- Email: [email protected]
Participant recruitment to the Lipid Research Clinics Coronary Primary Prevention Trial
3 Participant Recruitment to the Lipid Research Clinics Coronary Primary Prevention Trial Reagan H. Bradford
ABSTRACT: The Lipid Research Clinics Coronary Primary Prevention Trial, a multicenter
clinical trial sponsored by the National Heart, Lung, and Blood Institute, was undertaken in 1973 to evaluate whether lowering blood cholesterol would reduce the incidence of coronary heart disease. The study protocol limited entry into the trial to men 35-59 years of age who had type II hyperlipoproteinemia with plasma cholesterol of at least 265 mg/dl, who were free of coronary heart disease symptoms, and who were in good general health. Participants were initially recruited from physician and clinical laboratory referrals. After the first year of recruitment, yields from these sources were unacceptably low, and emphasis was directed to identifying new sources using mass screening. Six additional sources were used--blood banks, clinical studies, community screenings, mass mailings, media, and occupational (worksite) screenings. The target recruitment goal was exceeded although it required more than twice as long as originally planned. Characteristics of these eight recruitment sources and recommendations for those undertaking a similar study are presented, based on the experience gained from recruitment of participants for this trial. KEY WORDS: primary prevention trial, blood cholesterol, coronary heart disease, patient recruitment
INTRODUCTION The Lipid Research Clinics C o r o n a r y Primary Prevention Trial (LRC-CPPT) was a d o u b l e - m a s k e d placebo-controlled clinical trial that tested the efficacy of lowering cholesterol levels for primary p r e v e n t i o n of c o r o n a r y heart disease [1]. Twelve N o r t h American Lipid Research Clinics participated in this trial. The s t u d y protocol limited entry into the trial to m e n 35-59 years of age w h o had primary type II h y p e r l i p o p r o t e i n e m i a with fasting plasma cholesterols of at least 265 mg/dl. This cholesterol level a p p r o x i m a t e d the 95th percentile for m e n ages 40-49 years, based o n results from a pilot phase of the Lipid Research Clinic Prevalence Study. To qualify, m e n also h a d to be free of s y m p t o m s of c o r o n a r y heart disease and in good general health. A substantial n u m b e r of prospective participants w e r e excluded d u e to blood cholesterol level, h y p e r t e n s i o n , s y m p t o m a t i c c o r o n a r y heart disease, life-limiting or comorbid conditions such as cancer or nonatherosclerotic cardiovasControlledClinicalTrials8:31S-40S(1987) © ElsevierSciencePublishing Co., Inc. 1987 52 VanderbiltAve., New York, New York10017
31S 0197-2456/87/$3.50
32S
R.H. Bradford cular disease, use of certain medications, an accentuated responsiveness to diet, and unwillingness to participate in the study. Recruitment of prospective participants [2,3] began with an initial contact, at which screenees were required to have the prescribed cholesterol level of at least 265 mg/dl. After this contact, the study protocol specified a series of five monthly visits during the accrual (preentry) phase to evaluate the eligibility of prospective participants according to study criteria. At these visits, medical history data were collected, physical examinations of various types were performed, lipids and lipoproteins and other clinical chemistries were measured, and resting and graded exercise electrocardiograms were carried out. At the second preentry visit, a moderate cholesterol-lowering diet was specified for all participants. This diet was designed to provide 400 mg of cholesterol per day and a polyunsaturated-to-saturated fat ratio of approximately 0.8. A total of 436,679 age-eligible men were screened, resulting in 18,013 w h o qualified for visit 1. The number of eligible men dropped to 7731 at visit 2, representing only 43% of those eligible for visit 1. With each subsequent preentry visit, the number of eligible participants decreased progressively, so that at visit 5, a total of 3810 men were randomized into one of two treatment groups. The group that was finally eligible for entry into this study represented some 21% of those eligible for visit 1 and less than 1% of the ageeligible men w h o were first screened. Approximately 96% of those screened at the initial contact were excluded, due mostly to failure to meet prescribed lipid values. Between the first visit and entry to the study, an additional 79% of those evaluated were excluded. Almost three quarters of those excluded during these visits were ineligible due to low lipid values, with the remaining one quarter of exclusions due to medical and other criteria.
TIME COURSE OF RECRUITMENT
Calculations during the planning phase indicated that a sample size of at least 3550 participants would be necessary. Recruitment began in July 1973 with a goal for completion within 18 months. We planned to use referrals from physicians and clinical laboratories as our sources of participants, based on the experience with these sources by the National Heart, Lung, and Blood Institute Type II Coronary Intervention Study. These referrals were expected to provide participants w h o would meet the lipid, age, and sex criteria of our protocol. Thus, the pool of potential participants available by this approach would be enriched significantly over the general population. Furthermore, the initial determination of potential eligibility would be done by others, making this approach cost and effort efficient. The principal cost to us with this strategy was in the physician-intensive operation needed for interaction with community physicians and clinical laboratories. Although these sources were attractive insofar as cost and effort were concerned, they failed to generate a volume of initial contacts adequate to provide the required number of entries into the trial. Figure 1 shows the cumulative entries over the course of recruitment. Attainment of 3550 entries within the prescribed 18-month interval would
33S
Coronary Primary Prevention Trial Recruitment Date (month / year) 7'73
1/74 I
7/74
1/75
I
]
1/76
7/75 I
7/76
I
I
39OO 3600 330O 3000 =
2700 1
o
2400, 2100
target (18 months)
~ 1800 ~
[/
)
/
~ ~
1500 1200 9013 60O
240/month 180/month
//j
120moo~ 60/month 30/moflth
300 8
24 32 40 48 56 64 72 80 88 96 104 112 120 128 136 144 152 160 Weeks since initiation of recruitment
Figure 1. Cumulative entries into the LRC-CPPT, over time. (From The Lipid Re-
search Clinics Program: Participant recruitment to the Coronary Primary Prevention Trial. J Chron Dis 36:451-465, !983.) have required an average of 254 entries per 4-week interval, based on the observed retention rate. After the first year of recruitment with physician and laboratory referrals as the primary sources, the trial had attained a rate of only some 50 entries per 4-week period, i.e., 20% of the rate needed to meet our recruitment target. At this point, the viability of the trial was in jeopardy due to the low rate of recruitment. Over the next several months, a major revision took place in the overall approach to the recruitment effort in each clinical center. A very high priority was placed on identifying and exploiting new pools of potential participants. In general, the use of new sources proceeded in a stepwise fashion. Ultimately, in each clinical center this led to the adoption of mass screening strategies, using sources that inherently had less rich pools (i.e., proportion of persons in source pool who met initial eligibility requirements) and lower efficiency (i.e., proportion of initial contacts that proceeded to visit 1) than referrals. Thus, the use of mass screening was begun only after it was evident that this would be necessary to accomplish the minimum sample-size goal within a reasonable period of time. The response in recruitment during the second and third years is apparent in Figure 1. As a consequence of the n e w recruitment strategies, the rate of entries increased from an average of 41 per month during year 1 to 200 per month by the end of year 2. This rate declined later as several dinical centers approached their recruitment goals. On completion of recruitment after 37 months, 3810 participants were entered into the trial, surpassing the target recruitment goal of 3550 by 7%. CHARACTERISTICS OF RECRUITMENT SOURCES
Eight recruitment sources (i.e., a specific process of acquiring potential participants from a pool) were utilized--referrals from physicians or clinical
34S
R.H. Bradford laboratories, blood banks, clinical studies, community screenings (i.e., screenings at c o m m u n i t y events or locations, or involving community organizations), mass mailings, media, and occupational (worksite) screenings. From these categories, a total of 1857 individual sources were screened. The 12 clinical centers (see appendix) collectively screened 19 blood banks and more than 1100 occupational sites. Blood banks and occupational screenings were used at each of the 12 clinical centers. Clinical laboratories, clinical studies, c o m m u n i t y screenings, media, and medical referrals were used by 11 centers, whereas mass mailing was used by only 7 centers. Six centers used all eight source categories, five centers used seven sources, and one center used only three sources. Figure 2 depicts the time interval during which the various recruitment sources were utilized a n d the resulting flow rate of initial contacts. Medical referrals, clinical laboratories, a n d then clinical studies during the first year yielded a relatively low n u m b e r of initial contacts. Beginning early in year 2, Date ( m o n t h / d a y ) 7 73
1 74 1
7 74 •
1 75 I - -
7 75 I - -
1 76 l - -
!
40,000
3,000 Occupational Screenings ........
30.000
-A
.........
-
C ~: •
2.000 ~Z ~o
~- ~ "6 ~. 20,000
Blood Banks Commun'tyScreening s - - / - - ~ L
- -
.
.
.
.
.
. . . . . .
-~- . . . .
/
o ~
|
I
v
1 ,.ooo
<
E
,o.ooo
12
24
36
48
60
72
84
96
108
120
132
144
W e e k s since initiation of r e c r u i t m e n t
Figure 2.
Flow rate (number per 4-week period) of initial contacts, over time, and period of major yield for each category of recruitment sources (number per week). The data presented here include individuals who did not meet the age and sex eligibility criteria and reflect the temporal data that were available for 96% of the sample. The period of major yield depicted for each source excludes the time to acquire the first and last 10% of the total from the source, thereby minimizing both the initiation and termination phases. Temporal data were available for at least 90% of those from each source except clinical studies (88%), media (65%), and medical referrals (55%). (From The Lipid Research Clinics Program: Participant recruitment to the Coronary Primary Prevention Trial. J Chron Dis 36:451-465, 1983.)
35S
Coronary Primary Prevention Trial Recruitment
additional sources---media, blood banks, community and occupational screenings, and then mass mailings---were incorporated sequentially into the recruitment strategy. With this introduction of primary screening, a marked increase in the number of initial contacts was observed. This increased level of screening was maintained throughout the remainder of the active recruitment period. It is also noteworthy that the average yield per week of initial contacts from medical referrals, clinical studies, media, clinical laboratories, and mass mailings was much less than the yield from community screenings, blood banks, and, especially, occupational screenings. One measure of efficiency of a recruitment source is the ratio of the number attending visit 1 to initial contacts for those recruited from that source. Table 1 indicates that this measure of efficiency varied considerably among recruitment sources. A difference of up to 15-fold was observed between clinical studies and medical referrals versus mass mailings, occupational and community screenings, and blood banks. Another measure of efficiency of a recruitment source is the ratio of randomized participants to initial contacts for those recruited from that source. Table 1 lists this efficiency measure for each of the recruitment sources. Here again medical referrals and clinical studies represent the most efficient sources, while mass mailings, community and occupational screenings, and blood banks are the least efficient. When compared by this measure, a 28-fold difference exists between the most efficient source, medical referrals, and the least efficient source, blood banks. Table I also lists the proportion of total entries that were derived from each source. It is noteworthy that medical referrals and clinical studies together provided only 15% of the trial participants, whereas mass mailings, community and occupational screenings, and blood banks accounted for almost 60%. Another factor that had to be reckoned with was participant self-exclusion because of unwillingness to proceed to the next visit by men who otherwise met the criteria at their last completed visit. Self-exclusions due to unwillingness to proceed were observed at each of the preentry visits, although this was most evident at the initial screening when almost 30% of the eligible men declined further participation. Men recruited from mass mailings, media, and
T a b l e I Efficiency of Recruitment Sources
Number at Visit 1
Entries
Source
Initial Contacts (%)
Initial Contacts (%)
Percent of Total Entriesa
Clinical studies Medical referrals Mass media Clinical laboratories Mass mailings Occupational screenings Community screenings Blood banks
35.9 30.3 10.9 7.8 4.9 3.5 2.9 2.4
9.0 14.1 2.3 1.8 0.8 0.6 0.6 0.5
11 4 11 6 6 24 14 15
aApproximately9% of entries were not classifiablein one of the categorieslisted.
36S
R.H. Bradford Date
(month/day)
7/29
10/21
1/13
i
i
i
4/7
6/30
i
i
45 ' 40
35
c
3O
~. 25 1973-74 20 1974-75 15 10 5 .
4
8
,
,
.
12
16
20
.
.
24
28
,
,
,
.
,
32
36
40
44
48
.
52
Weeks since July 1
Figure 3. Proportion of participants proceeding from visit I to entry, over time. Open circles indicate periods representing less than 100 participants at visit 1. (From The Lipid Research Clinics Program: Participant recruitment to the Coronary Primary Prevention Trial. J Chron Dis 36:451-465, 1983.)
medical referrals were only one third as likely to be willing to participate as were those recruited from other sources. Another key variable that increased the magnitude of the recruitment effort with time was the proportion of men proceeding from visit 1 to entry. This proportion varied considerably during the course of the recruitment period but generally decreased over time (Fig. 3). Recruitment efficiencies not only varied among recruitment sources but also among clinical centers. If efficiency is measured by the number of entries relative to the number of initial contacts, the values for individual centers ranged from a low of 1 entry for every 217 initial contacts at one center, to the highest efficiency of I entry for every 36 initial contacts. One factor that accounted for some of this sixfold range in efficiencies was the relative mix of sources used by individual centers. The sociodemographic characteristics of the LRC-CPPT participants indicate that these men are not representative of the general population in that they are better educated, less likely to be a member of a minority group, and more likely to be married [4].
RECOMMENDATIONS FOR D O I N G A SIMILAR STUDY
The experience gained in the LRC-CPPT and other clinical trials suggests that a recruitment plan addressing specific critical issues should be carefully
Coronary Primary Prevention Trial Recruitment
37S
formulated and put in place before a trial or other study is undertaken [5]. The recruitment plan should take account of recruitment sources that are to be utilized and recognize that individual recruitment sources provide a differential access to the pool of potential participants and differ in volume, flow rate characteristics, richness, and efficiency. A significant factor to be considered with individual sources is the lag time from initial negotiation with the source until the first screening from that source is processed. Ideally, recruitment should begin with approaches to sources with longer lag times, followed by those with shorter lag times, to ensure that an adequate number of sources will be available for activation at any time. In addition, the flow characteristics of each source must be considered. Sources with high initial flow rates may be largely expended in a very short time after the screening of the individual source is begun. For example, three fourths of the total participants from individual sources of occupational and community screenings may be processed during the first week of screening. Screenings involving these individual sources must be carefully scheduled over the time span of a recruitment campaign to ensure a reasonable constancy of flow. Other sources with a much more sustained flow, such as medical referrals, clinical laboratories, and blood banks, should be activated early in the campaign so that sufficient time is available to process the bulk of participants from these sources. Other factors that must be considered in the planning are the prevalence of the target disorder in the population to be studied, the desired length of the recruitment period, the average lag time to get under way, the time required for completing preentry protocol visits, and the estimated refusal rate for participation. The size and complexity of recruitment campaigns vary considerably, from the small-scale study involving only a very few centers to the large-scale multicenter study. In the multicenter study, planning at both the collaborative and local levels is needed. At the local level, the planning should focus on the mix of sources that the clinical center expects to use. This has important implications for the workload of local staff members, as well as for their ability to meet the schedule. Once the plan has been developed, consideration must be given to the type of local organization needed to carry out the recruitment. Screening goals should be established that are within the present and future capability of the clinical center. Interim goals should be set that recognize the cumulative nature of participant preentry visits and the time requirements of the clinical unit for visits by participants w h o are already entered into the study. REFERENCES
1. The Lipid Research Clinics Program: The Coronary Primary Prevention Trial: Design and Implementation. J Chron Dis 32:609-631, 1979 2. Agras WS, Bradford RI--I, Eds: Recruitment for Clinical Trials: The Lipid Research Clinics Coronary Primary Prevention Trial experience--its implications for future trials. Circulation, Part II, 66(6), 1982 3. The Lipid Research Clinics Program: Participant recruitment to the Coronary Primary Prevention Trial. J Chron Dis 36:451-465, 1983
38S
R.H. Bradford
4. The Lipid Research Clinics Program: Pre-entry characteristics of participants in the Lipid Research Clinics Coronary Primary Prevention Trial. J Chron Dis 36:467-479, 1983 5. Agras WS, Marshall GD, Kraemer HC: Planning recruitment. In: Recruitment for clinical trials: The Lipid Research Clinics Coronary Primary Prevention Trial experience--Its implications for future trials, Agras WS, Bradford RH, Eds. Circulation, Part II, 66(6):IV54-IV58, 1982
APPENDIX: CPPT SITES A N D PERSONNEL, JULY 1976 I. Lipid Research Clinics Baylor College of Medicine, Houston, Texas Principal Investigator: Antonio M. Gotto, M.D. CPPT Director: David O. Taunton, M.D. Recruitment Coordinator: Virginia Elverson University of Cincinnati Medical Center, Cincinnati, Ohio Principal Investigator: Charles J. Glueck, M.D. CPPT Director: Moti Kashyap, M.D. CPPT Co-Director: Ronald Fallat, M.D. Recruitment Coordinator: Susan Kutchinski George Washington University ~k4edical Center, Washington, D.C. Principal Investigator: John LaRosa, M.D. CPPT Director: John LaRosa, M.D. Recruitment Coordinator: Marilyn McKeown, R.N.
University of Iowa Hospitals, Iowa City, Iowa Principal Investigator: Francois Abboud, M.D. Former Principal Investigator: William Connor, M.D. CPPT Director: Helmut Schrott, M.D. Recruitment Coordinator: Lois Rude Johns Hopkins University, Baltimore, Maryland Principal Investigator: Peter O. Kwiterovitch, M.D. CPPT Director: G. William Benedict, M.D. Recruitment Coordinator: Elizabeth J. Bruen University of Minnesota, Minneapolis, Minnesota Principal Investigator: Ivan Frantz, M.D. CPPT Director: Donald Hunninghake, M.D. Recruitment Coordinator: Florine Peterson, R.N. Oklahoma Medical Research Foundation, Oklahoma City, Oklahoma Principal Investigator: Reagan H. Bradford, Ph.D., M.D. CPPT Director: Thomas F. Whayne, M.D., Ph.D. Recruitment Coordinators: Marvin McDearmon Dana Pitts University of California, San Diego, La Jolla, California Principal Investigator: W. Virgil Brown, M.D. CPPT Director: Richard C. Gross, M.D.
Coronary Primary Prevention Trial Recruitment
39S
Co-Director: John Melish, M.D. Recruitment Coordinator: Nancy Rotchstein, R.N. University of Washington, Seattle, Washington Principal Investigator: William R. Hazzard, M.D. CPPT Director: Robert H. Knopp, M.D. Recruitment Coordinator: Kathy Stamm Stanford University, Stanford, California Principal Investigator: John W. Farquhar, M.D. CPPT Director: Michael P. Stern, M.D. Recruitment Coordinator: Cheryl Guerin Universities of Toronto and McMaster, Toronto and Hamilton, Ontario, Canada Principal Investigator: J. Alick Little, M.D. CPPT Director: Randolph Lee, M.D. Co-Director: Josephine Bird, M.D. (St. Michael's Hospital) CPPT Director: George Steiner, M.D. Co-Director: David Stinson, M.D. (Toronto General Hospital) CPPT Director: Maurice Mishkel, M.D. (Hamilton General Hospital) Recruitment Coordinators: Norma Mishkel Judy Wright Washington University School of Medicine, St. Louis, Missouri Principal Investigator: Gustav Schonfeld, M.D. CPPT Director: Joseph L. Witztum, M.D. Recruitment Coordinators: Betsy Boles Ruth Jacobson II. Support Facilities Central Patient Registry and Coordinating Center, Chapel Hill, North Carolina Director: O. Dale Williams, Ph.D. Former Director: James Grizzle, Ph.D. Deputy Program Director: C.E. Davis, Ph.D. CPPT Coordinator: Sandy Irving, M.S. Recruitment and Adherence Consultants, Stanford, California Director: W. Stewart Agras, M.D. Jacqueline Dunbar, Ph.D. Nathan Maccoby, Ph.D. Albert Stunkard, M.D. Steven M. Zifferblatt, Ph.D. III. Program Office Lipid Metabolism Branch, Division of Heart and Vascular Diseases, National Heart, Lung, and Blood Institute, Bethesda, Maryland
40S
R.H. Bradford Chief: Basil M. Rifkind, M.D., F.R.C.P. Former Chief: Robert I. Levy, M.D. CPPT Coordinator: Richard Havlik, M.D. Medical Officer: Alan Seplowitz, M.D. Former Medical Officers: Conrad Blum, M.D. Gail Morrison, M.D. Nutrition Coordinator: Virginia Keating, R.D. Laboratory Coordinator: Kenneth Lippel, Ph.D.
ABSTRACT: The Lipid Research Clinics Coronary Primary Prevention Trial, a multicenter
clinical trial sponsored by the National Heart, Lung, and Blood Institute, was undertaken in 1973 to evaluate whether lowering blood cholesterol would reduce the incidence of coronary heart disease. The study protocol limited entry into the trial to men 35-59 years of age who had type II hyperlipoproteinemia with plasma cholesterol of at least 265 mg/dl, who were free of coronary heart disease symptoms, and who were in good general health. Participants were initially recruited from physician and clinical laboratory referrals. After the first year of recruitment, yields from these sources were unacceptably low, and emphasis was directed to identifying new sources using mass screening. Six additional sources were used--blood banks, clinical studies, community screenings, mass mailings, media, and occupational (worksite) screenings. The target recruitment goal was exceeded although it required more than twice as long as originally planned. Characteristics of these eight recruitment sources and recommendations for those undertaking a similar study are presented, based on the experience gained from recruitment of participants for this trial. KEY WORDS: primary prevention trial, blood cholesterol, coronary heart disease, patient recruitment
INTRODUCTION The Lipid Research Clinics C o r o n a r y Primary Prevention Trial (LRC-CPPT) was a d o u b l e - m a s k e d placebo-controlled clinical trial that tested the efficacy of lowering cholesterol levels for primary p r e v e n t i o n of c o r o n a r y heart disease [1]. Twelve N o r t h American Lipid Research Clinics participated in this trial. The s t u d y protocol limited entry into the trial to m e n 35-59 years of age w h o had primary type II h y p e r l i p o p r o t e i n e m i a with fasting plasma cholesterols of at least 265 mg/dl. This cholesterol level a p p r o x i m a t e d the 95th percentile for m e n ages 40-49 years, based o n results from a pilot phase of the Lipid Research Clinic Prevalence Study. To qualify, m e n also h a d to be free of s y m p t o m s of c o r o n a r y heart disease and in good general health. A substantial n u m b e r of prospective participants w e r e excluded d u e to blood cholesterol level, h y p e r t e n s i o n , s y m p t o m a t i c c o r o n a r y heart disease, life-limiting or comorbid conditions such as cancer or nonatherosclerotic cardiovasControlledClinicalTrials8:31S-40S(1987) © ElsevierSciencePublishing Co., Inc. 1987 52 VanderbiltAve., New York, New York10017
31S 0197-2456/87/$3.50
32S
R.H. Bradford cular disease, use of certain medications, an accentuated responsiveness to diet, and unwillingness to participate in the study. Recruitment of prospective participants [2,3] began with an initial contact, at which screenees were required to have the prescribed cholesterol level of at least 265 mg/dl. After this contact, the study protocol specified a series of five monthly visits during the accrual (preentry) phase to evaluate the eligibility of prospective participants according to study criteria. At these visits, medical history data were collected, physical examinations of various types were performed, lipids and lipoproteins and other clinical chemistries were measured, and resting and graded exercise electrocardiograms were carried out. At the second preentry visit, a moderate cholesterol-lowering diet was specified for all participants. This diet was designed to provide 400 mg of cholesterol per day and a polyunsaturated-to-saturated fat ratio of approximately 0.8. A total of 436,679 age-eligible men were screened, resulting in 18,013 w h o qualified for visit 1. The number of eligible men dropped to 7731 at visit 2, representing only 43% of those eligible for visit 1. With each subsequent preentry visit, the number of eligible participants decreased progressively, so that at visit 5, a total of 3810 men were randomized into one of two treatment groups. The group that was finally eligible for entry into this study represented some 21% of those eligible for visit 1 and less than 1% of the ageeligible men w h o were first screened. Approximately 96% of those screened at the initial contact were excluded, due mostly to failure to meet prescribed lipid values. Between the first visit and entry to the study, an additional 79% of those evaluated were excluded. Almost three quarters of those excluded during these visits were ineligible due to low lipid values, with the remaining one quarter of exclusions due to medical and other criteria.
TIME COURSE OF RECRUITMENT
Calculations during the planning phase indicated that a sample size of at least 3550 participants would be necessary. Recruitment began in July 1973 with a goal for completion within 18 months. We planned to use referrals from physicians and clinical laboratories as our sources of participants, based on the experience with these sources by the National Heart, Lung, and Blood Institute Type II Coronary Intervention Study. These referrals were expected to provide participants w h o would meet the lipid, age, and sex criteria of our protocol. Thus, the pool of potential participants available by this approach would be enriched significantly over the general population. Furthermore, the initial determination of potential eligibility would be done by others, making this approach cost and effort efficient. The principal cost to us with this strategy was in the physician-intensive operation needed for interaction with community physicians and clinical laboratories. Although these sources were attractive insofar as cost and effort were concerned, they failed to generate a volume of initial contacts adequate to provide the required number of entries into the trial. Figure 1 shows the cumulative entries over the course of recruitment. Attainment of 3550 entries within the prescribed 18-month interval would
33S
Coronary Primary Prevention Trial Recruitment Date (month / year) 7'73
1/74 I
7/74
1/75
I
]
1/76
7/75 I
7/76
I
I
39OO 3600 330O 3000 =
2700 1
o
2400, 2100
target (18 months)
~ 1800 ~
[/
)
/
~ ~
1500 1200 9013 60O
240/month 180/month
//j
120moo~ 60/month 30/moflth
300 8
24 32 40 48 56 64 72 80 88 96 104 112 120 128 136 144 152 160 Weeks since initiation of recruitment
Figure 1. Cumulative entries into the LRC-CPPT, over time. (From The Lipid Re-
search Clinics Program: Participant recruitment to the Coronary Primary Prevention Trial. J Chron Dis 36:451-465, !983.) have required an average of 254 entries per 4-week interval, based on the observed retention rate. After the first year of recruitment with physician and laboratory referrals as the primary sources, the trial had attained a rate of only some 50 entries per 4-week period, i.e., 20% of the rate needed to meet our recruitment target. At this point, the viability of the trial was in jeopardy due to the low rate of recruitment. Over the next several months, a major revision took place in the overall approach to the recruitment effort in each clinical center. A very high priority was placed on identifying and exploiting new pools of potential participants. In general, the use of new sources proceeded in a stepwise fashion. Ultimately, in each clinical center this led to the adoption of mass screening strategies, using sources that inherently had less rich pools (i.e., proportion of persons in source pool who met initial eligibility requirements) and lower efficiency (i.e., proportion of initial contacts that proceeded to visit 1) than referrals. Thus, the use of mass screening was begun only after it was evident that this would be necessary to accomplish the minimum sample-size goal within a reasonable period of time. The response in recruitment during the second and third years is apparent in Figure 1. As a consequence of the n e w recruitment strategies, the rate of entries increased from an average of 41 per month during year 1 to 200 per month by the end of year 2. This rate declined later as several dinical centers approached their recruitment goals. On completion of recruitment after 37 months, 3810 participants were entered into the trial, surpassing the target recruitment goal of 3550 by 7%. CHARACTERISTICS OF RECRUITMENT SOURCES
Eight recruitment sources (i.e., a specific process of acquiring potential participants from a pool) were utilized--referrals from physicians or clinical
34S
R.H. Bradford laboratories, blood banks, clinical studies, community screenings (i.e., screenings at c o m m u n i t y events or locations, or involving community organizations), mass mailings, media, and occupational (worksite) screenings. From these categories, a total of 1857 individual sources were screened. The 12 clinical centers (see appendix) collectively screened 19 blood banks and more than 1100 occupational sites. Blood banks and occupational screenings were used at each of the 12 clinical centers. Clinical laboratories, clinical studies, c o m m u n i t y screenings, media, and medical referrals were used by 11 centers, whereas mass mailing was used by only 7 centers. Six centers used all eight source categories, five centers used seven sources, and one center used only three sources. Figure 2 depicts the time interval during which the various recruitment sources were utilized a n d the resulting flow rate of initial contacts. Medical referrals, clinical laboratories, a n d then clinical studies during the first year yielded a relatively low n u m b e r of initial contacts. Beginning early in year 2, Date ( m o n t h / d a y ) 7 73
1 74 1
7 74 •
1 75 I - -
7 75 I - -
1 76 l - -
!
40,000
3,000 Occupational Screenings ........
30.000
-A
.........
-
C ~: •
2.000 ~Z ~o
~- ~ "6 ~. 20,000
Blood Banks Commun'tyScreening s - - / - - ~ L
- -
.
.
.
.
.
. . . . . .
-~- . . . .
/
o ~
|
I
v
1 ,.ooo
<
E
,o.ooo
12
24
36
48
60
72
84
96
108
120
132
144
W e e k s since initiation of r e c r u i t m e n t
Figure 2.
Flow rate (number per 4-week period) of initial contacts, over time, and period of major yield for each category of recruitment sources (number per week). The data presented here include individuals who did not meet the age and sex eligibility criteria and reflect the temporal data that were available for 96% of the sample. The period of major yield depicted for each source excludes the time to acquire the first and last 10% of the total from the source, thereby minimizing both the initiation and termination phases. Temporal data were available for at least 90% of those from each source except clinical studies (88%), media (65%), and medical referrals (55%). (From The Lipid Research Clinics Program: Participant recruitment to the Coronary Primary Prevention Trial. J Chron Dis 36:451-465, 1983.)
35S
Coronary Primary Prevention Trial Recruitment
additional sources---media, blood banks, community and occupational screenings, and then mass mailings---were incorporated sequentially into the recruitment strategy. With this introduction of primary screening, a marked increase in the number of initial contacts was observed. This increased level of screening was maintained throughout the remainder of the active recruitment period. It is also noteworthy that the average yield per week of initial contacts from medical referrals, clinical studies, media, clinical laboratories, and mass mailings was much less than the yield from community screenings, blood banks, and, especially, occupational screenings. One measure of efficiency of a recruitment source is the ratio of the number attending visit 1 to initial contacts for those recruited from that source. Table 1 indicates that this measure of efficiency varied considerably among recruitment sources. A difference of up to 15-fold was observed between clinical studies and medical referrals versus mass mailings, occupational and community screenings, and blood banks. Another measure of efficiency of a recruitment source is the ratio of randomized participants to initial contacts for those recruited from that source. Table 1 lists this efficiency measure for each of the recruitment sources. Here again medical referrals and clinical studies represent the most efficient sources, while mass mailings, community and occupational screenings, and blood banks are the least efficient. When compared by this measure, a 28-fold difference exists between the most efficient source, medical referrals, and the least efficient source, blood banks. Table I also lists the proportion of total entries that were derived from each source. It is noteworthy that medical referrals and clinical studies together provided only 15% of the trial participants, whereas mass mailings, community and occupational screenings, and blood banks accounted for almost 60%. Another factor that had to be reckoned with was participant self-exclusion because of unwillingness to proceed to the next visit by men who otherwise met the criteria at their last completed visit. Self-exclusions due to unwillingness to proceed were observed at each of the preentry visits, although this was most evident at the initial screening when almost 30% of the eligible men declined further participation. Men recruited from mass mailings, media, and
T a b l e I Efficiency of Recruitment Sources
Number at Visit 1
Entries
Source
Initial Contacts (%)
Initial Contacts (%)
Percent of Total Entriesa
Clinical studies Medical referrals Mass media Clinical laboratories Mass mailings Occupational screenings Community screenings Blood banks
35.9 30.3 10.9 7.8 4.9 3.5 2.9 2.4
9.0 14.1 2.3 1.8 0.8 0.6 0.6 0.5
11 4 11 6 6 24 14 15
aApproximately9% of entries were not classifiablein one of the categorieslisted.
36S
R.H. Bradford Date
(month/day)
7/29
10/21
1/13
i
i
i
4/7
6/30
i
i
45 ' 40
35
c
3O
~. 25 1973-74 20 1974-75 15 10 5 .
4
8
,
,
.
12
16
20
.
.
24
28
,
,
,
.
,
32
36
40
44
48
.
52
Weeks since July 1
Figure 3. Proportion of participants proceeding from visit I to entry, over time. Open circles indicate periods representing less than 100 participants at visit 1. (From The Lipid Research Clinics Program: Participant recruitment to the Coronary Primary Prevention Trial. J Chron Dis 36:451-465, 1983.)
medical referrals were only one third as likely to be willing to participate as were those recruited from other sources. Another key variable that increased the magnitude of the recruitment effort with time was the proportion of men proceeding from visit 1 to entry. This proportion varied considerably during the course of the recruitment period but generally decreased over time (Fig. 3). Recruitment efficiencies not only varied among recruitment sources but also among clinical centers. If efficiency is measured by the number of entries relative to the number of initial contacts, the values for individual centers ranged from a low of 1 entry for every 217 initial contacts at one center, to the highest efficiency of I entry for every 36 initial contacts. One factor that accounted for some of this sixfold range in efficiencies was the relative mix of sources used by individual centers. The sociodemographic characteristics of the LRC-CPPT participants indicate that these men are not representative of the general population in that they are better educated, less likely to be a member of a minority group, and more likely to be married [4].
RECOMMENDATIONS FOR D O I N G A SIMILAR STUDY
The experience gained in the LRC-CPPT and other clinical trials suggests that a recruitment plan addressing specific critical issues should be carefully
Coronary Primary Prevention Trial Recruitment
37S
formulated and put in place before a trial or other study is undertaken [5]. The recruitment plan should take account of recruitment sources that are to be utilized and recognize that individual recruitment sources provide a differential access to the pool of potential participants and differ in volume, flow rate characteristics, richness, and efficiency. A significant factor to be considered with individual sources is the lag time from initial negotiation with the source until the first screening from that source is processed. Ideally, recruitment should begin with approaches to sources with longer lag times, followed by those with shorter lag times, to ensure that an adequate number of sources will be available for activation at any time. In addition, the flow characteristics of each source must be considered. Sources with high initial flow rates may be largely expended in a very short time after the screening of the individual source is begun. For example, three fourths of the total participants from individual sources of occupational and community screenings may be processed during the first week of screening. Screenings involving these individual sources must be carefully scheduled over the time span of a recruitment campaign to ensure a reasonable constancy of flow. Other sources with a much more sustained flow, such as medical referrals, clinical laboratories, and blood banks, should be activated early in the campaign so that sufficient time is available to process the bulk of participants from these sources. Other factors that must be considered in the planning are the prevalence of the target disorder in the population to be studied, the desired length of the recruitment period, the average lag time to get under way, the time required for completing preentry protocol visits, and the estimated refusal rate for participation. The size and complexity of recruitment campaigns vary considerably, from the small-scale study involving only a very few centers to the large-scale multicenter study. In the multicenter study, planning at both the collaborative and local levels is needed. At the local level, the planning should focus on the mix of sources that the clinical center expects to use. This has important implications for the workload of local staff members, as well as for their ability to meet the schedule. Once the plan has been developed, consideration must be given to the type of local organization needed to carry out the recruitment. Screening goals should be established that are within the present and future capability of the clinical center. Interim goals should be set that recognize the cumulative nature of participant preentry visits and the time requirements of the clinical unit for visits by participants w h o are already entered into the study. REFERENCES
1. The Lipid Research Clinics Program: The Coronary Primary Prevention Trial: Design and Implementation. J Chron Dis 32:609-631, 1979 2. Agras WS, Bradford RI--I, Eds: Recruitment for Clinical Trials: The Lipid Research Clinics Coronary Primary Prevention Trial experience--its implications for future trials. Circulation, Part II, 66(6), 1982 3. The Lipid Research Clinics Program: Participant recruitment to the Coronary Primary Prevention Trial. J Chron Dis 36:451-465, 1983
38S
R.H. Bradford
4. The Lipid Research Clinics Program: Pre-entry characteristics of participants in the Lipid Research Clinics Coronary Primary Prevention Trial. J Chron Dis 36:467-479, 1983 5. Agras WS, Marshall GD, Kraemer HC: Planning recruitment. In: Recruitment for clinical trials: The Lipid Research Clinics Coronary Primary Prevention Trial experience--Its implications for future trials, Agras WS, Bradford RH, Eds. Circulation, Part II, 66(6):IV54-IV58, 1982
APPENDIX: CPPT SITES A N D PERSONNEL, JULY 1976 I. Lipid Research Clinics Baylor College of Medicine, Houston, Texas Principal Investigator: Antonio M. Gotto, M.D. CPPT Director: David O. Taunton, M.D. Recruitment Coordinator: Virginia Elverson University of Cincinnati Medical Center, Cincinnati, Ohio Principal Investigator: Charles J. Glueck, M.D. CPPT Director: Moti Kashyap, M.D. CPPT Co-Director: Ronald Fallat, M.D. Recruitment Coordinator: Susan Kutchinski George Washington University ~k4edical Center, Washington, D.C. Principal Investigator: John LaRosa, M.D. CPPT Director: John LaRosa, M.D. Recruitment Coordinator: Marilyn McKeown, R.N.
University of Iowa Hospitals, Iowa City, Iowa Principal Investigator: Francois Abboud, M.D. Former Principal Investigator: William Connor, M.D. CPPT Director: Helmut Schrott, M.D. Recruitment Coordinator: Lois Rude Johns Hopkins University, Baltimore, Maryland Principal Investigator: Peter O. Kwiterovitch, M.D. CPPT Director: G. William Benedict, M.D. Recruitment Coordinator: Elizabeth J. Bruen University of Minnesota, Minneapolis, Minnesota Principal Investigator: Ivan Frantz, M.D. CPPT Director: Donald Hunninghake, M.D. Recruitment Coordinator: Florine Peterson, R.N. Oklahoma Medical Research Foundation, Oklahoma City, Oklahoma Principal Investigator: Reagan H. Bradford, Ph.D., M.D. CPPT Director: Thomas F. Whayne, M.D., Ph.D. Recruitment Coordinators: Marvin McDearmon Dana Pitts University of California, San Diego, La Jolla, California Principal Investigator: W. Virgil Brown, M.D. CPPT Director: Richard C. Gross, M.D.
Coronary Primary Prevention Trial Recruitment
39S
Co-Director: John Melish, M.D. Recruitment Coordinator: Nancy Rotchstein, R.N. University of Washington, Seattle, Washington Principal Investigator: William R. Hazzard, M.D. CPPT Director: Robert H. Knopp, M.D. Recruitment Coordinator: Kathy Stamm Stanford University, Stanford, California Principal Investigator: John W. Farquhar, M.D. CPPT Director: Michael P. Stern, M.D. Recruitment Coordinator: Cheryl Guerin Universities of Toronto and McMaster, Toronto and Hamilton, Ontario, Canada Principal Investigator: J. Alick Little, M.D. CPPT Director: Randolph Lee, M.D. Co-Director: Josephine Bird, M.D. (St. Michael's Hospital) CPPT Director: George Steiner, M.D. Co-Director: David Stinson, M.D. (Toronto General Hospital) CPPT Director: Maurice Mishkel, M.D. (Hamilton General Hospital) Recruitment Coordinators: Norma Mishkel Judy Wright Washington University School of Medicine, St. Louis, Missouri Principal Investigator: Gustav Schonfeld, M.D. CPPT Director: Joseph L. Witztum, M.D. Recruitment Coordinators: Betsy Boles Ruth Jacobson II. Support Facilities Central Patient Registry and Coordinating Center, Chapel Hill, North Carolina Director: O. Dale Williams, Ph.D. Former Director: James Grizzle, Ph.D. Deputy Program Director: C.E. Davis, Ph.D. CPPT Coordinator: Sandy Irving, M.S. Recruitment and Adherence Consultants, Stanford, California Director: W. Stewart Agras, M.D. Jacqueline Dunbar, Ph.D. Nathan Maccoby, Ph.D. Albert Stunkard, M.D. Steven M. Zifferblatt, Ph.D. III. Program Office Lipid Metabolism Branch, Division of Heart and Vascular Diseases, National Heart, Lung, and Blood Institute, Bethesda, Maryland
40S
R.H. Bradford Chief: Basil M. Rifkind, M.D., F.R.C.P. Former Chief: Robert I. Levy, M.D. CPPT Coordinator: Richard Havlik, M.D. Medical Officer: Alan Seplowitz, M.D. Former Medical Officers: Conrad Blum, M.D. Gail Morrison, M.D. Nutrition Coordinator: Virginia Keating, R.D. Laboratory Coordinator: Kenneth Lippel, Ph.D.