Two Recruitment Strategies for a Clinical Trial of Physically Frail Community‐Living Older Persons

To compare the efficiency and cost of two distinct but complementary recruitment strategies for a clinical trial of physically frail, community-living persons, age 75 and older. In the first recruitment strategy, potential participants were identified and screened for physical frailty during office visits to their primary care physicians; in the second, potential participants were identified from the patient rosters of primary care physicians and were screened for physical frailty in their home. Physical frailty was defined on the basis of slow gait speed and inability to stand from a chair with one's arms folded. General community in greater Bridgeport, Connecticut. Community-living persons, age 75 and older, who met criteria for physical frailty. Measures of efficiency included the number of persons screened for each participant randomized, the number of persons eligible for each participant randomized, and the refusal rate. Costs were calculated per randomized participant. One hundred eighty-eight participants, with a mean age of 83. 2 years, were enrolled over 22 months. One hundred and one participants were enrolled via the office-based strategy; 87 were enrolled via the roster-based strategy. Participants in each group had considerable quadriceps weakness and performed poorly on objective measures of gait and upper- and lower-extremity function. For each participant randomized, the number of persons screened, number of persons eligible, and cost were 15. 2, 1. 2, and 868, respectively, for the office-based strategy and 11. 6, 1. 1, and 764, respectively, for the roster-based strategy. The corresponding refusal rates for the two strategies were 14. 9% and 10. 1%, respectively (P <. 001). Although each recruitment strategy successfully identified older persons who were physically frail, the roster-based strategy was less expensive and performed modestly better on each measure of efficiency than the office-based strategy. J Am Geriatr Soc 49: 1039–1045, 2001. The recruitment of older persons, particularly those who are physically frail, into clinical trials is difficult and time consuming, 1–3 but there is a compelling need to develop and test interventions to prevent functional decline and disability among frail older persons. 4 We recently completed enrollment for a clinical trial designed to test the efficacy and cost-effectiveness of a home-based physical therapy intervention to prevent functional decline among a heterogeneous group of physically frail, community-living older persons. In this paper, we compare the efficiency and cost of two distinct but complementary recruitment strategies. The key element of each strategy was the use of two standardized tests of physical performance to identify older persons who are physically frail. 5 In the first strategy, potential participants were identified and screened for physical frailty during office visits to their primary care physicians; in the second, potential participants were identified from the patient rosters of primary care physicians and were screened for physical frailty in their home. By identifying potential participants from busy primary care practices, we hoped to enhance recruitment and retention and to facilitate implementation of our preventive intervention. The eligibility criteria for our clinical trial are provided in Table 1. These criteria were designed to be inclusive rather than exclusive to ensure that the results of the trial would be applicable to the broadest possible group of physically frail, but nondemented, community-living older persons. Physical frailty was determined by two tests of physical performance that have been shown in previous investigations to be most predictive of new disability in basic activities of daily living (ADLs) 1 year later. 5, 6 The tests included rapid gait (walking back and forth over a 10-foot course as quickly as possible) and a single chair stand (standing up from a hard-back chair with arms folded). Persons were considered to be physically frail if they scored greater than 10 seconds on the rapid gait test or could not stand from the chair with their arms folded. Poor performance on rapid gait and chair stands generally signifies impairments in lower-extremity strength and balance. 7, 8 Persons who were nonambulatory were considered to be too debilitated to participate in or benefit from our preventive intervention. Persons who had had a stroke, hip fracture, or hip or knee replacement within the preceding 6 months were temporarily excluded because they would likely still be recovering from their acute condition or surgery. 9–11 Because the physical therapy intervention included upper-extremity resistance training, persons who had had a myocardial infarction within the past 6 months were also temporarily excluded (for safety reasons). Eligibility was established using a step-wise process that was tailored to the specific recruitment strategy. In the office-based strategy, a project member screened potential participants during nonurgent (regularly scheduled, not acute) clinic visits at four different primary care practices in greater Bridgeport, Connecticut. To enhance efficiency, screening at these sites occurred during the days (or half days) when the largest number of age-eligible persons was generally scheduled to be seen. Before the visit, medical records were reviewed to determine age eligibility and to check for obvious exclusion criteria, (e. g. , nursing home resident, diag-nosis of dementia). To avoid unnecessary screening, a log was kept of age-eligible persons who were already enrolled in the study, permanently excluded, or temporarily ineligible based on the criteria in Table 1 or found not to be physically frail during an office visit within the previous 6 months. After verbal consent was obtained, physical frailty was assessed (as described above) in those who were still eligible. Persons who fulfilled criteria for physical frailty were sent a letter (cosigned by their primary care physician) and an informational brochure that described the study. During a subsequent telephone interview, a research nurse reviewed the purpose of the study, confirmed potential eligibility, and scheduled a home visit. Persons who scored 20 or better on the Folstein Mini-Mental State Examination (MMSE) 12 during this visit were invited to participate in the study. Those who agreed had a comprehensive baseline assessment and were subsequently randomized to either the physical therapy intervention or an educational control group. In the roster-based strategy, potential participants were identified from computerized patient lists provided by two community health centers and four primary care physicians. The previously described, step-wise process was subsequently followed except that physical frailty was assessed in the home (along with cognitive status) after potential eligibility had been established during the telephone interview. Persons who were temporarily ineligible based on the telephone interview were reevaluated at the end of their exclusion period. Our goal was to enroll 180 participants over a 2-year period. The roster-based strategy was originally designed to enhance the enrollment of minorities (from the two community health centers) but was later expanded when it became apparent that we would not otherwise meet our overall recruitment goal. In addition to demographic characteristics, physical frailty, and cognitive status, self-reported information was collected at baseline on 10 physician-diagnosed chronic conditions and on eight basic ADLs: walking inside the house, bathing, upper- and lower-body dressing, transferring from bed to chair, toileting, feeding, and grooming. 13, 14 For each ADL task, participants were asked two separate questions. (1) Within the last month, did you need help from another person to complete the task? (2) Within the last month, have you had difficulty with the task? Based on the results of an earlier study, 15 each task was scored as “0” for no help and no difficulty, “1” for difficulty but no help, and “2” for help regardless of difficulty; a summary score for ADL disability was created with a range of 0 to 16. As part of the physical assessment, maximal strength (in kilograms) of the extensor muscles of the nondominant knee was measured with a hand-held dynamometer (Chatillon, Model No. MSE 100, Greensboro, NC), using a standard protocol. 16 After a demonstration, two formal tests were performed. The best results were used in the analysis. Physical performance was assessed with three composite measures of gait, upper-extremity function, and lower-extremity function. These were, respectively, (1) the gait component of the Performance-Oriented Mobility Assessment (POMA), which included five items—step continuity and symmetry, path deviation, turning, and missed steps—and was scored from 0 to 8 (best score) 17; (2) a shortened version of the Physical Performance Test (PPT), which included three items of primarily upper-extremity function—putting on and removing a jacket, lifting a book and putting it on a shelf, and picking up a pencil from the ground—and was scored from 0 to 12 (best score) 18; and (3) a modified version of the Established Population for Epidemiologic Studies of the Elderly (EPESE) battery, 19, 20 which included the three standard tasks of standing balance, but used three timed chair stands instead of five and substituted timed rapid gait over 20 feet for timed usual gait over 8 feet. The quartile values for these two timed tests were determined from an earlier population-based study, 5 and a summary EPESE score was created with a range of 0 to 12 (best score). The Human Investigation Committee at the Yale University School of Medicine approved the research protocol. Because the telephone interviews and screening for physical frailty in the office presented no more than minimal risk of harm, verbal consent was permitted for these procedures. During the home visit, informed consent was obtained before the collection of baseline data. To estimate costs, we first itemized each of the specific tasks for the two recruitment strategies, from the development of the assessment forms to the confirmation of study eligibility and enrollment. We considered the costs of personnel and nonpersonnel, which included equipment, supplies, travel, postage, and telephone charges. For each task, we calculated unit costs for personnel, based on the average amount of time required to complete the task and an hourly rate that included salary and benefits, and for nonpersonnel (where applicable). For personnel we included the time required to develop and manage the database and tracking system; review and “clean” the computerized patient lists (e. g. , remove patients who had died or were living in nursing homes, locate patients who had moved) ; generate telephone call records and contact sheets; edit, key (using double entry), and file all data forms; travel to/from office sites and participants' homes; prepare correspondence; and complete assessments to determine eligibility (clinic, phone, and home). For tasks that were common to the two recruitment strategies (e. g. , development of the informational brochure and telephone-screening instrument), we divided the costs equally. Equipment included three personal digital assistants (PDA), which were used to facilitate the clinic screening. Supplies included letterhead, envelopes, brochures, assessment instruments, and other forms. We did not include the cost of software, nor the time required to develop our protocols, establish the office sites, acquire the computerized patient lists, pilot-test our procedures, or complete the baseline assessments. To estimate the total cost of each strategy, we summed the product of the unit-specific costs and the number of units used. A detailed breakdown of costs is available upon request. The baseline characteristics of participants enrolled via the office-based strategy were compared with those of participants enrolled via the roster-based strategy. T-tests were used for continuous variables; the chi-square statistic was used for dichotomous variables. The efficiency and cost of the two recruitment strategies were then calculated. Efficiency was assessed with three standard measures: the number of persons screened for each participant randomized, the number of persons eligible for each participant randomized, and the refusal rate. Statistical comparisons for efficiency were performed using the chi-square statistic. For each recruitment strategy, costs were calculated per randomized participant by dividing the total cost by the number of participants randomized. Statistical comparisons for costs were not possible because measures of variability were unavailable for these aggregate data. The analyses were performed using SAS version 6. 12. 21 All statistical tests were two-sided. One hundred eighty-eight participants were enrolled over 22 months: from October 1998 through July 2000. Figure 1 summarizes the recruitment of the 101 participants who were enrolled using the office-based strategy. During 300 half-day office sessions, 3, 449 age-eligible persons had scheduled appointments and 1, 519 were screened for potential eligibility. Two hundred and twenty five persons were ineligible for the reasons listed. Of the 1, 294 persons who were potentially eligible, 276 (21. 3%) met criteria for physical frailty; 127 (9. 8%) refused the frailty screen. Ninety-three and a half percent and 64. 7% of potential participants completed the telephone screen and home visit, respectively. The 101 persons who agreed to enroll in the study represented 6. 6% of those who were screened in the office, 36. 6% of those who met criteria for physical frailty, 53. 2% of those who were eligible for a home visit, and 85. 6% of those who fulfilled all of the eligibility criteria. The overall refusal rate for this step-wise recruitment process was 227/ 1, 519 (14. 9%). Recruitment of participants using the office-based strategy. Within parentheses are the number of persons excluded for the reasons specified. MMSE = Mini-Mental State Examination. Figure 2 summarizes the recruitment of the 87 participants who were enrolled using the roster-based strategy. Based on the telephone screen, 247 (24. 1%) of the 1, 026 nondecedents who could be contacted were potentially eligible. Of the 166 potential participants who agreed to a home visit, 128 (77. 1%) met criteria for physical frailty, were ambulatory, and did not have a terminal illness. Of these, 30 (23. 4%) were ineligible based on an MMSE score less than 20. The 87 persons who agreed to enroll in the study represented 8. 6% of those who completed the telephone screen, 35. 2% of those who were eligible for a home visit, and 88. 8% of those who fulfilled all of the eligibility criteria. The overall refusal rate for this step-wise recruitment process was 104 / 1, 026 (10. 1%). Recruitment of participants using the roster-based strategy. Within parentheses are the number of persons excluded for the reasons specified. MMSE = Mini-Mental Status Examination. The baseline characteristics of the study participants are provided in Table 2. The mean age was 83. 2 years, with a range of 75 to 97. The majority of participants were female and white; about half lived alone. On average, participants had an 11th grade education, had more than two chronic conditions, and had disability in at least two ADLs. Although the majority of participants met only one of the two criteria for physical frailty (i. e. , moderate), participants, on average, had considerable quadriceps weakness and performed poorly on objective measures of gait and upper- and lower-extremity function. Demographically, participants did not differ significantly based on recruitment strategy. However, participants who were recruited via the roster-based strategy had greater disability, were more likely to meet both criteria for physical frailty, and performed more poorly on the gait component of the POMA than participants who were recruited via the office-based strategy. The efficiency and costs of the two recruitment strategies are summarized in Table 3. Compared with the office-based strategy, the roster-based strategy was less expensive (by about 14%) and performed modestly better on each measure of efficiency. The goal of our clinical trial is to determine whether a home-based, physical therapy intervention is superior to an educational control intervention in preventing functional decline among physically frail, community-living older persons. Our criteria for physical frailty identified older persons with substantial quadriceps weakness and impairments in gait and upper- and lower-extremity function, which are the main targets of our physical therapy intervention. The average knee extension strength of our participants, which included 20% men, was at the 50th percentile for age-comparable participants of the Women's Health and Aging Study, a population-based study of the one-third most disabled older women living in the community. 22 Each of our two recruitment strategies successfully identified older persons who were physically frail. Although both were generally well accepted, the roster-based strategy was less expensive than the office-based strategy and was modestly better in terms of efficiency. These differences in cost and efficiency cannot be readily explained by differences in baseline characteristics because participants enrolled via the roster-based strategy were frailer on average than those enrolled via the office-based strategy. To minimize potential bias in our clinical trial, participants were randomized within strata defined by the severity of physical frailty and recruitment strategy. One potential advantage of the office-based strategy is that physical frailty, our primary eligibility criterion, can be established in the context of routine primary care. We successfully implemented our frailty screen in four busy, community-based office practices. Although our research staff (rather than the office staff) performed the assessment, which takes less than a minute, fewer than 10% of the age-eligible patients declined to be screened. These results suggest that the ultimate target population of older persons with physical frailty could be readily identified should our physical therapy intervention prove to be efficacious and cost-effective. Although some institutions may not permit verbal consent for screening in the current research climate, 23 assessments of basic mobility such as ours are considered an essential part of routine geriatric practice. 24, 25 Few previous trials have used an office-based strategy to identify frail older persons. Among the five community-based Frailty and Injuries: Cooperative Studies of Intervention Techniques (FICSIT) trials, three identified participants from a health maintenance organization database, 26–28 one used a voter registration list, 29 and another relied primarily on media announcements, community presentations, and advertisements. 30 Other trials of frail older persons, such as the Strong for Life Program, 31 have identified participants through a combination of mailings, self-referrals, and referrals from community agencies, professionals, senior centers, and senior housing sites. 31 The cost differential between our two recruitment strategies was due largely to the expense of having our research staff stationed at the different office sites. The efficiency and cost of this strategy depend on the volume and flow of age-eligible patients. On the busiest days, some patients left the clinic before they could be screened. On other days, only a few patients were screened because of cancellations and “no shows. ” The office-based strategy also tended to become less efficient over time as the “prevalent” cases of physical frailty were identified (data not shown). To enhance efficiency, we did not screen persons more frequently than every 6 months, and we did not rescreen persons who had previously met one or more of our permanent exclusion criteria. Although our estimates of absolute costs will likely differ in other settings, based on institution-specific factors such as salary levels, the relative costs of the two strategies should be comparable across settings. In our calculations, we chose not to include our development costs because they primarily tasks such as and development of the protocols, for and review and with that are difficult to and because we that these costs were comparable for the two recruitment strategies. Compared to the office-based strategy, the roster-based strategy provided with a more flow of potential Each we determined the number of to based on the telephone of persons who were potentially eligible based on the telephone which was and the number of potentially eligible participants who were a home visit. The rate of physical frailty among persons who had a home visit the efficiency and the cost of the roster-based strategy. We have previously shown that one of the asked during the telephone interview the past month, have you 1 / 2 or for 30 can older persons who are physically frail from those who are For each of our recruitment strategies, the overall refusal rate was Because our included older patients followed in primary care practices, we were to the of personal physicians. Our for were by the personal physicians and were sent on their office of our participants, with their before to enroll in the study. Our that the recruitment of frail older persons into clinical trials can be by the of primary care physicians. Based on current our target population of physically frail persons age 75 and older is to over the As interventions are they will need to be in randomized clinical The results of the current study suggest that our office-based and roster-based recruitment strategies can each be used to successfully identify and enroll older persons who are physically frail into randomized clinical We for review of an earlier of this and for development of the participant database and tracking system; for with data and and for with data and We also and the other physicians and staff of and the staff of and at and at the Health for our recruitment by from the from the on the and the for is a and the of from the on during the course of this study, was also a in Aging