AHRQ DESIGN AN EVALUATION OF THREE AOA PROGRAMS CHRONIC DISEASE SELF MANAGEMENT PROGRAM EVALUATION DESIGN CONTRACT # HHSA290200710071T CDSMP Evaluatin Design Technical Expert Panel (TEP) Cnference Call DATE: March 30, 2011 TIME: 2pm 4pm PLACE: Cnference call, 1 800 977 8002, Passcde 54253148# PARTICIPANTS: AHRQ: William Spectr, Prject Officer AA: Jennifer Klcinski, Aging Services Prgram Specialist Saadia Greenberg, Directr, Office f Perfrmance and Evaluatin Jane Tilly, Health/Dementia Team Lead Susan Jenkins, Scial Science Analyst, Office f Perfrmance and Evaluatin IMPAQ Internatinal: Daver Kahveciglu, Prject Directr and C PI Maris Michaelides, Research Assciate Sarah Ruiz, Research Assciate Leanne Clark, Gerntlgy Research Cnsultant Carlyn Crea, Research Analyst Abt Assciates: Terry Mre, C PI Rsanna Bertrand, Assciate Jhanna Hudgens Technical Expert Panel Members: Lynda Andersn, PhD Candace Gehring, MN RN Diana Scully, MSW Directr, Healthy Aging Prgrams, CDC Atlanta, Gergia Unit Manager, Hme and Cmmunity Prgrams, Aging and Disability Services Administratin Washingtn Directr, Office f Elder Services Maine Page 2 f 6
DISCUSSION: Opening Remarks by Administratin n Aging OMB and Cngress are very interested in hw CDSMP has actually wrked in the aging netwrk. OMB expects an experimental evaluatin. Despite a lt f research dne n CDSMPs, there are sme gaps in areas f: 1) the effect f the prgram n lder adults, 2) the effect f the prgram as it is implemented by the aging netwrk, 3) the effect f the prgram n healthcare utilizatin and cst. TEP Member Intrductins and Prject Overview by Daver Kahveciglu Daver Kahveciglu, frm IMPAQ internatinal, gave a brief verview f effrts t date and described where this call fits in the evaluatin design prject. Daver requested that the grup fcuses feedback n the methdlgy and feasibility f implementing the rigrus design. Discussin f the Design Use f Medicaid data Even thugh it des nt seem feasible t cnduct at this pint, it wuld be f great interest t states t assess the effects f CDSMP n the Medicaid ppulatin (r the dually eligible ppulatin). There are a number f new Medicaid initiatives funded by Affrdable Care Act that ptentially incrprate CDSMPs: Health Hmes (Sectin 2703), Medicaid Incentives fr Preventin f Chrnic Diseases Prgram (Sectin 4108), and State Demnstratins t Integrate Care fr Dual Eligible Individuals (Sectin 2602). The TEP members agreed that there wuld be great interest in making the case that the CDSMP prgram will save Medicare and Medicaid dllars and that CDSMP culd be integrated int nging prgrams. Due t lng delays in Medicaid data becming available thrugh CMS, it des nt seem feasible t use Medicaid data frm MAX r MSIS. Hwever, States d have this data current, but the challenges are getting individual states permissin and cmmitment f state resurces t pull the data and als creating a data system that is cmparable acrss states. A TEP member suggested that if the states were shwn the value f sharing the data (i.e., having an estimate f ptential Medicaid savings unique t State), they may be mre willing t prvide it. Results f any cst savings may als lead t Medicare funding f CDSMP. Furthermre, sme states are cnsidering integrating CDSMP in their Medicaid prgrams. Anther Medicaid relevant variable that wuld be interesting t explre is diversin frm (1) Medicaid and/r (2) nursing hmes. Insurance status (dually eligibles) culd be tracked fr this purpse. In light f difficulty t btain and use Medicaid data, there may be an indirect methd f assessing impact n Medicaid utcmes using Medicare utcme data. Sme Medicare utcmes culd be similar and relevant t Medicaid. Page 3 f 6
If significant number f participants turn ut t be Medicaid, then maybe we can later pursue Medicaid data. The mre buy in frm the States, the better; we shuld be careful that this des nt appear t be an impsitin frm AA. Strategies t buy in shuld be cnsidered in the evaluatin design. Fllw up Perid Recmmendatins The reprt currently recmmends a 6 mnth perid where the cntrl grup receives n CDSMP services. The TEP agreed that 6 mnths seems t be a reasnable amunt f time t delay access t CDSMP services. The TEP members stated that they thught 12 mnths may be t lng, and may pse ethical issues fr denying the prgram fr lnger than necessary. If a site receives a lt f participants frm dctr referrals, the pssibility f being placed in a cntrl grup may nt be appealing. Hwever, physicians may nt mind if their patients are guaranteed a spt in 6 mnths. A TEP member suggested identifying the prs and cns f using a 6 mnth cntrl versus a 12 mnth cntrl frm the literature. It shuld be determined if there are benefits t keeping the cntrl at 6 mnths (e.g., attritin). Participant Exclusin Factrs Exclusin criteria shuld be detailed in the reprt. TEP members identified the fllwing ptential exclusin criteria: participatin in similar prgrams, prxies/caregiver participants, and cgnitive screening. If using the riginal 24 grantees, the evaluatr shuld avid recruiting repeat enrllees; they shuld recruit individuals wh are new t the prgram. NCOA cllects repeat enrllee data. They are aware that sites allw participants t repeat the wrkshp 2 r 3 times. At the natinal level, 2% f participants are repeaters. The evaluatr shuld make a pint t track ther curses participants may have taken. Using apprpriate exclusin criteria will help t avid criticism f the utcmes. TEP members reprted that family caregivers, wh may be 60+, attend CDSMP wrkshps, and shuld be taken int cnsideratin. CDSMP is nt designed fr peple with dementia. Given that a high percentage f lder adults in nursing hmes have dementia and a high percentage f flks in assisted living d as well, we need t be careful abut inclusin f participants frm nursing hmes r assisted living. This may be an issue that needs t be addressed fr each state. The TEP members agreed that CDSMP prgrams generally fcus n ppulatins that d nt need extensive, lng term services. The percent f participants frm lng term care institutins shuld be identified, especially if the CDSM prgram is targeting these ppulatins. Page 4 f 6
Sample Expansin Currently the reprt recmmends an evaluatin f the riginal 24 grantees. If using the riginal 24 grantees, the evaluatr shuld avid recruiting repeat enrllees. Inclusin f Dual Eligibles We may want t recmmend that sites identify dual eligibles as part f intake prcess and suggest an ptinal task and analysis. TEP members agreed that dual eligibles are being served by CDSMP prgrams. The evaluatr shuld identify dual eligibles at intake and track them. Outcmes shuld be identified that wuld depend n the use f Medicaid data. The participant survey shuld include questins abut what services the participants receive. These questins shuld include generic descriptins f service categries. This will help t determine if participants are receiving Medicaid waiver services. Prject Challenge ARRA funding ends in 2012. AA requested that a Plan B, r a cntingency plan be recmmended in the event sites are nt able t enrll a sufficient number f participants. Additinal Cntrl Variables Insurance status f participants shuld be tracked. If we find a significant number f participants are part f the Medicaid ppulatin, we can decide t cllect Medicaid data t determine the effects n Medicaid. There are wrkshps that are prvided in languages ther than English and Spanish. Fr these wrkshps the lay leaders and their master trainers are bilingual and they deliver the wrkshp in anther language using the English curriculum. These wrkshps are typically recrded as generic CDSMPs. The evaluatr may want t determine via the survey, if participants receive Medicaid waiver and/r OAA services. It culd be that thse wh get these types f services actually have better health and better, mre apprpriate use f healthcare services. Ppulatin Characteristics The evaluatin reprt shuld include an analysis f participant characteristics and hw they cmpare t ther ppulatins. TEP members agreed that this analysis wuld be useful fr marketing purpses as well as explaining the variatin in utcmes. Asking questins abut ther Older Americans Act, Medicaid waiver services, r ther lcal aging netwrk services used is a relatively simple but interesting additin t data cllectin. Outcme Measures The gal f achieving cst neutrality may be a apprpriate in determining the percent reductin in Medicare spending that wuld cnstitute a significant enugh benefit. Page 5 f 6
The TEP members suggested lking int the average cst ne hspitalizatin and/r ER visit. Average csts may cntribute t the develpment f target cst reductins. Dr. Lrig included a self reprted cst savings f $500 in her research. The recmmended sample size wuld nt detect this level f savings. One pssible ptin wuld be t acquire CMS data fr the regins where the evaluatin is ccurring t get an average estimate f csts fr the types f peple participating in CDSMPs. The peple wh are likely t participate in a CDSMP have chrnic cnditins. Their cst f care will be higher than average s it is mre likely that reductins in cst wuld be fund in the cst analysis. Overall csts (nt including medicatins/part D) are recmmended t be included in the evaluatin. The desired/required reductin in csts may vary based n the different utcmes. Inclusin f Diverse CDSMP Sites TEP members felt strngly that bth rural and Tmand sites shuld be included in the evaluatin, t reflect diversity in the CDSMP prgram. TEP members nted that there are n majr differences between the CDSMP and Tmand wrkshps. The main difference is that the Tmand prgram includes a lnger nutritin class. Natinwide, Tmand accunts fr nly 8% f the CDSMP. IMPAQ/Abt pinted ut that there are tw ptins: Include Tmand in the study (can t identify variability) r d a substudy. TEP Members pinted ut that the English versin f the CDSMP is prvided in multiple languages. Instructrs implement the same training but they use translated materials. Tmand is the actually mdified slightly. Wrkshps delivered in ther languages may be recrded in the same manner as English CDSMPs. If CDSMPs are being prvided in ther languages, the evaluatin materials (survey and frms) will need t be translated. Budgeting fr translatin materials will need t be included in the evaluatin reprt. Discussin f Site Selectin and Crdinatin acrss Sites In rder t get an adequate sample size, it appears mre than 20 sites will need t be included in the study. If there are certain sites that are serving mst f the clients, they need t be included but the sample must als be representative. Selecting nly the largest sites wuld likely create an urban bias in the sample. TEP members pinted ut that they have been targeting underserved ppulatins. These ppulatins are ften in rural areas. These ppulatins wuld be missed if the analysis nly includes sites frm urban areas. Page 6 f 6
Class Zer A TEP member suggested that the evaluatin cntractr may be interested in recmmending sites hld an intrductry sessin where participants understand what they are signing up fr. This wuld als be a gd pprtunity t cllect baseline data. During the site interview calls IMPAQ/Abt did nt find that class zers were being implemented. Sme sites reprted that they read a script t interested persns wh call in. The script explains what the wrkshp entails and it is thught that this may help t prevent drp uts. The prcess was nt cnducted in a systematic manner. TEP members nted that nt all participants shw up fr the first wrkshp. Participants ften shw up fr the secnd class after hearing abut the prgram frm a friend/relative. A participant wh shws up fr the secnd wrkshp and attends at least 4 sessins is cunted as a cmpleter. Page 7 f 6