Objective: To document the recruitment of Medicaid and dual-enrolledMedicare beneficiaries with diabetes mellitus into a randomizedclinical trial.
Study Design: Randomized controlled trial with 1-year follow-up.
Methods: A total of 2242 Medicaid or dual-enrolled Medicarebeneficiaries with diabetes residing in King County, Washington,were recruited by direct mail for a clinical trial of diabetes self-caremanagement. Washington State Medicaid program databases wereused to identify the target population who received recruitmentpackets from the program director. Individuals who did not returna participation refusal letter were telephoned to determine studyeligibility. Subjects were screened during a study visit, and writteninformed consent was obtained. Enrolled subjects were randomizedto a self-care intervention group or a usual care group.
Results: Of 2242 recruitment packets sent, we were unable tocontact 40% of the target population, despite the fact that packetswere sent to the same mailing addresses used for monthly Medicaidcheck distributions. The primary recruitment challenges weremissing telephone contact information and a lack of interpretersspeaking needed dialects. Of the 146 subjects enrolled, 71% werenonwhite, 28% were non-English speaking, 69% were women,and the mean age was 59.8 years.
Conclusions: Research in Medicaid and Medicare populationsis possible but requires additional time, energy, and resources. Thefinding that 40% of the Medicaid population could not be contactedsuggests that Medicaid may want to revisit their contact informationprocedures to facilitate case management and otherprograms.
(Am J Manag Care. 2005;11:443-448)
Diabetes mellitus is a major public health problem,accounting for nearly 20% of annual UShealthcare costs.1 Diabetes is more prevalentamong individuals with limited means and in certainethnic minority groups. Among adult Medicaid recipients,17% to 25% have diabetes, as do 24% of the populationdual enrolled in Medicaid and Medicare.2-4 Theoptimal control of chronic diseases, such as diabetes,relies on healthcare providers and their teaching andsupport of patients' self-care management. Unfortunately,this strategy has not been widely adopted.Several randomized controlled trials of diabetes self-managementprograms have been tested in largely affluentwhite older women and have demonstrated reducedhealthcare utilization and costs.5-8 One program, theHealth Enhancement Program, showed an increase inself-care practices for chronic conditions but was conductedin a nondiabetic affluent older population.9,10Several diabetes self-management programs have beenattempted in low-income, minority, and aged populations,but none were conducted in Medicaid or dual-enrolledMedicare populations.11
The objective of our study was to test a diabeteshealth enhancement program in an underserved population,namely, Medicaid and dual-enrolled Medicarebeneficiaries with diabetes. Beneficiaries eligible underMedicaid (eg, the poor or indigent) or dually enrolled inMedicare (eligible for Medicaid and Medicare, such asthe indigent and disabled) have a unique profile that isdifferent from that of typical research participants.This article reports the recruitment challenges andsuccess in targeting a Medicaid and dual-enrolledMedicare population.
Individuals with diabetes enrolled in Medicaid ordually enrolled in Medicaid and Medicare residing inKing County, Washington, during fiscal years 2000 and2001 were recruited. The Washington State InstitutionalReview Board approved this research.
International Classification of Diseases,
Ninth Revision, Clinical Modification
Because Medicaid and Medicare protected healthinformation is confidential, the recruitment process wasdivided into 2 steps, the first under the jurisdiction ofthe Washington State Medical Assistance Administration(MAA) and the second under the clinical trialstaff. The initial eligibility screening and first subjectcontact were done by the MAA. The MAA used WashingtonState databases to identify potential subjects for thetrial. The inclusion criteria were coverage underMedicaid or dual enrollment in Medicaid and Medicarefor at least 1 year, a diagnosis of type 1 or type 2 diabetesfor at least 1 year (≥ 1 inpatient or ≥ 2 outpatientvisits with codes 250.00-250.93) or prescriptions for insulin or oral hyperglycemicmedications),12 aged 40 to 85 years (an agerange with frequent diabetes-related complications),and residence in King County. The top 10% and bottom10% (by reimbursements) were excluded from thepotential pool of participants, as we were unlikely tochange their healthcare costs.
From August 2001 to July 2002, the MAA mailed2242 study recruitment packets to our target population.The recruitment packet included a study invitationletter, a study description flyer, an informedconsent form, and a self-addressed stamped participationrefusal letter. The study invitation letter was signedby the director of the MAA and described the collaborationbetween the MAA and the University of Washingtonfor this clinical trial. The letter also explained thatMedicaid or Medicare benefits would not be affected byvolunteering for the clinical trial. Beneficiaries had 3weeks in which to mail the MAA a letter declining participation.When no refusal letter was received, the MAAreleased contact information to the clinical trial staff.Invalid addresses were checked by the MAA againstanother Medicaid database, the Automated ClientEnrollment System, for more current information.Study materials were resent to the corrected addresses.
The second recruitment step was performed by theclinical trial staff. Once a list of potential participantswas received from the MAA, subjects were telephonedto explain the trial and answer any questions.Beneficiaries with no listed telephone number were senta letter requesting that they telephone the clinical trialstaff with contact information. Searches for invalid ormissing telephone numbers were performed using telephoneand Internet directories. Transportation assistancewas provided, as needed.
Eligibility was confirmed during the first scheduledclinic visit, verifying aged 40 to 85 years, geographicproximity to the trial site, and the subject's establishedrelationship with a fee-for-service primary care provider(to capture complete healthcare use and cost data).Exclusion criteria included kidney disease (serum creatininelevel, >2.5 mg/dL [>221 μmol/L]; or dialysis),other serious illness that would preclude 1-year participation(dementia, terminal illness, or organ transplantation),pregnancy, or inability to give informed consent.Written informed consent was obtained from each participant.A blood sample was taken for serum creatininemeasurement. Participants were then randomized to aself-care intervention group or a usual care group.
Special efforts were made for the non-English-speakingparticipants. The MAA collects data on the primarylanguage spoken and assigns 1 of 34 language codes. Thestudy invitation letter and informed consent form weretranslated into the 15 most common non-English languages.Language-specific recruitment packets were sent.If the MAA list indicated that a subject's primary languagewas not English, an interpreter for the appropriate languagewas present when the clinical trial staff called thebeneficiary about the study and for all trial visits.Ethnicity was defined by the code indicating the primarylanguage spoken, as identified in the MAA database.
At enrollment, among the baseline data collected,participants were asked the following 3 questions toassess their motivation for enrolling in the trial: (1)"What is your worst problem with diabetes healthcareresources?" (2) "What is your biggest problem in controllingyour diabetes?" (3) "Who is chiefly responsiblefor your diabetes care?"
Differences in groups by ethnic status and languagewere analyzed with the χ2 test or Fisher exact test fordichotomous variables. Continuous variables were analyzedusing the test for parametric data or Wilcoxonrank sum test for nonparametric data.
The results of the recruitment process are shown inthe Figure. The MAA mailed 2242 recruitment packetsto the target population. After initial screening, the MAAsent the clinical trial staff 1566 unique names of potentialparticipants. The clinical trial staff attempted to callall participants, but no valid telephone contact informationwas available for 676 individuals. Mailing of lettersrequesting current contact information and repeatedsearches of Medicaid's Automated Client EnrollmentSystem databases were done to search for valid telephonenumbers. Of the initial pool of 2242 potential participants,900 (40%) could not be contacted, 895 (40%)refused participation, 67 (3%) were duplicates, 239 (11%)were ineligible, and 146 (7%) were enrolled in the trial.Reasons for ineligibility included major illness precludingtrial participation (25%), residence outside of KingCounty (12%), lack of current enrollment in a Medicaidfee-for-service program (11%), absence of a primary carephysician (10%), end-stage renal disease (8%), severemental illness (7%), death (7%), age outside of the range(6%), denial of having diabetes (5%), and others (9%).
Study participants were drawn from a target populationrepresenting many ethnic groups. Of the 1566Recruiting Beneficiaries With Diabetes Into an RCTpotential participants, 1019 (65%) were English speakingand 547 (35%) were non-English speaking. Therewere 29 of the 34 possible non-English-speaking groupsrepresented in our sample that required interpreters.The MAA could not supply interpreters for all the differentlanguages, resulting in 157 individuals who couldnot be contacted because of unavailable interpreters,as well as 42 whose primary language code was "other."As shown in Table 1, there were significant differencesin recruitment by the major primary language groups(χ26= 21.1, = .002). Significantly fewer Asians (16%)enrolled in our trial. Enrolled Medicaid participantswere more ethnically diverse than the dual-enrolledMedicare group. Significantly more Medicaid participantsdid not speak English (36%) compared with thedual-enrolled Medicare group (18%) (χ21 = 10.5, = .01).
Of the 146 participants who enrolled in the trial, 101(69%) were women, the mean ± SD age was 59.8 ± 9.9years, and 41 (28%) did not speak English. From self-reportedrace data collected during enrollment interviews,29% were white, 28% African American, 19% Asian,8% Native American, and 17% "other." Of the 146 participants,85 (58%) were enrolled in Medicaid only, and 61(42%) were dually enrolled in Medicaid and Medicare.
We used baseline data to analyze the motivations forenrolling in our randomized clinical trial (Table 2).Major enrollment motivators included perceivedincreased access to healthcare resources, additional diabetesself-care skills instruction, and diabetes educationalmaterial in the participant's native language.Overall, 37% of the participants reported that their primaryproblem with theirdiabetes control was lackof time with their healthcareproviders. Subjectsbelieved that, by enrollingin a clinical trial, theywould receive more individualizedhealthcare anddiabetes education fromthe trial nurses. Onlythose randomized to theself-care interventiongroup would be offeredparticipation in weeklysupport groups, additionalclinic visits with the studynurse or social worker,and increased telephonevisits.
For the non-English-speakingsubjects, the most common problems werelack of diabetes educational material in their language,difficulties with their self-care, family obligations, andlack of money. Because the intervention arm would beprovided with food pyramids, diet information, serving-sizecalculations, carbohydrate and calorie contents ofvarious cultural foods in the participant's appropriatelanguage, and interpreters during all trial visits, accessto these resources was a major motivator to enroll.
This trial focused on developing an individual's diabetesself-care skills. Therefore, the participants wereasked who they thought had chief responsibility fortheir diabetes care. Most English-speaking subjects(74%) believed that they were responsible for their self-care.Asians indicated that fate or God (45%, <.001)was chiefly responsible for their diabetes care.Europeans placed the responsibility more frequently ontheir physicians (69%, <.001) compared with theEnglish-speaking group (17%).
Recruitment is a challenge for most clinical trials.The most striking recruitment finding in our trial isthat, despite mailings and repeated telephone calls toMedicaid beneficiaries receiving monthly financial supportfrom the MAA, we were unable to make contactwith 900 individuals (40%). Addresses for Medicaid beneficiarieschanged frequently and necessitated repeatedsearches for valid data. Medicaid does not require a telephonenumber as contact information, and manyMedicaid beneficiaries did not provide this information.
We successfully enrolled an ethnicallydiverse group of beneficiaries with diabetes. KingCounty, which includes the city of Seattle, compriseswhite (75.7%), Asian (10.8%), Hispanic (5.5%), African American(5.4%), Native American (0.9%), and Pacific Islander (0.5%) ethnicgroups.13 Ethnicity has been found to influence healthcare utilization andinvolves differences in effective approaches for recruitment and participationin clinical trials.14-16 In our trial, interpreter services were limited tomajor language groups, so some ethnic groups were excluded from screening.More than one third (35%) of the potential subjects who were recruitedcould not speak English, even though they had been in the UnitedStates for at least 1 year. Interpreters were necessaryfor the initial telephone contact, because non-English-speakingpersons typically hung up if an English-speakingperson called without an interpreter to explain thepurpose of the call. Enrollment of ethnically diversesubjects requires additional time during trial contacts(visits and telephone calls) when interpreters are used.
The reasons why subjects enroll in clinical trials areimportant to consider when designing an effectiverecruitment strategy. In our trial, motivations to enrollincluded perceived increased exposure to healthcareproviders and diabetes educational material, along withthe opportunity for individualized diabetes care plans.For non-English-speaking subjects, the availability of aninterpreter and language-appropriate educational materialwas appealing. Reports of clinical trials rarely providereasons why subjects enroll in the trials. One trialexamined the reasons why older people participate inclinical trials and found that anticipated personal benefitsand ongoing health monitoring were the strongestmotivators.17
The strengths of our study include the collaborationof multiple agencies, a balance of Medicaid and dual-enrolledMedicare beneficiaries, and an ethnicallydiverse study population from one defined geographicarea. We were able to document the target population,potential participants, eligible participants, and enrolledparticipants, as recommended in the ConsolidatedStandards of Reporting Trials.18 Of subjects who enrollin clinical trials, most are typically young well-educatedwhite women.19,20 One study19 using mail and telephoneinterventions for weight control attracted mostly well-educatedyounger women. Typical diabetes self-managementprograms have recruited largely affluent, whitesubjects into their studies.5-10 Although our enrollmentrate of 7% seems low, studies21-24 using mass mailings forrecruitment typically only enroll participants from 1%to 7% of the letters sent.
The major recruitment limitation was the inability tocontact 40% of our potential participants. Therefore,how representative our trial population is of the totalWashington State Medicaid and dual-enrolled Medicarepopulation with diabetes is not known. Of the 2 majorrecruitment challenges in our study, the lack of contactinformation reflected the Medicaid population as awhole, and the lack of an available interpreter reflectedRecruiting Beneficiaries With Diabetes Into an RCTthe ethnic diversity in our Medicaid and dual-enrolledMedicare population. As already noted, Medicaid doesnot require a telephone number as part of the contactinformation for beneficiaries. Many clients may not beable to afford telephone service, or they may have nonpublishedtelephone numbers or cell phone service. IfMedicaid required a telephone number and 2 contactpersons (family members or friends with telephonenumbers), the likelihood of maintaining contact withthese individuals would be higher. Medicaid may wantto revisit their contact procedures to facilitate casemanagement and other programs. Two beneficial programs(transportation assistance and interpreters) thatMedicaid provides were useful for recruiting subjectsinto our clinical trial. Transportation assistance wasused by 67% of the enrolled subjects, and interpreterswere required for 31%. Washington is one of the fewstates where Medicaid covers interpreters to be presentduring medical clinic visits. In states where theMedicaid population is ethnically diverse, similar programswould be useful.
For studies recruiting Medicaid or dual-enrolledMedicare participants, contact information may be lackingfor many potential participants, and extra resourcesmay be required, such as study staff time and transportationassistance. If the study will be drawing froman ethnically diverse population, specially adaptedmaterials and interpreters may be needed. Our studyfound that, despite significant challenges, it is possibleto recruit Medicaid and Medicare participants into aclinical trial.
We thank Nancy Anderson, MD, for her assistance with Medicaidrecruitment, South East Senior Services, study staff interpreters, andparticipants of the trial.
From the Department of Health Services Research and Development, VA Puget SoundHealth Care System (LVM, GER), and Department of Health Services, University ofWashington (GER), Seattle; and Diabetes Prevention and Control Program, WashingtonState Department of Health, Olympia (JEN).
This study was funded by the Sandy MacColl Foundation, Seattle, Wa, and The RobertWood Johnson Foundation, Princeton, NJ; Aventis Pharmaceuticals Inc, Bridgewater, NJ;Washington State Department of Health and Medical Assistance Administration; and VAPuget Sound Health Care System.
Address correspondence to: Lynne V. McFarland, PhD, Department of Health ServicesResearch and Development, VA Puget Sound Health Care System, Metropolitan Park West,1100 Olive Way, Suite 1400, Seattle WA 98101. E-mail: firstname.lastname@example.org.
1. Hogan P, Dall T, Nikolov P; American Diabetes Association. Economic costs ofdiabetes in the US in 2002. 2003;26:917-932.
2. Karter AJ, Ferrara A, Liu JY, Moffet HH, Ackerson LM, Selby JV. Ethnic disparitiesin diabetic complications in an insured population. 2002;287:2519-2527.
3. Riste L, Khan F, Cruickshank K. High prevalence of type 2 diabetes in all ethnicgroups, including Europeans, in a British inner city: relative poverty, history,inactivity, or 21st century Europe? 2001;24:1377-1383.
Health Aff (Millwood).
4. Foote SM. Population-based disease management under fee-for-service Medicare.2003;Jul-Dec(suppl Web exclusives):W3-342-56.
5. Norris SL, Engelgau MM, Narayan KMV. Effectiveness of self-management trainingin type 2 diabetes. 2001;24:561-587.
6. Renders CM, Valk GD, Griffin SJ, Wagner EH, Eijk VJT, Assendelft WJ.Interventions to improve the management of diabetes in primary care, outpatient,and community settings: a systematic review. 2001;24:1821-1833.
7. Brown SA. Interventions to promote diabetes self-management: state of the science.1999;25(suppl):52-61.
Patient Educ Couns.
8. Glasgow RE, Toobert DJ, Hampson SE, Brown JE, Lewinsohn PM, Donnelly J.Improving self-care among older subjects with type II diabetes: the "SixtySomething?" Study. 1992;19:61-74.
J Am Geriatr Soc.
9. Leveille SG, Wagner EH, Davis C, et al. Preventing disability and managingchronic illness in frail older adults: a randomized trial of a community-based partnershipwith primary care. 1998;46:1191-1198.
J Am Geriatr Soc.
10. Phelan EA, Williams B, Leveille S, Snyder S, Wagner EH, LoGerfo JP.Outcomes of a community-based dissemination of the Health EnhancementProgram. 2002;50:1519-1524.
Diabetes Metab Res Rev.
11. Eakin EG, Bull SS, Glasgow RE, Mason M. Reaching those most in need: areview of diabetes self-management interventions in disadvantaged populations.2002;18:26-35.
Am J Med Qual.
12. Hebert PL, Geiss LS, Tierney EF, Engelgau MM, Yawn BP, McBean AM.Identifying persons with diabetes using Medicare claims data. 1999;14:270-277.
13. US Bureau of the Census. King County census 2000 results [King CountyBudget Office Web site]. Available at: http://www.metrokc.gov/budget/census 00.Accessed March 30, 2005.
14. Fiscella K, Franks P, Doescher MP, Saver BG. Disparities in health care byrace, ethnicity, and language among the insured: findings from a national sample.2002;40:52-59.
15. Fitzgerald JT, Gruppen LD, Anderson RM, et al. The influence of treatmentmodality and ethnicity on attitudes in type 2 diabetes. 2000;23:313-318.
16. Murray-Garcia JL, Selby JV, Schmittdiel J, Grumbach K, Quesenberry CP Jr.Racial and ethnic differences in a patient survey: patients' values, ratings, andreports regarding physician primary care performance in a large health maintenanceorganization. 2000;38:300-310.
17. Tolmie EP, Mugall MM, Louden G, Lindsay GM, Gaw A. Understanding whyolder people participate in clinical trials: the experience of the Scottish PROSPERparticipants. 2004;33:374-378.
18. Gross CP, Mallory R, Heiat A, Krumholz HM. Reporting the recruitmentprocess in clinical trials: who are these patients and how did they get there? 2002;137:10-16.
Am J Manag Care.
19. Jeffery RW, McGuire MT, Brelje K, et al. Recruitment to mail and telephoneinterventions for obesity in a managed care environment: the Weigh-To-Be Project.2004;10:378-382.
J Clin Oncol.
20. Wei SJ, Metz JM, Coyle C, et al. Recruitment of patients into an Internet-basedclinical trials database: the experience of OncoLink and the National ColorectalCancer Research Alliance. 2004;22:4678-4684.
21. Whelton PK, Lee JY, Kusek JW, et al. Recruitment experience in the AfricanAmerican Study of Kidney Disease and Hypertension (AASK) Pilot Study. 1996;17(suppl):17S-33S.
22. Fujimoto WY; Diabetes Prevention Program Research Group. Backgroundand recruitment data for the US Diabetes Prevention Program [published correctionappears in 2000;23:876]. 2000;23(suppl 2):B11-B13.
23. Cosgrove N, Borhani NO, Bailey G, et al; Systolic Hypertension in theElderly Program Cooperative Research Group. Mass mailing and staff experiencein a total recruitment program for a clinical trial: the SHEP experience. 1999;20:133-148.
24. Sung NS, Crowley WF, Genel M, et al. Central challenges facing the NationalClinical Research Enterprise. 2003;289:1278-1287.