Financial Incentives, Workplace Wellness Program Participation, and Utilization of Health Care Services and Spending

Offering financial incentives to workers to participate in workplace wellness programs has a big impact on the programs’ success, according to new research by EBRI.

This paper analyzes data from a large employer that enhanced financial incentives to encourage participation in its workplace wellness programs. It examines, first, the effect of financial incentives on wellness program participation, and second, it estimates the impact of wellness program participation on utilization of health care services and spending.
The Patient Protection and Affordable Care Act of 2010 (PPACA) allows employers to provide financial incentives of as much as 30 percent of the total cost of coverage when tied to participation in a wellness program.
Participation in health risk assessments (HRAs) increased by 50 percentage points among members of unions that bargained in the incentive, and increased 22 percentage points among non-union employees.
Participation in the biometric screening program increased 55 percentage points when financial incentives were provided.
Biometric screenings led to an average increase of 0.31 annual prescription drug fills, with related spending higher by $56 per member per year. Otherwise, no significant effects of participation in HRAs or biometric screenings on utilization of health care services and spending were found.
The largest increase in medication utilization as a result of biometric screening was for statins, which are widely used to treat high cholesterol. This therapeutic class accounted for one-sixth of the overall increase in prescription drug utilization. Second were antidepressants, followed by ACE inhibitors (for hypertension), and thyroid hormones (for hypothyroidism).
Biometric screening also led to significantly higher utilization of biologic response modifiers and immunosuppressants. These specialty medications are used to treat autoimmune diseases, such as rheumatoid arthritis and multiple sclerosis, and are relatively expensive compared with non-specialty medications. The added spending associated with the combined increase in fills of 0.02 was $27 per member per year—about one-half of the overall increase in prescription drug spending from those who participated in biometric screenings.

Figure 13 The existin inclu Note: A Figures F References Appendix–Econ biased igure Health ded the g estimat 20 shows the reg ll mo literat In 1 dels incl surance ure 5 e,31 s of t is replete 2 u u hde nion e o Porta ef metric Analysis d t ression fec m h with problems be fol e ility a ts of HRA mbers model r lowin nd Acco ex g and posed to e co sults for untabi variates: geograp of incompl BIO. the lity Act Whil biomet ne e th ete w of 1 ric screen e d financi inc ata. In some cases, p hic 99lusio 6 re a (HIPA l i gion, ing partic n of ncen A in household siz ) tive for dividual prohi ipatio b part its di rescrip n. A fixici escrimination a e d sp , hea expecte effe tion ation cts in ldru th in H insur g d cl , re the RA ai gance quirin s ainst i ms were not only, begi model sa plan g ndivi non type, fegua du -nni u als nion ng i rds n Paul Fronstin is director of the Health Education and ReFi search Pro gure 16 gram at the Employee Benefit Research Institute Financial Incentives, Workplace Wellness Program Figure 1 Sample Means, by Cohort, 2011 Figure 6 availabl against t based on plan annual employees year wa e, an his po heal 20 ge to partici 1 d a 3 th status, but reversals an tential m . T ount, num est Gro pate in biometric endoge up d ber of does allo 2 consiste adjustments o neity years o due w sc d of to ti em re f enin tenure ployers to the 4 ccurring aft me-invari gs in 0 wit ,5 addi 47 ant hpro no th Figure 10 Figure 19 er Figure 2 Figure 8 Figure 4 tion to HRAs u v e em n-un clai id nobservabl e r ms we ion ployer, ewar emplo re s ds i ea n s, or fina n u order to y d t bmitte dynamic co ees who he C ncial d h rec arlso were w inc e e nfo rentives fo ive th n Comorbidity not r e of unders fere e econcil $2 d t may re r em 0/month premiu he ed ployee new Inde . On main. T fin e study x score. parti ancial here c m ipation did not fore, (EBRI). M. Christopher Roebuck is president and CEO of Figure 17 RxEconomics, LLC. This article was written with assistance Baicker, Katherine, David Cutler, and Zirui Song. "Workplace Wellness Programs Can Generate Savings." Health Affairs Figure 1, Use of Specific Health Management Programs, by Firm Size, 2013 .............................................................. 5 Use of S Mp eec an S ifipe c H ndi ea ng on H lth Mana ealge th C men aret S P erro vig cr eams s, by, by Cohor Fir t,m 20 Si 11ze , 2013 Financial Incentives Effects on Participation, and Utilization of Wellness Program Participation Test Group 1: Select Health Care Services and Test Group 2: Non- Control Group: Other Individual Impact of Participates Financial Incentiv in Wellness Program Offered es on Participation by Employer, Percentage Reasons for Percentage of Percentage of Firms Offer Participating in of Firms Offering IndiviSpecific Financial duals Employ Reporting Health Risk er Wthat They ellness Incentiv Assess Program, 2014 es Would to Employ ments ees an instrum incentive in a have h discount r weellness alth for eent sult ca participatio program. al varia ed re claims, so in ab 5 les Whil Participation in 5n in -per (I a V) analysis e HIPAA HRAs in centage-point verage spen allow plan wa Health di y s s also for i ng increase ear 2 was im n conduct 0 centiv Risk 12, an in put pa e Ad fo s e ssessme e up to rticipation. d to d based on r both obtain unbiased estim as mu HRA n As ts, u ch as 2 t sexpected, ilization. by and biometr Cohort, 0 perc And, the eates of n 201 i in some c screeni t of change i 1 the ?2013 ? an total studi n ngs f d ina s icost of covera n e . Specific ncial i pl s, data an y ne cent on H ar ally, 2013. ive since RAs ge, a s had from EBRI 29, no. 2 ’s research an (February 2 d 0 edi 10t): orial staffs. 304-311. The authors wish to thank Robert Kaestner for valuable comments and Union Members Union Employees Union Members Test Group 1: Select Test Group 2: Non- Control Group: Other 90% Among Those Offered a Wellness Program, 2014 To e Spendi xamine thng e effects of financial incein Health Risk ntives on wellness pro F Aissess gure 21 gram men parti ts (HRA cipation, s) a difference-in-differences (DiD) Who Participate Probably or Biometric Screenings, by Participate in Employ in Wellness Programs, by er W Firm ellness Size, Firm Size, Program, 2014 2014 Descriptive Statistics (n=15,312) (n=40,547) (n=16,123) Variable Finally, maximum o and Figure financi a small im biometr a 2, the l incen Perc pact co fi c 3entage of Firm hort of tscreenin 0 (5 ives wer percent percentage poi 16, gs w e is im 12n 3 e plemente o union s Offer re oft w allo ne ts) t em n missi we ing Hea d d h ploy for s e year befor unng. ees w der le th Risk lecte PPACA, hd o n co Assessments or Biom ee th ho ver we rts in ce hich a requirem receiv lso allo ed the rtain ent t ye w $20 s o ars, indivi up to etric Screening get /month the 50 pr d per re ua em duction lcsent i ium discount were s, bn y i in essenti F nicenti rm hea Size, lth insura took ef vally pse es for i 2014 fect. n unce ............... 7 terve do - ntions suggestions. Any views expressed in this report are those of the authors and should not be ascribed to the officers, 100% Union Members Union Employees Union Members 60% 95% August 2015 • No. 417 60% by Various Financial Incentives, 2014 93% 40% Baxter, analytical Siyan, appr oach was used with Kristy Sanderson, Alison J. Male person V -level enn, Le fixe igh 86% d ef Bl fe izcts. Speci zard, and An ficall drew J. 71% y, the Palm followin er.g "T models he88% Relat wer ion e spec ship B ifie etw d:een 70% Wanted to improve health premiums des randomized to Similarly, igned to unio duri prn mem HRA and ng t event or he t bers h re ree BIO. duc di -y d not e Hence, ear study tobacco Imp rece time-by-gro ive peri usany fi e. od comprised act About o nancia fu W one p interac el i lln -n ha centiv ess Pro the Contro lf tion terms s of lar e to g gerpa l employers off am o 45% Gr rticipate oup erved as instr n Ut (Figur in iliz the at e er a io 1 biometr 2 uments in n f ). o inancial f ic scr a inc 35% two-stage I eent enin ive for g progra compl V m. The eting 80% Variable (n=15,312) (n=40,547) (n=16,123) trustees, or other sponsors of EBRI, Employee Benefit Research Institute-Education and Research Fund (EBRI-ERF), or Sample By Paul Fro meansn are sho stin, Ph.D wn in ., Em Figure plo 13 for yee each o Benefit f the th Rese ree co arch Institute, and M. horts as of the beginning o Christopher Roebuck, f the baseline year (2011). 79% 80% Age (years) 47.2 47.1 65% 46.5 80% Arguably the Figure Return 3, Perc on I 90 most critical m entage of Firm %nvestment an 36% es That O thodolo d Quality of g ffer E ical Stu challen mployees t dy Metho ge that hd e olo pla O 20pportunity to 12 gy in gues 20 Workpla the 13 field der Complete a ce Healt ives fro h Pro H m e m the alth otion Pro volunta Risk Assessment, b gra ry nature ms." American ofy Firm Total healthcare spending Health Care Serv $3,382 ices and Sp $3,679 ending $3,588 HRAs at an estimation fra regression res anm ults show t nual val ework. Mo ue of hdels (1) an e lfinanci ess than al id $500 for ncen (2) comprised th tives increase nearly two-t e first d S partici m hirds all emp stage, an p loy ation er d s a more than n by 2 perc La d rt gh e e e seco mploy $500 enta ernd stage s ge points in pl for on is given e-third of these em be an low. year 2012, ployers and their staffs. Neither EBRI nor EBRI-ERF lobbies or takes Small Empl posi oyerstions o Largn e E sm pp ecif loyeic rs policy proposals. EBRI invites comment on The sample was predominantly male (86 ?88 percent of union; 71 percent of non-union employees). Mean age was Ph.D., RxEconomics, LLC 62% Region 79% Sm all Employ ers Large Employers 1 50% * 2 * 1 * 35% 72% 51% Figure 12 51% workplac Journal o Size, 20 e welflness pro He 08 al ? th Promotio Inpati 2014 .......................................................................................................................... gent hos rams. Employees pi ntal 28, no. 6 (J who o uly/Aug pt to 251 ust 2 take part 014): 34 in HRAs 7-36 251 3. or biometric screeni 262 n88% gs are likely to be di ............. 7 fferent To maintain current health status 32% 41% 50% 70% (Fi reduc gure e 5 d partic ). ipation by 3 percentage points in 2013—findings perha HRA ps also attribut Biometric ed to Screening changes in member this research. comparable60 ac 80 %ross the thr % ee cohorts at about 47 years, and the majority of subjects resided in the western part of the Northeast 0%2 68%%2% 50 70% % 2 68% * 67% (1) 70% Emergency depar tment 142 102 ¯ 117 66% 66% from those who choose not to participate. Economists call this selection bias. If those differences in individual (4) Fin U a tiliz nc ation of H ial Ince ealth C ntiveas f re Ser or H vic ees alth Risk Assessments (HRAs) and Biometric messaging. Midwest 1% 23% 9% United States. Household size averaged 2.5 ?2.7 individuals per family, and tenure was in the range of 15 ?17 years. Charlson, M. E., P. Pompei, K. L. Ales, and C. R. MacKenzie. "A New Method of Classifying Prognostic Comorbidity in 30% Specialist physician 155 177 182 Figure 4, Percentage of Firms Offering Specific Financial Incentives to Employees Who Participate in Wellness 66% Inpatient hospitCo al adm nvenienc iss e to ions work 0.001 -0.0002 Financi The characteristics 2014 EBR 70I/Gr are al Incentives, Workpl % een unmeasur wald & Associates Co ed and also consumer En rrelated with o gag ace Wellness Program e ume tcomes nt 32% in H of int ealth Care Surv erest such as ey (CEHCS utilizatio 38% n of h ) addressed a n ealth care ser umber o vices f Screenings (BIO), by Employee Group and Plan Year, 2011 ?2013 South 0% 17% 1% Introduction 60% Health status was essentially the same across the three cohorts wh en measured by the Charlson Comorbidity Index Primary care phy sician 309 216 260 Longitudinal S tudies: Development 1 and Va * lidation." Jour 2nal of Chroni * c Disease 40, n 1 o. 5 (1 * 987): 3 73- 83. 40% 61% 60%Programs, by Firm 50% Size, 2014 .................................................................................................. .................. 8 Inpatient hospital days 0.002 -0.003 Equation questions and worker (4) differs fro pr m oductivity, n the individu from West (3) in t aï al perspectiv ve observatio hat predic e n regar ted—rather th al analysis o ding 99% part f icipat we an actua llnion ess pro in l—val HR grues of As and am impacts 58% HRA a biomet wi nd BIO ric screen ll yield obta 88% incorre ing programs. Th ined from ct estimates. I (1) ane d n survey (2 the ) Wellness Program Effects on Utilization of Health Care Services and Spending Copyright Information: Study Group This report is copyright Plan Year ed by the Em 2011 ploye Plan Y e Be earnef 2012 it Research Institut Plan Year 2e 013 (EBRI). It may be Participation, and Utilization of Health Care Services and Other outpatient 1,515 1,785 1,733 24% Employer spe (CCI), a proxy 2540 %%n measure of h ding on employme ealth status nt-based healt derived from h ben di efit agno s in th sis codes e privatfrom med e and public s ical ¯claims ectors reach data (C ed $ harlson, et 660 billio al. n in 2 1987) 013, 60% 2 * (2) Emergency department visits 0.002 -0.003 Household size (indiv iduals) 52% 2.5 2.7 2.6 To learn more about own health risks 31% 38% enter the model. The IV estimates are ? and s . Again, bar accents distinguish coefficients in (4) from their absence found that of 62 a randomiz percent of in ed co dintrolled trial, viduals report aed dvance partici d e pcon atinometri g in em c tec ployer hni ques mi HRAs and 65 ght bepercent used in an partici attp eated in mpt to biometric correct for 50% Test Group 1: IV IVHRA 3 : 1 HRA: HRA: Deyo, R. A., D. C. Ch Prerkin, escriptia on dr nd M. ug A. Ciol. "Adaptin 1,009 g a Clinical Comorbi 1,147 dity Index for Use with 1,034 ICD-9-CM Administrative used without permission, but citation of the source is required. accounting for one-quarter of national health expenditures. Plan sponsors are increasingly interested in returns that Figure (Dey Figure o, Cherk 5, 21 re Perc po in an entage of Large Employers O rts the im d Ciol 1 pact 992 55% s) of HRAs (Quan, a et a ndl. 20 biometr fferi05 ng Fi ).ic s nac nci real Inc enings e on ntive uti fo lization r HRA Com of heal pleth care tion, an an d Ma d spendi ximum Val ng as esti ue ofmated Spending Specialist physician visits 0.01 0.002 Tenure (years) 15.3 16.7 17.3 50% 57% counterparts i this selectio screenings w n hbias, but u n en th (3),ey anwere d or nde di rstandin availabl nary least s e g th (Fiq g e u udif ares was re 6 fere ). O nces nly a empl be btween out on oyed to ( di participants and non e-hal s estim tributi f o on by ate the model f indivi cohor duals r t) pearticipants with ro ported that Premium bust st Inc is o ent thiandar v eir em f gr e eat dployer erro import rs oance i ffered n 4050 %% EBRI Employee Benefit Research Institute Issue Brief (ISSN 0887 ?137X) is published monthly by the Employee Benefit Research Institute, Databa 30 ses." % Journal of Clini 32% cal Epidemiology 45, no. 6 (June 1992): 613-9. 20% by the might multiva be generiate r rated S C el harls ec e from t gressio t U on c nion P om hriim s E n orbidit large m models. No ary c ploy yare phys e index invest es m ic signific ient in an vis $50 gi it terms ant s f0. t c 19 eff ar of d ects of improve partici 0.01 $ d50 gi member pation in 0. f19 t card hea HRAs lth -0. an on util 001 d enhan ization 0.19 ced of he worker alth pro care ductivity. Individuals Incentive, and time Lar are ge Em indeployers, xed usin 2014 g i (from 1 to .............................................................................................. N) and t (years 2011, 2012, and 2013). The dichotomous ................ 8 dependent 18% Offered incentive prizes 34% 27% 41% 1100 13t40 h S %t. NW, Suite 878, Washington, DC, 20005-4051, at $300 per year or is included as part of a membership subscription. Periodi- Total healthcare spending 100% 100% ($20/ 100% month) clustered by Recommended C person. Sur itation: prisi Paul Fronstin ngly, the IV c and oeffic M. ients Christo werp eher quite Roe 41% similar buck. in “Finan both 41% cial Inc magnie tude ntives, Work and signiplace ficance Welln to tess heir naïve some type By Paul Fro such pursuits. of f n inancial stin, Ph.D incent., Em ive forplo part yicipating (Figure ee Benefit Rese 7). arch Institute, and M. The main reason individuals partici Christopher Roebuck, pated was to improve All three co 30 hor % ts had relatively high earnings in 2011. Non-union employees earned an 40% average of $98,346 in 2011. Prescription drug fills 0.08 0.31*** Annual earnings $62,731 $98,346 $94,542 Recently, there has been growing interest in workplace wellness program as a means to realize these benefits. The services an cals postd s age pra en te din paig d i n were Wash found; ington, D that C, and is, all co additional m efficie ailing n o tsffi were ces. P statisti OSTMAS cally in TER: Send distin addr guishabl ess change e from s to: EBRI zero. Issu To assess e Brief, 1100 variables, HR40 A and % BIO, indicate whether or not person i for plan year t received a health risk assessment or biometric 40% Inpatient hospital 7% 7% 22% 7% Fronstin, Paul. "What (n= to Ex 15,312) pect During Open Enrol BlIment O: Season: Findin BI g O s: from the SHRM/EBI BO R:I 2014 Health Benefits 26% health— counterparts, Program Parti 45 percent cit c suggestin ipation, Source: e an d thi EBRI g t d Utilization h anal at confo s as a major r ysis based on adm undi of Health ng e ason an due inistr to time at Ca ive cl re S d ai 35 me s -vary rvices dat pea. rcent report ing anfactor d Spendi s was not ed t ng. his as a ” EBR at is mi I Issue sue. For nor reason Brie br f, no evit (Fi . 41 y and since gure 8). Oth 7, (Emplo IV e yee r Ph.D 13t., RxE h St. NWc , S onomics, LLC uite 878, Washington, DC, 20005-414% 051. Copyright 2015 by Employee Benefit Research Institute. All rights reserved. No. 417. 15% Spending Figure Members o 6, Indiv f unions that idual Partic diipa d not bar tes in Wellnes gain in sthe e Program nhanc Of ed fefred by inancial i Emplo ncenyter, Among T ives earned a hn ose Of avera fere ge of d a Welln $94,542, ess Pr where ogram, as 30% 40% wheth Patient e Protec r thesetion an non-fin d A din ffor gs d were able pote Carential Act of ly du 20 e to 10 (P the PA st CA) al udy’s use Testlows Group 1 of employers to ordinary least provsquares mo ide financial 13% del incentives o s with the non f as much - screening, r 30 e% spectively. Emer genc TEST y depar 1 and tm TES ent T2 record th 4% e individual’s Test Grou 3% p assignment (as 3% previously described). The 20% Reduced prem B ium ook s Mov 36% ie Tickets Draw ing 20% No Incentive 32% Notes: 18% 32% Survey." EBRI Notes 35, no. 12 (Employee Benefit Research Institute, December 2014). estimators are less efficient, Benefit Research Institute, TA otu only the (n agust 201 l 5on- ). IV) linear fixed effects resul -$2 ts from equation (3) are pr $53 esented in the paper. wellness program enrollment reasons included: 1) to maintain current health status, 2) convenience to work, 3) to 30% 30% those who did earned an average of $62,731. 2014 ........................................................................................................................................................ 9 Specialist physician 5% 5% 5% as 30 percent (up from 20 percent) of the total cost of coverage when tied to participation in a wellness program. This normal distributions of Test Gr the oup 1 = depen No incent dent vari ive for bi ables, th ometric scre stu eening,d incent y also Tes ivte f Gestimated or rou heal p 2 th risk assessm non-linear ents in 2013. fixed effects Poisson models, but Control Group and year 2011 are the excl 28% uded categories. X represents a vector of time-varying, person-level 30% Data and Methods Test Group 2 Inpat :ient hospital HRA: 6 HRA: 8 HRA + BIO: 20% 26% learn mor10 e a %bout own health risks, 4) offered incentive prizes, 5) reduced premiums, and 6) to avoid premium 9% T P est rim Gr aroup 2 = y care Incent phys ivic e f ian or biometric screeni 9% ng in 2013, incentive for heal 6% th risk assessments in 2012 and 2013. 7% 20% 8% Fronstin, Paul, and M. Christopher Roebuck. "Financial Incentives and Workplace Wellness-Program Participation." EBRI results we provision o re f PnPoACA is e t meanin xpec gfull ted to y dif 11%fer driv ent. R e ine creased ado sults from th pti e on mo of this stra dels estima tegy, an ting the d t impac here is alr ts of biometric eady evidenc screeni e to suggest ngs on covariates including geographic region, To avoid p ho reusehold size, mium increases health insuran 31% ce plan type, 18% annual wa23% ge amount, number of 20% Control Group 20% Emergency department 1 -4 Report Availability: This report is available on the Internet at Pww remw.e ium I b nc ri.org entiv e Premium Incentive AT A GLAN CE With Figure resp 7, Em ect t plo oyer Offers Ca enrollm Control ent Group = by sh Ince type o No incent The Em nt f ivhealt ive o e forplo bih r Re om pl yee etan ward for ric scr Benefi by th eenit Research Institut ng, e P no i three articipati ncentcohorts i ive fn org in heale t n h r Well (EB 2 isk assessm 011, ness Pro RI) was founded in 1978. Its regar ents. g dram, Among less of cohorWorkers t, the majority mission is to Whos of e increases. The survey also a Other outpati sked in ent dividuals about 45% their reasons for no 49% t participating in th48% eir employers’ wellness 10% Nonunion Employees $50 gift card 19% Major Reason that addin Issue Brie g we f, no llness . 41 progra 2 (Employ m ince ee B ntives i enefit s by fa Reser more arch In stitute commonplace , March 20 among 15). employers than other changes (Fronstin utilization of health care services and spending were similar with one exception: Biometric screenings led to an average years of tenure with the em Spec ployer, a ialist phys nd th ician e Charlson Comorbidity In 1 dex score. Individual 1 fixed effects are included in 20% ($20/month) ($20/month) Study Sample contribute to, to encourage, and to enhance the development of sound employee benefit 5% Prescription drug 30% 31% 29% employees were covered under a preferred provider organization (PPO) (Figure 14). However, among members of programs Emp . A 10b % lout 6 oyer Of 0 perce fers W nt res ellnep ss P onde rog d t ram, 20 hat they 14 di 9 d not participate because they could make changes on their own 3% 10% Who we are Prim Tary c o addrare phys ess specific ic he ian alth problem 2 -2 increase 2014). of 0.31 annual prescription drug fills (p<0.01), with 17%related spendin 32% g 2% higher by $56 per member per year the mo dels as denoted by a and dprograms and so . Moreover, e und public policy and ? are the i through objective diosyncratic error t resear ech and rms. Fin educati ally, Mino on. EBR ßr Reas , ß , on ß I is the only , ß , ? , ? , (n=40,547) BIO: BIO: i i it it 12% 1 2 3 4 1 2 10% So urce: EB RI analysis based o n administrative claims data. 1% 11% For this unions Goetzel, Ro with study, n th 10% Z., et e inc data al. entive, we"Do re th used Workpla ose enroll from a ce H lar ees ealth Promotio ge manu not in the factur PPO n (Wel F in wer igure 14 g employer lness) Pr e in the h ograms point of eadquarter service Work?" ed i (PO n the JourS) pla nal o Midwest fn O . I ccupation n, contrast, am with em al a ploy nd ees ong (Figure 9)—28 percent cited this as a major reason and 33 pe 10%rcent reported it as a minor reason. Lack of time was the ? This paper analyzes data from a large employer that enhanced financial incentives to encourage Other outpat privat ient e, nonprof it, nonpartisan, Washington, DC-b -1 ased organi -6 zation committed exclusivel 9% y to (p<0.01). Notes: Book Movie Tickets Draw ing and ? 10 are par % ameters to be estimated. Note that bar accents are used to distinguish coefficients in the BIO equation 0% Figure 8, Reas This Issue Brief 0% 0%0% ons was con for Participating ducted through th in Employer Wellness e EBRI Center for Program, Research 2014 on Healt ........................................................ h Benefits Innovation (EBRI CRHB ......... 11 I). The Distribution by Type of Health Plan, by Cohort, 2011 non-union members and the Control Group, very few were enrolled in the POS plan. Instead they a were enrolled either second-most-popul located throu Environmental ghout th Medicine ar reason for e Unite 5 d6 Stat , not no. 9 public es. Th parti (S poli ept e cipatin study rec em cy r bg, a e er search 20 n ed iv 14 others stated t ed healt and ): 92edu 7-9 cat h34. iion on nsura hat they nce e econo nm were rollme ic securi n already t informat ty abnd em heaion, lth plo y. y medical ee benefi an t is d sues. Prescription drug -11 56*** Despite the participati ir rise in on i ponpulari its workplac ty, workplac e wellness pr e wellness pro Statem ogram g en rams s t o . It ex f face Owamines, first, nersh challie p nges. While the effect they may com of financial incen ply with Itn ives on ternal Any Financ Te ia s l I t G nc S r enti o ele up v ct e to 1 Uni = 20 N 11 on o in Mem Wo cerk nt ber ers ivs e ( Pa fT oes r y Sm bio t Grm a oup lle etrr 1) ic scree Wo nin rk g, ers in Ha ce 20 nt ve 12 iv Sm e fo alrl healt Re hNo c re is in- v ke Hi as union s gh eer HS s Em sm plo eA o n yte sr es in 2 HRA (T0 es 1Re 3 t G .20 cre oup 13 ive Gi 2) ft Cards, Travel, If employer made If employer offered RequirIf emp ed by e loy mp er loy offered er If emp 19% loyer If emp 17% loyer reduced If employer offered If required to Control Group: HRA: HRA: HRA: from their counterparts in the specification of HRAs. Ordinary least squares was used to estimate the models with Test Group 1: Select Test Group 2: Non- Control Group: Other following organizations Participate in provi Wellne dss e thPe e rfun centad gin e of g for E the PremB iuRI C m RHBDe I: duc Am tibleerican Express Co, A Contributionsmeriprise Me Fina rchandis ncial, Aon e, or Cash Hewitt, additional discEBRI’s member ounted $250 cash ship includes a incentive increased p cross-section of remium contributpension funds; businesses; trade associations ion to time off participate to enroll ; TeS source: t Gro u pEB 2 = RI I nanalysis centive f based o r bio m on et adm ric scinist reenrat ingive in 2 claim 01 3, in s cdata. entive fo r health risk assessments in 2012 and prescript in the exiclusiv on dre provi ug claims, as der organization United States Po well as w e(EPO lln stal Service Stat ess-pr ) or a h ogram ealth ement -pa ma rticipation data covering t of intena Ownen rship, Managemen ce organization t, and Circulation h (HMO). Overal e period 2011l ? , very f 2013. The ew em em ployees ployer To inv Revenu estigate the main finding e S wellness prog ervice (IRS) an ram participation, and sec d Departm furthe ent o r, thfe pr Labor escriptio (DOL ond, it estima ) r n drug fi ules, w lls an e tes the imp llness d sp programs endiang models w ct of wellness p fall under ere r thre um e ogram partic -estimated brella ofby the ipation on 0% Other Union Employees $50 gift card $50 gift card $50 gift card contribut Pr ion og to ram HSA premium Union Memb fo err n s on-participU ant nion Em s HSA* ploy or HR ees A* Union Members in pr efered health Figure 9, Reas 0% 0% ons for Not Participating in Employer Wellness Program Among Those Offered but Not Participating in Grossmeier, J robust standar Source: EBR ed ssica, Paul errors C I analy clustered o ntro E. Terry, sis based on l Gro up =by N administrativ labor un Davi o inc person. end tiv e ions; h R. Anderson, claims e f o r b data. ioem alth etric car sce prov re and eningDavid iders and insur , no inceWri ntiveg fo ht. r h ee rs; "Fi alth government org n rancia isk asse l Impact ssments. anizations; of Population and service firms. Health Blue Cross an Source: EBR d Blue I Shi analy Notes: eld sis based on Association, administrative claims Boeing data. Co., Deseret Mutual, Federal Reserve Employee Benefits System, A number of questions were Publication Title: asked of nonparticipants EBRI Emplo 0% to yee gaug Benefi 10% e th t Res eir 20% lik earc elihood of h30 Ins % titut 40 e Iss part % ue icipatin 50 Brief % g 60in r % esponse to 70 % 80% various Health Risk Assessments Biometric Screenings chose the consumer-directe or HRA* d health plan (CDHP) in 2011. plan had been offering HRAs since at least 2004 and introduced biometric screenings in 2007. Individual responses to HRAs Notes: Variable Health Risk Assessment (n=15,312) (n=40,547) Biometric Scre (n=16, ening 123) Americans utilization of health wit Dis he Disa ase bilitie Healt s Ac ct (ADA hare services and Tel). Em ephoneployers have or spending. Health be On en -site waitin fitnesg y s Fit eness ars for th facility e Equ Face-to-f al Employme ace Sleep disor ntd Opport er Resiunity liency therapeutic cl Noass, and th tes: (n=16,1 e c 23) oefficients sorted accordi BIOn : g to magnitude (wh BIOe : re statistically sign BIO ifica : nt). Figure 22 presents Program, 2014 ........................................................................................................................................ 11 Test Group 1 = *** No S incentiv tatistically e for biometric significant screening, at the incentiv 0.01 e for health level. risk assessments in 2013. Management Programs: R a eevaluating the Literatur Publication Number: e." Population 0886-137x Heal th Management 15, no. 3 (2012): 129-134. Source: Test Grhttp://kff.org/rep oup 1 = No incentiv ort-se ect for biometric ion/ehbs-20screening, 14-section-incentiv twelve-w e ellness- for health progra risk assessme ms-and-hea nts lth-r in 2013. isk-assessments/ types of General M finan ills I manag cial in nc., Health ement centives. asswa essm ys, Internation Th ente survey web-based coal Busin ncluded that advoess M cate achines Cor betw fac een ility 66p p .,e d JPMorga rcent iscoun and ts n he 7 Ch al 9t h ase, Mercer, perce /lifestylnt of e diag in nosis a a dinviduals d nd Pfizer I pr woul ogn ram c.d PPO 61% 73% 75% as well as results from the biometric screenings were also obtained. a Test Group 2 = Incentive for biometric screening in 2013, incentive for health risk assessments in 2012 and 2013. Book Movie Tickets Draw ing No Incentive selected result Commission (EEOC) to T Hest s from this eatlh sav Group 2 = ings accounts. Incentiv prov ex e ide formal ercise. T for biometric screening h gu e lar idg in 2013, ance on est inc incentiv rease well e for health risk ness pr in medication assessmen ograms as they r ts utilizatio in 2012 and 2013. n ewas for stati late to the An Ds, w A. Yet, in 2014, t hich are widelh ye EEOC used Since the fin Source: ancial i EBRI/G ncent reenwald & Associates ive for partici Consum pati er Engagement on in HRAs w in Health C aas implem re Survey, 2014. ented in 2013 for Test Group 1, and was in effect in Source: EBRI/Greenwald & Associates heEBRI’s work advances knowledge and unders al Ctonsum h/lifes er tyEngagement le servic in H es ealth Care Survey, 2014. tanding of emplo coaching trea yee benefits and their tment b 1) Filing Date: 08 b C /15/2015. 2 ontrol Group = N ) Issue Frequ o incentive for biometric ency: screening, Monthly no incentiv . 3) Nue mber for health of Issues Published Annuall risk assessments. y: 12. 4) Annual Subscription Price: $199 per Source: Kaiser Family Foundation, 2014. Source: Health reimbursemen EBRI/Greenwald & Associates t arrangements. Consumer Engagement in Health Care Survey, 2014. Despite ther * Asked only e being of EPO no health sav diffier ngs account (H ences in SA)/heal the C th reimbursemen CI by cohort, t t arrangement (H here RA wer ) enrollees. e some small differences in utilization of health care participate if a financial incentive was introduced, an 0% d nearly 80 percen11% t would if the employ 16% er made additional ? The Patient Protection andcoac Affordable C hing are Act of 2010 (PPACA) allows employers to pr programovide financial Participation in HRAs increased from 57 percent in 2012 to 88 percent in 2013 among members of unions that Mattke, Soer y file ear d its or is included as part of a me first-e ev ner le , et al. gal chall Worke place Welln nges mbership to importance to well subscription. 5) Complet ess Programs St ness programs, the nation’s econo udy: cl eaimin Mailing Address of Final my g Report. t h among policy at they violated t Santa Known O m Monica: RA akers, ffic he e ADA the news of Publication: (Not prin ND (Pomeranz Co media, and rporation, 2015). A the public. It te 20 r): 1 E3 m . U.S. ployee to treat Figure 10, high c Perc hentage of olesterol. c In This therapeutic dividuals Repor class accounted ting that They Wo foruld Pro one-sixth bab of t ly Part he ov icipat erall e in Em increase ployer in pr Well escription ness Pro dr gram, ug 2012 and 2013 for Test Group 2, then ß , ß , and ß are estimates of the impact of the financial incentive on 2 3 4 HMO th 0% 13% 8% services. There were 14.2 (per 1,000 individuals) hospital admissions among members of unions that bargained in the contributio The study exa ns tm o an HSA ined the or set he of em alth rployees eimbursement a who were rrangeme full-time active nt (Figure wo10). rkers, 18 About 7 i ?64 y nears of 10 woul age d pa (articipate s of Dec. if 31, the 2013), Benefit Research Institute (EBRI), 1100 13 Street NW, Suite 868, Washington, DC 20005. 6) Complete Mailing Address of Headquarters incentives of as much as 3 does this b 0 percent of the total cost y conducting and publishing policy re of coverage when tied to search, analysis, participati and special reports on on in a wellness bargained in the incentive (Test Group 1). Interestingly, participation increased from 40 percent in 2011 to 57 percent What we do d utilizatio District Ju n. S S by Var odge urce e: M c deni oi eous nd rcered were Nationa Financial I EEOC’s SS l Survey RIs/SNRIs req n ofcent Eu mes ployer- ives, 2014 t to (for Spblo ons depr o crk one o ed ......................................................................................... H ession an ealth Plan f th s, 20 e d an programs 14 xiety), AC from E inhi impl bitors ementa (for hype tion. Althou rtensio th ghn the ), follo EEOC rec wed by .............. 12 ently p To e artix cip amin atioe th n ine e HR ffects o As. An u f fin ndancial erlying ince assu ntiv mp es on tion of well Dness iD mo pro deling gram is that the trends in participation, the a the dependent varia nalysis estimated multivariat ble are e POS 39% 0.2% 0.3% or General Business Office of Publisher (Not printer): Employee Benefit Research Institute (EBRI), 1100 13 Street NW, Suite 868, employee benefits issues; holding educational briefings for EBRI members, congressional and and co incentive; ntinuo 15.u2s a ly enroll dmissions ed in among the com non-u pany nion health employe plan e du s; and ring the 15. thre 4 ade-year period 2 missions among m 011 t embers hrough o20 f un 13. Em ions witho ployees ut the in employer: Nyce, Stephen 1) dis , Jessica counteGrossmeier, d premiums,David 2) offere R. Anderson, d a $250 cPaul E. ash ince Terry, ntive, and 3) incre Bruce ased premiums, 4 Kelley. "Associatio ) ren Bet ducew d contri een Ch butions anges program. e in 2012, even before the change in the financial incentive took place. This may have occurred because of stepped-up Washin issued gto propos n, DC ed r 20005. egulations 7) Full Names and Com that address welln plete Mailing Addresses of Publis ess programs and the A hD er, Editor, and A, until final M regulat anaging Editor (Do not leave bla ions are released the nre is k): regression thyroid hormo models usi nes for CDHP hypot ng a di hffer yroidism. Also ence-in-differenc signific es antly 0% (DiDhi ) analytical gher was t a hp e proach w utilizatio 3% in of th per biologic son-level 1% respon fixed ef se modi fects. As fiers and equivalent across groups prior to treatment. Thus, 0 represents a test of our identification approach, since both federal agency staff, and th e news media; and sponsoring public opinion surveys on employee th incentive (i.e., the Control Group) (Figure 15). The number of hospital days (per 1,000 individuals) was highest among health to HSAs or h in th pla e Hea ns t eh alth reimbursement arra lat pa th Risk Status an id claims on a d Cha prepai n ng ges in ements, 5) o d or ca Future Hea pitated ffered t basis were excl lth Care Costs." ime off, or 6) uded, sinc Jorequir urnal o ed participatio e m f Occ onetary amou upational n to an nts wer enroll d Enviromental in t e nh ecessary e Publisher, Employee Benefit Research Institute – Education and Research Fund, 1100 13 Street NW, Suite 868, Washington, DC Figure 11, Percentage of Source: EBRI In anal dividuals ysis based on adm Reporting t inistrath ivat e clThey aims dat Wo a. uld Probably Participate in Employer Wellness Program, communication efforts that reached all employees through banners, posters, internal mail pushes, supervisor crew immunosuppr still uncertainte y regar ssants. Thes ding ae spec number ialty me benefit issues. of pro dications a visions, whic EBRI’s Ed re used to h ma ucation and Re y dam treat pen autoimmune diseases, such employers’ ent search Fund (EBRI-ERF) performs husiasm foras rheumatoi wellness the char progra dit arthritis able ms. , Test previous Group 1 ly not and t ed, the he tContr hree groups ol Grou wer p still e e had th xposee same d to the i F nienhanc centive gure 22 ed fi struc na ture ncial i for HRAs ncentive in 2 s for 01dif 2. Sim feren ilarly, the t programs at th finan cial Table of Contents 20005. Editor, Dallas L. Salisbury, Employee Benefit Research Institute – Education and Research Fund, 1100 13 Street NW, Suite ? Participation in health risk assessments (HRAs) increased by 50 percentage points among members of for the ana non-union em lyses, whic ployees Notes:h i (4n 1.8 cluded h ), compare ealthd cwi are services th 38.3 amon spen g mem ding as o bers o utcomes. Spous f unions withou es, p t the artn ince ers, a ntive nd oth and 3 er 7.6 dependents among preferred h Medicine eal 54, th pl nan. Also, o. 11 (Novem as shown ber 20 in 12 Fi ): gure 11, 59 1364-1373.perc ent to 68 percent of nonparticipants would participate if by Various Cost-Sharing In educa centives, 2014 tional, and scientif ...................................................................................... ic functions of the Institute. EBRI-ERF is a tax-exempt organizatio ........... 12 n Wellness programs usually include health-risk assessments (HRAs) and biometric screenings. HRAs are meetings, and other means, which likely brought about a greater focus on health. Participation in HRAs among Yet, use of financial incentives to encourage participation may continue to grow as employers seek ¯ ways to avoid the 868, Washi dif and m fereultiple s nt times ngton clerosis, and a ., DC 20 Moreover, m 005. Mare rel nagi embers ng Edi atively ditd no or, S ext choos pe teph nsive com en Bl e akel whic Figure 20 Figure 18 Figure 1 p y Figure 7 Figure 9 Figure 3 h gro ,are Emplo d u with p y 1 eto belong to—the e Be non-spec nefit Res ialty me earch I decision n dications. I stitute – Ed to bar n uca fact, the gtain in the inc ion and Re adde searc d entives h incentive for BIO participation was only in place for Test Group 2 in 2013. Thus, reports the incentive effect on BIO Impact of Biometric Screening Participation on Prescription Test Group 1 = No incentive for biometric screening, incentive for health risk assessments i n 2013. Introduction .................................................................................................................. Figure 5 ........................................ 4 th members o unions th f the uat b nions argained in th with the incee incentive, and increase ntive. Utilization of emergency d 22 percentag departments e points was also hi among non-uni ghest among on employees. non-union were also excl member cost-sharing for o uded from thue analysis. tpatient supported b physician visits After applyin y contributions g or th pr eses e criteria, and g cription rants. dr the ug fina s was l a n increas alyticaed or l file was com reduced. p rised of 71,982 Fund, 1100 13 Street NW, Suite 868, Washington, DC 20005. 8) Owner: Full Name: Employee Benefit Research Institute – Education members o questionnaires f unions that that indivi di dua d not bar ls can com gainp in let the e to premium evaluate t discount a heir health lso i risnkcreased from s and quality 55 of li percent fe. HRAs c in 201 ollect 1 to data o 61 pe n rcent T ¯ est Impact of Gr ¯ oup 2 = Reasons Percentage Percentage of Incent ¯ Financial Incentiv Employ Biometric Screening Participation, ive f for Not Participating or bi of Firms Offering omer Offers etric scr Indiv eening i iduals es Cash n 2013, on iReporting in Incentiv Employ ncent Biometric Screening Participation Employ ive for heal ees e or er tthat They h rthe Opportunit iRew sk assessm W 201 ellness 1a ?rd 2013 ent W for s i Program ould n 2012 and 2013. y to spendi excise t ng associated wit ax on high-cost h he th ale combine th plans under d inc PPACA, rease in f whic ills h of 0. takes e 02 was ffect $ i 27 n 2 per 018. mem Moreo ber per ver, ther year—a e is ba out one growing -hal body f of the of partici was not m pation; ade and by indiv , idua , and ls, but D are eac collect rug Utive h e iliz lx y at th a pe tio cted to ne u an nion b d C e e leve o qs ual ts l. b to z Ty here T ero. hfe o rre, the apeu study tic Cla design ss is quasi-experimental in Pomeranz, Jennifer L. "Participatory Workplace Wellness Programs: Reward, Penalty, and Regulatory Conflict." Milbank Percentage of Large Employ ers Offering Financial Incentives for Completing and Research Fund. 9) Known Bondholders, Mortgagees, and Other Security Holders Owning or Holding 1 Percent or More of Total Amount members, employees Figure 12, Fi (1 whi n22.4 ancial I ch inc per luded bo ncent 1,00ives for H 0 in th unio dividuals), n a ealth nd compare Risk Assessments non-union d em withployees. 111.6 (HRAs) an amon g mem d Biom be ers o tric Sc f unions reenings withou (BIO t the ), by Employ incentive e and e Background on Wellness Pr Control Group = ograms No incent ............................................................................................... ive for biometric screening, no incentive for health risk assessments. .......................... 4 health status and behavior, as Complete Amo Probably Participating in ng wellThose Offered but Not as Health Risk me Participate in Employ dical history details, inc WA ellness ssessmen Program, Participating tludin s, er by W g those ellness A Firm Size, mon in of g Program, 2014 Program, the Wo2008 rkers individu ?2014 al’s family. It is common in 2012 (Control Group). However, participation fell to 41 percent in 2013, a finding that is yet unexplained. evidence that sug 100% gests workplace wellness programs may reduce health care spending, although the extant literature that mem overall increas bers wer e in pr e essentia escripti HRA on dr lly ran s, and Maximum ug domize spending from biom d to the Value of Incentiv three groups. T etric screenin he eestimate a , gs. Large Employ pproach ers, , wh2014 ich is explained in detail in of Bonds, Mortga 60%ges or Other Sec Prescriurities: ption D None. rug F 10 ills ) Tax Status (For completion by nonprofit organizations authorized to mail at nonprofi Statistical Impact t Quarterly ? Participati 93, ao no. 2 n in the biometr (June 2015):ic screening p 301-318. rogram increased 55 percentage points when financial incentives EBRI Issue Briefs is a monthly periodical with in-depth evaluation of employee benefit issues Preferrred provider organization. 106.7 for Group an the Control Group. d Plan Year, Visits 2011to ?2013 ............................................................................................................. specialists and primary care physicians were about the same across the three 13 Whose by V Employ arious Cost-Sharing Incentiv er Offers Wellness Program, 2014 es, 2014 rates) The purpose, function, and nonprofit status of this organization and the exempt status for federal income tax pur 55% poses: Has not practice for employers to contract with third parti Sm eal s to l Em admi ployers nister H LaR rge E As in or mployers der to receive de-identified results on their b Prior Literat Wellness on this topic Program Participation Effects on Utiliza ur is not e .............................................................................................................. without Ther limitations, as apeutic Classes will (T be reat discussed ment Condi tion of in tiogreat ns) e Health Care Services and r detail below. ........................................ 6 89% Spending the Appendix, should yield unbiaseand trends d estimates of t , as well as he im crit pact of ical anal the e yses n of hance emplo d fy inanc ee beial inc nefit polici entives o es and pro n HRAs an posalsd . EBRI 70% were provided. Exclusive provider organization. Our Prior Literature Study Desig changed during 70% preceding 12 mo n nths: 501(c)(3). 11) Publication’s name: EBRI Employee Benefit Research Institute Issue Brief. 12) Issue cohorts. The n 90u %mber of prescription drug fills was highest among non-union employees (16.2), compared with 15 To continue receiving the $20/month premium discount i 20 n 12 2013, 20 non 13 -union employees not only had to continue Quan, H., et a 70%l. "Co c ding Algorithms for Defining Comorbidities in ICD-9-CM and ICD-10 Administrative Data." Medical employees in return. The information collected is used to provide custom feedback to participants about their current Statins ( Notes Dysliis a monthly pidemia) periodical providing current information on a variety 0.05 *** of employee benefit Health maintenance organization. biometric screening participation. Date for Circulation Data Below: August 2015. 13) Extent and Nature of Circulation 64% : a. Total Number of Copies: Average No. Copies Each 65% To inv Figure estigate 13, Sa 50mple M % wellness-program-parti eans, by Cohort, 2011 cipati............................................................................................................ 15 on impacts on utilization of health care services and spending, the study Data and M 60e % thods ............................................................................................................................................... 10 You can make changes on your own 68% There among m is ae small mbers of (and oft d unions en without controversial) the incentive but growi and 1 ng bo 4.8 amo dy of literat n 28% g meumbers of re that has unio evaluated variou ns with 33% the incentive. s aspects of participating in HRAs, but they also topics. had tEBRIef o partici is a week pate in the biometri ly roundup of EBRI r c screeniensearch g program. Accor and insights, as well ding to the biva as updates on riate Activity in t This paper an he alyzes data employer Point of’s ser HR fro vic S m e heal As S a lar R and Is/ th pl Sg N an. biomet e em RIs (D ployer that eric s pressi creen on, en ing Anxi h p a ernced fi tograms occur y) nancial inc red entives to primarily during open encoura 0.03 *** ge par entrollment season— icipation in its Issue During Pre Conclusions Care 42, nco eding 12 Month . 11 (Novems b: e393 r 2; No. Copi 005): 11es of Single Issu 30-9. e Published Nearest to Filing Date: 393. b. Paid and/or Requested health publications risks, and an action plan for addressing them. ? Biometric screenings led to an average increase of 0.31 annual prescription drug fills, with related 4 80% 61% Test Group 1 e used the following model to estimate using ordinary least squares with robust standard errors clustered by person. surveys, studies, litigation, legislation and regulation affe 2 cting employee benefit plans, while 60% Consumer-driven health plan. Circulation (1) P workplace welalness pro id/Requested gram Ou s. Recent tside-Count revi y Ma ews in il Subscriptions 59% clude Baicker, Cu Stated on Fo tler rm and 3526: Average Song (20N 10), o. Copies Each Issue During Prec Baxter, et al. (2014), Goetzel, eding et findings, participation in the biometr ACE inhibi ic scr tors (eH enin yper g teprogra nsion)m increased from 22 percent in 20 0.12 03 to 8 *** 9 percent in 2013 Study Sample .................................................................................................................................................. 10 the mont workplace hs 60wel of % October, lness prog Nov rame s. It has t mber, anw d D o r ee cemb search o er prec bject edin ives: First, to g a new be nef examin it year. e th Although the employer e effect of financial incentives o had these n The study also investigated wellness program participation impacts on utilization of health care services and spending. This Figure paper co 14, spending higher by $56 per member per year. O Dist 60%nfirms that ribution by fi Ty nancial pe o f Health incentivPlan, by es—on the Cohort order o , 2011 therwise, f $240 ............................................................... per em no sign ployee per year— ificant effects of participati were successful on in HRA ............... 15 at s or 12 Months: 253; No. Copies of Single Issue P EBRI’s Blog T ublished Near est Group 2 supplements our regular publications est to Filing Date: 253. (2) Paid In-Count , off y Subscriptions St ering commentar ated on Fo y on questions rm 3526; Total spendin g and spending by type of health care service is shown in Figure 16. On avera 55% ge, baseline annual al. (2014), and Mattke, 40% et alT . (2013). hyroid hor Pu mone blished stu s (Hypotd hy ies roiidi nclude e sm) xamples of both successful and 0.02 *** unsuccessful programs, among non-u Biometric nion 70% employe screening es. Again, t programs his was the collect kin information d of effect on that physical would char have acteristics of the i bee55% n expected fro ndivi m the dual chan such as ge in welln wellness pro ess prog gr rams in am partic place for several ipation, and secon yead rs prior, t , to estimate he firthe m markedly impact o al f welln tered the inc ess progentives offered ram participatioto members fo n on utilization of r their Dependent va Study Design 50% riables .................................................................................................................. included six utilization-of-hea 53%lth-care-services count variables. These were ................................. 10 the annual number of: Average No. Copies Each Issue During Preceding 12 Months: 46; No. Copies of Single Issue Published Nearest to Filing Date: 46. (3) Sales encourabiometric scr ging widesprea eenings d Ypa ou d rticipation o non utilization of h ot havrece e en in oug ived from t h tihis employer’s me to par news tie cialth car pat reporters e workplace e serv , po 27% lic ices and y we makers llness pro ,spen and others ding were fou gram, speci . The 33%EBRI Databook fically nd. health risk on Employee 52% 52% (3) 51% spending was highest among non-union employees ($3,679), compared with $3,588 among members of 51% unions that Biologic resCo pon ntrol s Gr e m ouodi p fiers (Some Cancers and Throug and ma h Dealers in findi and Car ngs from riers, Str this research a eet Vendors, C 51% re o of unter Sales, and ten at odds. For exam Other Non- ple, USPS tw Paid Distribution: Average No. o recent reviews examined the retur Copies Each Issu n-on- e financi height, w al ie nicen ghtt, body ives (Figure mass in 18 dex, ). M bloo embers o d prefs un sure, choles ions did nto et receiv rol, and e any glucose level. financial inc The entive fo screenin r gs are participatin used to i g in th dent e ify Benefits is a statistical rF efer igur enc e 15 e work on employee benefit programs and work force-related partici health Figure p care services an 15, ation Mean Utili 50 during the 2012 an % zation d spendin of Heal d g. 2013 th As Car sho pl ean years. Servic wn in prior res es, by Cohort, 2011 earch (Fronsti ....................................................................... 15 n and Roebuck 2015), voluntary wellness 1) inpatient hospital admissions, 2) inpatient hospital days, 3) emergency department visits, 4) specialist physician assessments and Descriptive Statistics biometric ........................................................................................................ screenings. In other work, EBRI has reported that these incentives ................................ 14 brought in less-healthy 50% Autoimmune diseases such as Rheumatoid arthritis and During Precedin 50%60 g 12 Months: % 0; No. Copies of Single Issue Published Nearest to Filing Date: 0; (4) Other Classes Mailed Through the USPS: did not bargain in the 47% financial incentive and $3,382 among members of unions that did bargain it in. Otherwise, the investment (ROI) from wellness programs. issues. The 2010 review estimated that medical spending declined an average of biometric individuals scre at 30% high enin risk for g progra c m, yet th hronic co ere nditi was a sli ons such as ght indi crease abetes, in parti hypert cipation ension, a rates. nd dyslipidem The partici ia. O pation rate incr verall, about o eas ned from e-half programs ? The largest in disproportionately crease in medication utiliz Mattract relativ ean Utilizatieo ly healt n of H ation h ey in alth as di C vidu a resu are als. Se lt of biometr This s rvicees lection , by C ic screening bi ohort as plagues , 2011 was f all relat or statin ed obs, whic servatio h are nal visits, 5) primary care physician visits, and 6) prescription drug fills (30-day adjusted)—including overall and therapeutic indivi Y Average No. C is a d vector uals—th o p of six o ies Each Iss se argu healt ably h ue D c i are services C n uring Preceding r most ne ohn's) ed o utilization 12 Months: f the program count 0; No v . Copies of Single Issue P (Fro ariables: nstin an 1)d in Roebuck patient hos ublishe 2015 pid Nearest to t). T al admissions 0. his 01 f*** inding Filing Date: , 2) in is not sur patie 0; c. To pn risi t tang, l Paid Endnotes it Results ............................................................................................................................................................... 14 distribution 40 o %f total spending was about the same across all three cohorts. Whe Figure n HRAs 16, Mean S were first intro p Yen ou d din o not n g o duced, eed n Healt it becau employe h se Car you ae re al Services, ers wer eady heal e o thy by C ffere od a hort, 20 19% $50 gift 11 ....................................................................... 16 card to com 29%plete one. Beginning with the 2012 about $3.27 for every $1 spent on wellness programs (Baicker, Cutler and Song 2010). The more recent (2014) paper 9 percent and/or R of larger equeste employers offer HR in 2012 d Circulation [Sum of 15b. (1 to 17 percIm As or biom ent amon munosupg m ), (2 petri ressa )e ,c screeni (3), an mbers of nts (d (4 Aut n un )] Average No oi g pr mio mograms, wh ns that une disea . Copie barga ses er s sieas ned in Each Issue During Preceding 12 uch about as the one-thir $20/month d of smaller em premi Months: um discou ployers 299nt ; No foffer o . r widely used to treat high cholesterol. T Th est is ther Group 1apeuti : Selecc t clas Tests acc Group o 2: N unted for one on- Control G -sixth of th roup: Othere ov erall increase studies and m 50 u% st be accounted for. A key strength of the present analysis is that it makes use of a quasi-experimental 1 class-specific medication use. The study also examined seven spending measures derived using allowed amounts from hospital days, 3) emergency department visits, 4) specialist physician visits, 5) primary care physician visits, and 6) given that r40 elu %ctant enrollees may require compensation for their opportunity cost of time. See Sponsor highlights spreadsheet Table 1 at http://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends- Copies of Single Issue Published Nearest to Contact EBRI Filing Date: 299 Publications, (2 . d. Free Distribution b 02) 659-0670; y Mail (Samples, complimentar fax publication orders to y, an (20 d other f 2) 775-6312. ree): (1) Financial I 40 40 %% ncentives Effect o rhen um Wellness Pr atoid arthritogram Pa is, mu Uln tiipl on e Me rticipation scmb leros ers is , an ................................................................. d C Unir on ohn' Em spl )oyees Uni0. on M 01 em ***bers ............ 14 plan reported a year, t nh a e vannual erage ROI o financi f $1. al inc 38, entive and th we authors as effectively also r note aised to d that $240 scie in th ntific 38% e form quality ofamong t a $20/mo 36% he nth r papers surveye eduction in th d had e 32% partici HRAs and one-quart pation in HRAs. It er offer incr e biometric ased from screenin 17 perc gs (Fi ent to gure 2) 29 perc . However, ent amonthe g mem percentage of smaller bers of the Control employers offer Group. Increase ing d design to in prescription drug utilization. Sec derive unbiased results due to this pote ond were an ntial confoun tidepressants, followed by ACE inhibitors (f ding. or claims data. These were spending on: 1) inpatient hospitals, 2) emergency departments, 3) specialist physicians, 4) 20% Outside-Cou prescription dr nty as Stated on Fo ug fills (30-da rm y 3526: Average adjusted)—in No cluding overa . Copies Each Issue During P ll and theraperu eceding 12 Mon tic class-spectific m hs: 35; No. Copi edication es of Single Issu use. Y also e Subscriptions to EBRI Issue Briefs are included as part of EBRI membership, or as part of it a Results and-Reports/NationalHealthExpendData/NationalHealthAcco (n=15,31 untsHisto 2) rical.html (n=40,5 47) (n=16,123) Figure 17, Participation Variab in le Health Risk Assessments, by Cohort, 2011 ?2013 ............................................................. 17 40% Prescription Drug Spending person’s healt Orders/ h insurance premium for non-union employees. The intent was to increase participation in HRAs. improved Wellness Pro over time gram Effects on (Baxter, et al. Utilizatio 2014). n of Health Care Services and Spending ..................................................... 19 member Published Neare HRAs has bee -communication n st to Filing Date: growing r ef ecfe o35 ntly, rts that reac ; (2) I whil n-Co e it unt hed all has be y as Stated on Fo employees may en mostly rmflat 3526: Average for have larger em contribut No. Copies Each Issue During P ployers. ed to these Since 2008, higher th partici e percenta receding 12 Mon pation ge rates. of ths: 9; Of course, hypertensi the pressin on), and thyr g question is: Are oid hormones (f these incentives or hypothyroidi worth it sm). for employers? In the one year following the 30% $199 annual subscription to EBRI Notes and EBRI Issue Briefs. Change of Address: EBRI, primary care physicians, 5) other outpatient, 6) prescription drugs, and 7) total healthcare. contains the following T seve he progn spe ram is not ndi cong me nveniently asures derive located for you d using allowed amounts from claims data: 1) inpatient hospital 15% 27% Immunosuppressants (Autoimmune diseases such as Inpatient hospital admissions 30% No. Copies of Single Issue Published Nearest to Filing Date: 9. (3) Other Classes Mailed Through the USPS: Average No. Copies Each Issue Additionally, a subset of unions bargained in the premium discount for its workers beginning with plan year 2013. Other 2 1100 13th St. NW, Suite 878, Washington, DC, 20005-4051, (202) 659-0670; fax number, smaller em30 plo %yers offering HRAs has increased from 8 percent to 32 percent (Figure 3). Among larger employers, comp Conclusions letion of 30% ............................................................................................................................... health risk assessments, no effect was seen on utilization of health care services and spen .......................... 19 ding. Biometric 23% Although health risk assessm 30% rhe ents and bi umatoid ar ome thrittiric screenings were generally complete s, multiple sclerosis, and Crohn's) d during the fourth $15.31 quarter of a given spendi Figure Subscriptions ng, 18, 2 Bio ) emetric Scr mergen (per 1,00 cy department spending, eeni 0 indi ng vPartici iduals)pation, 2011 3) spec ?2013 ........................................................................................ 14iali .2 st physician spendi 15.2 ng, 4) primary care 15.4 ** physician spending, 5 17 ) Financial Incentives Effect on Well During Preceding 12 Months: 0; No. Copies of Single Issue P ness Program Participation ublished Nearest to Filing Date: 0. e. Free Distribution Outside the Mail (Carriers 10% 29% Prior investigators have acknowledged several common methodol 21% ogical challenges of program evaluation. First, many (202) 775-6312; e-mail: subscriptions@ebri.org Membership Information: Inquiries Figure 19 shows results from the multivariate regression model o 22%f participation in HRAs. Separate estimates are union m ? Biometric scr embers continue eening also led to significantly higher d to receive the $50 gift card when H utilization of bi RAs were complete ologic response modifiers d. and As previously Background on Wel discussed, a ke lness Programs y concern in this analysis was selection bias. To safeguard against obtaining biased 51 percent offered HRAs in 2014, up from 47 percent in 2008—though most of that growth occurred between 2008 and screenin of other me gs, on ans): Average No. C the other han op Bd iies ol , incr oEach Issue D gice rased prescription esponse m uring Preceding odifiers dr (Su o12 Months: g m ut e C ilizatio ancer0 s n a ; No and n. Copies of Single Issue d spen 5% ding in the first Publishe year post-partici d Nearest to Filing pation. Inpatient hospital days (per other calendar year ( outpatient e.g., 2011), pa spending, 6rticipation in those programs resu ) prescription drug spending, and 7) total lted in premium discounts in the following plan ye healthcare spending. As noted above, ar (e.g., X References ............................................................................................................................... regarding EBRI membership and/or contributio 18% ns to EBRI-ERF should be dir ........................... 21 ected to EBRI 20% studies have small sample sizes. Evaluatio 17% ns based on too few observations are often underpowered, which leads to presente As described a Date: 0. f. d Total for Free Distribution members o bove, financ f u ial nions that [Sum of 15d ( ince 2%ntives fo bar 1), g r (2), ( HRAs w aine 3) and d in t e (4 he re c )]: $20 Average No. C han/m geonth d from a pr oemium pi es Each I $50 gi discou ft ca ssue D rd nt for to a $ uring Preceding partic 20/mont ipation in 12 Months: h premium HRAs 44 dis starting ; No. count immunosuppr You j esus sant t do no s.t k The now eno se speci ugh about athe lty m progredication am 9% s are used to 27% treat autoimmune diseases, such as 20% Autoimmune diseases such as rheumatoid arthritis and estimates of t 2009. The lar 20% g he dro e im 1,pact o 000 i p in nd 20 fi t v12 ih dua e and subse well ls) ness pr que ogram, this nt growth 37 stud in .62013 y con is lik duct ely a ed two 41.8 statistical ano analyses. First, it maly. 38.3 estimated linear fixed Employers of all sizes utilize various types of health management solutions. These include wellness programs designed Unsurprisingly Figure 19, Impact of , statins w Financ ere ial Inc the most entives o positively af n Particfecte ipatio dn in H therapeutic ealth Risk class, which Assessments (HRAs) one might expec .................................. t since cholesterol . 18 20% 2012). Using the latter notation, a contem President Dallas poraneous model structure was adopted because HRAs and Salisbury at the above address, (202) 659-0670; e-mail: Majobiometric scree salisbur r Reason y@ebri.org ning represents a vector of time-varying, person-level covariates including geographic region, household size, health Copies of Single Issue Pub 20% lished Nearest to 14% Filing Date: 44. g. Total Distribution (Sum of 15c. And 15e.): Average No. Copies Each Issue Prior to few statisticall 2013, 0% y signi all employees ficant results. Furth were offeree d a rm nominal ore, analyses reward on subpopul for their pa ations (e.g., rticipation in patients w biometric scre ith che ronic con nings inclu ditio din ns) g for non-union employees in C 201 roh2, a n's)nd for members of unions that bargained in the incentive $12.10 starti ** ng in 2013. There in 2 Endnotes 013 (T ...................................................................................................................... est Group 1), and non-union employees who started 17% receiving the premium discount ...................................... 21 in 2012 (Test Group 2). rheumatoid arthritis and multiple sclerosis, and are relatively expensive compared with non-specialty effects models, a met Emer hg od t ency h de at has th partmen e a t vid siv tsantage of controlling for any time-invariant personal characteristics that may 17% to promote levels were he include alth an d in t d preve he biometric nt diseas scre e, as enin well g. Oth as disease mana er evidence of geme chrn ot nic disease interventions may also have that are designed to mana been uncovered ge During Preceding 12 Months: 343; No. Copies of 16% Single Issue Published Nearest to Filing Date: 343. h. Copies not Distributed: Average No. insurance pl results received by the memb an type, annual er at the end o wage amount, num f the previous ber of ye year ars of te can be nure expected to largely impact utilization of health with the employer, and the Charlson Como care rbidity 10% 10% So urce: EB RI analysis based on administrative claims data. Minor Reason books and random drawings for movie tickets. One year after the $20/month reduction in health insurance premiums suffer the same fat (p ee . S r 1,00 econd, investigat 0 individuals) ions that examin 111.6 e only one employer 122.4 do not generate r 106.7esults that are B was no Since oth the partic chan 20ipa 12 an ge in tion is volu d 20 financia 13 ef nta l i fects are nry, employe centives refor lati rs ve HRAs for mem sometimes off to 2011. Onb ly the co er er fi s of nancia un efficients from th ions l in that centives to did note r per bargain esgression uade in employ the rela premium ted to ees to take the impact of discou part in nt. Copies Each Issue Du Figure Editorial Bo 20, medication Impact of ard: Dallas L ring Pr s. Financ The added .eceding 12 Months: Salisbur ial Inc y, publisher sp entives ending associ ; o Stephen Blakely 50 n; No. Copies o Biomeated with tric S , editor fc Single Issue P reening . Any the combined inc views Participation ublished Nearest to Filing Date: expressed in this p ................................................ rease in fills of 0.02 was $2 ublication and th50 ose o . i. Total (Sum of f the author 7 per s sho ......... 18 15f. And uld 8% confound the analysis. Unmeasured variables th -3%at change over time such as a person’s diet and exercise, however, may patients with You a c re wo hronic co rried that yo nditi ur emo plo ns. Among yer will know your la p rger ersonal employ health ers, about four-fifths use disease management and health Appendix–Eco during the 10 bio % nometric A metric scree nalysis ning process, le ................................................................................................. ading to initiation or improved adherence on pharmacoth........................... 22 erapy. 10% services and spending in the next plan year. Index score. IndividualNo fixed tes: effects are included in the model 10% as 9% deno25% ted by p, and u is the idiosyncratic error term. 10% i it 15g.): Averag not be ascribed to the officers, e No. Copies Each Issue During Prec trustees, members, or eding 12 Months: other sponsors of the E 393; No. Cm opies of Single Issue Published Neare ployee Benefit Research Institute, the EBRI Educ st to Filing Date: ation and 393 . j. Specialist p inh fo ysi rma ci tio an n visits 1.2 1.7 1.4 was intro generaliza duce ble td for o broa HRAs der populations. T (in 2013 7%), non-u hird, the t nion employe iming of es th were e evaluat also rie oquire n is also o d to com ftenp an lete issue. Sometimes, workp biometric screenings to lace According to t 10% he bivariate results, participation in HRAs increased from 66 percent in 2011 to 95 percent in 2012 the workplac financ e ial in wellness pro centive og nram pars. These can ticipation are be provi presente ded i d. Par n a num ticipation ber in of differ HRAs increase ent ways, such d by 50 as discou percentage nts and points surch amarges ong member per year—about -10% one-half of the overall increase in prescription drug spending from those who still pose a problem under this approach. Consequently, the study also estimated instrumental variables (IV) regression 0% assessments, two-thirds use SSRI = S tele elecphon tive see or roto niweb-base n reuptake id nhhe ibito alth rs; Sor NR lif I = S estyle coachi eroto nin-norepi ng, neph on rine-hal e reupf t ak employ e health advocate Percent Paid and Research Fund, /or Requ or theirested Circ staffs. Nothin ulation: Average No. Copies Ea g herein is to be construed as an attem ch Issue During Preceding 12 Mo pt to aid or hinder the adoption nths: of an 87 y pending le %; No. Copies of Single Issue gislation, regulation, All other variables are as previously defined. Select Union Members (Test Group 1) Non-union Employees (Test Group 2) Financial Incentives Effects Primary care pon hysi Wellness Program Participation cian visits 2.3 ................................................................ 2.0 2.0 ............ 22 2011 2012 2013 continue Figure 21, to r Im epact of ceive thWell e disc neount. Union ss Program on em Utilizat ployees iodi n of d not re Health Ca ceive re finSaencial rvices an incentiv d Spendi es in ng pla..................................... n year 2013 to complete 20 wellness programs are studied too soon after implementation. Utilization of health care services and related spending members o 3 among to premi non- ums, f u union nions that reductio employe ns in in bar hibi c e tog s ro (T aine sst sharing, ; Aest Gro Cd E = in t A ng he up io gi ti en ncentive 2), ft car sin- and cod ns, pri was v for erting pl 9 z e e3 nz an year givea p ym ee. rcent w 2 ays, an in 013 2 013—a (Test d contri Grou n e butio ffp ect 1n ) that , an s to a h d woul increas ealth d be savings acco ed ex 22 pecte percent d (Fig un age ure t Published Neare As yet, this st participated i udy is st to Filing Date: una n biometric screenings. ble to exami 87%. 16. Pu ne t blication of Statement he wellness program’s of Ow effect beyond nership Publication: Will be one year. printed in the As more data become availabl August 2015 issue of this e, or interpretative rule, or as legal, accounting, actuarial, or other such professional advice. www.ebri.org models. In this analysis, explicit use is made of the quasi-experimental design of the financial incentives rollout in order services, and The CCI is anabout t index that the wo-fifths off oretically ranges from 0 to er on-site fitness facilit 37, with higher scores indicating greater disease ies, gym discounts, and face-to-face health or l burden. ifestyle coaching 0% 10% 20% 30% 40% 50% 60% 70% Prescription drug fills 15.0 16.2 14.8 0% publication. 14) Signature and Wellness Pro 0% gram Partici *** St Title of Editor, Publisher, Busi ap tis ati tico an Effects on lly significant at tUtilizatio he 0.01 le ness Manager, o ve n l. of Health Ca r Ow re ne Sr: Dallas Salisbury, editor; Emplo ervices and Spending .................................. yee Benefit Research 22 biometric may be ex screenin pect 0% ed to g, tho incre uase gh they did recei during the firv se giv t year of eaways i a wenllne pr ior years. ss progra m, when health issues are first identified and 17). points amon (HSA). In 20g 14 non-u , 36 perce nion em nt of lar ployees ger empl for plan year oyers and 1 2012 8 (Test perce n Grou t of small em p 2). Note ployers again t offer his was ed a f exina pec ncial i ted, sinc ncent e th iveer to e was it will be possible to detect whether or not HRAs and biometric screenings lead to changes in member behavior, to derive effect estimates unbiased in the presence of all confounders whether fixed over time or not. Surprisingly, the 0%Source: EBRI analysis based on administrative claims data. (Fi HRA and BIO gure 1). Sm are in aller Pemploye er dicators cent Offering rfor wheth s also employ Financial Ince ent r a ive to n in health dividual mana cogement solut mpleted Less Than $either 500 ions, althou program. B gh t echause HRA ey are $500 less l or Mo and BIO a riekely than re com larger pleted So20 ur08 ce: EB RI analysi20 s ba 09sed on adminis 20 tr10 ative claims data. 2011 2012 2013 2014 Institute, publisher; Stephen BlakelyHe , managing ed alth Risk Assess itor. Date: 08/ ments 15/2015. Biometric Screenings Figure 22, Impact of Biometric Screening Participation on Prescription Drug Utilization and Costs by Therapeutic EBRI Issue Brief ** Stat is is rte ic gister ally sed in the U. ignificant atS. th Patent and T e 0.05 level.rademark Office. ISSN: 0887 ?137X/90 0887 ?137X/90 $ .50+.50 4 Notes: Source: EBR Reduc I ed analy drug sis based on cost sharadministrativ ing e Inclaims crease data. d drug cost sharing Reduced office visit cost sharing Increased office visit cost sharing Complete Health Risk Assessments treatments and/or preventive services are sought. Goetzel, et al. (2014) and Grossmeier, et al. (2012) suggest that no chan partici Given that t pate ge in in h fa we e dependent inancial llness incpro ent vag riables had no ives in ram (Fi pla gur ne ye 4). Most n- ar 20 normal distributio 12 empl for m oyers used a c embers o ns, fixed effects Poisson models f unio ombination ns that barga of premium ined in were al the disco prso run, but res eunts an mium discount d giveu aways. lts for improvements in healt Notesh :, and ultimate ly, reductions in utilization of health care services and spending. IV coefficientsT wer est Group 1 = e quite No incentiv similar in e for biometric botscreening, h magnitu incentiv de e for health and signific risk assessme ance nts in 2013. to their naïve counterparts, suggesting that employers prior to the to start of No do s tes: o. a given plan year, the study model precludes obta ining incorrect estimates of ? and s due to Given that Class ......................................................................................................................... the changes in financial incentives for participation in HRAs and biometric screening pro ............................. 20 grams occurred for Test Group 2 = Incentive for biometric screening in 2013, incentive for health risk assessments in 2012 and 2013. Test Group 1 = No incentive for biometric screening, incentive for health risk assessments in 2013. Source: http://kff.org/rep Test Group 1 ort-s =Ne oct iion/ nce ehbs ntiv-20 e fo 1r 4-s bie o cm tion et-rtw icel sv c e-w ree ellness- ning, in progra centim ve s- and for -hea health-r lth riisk sk -as as ses se sm ssen mts en / ts in 2013. three or more years may be needed in order to detect population health effects. Nyce, et al. (2012) suggest that it may Some made H Source: SA contributio EBRI/Greenwald & Associates ns, while Consum few used er Engagement altered patient in Health Care Surv eycost-sharing. , 2014. plan were quantitatively similar. year 2013. However, it was found that participation increased 11 percentage points in 2012. As mentioned above, confoun I certify that all information fur ding due to Source: Control G EBR time-var roup = N I/Greenw o nincentiv ished on this form is true ald & Associates ying e for biometric factors w Consum screening, er as not at Engagement no incentiv and comple in H issue e for ealth C health .te: Stephen Blakely For are Surv risk assessment brevity ey, 2014. a s. nd sinc , Editor and e IV estimators are l Director of Commu ess effici nications. Date: ent, the Source: Source: Test http://kff.org/rep G EBR roup 2 = I/Greenw Incentiv ald & Associates ort-s e e for biometric ction/ehbs -20 Cscreening onsum 14-seer ction Engagement in 2013, -twelveincentiv -wellness- in He ealth C for health risk progra are Surv ms-and assessmen ey -hea , 2014. lth-riskts -as in ses 2012 smen and 2013 ts/ reverse causality. However, because HRA and BIO are voluntary programs, participants may differ from nonparticipants Test Group 2=Incentive for biometric screening in 2013, incentive for health risk assessments in 2012 and 2013. different groups in different years, this study constructed three cohorts for use in the ensuing analysis. Test Group 1 © 2015, Employee Benefit Research Institute ?Education and Research Fund. All rights reserved. 08/15/2015. take one year or longer for health spending patterns to change after worker health risks are impacted. Fourth, the stepped-up communications efforts that reached all employees may have contributed to this higher participation rate. study presents only the linear fixed effects results. See the Appendix for more details. on unobserved characteristics, which if correlated with utilization of health care services and spending may lead to Co ntro l Gro up=No incentive fo r bio metric screening, no incentive for health risk assessments. ebri.org ebri.org ebri.org ebri.org ebri.org ebri.org ebri.org ebri.org ebri.org A monthl ebri.org ebri.org ebri.org ebri.org ebri.org Issue Brief • August 2015 • No. 417 ebri.org Issue Brief • August 2015 • No. 417 ebri.org Issue Brief • August 2015 • No. 417 ebri.org Issue Brief • August 2015 • No. 417 ebri.org Issue Brief • August 2015 • No. 417 ebri.org Issue Brief • August 2015 • No. 417 ebri.org Issue Brief • August 2015 • No. 417 ebri.org Issue Brief • August 2015 • No. 417 ebri.org Issue Brief • August 2015 • No. 417 Issue Bri Issue Bri Issue Bri Issue Bri Issue Bri Issue Bri Issue Bri Issue Bri Issue Bri Issue Bri Issue Bri Issue Bri Issue Bri y research report f e e e e e e e e e e e e efffffffffffff • August 201 • August 201 • August 201 • August 201 • August 201 • August 201 • August 201 • August 201 • August 201 • August 201 • August 201 • August 201 • August 201 rom the EBRI Education and Re 5 • No. 417 5 • No. 417 5 • No. 417 5 • No. 417 5 • No. 417 5 • No. 417 5 • No. 417 5 • No. 417 5 • No. 417 5 • No. 417 5 • No. 417 5 • No. 417 5 • No. 417 search Fund © 2 015 Em ployee Benefit Research Institute 17 12 20 15 11 18 9 7 8 14 19 21 16 22 10 23 13 2 5 3 4 6