to all parties. Investigators must be
certain that necessary staff, facilities, and equipment are
available at appropriate times in the proper places. Keeping
potential participants waiting is a poor way to earn their
confidence.
Investigators and staff need to keep
abreast of recruitment efforts. Conducting regular staff meetings
and generating regular reports may serve as forums for discussion
of yields from various strategies, percent of recruitment goal
attained, as well as brainstorming and morale-boosting. These
meetings, useful for both single- and multicenter trials, also
provide the opportunity to remind everyone about the importance of
following the study protocol, including paying careful attention to
collection of valid data.
Record keeping of recruitment
activities is essential to allow analyses of recruitment yields and
costs from the various recruitment strategies. Recruiting large
numbers of potential participants requires the creation of
timetables, flow charts, and databases to ensure that screening and
recruitment proceed smoothly. Such charts should include the number
of people to be seen at each step in the process at a given time,
the number and type of personnel and amount of time required to
process each participant at each step, and the amount of equipment
needed (with an allowance for “down” time). A planned pilot phase
is helpful in making these assessments. One positive aspect of slow
early recruitment is that the “bugs” in the start-up process can be
worked out and necessary modifications made.
Several additional points regarding
the conduct of recruitment are worth emphasizing:
First, the success of a technique is
unpredictable. What works in one city at one time may not work at
the same place at another time or in another city. Therefore, the
investigator needs to be flexible and to leave room for
modifications.
Second, investigators must maintain
good relationships with participants’ personal physicians.
Physicians disapproving of the study or of the way it is conducted
are more likely to urge their patients not to participate.
Third, investigators must respect the
families of potential participants. Most participants like to
discuss research participation with their family and friends.
Investigators should be prepared to spend time reviewing the study
with them. If the study requires long-term cooperation from the
participant, we encourage such discussions. Anything that increases
family support is likely to lead to better recruitment and protocol
adherence.
Fourth, recruiting should not be
overly aggressive. While encouragement is necessary, excessive
efforts to convince, or “arm twist” people to participate could
prove harmful in the long run, in addition to raising ethical
concerns. One might argue that excessive salesmanship is unethical.
Those reluctant to join may be more likely to abandon the study
later or be poor adherers to study interventions after
randomization. Effective work on adherence begins during the
recruitment phase.
Fifth, the recruitment success is
closely associated with the level of commitment and effectiveness
of communication of the investigator and the study
coordinator.
Sixth, electronic health records and
social media provide important opportunities to use electronic data
and communication for certain types of trials.
Monitoring
Successful trial recruitment often
depends on establishing short-term and long-term recruitment goals.
The investigator should record these goals and make every effort to
achieve them. Since lagging recruitment commonly results from a
slow start, timely establishment of initial goals is crucial. The
investigator should be ready to randomize participants on the first
official day of study opening.
The use of weekly and/or monthly
interim goals in a long-term study orients the investigator and
staff to the short-term recruitment needs of the study. These goals
can serve as indicators for lagging recruitment and may help avoid
a grossly uneven recruitment pace. Inasmuch as participant
follow-up is usually done at regular intervals, uneven recruitment
results in periods of peak and slack during the follow-up phase.
This threatens effective use of staff time and equipment. Of
course, establishing a goal in itself does not guarantee timely
participant recruitment. The goals need to be realistic and the
investigator must make the commitment to meet each interim
goal.
The reasons for falling behind the
recruitment goal(s) should be determined. In a multicenter clinical
trial, valuable insight can be obtained by comparing results and
experiences from different centers. Those clinical sites with the
best recruitment performance can serve as role models for other
sites, which should be encouraged to incorporate other successful
techniques into their recruitment schemes. Multicenter studies
require a central office to oversee recruitment, compare enrollment
results, facilitate communication among sites, and lend support and
encouragement. Frequent feedback to the centers by means of tables
and graphs, which show the actual recruitment compared with
originally projected goals, are useful tools. Examples are shown in
the following figures and table. Figure 10.1 shows the progress of
an investigator who started participant recruitment on schedule and
maintained a good pace during the recruitment period. The
investigator and clinic staff accurately assessed participant
sources and demonstrated a commitment to enrolling participants in
a relatively even fashion. Figure 10.2 shows the record of
an investigator who started slowly, but later improved. However,
considerable effort was required to compensate for the poor start.
Clinic efforts included expanding the base from which participants
were recruited and increasing the time spent in enrollment. Even if
the clinic eventually catches up, the person-years of exposure to
the intervention has been reduced which may affect event rates and
trial power. In contrast, as seen in Fig. 10.3, the investigator
started slowly and was never able to improve his performance. This
center was dropped from a multicenter study because it could not
contribute enough participants to the study to make its continued
participation efficient. Figure 10.4 shows enrollment in
the TASTE trial [17], an example
of a trial that enrolled the majority of eligible patients in an
entire country, and even in this highly organized trial, enrollment
started gradually and the rate increased over the first few months.
Fig.
10.1
Participant recruitment in clinic that
consistently performed at goal rate
Fig.
10.2
Participant recruitment in a clinic that
started slowly and then performed at greater than goal rate
Fig.
10.3
Participant recruitment in a clinic that
performed poorly
Fig.
10.4
Participant recruitment in the TASTE trial
[17] using a national registry for
participant identification. PCI = percutaneous coronary
intervention
Table 10.1 shows goals, actual
recruitment, and projected final totals (assuming no change in
enrollment pattern) for three centers of a multicenter trial. Such
tables are useful to gauge recruitment efforts short-term as well
as to project final numbers of participants. The tables and figures
should be updated as often as necessary.
Table
10.1
Weekly recruitment status report by
center
|
Center
|
(1)
|
(2)
|
(3)
|
(4)
|
(5)
|
(6)
|
(7)
|
(8)
|
|---|---|---|---|---|---|---|---|---|
|
Contracted goal
|
Enrollment this week
|
Actual enrollment to date
|
Goal enrollment to date
|
Actual minus goal
|
Success rates (3)/(4)
|
Final projected intake
|
Final deficit or excess (7)-(1)
|
|
|
A
|
150
|
1
|
50
|
53.4
|
−3.4
|
0.94
|
140
|
−10
|
|
B
|
135
|
1
|
37
|
48.0
|
−11.0
|
0.77
|
104
|
−31
|
|
C
|
150
|
2
|
56
|
53.4
|
2.6
|
1.06
|
157
|
7
|
In single-center trials, the
investigator should also monitor recruitment status at regular and
frequent intervals. Review of these data with staff keeps everyone
aware of recruitment progress. If recruitment lags, the delay can
be noted early, the reasons identified, and appropriate action
taken.
Approaches to Lagging Recruitment
We have identified five possible
approaches to deal with lagging recruitment, in addition to the
strategies to enhance enrollment reviewed above.
The first is to accept a smaller
number of participants than originally planned. Doing this is far
from ideal as the power of the study will be reduced. In accepting
a smaller number of participants than estimated, the investigator
must either alter design features such as the primary response
variable, or change assumptions about intervention effectiveness
and participant adherence. As indicated elsewhere, such changes
midway in a trial may be liable to legitimate criticism. Only if
the investigator is lucky and discovers that some of the
assumptions used in estimating sample size were too pessimistic
would this “solution” provide comparable power. There are rare
examples of this happening. In a trial of aspirin in people with
transient ischemic attacks, aspirin produced a greater effect than
initially postulated [40].
Therefore, the less-than-hoped-for number of participants turned
out to be adequate.
A second approach is to relax the
inclusion criteria. This should be done only if little expectation
exists that the study design will suffer. The design can be marred
when, as a result of the new type of participants, the control
group event rate is altered to such an extent that the estimated
sample size is no longer appropriate. Also, the expected response
to intervention in the new participants may not be as great as in
the original participants. Furthermore, the intervention might have
a different effect or have a greater likelihood of being harmful in
the new participants than in those originally recruited. The
difference in additional participants would not matter if the
proportion of participants randomized to each group stayed the same
throughout recruitment. However, as indicated in Chap. 6, certain randomization schemes
alter that proportion, depending on baseline criteria or study
results. Under these circumstances, changing entrance criteria may
create imbalances among study arms.
The Coronary Drug Project provides a
classic example [41]. Only people
with documented Q-wave myocardial infarctions were originally
eligible. With enrollment falling behind, the investigators decided
to admit participants with non-Q-wave infarctions. Since there was
no reason to expect that the action of lipid-lowering agents being
studied would be any different in the new group than in the
original group and since the lipid-lowering agents were not
contraindicated in the new participants, the modification seemed
reasonable. However, there was some concern that overall mortality
rate would be changed because mortality in people with non-Q-wave
infarctions may be less than mortality in people with Q-wave
infarctions. Nevertheless, the pressure of recruitment overrode
that concern. Possible baseline imbalances did not turn out to be a
problem. In this particular study, where the total number of
participants was so large (8,341), there was every expectation that
randomization would yield comparable groups. If there had been
uncertainty regarding this, stratified randomization could have
been employed (Chap. 6). Including people with non-Q-wave
infarctions may have reduced the power of the study because this
group had a lower mortality rate than those with Q-wave infarctions
in each of the treatment groups, including the placebo group.
However, the treatments were equally ineffective when people with
Q-wave infarctions were analyzed separately from people with
non-Q-wave infarctions [42].
The third and probably most common
approach to recruitment problems is to extend the time for
recruitment or, in the case of multicenter studies, to add
recruiting centers. Both are the preferred solutions, requiring
neither modification of admission criteria nor diminution of power.
However, they are also the most costly. Whether the solution of
additional time or additional centers is adopted depends on cost,
on the logistics of finding and training other high quality
centers, and on the need to obtain study results quickly.
A fourth approach to lagging
recruitment is “recycling” of potential participants. When a
prospective participant just misses meeting the eligibility
criteria, the temptation is natural to try to enroll them by
repeating a measurement, perhaps under slightly different
conditions. Due to variability in a screening test, many
investigators argue that it is reasonable to allow one repeat test
and give a person interested in the trial a “second chance.” In
general, this practice should be discouraged. A study is harmed by
enrolling persons for whom the intervention might be ineffective or
inappropriate. However, in some progressive diseases, waiting a
year to recycle a potential participant may prove to be
useful.
Instances exist where, in order to
enter a drug study, the participant needs to be off all other
medication with similar actions. At baseline, participants may be
asked whether they have adhered to this requirement. If they have
not, the investigator may repeat the instructions and have the
participants return in a week for repeat baseline measurements. The
entrance criterion checks on a participant’s ability to adhere to a
protocol and their understanding of instructions. This “second
chance” is different from recycling and it is legitimate from a
design point of view. However, the second-chance participant, even
if he or she passes the repeat baseline measurement, may not be as
good a candidate for the study as someone who adhered on the first
occasion [43].
The fifth approach of broadening or
changing the pre-specified primary response variable is very
common, and is discussed in more detail in Chap. 3. Enrollment was slower than
expected in the Prevention of Events with Angiotensin Converting
Enzyme Inhibition (PEACE) trial [44]. This, combined with a narrowing of entry
criteria to exclude patients with diabetes and either proteinuria
or hypertension and microalbuminuria for whom benefit was clearly
established from other trials, prompted a change in the primary
outcome (of death from cardiovascular causes or nonfatal myocardial
infarction) to include coronary revascularization, reducing the
sample size from 14,100 patients to 8,100.
References
1.
America speaks, polling data
reflecting the views of Americans on medical, health and scientific
research. Available at
http://www.researchamerica.org/uploads/AmericaSpeaksV14.pdf.
Access January 21, 2015.
2.
ECRI Health Technology
Assessment Information Service. Patients’ reasons for participation
in clinical trials and effect of trial participation on patient
outcomes. ECRI Evidence Report. April 2002, Issue 74.
3.
Wright JR, Crooks D, Ellis
PM, et al. Factors that influence the recruitment of patients to
phase III studies in oncology. The perspective of the clinical
research assistant. Cancer
2002;95:1584–1591.CrossRef
4.
Cox K, McGarry J. Why
patients don’t take part in cancer clinical trials: an overview of
the literature. Eur J Cancer
Care 2003;12:114–122.CrossRef
5.
Sharp L, Cotton SC, Alexander
L, et al. on behalf of the TOMBOLA group. Reasons for participation
and non-participation in a randomized controlled trial: postal
questionnaire surveys of women eligible for TOMBOLA (Trial of
Management of Borderline and Other Low-grade Abnormal smears).
Clin Trials
2006;3:431–442.
6.
Barnes K. Patients provide
insight into trial participation. Outsourcing-Pharma.com, July 4,
2007. Available at
www.outsourcing-pharma.com/content/view/print/135930.
Access January 21, 2015.
7.
Mills EJ, Seely D, Rachlis B,
et al. Barriers to participation in clinical trials of cancer: a
meta-analysis and systematic review of patient-reported factors.
Lancet Oncol
2006;7:141–148.CrossRef
8.
Lovato LC, Hill K, Hertert S,
et al. Recruitment for controlled clinical trials: Literature
summary and annotated bibliography. Control Clin Trials
1997;18:328–357.CrossRef
9.
McDonald AM, Knight RC,
Campbell MK, et al. What influences recruitment to randomised
controlled trials? A review of trials funded by two UK funding
agencies. Trials
2006;7:9.
10.
Caldwell PH, Hamilton S, Tan
A, Craig JC. Strategies for increasing recruitment to randomised
controlled trials: systematic review. PLoS Med 2010;7:e1000368.CrossRef
11.
Hanson LC, Bull J, Wessell
K, et al. Strategies to support recruitment of patients with
life-limiting illness for research: the palliative care research
cooperative group. J Pain Symptom
Manage 2014;48:1021–1030.CrossRef
12.
Huynh L, Johns B, Liu SH, et
al. Cost-effectiveness of health research study participant
recruitment strategies: a systematic review. Clin Trials
2014;11:576–583.CrossRef
13.
Embi PJ, Jain A, Clark J, et
al. Effect of a clinical trial alert system on physician
participation in trial recruitment. Arch Intern Med
2005;165:2272–2277.CrossRef
14.
Thadani SR, Weng C, Bigger
JT, et al. Electronic screening improves efficiency in clinical
trial recruitment. J Am Med Inform
Assoc 2009;16:869–873.CrossRef
15.
Hawkins MS, Hough LJ, Berger
MA, et al. Recruitment of veterans from primary care into a
physical activity randomized controlled trial: the experience of
the VA-STRIDE study. Trials
2014;15:11.CrossRef
16.
Weng C, Appelbaum P,
Hripcsak G, et al. Using EHRs to integrate research with patient
care: promises and challenges. J
Am Med Inform Assoc 2012;19:684–687.CrossRef
17.
Fröbert O, Lagerqvist B,
Olivecrona GK, et al. Thrombus aspiration during ST-segment
elevation myocardial infarction. N
Engl J Med 2013;369:1587–1597.CrossRef
18.
Kääriäinen I, Sipponen P,
Siurala M. What fraction of hospital ulcer patients is eligible for
prospective drug trials? Scand J
Gastroenterol 1991;186:73–76.CrossRef
19.
Sheldon T. Dutch neurologist
found guilty of fraud after falsifying 438 case records.
Br Med J
2002;325:734.CrossRef
20.
Ross DB. The FDA and the
case of Ketek. N Engl J Med
2007;356:1601–1604.
21.
POISE Study Group. Effects
of extended-release metoprolol succinate in patients undergoing
non-cardiac surgery (POISE trial): a randomised controlled trial.
Lancet
2008;371:1839–1847.CrossRef
22.
Hunninghake DB. Summary
conclusions. Control Clin
Trials 1987;8:1S–5S.CrossRef
23.
Kingry C, Bastien A, Booth
G, et al for the ACCORD Study Group. Recruitment strategies in the
Action to Control Cardiovascular Risk in Diabetes (ACCORD) Trial.
Am J Cardiol 2007;99
(Suppl):68i–79i.
24.
Hulley SB, Furberg CD,
Gurland B, et al. Systolic Hypertension in the Elderly Program
(SHEP): antihypertensive efficacy of chlorthalidone. Am J Cardiol
1985;56:913–920.CrossRef
25.
Cosgrove N, Borhani NO,
Bailey G, et al. Mass mailing and staff experience in a total
recruitment program for a clinical trial: The SHEP Experience.
Control Clin Trials
1999;19:133–148.CrossRef
26.
Bryant J, Powell J. Payment
to healthcare professionals for patient recruitment to trials: a
systematic review. Br Med J
2005;331:1377–1378.CrossRef
27.
Freedman LS, Simon R,
Foulkes MA, et al. Inclusion of women and minorities in clinical
trials and the NIH Revitalization Act of 1993—the perspective of
NIH clinical trialists. Control
Clin Trials 1995;16:277–285.CrossRef
28.
Cook ED, Moody-Thomas S,
Anderson KB, et al. Minority recruitment to the Selenium and
Vitamin E Cancer Prevention Trial (SELECT). Clin Trials 2005;2:436–442.CrossRef
29.
Glickman SW, McHutchison JG,
Peterson ED, et al. Ethical and scientific implications of the
globalization of clinical research. N Engl J Med
2009;360:816–823.CrossRef
30.
Shibata M, Flather M, de
Arenaza DP, et al. Potential impact of socioeconomic differences on
clinical outcomes in international clinical trials. Am Heart J
2001;141:1019–1024.CrossRef
31.
Orlandini A, Diaz R, Wojdyla
D, et al. Outcomes of patients in clinical trials with ST-segment
elevation myocardial infarction among countries with different
gross national incomes. Eur Heart
J 2006;27:527–533.CrossRef
32.
O’Shea JC, Califf RM.
International differences in cardiovascular clinical trials.
Am Heart J
2001;141:866–874.CrossRef
33.
O'Connor CM, Fiuzat M,
Swedberg K, et al. Influence of global region on outcomes in heart
failure β-blocker trials. J Am
Coll Cardiol 2011;58:915–922.CrossRef
34.
Mahaffey KW, Wojdyla DM,
Carroll K, et al. Ticagrelor compared with clopidogrel by
geographic region in the Platelet Inhibition and Patient Outcomes
(PLATO) trial. Circulation
2011;124:544–554.CrossRef
35.
Vickers A, Goyal N, Harland
R, Rees R. Do certain countries produce only positive results? A
systematic review of controlled trials. Control Clin Trials
1998;19:159–166.CrossRef
36.
Ni Y, Kennebeck S, Dexheimer
JW, et al. Automated clinical trial eligibility prescreening:
increasing the efficiency of patient identification for clinical
trials in the emergency department. J Am Med Inform Assoc 2014; epub ahead
of print: doi: 10.1136/amiajnl-2014-002887.
37.
Köpcke F, Prokosch HU.
Employing computers for the recruitment into clinical trials: a
comprehensive systematic review. J
Med Internet Res 2014;16:e161.CrossRef
38.
Durkin DA, Kjelsberg MO,
Buist AS, et al. Recruitment of participants in the Lung Health
Study, I: description of methods. Control Clin Trials
1993;14:20S–37S.CrossRef
39.
Daly M, Seay J, Balshem A,
et al. Feasibility of a telephone survey to recruit health
maintenance organization members into a tamoxifen chemoprevention
trial. Cancer Epidemiol Biomarkers
Prev 1992;1:413–416.
40.
Fields WS, Lemak NA
Frankowski RF, et al. Controlled trial of aspirin in cerebral
ischemia. Stroke
1977;8:301–314.CrossRef
41.
The Coronary Drug Project
Research Group. The Coronary Drug Project: design, methods, and
baseline results. Circulation 1973;47:I-1-I-50.
42.
The Coronary Drug Project
Research Group. Clofibrate and niacin in coronary heart disease.
JAMA
1975;231:360–381.CrossRef
43.
Sackett DL. A compliance
practicum for the busy practitioner. In Haynes RB, Taylor DW,
Sackett DL (eds.). Compliance in
Health Care. Baltimore: Johns Hopkins University Press,
1979.
44.
The PEACE Trial
Investigators. Angiotensin-converting-enzyme inhibition in stable
coronary artery disease. N Engl J
Med 2004;351:2058–2068.