This study examines whether disenrollment from health savings account (HSA)-eligible health plans is associated with risk selection, such that once enrolled in an HSA-eligible health plan, less-healthy enrollees are more likely to disenroll from the health plan than healthier enrollees. Such a phenomenon contributes to adverse selection and death spirals. The paper focuses on the presence of health conditions and use of health care services as key determinants of disenrollment.
Key findings:
- Overall, 5 percent of HSA-eligible health plan enrollees in 2013 and 2014 switched to a different type of health plan in 2014 and 2015.
- Among those disenrolling from an HSA-eligible health plan in 2014, 6 percent were enrolled in an HMO or EPO, and 94 percent were enrolled in a PPO or POS plan.
- There is evidence that individuals who disenrolled from HSA-eligible health plans were more likely to have certain health conditions than those who remained enrolled in the HSA-eligible health plan.
- Among individuals with no health conditions, 4.2 percent disenrolled from the HSA-eligible health plan.
- Five percent of individuals with dyslipidemia disenrolled, along with 5.3 percent among those with hypertension or depression, 5.6 percent among those with diabetes, and 6.1 percent among individuals with schizophrenia/bipolar disorder.
- Individuals with multiple conditions were even more likely to disenroll.
- Disenrollment was 7.1 percent among those with a combination of hypertension, dyslipidemia, and diabetes.When depression is added to those conditions, disenrollment jumps to 8.2 percent.Among individuals with those conditions and others, disenrollment was 13.4 percent.
- Individuals with claims related to childbirth were twice as likely as those with no health conditions to disenroll from HSA-eligible health plans.They were predicted to disenroll at a rate of 8.4 percent.
- Individuals with ulcerative colitis and those with metastatic cancer were more likely than those with no conditions to disenroll.Otherwise, there was no statistically significant difference in disenrollment rates between individuals with rare and costly chronic conditions and those with no conditions.However, newly diagnosed individuals with rheumatoid arthritis or multiple sclerosis were more likely to disenroll from the HSA-eligible health plans as compared to individuals with no conditions.Predicted disenrollment rates were 6.2 percent for individuals with rheumatoid arthritis and 7.3 percent for those with multiple sclerosis.
- Use of health care services was a statistically significant predictor of disenrollment from HSA-eligible health plans.However, the magnitude of the impact was quite small.For instance, 4.2 percent of individuals who used no health care disenrolled from HSA-eligible health plans, whereas 4.4 percent of those at median use disenrolled, and 5.2 percent of those at the 90th percentile of use disenrolled.
Figure 8
Figure 1
Figure 11
Figure 12
Figure 6
Percentage of Employers Offering HSA-Eligible Health Plan/Health
Use of Health Care Services, by Enrollment Status
Predicted Probabilities of Disenrollment From HSA-Eligible
Figure 9
Percentage of Large Employers Offering a CDHP as the Only
Presence of Cancer, by Disenrollment Status
Reimbursement Arrangements (HRA), by Firm Size, 2010 –2016, With
Remained in HSA-Eligible Health Plan in 2014 Disenrolled From HSA-Eligible Health Plan in 2014
Figure 3
Fi he inc Fi R na g e a rur e lt fe lla h s e yses , r 6, w e te ae nce P ligib tw us rd ould eid senc inde le s not he ha e p a a v e of lt e llow nde h p C ra ant nge la n for c ns v e d a rt , r he afr ia b s om y b c p le om D oss s w is 2.9 p e ibilit a nr a rt s e ollm o d y c3. r t of eo 7 e a nt tindiv swit p ee dSt r c b ca e idu y hin tnt us ya .e g ls a ................................ W fr r. w o e m it G d h no e id a nde n HS not crond , A a a- c ge c it eligib oun ions , a nd tle . for ................................ g he Pe r e a og ad n ltic y h p ra toff p ela hic dsets n to a d is re e g fr nr ion w o n HR ol mlm highe e e Ar nt ........................... e or c r ra a t he t e pe m ts w u p p re oss loy e dr. e e ibili e 6.2 ty of 12
Paul Fronstin is director of the Health Research and Education Program at the Employee Benefit Research Institute
The Impact of Hea Helth alth S Plan, ta by tu Us s e a of H ndea Us lth Car e of e Se He rvice a slth Care Services
Predicted Probabilities of D Figu ise re nrol 4 lment From HSA-Eligible
Appendix Figure 6
45% Health Plan Option, 2010-2019
Projections Through 2019
5 Remained in HSA-Eligible Health Plan in 2014 Disenrolled From HSA-Eligible Health Plan in 2014
100% Variable Means by Disenrollment Status
60
See www.kff.org/report-section/ehbs-2014-section-seven-employee-cost-sharing/
A p dis r ded em it nr iona ium ollm p lp ly e ea rn , c yte e m nr nt freom oll nt for s. m H indi eRA nt s. v inf idua o W rm e ls w a ini tion a it tia h r llyhe nd eum xc alm a aim ine toid s d d a swit artta hr on cit hin isa a g lFigu l sp nd in a ouses, 7.3 p nd re 2 out erc c hil e of nt dH rfo e RA n, r s b those and ut ot foun whe ith mu rd d te ha plte tipl nde thr e e nt sc es of le -qrua osis trhe te . r p s of olictyhe hol H dRA er
BBuc(hm EBRI ue)lle . M. r, Thom Christa op s he C. r19 Roe 98. buc "Dk oe iss a pr e Fisi xe de dnt -D olla and r C CE on O tof ribut RxE ion Low conomeic r s, L SpeLnding C. Thi ?s " IH ss eu alt e h BA rie fffa ir wsa s 228 wr-it 235. ten w ith
Health Plan, Varia b by le M More Comm eans, by Disenron ollm Hea ent Sta lthtu Con s dition
Presence of Common Chronic Health Conditions, by
6%
Examples of Adverse Selection and Death Spirals
on Disenrollment From HSA-Eligible Health Plans
Employees
3.0%
Figure 7, Presence of Childbirth, by Disenrollment Status ....................................................................................... 12
Health Plan 53.9 Choi 53.9** ce, Among 5 Percent of Individuals Who
w dis ee re nr a ollm ccess enetd s r te o si dd ee te dr in a mine si w ngle het he sta r tt ehe . W polic e wyehold re ce onc r he ard ne indiv d tha idua t the l ose r fa dm atil ay w could over ask ge e w in e our a 39% c h ye result ar s if t . A he dic la ho rg tom e ous
6 40%
assistance from the Institute’s research and Dise editor nrol ial s lme taffs. nt AStatus ny views expressed in this report are87% those of the
90%
See https://www.ebri.org/docs/default-source/fast-facts 10-499 /ff-311- 50 hs 0+a-30july 5,00 18. 0+pdf
From 2013 to 2014
? Use of health care services was a statistically significant predictor of disenrollment from HSA-eligible health
Buchmueller, Thomas C., and Paul J. Feldstein. 1997. "The Effect of Price on Switching among Health Plans." Journal of
All Condi Dise tions Bnrol elow led From HSA-Eligibl Fro e mHe 201 alth 4 to 2 Plan 015
2.6% 5.2% 13.4%***
v num ariab be le r of ford h isaevnr ing olla m c eov nts w ered er spouse e due t o and an e a m coun ploy te of r th the at d num iscon betrinu of ec dhil itd s H ren R/ot A. he r dependents were also constructed.
authoB rs a uchm nd ue shlle ould r a nd notFe bldst e asecin (1 ribed 99 to 7 a the nd offic 1998 er )s, t looked ruste a ets, or ot the eff he ecrt sponso of movris of ng to EB aR fix I, e Ed m cpont loyreibu e B tion a enefit t Research
Remained in HSA-Eligible Health Plan in 2014 Disenrolled From HS 80% A-Eligible Health Plan in 2014
18% 35%
50
Remained Disenrolled
Septe Fi 7m gb ur er e 20 8, 1Use 0 • N of o H.e a 3lt46 h C are Services, by Enrollment Status ................................................................................... 13
November 12, 2018 • No. 465 2.5%
35% Remained Disenrolled
Im By Calc Pa pli ula ul Fro c tp H e at la d e a ns fro ions lt 80h . % ns m E HFi ctow ono in, g for ur ev e m Ph.D. e 6 ic P r, s in la t he 231 ht n S , tp E m -24 samplo :po g /7. /nit w ns w ude w ye o .me of r es Ben rc t he and Publ er. c Be im o em/ fi ptw a tne R ctee w w ess a n ric Po o s q e20 oa m/ uit rch I 13 me eli sm and 2 rc cns y ea r l-t l. na itut 01 tional For e 4 , ins -a he tnd aa nlth ce M. -, surve 4.2 p Chris ye -e rmp cteop nt loy he of ers indiv r R -find oebuck, idua ing-ls ne w who -ways-
5% 16.6%** 33%
It is worth noti 2.5% ng that unlike for the policyholder; spouses, children, and other dependents were not required to have
Institu the te- Univ Educeartsi ion ty of and Ca Re lifo sea rnira c h Fund (UC). B( eE for BR eI -19 ERF 94), , UC or tset i heir tst s a cont ffs. N ribe ut ition e her Eq Bua RIl nor to t he EB RI cos -E t RF lob of theb he ies or alth takes
32% 73%
Hypertension, Dyslipidemia, Diabetes, & Depression in HSA 8.2 -%*** From HSA-
72% 72%
in HSA- From HSA-
to Our -ho fin ld-dthe ings us-e lid , n e how no -onhe e -he va ea lt rlth , h ca c-le ba r ee r ne l y dfit is su -ecg nr og soll e t-st g ero d th w fr ath. om t dhtml is H eSA nr ol -elm ligib ent le fr he om alt H h SA pla -e ns ligib , wle he he rea as 4.4 p lth plan es d rceue nt t oo f th heose alth st at a m tu es m dian use ay
15.5%
16% 30%
There are ways for employers to mitigate adverse selection if offering a choice of health plans is something that the
c Ph.D. omple,t R e yx eEco ar eligib nomics ility (, i.e L .,LC to accommodate intra-year qua 4.4% lifying events like marriage and childbirth).
positions plan w on itspe h th ce if ic la r pg olic esty m pe rop mb osa ersh ls. ip, EB b Rut I inv in 199 ites c 4,om UC m reent duc on edt his the r e esea mpr loy ch. e r premium contribution to
Eligible Eligible
Figure 9, Pre70 dic %ted Probabilities of Disenrollment From HSA-Eligible Health Plan, by More Common
30% Eligible Eligible
th HMO/EPO, 6%
Bundorf, M. Kate. 2012. "Consumer-Directed Health Plans: Do They Deliver?" Research Synthesis Report No. 24 (Robert
4.2%
Hypertension, Dyslipidemia & Diabetes 7.1%***
40 63%
contr ibutd eis m enr ucolle h ledss , atnd o p5.2 p otente ia rl d cent ea tof h sp those iralsa ttha the n p90 lan sp peonsor rcentil d ee of sigus n d e e dcis iseions nrolle suc d.h a s those described above. This
8
employer desires to do. Strategically setting the employee share of the premium can be an effective way to avoid risk
Variable Total Heal61% th Plan Health Plan
See Figur the e 7. a 2 m in oun http tV c s ah : r/ ia a /r w bg lw e ed w .b ky ff .to he rg /le reapst ort c-o sst ecly tion/ plae n a hbs v-a2017 ilabT le -o s ta e stcla tion tew -H 7 ide - ee am . lth Thi p loy Pl s p a en eolic -c Ho ey sa t l- le th sha d Pl tring o aninc / re ases in
Health Condition ..................................................................................................................................... 14
59% 59%
2.0%
The I Wmp ood 14% Johns act on of Founda 35.2 Hea * tion). lth http S ://wt wa w.r tw u jf.o s rga /cn ont d e Us nt/dame /fa of rm/ He reports/ a relth port Ca s/2012/ rre wj Se f40240 r5v ices
4%
13.2%***
Sinc has e im w pe o rhy tant pot im hepsi lic ze ad tion tha s fo t he r a blt ot h s h th tae tus w a w yould emp lik loy ee lyr s off be ce or r rhe ela atlt eh p d wla itn h s a HSA s w -e eligib ll as fo le he r p aublic lth p la polic n dy is.e nrollment, we
Sugges60 ted % citation: Paul Fronstin and M. Christopher Roebuck. “The Impact of Health Status and Use of Health
N= 242,359 229,953 12,406
selection. La 25r% ger pre 33.0 mium diffS ecr he iznt ophia ren ls ia/B bie pol tw areen more and less generous p 6.1%***lans will drive healthier employees to the
N= 309,143 294,725 14,418
employee premiums in higher cost plans and a high degree of plan switching by UC employee.
9
? Disenrollme 51% nt from HSA-eligible health plans, w 22% hich includes a disproportionate number of individuals with
See Figure 7.8 in https://www.kff.org/report-section/ehbs-2017-section-7-employee-cost-sharing/
on Disenrollment From HSA-Eligible Health Plans
include Carde tSe hre rv eic sets es on of Ma D su le isc eh nrm ollm easu ent re Fr s. oFi mr s HtSA , w-e E ligib deriv lee d H e the alt h C65% P ha larns lson .” E CB om RI 65% oIrss bidit ue y B rIie nd f, e65% no. x (C465 harls (on, Em peloy t ae l. e19 Be 87 ne ) fit (D eyo,
Male 64% 64% 55% ***
le In sst r go ene duc rous p tion lan a 11.5% nd less-he 48% althy employees to the mo 48% re generous plan. Minimizing premium differentials will
Figure 10, Pr12 ed %icted Probabilities of Disenrollment From HSA-Eligible Health Plan, by Childbirth and Less Common
Buntin, MUlt elin 30 im da a tB ee ly e, uw it d ke rov s, A e 20% m one elia p la M. n o Ha ut v ilof and, the Rola UC n int d eM rna cDl em vit atr, ka end t. Of th Neerose aj Sood emp. loy 20e11 es w . "H hose ealthpcra erm e ium Spes nding and
50% Congestive Heart Failure
5.7%
19%
healt 20h co % nditions, increases employer spending on health coverage by 2.5 to 3.2 percent.
1.5% AA gg ee 4 43 1..0 9 44 32 .0 .0 42 4.0 4.6 *** ***
Cherkin and Ciol 1992) (Quan, et al. 2005), which is widely used in the extant literature as a gauge of general health
Research Institute, November 12, 2018).
t hereforeH reeadltuc h e C ond risk itsele ion c ................................ tion. Similarly, offering ................................ a choice of only highe ................................ r-deductible plans, ................................ some with HSAs and so ..... m e14
Both the num did not ber inc of rh ee aa se lth bs ea tw ving een 19 s acc 17% 93 ount and s (19 HS94 As, ) ra oug nd hly enr 5 ollm –6 p ent er cin ent H SA swi -e tc ligib hedle p la he nas. lth p lans have grown
Prev 3% entive Care in High-Deductible and Consumer-Directed Health Plans." American Journal of Managed Care
39% Appendix Figure 5
Region
10% Region
Appendix Figure 1 22.0**
36% Diabetes 5.6%***
status. For fa 40m % ilies, we retained the maximum score A A p pa p pm e en no d dng iix x F F aiill m g gu ure ree 3 4 mbers linked to the policyholder.
Appendix Figure 2 21.0
without them, will reduce risk selection. Employers can also use contributions to HSAs to reduce risk selection.
significantly since HSAs first became available in 2004. In 2017, enrollment in HSA-eligible health plans ranged from
22215 -230. % Marginal Effect Estimates From Multivariate Probit Models of Disenrollment From
By Paul Fronstin, Ph.D. Northe32% a , E st mployee Benefit Research I 16% nstitut 16% e, and M. 14% Christop *** her Roebuck,
Northeast 16% 16% 20% ***
Copyrig 20ht Information: This report is copyrighted by the Employee Benefit Research Institute (EBRI). It may be
Diagnosis Codes for Chronic Conditions
MEs arg tiim na alte Effe d Cc ot eEs ffic tiim en ats te s F r F o rm om M M ulu tilv tia vra ia rite ate Pr Pr ob oib t iM t M od oe dle sl s o f oD f D isie sn er n o rlo lm llm en et nF t rF o rm om H H SA SA -El -El igiig bilb el e H e Ha ela th lth Pl Pl an a n
Figure 11, Predicted P Es rob tim aa bte ilit die Cs of oeffiD cis en ets nr F ollm rom e M ntu Fr ltiv oa m ri a H te SA Pr -E oligib bit M le o d H ee ls a lt oh P f Dils aen, nrb oly lm Use ent of FroH m e H alt SA h C - are Services ...... 15
Similarly, Cutler and Reber (1998) examined changes that took place at Harvard University. In 1995,
Employer contributions that are based on healthy behaviors may steer less-healthy employees to the more generous
1.0% H De SA pres -El sii on gible Health Plan, First Di5.3 ffe %** re*nced Models
21.4 million to 8%33.7 policyholders and their dependents (Fronstin 2018). The number of employers with 500 or more
Midwest 24% 24% 24%
30% Midwest 24% 24% 18% ***
Ph.D., RxEconomics 10% , LLC 34%
Second, we generated a vector of indicator variables for the presence of health conditions using diagnosis codes.
used without permission but citation o Eligif blte he H e sour althc Pl e ais n ,r F eiq rs uir t D eid ffe . renced Models
23%
2% 10% 6.6%
Charlson, H M. arv Ea ., rd P . m Pov om ep de ti, o K a. fiLxe . A dle cs, a ontnd ribu Ct. ion for R. Ma c he Ke anzi lth b e. e19 ne87 fit. s. "A C ont New ribu Me tions thod w o ef C re la set a ssify t ing 85 p Pr eog rce nos nt tof ic Comorbidity
plan, if they are less likely to qualify for those employer contributions. Finally, the easiest way to avoid risk selection is
South 35% 34% 43% ***
e mployees offering an HSA-eligible health plan or healt 6.2 h r %** eimbursement arrangement (HRA) has increased from 23
So Cu on th dition ICD-36% 9-CM Codes 35% ICD-10-CM 47% Codes ***
28%
D Diis se en nr ro ollllm me en nt t iin n 2 20 01 14 4 6.0% D Diis se en nr ro ollllm me en nt t iin n 2 20 01 15 5
Hypertension 5.3%***
Figure 12, Percentage of Large Employers Offering a CDHP as the Only Health Plan Option, 2010-2019 ..................... 17
Specifically, an individual was considered to have a given condit 26% ion if he/she had 1 or more inpatient, or 2 or more
6% Disenrol 25% lment in 2015
20%
in L theong leait st udin -cost al Stu lW y e ps olic d t iey s: D fore e vm elop ploy me ee nt s e aa nd rnin Va gl ida betlow ion." $4 25% Jour 5,00 na 0 a l o y f e 26% C ahr r, oni 80c p D er is ce e 19% a nt se fo 40 r e (m 5*** )p: loy 37e3e-s 83.
10 5.2%
to offer only one health plan. 23% 5.1%
V Va ar riia ab blle e West M Mo od de ell 1 1 M Mo od de ell 2 224% M Mo od d 5.0% e ell 3 325% M Mo od de ell 1 1 15% M Mo od d*** e ell 2 2 M Mo od de ell 3 3
percent to 61 percent between 2010 a22% nd 2016 (Figure 1). And the number of HSAs reached 22 million at the end of
Report availability: This report is available on the Internet atD w isw en w r.e ollb m re i.or nt g in 2015
0. 5% 5% Variable20% Model 1 Model 2 Model 3
outpatient (on different dates) claims with an associated Candidate International Classification of Diseases (ICD-9-CM
Unknown Region 0.02% 0.02% 0.02%
1 Dyslipidemia
5.0%***
earning be17% twU en ek n $45,0 nown R00 eg io an nd $70,000 a year, a4.3 nd 0 75 .0 6 4 1p 4 % .e 2r 1cent for 0.0 e 1m %ployees e 0.0a 1r% ning mor e than
Variable Mod4.1 el 1 Model 2 M0.3% odel 3
M Ma alle e 0 0..0 00 11 77 4* * 0 0..0 00 11 87 0* * 0 0..0 00 22 21 3* ** * 3.7 -0 -0..0 11 32 85 3* ** ** * -0 -0..1 03 16 21 3* ** ** * -0 -0..1 03 14 21 1* ** ** *
2017. A T A G L A N C E 3.4***
Appendix Fig10 ur 4% %e 1, Diagnosis Codes for Chronic Conditions ........................................................................................ 18
Male 0.0000 0.00.2% 000 0.0000
1%
or Cut Ile CD r, -D 10 a,v id beM., ginna ing nd in 201 Sarah J 5) . Re cod be es r . (19 see 98 A. p"p Pe andix ying Fi for gur He e a 1lt )h I . 6C 4n 5 ond su .11ra itn ions ce: in The clude Tra dd c eon off b gest etiv we e e he n C arom t fapilu etrit eion a , nd
Public policy could contribute to risk selection in employment-based health plans, depending on how employers
Thi A Ag ge es study was conducted through th -0 -0 e.. 0 0 E0 0B 0 5RI 5 0* ** *C * * ente -0 -0 r.. 0 0 for 0 00 5 5 7R * ** *e* *sear-0 c -0 h o ..0 00 0n 0 56 8H * ** *e * *alth B -0 -0e ..0 0 ne 0 00 1fit 1 0* *s * * Innov -0 a 0 .t 0 .ion, 0 00 01 with t -0 -0he ..0 00 0 1 0 fun 3 1* ** *d * *ing
$70,000 a yM ea alr e. Non-union employees experienced 0 the .00 change in 199 0.005, while uni0 on .00employees were
0%
Asthma/COPD 4.8%
Age 0.00 60 45 0.21 0.0000 0.0000
Individual Coverage 37% 36% 44% ***
0
hypertension, dyslipidemia, diabetes, asthma/chronic obstructive pulmonary disease (COPD), depression,
Adverse Se 20 le 10 ction." 20 Qu 11arterly20 Jour 12 nal o 20f E 13 conom 20ic 14s 433-20 466 15 . 2016 2017 2018 2019
M Miid dw we es stt Individual Coverage -0 -0..0 22 54 98 4* ** ** * -0 -0..0 22 54 85 4* ** *36% * * -0 -0..0 22 65 53 7* ** *36% * * 0 0..0 00 33 74 7* ** *41% * * 0 0..0 03 07 3 *** 0 3* ** ** * 0 0..0 03 04 30 1* ** *
respond. On 0.0% 0% July 11, A20 ge18, the House Ways and Means Com0m .0i0 ttee approve 0d .0 0 a number of0 .he 00alth-related tax bills with
support of the following organizations: Aon Hewitt, PP B Olue /PO SC , 94 ross % Blue Shield Association, ICUBA, JPMorgan Chase,
While the not size 2% a of ff Ine pat c tihe t en et Hos d H unt SA pital -il 19 e Da lig ysible 96 (pe, r w he Ehic me alt rh a ge h p ncllow yla Depa n ma erd tme tr nhe k t e trp e Plrsea a ima ce ryr Ca c ha he res g P rs t hyrsow io cian look a n subst 's Spet c iaa a lis nt tt P rie ha ya ll st iy cm ian in t e 'snt Ohe ffi g ce rp oup aP st re s(15 ctrhe ip tiy o n Drug e 19 ar95 s, it Fil non ls could - be
Appendix Figure 2, Estimated Coefficients From Multivariate Prob 649it .8 1 Models of DisO e6nr .0ol 1xlm - Oe 6nt .02 Fr x om HSA-Eligible
Midwest -0.0134 -0.0135 -0.0134
Has Covered Spouse 48% 48% 38% ***
2010 2011 2012 2013 2014 2015 2016 Very likely to
Non-Metastatic Cancer Metastatic Cancer 0.9%
100 enrollees) Visits (per 100 enrollees) Office Visits Visits
This st So Soudy e u utth h xamines whether disenr No ollm Cone dint tion 0 .0 s 0 f.r 0 0om 0 19 health s 0.0 0.a 0 0v 0 1ings 9 4.2% account 0.0 0.0 00 0 1 (HSA)-e 0 0..ligib 0 11 32 96 7le * ** * * *health p 0 0..1 03 1 la 8 2 0.7% ns 7 5* ** * * *is assoc 0 0..1 0i3 1 a4 2t2 1 e* *d * * * *with
schizophrenia/bipola H rM a d s id is w C oe o rs v dte ere r, dr he Spum ousa etoid arthritis, multiple -0 sc 49% .1 le 4rosis, Crohn’ 49% -0.s d 14 isease,44% ulc -0 e.r1 a4tive *** colitis, psoriasis, non-
648.82 O4.80
the intention of expanding the number of employers offering HSA-eligible health plans and the number of people
offer in 2019
much l Kaa iser P rger e tod rma ay ne if indiv nt So e, uMe th idua rcels r, w Pe fiz ree rnot , and disP ehRMA nrollin.g from t-0 hose .0063 plans. Dis -0 e.n 00 rollm 63 ent ma -0 y. 0 b0e 6 3 responsible for what
union workers) and a control group (the 1996 union workers). The study found evidence of adverse
Health Plan ................................ Number of Other Covere ................................ d Dependents 1 ................................ .0 1.0 ................................ 0.8 *** ................... 19
0%
Deyo, R. A., D. C. Cherkin, and M. A. Ciol. 1992. "Adapting a Clinical Comorbidity Index for Use with ICD-9-CM
W We es stt 6 South -0 -0..0 33 20 28 3* ** ** * -0 -0..0 33 10 93 4* *-0 * ** *.0 67 50.x -0 -0..0 33 20 38 5* ** ** *-0.07-0 -0..0 11 10 2 O 2 9 4* *.* * 8* *1 -0.-0 -0 07..1 01 13 04 2* ** ** * -0 -0..1 01 18 07 7* ** ** *
0% Stat Siou stiN r ccal u e: m sNati ign bie on ficr an alo c Bfe us O de ine tno h se s ted r G rC as ouo f p ol v on le ow re Heal s: d * *th, D *p< e La 0.0 p rge e 1; nE *d *mp p< en 0.0 loy ts ers 5; *' p< 200.1 18 0, Heal are ba 1.th 0Car sed e S on trate Krus gyk an al 1-. d W 0 P al lan lis Des equa ign lity S of urv 0 po ey .pu 8 and lations tes*** t of
r m ise kt a sele statcitcion, canc su ec r, h th and atm on etcaest e anr tic olle cad n cin a er. n HSA In ad- deitligib ion, le a n he indic alth p atla orn, for le ss chil -he db air ltth h w y ea nr s c olrle ee as a tedr e b a m sed ore a lik lso ely on to dia dis ge nos nroll is
No Use of Health Care Services Median 90th Percentile
West 0% 2% -0 4% .0088 6% -0 8% .0085 10% -012 .0% 087 14% 16%
covered by them. If growth in the HSA-eligible health plan marketplace occurs in an environment where workers have
Source: Figure 5 in http://www.mercer.com/newsroom/national-survey-of-employer-sponsored-health-plans-2016.html.
appears to sele bec t tion: he reYcoung ent se ta r le lem d pgloy row ee th in s wetrhe e found size of to the be H m SA or- ee lik ligib ely le t o heswit alth p ch to le lan mss ar k ce otst plla y cp ela (ns Fron tha stn o in 20 lde 18 r ). We are
diffhtt erenc ps Hy ://w es pe w irn w tem .bu ns ea is oin n ne vs al sue grou by ph di e Dy sal en tsh.o lriol pirld m gem /ne enit w as stat /nb us gh . -newsDi /pres abe stes -releases/pressA -rs el th ea ma se /CO -deta PD ils/?ID=348 De pression Schizophrenia/Bipolar
Statistical significance denoted as follows: ***p<0.01; **p<0.05; *p<0.10, 6are ba 51.01 sed on Kruskal-Wallis equa Ol7 ity 5 .of 8 2 populations test of
U Un nk kn no ow wn n R Re eg giio on n 0.0 0.0 00 00 0.0 0.0 00 00 0.0 0.0 00 00 0.0 0 .0 02 35 00 .0 .0 04 35 00 .0 .0 02 21
Administrativ W e eD sa t tabases." Journal of Clinical Epidem -0iolo .09gy 45 (6): -0 61 .03 9-9. -0.09
from the health plan t In ha div n he iduala C lto hie vera r e gnr e ollees. Such a phenom 0.0e 00 non cont 0 ribut 0.0e 0s t 00o adverse 0.0 sele 000ction a Disordnd er death spirals.
codes on mother’ dis fferenc claim es is fo n mea rn d val eue liv by e r diie sen s. rollThese ment stat us fla . gs were coded to indicate if any family member had the condition.
Charlson Comorbidity Index^ 0.31 0.30 0.34 **
a A p cp hoi endix ce of Fip gla urns e , 3,e m Ma pr loy gina ers w l Effill e cne t E est d im to atta ek s F e st rom ep s t Mul o tm ivia nim riatize e P rris ob k itsele Mod cte iols n. of HD ow iseenr ve ollm r, ris ek nt sele Froc m tion cou HSA-Eld ligib be le
Childbirth 651.11 O80.x
una ware of any re Sse tatisa tic ral c h o signin fican the ce de d no eted te r as m fina ollownt s: *s of **p<0.0 d 1; is*e *p< n0.0 roll 5; m *p< e 0.1 nt 0..
employees. This resulted in a 16 percent premium increase in the high-cost plan in 1996. Non-
IIn nd diiv viid du ua all C Co ov ve era rag ge e -0 -0..0 11 00 72 0* ** ** * -0 -0..0 00 88 95 2* ** ** * -0 -0..0 00 99 95 9* ** ** * -0 -0..0 00 22 97 8* * -0 -0 .0 .0 02 14 -0 -0..0 02 08 28 6* *
Individual Coverage 0.00 0.00 0.00
Charlson Comorbidity Index^ 0.26 0.26 0.28 **
Has Covered Spouse 0.0071** 0.0066** 0.0078***
651.21 O30.x
The paper focuses on the presence of health conditions and use of health care services as key determinants of
m Tabl itigate e d H o te f o aC lt the h P o nt d Sla tat en e g is ................................ tirnt ce ale s s it gn ha ifict an p ce olic deno yted m a ask fe olr low s a ................................ s: llow ***p<0.0 fi 1; rs **tp< d 0.0 olla 5; *r p< c 0.1 ov 0, ................................ e are ba rage sed for on K trhe uskal t -r We ala lit s m eqe ua ................................ nt lity of of pod pu is lati eon ases s test o out f sid................... e of the 20
H Ha as s C Co ov ve ere red d Sp Spo ou us se e -0 -0..0 10 09 17 8* ** ** * -0 -0..0 11 21 04 1* ** ** * -0 -0..0 11 11 93 3* ** ** * -0 -0..0 11 54 51 3* ** ** * -0 -0..1 06 17 58 1* ** ** * -0 -0..1 06 12 49 7* ** ** *
random dise Hn arsollm Cove en re td c Sp ono tin usue e d. Within three yea 0r.s, t 076he 1** high-cost 0.0 7 p0la 1* n w * as no 0.long 0834e **r* offered
Fronstin, Paul. 2018. "Has Enrollment in HSA-Eligible Health Plans Stalled?" EBRI Issue Brief no. 441 (Employee Benefit
differenc N es um in b me ea r n ov f al O ue th by e r diC sen orv ol e lm re en dt D stat eus pe .ndents 0.006 35 01 *.* 31 0.0054*** O31.11X0 0.0019
Thi rd, measures of he Ha elatlh se th Co rn vd ic ite ios ut ns~ilization were also constructed. Specifically, the numbers of inpatient hospital
d isenrollment.
UndeN N ru st um m ab b nding e er r o off O O w tth hha e er r tC C d o or v vie e vre re es e d d D Dnr e ep pollm e en nd de e en nn tts s t and -0 -0..0 0 d0 6 is 6 2e 0 7nr * ** ** *ollme -0 -0 nt ..0 0 0 5in 5 75 6 to * ** ** *and out -0 -0..0 0 0 6of 6 63 3 H * ** *S * *A-eligib -0 -0..0 00 2le 2 86 3 he * ** ** *alth p -0 -0la ..0 0n 2 04 2 s is 6 2* ** * * *impor -0 -0t..a 0 0nt 2 08 2 6 6t* *o * ** *plan
deductible in HSA-eligib Num leb he er o aflt O h p the la r ns Co.v ered Dependents 0.0318** 0.0580*** 0.02
Introduction .......................................................................................................................................................... 5
Findings From Probit Regression Equations 651.41 O82.x
Re be sea caus rch I e of ns Ha e td it alv ut th e e rC se ). on ht sele dittp ios:/ c ntsion / ~w.w w.ebri.org/docs/default-source/ebri-issue-brief/ebri_ib_4411.pdf
Charlson Comorbidity Index 0.0018***
Congestive Heart Failure 0.3% 0.3% 0.3%
C dao ys, e nclus merg ion enc yand L departm im ent it v at isit ions s, spe cialist physician visits, primary care physician visits, and prescription drug fills
Appendix Figure 4, Estimated Coefficents From Multivariate Probit Models of Disenrollment From HSA-Eligible Health
C Ch ha arl rls so on n C Co om mo orb rbiid d C iitth y y a IIn n rld d se e ox x n Comorbidity In 0 0d ..0 0 e0 3 x3 98 6* ** ** * 0.0192*** 0 0..0 00 33 96 2* ** ** *
sponsors and policy makers, since enrollment in these kinds of 6p 51 la .5ns 1 is still expected O 3t6 o .4g xrow. In 2016, 25 percent of
Overall, in examining how the use of inpatient services, emergency department visits, primary care and specialist office
CC on og ne gs ets ivte iv e H e H ae rta F rta F ilu are ilure 0.2% 0 0..2 0% 052 0.3%
When the Patient ProteH ctyio pe n a rten nd sioA nffordable Care Act (ACA) pa 14% ssed in 20114% 0, the increa 15% se in employ er adoption of HSA-
Data and Methods ................................................................................................................................................. 7
Predicted probabilities of disenrollm ent by more common health condition are presented in Figure 9. The predictions
K we ey r efin su dm ings: med for all family members. 656.4x
C Co on ng ge es sttiiv ve e H He ea art rt F Fa aiillu ure re 0.0 0.1 10 17 00 .0 .1 01 96
In this paper, we exaC m oine nged s ttiv he e H re ea la rtt F ions ailuhip re between health status, use of he 0.a 0lt 6h care services, and disenrollment from
At least one Plan, suFi rv rs ety D indic iffera etnc ese tdha Mod t lae rg lse ................................ employers seem to unde ................................ rstand that offering ................................ only one type of he ....................... alth plan 21
employers with 10?499 HH y p y e e pm rt ee rt pn e loy s nis oe n io en s and 61 percent of employe 12% rs with 500 o 11% 0r .0 m 01or 4 e emp13% loyees offe *** red an HSA-eligible
Fronstin, Paul, and M. Christopher Roebuck. 2013. "Health Care Spending after Adopting a Full-Replacement, High-
visits, and prescription d Dryug slipfil ide lsm im ia pact disenrollment from HSA 19% -eligible he 19% alth plans, 18% we found t hat while the estimated
eligible health plans was expected to accelerate. Health insurance premiums were anticipated to increase because of
669.70
wereH H g y ye p pe e ne rt rte e ra n ns s te iio od n n from the regressions presented in the0 0A ..0 0p 0 8p 8 4e 0 3ndix, * ** ** * which were evaluated at the 0 0 ..m 0 02 0e 9 2a 1 6n * ** *s of the non-health
Hypertension 0.02
m HSA ini D- a m e tia ligib ze ................................ s r leis he k saeltle h p ction. lans. BOur etw ................................ eeexpe n 2010 ctation w and 20 as 18 tha , t t................................ he less pe -he rce ant lth ay g e indiv of la idua rge ................................ l s w emould ploy ebre s m offe orr eing like aly ................................ C tD ha Hn he P as t alt he h ie only r plan . 7
DD ys yls ip liip die dm em iaia 16% 016% .0036*** 17% **
health plan or HRA, colle Dic atb iv ee tely 3 s known as consumer-directed he 6% alth plans 6% (CDHPs) (Fig 8% ure 1). B*** y 2019, 34 percent of
Deductible Health Plan With a Health Savings Account: 66A 9. 7Fi 1ve-Year Study." EBRI Issue Brief, no. 388
coefficients on the models were statistically significant, the magnitude of the impact was quite small. For instance, 4.2
various ? D D y ys s p lliiOv p p ro iid dv e e eis r m m a ion iill, 5 a a s in t perhe ce nt AC of A.H These SA-eligib pre lem he ium alt h p incrla en e ases 0 0..nr 0 00 5a olle 5 3long 1 8* ** *e * *s in 201 with th3 a e end xcis 20 e 14 tax o swit n h che igh d- c to ost -0 a 0 . 0 .he d 0 0iff 0 0a 1e ltrh p entla tn yp s (t e of hehe soa-lth
Dyslipidemia 0.0389***
c A Me ond ppteho it ndix ion v dsFi ag rur iab ele 5s,. Ma Jurst gina ovl Ef er 4 p fect e rEcst ent im a of teindiv s From idu a Mul ls w tiv ita h no riatehe Praob lth condi it Modetls ions of w Dis ereenr pollm redic etnt ed Fr to o m dis H eSA nroll fr -Eligib om le tH he ea lth
Diabetes 0.0087***
individuals to disenroll f D A ria om sb th e m te H a sSA /CO -e PD ligible health plans. We found 6% 5%evidence6% 5% that individ6% ua 6% ls who d is **enrolled from HSA-
op tion increased from 10 percent to 39 percent (Figure 12). However, the same survey expects it to fall to 30 percent
Figures 4–6 show how the presence of health conditions varies by disenrollment status. There is evidence that
employer(s w Emp itloy h 10 ee ?4 B99 ene efit m p Re loy sea eer s a ch nd Ins 72 tit u pte er)c.e nt of employers with 500 or more employees said they were very likely
perVa cernt ia b of les indiv ................................ iduals wC ho ong us ese tid ve no He ................................ ahe rt Fa ailt luh ca re re were pre................................ dicted t 4o 28d .xisenroll from ................................ HSA- I5 e0ligib .x le health p ........................... lans, whereas 4.4 7
D Diia ab be ette es s Diabetes 0 0..0 11 00 50 6* ** ** * 0.0931*** 0 0..1 04 10 23 6* ** ** *
c Th alle e d p ur “C pa ose dill aof c T this ax”) w Issu ee re B e rixpe ef is c tteod etxa o c m aine use tm heor dee teem rm ploy inae nt rs t s of o offe disernr Holl SA m -e eligib nt frle om he H aSA lth p -eligib lansle . The heaout lth p loo lan k s. for I f HSA-
plan in 2014 and 2015.
HSA-eligible health pla An st, hm acac/or CO dPD ing to the regression model. In general, the -0. 0r0 e0g 6ressions show that the presence of the
Plan, First Differenced Models ....................................................................................................................... 22
AD st eh pm rea s/s C io O nPD 7% 5% 7% 5% 7% 5%
e in 201 W ligib e exa le 9 , m he w ine a hic ltd h p h the rla ai ns ses re la wq te ions ue re st m hi io or p ns e b e lik ats t w ely e o e w tno hy H ha SA e vm - ee p c ligib lo ery te a le in he rs w hea oul a ltlh p td h condit b la an c kd off of is ions thenrollm tful hal r n th en etp ose in 201 lace m who e 4 na tr. nd eIm t 20 is aine p 15 os d a si es a b nr le olle fun tha dc ttin t io the n o he y f th tH hin SA ek -
individuals who disenrolled from HSA-eligible health plan Figu s we re re 7 more likely to have certain health conditions than those
to offer such a health plan. However, only 9 percent of employers with 500 or more workers offered an HSA-eligible
A As stth hm ma a//C CO OPD PD Asthma/COPD 0.0 0.0 03 30 -0.01 0 0..0 04 02 39 9* ** *
percent of those at median use and 5.2 percent of those at the 90 percentile of use were predicted to do so (Figure
Hypertension 401.x - 405.x I10.x - I15.x
d eligib isenr le ollm hea elt nh p t from lans H rSA D em e- pa e re ins ligi ss b ist ole nr ong hea, ltas h p ela m ns ploy is e arss s c oc ont iatinu ed e w tito h r gris ak p p sele le w ctition, h e 0.0ff 0t0 he or 2tn i s tto m m aa yn c aon get rtibut he re is ting o ac do vst er se of
D Sc eph re izs os pih ore nnia/Bipolar 1% 6% 1% 6% 1% 7%
varMe ious thod hes a ................................ lth conditions is a strong ................................ predictor of disenr Figu oll ................................ mre ent10 , resulting in hig ................................ her disenrollment. Sp ............................ ecifically, 5.6 8
2
Fronstin, Paul, Martin J. Sepulveda, and M. Christopher Roebuck. 2013a. "Consumer-Directed Health Plans Reduce The
eligible health plan. We found Pre even hig sence he rof C disenr hiollm ldbi erth, nt raby tes Dise among nrol indiv lme idunt alsStatus with multiple conditions. And while
offe inderp D ing De e ende p pre re a s s c nt s s hoi ii o ov n na ce ria of ble he s in 201 alth pla 3 a n is v nd a 20 lua 14 b, le r easpe nd cttha ive t ly the . 0 0 ..y Fo 0 0 0 7 t7 r 9 hin 5 5e * *a * *k * *c h o thef th y caen 2 ye mitiga ars, tea r is pr kob se itle m ctod ion e 0 0. l w ..0 0 3 0 A 4 3 a nd 9 1 s spe * ** *it sh cifi ould ed for be each of
who remained enrolle D de p in t resh se io n HSA-eligible health plan. Those who disenrolle 0.d 0 0were more likely than those who
health plan or an HRA as the only plan option in 2016.
? Among those disenrolling from an HSA-eligible health plan in 2014, 6 percent were enrolled in an HMO or EPO,
11). Schizophrenia/Bipolar Figure 5 0.0039
A sele pr pov pe ciding tndix ion a he Fi nd gaur lt dh b e e a6te ,h sp ne Vafit rir iR a D s. Fu ab h y ls e s le lu ia p m Me s m ir dta ehe m ta ouc iins a r dm h A , or rt b a hy e s ini ri , tD is pis ublic te ianr l p oll la pm olic n ch ent y oi cSt h ce a a. n tus g The e 2 7 s m ................................ 21% .0p xaa -y p 2 e a 72 r lso d .4 foc x us riv 1% e es fu on tur E t................................ he 7 e8 .g x p r-ow 1% rE e7senc 8 th. .5xOn e of July hea 11 lth ...................... , c 20 ond 18it , ion thes and 23
Schizophrenia/Bipolar 1% 1% 1%
Predicted Probabilities of Disenrollment From HSA-Eligible
percent of individuals with diabetes, 5.3 percent with depression or hypertension, and 5 percent of those with
Sc Sch hiiz z Lo o ong p ph hre re -Te n niia ar//m Bi Bip p Use o olla ar r Of Outpatient Physician Visits A 0 0..0 1 nd 1 43 69 7P * *r * *e * *scription Drugs." Health Affairs 0 0 ..32 1 00 09 9 ( 4 86) * ** ** *: 1126-1134.
Schizophrenia/Bipolar 0.04
t us rhe ec eog 3 se of niz he tes of a dlt th c ha he ta a rte lhe t h st ser re v a is ic t u c es m o s nfl we a ic a s a tsu ing rsete s (i avtide is.e tic ., nc a6 c lly e on r si oss gtnif he -se ic d a ce n tg ion t rp ea r ee l mod t do icw tor hic e ls of h e ). d is m Coe epnr loy ffoll ic eie m rs ha nt ent s v a fe r re om a rde op H por SA tetd e -e dC ligib in DHA le Pp s a p he es t ndix althe h p Fi only la gns ur e , p t l2 a h,n e
Fi rending maine sd ................................ to have common chr................................ onic health conditions su ................................ ch as hypertension, ................................ dyslipidemia, and dia................................ betes. While the 8
Rheumatoid Arthritis 0.0133**
and 94 percentM w ule tip re le e Sc nr Pre le olle rosse d is in a nce P P of R O orar POS e Chroni plan. 0.c He 2% alth 0 Cond .2% ition0 s, .3%
Rheumatoid Arthritis 1% 1% 1%
us House e of he Wa ay lts a h cnd areMe ser ans vic D e C ias o be m ats k em s it etye e d e atp ep rm rov ine ad nt a s of num dis be enr r oollm f hee ant lt 2h 5 .0 -.r x elated tax bills, inc E10.x l-uding E11.x the Increasing Access to
Health Plan, by Childbirth and Less Common Health Condition
Rheumatoid Arthritis 0.06 0.06
dyslipid Rhe eu m mia at o w ide A re rt h lik rite iR sly he tuo md aitsenr oid Aoll fr rthritio sm HSA-eligible he 0.a 0lt 05 h p 9 lans. 0.1416** 0.0050
magnitude of the impact was quite small.
a op nd tion. mar g Aina diff 3. l e 5% erff ee nt ct su eM st rv u im e ltiy p a, le te w Sc s c hic le ah ro lc s inc ul isalude ted a dt m the idsize med a n o com f a pll r anie eg sr,e indic ssors a ta ers t e0 rha .0 e2 pt0 or 7 a*b t*eout d in A 15 p ppeerndix cent Fi of gm uridsize e 3. d and large
differences are statistica Cll ro y hsi n'g snif Diic sea ant se, they are not large. For 0 .in 3% stance, 13 0..2 p 3% ercent 0 o .2 f HSA % -eligib le health plan
by Disenrollment Status
Multiple Sclerosis 0.2% 0.2% 0.2%
Findings From Probit Regression Equations .........................................................................................................13
Lower Premium Plans and Expanding Health Savings Accounts Act of 2018. The Act, which would increase statutory
Frons M M tu u in, P llttiip plle ea Sc Sc ul, M lle ero roa s sriis s tin J. MultSe iplep Sc ulv le ero da si, sand M. Christopher 0Roe .0 0.0 02 3b 6uck. 2013b. "0Me .22d 0ic 6*a *tion Utilization a 00 .0 .1 0nd 1 96 Adherence in a
491.0x - 493.2x
7
Crohn's Disease 0.0079
Ulcerative Colitis 0.3% 0.3% 0.4%
emp ? loy erThere s offeris onl evy ide a nc CD eH tP ha , ta m indiv ong id tua hose ls woffe ho d riis ng enr a olle CDd H P fr.om HSA-eligible health plans were more likely to have
d Fin isenr anc olleial es ha Im d hy pli pec rtat ensi io ons n, w o hilf eA 11 dver .5 persce ent S of ele thos cte ion who remained had it (Figure 4). Similarly, 6.2 percent of
3.1%***
Crohn Re 's ma D ine is de in a HS seA-Eligible Health Plan in 2014 0Di .3 s% enrolled From HS 0.A 3-% Eligible Health 0 Pl.a2 n i % n 2014
1.0% 493.81
C Cro roh hn n''s sCh D D iliidb s se e irta a h (i s se e n poC st ro year) hn's Disease -0 -0 .0.0 06 75 0.08 -0 -0 .0 .0 06 51
Indivi Ba du ca kls g rw oit uh m nd:ult iple conditions were even more likely to disenroll. Disenrollment is likely for 7.1 p 8.4 e%* rc*e *nt of those
limits on annual HSA contributions, was passed by the U.S. House of Representatives July 25, 2018. The bills would
Health Savings Account-Eligible Plan." American Journal of Managed Care 19 (12): e400-e407.
UlcA es ra thtm iva e o C rolitis 0.0080
The Fi There nan 20 cia a 15 rl I e m m som od plic e ea ls lim t ions wit ea r e tof ion re A- s t d ev st o eim rtse he at Se e dd ale ton ac tus ion the ed ................................ su fob r -ou sar m a pn le a ly of sis. polic Fiy rs hold ................................ t, wehil rs w e w ite h 2013 have aa n re dla ................................ 20 tiv14 ely d la atra g e ( i.e cla ., im cont s dinu a............ tao se us t, w e15
Psoriasis 1% 1% 1%
cert3. a0% in health condi Ulcerattiions ve C o tha litisn those who remained enr 0.4 3 oll 9% 3.e 8d 2 in the0 .H 2SA % -eligib J41.xle -0 J .he 4 3 5% .x alth plan.
disenrollees had diabetes compared to 5.2 percent among those who remained in the HSA-eligible health plan. There
U Ullc ce era rattiiv ve e C Co olliittiis s 0 0..0 21 09 88 9* ** ** * 0 0..1 08 14 66 6* ** ** *
Our Da tra and M esearch and etthe ho U rlc e ds e sea ra tir vc eh o Cof o litisthers have found that HSA-eligible health p 0.0la 9 ns attract healthier enrollees and cause
Chronic Obstructive Pulmonary Disease
w als ito h a low co em r tbhe ina etxc ion o ise f h Ps tax o y op ria en srits d eis ns trion, ibutions dyslipid fore non mia- q aua ndlif die iad b e etxpe es. ns W ehe s, c n d rea et pe 0 r .e 0 ass 0 g 1ion is a 7 race pe drd iod ed for to top hose enin cg ond an HSA itions,, and
Non-Metastatic Cancer 3% 3% 3%
do not know if individuals had a choice of other plans. Presum4 a 9b 3ly .9x some of the individuals in our sample did not have a
Ris eligib k sele le in a ction m ll 3 staudy ye y be m ait riga s). t e In cont d by ar num ast tb o etrhe of cot rohe ss- r sec facttiona ors. l mod For ins elst, atnc heese , pc eon rha ta pine s sd ur tp he ris c ingly, in the hange (from c r20 oss 13 -se to ctional
Psoriasis 0.4% 0.4% 0.5%
Ps Pso ori ria as siis s Ulcerative Colitis 0.0 0.0 05 51 00 .0 .0 07 67
w I m ap s no licatd ions iffe rfo enc r P ela in t n Sp Ps he o onsor ri pars eis va s a lenc nd eP of ublic as tP hm olic a/ yC ................................ OPD, depression, or sc ................................ hizop 0h .0r2 enia/bipola 6.9 r %** d................................ is *order between those.... w ho 16
In 2003, the Medicare Prescription Drug, Improvement, and Modernization Act (MMA) included a provision that
fewer healthier enrollees to disenroll from the HSA-eligible health plan. We estimate that this disproportionate
Kruskal, W. H., and W. A. Wallis. 1952. "Use of Ranks in One-Criterion Variance Analysis." Journal of the American
Non-Metastatic Cancer 0.0061**
predicted disenrollmentM je um tasp ta s t tico C8.2 p ancerercent. Among individua 0.3 4ls % 96 .w xith all of 0.3% the condit 0.i3 ons in Figur % e 9, likelihood of
make it easier for spouses to make catch-up contributions. In addition, some services and benefits related to the
m 20 choic od 14e) el, in a of indiv pll inde laidua nos, a p ls A e nd m w nde it ong the h hig ntr e indiv vfo a h rric e aidu ost b w lee a s. c rls e ond These w not it ith n ion be o s, suc m ah b od e le a e lt h a t ls o h condi w e s r a itsi h fir he lyum t ions d st is-ae d t, nr iff oid 4.2 p oll fr e ra ernc te o hr e r m c id te is tnt rhe , ig p ht d ir so is -H ha e rSA ia nr nd si olle -s, m eligi sid d b e ult frle om viple ahe r ia t a he b sc ltle h p le H s p rS o la A si rn. ov -s, a eligib ide Thi nd a le s m d C d he rohn’ it aion a y ltbh p s a el lan.
Non-Metastatic CancN eo rn-Metastatic Cancer 0.01 3% 3% 3% 0.00
Non-Metastatic Can N co e n r-Metastatic Cancer 0.0005 0.0654** 0.0004
2.5%
disec nr re olle atedd a 0. he nd 8% alt those h savw ings ho raecm cou aine nts (H d in t SA he s) H aS nd A-e Hligib SA-e leligib hea le lt h p hea la ltn. h p lans. HSA-eligible health plans have deductibles
Data Metastatic Cancer 0.0011
disenrollm Ste an tis t t of ica indiv l Assoc idua Cih alit ls w d io bn irt it h 47 h th (in : p 58 eo s v 3ta -y r 621. e ious ar) health conditions 2% examined in t 2% he paper inc 3%reases *** employer spending on
F32.x - F32.8x
Conclusion and Limitations ................................................................................................................................... 17
d m is ae nnr ag ollm eme ent nt of is 13 chr .4 p onic e rd cis en eta.ses could be excluded from the deductible.
Metastatic Cancer
reduc M M ing e etta as sttthe a attiic c p C C ea a rc n ne c ce nt er ra M g M ee te a tof s ats aindiv ttia ct iC c a C idua na cn ec rls e rwho disenrolled0 0 . ..0 1 1 98 77 5* ** *0.3 2 0% 0.4x 0.2 0.% 01 06.8 .3% %** -0 -0 .0 .0 0 1 07
ins dise ight ase , a ws t erhe e not y m m ea or su er e lik te he ly im to p da isce t nr of oll fr chao nging m HS fa A-m eligib ily cle ons he tealla lth p tion, lans he a as c lth st om ap tus are , d a n to d indiv healtidua h sels rv w ice its ut h no iliz ca ond tion o ition n s.
Childbirth (in post year) 0.0178***
F33.x - F33.7x
that are higher than what individuals are used to seeing in more traditional types of health insurance. In 2018,
health coverage by 2.5 to 3.2 percent. We made a number of oversimplifying assumptions in generating this estimate.
296.2x - 296.3x
This study makes use of the Truven Health Analytics MarketScan® Commercial Claims and Encounters Databases
C Ch hiilld db biirt rth h o (i (i n n p po o Fi s sv tt e y yC e ep a a he r) r) ild rc be irtnt h (iof n pindiv ost yidu ear)als with dyslipid 0 0..0 3e 3 1m 0 60 6i* * a* * * *disenrolled, 0a .1 long 903* *w *ith 5.3 perc 0 0e ..3 0 nt 6 30 2 a 4 5m * ** ** *ong those with
Childbirth (in post year) 2% 2% 3% ***
Re su Thi b fe s m seq renc a ue ye nt b se ................................ H dSA ue- e to ligib the le fa he cta tltha h p t la t................................ he n se disindiv enroll idua mels nt . a rC e oe lik ................................ ff eilc yie to ntr s ea arce h th rep eor ir tou edt ................................ -in ofA -p poc pe Fkndix 3e 3t .9 mFi ax gim uru e m 4 ,r .......................... a end garm dle as rg s of inal e the ffe c24 t
There were also no diff Inp earte ienc nt e Hs in t ospita he l D a py rs evalence of individuals with rare chronic conditions 0.00 , 0su 3** c*h as rheumatoid
Quan, H., P. Halfon Sund Healta h rSe ara rvjia cn, es A U. tilFong, ization †B. Burnand, J. C. Luthi, L. D. Saunders, C. A. Beck, T. E. Feasby, and W. A.
2.0%
HSA-eligible health p Dla epn res w ssioe nre required to have deductibles of 311.xat least $1,350 for employee-only coverage and
Figure 10 shows the probabilities of disenrollment among individuals with less prevalent conditions, which are often
A consistent fi Rh nding eumatoi d A in t rthri he tis research literature on CDHPs (discussed in more detail below) is that HSA-eligible health
IIn np pa attiie en ntt H Ho os sp piitta all D Da ay ys s 0.0 0.0 00 01 5.2% 0 0..0 00 03 00 3* ** *
Inpatient Hospital Days 0.0032***
( Se cop cond yright , w e© d o Trnot uv hy e n He k pnow ertea ns lt if in h ion o A dna ividua rl yd tie cp ls s, a r e vss o ll r lun ion, ight ta5.6 p rs r ilye d ser is erec vnr e ent d oll ) a e for m d ong fr the om ty hose te he arir s 201 w HS itA h d 3, 2 -eia ligib 01 be4, a tle e s, a F he 3n 4d a .nd 1lt 20 h p 6.1 p 15 la. n. e Using rIcte is nt t pa he oss m ong ful ible l that
Emergency Department Visits 0.0002
he estaim lth p atela s c n aclhosen. culated a The t It nhe prae tis m eult ne t a s f Hn o os rom pf a ita lll tD he ra ey gfi srrest ss d or iff se arreenc re ep m orod 3t0 e9 0 e d .0 .l w 3 x in A 3 - he 30p 9rp .e 1e xw ndix e0 found .3Fi 3 gurteha 5t. 0 indivi .39 duals newly diagnosed with
arthritis (RA), Crohn's disease, ulcerative colitis, psoriasis, and multiple sclerosis (MS), by enrollment status (Figure 5).
Gha 0. li. 200 6% 5. "Coding Algorithms for Defining Comorbidities in ICD-9-CM and ICD-10 4 Administrative Data."
To generate the 2.5 percent estimate, it was assumed that disenrollees did not increase their use of services in their
Endno $2t,700 es ................................ for family coverage. ................................ In ad1.6% dition, maximum ................................ out-of-pocket amount ................................ s cannot exceed $6,650 ............................ for employee-only 24
Emergency Department Visits 0.0085** F43.21 0.0190***
p ve larns y Em c a ond st erg lyH e tn R o cA yts rD ein e ap tt. a rod rt The Em mu ec n ee trg p V p r e ie s osit nd ic ts ic y iv t D ee e d p ra p isk rt rm ob sele ea nb t V ili cit t sion w y it sof dhe isen w nroll orm ke er nt s a arsesoc g 0iv .i0ae 0t0 n a e8 d* * w choi ith ch ce ildb of pirla th is a ns.0 .0 N l0 so inc amelylude , hed a ltin Figur 0 hie .00r1 7**e * 10
Health Services Utilization†
C those ommw erho ciad l C isle anr im oll s a ed nd wE en re c orunt equir ere s D d tao tacb hoose ase, w ae d ciff onst erernt uc tty ep de a of n ahe naaly ltth p icala l d n e ata itset of her beindiv cause idua the ls le ir ess m p th loy an 65 ye er dropa pre sd of
indiv Sp idua ecila s w list it Ph h sc ysic hia izop n's hr Ofe fic ni ea V /bip isitsolar disorder 3.0 9.28 0.0003***
Emergency Department Visits 0.55 0.55 0.57 ***
RA and MS were more likely to disenroll may be due to newly diagnosed individuals being less likely to understand that
Similarly, there were no diff 0.5% erences in the percentage of cancer patients by enrollment status (Figure 6).
Ps0.5% oriasis F43.23
new healt Me h p dic laa n. l C H arow e 42 ev e (r 11 , p ): re11 vious 30- 9. re search has found that new HSA-e5.1% ligible health plan enrollees reduce their use of
cSp Sp ove e ec criia a ag lliis s e 1. tt 5% Ph Ph and y ys sii$1 c ciia a3,30 n Sp n''s se O O c0 fo iffa ffiilc c is e etr V V Ph fa iis sm y iitts s sily icia c no 'sv O erffa ic gee V . iC sie tsrtain primary pre-0 -0 ve ..0 0nt 0 00 4 iv 4 5e * ** * * *services — ty0 p.i0 c0 a2 ll8 y* *t*hose 0.5% deem 0 0e ..0 0d 0 0 2 0 t0 o 2* ** * prevent
given the high associat In ep da c tie ost nt s. Hos Indi pitav l idua Daysls with claims relate0d .3 t3o childbirt 0h .3w 3ere twice 0 .a 3s lik 5 ely a*s those without
individuals are more likely to choose a CDHP, which in turn raises the health risk in the other plans. Risk selection is
Primary Care Physician's Office Visits 0.0002*
Specialist Physician's Office Visits 3.7 3.7 3.6
a the ge H wSA ith -e eligib mploy le m heeant lth p -bala sed n or he b ae ltch aus coe v ear aw gor e. k e The r chuni ange t of d ja ob nas, w lysis w hich r as t eq he uir peolic d ay hold change er, a in he nd the altse h pe la nr n. olle One es w coul ered
they are likely to reach their maximum out-of-pocket limit. Similarly, risk selection may have b 0.4% een mitigated in our
Pri Prim ma ary ry C Ca are re Ph Phy ys sPri iic ciia a m n na ''s sry O O C ffffa iic cre e e V V Ph iis sy iitts s sician's Office Visits 0 0..0 00 00 22 3* ** * 0.0022* -0 -0..0 00 05 01 5* ** ** *
healt the h c onset of are servic de is s ea and se — re dcuc an b e ov e e arnd all oft spe ending n are ;e txe he m re pfor t freo,m it tis he re d ae so duc natb ible le t o and expe cov cte r deis de in full nrolle. es t Oto heinc rwris ee a, se allt he heia r lt us h e
Emergency Department Visits 0.54 0.54 0.54 **
Prescription Drug Fills F20.-0 x .0001***
c cr hil itic d ? b a ir l in th o Indivi em r a pn loy dy u a m ot lse he nt writ - b he h m Pri ased a m lt ult ah ry iple he c C ond a a lt re ch p it on Ph ions d la yit sns ions i ctio ab nd e 's w is c a O e eus r nr ffe ic e oll fr e e it v Ve iinf s n mor o itm s lu e H nc SA e elik - 4 s w e .e 1 ligib ly ho to le is d he e isnr e anr 4 lt olle .h p 1 oll. dla in a ns: p Th la 3e n a .6 y w nd, er et he p *** rreedfo icrte e, dt t he o d cis os etnr of oll a a t
inc arglude ue, d how if te he veyr ,w te ha ret ful a jlob -tim cha e e nge mp loy could ees w beit rh cont elatedinu to ou ths e e aligib vaila ilibtili y tin e y of ahe ch (b alth p ut no lan o t ne pt cions essa . r ilFu y e rtvhe erry )r ecse ale anda rch sho r yeuld ar.
Findings Multiple Sclerosis
When it comes to use of health care services, there were differences in t 4.8% he percentage of individuals giving birth by
Pre Pres sc cri rip pttiio on n D Dru rug g F F Pre iilllls sscription Drug Fills 0 0..0 00 00 12 9* ** ** * -0.0014*** 0 0..0 00 01 00 1* ** ** *
st udy by ded 0.uc 4%tibles in non-CDHPs, as there is evidence that deductibles have been increasing in those plans as well.
295.x F30.x - F31.x
Specialist Physician's Office Visits 3.6 3.7 3.4 ***
of health care services when they move from HSA-eligible health plans to HMOs and PPOs because of lower patient
care serv 1.i0% ces must b Pre e s su cri bp je tio cn t t D o rutg he F ilp lslan deductible. 22.5 22.5 0.3% 22.3 *
a rate of 8.4 percent. Sim Schiila zorply hr, en indiv ia or B idu ipoa lals r with ulcerative colitis and those with metastatic cancer were materially more
health plan. For example, if all healthy employees choose one plan and all sick employees choose a second plan, then
Notes: Reported are marginal effect estimates from probit models calculated at the
I ens xplore urC eo en s w w sthe an etr the e erxc wlude orked r s c if thhe ange y did job not s b ha ec -1 v a.e us 2 0be 7ot 7 of *h me **thed -1 he ic.a 2 al1 lt 6 a h p 7 nd *** la pn ch har -1 m oi .1 ac 9 ce 5 ys a 2 * c*o * vvaeilra ab -1 gle e .5, 1 aor 5 nd 2 *if t *w * he heir t-1 he he .5 r 4a t9 his l0 th p ** *isla an t ss -1 o y.c p 5ia 0 e5 tw 3 e*d a * *s more
enrollm Note ent s: R st ea ptous rte. d I andivi re C o m na sdtrg a ua n in tla s w l effho ect d eis ste im nr aolle tes d fro fr mom pro H biS t m A- oe d -1 ligib e.l7 s0 c 2 1 le a 9 9l6 c *he .* u 1*x laa te ltd h p at-1 la th .ns 7 e0 m 5 w 4 e* a e *n r *e s o tF w f 2 aic 5ll. 9 -1 e in .d a 6e s lik 9p 9e 6n *e d ** e ly n ta vs t ariho abse les.w ho
Overall, 5 percent of HSA-eligible health plan enrollees in 2013 and 2014 switched to a different type of health plan in
The percentage of worPri kem rs w aryit C h e are m Ph ploy yse ice ia -n only 's O c ffov icee V ra isg ite s and a 4 .d 2eductible 4 inc .3reased from 4.1 55 perc ent to 81 percent
cost-sharing in HMOs o Dise nr and ollm PP ent Os. wa In Fr s 7.1 p onst erin cea nt nd am Roe ong buc those k (20w 13 it)h a , w e c ofound mbinattha ion o t ad f h op ytping erte ans n HSA ion, -d eyligib slipide le he mia alt , h p and lan
means of all independent variables. Statisti0.3% cal significance denoted as follows: ***p<0.01;
Figures Non-Metastatic Cancer
lik the e ly d iff thearn t enc ho e se in p wla itn cos h no c ts on wdill it ions be la tro ge d is ae nd nrol the l. se Otla he rg rw e is de iff , etrhe enc ree w s in c a 4.6% s no ostst w at ill is t lik ice aly lly F ps 3r9 ignific ompt a dnt iff e driff en ec re en s in e ce in mp prloy edic ee te d
Notes: Statistical significance denoted as follows: ***p<0.01; **p<0.05; *p<0.10.
Statistical significan N co et e ds e:n St ota etd is a tis c a fo l s llo ig w nsif:i c *a **np c< e0 d .0 e1 n;o *t* ep d< a 0s .0 f5 o;l lo *p w < s0 : .*1 *0 *p . <0.01; **p<0.05; *p<0.10.
w cait ph ind itated iv idua or non ls w -bitah he sic/m al8 a th c jor ond med itic ions al. . After imposing these criteria, approxim 0.2% ately 9 million policyholders remained,
remained to have a claim for childbirth (3.1 percent vs. 1.6 percent) (Figure 7). Figure 8 also shows that those who
2014 and 2015. Among Pre ts ho cri sp et id on is e Dnr ruollin g Fillg s from the HSA-eligib 2le 1 0.2% . 1health pla 2n in 1.0 2014, 6 p 22e .0rcent w** ere enrolled in an
betw Indivi een 2006 duals aend nrolle 2017 d in HSA . Ded-uc eligib tible le s for hea e ltm h p plo lay ns ee a -r only e eligib covle er a to ge op inc en a rea nd sed fun from d a $4 he7 alt 3 in h sa 20 vi06 ngs to ac $1 coun ,046 t (in 201 HSA), 7 i a tn ax-
Notes: Statistical significance between those who remained in and disenrolled from the
**p<0.05; *p<0.10.
reduced total sped nding iabet e bs. y 4W –8 p hen d erce ep nt r eov sse ion is a r the long ddee dr tto er tm hose . If w cond e a itss ions um , ed is tha enr t d ollm isenr enolle t jum es inc ps tro ea 8.2 p sed tehe rce irnt us . e A of mong
0.5% 0.2%
disenrollment rates between individuals with other chronic conditions and those with no conditions.
contributions, which will cause even further r0.2% isk selection. This process can lead to the well-known phenomenon of the
Figure 1, Percentage of Employers Offering HSA-Eligible Health Plan/Health Reimbursement Arrangements (HRA), by
9
each with 1 to 3 years of eligibility. Next, given the primary focus of the paper, we retained only those policyholders
disenrolled fr 0.om 2% HSA-eligible health plans used more inpatient services, used more prescription drugs, and had more
Endnotes HSA Rh -e eu lig mib atle oi d h e Aa rtlh th ri tp islan denoted as follows: ***p< 70 1.4 0 .1 x; **p<0.05; *p<0 M .1 00 5, . xa -re M b 06 a.s xed on
HMO exe orm Ep PtO tra und st or 94 cp us ertc od en ia t l a we cc re ount enr olle thad t is in a fu nde PPO d or wit P h cont OS pla ribut n (Fi ion gur s a e nd 2). a s Sa sem ts t ple ha m t e aa n i nsndivi for d the ua l c ent an use ire sa m to ple pa a yr for e
PPOs, and fromCr $3 ohn's 52 Di in 2 sease006 to $1,175 in 2017 in HMOs.
3.9%
health care servicindiv es by idua fouls w r pe itrh th centose in t che ondit ir ne ions a w hend alth p othe lan rs, d , em isp eloy nrollm er he en atl tw h c as 13.4 are co st pe s w rce ill n tinc . rease by 2.9 percent. If
“ Thi deradt, h s pla pn t iral,” yp e w w he ars p e ar e he -ca od lth p ed la in t n e he v eTr ntuv uaelln d y ba etcaom . Ie t s s is o poss cost ible ly tth ha att it som is no e he long alth p er w laor nst h o weff re e rcing. ategIor f e ize md p loy inceoers a recrte ly .
Firm Size, 2010–2016, With Projections Through 2019 ................................................................................. 6
Kruskal-Wallis equality of populations test of differences in mean value by disenrollment
who were enrolled in an HSA-eligible health plan in any of the 3 study years. This resulted in a study sample size of
Notes: Statistical significance between those who remained in and disenrolled from the
emergency department visits than those who remained. However, the differences were generally small, and in the case
show he n in alth c Fig aur ree e 3. xpeThey a nses. rIe ndivi show dua n se ls w pa itrh HS atelyA b s b y e dne isefit nr o frllm om e nt a ts riple tatus -t aax a s w de vll. ant St agae t: ist Eim cap l signi loyeefic cont ancre ibu of tions diffe tro entc he es in
Multiple Sclerosis 340.x G35.x
w The e afin ssum dings fr e an om 8 p e the rce fir nts inc t-or re da ese r din use ifferenc of e he mod alte h c l yiaerlde e se d rd viff ice es, e rent m fin ploy dings. er spe The nding first w -o ould rder inc diff re ea re se ncb ey m 3.2 od e pl w erca es us nt. ed
0.0%
Similarly, some of those statuhe s.alth plans may also have high deductibles, even though they are not coded as such. PPOs
very sensitive to differences in premiums, then this selection can be problematic in terms of offering more than one
No Con HSA dition -e sligible health plan denoted as follows: ***p<0.01; **p<0.05; *p<0.10, are based on
42 1 2,808 unique policyholders: 242,359 who were trackable from 4.2% 2013-2014 and 309,143 who were trackable from
of emergency room visits, the magnitude of the difference was not meaningful.
account are deductible Re mfr aio nem d i nt HS ax Aa -E b lile gib le inc Heom alth P e la, na inn 2y 01 int 4 erest or otheDi r sc ea nrp olit lea d Fr l e om a r HS nin A-E g lis on gible Hea ass lth P elt as in t n in 201he 4 HSA build up tax
means ? ac Indivi ross d Hu SA als -e w ligib ith c lela he im as r lth p elala te nd d tis o ecn hil rollm dbire th w nt st ea re tus tw w ica es a as lik ssess ely e d a s us tho ing se the w itK h no ruska he l-W altahll is cond equa itions lity of to disenroll
See http://devenir.com/ C w rop h- nc 'so D nte isent/ aseuploads/2017-Year-End-Deve5nir 55.-xHSA-Market-ReseaK rc 5h 0.- xReport-Executive-Summary.pdf
to identify individuals newly diagnosed with less prevalent chronic conditions. The model identified individuals enrolled
^Maximum index score among all covered family members.
Figure 2, Health Plan Choice, Among 5 Percent of Individuals Who Disenrolled From HSA-Eligible Health Plan Between
a tynd pe H of MOs o healtft h p en h lan. av Kru On e d se tkhe d auc l-W ot tible a he llis rs of ha equ nd, $1 alit ,000 yif e of m p or o pp loy u m lae o tie r oe s a n s for r te e s indivi not t of d vd ie ffua r ey re l co sensit nce vs e r iin a v g e m e t e .o an d iff vae lu re e n bc ye d s in p isenro re llm me ium nt s, then it is more
0.0%
2014-2015.
free, and distributions for qualified medical expenses from the HSA are excluded from taxable income to the
populations test (Kruskal and Wallis 1952). In the remainder of the paper, only the findings for disenrollment between
A s shown a from bov HeSA , 6 p -eligib erce le n the of alt dh p isenr lan olle s. eThey w s enrolle erde in a pren HMO dicted ta o nd disenr 94 p oll a erce t nt a rean te rolle of d 8.4 p in ae P rc Pe O. nt. W e assume that
~Amo0% ng all cove 1% red family 2% members 3% . 4% 5% 6% 7% 8% 9%
2 RheumU ato lcid e A rartth ivre iti s Colitis Multiple Sclerosis Crohn's Di5 s5 e6 as .x e Ulcerative ColiK tis51.x Psoriasis
in an HSA-eligible health plan in 2013 and 2014 who were diagnosed with these less prevalent chronic conditions in
2013 and 20 st1 a4 tu ................................ s. ......................................................................................................... 9
See Figure 6 in http://www.mercer.com/newsroom/national-survey-of-employer-sponsored-health-plans-2016.html
likely that a firm can offer a variety of plans. The upshot is that the responsiveness of employees to changes in
There was no diff Se tatrie stinc cal e si gn in u ificanse ce de of nop ted rim as fa olrlow y sc: a **r *e p< 0.0 phy 1; *si *p< c0.0 ian 5; *of p